From ceff1ba5bafe0d961e46035ff387720b35bd2af1 Mon Sep 17 00:00:00 2001
From: tereom <teresa.ortiz.mancera@gmail.com>
Date: Thu, 5 Dec 2024 15:41:19 +0000
Subject: [PATCH] =?UTF-8?q?Deploying=20to=20gh-pages=20from=20@=20tereom/f?=
 =?UTF-8?q?undamentos-2024@0340e4cdd29e55e3c3ada81fb16d2438bf9d6024=20?=
 =?UTF-8?q?=F0=9F=9A=80?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 01-exploratorio.md                            | 294 ++++++++---------
 .../figure-html/unnamed-chunk-11-1.png        | Bin 49770 -> 49742 bytes
 .../figure-html/unnamed-chunk-30-1.png        | Bin 152363 -> 151685 bytes
 .../figure-html/unnamed-chunk-31-1.png        | Bin 179698 -> 180053 bytes
 .../figure-html/unnamed-chunk-33-1.png        | Bin 169812 -> 170911 bytes
 .../figure-html/grafica-pcr-1.png             | Bin 36040 -> 36149 bytes
 14-intro-bayesiana.md                         |   8 +-
 .../figure-html/unnamed-chunk-6-1.png         | Bin 9788 -> 9580 bytes
 .../figure-html/unnamed-chunk-8-1.png         | Bin 19561 -> 19227 bytes
 15-bayesiana-calibracion.md                   |   3 +-
 16-bayes-mcmc.md                              |  46 +--
 .../figure-html/unnamed-chunk-13-1.png        | Bin 40661 -> 42056 bytes
 .../figure-html/unnamed-chunk-14-1.png        | Bin 57231 -> 57704 bytes
 .../figure-html/unnamed-chunk-19-1.png        | Bin 10040 -> 10179 bytes
 .../figure-html/unnamed-chunk-20-1.png        | Bin 13613 -> 13377 bytes
 .../figure-html/unnamed-chunk-21-1.png        | Bin 104955 -> 106383 bytes
 .../figure-html/unnamed-chunk-24-1.png        | Bin 9916 -> 10083 bytes
 .../figure-html/unnamed-chunk-25-1.png        | Bin 17969 -> 18202 bytes
 .../figure-html/unnamed-chunk-26-1.png        | Bin 15940 -> 15940 bytes
 .../figure-html/unnamed-chunk-27-1.png        | Bin 44698 -> 44703 bytes
 404.html                                      |   2 +-
 89-transformaciones.md                        | 110 +++----
 .../figure-html/unnamed-chunk-3-1.png         | Bin 36406 -> 36380 bytes
 .../figure-html/unnamed-chunk-4-1.png         | Bin 116061 -> 116640 bytes
 "an\303\241lisis-exploratorio.html"           | 296 +++++++++---------
 ...251ndice-principios-de-visualizacion.html" |   2 +-
 "ap\303\251ndice-transformaciones.html"       | 112 +++----
 "bootstrap-param\303\251trico.html"           |   2 +-
 ...3n-bayesiana-y-regularizaci\303\263n.html" |   5 +-
 ...263n-por-m\303\241xima-verosimilitud.html" |   2 +-
 ...-y-distribuci\303\263n-de-muestreo-1.html" |   2 +-
 index.html                                    |   2 +-
 intervalos-de-confianza-y-remuestreo.html     |   2 +-
 ...ci\303\263n-a-inferencia-bayesiana-1.html" |  10 +-
 ...bas-de-hip\303\263tesis-e-intervalos.html" |   2 +-
 ...dos-de-cadenas-de-markov-monte-carlo.html" |  50 +--
 "propiedades-te\303\263ricas-de-mle.html"     |   2 +-
 "pruebas-de-hip\303\263tesis.html"            |   2 +-
 reference-keys.txt                            |   1 -
 referencias.html                              |   2 +-
 search_index.json                             |   2 +-
 temario.html                                  |   2 +-
 tipos-de-estudio-y-experimentos.html          |   2 +-
 43 files changed, 482 insertions(+), 481 deletions(-)

diff --git a/01-exploratorio.md b/01-exploratorio.md
index f08ea99..04562ab 100644
--- a/01-exploratorio.md
+++ b/01-exploratorio.md
@@ -67,23 +67,23 @@ slice_sample(propinas, n = 10) |> gt()
 ```
 
 ```{=html}
-<div id="xddpgkkgug" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
-<style>#xddpgkkgug table {
+<div id="qcaxfnkqcl" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
+<style>#qcaxfnkqcl table {
   font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji';
   -webkit-font-smoothing: antialiased;
   -moz-osx-font-smoothing: grayscale;
 }
 
-#xddpgkkgug thead, #xddpgkkgug tbody, #xddpgkkgug tfoot, #xddpgkkgug tr, #xddpgkkgug td, #xddpgkkgug th {
+#qcaxfnkqcl thead, #qcaxfnkqcl tbody, #qcaxfnkqcl tfoot, #qcaxfnkqcl tr, #qcaxfnkqcl td, #qcaxfnkqcl th {
   border-style: none;
 }
 
-#xddpgkkgug p {
+#qcaxfnkqcl p {
   margin: 0;
   padding: 0;
 }
 
-#xddpgkkgug .gt_table {
+#qcaxfnkqcl .gt_table {
   display: table;
   border-collapse: collapse;
   line-height: normal;
@@ -109,12 +109,12 @@ slice_sample(propinas, n = 10) |> gt()
   border-left-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_caption {
+#qcaxfnkqcl .gt_caption {
   padding-top: 4px;
   padding-bottom: 4px;
 }
 
-#xddpgkkgug .gt_title {
+#qcaxfnkqcl .gt_title {
   color: #333333;
   font-size: 125%;
   font-weight: initial;
@@ -126,7 +126,7 @@ slice_sample(propinas, n = 10) |> gt()
   border-bottom-width: 0;
 }
 
-#xddpgkkgug .gt_subtitle {
+#qcaxfnkqcl .gt_subtitle {
   color: #333333;
   font-size: 85%;
   font-weight: initial;
@@ -138,7 +138,7 @@ slice_sample(propinas, n = 10) |> gt()
   border-top-width: 0;
 }
 
-#xddpgkkgug .gt_heading {
+#qcaxfnkqcl .gt_heading {
   background-color: #FFFFFF;
   text-align: center;
   border-bottom-color: #FFFFFF;
@@ -150,13 +150,13 @@ slice_sample(propinas, n = 10) |> gt()
   border-right-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_bottom_border {
+#qcaxfnkqcl .gt_bottom_border {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_col_headings {
+#qcaxfnkqcl .gt_col_headings {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -171,7 +171,7 @@ slice_sample(propinas, n = 10) |> gt()
   border-right-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_col_heading {
+#qcaxfnkqcl .gt_col_heading {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -191,7 +191,7 @@ slice_sample(propinas, n = 10) |> gt()
   overflow-x: hidden;
 }
 
-#xddpgkkgug .gt_column_spanner_outer {
+#qcaxfnkqcl .gt_column_spanner_outer {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -203,15 +203,15 @@ slice_sample(propinas, n = 10) |> gt()
   padding-right: 4px;
 }
 
-#xddpgkkgug .gt_column_spanner_outer:first-child {
+#qcaxfnkqcl .gt_column_spanner_outer:first-child {
   padding-left: 0;
 }
 
-#xddpgkkgug .gt_column_spanner_outer:last-child {
+#qcaxfnkqcl .gt_column_spanner_outer:last-child {
   padding-right: 0;
 }
 
-#xddpgkkgug .gt_column_spanner {
+#qcaxfnkqcl .gt_column_spanner {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
@@ -223,11 +223,11 @@ slice_sample(propinas, n = 10) |> gt()
   width: 100%;
 }
 
-#xddpgkkgug .gt_spanner_row {
+#qcaxfnkqcl .gt_spanner_row {
   border-bottom-style: hidden;
 }
 
-#xddpgkkgug .gt_group_heading {
+#qcaxfnkqcl .gt_group_heading {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -253,7 +253,7 @@ slice_sample(propinas, n = 10) |> gt()
   text-align: left;
 }
 
-#xddpgkkgug .gt_empty_group_heading {
+#qcaxfnkqcl .gt_empty_group_heading {
   padding: 0.5px;
   color: #333333;
   background-color: #FFFFFF;
@@ -268,15 +268,15 @@ slice_sample(propinas, n = 10) |> gt()
   vertical-align: middle;
 }
 
-#xddpgkkgug .gt_from_md > :first-child {
+#qcaxfnkqcl .gt_from_md > :first-child {
   margin-top: 0;
 }
 
-#xddpgkkgug .gt_from_md > :last-child {
+#qcaxfnkqcl .gt_from_md > :last-child {
   margin-bottom: 0;
 }
 
-#xddpgkkgug .gt_row {
+#qcaxfnkqcl .gt_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -295,7 +295,7 @@ slice_sample(propinas, n = 10) |> gt()
   overflow-x: hidden;
 }
 
-#xddpgkkgug .gt_stub {
+#qcaxfnkqcl .gt_stub {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -308,7 +308,7 @@ slice_sample(propinas, n = 10) |> gt()
   padding-right: 5px;
 }
 
-#xddpgkkgug .gt_stub_row_group {
+#qcaxfnkqcl .gt_stub_row_group {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -322,15 +322,15 @@ slice_sample(propinas, n = 10) |> gt()
   vertical-align: top;
 }
 
-#xddpgkkgug .gt_row_group_first td {
+#qcaxfnkqcl .gt_row_group_first td {
   border-top-width: 2px;
 }
 
-#xddpgkkgug .gt_row_group_first th {
+#qcaxfnkqcl .gt_row_group_first th {
   border-top-width: 2px;
 }
 
-#xddpgkkgug .gt_summary_row {
+#qcaxfnkqcl .gt_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -340,16 +340,16 @@ slice_sample(propinas, n = 10) |> gt()
   padding-right: 5px;
 }
 
-#xddpgkkgug .gt_first_summary_row {
+#qcaxfnkqcl .gt_first_summary_row {
   border-top-style: solid;
   border-top-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_first_summary_row.thick {
+#qcaxfnkqcl .gt_first_summary_row.thick {
   border-top-width: 2px;
 }
 
-#xddpgkkgug .gt_last_summary_row {
+#qcaxfnkqcl .gt_last_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -359,7 +359,7 @@ slice_sample(propinas, n = 10) |> gt()
   border-bottom-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_grand_summary_row {
+#qcaxfnkqcl .gt_grand_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -369,7 +369,7 @@ slice_sample(propinas, n = 10) |> gt()
   padding-right: 5px;
 }
 
-#xddpgkkgug .gt_first_grand_summary_row {
+#qcaxfnkqcl .gt_first_grand_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -379,7 +379,7 @@ slice_sample(propinas, n = 10) |> gt()
   border-top-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_last_grand_summary_row_top {
+#qcaxfnkqcl .gt_last_grand_summary_row_top {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -389,11 +389,11 @@ slice_sample(propinas, n = 10) |> gt()
   border-bottom-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_striped {
+#qcaxfnkqcl .gt_striped {
   background-color: rgba(128, 128, 128, 0.05);
 }
 
-#xddpgkkgug .gt_table_body {
+#qcaxfnkqcl .gt_table_body {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -402,7 +402,7 @@ slice_sample(propinas, n = 10) |> gt()
   border-bottom-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_footnotes {
+#qcaxfnkqcl .gt_footnotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -416,7 +416,7 @@ slice_sample(propinas, n = 10) |> gt()
   border-right-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_footnote {
+#qcaxfnkqcl .gt_footnote {
   margin: 0px;
   font-size: 90%;
   padding-top: 4px;
@@ -425,7 +425,7 @@ slice_sample(propinas, n = 10) |> gt()
   padding-right: 5px;
 }
 
-#xddpgkkgug .gt_sourcenotes {
+#qcaxfnkqcl .gt_sourcenotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -439,7 +439,7 @@ slice_sample(propinas, n = 10) |> gt()
   border-right-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_sourcenote {
+#qcaxfnkqcl .gt_sourcenote {
   font-size: 90%;
   padding-top: 4px;
   padding-bottom: 4px;
@@ -447,72 +447,72 @@ slice_sample(propinas, n = 10) |> gt()
   padding-right: 5px;
 }
 
-#xddpgkkgug .gt_left {
+#qcaxfnkqcl .gt_left {
   text-align: left;
 }
 
-#xddpgkkgug .gt_center {
+#qcaxfnkqcl .gt_center {
   text-align: center;
 }
 
-#xddpgkkgug .gt_right {
+#qcaxfnkqcl .gt_right {
   text-align: right;
   font-variant-numeric: tabular-nums;
 }
 
-#xddpgkkgug .gt_font_normal {
+#qcaxfnkqcl .gt_font_normal {
   font-weight: normal;
 }
 
-#xddpgkkgug .gt_font_bold {
+#qcaxfnkqcl .gt_font_bold {
   font-weight: bold;
 }
 
-#xddpgkkgug .gt_font_italic {
+#qcaxfnkqcl .gt_font_italic {
   font-style: italic;
 }
 
-#xddpgkkgug .gt_super {
+#qcaxfnkqcl .gt_super {
   font-size: 65%;
 }
 
-#xddpgkkgug .gt_footnote_marks {
+#qcaxfnkqcl .gt_footnote_marks {
   font-size: 75%;
   vertical-align: 0.4em;
   position: initial;
 }
 
-#xddpgkkgug .gt_asterisk {
+#qcaxfnkqcl .gt_asterisk {
   font-size: 100%;
   vertical-align: 0;
 }
 
-#xddpgkkgug .gt_indent_1 {
+#qcaxfnkqcl .gt_indent_1 {
   text-indent: 5px;
 }
 
-#xddpgkkgug .gt_indent_2 {
+#qcaxfnkqcl .gt_indent_2 {
   text-indent: 10px;
 }
 
-#xddpgkkgug .gt_indent_3 {
+#qcaxfnkqcl .gt_indent_3 {
   text-indent: 15px;
 }
 
-#xddpgkkgug .gt_indent_4 {
+#qcaxfnkqcl .gt_indent_4 {
   text-indent: 20px;
 }
 
-#xddpgkkgug .gt_indent_5 {
+#qcaxfnkqcl .gt_indent_5 {
   text-indent: 25px;
 }
 
-#xddpgkkgug .katex-display {
+#qcaxfnkqcl .katex-display {
   display: inline-flex !important;
   margin-bottom: 0.75em !important;
 }
 
-#xddpgkkgug div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
+#qcaxfnkqcl div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
   height: 0px !important;
 }
 </style>
@@ -528,66 +528,66 @@ slice_sample(propinas, n = 10) |> gt()
     </tr>
   </thead>
   <tbody class="gt_table_body">
-    <tr><td headers="cuenta_total" class="gt_row gt_right">9.78</td>
-<td headers="propina" class="gt_row gt_right">1.73</td>
-<td headers="fumador" class="gt_row gt_left">No</td>
-<td headers="dia" class="gt_row gt_left">Jue</td>
-<td headers="momento" class="gt_row gt_left">Comida</td>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">15.81</td>
+<td headers="propina" class="gt_row gt_right">3.16</td>
+<td headers="fumador" class="gt_row gt_left">Si</td>
+<td headers="dia" class="gt_row gt_left">Sab</td>
+<td headers="momento" class="gt_row gt_left">Cena</td>
 <td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">13.42</td>
-<td headers="propina" class="gt_row gt_right">1.68</td>
-<td headers="fumador" class="gt_row gt_left">No</td>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">16.47</td>
+<td headers="propina" class="gt_row gt_right">3.23</td>
+<td headers="fumador" class="gt_row gt_left">Si</td>
 <td headers="dia" class="gt_row gt_left">Jue</td>
 <td headers="momento" class="gt_row gt_left">Comida</td>
-<td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">11.02</td>
-<td headers="propina" class="gt_row gt_right">1.98</td>
+<td headers="num_personas" class="gt_row gt_right">3</td></tr>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">38.01</td>
+<td headers="propina" class="gt_row gt_right">3.00</td>
 <td headers="fumador" class="gt_row gt_left">Si</td>
 <td headers="dia" class="gt_row gt_left">Sab</td>
 <td headers="momento" class="gt_row gt_left">Cena</td>
-<td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">22.49</td>
-<td headers="propina" class="gt_row gt_right">3.50</td>
-<td headers="fumador" class="gt_row gt_left">No</td>
-<td headers="dia" class="gt_row gt_left">Vie</td>
-<td headers="momento" class="gt_row gt_left">Cena</td>
-<td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">11.59</td>
-<td headers="propina" class="gt_row gt_right">1.50</td>
+<td headers="num_personas" class="gt_row gt_right">4</td></tr>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">3.07</td>
+<td headers="propina" class="gt_row gt_right">1.00</td>
 <td headers="fumador" class="gt_row gt_left">Si</td>
 <td headers="dia" class="gt_row gt_left">Sab</td>
 <td headers="momento" class="gt_row gt_left">Cena</td>
+<td headers="num_personas" class="gt_row gt_right">1</td></tr>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">12.46</td>
+<td headers="propina" class="gt_row gt_right">1.50</td>
+<td headers="fumador" class="gt_row gt_left">No</td>
+<td headers="dia" class="gt_row gt_left">Vie</td>
+<td headers="momento" class="gt_row gt_left">Cena</td>
 <td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">24.71</td>
-<td headers="propina" class="gt_row gt_right">5.85</td>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">8.52</td>
+<td headers="propina" class="gt_row gt_right">1.48</td>
 <td headers="fumador" class="gt_row gt_left">No</td>
 <td headers="dia" class="gt_row gt_left">Jue</td>
 <td headers="momento" class="gt_row gt_left">Comida</td>
 <td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">9.60</td>
-<td headers="propina" class="gt_row gt_right">4.00</td>
-<td headers="fumador" class="gt_row gt_left">Si</td>
-<td headers="dia" class="gt_row gt_left">Dom</td>
-<td headers="momento" class="gt_row gt_left">Cena</td>
-<td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">19.65</td>
-<td headers="propina" class="gt_row gt_right">3.00</td>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">11.61</td>
+<td headers="propina" class="gt_row gt_right">3.39</td>
 <td headers="fumador" class="gt_row gt_left">No</td>
 <td headers="dia" class="gt_row gt_left">Sab</td>
 <td headers="momento" class="gt_row gt_left">Cena</td>
 <td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">18.04</td>
-<td headers="propina" class="gt_row gt_right">3.00</td>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">50.81</td>
+<td headers="propina" class="gt_row gt_right">10.00</td>
+<td headers="fumador" class="gt_row gt_left">Si</td>
+<td headers="dia" class="gt_row gt_left">Sab</td>
+<td headers="momento" class="gt_row gt_left">Cena</td>
+<td headers="num_personas" class="gt_row gt_right">3</td></tr>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">38.07</td>
+<td headers="propina" class="gt_row gt_right">4.00</td>
 <td headers="fumador" class="gt_row gt_left">No</td>
 <td headers="dia" class="gt_row gt_left">Dom</td>
 <td headers="momento" class="gt_row gt_left">Cena</td>
-<td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">30.40</td>
-<td headers="propina" class="gt_row gt_right">5.60</td>
+<td headers="num_personas" class="gt_row gt_right">3</td></tr>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">15.06</td>
+<td headers="propina" class="gt_row gt_right">3.00</td>
 <td headers="fumador" class="gt_row gt_left">No</td>
-<td headers="dia" class="gt_row gt_left">Dom</td>
+<td headers="dia" class="gt_row gt_left">Sab</td>
 <td headers="momento" class="gt_row gt_left">Cena</td>
-<td headers="num_personas" class="gt_row gt_right">4</td></tr>
+<td headers="num_personas" class="gt_row gt_right">2</td></tr>
   </tbody>
   
   
@@ -636,23 +636,23 @@ bind_rows(head(cuenta), tail(cuenta)) |>
 ```
 
 ```{=html}
-<div id="zwocxfeiud" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
-<style>#zwocxfeiud table {
+<div id="phufidpmfh" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
+<style>#phufidpmfh table {
   font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji';
   -webkit-font-smoothing: antialiased;
   -moz-osx-font-smoothing: grayscale;
 }
 
-#zwocxfeiud thead, #zwocxfeiud tbody, #zwocxfeiud tfoot, #zwocxfeiud tr, #zwocxfeiud td, #zwocxfeiud th {
+#phufidpmfh thead, #phufidpmfh tbody, #phufidpmfh tfoot, #phufidpmfh tr, #phufidpmfh td, #phufidpmfh th {
   border-style: none;
 }
 
-#zwocxfeiud p {
+#phufidpmfh p {
   margin: 0;
   padding: 0;
 }
 
-#zwocxfeiud .gt_table {
+#phufidpmfh .gt_table {
   display: table;
   border-collapse: collapse;
   line-height: normal;
@@ -678,12 +678,12 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-left-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_caption {
+#phufidpmfh .gt_caption {
   padding-top: 4px;
   padding-bottom: 4px;
 }
 
-#zwocxfeiud .gt_title {
+#phufidpmfh .gt_title {
   color: #333333;
   font-size: 125%;
   font-weight: initial;
@@ -695,7 +695,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-bottom-width: 0;
 }
 
-#zwocxfeiud .gt_subtitle {
+#phufidpmfh .gt_subtitle {
   color: #333333;
   font-size: 85%;
   font-weight: initial;
@@ -707,7 +707,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-top-width: 0;
 }
 
-#zwocxfeiud .gt_heading {
+#phufidpmfh .gt_heading {
   background-color: #FFFFFF;
   text-align: center;
   border-bottom-color: #FFFFFF;
@@ -719,13 +719,13 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-right-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_bottom_border {
+#phufidpmfh .gt_bottom_border {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_col_headings {
+#phufidpmfh .gt_col_headings {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -740,7 +740,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-right-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_col_heading {
+#phufidpmfh .gt_col_heading {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -760,7 +760,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   overflow-x: hidden;
 }
 
-#zwocxfeiud .gt_column_spanner_outer {
+#phufidpmfh .gt_column_spanner_outer {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -772,15 +772,15 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   padding-right: 4px;
 }
 
-#zwocxfeiud .gt_column_spanner_outer:first-child {
+#phufidpmfh .gt_column_spanner_outer:first-child {
   padding-left: 0;
 }
 
-#zwocxfeiud .gt_column_spanner_outer:last-child {
+#phufidpmfh .gt_column_spanner_outer:last-child {
   padding-right: 0;
 }
 
-#zwocxfeiud .gt_column_spanner {
+#phufidpmfh .gt_column_spanner {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
@@ -792,11 +792,11 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   width: 100%;
 }
 
-#zwocxfeiud .gt_spanner_row {
+#phufidpmfh .gt_spanner_row {
   border-bottom-style: hidden;
 }
 
-#zwocxfeiud .gt_group_heading {
+#phufidpmfh .gt_group_heading {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -822,7 +822,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   text-align: left;
 }
 
-#zwocxfeiud .gt_empty_group_heading {
+#phufidpmfh .gt_empty_group_heading {
   padding: 0.5px;
   color: #333333;
   background-color: #FFFFFF;
@@ -837,15 +837,15 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   vertical-align: middle;
 }
 
-#zwocxfeiud .gt_from_md > :first-child {
+#phufidpmfh .gt_from_md > :first-child {
   margin-top: 0;
 }
 
-#zwocxfeiud .gt_from_md > :last-child {
+#phufidpmfh .gt_from_md > :last-child {
   margin-bottom: 0;
 }
 
-#zwocxfeiud .gt_row {
+#phufidpmfh .gt_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -864,7 +864,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   overflow-x: hidden;
 }
 
-#zwocxfeiud .gt_stub {
+#phufidpmfh .gt_stub {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -877,7 +877,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   padding-right: 5px;
 }
 
-#zwocxfeiud .gt_stub_row_group {
+#phufidpmfh .gt_stub_row_group {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -891,15 +891,15 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   vertical-align: top;
 }
 
-#zwocxfeiud .gt_row_group_first td {
+#phufidpmfh .gt_row_group_first td {
   border-top-width: 2px;
 }
 
-#zwocxfeiud .gt_row_group_first th {
+#phufidpmfh .gt_row_group_first th {
   border-top-width: 2px;
 }
 
-#zwocxfeiud .gt_summary_row {
+#phufidpmfh .gt_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -909,16 +909,16 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   padding-right: 5px;
 }
 
-#zwocxfeiud .gt_first_summary_row {
+#phufidpmfh .gt_first_summary_row {
   border-top-style: solid;
   border-top-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_first_summary_row.thick {
+#phufidpmfh .gt_first_summary_row.thick {
   border-top-width: 2px;
 }
 
-#zwocxfeiud .gt_last_summary_row {
+#phufidpmfh .gt_last_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -928,7 +928,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-bottom-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_grand_summary_row {
+#phufidpmfh .gt_grand_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -938,7 +938,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   padding-right: 5px;
 }
 
-#zwocxfeiud .gt_first_grand_summary_row {
+#phufidpmfh .gt_first_grand_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -948,7 +948,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-top-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_last_grand_summary_row_top {
+#phufidpmfh .gt_last_grand_summary_row_top {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -958,11 +958,11 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-bottom-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_striped {
+#phufidpmfh .gt_striped {
   background-color: rgba(128, 128, 128, 0.05);
 }
 
-#zwocxfeiud .gt_table_body {
+#phufidpmfh .gt_table_body {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -971,7 +971,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-bottom-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_footnotes {
+#phufidpmfh .gt_footnotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -985,7 +985,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-right-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_footnote {
+#phufidpmfh .gt_footnote {
   margin: 0px;
   font-size: 90%;
   padding-top: 4px;
@@ -994,7 +994,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   padding-right: 5px;
 }
 
-#zwocxfeiud .gt_sourcenotes {
+#phufidpmfh .gt_sourcenotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -1008,7 +1008,7 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   border-right-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_sourcenote {
+#phufidpmfh .gt_sourcenote {
   font-size: 90%;
   padding-top: 4px;
   padding-bottom: 4px;
@@ -1016,72 +1016,72 @@ bind_rows(head(cuenta), tail(cuenta)) |>
   padding-right: 5px;
 }
 
-#zwocxfeiud .gt_left {
+#phufidpmfh .gt_left {
   text-align: left;
 }
 
-#zwocxfeiud .gt_center {
+#phufidpmfh .gt_center {
   text-align: center;
 }
 
-#zwocxfeiud .gt_right {
+#phufidpmfh .gt_right {
   text-align: right;
   font-variant-numeric: tabular-nums;
 }
 
-#zwocxfeiud .gt_font_normal {
+#phufidpmfh .gt_font_normal {
   font-weight: normal;
 }
 
-#zwocxfeiud .gt_font_bold {
+#phufidpmfh .gt_font_bold {
   font-weight: bold;
 }
 
-#zwocxfeiud .gt_font_italic {
+#phufidpmfh .gt_font_italic {
   font-style: italic;
 }
 
-#zwocxfeiud .gt_super {
+#phufidpmfh .gt_super {
   font-size: 65%;
 }
 
-#zwocxfeiud .gt_footnote_marks {
+#phufidpmfh .gt_footnote_marks {
   font-size: 75%;
   vertical-align: 0.4em;
   position: initial;
 }
 
-#zwocxfeiud .gt_asterisk {
+#phufidpmfh .gt_asterisk {
   font-size: 100%;
   vertical-align: 0;
 }
 
-#zwocxfeiud .gt_indent_1 {
+#phufidpmfh .gt_indent_1 {
   text-indent: 5px;
 }
 
-#zwocxfeiud .gt_indent_2 {
+#phufidpmfh .gt_indent_2 {
   text-indent: 10px;
 }
 
-#zwocxfeiud .gt_indent_3 {
+#phufidpmfh .gt_indent_3 {
   text-indent: 15px;
 }
 
-#zwocxfeiud .gt_indent_4 {
+#phufidpmfh .gt_indent_4 {
   text-indent: 20px;
 }
 
-#zwocxfeiud .gt_indent_5 {
+#phufidpmfh .gt_indent_5 {
   text-indent: 25px;
 }
 
-#zwocxfeiud .katex-display {
+#phufidpmfh .katex-display {
   display: inline-flex !important;
   margin-bottom: 0.75em !important;
 }
 
-#zwocxfeiud div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
+#phufidpmfh div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
   height: 0px !important;
 }
 </style>
diff --git a/01-exploratorio_files/figure-html/unnamed-chunk-11-1.png b/01-exploratorio_files/figure-html/unnamed-chunk-11-1.png
index 34032e0582a1046e66dfd5bd7b5b7c2751a10592..4d98c488cacb2a2b5fafe770a31a0294f9b3c919 100644
GIT binary patch
literal 49742
zcmd43^<Pw77dC!om?4G^2?0S`TBRFNy1PR`QaYsuK@>?zrKJ_6L}?g732Bh-knWI%
z_Y8PH@BKXAKj8cR;AcF;IcKl6_lj#>YwrnDSCzv@Tt@%^fUh8bUlRZ@!T(|a*cjl)
z0JiZt`a)e<`vG_ffk2>8C<X=w3<iV4;h311SXfxt*w{EYI0ys+7Z(=~4-X$7pMZdX
zkdTmwh=`b&n1qCcl$4Z=jO^OAYvknQ6ciMcl$2CdRM)Rxr>3T+p`oFrrM+?E1|1z8
zJw5%+n>QI47#JBDnV6WEnVDHwSXfzE+1S`_-MYoj&d$NX!O6+V#l^+V&CSEZ!^_Ki
z`}XZSckbN1dzX)okDs4kKtMoHP*6xnNLW}{L_|bXR8&k%Ok7-CLPA1PQc_AvN?Ka_
z-o1M=GBWq?-+%Dnfvl{ooSdAzyu5;ff}*0Ll9H0Lva*VbimIxrnwpxry1IsjhNh;b
zmX?;bwziIrj;^k*o}Qk*zP^EhfuW(Hk&%(Hv9XDXiK(fnnVFfnxw(af#lwdWA3b_x
zX=!O?W%c;+V{2<`8yg#2TU$FjJ9~S32M32IPo6kBIyyNyIXgSMxVX5wy1Kc!xx2f2
zczAevdU|<zd3$^N`1m}1`qbCg*U!(--`_tVARsU>5Q#(v1qD5O_AEF!I3y(G`Sa(Y
zp`kBcym<NYWms6)t5>hW!^0yYA|fLrqoSgsqoZSDVqU*~9UB`P7Z(>FAD@trkeHa5
zl$7-5&6~Gx-zFy~r=+B$rlzK)rKP8*XJlk#W@f&7_bw|dD?2+oCnqO2H#aXYFF!xO
zprGLW`}ZF{d?+j|{P^+Xr%#`Xii(Phi%UvMN=r-2%E~@}{#;&O{^iS;uV24bR8&+}
zR#sJ2RaaNn)YR10*4EY4)z{ZIG&G=4sBhoCH8wUjH8p+z{{6>~AI;6pEiEmrt*vcs
zZ9jkhY;SMx=;-L|?Ck35>hA9D>FMe1?d|L9>+kRX_3PKbz`)?(;Ly;}@bK`+$jIpE
z=-AlU`1tt5#Kh#}<kZyE^z`)1%*^cU?A+Yk{QUgy-@g|Y78Vy5mzI{6mzP&oR#sP6
z*Vfk7*Vi{THa0gmx3;#nx3_n8c6N7n_xASo_xJz&`Ezh^aCmrlbaZrle0*|pa(a4t
zc6N4tetvOrF<WBd5BdO(v%G;D0MNcb|AR#T-nIe&dO+d6w6=HZTAG(R=>Q$7p+`kY
zFeNt81Va3ivz+J2EvsAyHed3XIz(=KeVH~93@)wVe4!5e7KD3^H7n~GHT(fFtdWyi
zakrg$xq3cf@7Psgqdr4&bbO-ScPhZwCu966As+TK`j_kwYb+!^_A?<YG61|(l!ky8
z1!Vv4U%WN}d|4)`7ib5pW?MQ)qA~Vro&+4pL~mu%V#-{1kNKr5WU2Jj%tr`!t9)Be
z0>y57ZQM`dlZCoedq82aOQh1Bz6%nV^f=yI_|}=fy8Nm2*KnF*M4i)HvvZ1voChDn
z5z~j~7>&iR5YwMp8$*5te7zSG?<@L9905pc2iHQq`>E+Oc}zqDk&?&*EP?jDXQqN&
zJbD<GH!7#Ipsk_KGkS3N)_p5v^IhG3i)UD-Me_8&C%g2cp;ANx^o79WK4vuBCogG{
zLigE+sS&o%*PfF7xd>HI-;LTjnSJjTI~peg!1Gmc79E*9Q!>1zmUR0IjiQf&FVwDK
z=ZueO0N7cc^JG${TwC6CHp@6jQvpO>*|u!&>UK}<J^Y_H0o=aoPJ0hj!;)CX1t6;@
zMs!k<4G`U5Z@q=|0S=P%*I?=Suab7uj2gb}!<<*`q-T_FK)q?qO}90bu&Z=tq&}jp
z;UP5@JcbVQrnM61W*XnJ(o6+wt^M$1-!9I~{tHrtNaT`fO~_qI+Xee<W*#V=V)3RC
zSAfZka8_)t($3@P>k_X?2`F%;fBqzR2RTqZR-w^oTTz8L2o_JkojYXrAzOpLw{^aq
zq?ZO#g?!dx8N6QfvKaF6dQD}>ulKHM4c(new1A@Nkqsh%xweA+&JJ!$P<&A=F*YL|
za|K<-=r764<zG#+`?-@ZE+;k{kigWpw6D_uRwGSbYXB9YWRt}G;!}yhjXmCCyN4?j
zN=e)ce`t0M8-_TchxJ814|1^-1&u$56b9%S`vbKPtDd)6n^8fD`2YC4%#9t%`_TtI
z#4`V=Bk~bW|KVz2kz<p03qz*%_*S)qa3uW)X;}y-W{J;k1FL%UD4~O>Ew31?=Wr&i
zSo{~|G_i%!%q{wGARS&Hyg86xLtW!ABMfm_zSoQqq8zXhq&mI$gak$Ph)(RVP4CV#
z;0C8bk-LZPSyAONP(avA{bzFnpNN^$s4Md&jNWdoVP#s<e|42`5aO*hY-|YdjIP*!
zf)clMV~4zrZd@<M@K*C)7{QqYMBBaTE^Rde${VmKW?^5)>dj)pSJx+{)h~W8iQ%Ts
zp@|_dw0duEs&N*ITy%*4x}{}hv-4v6AkDUUv-}y57qrN2(PXBA+SGRu*7z$D@X7Xp
zBr+Q_5x5`u1ap{?^TvMY?Z(Y^3Z6-Bf*ENft?qIO;y#TeBayBIqGD0-`IdiUc(=?<
zy5FA@b<k$7Taro90qspiEr@R6H}%85_pEl;qXijOBp!3~j*jmpk)u``ow_Sz%;O{b
z$T5L6!8*3e7A9mvtwoei`wIX!yglp#=|PR|02Z>PJMXs_FnO03M(^lw-8+S6@Nu-M
zUD|75V9=}a6-Mc+3}S2!y60n&SYoTM&UYw=H)SC7FYRJ9fadfOSA2a=q>xK*`d3&F
zezJi;!)Xt$fTB?WjtOUR`W>8U7vx;xwMZ!=V^!_6Nc-a4LP9pYT?(WZJ|5&HwGlx{
zSla8i0M;kne~r_Bb9i>D9l{%SCILp!0!$w@$lHdh>)etSS&!ZUvI%TDTbtG?$H80>
zKIDw9M<VAF7Q$B>XJQUdg-KV)99K=s+DUb+wE*TqLZr(=LPNq$z~xcH2}ZfWAD``P
z{9OIV-M8oq52=E9cQney-yOQF=*Dye{IkmX7S-ALurpk^x_iaBnwHV=+i3icaOW{n
z^7tZk?a>Y|z=)HlI0e>&tEgU8kXk8X4|(~4kWDO5I}!MJeXJklUZa{6ndD$;c^CDQ
zBpg0gu~|y+{ez+`eS{y`lW1glBftMUi4%V|0H%R|2VpLW>e1vdq(GRzwfcb|1U3hQ
z-6pB+WdJ>ht5Thb@n}ak&`nlf#DJaoq>X^=75BGSg`Y`~mB+|?Y6;B1YO~`SrHPie
z$qD0X1f~6Tou{Q09`w38cc2#Y*&#qESrHZXmy!Q0ScN@Ot*yoX;yn`y>P2&Lrc-iP
zqH>uBqTuV1Z1TO2iYZm5@0XCX5zE--MDFEOQ3chr0bb$nr@nlc<c178t{vo8V}xXU
znN+9nyWlb%g-LzFDE;frs*q~~{g~K)6>}wkm%_~b!hUEI;h4h?4VptKos<AW8+)I{
zG6C@?@n6-)+@Bdu`^N#~(vgiCq&vGxVvHOgx`sFF?MzdhR=hras89lCo1<ty<2|Y%
znC4%N;Do=?ul+?lZL})&an>5v_){({95vL*Q3Ssy^3SM<X+JKc+^f(MpX=YD@@%y{
z`O^|cP<>$Qx$SloCgv#H6ANsV-h!TqGISVT(q8+q*l<TcUBja}z$?;r`*HkVO<z(M
zY6(qG6#qzKQFOjdS^G*AU!1U~&z$z9&S`{{{t?@_qgcAWlWFN7&7HZcl`JE`It#Eu
zTJDNRF17%_h;is?{Z#T_7O6pAlAFepqpHj+H*-pC#A3aF%G%;K_(+v+`Ncc;8Z5En
zj@~O+#+gqMDR9rJOf9Gug%B~Oe&%Tfv>NPn>O2~#88{Qjk(IY;56^9$YO$brn$3oa
zjBoKf{2)NGN2d@dA=3Bwrv`ZlzOx6&(tq?Lb8MJ^eR$h(Xhq3a-}%?BYe^eVyKzy)
z)*HtWMc>qYnq*%Co>0EAr#73}``_r;8&a(K$-dhm4M#-HmauwmP050h6%x|vZTm!T
zD@?#6;e<VCs5R2*lda*3Ft3Y8Xgc6hOgBD?&z^E@Qn>Qt%#mCB>3V+LENjq%$U2<q
zyX!|@A=@mm&mYnCnmgx-O5c3A!)X8bdc%fHlx+eV(WU$CT8wa$ZKWOeDQWp}z95S@
zcn2m{{#G&OdFuo6$*#UbjHd3{8f3G<hDT~e=i%q*1ORW<zXQH;(yqIFugJ(Tb#Ef^
z@dmqO!J}KkpWJzK3s3TsnadIuz#LRpIpr@iY$r~cT2OkF!1=|KH`3FB8E6Whc}}%G
zWn;=naNn4jk?1iPU8x7Xla)d?IKP<gT?L#f*5}<*{8<X-37BVJCwVCmH+&TIbvS{h
zE^cCp*K+~9<8LX~1SbrE(BP)kqx+t|VpEToADXUfOw@F>X5f)SD7(7tiS@ti+Y}ys
zRn+rxR8<N@W-fLa9-C<F&jt*Y7|&@fK3JW}d=Tj`_)*73D!jg?Q)v*&Yiv5ik@d=H
zy@(_Hu{bFf1KPDj!N`l>pyIf;Ic2%hkhPyXb(Nwk0OQm-to8KRAElMbAlzi5`2Qbn
zQV^liPEVmX_P5^upu~^1Gv-K!s#0||G-abd7f2&lQLDkzU1a72^PNGs*z{_}4V~FH
z&iHS9G5)pcv&Z4JCn)^up-}PAZoz^;KHh3K_=yRA^0*c>Kd<~aH0A1v*R}A$G_Yyp
zwonT7(=(ls>oSm+Qjij9h!6<35mEr09-j;#B}0lLp;8d+Odux+YW@uBjHM2+0T41I
zpE{td4r~Aj@J1vw1`~nAL7-@*AejJ69q5yS@Jd4}G1veyB>1lDYyb-z;03^uP$DUa
zJO&&EgHID=0_p&i4Y17w-edyfkpHI#C1XQ!$Pv=WLfE1hHv_#J(X&>&NqE|pO<0CY
zhQuZVsE3GAf!&J>mX>0B;?d5=m}IEQZ?|Qe*;l6wRd-D9Zj2akNVFy83t!!xM2vD$
zQWy&I*AnR$#Px>FGZdHtYTw`3o>YHkbh%-z8++eGGA+^ZwY8bl<+`|BgBTm;(I!1Q
zP3NoPI7^h1;(j&?-?CM{xgOx<$=StlaXa=a${WU8OyRGUh^YZEr~?kh4}i^B;`r<E
z7-6LvS?L#m9R;d@0rH*G8GK@vXkF=ZBs9~&gxoAKTlz{aJz|u|I`lJO!!F**$S@@J
z8(KM>rw@D|`6d)jYx%(daHkY!lM7}8h}i&vA_nAd6Sg!#i@uM_A|dQcmfuP-uy#^u
zgGq&Hw4es{XVvsZt$yJ^UmL3{-%BVFXhC!AP;a)^ebT`C-DUG**65eU!KDD5kSHeW
zj-CRfVBZ3{C?EO&m}^^f<!gNa@T_v4a&Vdi+Kr<+lGYo-ao{LSxOsSf8_g)LREk2)
zZ_`=g*DCB^^^{O>IK4H2<9G6d3lXd@H{k{c!4sCx9sn527IOTWB-~JaqdO0dmqjL=
z(;R~)<UWJqvjIka3MUIpz+Nkjn!ql0tH8CfLgwjb&=`KK-EjC;L2z=+vyZ|;vNc?M
zIOcht&p9bVDzWLc4o*Bs8@Z?qs+H?zzFSUQ_nnG_ng>Hmeyn@sIB_5+ZAY0~d|;L-
zuhs`hoTVYH$-xbCh?5J)DA6Jg$(;?X;+`8?3y(Y0)MA$jthu3-pK46^j`b#pzD4C9
z=cJeJPF%{CV(gu&i)Az7bHpPyYoT6RX9(qH1AN@JJhFbndWf<3>2ZAAoOd5wr%NJI
zBX4)bc@qsd=0&lY4u3Rya$F~M5Sr#BhxX^;g~6m=VAr$`#YlsBZ5-6_DKthDYjJRY
zwa210kt6DZ8Nzijb2^`EWcSAS&7qg_Vz8}f?riUyN)<zUCptCJv7jlb1DqT1!<Ki!
z-*lXd6OpSpQs*?4lQx+<i>P&3j5EUuzms1<2gPFSoDqX(HilJtf)(2DPFFd(DKAVH
znq4!y=QP4g+v}R&CEFjj)7Vu7s1`43bn2*54Q9+~bXwh@D#Fp(TYPIIJf3SYWcs91
zW4j^e<!xZo^<z}<?t1$-l_+VmzTKl~{b^^~fqI(pS;{@Dx^t42m7nJ<KV<-?CHLz3
zZ1-x3MoLu-j}~Q)hF%qJou#}R(v>dFxC}psJ7&y#it0w4qvodg5Yrw}11>_ZADR(x
zJL4tarEwfA<-zjlI#(Kr_fL<10SGLI`uU7G|F$*Rl*9{_5ZP@yl3Z<0U=M*x3Aw^D
zb*NH=u~9{JhBMF-9;~0_NPl7VFTnZfpkLOp#e{3H#r@Ze)`{%#fsxN$E~*;G^$GTp
zynnfx7B~ESmt1(M{^K<>Z*8~dTA}-k-r8|3H=AAnvHOc}Q-o8`OYDmUjElF-G&?Af
zUP8JpGkY)mg)6_^UJ7qkO?l)&!~K0EDL92DSm5V81E_v|UZWe^OF9@=SI?QH;c%gm
z!{XQ$121(gcE_|ElFuhEB)$@Kk9}?^1%nxZ@g{@w&7I$<b@0;FaHs+$<&lw0ZrcB;
zV)x0V*<p%8@_sS`VZL8r#(DsFrw5$gj9^>^(3|pZ7h^l=uBK{CD~o<hUf@`fws~n_
zyu&alT6W~C4ba?$AwoB?xLE<#dDH|<o!8h4wfC^jG5Fh7wqwo~UG1;phB}IOm*#|*
zi1E>_0AQ~Mh5HSg4xy<`1yW-$yhxE2g^q{P!86C<?`59jcDZ?Q?BWZOFRk6^zVv>S
zIScB7#uriW3@xH6pxPSLj)TGxb{2)qJM}JyMkx*8^S|Gjcb=GzIIj*9y3PisEWA)?
z?WR6N`~RA5IxycIG=_sRKNPWELJ6xGN<=RXMsA2Yem)C?B9Qn9r1Tw>S5!T=@jKA-
zidcaoH?-eKVmY!Q5nvz-hMGImp{91m;Y@TzY(SOE14t|WRjT_*Dn+~1pAE!gp_Ah^
zFpK`iL1FQ+0q|#FDwT)BQTJV-hlKQ)@JuPy!2K_btUk`TFo>~)OY^&3HHx6CK*<0-
zv|jxf1VpAE(ZVx%>uLL0Sn?*yr_yL&M39{xjimFaB^r9jT(lV~F7FJ{`V?y1{2P_&
zk<VmC#8+W(sCKl;Ax5-~G1dU?^mBoQTQ|qvv*HJe%i_FqZ?9^+=5>B{-ag<q`*d$J
zPG0wX@es_&muW)^kV|A8Bq%h@CYK7~78uH1Ao$1&=8p45;h9IHXX|VFma|(VeZJ#4
zz2dEi_QlgAxm&cfW0@J-*yb}tJJ>IXO*WwjDHsA7xeMPiQEe(ILH11ZB4cs_(h>&D
zK9~S?;V`1RZ6l?vP5TL>IL$@r3wX>aDvSjF(h4gt2Q9wLVd4~G^_&q9Lg4gJII4LW
zDpZjplOmhUs}e#d^=X(mO>hgl#Y(*j*cm0m%}o0o{Q3okJ1>8K9O5%7qg5qU*801k
zqftf<wO$uXx2PEGD!t|7nhPrQ@(pR@{(&Jwi7~#k`-IHhDKG}ie#A>ft!8vtW4(1G
zMp2e^SJs*=9k4bFhCiVj>8x}mcA1Znd3-fJsoY2pOJtsvJoBRi!i3}pdn|-ISF(5}
zkfCg?<rs0h<M6_P4c1NSBs~_+EcaLqz$LQHPw{h`$cq;s2tp@9cZgUFpK`zU(s|vP
zy<qgxfDb7eHYDIq2ggFE%d)G)ieuiALn2r+uHzk5H6ENBY~w$g+aUhhL4V=qjeWJ&
zsuZXXAz?W;ka}Dy{3_YQhaKJCR|*9tkO4_*(e1lbv=+otW>0^av*{;u)~ZI?&{C)@
z^rUJZ*EPxfg%;I7I;&Pq2*%;yM+(h+NB0)2gaH+BQYuC%4lYbmX(Pmpjgy7ie_g0C
z*+0h$%wTeR{(Si9!2@PX>pf<*t)*8#G5@nl^1dn#*hC~to9o3?b!%*B<4eFr#h7p+
z^HXWVmDjRy7$0}0NXXaB7x<KV5#)}@i0BYTDcs!KR*H%S-Q{Yr*@o{u<s*enQ(qHw
z8fp+C3&|2n(;mJ!sM}c6pMN-|`7oqN%)zj_?)6fh6;EpXc=_1rc#EXh=_7MCfCCIK
zsoX%*mqnW=fJ(*;k6Cqa@V8r$O^I^fWLtS*&kAD>Mg|v6&+<w>xB4cNkkF-+&VKjc
z&KRx<zFHbT2qhw59lK<HclBPV2B0{1{Po0lKfTl_ufF?Q)$UyTs;=aDkso7L7H~fk
z7_LNu@r4_>z-|EVg)L&Bj)0tJQ1dJ%NI8@@67mA7&VUBO+8|<%XF@)mesmo9-(LX*
zPyXvGGA3jSz@^Ni4!#b=%rOWAGVB@5nFRU9G!ux&{wJ!yai-;v0eKW`0P}wWi!}5Q
zMnAC!g=gMJ1NHYHxMq_FiXk0>by5&EVl>ER2hli!1f>OVsQk|eCD3Wxeh|)*NP{r`
zQU!WQ2gb1N)1ZIJ3ktUlWF-HOB$+%Ywf~jmB^GZqdGvojx&SP6Wc`;8ndI914gW10
zlF+8(zeIC|Ia6hRIl6|B3PV>7n$i$njR>G<1G40BZln$%m^RZQQ~rDy2NoNiuGUE+
z=I!vWENKK<dA89P26XDul98(euNW`FDk&Ki#w=IQjicLn79Y1fwtR9Zm-pwdXLBOQ
zy8Ciq$YaCfFbF(TY@;Kxv}APTolvBbz5LKS#=fU9%nj}R^jEPh4!(u6US$qQokrx0
zld=K7OU3hE`gcYBFgT6IGmcMt{W_t@<Xf(24ab$Yc3nfICE)DZsv*flg4RHtqXT|G
z->j_Zr+ybsE_;c9X4<+paB^K)m5?5gZ$qa@3LN*Xh3K#WKV0%EFQS(V`U-)mKH^`T
z$hVe-|GHN$V(OKRFY;Zyr-?BVs(!s8fk>)@*;;OnM&jl&Im9wH-_n-&YEr78LaJEm
zA0TgJZD)hMVVOebw|XxsewQ^=)153+h64++i&p>l-pbPQ_ADB@t+sb}TS3(P!$tI+
zK49BL(5yg{UB?BbRQl7}m*od#Gq)f^aiNi3!_RE;Y+y%NJFd9l+q*<)HUJ`>a5U2K
z(RMR=7+UfAT|dCE+?c-Sd$Z{{e637%s*u^740R)YzsiU}mkf6rZt78OK(Ir{OuBq8
z8@N+FpFXQ?Jt7>V0toJ7Yu=3N8`cFr$uj>al5|VT+8MuRJx|lp;6#R6r75go&;v0Y
z;?O3#@jS3qK~-m=Eg@J;6++j#_RFiU<Ce_q-LGyRGy>mspsYwyY1Et%C*3;#;L~l<
z0}p}$;~}#d^>+M67meA;_NjM1O15Mtw`<-#?PjEfBsW_(I4_6BSe*GhZu6`1Tutda
zCrOEbcqcvbC`*%V_`xyDXuEJ8ID5;K|7T+GSS`c3g@LEBgTvxlh0oqS7RLlCAyW*Z
zpdnyb=t#QcnWTgbv}C@sL0QQgU&?2EM5J74XQ@KxDPh6PLiA?CejD9Mx>4PR3OAjv
z@84j(V<NT6whb+oJHNw|1(3B2{CJ{52>GbBC@zJir+yE6VuvmYlN|>{9n9F%n^rkq
z*QMADT)&q`fJPzMLSc%bpV|Bfd8RFD)A!3oyj4aH^e3=f*FXY0bc~Od>5-nxqXT)L
zli7gWZRKqfsZC)~@+08}J{J-G*f{bF7(xnx4j@FR&2paHZQ#8-a3-w}1_(;HcsM*}
zW6gYi5#!FQWc^<rlf-05kPJ{Jqel|9!W5|?%2{!3MvVwEWaDA+_c+wLk00hlxV3yJ
zSXIh5=d%Gg=x8McIK|k;>%yZSxRZ~td<4BD!b~X&o>LO|iB?RN+z`eq@`v_6FBZ2X
zNyAbP?z=UwVgW};S){TFlgBs^x>n+)w26fSp^yC{pfLa#KcTQK+;ufMjBzEf?o2BI
z10?T;I~O^M%^HhQax=Dc;arlHd5#w$1yT}d?LG~RDu@HIOJwV>R?#B&dDrR#HDC%m
zO*KvB0~-Lu+pAmL^tY}=4xOT~rv~<}f9yPKyPRpg{#Pt8tto!<U<)1<8VUbDXLW65
zQ-7!W8G!eKt_Z6qz36bH_9axEW-;$=^@TuHS>*3UMKZtoUwcej#|_&G<_hSy-vdI8
zbiO$)RtRz`ff&-&rZZxMI!Z_sVeN9%PRp}-zTbsUhGar#mz&5YfHph{%o|Lb6;nD#
z`wdKA;W2^BWsI1uf6XTkD5|fpqm*ha&AJ9gtnHR(m2cjJ<g$XW(yf#k8Cb!cy*E?A
z<K~2S6)eJ_LUqYD*^uiC0fv<zVilU)u&KvXuJ`n0QvDGM)Eg?tx{H6Y{M}GXeMKO-
zqyvr;_b|C^aC$hx42vMyppAB~BQ~M0M1O;kF6;{sHP4c_sGiIp@^Rk(+23J&poh;0
zGRw(P<F@k}?d;%uzcD<eL1BQw^sPSbw1R8H(alo6<C}}ke<L+3veS~Fb324RK2Ni~
zLI$*{%F%j#z$A?};%+Gkig|S6j^cN}^|{plE0!*^x(Gh~t3Hc8y<KpGMj=C2UZJCn
zV0ZD4Fy{z@GHfIZ&u>2tqQAfuwpF45G(kqOp@%<e5um+9Wda@gIMY02NO5poD(irB
z7B|}uyd*>_xfgHs-T`FI37$XB0?R%JyRhg&E9HIlk1v$hpLRvQyMzfsYCuMBkT!15
zi>{ucQa|`T2wUVyIK1T7NU0F%YB-9zjjY}Hwk6@eRHQCcFyuhKj;ndI7|aKz{+4%N
zpPvOTlK!2QgYoZS!tvZRvbx4I<CIrolhVOXoS1PimhQVsvKt43##O_40QuoRvnHSV
zDiZ5keJgXDzb0v$OyI8L-?@J)BW*lsLQTK=b~dO{9r!u7Xh0h|agi06ina-m%`RO`
zQa_%4u235B&t8Dl(*y4-M7>CWNkI_A?Dh)G^ArT<s1v3aV-~)=l}5+7QE!<4qWAaE
zN)4QiTUB5TV$j<^ruO2=b#NB4u|=D=G$0326sJQ!{~we@k5q#+#~Xum+kX&?A`$8W
zC{yAF%c}nXATYpLAS0QR(6H+toJ0cSC7te!ZT~kY!hup^Z<(}N{Ttc9Anq_B6IZH#
z0j0_Yz>%&H$24~`6sk^NanNx3^QN5G<~)J%`TE#YNdEK{vFYLkAzL$gI|uo;8pu;h
z0J;R?Yd*FNWPe6{{hd)v1y_=0_2B?0Kp5IessMy(?&uBR9n-<;YyucmHA-ro`FgBV
zX<wbD8a;^Zm(632(^=F2(447@8^lkgrc(vWMu6F3`+XMr{%zG>ci;&D7=^2D4d2J)
zxpM=;d0jcZItm&C)>mOExaN_b1S%Ly@-452Ug?6^O4e{Yg!gkAZTR@9g74dWzqaKE
zgxLTSSaZxyLWPt(;{O2h??!M6Uk!rkqq&Mi3+lBXw)?4l7ot$nar{PgQW_!;qVVr^
zaPwr+8b%bLsO1_y39b+23nf)VVB1J<YlrHc2?pe{B{f2d1PqrOT*yG%i|4ydKm^@u
z?)Iw}Pca`~2pCI<JEA&w7nTVp8ML^dZxlVV{*k28Gmy^~gD6>MF-GNqy#|1odu&_^
z91Dw5`W(7wap&4g)H9DS@^&yFE%w~5;@h|(P+@kRNL!SKUNdPRCBUHkIy{rl74)Iy
zjogWhXKC>{dmHa^S9kG}l9;=0?_ao)B&I#M(&iI<i+sJ$U=BC)eSfk;2ah3$!LMhK
zWk8Pm!7B7y3*Q5wZOH~6pAV=lDwXRMNN&oc`wghYOWJ$33tCxH+5XJH1xQa;-*GnG
zeu@rI#(uDpaqwYO<Zqna4zGZfrDDt`xvKm02aI0+B!|TTiAN$ulKhV53++c%N!$=L
ztAN@Xh=7?y`vO_j%;i1}-=F%x=zTywtq>uVM`LWO$BDEA*-I*{`DqG~Vb{^XJ0rHn
z68nRp1JsiuRzhFrUSCT&Fj?$*Chkm%>U=Na_G+)+qk2YU;#qA??K?CLO@{jPJbCnb
zq#(*%B;0|cRGJ(j58^9u4@e9LmYm`wCi|40d5!aZf_nNna6#0*l6|WClGfbP+}xWW
z$1!>NVwN=K2mfL>!_aKU;nY@)uZSFmH%phxGazRZ<a#t9tqB-<<VZ$Wbxj~8e6U_w
ztGW9P{}WyV&#YOMVF(YzFGXZLHOtIH1}2)m68E&Zt0D^+28r&5-r-8$xkA8(udXF7
z_AHwQT7WX;f((%%9fX}{w4?~yf)2@I!X#}fKCVSXaEc^~naq%{1;qi}ku*nNg&(AN
z6sp*hFHM&0)`6te2MUzH9aEzQ2vX7(T(#(=VTNFUy$s})nBnw~2#o(Qlss~Eeq?Y@
zc4kYmfxKv3l|?HKupxZ<wGot(6~Z~>zx!Mm)VMB>zj)L?J_Kr>{S+c5TlS*a2U<?w
z8#(64RrLT9qHPH-loCR!ndVgA9h>MN<B?&^%V5_nh2AHbY83eXdqM^_#3<flw;w5$
zqJFTgB9Qlr&FZbDQz`tjvXIEG>74R|x$Yd(?U8Vg*^EXzyzZRkg`ML{u}i`4Kf>8G
z!p^~0e4yJG8Y)u-a;`P(>7g8p$ir04i|Lf7XYhrwN@aGd7Sks&>2UjIptieIWW+V5
zn`735w&N`Dl%m#d-nAt*+tu0rB<rx~{X|<<OBcl_Bb71=D=SM*{c@Y{Sx)n<5DG{t
zrO>jh#VgE}sJDau*(t^!BDQy5(f_v6S=);k1R@_kGQv9<J+IgnqAa&A7D$~ttcZeo
zr#<q}oB#O@?szkhKigkf*<nF#oPFM<Db_@|HI8i3+I|5X#P`f{JDI<<|4zqV7hB7@
zDLKW@(^YS5`asW&rqD*R{oc3v8|MasMcOyX0EVC^rFAWs7&&SwBuO4S8$WPo=z*F)
zcF%E9^)y?gNYe%Zmd6||D?-@dwLEwoI2+yY+$=xv&|!+MF`TwExZ~`iqY}<~IFH(f
zdP^Tj%4ocjB@=di{Qh3Wu0ExtOwt?b_MmfUw4u-gH-*60GF(2^kte)Z+%7*$&N-v)
zxDy$ZE8_UM04Pi6a;drAGFQ^xv0F!lOC|6K%|UbW;|NgVmw=?JAgH)blyq@)m`?&o
z!Pa9BCk5MI3)|^?(N0e3vzAm_8%XgZvc04Ajbug3vmv41nQNkU!r0UDo5DvwZY94v
zX)E_Eb}UWXYzap>Xa^8iPUDZ}+3Eln3s08&D-@mFwjNAh@~2hdDAm?)q_=Muixzp}
zw8jLB7HdA4nvtoA`yok*CpjORZYzVu(yOu)@v>7I$XI!G)P!bm2kY~UoEA?s00(Vu
zWN`WF+A=?+k%9Drb>)JV>%W;<D%UOMbpgFfTK8w3@3(h7-2FXe9R~X<tdfa?f`R(C
z<_~j<E<$ix@KQXY-wtX{IEqe1gAU;e`N(?LHj2(W;yO`<Ol{{kXu%TbvPq;!QrBr$
zsIwmdLf3XNnGT!!L5Lyl`RR0;la((?@|3xAVsH+f2jla&=%nNQiUS5RkQA(jJHP&u
z<1QxUpPW{rS*wx$ZJG%kQG_A-F?Z4QF)l&Jw_BrQJNKOv2XiO{Y4Lx>GBg=a{Pb&z
zgjdV0`!im?WQ&ki-WDoFp;)7_Z3gaP``G~c{0}(25`nQX5Y_r^)BWWGAdyOK!WdwE
zv5INEH>C2tS4ue35ppVbqG5<-PLgXvLGe=HCkN6MJ-i2`^)-eb1roz`%ewcC6raoU
zlEzNBxlo0p`;<${`UNS-EdAm_rYOxd<9&ol$3Kq2pN-r>WR9bmXcNG79yI`;M$#B}
zskGfbxcJ=BbZ?9VRplgcc&NASPb7w-`i7>9TDgF;`zn#VKxap4z@FB&Y$~yGSmg(R
zrtP!kM<4Y{?c>N22QH-jnA@$J?!OOCspT#kfUgaJe7aXpH2di?pSD<*b*+pMf^<lM
z2EdwR`=Tpb_^$grPJg912t4b9+-Yy$@<%2HG}pOYp)?=3dPx{w#_M@Vf*q!pf5)+p
zCkTc#vH`q6vHJ>Hi9Pw;DdeAETja|rTr><F^2UCzyvD9biyg!?-Dr3p-yB5BwFYNO
z-wBP%bPy=#qb63pk56YZe?PiOj_R~hFjxqO&9$b~rpR8&s|RGfc8UKAm(SG-w^n8c
zYlr*4ZXx?#!ng9B1<Q}5A&qA@9sUt-i~FszEfLB=B&pV>?h2F*LJz|Tc>FWi)UF`^
z{54byleo2*6s@9l%=%!;CF(QhLP{=WI=Edl$hV)}6uWwz0X*iyVv~iIA4n_M#_f?<
zT?f%%mMX;f`1e@TkGdpzph9wOfL_8D#hDGVw8cA)*Ds45Sy>2~UtPH`MuwCc^6MjO
z7S4+8Y38e0UNr0A_uNo2<ymV`#HDv)j3jtZo7)?e_x9Ev_Ga@{yLFya>f5StWRtz#
zF2>d?o8G8^C)wEO$JK!gae2+q2%6OUH||cs4GjT!ihRa<6F!xcfz{y}T>nVl)xWci
zGflrfT#92$jXABQ)j!8!^Z=~p?)yVq10(P#LZ9gmXVhAnYAFJenkEEkW5G5nn1Y!#
zRm!erV2z5-|3|Pbq2)YVHF+W_^%F2QI$V<m*J~f73P*=rpI1<eKQaWLE{?HnWFwC2
zKQ5;_O2bJ0YhfyYPr00zO1_8$V5*_OxJ3B%Cs=g6JcezFynR^J;>LH@SDqIS2#~bP
zn7fmm4MIAzFDsi?79?qTU_~pYd5$OJoroLdI+u&hilhj$lHh$R=`)3X>;@}t_^J7v
z&mrQ9W@StPx_avU`S>BwED^z;KiMq~34dk1cBj3sNsZoD#a<C-n)yzDnGgM0)4xU=
zYmbwSlR{GVDa7TNSnT#upeI=k@NIjOb>rN+KUoucDEo#?fu3;&l1i8f?o&_(8(5)3
z@`ieJ$fz9KldtX#onW=WoFxHSqVRfay)t8Mqxa?{LK9k9mzaaY8uU0%$ehy#1A#(Z
zPN}mCeEBz4Po}3$(*viMtigO>UB$3MikvY?7P}e#I7&4q@yJwB65I3GB6iqXitwJ-
z@Hve((bbAjWxym#L!TmFgzgzW*(ImPKuEm>EyBxFi=iZEOJ0wU?E7!QB<g5jaIBUz
zYc}(0N`=XLItd{w4t^OEEnv;_5&<{orJ4E?uur;~&S6}hOvG}r*FeMijt6Wyq5CW!
zgTI{-^nChwY3hoQmnHnSYMZWV{?|G{1~h`(UKE_Af9O*eIfw&BE9d`yCx%(__3{<}
z5X#?a;Uz@VzstV<yYBKLLkS_H+ef4RRkMw20OF7Uj{kjOIVRiZV7Y(QaT{#N(P^Uv
zbolv)Wu6U%R?fjStSCUPhv*-M8f>%m1GMd&AeZ@{x^P?_cmW(I8UX(;+R^|=e>KQ)
z($VcJ5?KbgsIBl=O%FU0485fDK|6~o%2I)pw-d%|pbb+U5P&=HGxx{z?j@}LLtDI7
z0s^^iVmyS5GWQgZJO`Z)1~*p%D$?mE`mMWfWZX;tDlBdX|2v}!fY%Mi<#I-;(TEVM
zHJ6^NSfm7G^h}O4etbP5QXTE~=*pwPt($72FZvdMxn<{L_-C@qMR33||KwTk^Z2mJ
zQv9EK-;xAnZ7G(k0o+|yE{ik(83(kaFz4rga>%PSxX%=C|MNm~!neG>hSv&GRqnRB
zb!c9ey1d3mb0?2-;m*23OagI=!kc^l+&x|Q_>+PgQY2oLb?j{}I-2Z54}Fy!J!P9i
zC6$V5WNRo8^vnn3Q-4YDd16!;Jb7!09+`*X*0&_s@eu7Po)8*+WzFM!mge692CnOK
zmQhs~843a}jzm|XlN>T`tTJk!tP}ckU*Ur%{3H-v(X&&a!hYb8z6bk)eh)WFNeUbC
z#L2PG0v`(=i!YL2MNLa$RFeDjcn&YJByk{Xsd&v5J)SGXfq7Y*NNSL2$1iNa%QVL%
zRINTxpTs{G=@x#aS{e5A@^EjQ>~pC0(0JUA2Gwu0aud+X&z!efn2;lJq8+uy51ZC2
zl6tz*$`Wh4PlS2Nkd&ZDx8P<@pWv6?+`oPw;}DN(glBQ=jvYGoc;HUo8NiA5RoNhy
zOOE~MrDVKWclPJwOcU!z3DgMy_NOqxW@In4(rvr#R%Bx>6nYA*MCCEIi_L_A>aKtZ
z;I~-&(DAd2`m|UFGE{sQF}~q_zVQ5%SI04h)bGDqXKaMjhsa2S+oO10a3QH9=>rO+
zGz*)grZptmuX?)HWTKNaaM9sR1)!KRmR{b`JHJ~mRvFsI{dyvgTRFmpn1U`L{;lE^
zJLeyoW%2HUP=kr(u~ZB+Io|zQsBkFbks~M8@n<o+8FKt2!-(Fkf(5aB!^51z)J~f}
zTn^dWyE`A&=GIqnr|D>458L{@73SnfrgmK)q+ov&9vcDm4u0feTu(S6bE745=*M@5
zIsJ)iRj$qt4E<i%<X6T!Zl{s^xdxn^s!m{e61*S0m8<3MkpGz2f3ch8Eo<iwf-beL
zH$1%Imggpu0>e$VIAC%Bx9~({%!^$%2U@ce-JYIVfi(NVt{_$RjE2e$^!~J+f`!;m
zcyyf`<JG#6TT)JR#XkXc{%T$A!Jsm`DhED(H^#+$t6}3$&H}HfFt!B1svw#QwII@x
zNf#Y9(ghYm_1NPvWRvfIxY+KNmEDp1!84wgAjHgJuX^v(N~B$dZJ?U?`r5~jLC-Ug
zCN7UYci^=hf2@s+7k|c=4kZc7M-B_^ko62!et%OP#Q)5XgQ_NQYX_fjV1w&3xfnd>
z{x1*2678LiQ{)irPO>toX@_Fe_U_RAU#oYTDtSG-!m*-pSkEeAM-o|sSYM0#{nJ{_
z1d8ySdYB55lZq>P&SY}!q&tl|Z;3ueqpu`j(B?BItuMusvZDusFO$E~#)e{d`>Q;W
zW6`YMX@`;_P0$QdEl^a%G+=jhlHz;8{*Jge88X`Pvj`4`2W@8L>3DqO?SQARvt|ND
zZ9=WEO)2y7?Z=%gCAs`RO^eSQ@sc*#7*xnA{q3x}K!+#a+s-uyHMRwA?cshXdM~HB
zG+Abkw*3U0Oru`5jGR{ZsQ0>ZHt^Vf^%rptui*256Tx-IWTMc=EL^uvk{zq0uQsH;
z0k{`f>{kVM1^%3;PfnT!y|dsqifh@3l|m~A?l6reOM|W?=kR-N^J$4PFuM2&teey&
zdRCV_{JO1w$efj8?<bs1?VsWHx0y{qit=*zmQ-Pq%I*1Z!!+IW#=FQe0n<t)E_6WP
zhODtP;i|2xG0SxFVD$)QbiHKE32<&aOYE1~YuwLgTTV~Dw<K^QO-tblFVe6e)#Q&;
zVUj?@D8K({V`G+__{D@thWCB6qvK5XVT`_R<~bK$#;KAlqQ_WN7b)a1aMQTxR9d|;
z_@%R)h88a;kb0-VSj~1LkFN5czCABwjd5;?$*F*BVR6-fi{&6w{0kZ-Rm04MjsXSb
z^@-=4=gnK2jPXoNMAaJ~Z53OeF}FRZfFEnjNLw`R??!o7gL~D`eQpb=(8KXI3IJ9f
z8C7ume4-}p^fnq<u!G3rv<r*P<hZY)IZ03=@Ob7$#?eZmmFje)Wl`0I*)Oa27#Yid
z`Sv%!j-GA4D)2I0@Uzkej_BoLe!)Ri(wPZy96Kk&w}jW@9+9A&cRpaGTp>M|@h6#*
zi!9$UzN@Y&PL)@XqgN~g_;?$4`p+Db$8lEb2u7QXv4jWS->E!g+6u~j4X9vGSRFW5
zTqYbgGUQAji(*8N+_{J4-&_|(gS+HMfiw_}HgGZrdqNCsNCpVy4q;vSpV!t0!k$t!
zW$==pEUBP(lxRc)c<T_=gnabi(hxFSDe|A>z;FoK8U)x$q}X5yg;#&N>P63DLVGaE
zZ|R|p0o;dr3o<`;b*~+<)cGbgaQx5hl#+BUf9lVA3Gb*Zg;E4P081_*ML;ftX0U=A
z@Y&WBzKrys^nZ+S@62MGoV&bNYx|*AAyIJ#T(CC&<U-a;+!X?s+dLZ+VjRFNen&_Q
z$Q1U;p~1XF7^pOg&vVDoDQ^<x%{WS3H%a=r-VxHxOc+<ew+OUnXu#b=kqFpUi@u8f
zz)N2l0PHy!trw&HRUI(vb-WvbV{u1wse6IHLSl?Jl#iZtL4{gA@$^TPa1AWxLf*~|
zw>RPDm1;jJP*~ag-|8XK*XgnW+1&Y2K?1Owg@$p1w$_i|JcgyHetn~BtS=|Zj~Cv<
zl%tibI_PqPaP$UbeA#q`B~!tVQ3?_tGz;J2e_~vtp*~R^lz`$W8*{cBlFDSc!UC1b
zzLqI)sW=MUjJVVN9JAy+E5Aavb!|o8epfLLTf#rs)BMQRUU5jX>h(Z6Sgj#S5K1OR
z#byFj+)a6)Q#fPR*>w0L%Z~9*vY&<5rLWpS=IKVO9et)k;N+0KtfWYehIjt1$F?FQ
zeY*W=eWmVnQv&P6QffASZRSf@n^bskEpv$}u(NOV&KcarML#1Q5RjlOm*|laQ|_wl
zgWfJ-M~?czdh2EyUu2EyZCrY0y~AG;a*ul_lT|}vnk~80<Rxau3dEr{P>x8+j1Zc+
z?$v$qfO%>NA`_2{8)-6kov(;u<<Z(%oeV|_Uyo=0pJ+_9fwF_8s|S;Z;h(HZJK|BQ
zXJGRizi~ybT-Y6%2OiquY9$kS70_9s;|rT66pfUWe9mVvqanX|`QCVt_I4W^upemc
z9=`Amxepr<4CwUfJ2ynn>ksH*$RAP`fKqxzc8?!HZzR`1rd!;nDU!}7h?q9LBb+ZP
ze)(SY2`1zm5hJG`Oyg>?%HUf9R8q-;g|O+x-*l$@V%43lzShCF!5|a%3SOcrswT7@
zH<4{>v%2ioU-LR&<q@}-${icsq*ls%$*~5u^9Eg?Mj4UbM3_L`+$z_eE!MZBOQ-rt
zj{0orc`sfPspovniDzmQL+`>Q>QdS|Xlu91Z8b=h)DUkL#!dIaD#?=&f;{(Ey8Y!o
z>jD$4O|`5qO6363OqcX+Jtgn<!(HZr7GW5OrL1txwbpo_Fd@CfBO9!p3&)+>uQ8zk
zj1DkF;6V4wF?j?S2|LqkI8>cUdbi5aw&fsFXLLfOBI4h5zG%7j{RU(WVsrfF8!y~2
zRUp+4v+XK8lmm`kgTIJx0Jx4M*EAo^Z@UgYLoe7hlwkX<FhtbmZ4l#eQ!%UUUF^~;
ztO%C)*4%EFxV35honY-Li8r>CN$?CS;g<aZF6x91P3c-fXx<L9Fjbmb6_E8Z70vXx
znvH=yIARaPG;T7lrDBEQ>*Sm&S%`ADiGiKa67E>i^&yFpV`;@9{|?^(-Wu>s;!Ty1
z?ID@Gu`ewyQ2<;W6G4u`n&Y>asc*(W+To9SL<FuEZIYlv_y;o7QEu4XTWRkj?LH{v
zx~sZ2vt}Iuj%No4lRubB<q+@eyLX<9O@5nJo;@<ZfA5&Pew`X^|HcqMnUukeVTqp@
znNRy;HmHk|YA(ZbE+D~>9ik<f9hN8C{_U)MgeSE$nP>s*0d9x0r440>Jq@=X6Qa8$
z0iwV<mZ4ro=3X*}_F(xFG3|RGPtKgB2oY0TNG>{`%;<|%&g75;PmeMV&smloAzUPP
z`yt~ca)A9%qP|eop9e+Y4xpd+N})pcwC1Q=h6-WopLD%w6P%61uat|k465?#rXu<L
zyLe0>FGNhqw>R~9GZ^e%I4q2Q6^F9%<8G-n?wc2Z(U(mZIBD^CiT4Qs=#1LBPrMHa
zq(vKuCh%Mih>s$}Y$^jA*X7omu&nijbQ#hc3!|P@@DdEaN=zh_4L27ty#m$zoLpRb
zK^Pt0&lI;$av8yO5av8ow!+XQ)$U<36pt+z-#)cqPt32i@7IjS>cato5bE0j%Hi1W
zVXuubz$o}!6Tamz#-d_~1wFJe{8BuWypg1S3AHr=ibR$UUL12FpJsC{U`%1(GbAP|
zGOS2!&T8bF39N+fqYu&^U_+Lv7XSqP8#o%Y>=+^N!N>y1?GZnm%k<9<H$Qx3hozRD
zUc|@|5?%8JDB`x&E)M3s<?Pq=`0j!0lVfhEQ2G-Xea)M3!OE{rXo>E@w?wJ}$#5{X
zhP2;n*0iW9yuY$Y)Ye3>_M6=^oc_6(L8y>?P26$gO)0<ZOc&$oDu$o__((}GOCqkJ
z?pm650Zu?9OS$63RPMl)hahK~4{FTD+<u9TWPJh6hV%ugN|SH0>_Ftx2&bk$d+TxD
z-B9D}pz}UQnB%@xYr%k#_I>?vLUIKi5SdIM_ZFj^qxF<38X)oYn)RcQkicdEAxmfq
zV9KavLxS}5l)auH@{Q&)7dhZe8;PW9CiDAlWGPq*PBpKQNuA@y#jXyqET>etGv!Ky
z>AnC6$ayb}5A-dce$Umsu)j>>CCo^r@R~Pt*NlF}oqwyieF7`FTfIe|*(^jJ1A^b-
z@t%w~eI->`-d3F}s?1(KR}cA6FKs+1ov8L)A89Z;O@!7*jxZzno6{vTb*6%zKqGlP
zckn^?Bv&Mp`bL3U<AuQI-OxO6&*4#06}S=@F+Gn>LPIu9G0uAg@6viU!iY2zzv;ej
zH8b$#iv^Kxl#kImsIF`fv~pMRv87EKrYuCefA2{7_<-sXt_=GTnDZ;xyu(2dAnb;Z
zf*u)Df0hH{zMbqg4xQ9mBzuH8xwR0cex1CNHMF1KjNjoma&GMl^2cOs$hr|OLp}y3
z%FHK$nn5oVX?Tzb^hp75192LL9vM<uxN^Un?jm{R3pvWyU0o#1ZXk@WpDFg5F&zx_
zP?gN@{+V&_-s=QegInI0{oOGFu95%K1r+nWgjT7*#%bSY)rriP#|OHYgm(mHq?fBa
zD}jFwO862Os{=dp(|e&O`Hy?CCL654v)zT|xYe?rOd;PNcq(N3`h2yI22N+HW}j2s
zl&U>*w#y)Nq$>K@qe28r%x&k!9t3|Z;Och)eo?C`4}K~YL??q>oay}EY0pjzy|Ip-
zVuaWn=V~s;8H*YM|C~FJ!~(Zc8@M{VxV>UwuYTZrPtf=?If{SOjlh;24=#N@?yVmd
zMwuL-qllP>Z&5gz75f?-dia-Yy!p0O==jpSMay5adzcE1vC3f1E*{;^4eUB+V&c!K
ze_pqFx|jQBgK1Fce})rw%{1#3Vw#;?*k?2pD3*M@NL&hVYfO;!H8LUdEKQ+9Gg2TL
z$*xI-Cg<O!rKMU}k`Sa6Sk+A-2e+MzH-v-9g9b!26C&?bVxlvmqXVxIJ~}WHE~v}X
zQ*=Daf5!Q}7g{ilZeWrkPlcVuA?Ush^%%d?eR+pDqRk$46^B2%z40+eO0oH|<g_XL
zJ@<8-KJ!{!Ysp)$#a{bdytN-d_lAA)d>}v18#i<%fF$bmpJd&7y8{gpP|=65t%+9w
z>%nzmwc@8|wRMqafdki*A16D`-&+3mzsiE`9ntm7u&^hZ9IgGV1&$7zIT25?!cpCE
zx0x93unFaSP>YMNv=$P0k_4RN1Z{cb726FB>0uN%eN{3EG#31VGN=4>S8}XDPjIM<
z2sGGu25>v(=cQ@(fb@WLiM)QvbZP$Jm9LxffpIq?#43UmCofNw@G@Ex7^?w~X_AAz
z6ev_y3+N@C0?pJKVx-+>7wLFRrqz;vu@u;Gsieh-yjD30_)%(masxJId)kWWC9kwf
z>EKs7`1th;mOs6}vC{^m(vBnsoZRD-u2sI5({A^oT7Sf_5(+zhs#C<YcX1;<AV?wz
z|L#}e57Me$KPNoUJxa<wHsp}7uGfPi8slloz3Mk_l7w|zmLI3pEYcS<XgG$c6d4Kf
z@|7r>BcmJCY9h)b1n)C@c7B>$q@W?8?9I_|PU30T^Bp-b7#1Y^Uc~ZTMw-t;3_Dd^
zIX}g-=P)a}i!blXjP*pNCR`n0UuYXHB93JIptJwMVdb3T<mt)r1NWat!xs3iBq);u
zyKC6f8^Y@Kz(#s70%!VY*p$EUh0gA_AcCH)^7njLsnPcbrV4$(aPbg@w8Ufe6T80)
z(_R30N8LHB_wO_1I9X&UR(@st(#?a}^pZiYi}TCTf)Ba%KG>a%{EbxO;-&x(1$Ma{
zr|enh*4<87V#q3d`lkM#_8)=+T$~TrdN+N;5=KZn@lQNgYuu8+V=wO>2Fl#JB6r5q
zDXNVKc7!mWkf8*?b1<>f6K>AA3>x)<4Ql01oputCAe|4GzY0YCT-R7!UK1`HP$!d(
zy>}b3=okIOVYY6i6tn2uqqcP84&`I)Aw|4KgR&o<_4vhLjBn;gQkTn_srYLhn6>qf
ziHenYb|&tnATSw+)QLzA>~Jbq_yJ+J=?_ByXeZ%Hq+>An3u^fEx?b`7$4hU9X=fQu
zT7@ilduS9XVpM2SzA6FsO6h+jt8~Q}r6(z--kPoTs|ZpSY-sg2+E?#;hN6o!Jjl4g
zxW`Kt8dU#K`&h-RXr~0^F|I_B4KD$?)C2UOmj_G0P9>-C;Tr+MBHE?7<z=na$-!je
zi97WaNr!LqH177dafqdGAH>H<N}Kd9^j8Nh+^qYvQubWbs?#k|BdzW(1sci70HOGn
zoZC?B_oU$Ot<0?!_C6^(eEIDEI}qFOT^2U6@u|B~{XzFz4qlr07#Pov#^xx029A3-
zy#-7hA!`6#@4_<+k<TT%D+1Jbk}I+ZR&$>HRj17(L3C=yp8g!Ro(=Ki`xaCWnfY=o
z(K1I?7YtYLMV|z_JYn-2Mz2K>Ks$JUj4RJk`*Yp^wO7~a)Nc3YY5ktuAMsJfVD6u7
zH}2i|Epltf1m;XXjxFU)d^3Sw(|E~eqA~+7G=*mw^W54vxzOmXQ)Q{tK|$S3w3gBU
z9E)zbSDx~;g_X0FC&z6?E70v{$)n<{12k91ZG^$yAHmT!eZ#^Sq7&5b)9zVYeJ8%B
zccJ5Yvc7oNh|ivU;&9R`zv`<m1!N7&{ss|6b;A38{2Nn7f++WZwSS+_0XK*m@mT&M
ze1BsWK27^e%wl%e8eeey-hXF}x3212(nY+Z)-~^ICflQ^g{(GyK`fUgng0)2Ul~^A
z+JyVA#iBt_knTo6l#niIq&uVpq#KlG5rTl!qPt5<=>|mvq+1%KQ@Z24=-&G~=Q@A5
z__Lmzxo4iaXXeGcdR8Q6+K!?U@S?o*$=myF`!=+|WXJLp5sdaWoKGC3U)O%4xWANi
z_U+8;{nTe*Et?HErIuw6|0%)^r*QP+#$fronTBBC(Vt2w7?tlGlvS)O95pJH`@XiH
ztEWdlH8)j}>x}MvA)ny#I_b@~-i^vtY-$|GIHme>$Q~*f&ACQ)@Mq}Y{=irYxq9&Z
z!N`YUI~rip5i9_V0KLrD!ZfrT^3Y6v-B%LtwYn@0B^R;+w~oy{FHU4*N;FF`R|)To
z+9jqD&c;ZK<E4}zz}?MTA*THf`U#|X-a+dwCMXMD^o3o}Bw2zLDh;g`J3D5wVZG2U
zIk!af%LlD9<-gpNL4{blf&Lt<dzT-S;IPA_%c*KjB2S2HM2ZjJggLBsyfn$U>6-9|
zV)P1qAE9fffurZ187DbcMXvAxQvq_6;<F@wXseV1%!TET1j2rhmak~IuG1DMDU6)Y
zAwNL^<r#!bU#j6A3F2d3`P_xa$g7LC9$!){y>6E<u&IQb?hV$C9uqu0cI_eB-=Ul^
zjc%`Gd^1Gn<p7P%YFk;4=(mB@Ou%#|pw0j<#<7xKBWgV2o2}qrmfek)aT4M15*q<G
zwqfw9rHM4GQTO6jG6V-dR>j=?Z|sq|9PsJ0RbkP^(|y>Mqo~||v9pwrcmuB77Q|+}
zHUo~)nIdzp(s}Ee&pjDG-%->ZP!Y4CQJJGLKgBCw)ZyrU`G(a!&K&D6k+6c-CSqSO
z{?UN@<C{!WxR$n&^2qAZixHH6$5S&+uUE?;nYBR4I}CcWMZ~?(BR=_As8gO|BtX#X
zruEd|XT?&~6XNDYZM$qEc<TNX{mTs4-}oFeSeDs@7~$<28R;3Q<q$P12g)<ca40Hl
zz|({wXL3MFhIJLUu*&96eE6AF@N`-mUQgi}vT+NcU-c$U6g|Na^xn=@cJqBKOFSr&
z#Cq;${qM4IMZZ*Xt)?#wel*CJ1wJB=g>1H*#+#91w1Ks)aUjL?tdkW!x|vqcJ1AB2
zf%eq(J>!<5*YU2WVJeqRPWRGkf5udTpo=Ec^+fzjIl@-uePjF=f6LF*?m@hiVwSeX
zizyz>0Gles-uku&i0u1d4rGlOq);AUM-ab;tUnhpKN_xj>N9IE`TeB-MwESK?NDH+
zrmf1A*S-7Y&E~$Z`y9jy`X}6(PX_(@9K|Bp^11jU$eHHEO!0=qjaC)*AKCl?+Y=f~
z%Gi~=CzmKqg1+XN9L>zRG?4=P6fXa4i@LzNY;13E<}drk91O%WVh>mCKU07ZYtu4i
zOBS!t+gXbLbDT5oh}`~cE^Bo?+aKz~Ge2+NhRSItipM9y_6o-jw%HYOe0f>HtG#2(
zFOeM>dCcJXSw**}hv<7J<*nN?`%?d0JACmVCYFxDx*H)s%vOd>O-_o4j<s@@hiXSF
zpRG?gT;#Mr|7S&=_uE1Pv%b<rwFrx8K%xFiHSvN$lK*H67yz|#H|ox@Pa7%UB+|G{
z{~n#|doPW!b&3vtOxHsL_d8MwmW2NyE5UPbdLHz^9)3w836tg~Q@=nX_(yq_zs2!1
zNPb7yZG3ge*e_24vo(mViEr2}CrcYO!7{_U4W&4tEfDQwBk$3sk=EE!vL|<t{Q0#9
z$R2w=op2fAG5_#pW1ay5Cb9GCuip}$WZ4Up>HPQGgZ2gzDTu`A`v`+7m7_?oco@|8
z*I>SD=G17Fn7-`l_lGXW{YxnA!XI};e17ejUce&xs)#fHaMnMW;U<lgL~e#DQyRcW
zt`N!^m&miThkq6jitMm<$rhuAK<(?-=VBMGQ6ua2ZftW{U+lficD<$)+-iH%GaJ2j
zoMe1;4tBS=^E^SicOoNoxG<vJd`#IM9?!PE?^O)un}V=0$^s`HP)5WK*U$D!TMn-$
z5g#(pjR>K%n{)136OV*E7I8UVFTXuF8`M^_gES#&{E6o-y@4{FS)mna6o)ElyBX2%
zQOt!=R6GVtw?Tw$hPp)`rHXB29|h(<tE_ytn7KyJCVWb-K*Y!BJU-Ze{jL~^Mc6eE
zy8S@(hPJ4WO+qUOxBt#w)+GVEslKn2O9!nqyo3U_U`S_-GfD<kkN-k<H5f!`G*8(*
zBPN*SaPF5k=3Q4gYBoc!40NsJhgl&mPsWDV>byc<m_!1*F2M-x0o<2Q@3t=oB+PXw
zd^kGAJjTX4*grXh7D|4%ru_Uk_|Jl?p>8=2Rk!!Q98Yz0GDP%(^*Q1$i5rhI-%yJs
z*W^ljC|yw8#_!y3vj(DbJfR=bxNd~e+~U@}xS(L{U>A%!vb-YI+&?+lJKifXz4<sO
z!O80Q&J}C~0D|bphiwEL^MbQlHlkw<3Y*g3HVFC|oZmZ;S}10J7x~%hV-o&Gepq$a
z<O}%pP9Ho&d>O>~aKoGkJ#@c|oG(d1>uUU~iuBn8e|Gi^Uwa156%)%m(@l{0BT=f5
zmpz^c(`49_Wns!&&Hc|iB$gKnG)bm4j$?Het-93#x=}UJQ^H4DySCG`aE4FAXNR&~
zPj8wG<^xXa@WJKV(ZOwzlCAv3{R6xMZ(0!r=Aw1^^$hQt?DZe~5jYZL2ipPK*nG*z
z`xm$8x(d~U&-OomP9s5-9iflJ80)t~x5lunWMx<s9$bfd9-g4JQNAJZ^})#36QO+P
z^7mdNftJmO>h0`p)8dH$ASl%2?H`asQng!kg>Y5*3ljKkNz#Qi(&cZGEsSS^_jBE=
z7xL3gOqO&tTHh{FPh|{asnQl2B<3SgJ4V}X)Xq%TN&{mGP~Lk_zkZd*?f4Txt-as2
zJC~(Ve2$#bf2|IZvny*Gdd<4)LSeq2RsL%K1<b)3n>dj<^%BQUIOOs6g9rF3wYtsu
zeY-<2Y!OdSY+{9inD?23ncu&@gkUERx7MQ?MFPK!<M!9c4ehg(1khUkL{%N}Aj_@8
zoJ?KpZS5{6^X=2q1MNKmyn-!jK}$7qIe-lqyUYMi(B}`rdw1|hbG3@m1~QY6KT6kp
z2;XuWa7JmL?~y4NE8PmbknF&K7Lry6-NpbXXhu-3e$E>!bbDgo7TED$Wn+N;6lyg#
zUCJb(_<Hm^T};EjsToRgn%=MyyEFV(iy7f;rV-~i1IV`3=*fj$L3Ax~4EeXYLlQq9
zDhkUIN{!o-Xtt75JNEf|9=4*`BmgG>mv^sGB9u*$RFONt^zStV1O4o~qxCvH9>PG}
zRe+$kR{&I3;d)c#HHpd}Yo!DfSG}uj!6N7Wc<Ew$-5>Z(!X+)?`7O8Y?OWeBTQ}bC
z87f*BSQq(FVMY@%ZQ{Tt36_XE+>p}{n7gF{&BQ5goA6Wu*VY2d%WLey)itdbBbUTm
z_u-*`0a6$u_>ya9;)(t=7XJq>8>L@cwX*q=;)Pq3Q~^@QJHuHzhht{mg}JX106U@F
zkS8}DT`nQ^WWWD_9N4=c+A)m;jS%kb&En}ESZ#3Sm^7C8&sBgWCT!64**wlXzUz>D
zu_coGr}oSgkc*#><bJ&p80ym^yAN*;_>%Nb5PA0x9t8sic5|Oq0J!w^MfH{VZ}Qel
zm&yNz6wX2FyS~M;9iUr0%{7=hTYt<RcNO^3_l>FC>5cQ;$}Q0+u>&ta5DeOZ846?>
zxZ0dQ2&wAmEmd~r;@)2K{JS)}5*l#=lFMu6ki#{-pn$IQl+=Bt5Sq7dl!u$kAN!S6
zd-CTyfl@eL#`f94sWg0_oaHW2NdRr(l=}XdI1LNZS&(#)WrfY$xemr3ZMw^-T9!j3
zP8E2Ib2ypi#FE88^1lnaRZE0KTK<q<e3k_<4M6%HP$LZ}R}B6eUas>4=6>sQmgHyf
zz<p;kH&vW%1<zT)HX)$P3uaAqc^-b*o||IGRdVuHU>x}(s`|YEw14Mz^I2g*DLr_r
zNlj5BDK~Ac=b)R4dggEs0{ZeR$PSKi&t?R&4HPKh3qe_FyFNFaAg+OwQ4a9I^MKRY
z6D`5NlV=E2XkQ*gvv?x@&CzQ)^gi!%fZV2DUNF4rN9%qg+Jb1^efS>D^G9NIYj?K|
zW{H|>J-dgFQ+R(9Q(0%EVr)HQr|9om`H}Z16?81XT(oNW-pu+GpHWiey#c3TlW#wb
z^}CPGP1vB?qR;m2uO35x;*1Nn{lN}G!nS(UU2ElXgHQP%9c$mem!Wu$Gpz;|8h|b!
zNOJdg=P(}pPUWqtZzp;|DB>|h%2-AsLU~bKI}xRlj}Aqt-c)0P%OBw1Qti^fmnNlv
zim-0)fVX!fY#A$@v%FWXl=1-?(J$H$fNr(Ib)dmfiO@fy6a(1pfcdJI&R?JNYvU!@
z?fxo26r6g$kpW9FapIr)Tim1O4+HX}$gkR&PPu`bJ5yN(^o#JM!eE*w2M2ou_b`zH
zbVu!#C~AxU@!cP^RfltBDe1lhkz4n<n?A?4D=vos>e#_lLZz49Y4V?4E_1SVQ-U#d
zN5iH_^lJToj?^ms62f$_ygOY@;NoHF)<ntT*5VF}*L6AmOP)f@Qil3BX^`iTC!&x-
z8Gt=zE{RX=7Y(VEs<d@757jBAmY<#!;`4?$9;ar8Z;?Y3FMs{mo~iMx#1F#DmUET+
zBzs~{?rO?G7sggGU`lN4BkWxL{_1V%_7dqfVF_{de2djzuUHx=u>7(8MRrKcHQd{`
zp+-zkfn9E`Be)80_bhdu&19n3G+z?H>%x$0INDTP59+yx_DvB&=KXv2)nCZWhVrI}
zKNr8g7X0Yusrk(376$_Zz1YC>C*>Q>zg~(E^epvAVN~Ls7W%!9YUXED!@QxULoHdS
zc}Wz=>BQ0^xwK$SuBh+zQhi=**=1^vx%LNr=o!t&QCrc!7C-?q<8#zvHI!@xc~{wb
zi5vv8BACCvZ-n_5l4;?!0yU9M<&_Lo2H|ZBbCb=oZtZrnB$h5=^3X1sz6&2&gnMkA
z<y5deG1j^J%LY8spCc)^8%RuWr)&#|7vE=DZHelS@v~pb<&+&9wm%9s7XL*_3P=;7
ze&fTCi9jo^Md38d{xi>R!Kw1_a{=Es-`h$kPlj_pa|XY)pLban4n}#UWG_sTIWN$9
zr3w41I;udh=Q!30I<DPK(e{Q`M^31ch0M(KO5tDi@qm32$zq_+HFgNh53&z)Tw<_H
zST7fmBhY~jvTr$j39c`t4e<BReiiSv*d@AsUzF>p$W0uq6G$D|Vt>jLk93VpQSA`A
zinyT();;LoJQn;||Bbo*g3`cHkVA86?LNUe<IAh#(g*meLM38{t1LZgeaLt1YfS81
zMj@3C6Zo#^-w}bY@>T8G@J8u@;9{}>14_KZN1dB>UW`@2)+<=sYiD~WX#%x{ga)-%
zar~c;cZW<3yFAOBFQ9(C^&y8!%PY-E<8=y7Q`hZN$BKrc85f(=6Y0|pr?G)#FWER9
z1YpVw*o`ddS$c#zM2$0;U@%bBH(i>|++DIgkHCZrTyh&@a%)E7<fpQ7X)_!gUqh2y
z!|u+W>3NYR%^t<c*0_S?!<X6hnG+dicClc-FWC*^IT#iUGjY8~EIdb`^>0)#ZuZ=9
z*$NEp@K-si7b|`~M)LaT5zIXlIFZr3G2_?!*qI}|mlD5m|5t}vy<<$^J3re18oXcA
z^rsu6_0F$P(ht@$x7|&*aofz`<-B>_3~m_j!(XI98$NS9Yj|A^h$xBRP}-Z^QZ6j+
z=u8&8Zg*$dZochQ#Yu-&pr=Ds2beQ&E8kMUZPibg5_aci8?#0%^$_%F%gv_-1B$8S
zX6zNHNp#enj0rF_&v;d)oIKH<N5vm*GGiZ~wBjd2k-w&Yu|Whyf4dc!)cL;Fb<q9!
z*Qsn+XSdS4r?9Wc+uw8E^QQu#(fIP%+HCk$au(~u;A^k+ZCAir`E;F^O<(+5Jv~Dk
z7AOy9UM^2+eg0HDpUX-`k-~wY>peOQh?hOb%i_bvbbtW<%v%n?k?)pqY+$89rM7%R
z<c8NLdfJvZab2CVrlSp{1t1Ul!UjDslQ~+5u}d60Y$QzvW*(K>HXZFno_5I?_tdj9
z`>ObU5G1X!6fNYMHPHOkuVDF-wx<|ncaWZ%j5gU5IK-OVcY`gWX#-P7^pK_}>Q!P9
z%i8>^SAJ{W&n7--E6RKV_)#%4DbW|M$rTQ3yAMQbxiU*<Zkcq>)ovFSQ+(VzM<|O{
zGATuSMmyU`=w#t<h6sMdk8V0R59VBazaI{?FmrHqWMOm+t$qf$HaMYYyR8#A)q462
z)|WHBNX$-;d)g8sq*&qoU!d#KPz!HkMk@+V{tkX4%jx9-*@NlQ(%@4;r_0&M)-sHR
z0hv-<p7xDu{%>3vk}d+MWw9%+>kQuQ@1+2ox#(9^TWhw$s<`U&XrF+FOeM611?$?!
z<OUlA7NJ2Bo)?5AkDqcM9wbh+L)Xbz;Wu>fQ@dm#mv3CLtu@yn^Nt77rG_udBOs*$
zTL~GpH7#t#8luu+t8o1v_e&W_mNSiec%*e4o?2xWql8kzwjLLNp1ik_V~kp0RIVj3
zRxgr!e~=(JRz<hzdNxt_+uuoO;o>-tidwSrU%Tx8I4NjFNxy98Nyv)EG{<3$vSOIK
z?pWL!9JO_-w0Ik`LH{g0YuzY+lo#__ZeBFIdo};wIw^!elrYr;vfW(AYFhm%N)yxf
z<JxVT#OF3sIj5j;k^6cv|4PH&dqImY4DRjQPNSdbY2ciLLZzqKveWsr%-ns%B|nc9
zd)`^Be~`mt4Ek6}P5=BUWgK$Y7R3ZlltD2{gMB1H`S#p(!|HdtPnWG8RR;&Vv-azQ
zy`+tXBUy1mx0B6!k*^IR>w^uaRnr$`2}-!AjOlJ;qEM<r)?UVq+~QiN@E`B(CUEmG
z<dP_gL%e)#d6Eou8!A#n4vN*9W1jA9M64c1H?<0KnyvrBPKghfB=o(W<?GWU$chaO
zO7>sYb@_6`ZWq16Rj^)Gf^&d7eDWZ$g3Z~dg7xjdIyc1Y!dxEeWGwE4nEZwT=O7xp
zY^6g7AViIx-{LzJU2XJnU5y)BzNv<-uov*7q)lp(_zo7$brYR&@>gIi*L^lGQWIIp
z*s=B2N~glX*5t;AUB9wAf4*Y6Q0Z&SYWMsLenH#wM0MqO(x_)!WU_BzgE@XAtZ<2C
zk8gU^(!`Oy*IB%wtF}ehm5*C))q}_B(V5})db&QmV^7zy2t?Or{4^=pPJOq$Ck#E;
zTJP1=+-RjYpqP#4y?D)He)!@r*1lTUFTc;`aU23_F!Ug5ux<F~s=ftY9xyjKvrdIo
z8vN{c$Z>efba97(=ZRNt{b$l`cf&}{xw*<UJo4mo(}7_As*2N6A<rX=)f^(vvpm<L
zGMmcdu36ul?(G9))%e|A9WKiC(RR<d@tL32+(t*sNBC{)w_z+y)W{B)x+9O_;FJEO
zl`+yV=zv4gvqDL}v_g7S+qbgVaj|AHPRBJ#qt9I4u!=%I$C+*j-H34d+4WlHm%>we
zv=;Q4hXrR0d$xsB3BR={FS&uBEO?tI`Ipk)$4^MS@amiul_0(Ee+yvA$w0_zm&?6e
z$eStqNk5^P$(+FF*;c8TZP!}H*@eAcqhcsA9|R4`9KYVDaf$4YmY|0p#=@e6<sQO~
zp0G~vL0RFCd8?skJ}R#x*7IZ4d{KaL8;-Ems4540r5q3@K;0#hk%w%@ch;=jmWnSc
zMf#i?yPF1R9myGlscW!l*flRcIDTkd(@m_}t<HUGPe#+42W|*$K_iFyaA0!@kV9aT
zzE5zy2LaptsXc-RXvOLw$TzJ-p}9<Ld)W$)CeK|X|F4<InlKcmold?=vV6~)C%@eF
z%gA{d{kBoUm*R&M?T9&dP-mn~EW7tAPR|x&>CY;k8>91659tTt00GgrTD@u5;DvX*
zaaljL5yZk719@*ne95U-v)hh~2(ZpZ6Xl$%isRXea`}7*wNxk%)PPdT-;aKDvGii+
z+0RZ~v=UOd@BOAMo}YbtIXMZ(fgKb(=Vpb!D&^O3*^!~&YISwYl~47W7i&fekpkR<
zT%!}D3_O#v<kxUN0abiv(6hxa58r3G>8d{l08`{dbs@VQdYpBq<6z`LK5gJ>R3?<y
zWvNd2jj`D4x%jmVK_!LV1KVi-6wd+03QzGYfTV$!-PYJN*5Ytshplv8=y2L_W^WMn
z0u|h$TRZvPbeEZ5N}TV^Na`a9*+t)~OJh3&wBP7S9YWGKl>f-tRAGU_|G(+nbSd7P
z=wQa%8Vqwe+l~;2CG-=A`VbU-Y%y1mRa59&nABrf?ggvt{>cT<ljEtKWlUu2bX4iG
zfEjULaU2YHS(iV3&mRm&&g?6C&{mle8j9$%%jmXv^5hGK*0{UXW+(JL%iZo0p}bzQ
zOC;{op~!~`N*DxX5i{5nkreN38i>PRR(ZrUmB=Iy3ahNM=C+4^(3PNtW3<P1y`U{m
zvt%0OgRug(ithm7%TD{Zu~UK!jQG7)`#dqTw2JRb&fg*9YJCD;Q9VWeq(#BTPLh3Z
zYc1vD^zhZH-<pK#_nWO}4++7~M=15`^ZWiv+;@HkcB$m`i3X>@#73zWRg(aGv#>~A
z{z#z#VO!*_{fKt#x93o0`P?H;<2PaT^{wGH-~QXiPg<p<cm6u>B>A2_tG$-+&2FG-
z%hIB#b90b&qT=2B@Rfs3_JLJ0k9foqqG<@tC}RK^_sYar*z<lCptNQs0k*LioQAsn
zu6T-QtK7_0;y~QGr^M>%vDC6%g#EXby)wb<r|Lxp5E*p~u@oS`o4wY@BmCuk{2k<g
zwMdqGA(L^{8pvO!D##rL2Xj1dw~?!f;PXa>iX22A8VFgxK?m~_koUf4$C^qb+K^;M
zqK7k@rWx`~HLvq+ISE8xeu2LC`-b0k`vzaetdK^8aYA7_m|W+<b}2i%6s`4@ySwUj
zt}S;ed#T)0iRk8yU0$gN`poZ}d)7~!o;+0AqJV;~Q8P1~@^=2ji{*uL6pu50(aqPa
zkY(*+-lK#auv^T(mon&~*QJ*BI3%0t$+Ifz3L5#tex$dW<70Uq2P`aVgYwJWj}lZl
zRCB+E?3dgpwdEeH^=Wx!dY)T)N<J=fckv!7K*p_CS)?CVJG7$tjM3Q1!3$x&JS&|p
z4DRBOaAiWz95T7PZBFah$jpD}=Plpb#thBQE6SqRWy)gV$IFDP)^O3xyib-Ika+v;
zX_SE;jwtEf&`Q93S@%Kcz2GE<h|wE5UIfJEdqhG+3koJxnGxihs6WQSpj>1Z3n8HT
z8pRwtquFXv#${5=DV+$>n@%@m_<%L$Ag}wNrf+)Lb$!AS3x=7Abgc#iXHe)diN<s7
zXR@7>GNZ-C;BCDRjx4?o_~%r_kOw`Imb>K@dVVmoR%kO8D?tr!zn3`$;kjF2*Yz=H
zaWKC~o~yAS;vYK&<~HDeFl*!m)a|qhf@Ss&iaL7e>rbO&!O7Uj88)7}x5xL-l}bEp
zEA5(4&LEAIP;}`#zuxinIxK#1P^#EC@WtU?Yu|5D_>6_D1Q4^bwnQY&1%2Wo`Ujpi
z$RSNFoFjIeDUgTf#|Stw`~g%<spTpd&mj1pHLD=+UA@nBVK#ErmVfN{zJabysE`Bq
z6Vq*uyj{XHCvpgV^vW+nE&~^;2uCA`OC}}8LYrhYeGy`IDE^i@GtM_^tz#71m-jQv
zC*T(bfhZQ;N8qOMGYzFvHdCGkgZZwF<@Ijr>V4I+Wi+?G+-{6YsZp#lGGWjCrXh>9
zt^E(Q$u>Uo_w95kIdBPm^Z&ryQ7Dj}5S;0sDOq|c1++cIN6*Z8)ipCMKtTDg!WKe*
z*5=gD#KouUPbt%3OX1YEuq!1}#TA5&(-AH+JymPj&S*uLxmLS12hS+wrW72`mp&!(
z$@zEu-H%pXh?31wrD4|R(W|LudcS>YV&1j>Zx6b^KlXNOQX-dOai@ouS$d17b5O2~
zuEFS|Wu0)&u6;od#OLV9#MgK0=eJ95r8%F&w1G`QiqYBq)OK;K?0#wHJ5{2Fm*;zs
zt#)M(;Hql&$EH?mnv6-a!t4Bzagoqj;P&IPbJ+*OGgQeeJo=*-(SmP!jEZo_;+Gqr
zKTNnetj;j4ynSnDItxF%&9Nq$r5L;|{jxXQZ8$~_+MbX`;9je<S-?BTWZ3{eBZ-J?
z(c!AC>N&iBg#0T;=5dC};pi*-KFsj}cHQIMNJ1`K?Wdx0J?!%LU!au7tfZZ0xRu@B
zF6UdajI(D9*Ihpp%2}@g47i9~3=&X4^X&$9Hp~cr^qMzx2@561;)PRPYMxee@g(Sx
z{lywEvv-=*Tq&)^?fTny4##Y-F}8z31?x^Kb#fV5`&2>e1JU!3-kr@J^JV#!w3W{j
zxC~0xREBLKGv*7fUY-v@ZT-axukG(1_>@e>nu(3F@PLfqHq`gc(Ep|o-D0x{Vznm$
znF9JH>8w(LWw8;$J_^<GC=ts(2Aw&tuO9oQ13hZfJyJ?ht~!a;X}cH;ogU(!Bkarl
zkuw#%z|XVLAJb>SHRSwJ0{q?#%g?TAHCp`(22jk*6)!HkNg~51$O?!R$#8++KePHT
z_&jfZR$KEFSqEZAN2WNqi|CF&hD)wLB@Y1KLvJZp%6eX41#$DYG-H4lz5lQcbF28^
z%C0y}<v>I-%x`#}e0$qW`d=h#gIZiKVAMahh#QP8ja~t6F2+PCeSgIX3?ZIDx;Uo^
z{U=Ya0#1c*oU1DIffqxRio<qq*KY!h`Un0!rAKheS`=r=^w8KFI<J$jQ+@L;v;Urv
z1cB&Q<ZXkC`q)uda@&EGi}SK4Pmab8R6B6ATI-$&andy>x3fVbu$m*XMfChFvJuIC
zdc<#vI}hy)eWA3m-&!ueRFhiu`U}mfY+txmdM@$5QpyG&A1@f1IcaJkz7{-4o6oA8
zDANZ&V`YK+$o={Gg_4tkx3{fpM7fdpG^TX7{RWXJ7xZkI_vmb7=kB5Hc)x9Nf?|fd
z;;VI*EKS8;LbR*<H&)+}t6tgUe@|xt>Sp02dAL_(9oy7GF^hZNzhp5!^$sZ-AB%sm
zhFw=A7V{vjpU%Fmok#NE7V8L8DCWsy)LFj~Lx4?0FKO2w7aE#lA=iET+Nh#VKg+B(
z&8GA7KWJ11+!OkwTY*Yn<QTy^P~`jMpt*S(wZ9x^c{~%t(5^jh{+4>yggfL_+wac}
z!2|k&<}m(S6Cv?7=z5za8@!$J_VgVo%H>OjbiC0}dn{WnqRep4p_lEsw*njXkbhL|
zIp%_~yRvs?T^X;s01TOhaOP|S=Xc9RKa!qPjFYmh@h8W9w{!LvtRHA6*IeI2Jiq&Q
zH9#^pNuIkv&-Pb!Y6~f)<(sm$+-W|8E?0Ou^z3LcJO@=HUw|nnfUff?+*F1hd<?!D
z<XiP;auWF0DiE|42qaIG+aR1+Fj*Pq(#a4%7<ww2ob|(rqUee&Tbl0aD|hfVDW>!p
z<E5=v_zIrH_A4z)F~Vy(v4b#2m409a@y~{h7(*;_KmuuAkY}CSkxt>!tqCMoZ2D^}
zO8hVVq_yVz3Y^-2Bl4Z|tOF=z@>Mi>;U4G#9g`y!B_FCye+2qzaCeBKM5ea*gt6iz
zt>m%NsN?b#-%jy<!v!N0Y6q>yZ;mUS=>xK~e_ytNUhFyzT)cet^DFrln2yS_wyNwk
zT#&yQ(@EIVuDh(A(LV1XCRP8PnWn*9=<k}jVKX03r0#x?h#|DN&2(o8zQp9|W?{gR
z2&XNUMg=-BvMpAbzF$xOnXZh3=vj5adVl~{kl{Mnojc=!wvfg`02`yZ#qW#4GHM%g
z_N9er7sOIB#pk4IfB44?Mk70rT6w}UyCe|LA_KDcP_$K+)?;HkDoOK{b>*nk$xPuo
z>RI9aolCglT+l*JERmCrpG&QS$an4r1N{2d{c0*L%zEbW&{s(*Zy6(4;AhZBAP&uQ
zd&+0Q*6{d;_P+x85k&3dgj<2uM-`jDX+rXQ?Yb4B;jabxim<7jkuc((Wj4I@tcWoo
zn#0%0V|Jaue%qY+;-g>$=$iA)xNV1}U}t5t{MzYRi%9WR^h_E?nixo_TRd5}Coq9n
zj37a-eyg~<RnLCmfKf4bi3R0;ke2oVWLf69I0dV;lYQ}QJC8XbN9%<Vel1LJijyyv
z)NnU4)Zf?gQiWs>KPCXi(RmUXri`+PRd2InVm`&%@{f-MmaB7LYs(7`Amv1JJ$U)_
z`N2cVlcQ24IueLz`)p@RU1!_XGO{lv-n{`*N($GM3(1}X2R9GirhPgy<On|ygSVk)
zMkAx9LlSBq5L*zsqNU;1Xdo(!Nt8J_MZ|n}M5T7J6*gFz!;f;233yDQ76bsl9Lbuz
z?~!<~BI?gyElyU<nCAJ>F06#(174YAV{Y4%{h|*BD&8oe_aM6^=TckJombHJfZdo~
zlL(w*Env7@HoPYt3<0B#ms(5hgx+%=dvhu|qj|5Mz2m8>@~NZqa<963GvXBk_pU7i
zm+!NcSU~C2`%fa5eOBni;~?VCL!V))-i)%f-q2lWUd90)Zt9()^LV0>mkV6DqJj|t
zl<d;cKZLD+d@JcTUq&Odt$fcVhD*mjmKn7CRm>gHL7S&FYBuCmL7Xg)1=ozU54@n-
zE4|p(5&+*1>0i97fWcsj_TGr;-x%d#FCWYj371zNkL?kEBo`s(d^CkhaA)LIfM6dU
zp7l3jWD6GG(IAd=*DQ?Jbo6^PNEXb#2gxZ0XTK+AEo%(XGlQjFv5tfwM|C)gNRV$!
zyly%a^O_Up;_gcow&Bc{tlbW=Ll$%j#Asu@m;)t8gs~G;ft!7=M7huzNVh6-UuTXT
z)KE{ij+wPifiGQlM9{WKwfBW-w4eLurIW*U9BJVelJuODk-TX3XF-(jIqsbJS9ooN
zayM`D67?C56Ztu|#g*Zz*rFpJEEF*^uD4NV=LUS};S>~OPX|w7gG<7^*f8A4r3K@v
zx@Opne}iFbRZdGZJ%EIk6a7h(D;60^BZreiB~UJMWa&pA3r6U(qB>l*s1;@>^?_}P
z7wS0vJE|NqyGP;EYq2aDFArn-)_Tw1E-$6;$sdz!gme+UmGq7a1f}g1*<pb)L6Kse
zP{Qclp7wn$|6j3TGBQyxNa=}a<<(18fyr{%VEdi4r2`%_s0Ds^(l!&WS%qmt+;ECq
zud>W5bv{A|;35Lb;GD{KjtSDRsPGD>Zy&y9(@Fi1>KJ~hlm7Vk?!GX&v`Ze9&yD~>
zwj{x_b+cwKVtjmZvF0otO}fzm3&fvUu!8G_biKllbC~4U8fV`e3;?T-gCWM^I4)9X
z3(?r{bk|q#8Z35vzkjEbjBwg-e-;#xx#qY>#lO{cU{Kw&LQ))U#2djNBOS`k!#Eu;
z^y$0UrJS<2?31E=A5&fK=Ns`}fd@l0u997~$3N3(C3LWM+w#QK{HW_4)d-0fIvkBV
z;_%vSt7+14*!hgNa15>XlMq^5JZ}Qx2aYWyYb{(XB}4KB*nO7-=^O^Jh)E<+ZqQ{0
zvD7nw99ihWoTK&4Sd4|PuJBfOB;{h5-A)XHzT}~Pi7(Eq;77gIaWk!{vzG<p<>dPP
z!Py%D_SGh5p9o=tgOZfn0`5pEP6*+j6QzcIh7*x`^HO#7Zf-d-t3Wpz!+81<^COpw
zWG%LS{4BOf>gO*Ujb^5zE6wj6KO7y0ZE2PnPJTKbv1^`4(4Xg6m*O6zJbhbz+?XOA
zJmz%Y<f|vGVJlhs+pj0T^`1Dc<^(f(y0MtQhL~pZ?=v236WYFlB^x&}hdY<uOwXy2
zo3LkT9SBUS%wC)q&8gWQ$@wtu1h7||xYhE&z@@HpTc757?<v&!a8ufp3OVm2o4wHL
z>$WEoZe^I<c1ZlWOj9e;Z9~W+o!b1yRt*v6`o3S;r=qcj%lRy{aueD?e4}Emg{ALl
zUOOkp7qIO#aP$cEI+AZO#8+J;$#FEAMbQn6H>bL9BzG4@<r8(OaT&ulIa=|%ap^K8
zhZw_kudbvE_`e!RFhWv~3U^VVyzZ2#H+O$slmIqgJuMCJ!=8vZ1H&d8gh-G5>E$<3
zS1~6`IoIRtYK%&&nWpNxhDRD$+Ox7ZAE?`%Sx+iBzw?wdZ@2F@Ym}ocY+-&~#GIaS
zdvhid9jR0({<_my8#C{>&*02YY~7mzX%@rkDzee6qyV=0jlmy_?7F(yUMrl_hfp%4
zq9q3J=Q8)<c&u}=$fUcs2Em{(v<zm&wWGRhF2<Kv)utbS>eOOuIkZ&k5-U82I)CzA
z2rfK_ns(4hjp*2wLDx|4ipdLA57Tmz;f}lcN?%;ilQPEbXKo5uYvvEU>SI07ibKu&
zMIXR8O6}fxjoNUjsekGR66XQ}UgO0(89j@wlEQO%E&M++zVrB1hg>l1=}lY9HNP%E
zixfUYmI%WR<t8n!KaS`$bd%1Y*wy0k^Lzs$Hf0m~%!n{AxgE4~iTCGQvFlIhCX53a
zq;iHHLFy_LRdkcNp)a>aNtc-q2s8jamNTwEj+Y&1d0Bmja%<PsW&<(4?kcB@L&vyn
zH<F7ikreejLFJ)dvF$jL=S!eFBf$nI9a<1gUMCPU9cC2&9Azvi-S@r55w1w^F7n;~
zhkpr63r}5w_)#r-Rv-23ajHNbVTY6uIb`*;P6?apNdTljc1$K%C3Mi3Zw*-nqHc#O
zX9u6-KxTZejt6)5<XthoPb__x7C;FBajpvb(^u0VC{aS(^{i6{LZGYgSWj~|^aR!W
zGti)f6S+yW?g?FYq)bJDzVC5#a9u)iA335%TS>yu;(BMLY+EB_uKcCJUe{iXlNPF-
z5&tG$Q2LPyEGhzQuQf<6G>y%6gxjXGm~gK;$E%vbYv{}P?`J{fA5K8;x+(<ZcI`y(
z33OIXcxF;}I6byk3BQuRe8URg9r=He)rl;1nzfvS4tZqwx`|tmG%r>?e>~DY{&sxo
zV&gHF-Iq`e%pj~DmHx{-jzFsbaZ`7;2AzcFb98=?V_M(4Przj8!zCo>zH_$s{3W(J
zYBqjbhGs#2e#+R0FQUK#1sBHk?$H}zcT><YuHD#BAxH971cZJEd?OU~iU@E{;lV*1
zY~ot&X?q2r5b(2R307S{J@(o0Ot1w%A}f7RuP1Gr#P~2Ph?VO_d_qp6(v*cf*hnR^
zozX>QYfU6Y=*?WXzA%>q_$9R|2}Za*51~0dGPCP1MspL12uy--L&u$^&Q)Jyh)YWP
zouylY4IT>1ctO6c@t*M|`#aHAQw%$+Np%-9MCy2qc~eqc4N%(MUUM1>yI<2!2%ham
zPGlL6SgFd5<lWaZbFa8+x+gG#?jeLah;$cWt_MzpJO}ELuI>C@W25A}fMapFO9@XI
zZxRxS(%env{J;>{Uv#Xe(=J}U<*-~H98+pNoAdKcZFG=348+Jyy|``8lYH2Fqh}UN
z&A-)Yhgf+RPDuV<iB20Jmtlf<-3AKgPQ6Bfbrd0qw(SO{3RsUR`GBwE`)&V36CYzV
zj6pIH%1Z|!2aToWB_vE3Sj&gfsobTOJ>qP(T13BR1R3WamJo59V@>VrrqX~kf+!98
zLQk<~Pzh84@&H3Q?Z(94bnb_M55E}rX!>Vl(Jj<~(l1B7QKP(u>2$4!8I=KLA2O*h
zYC;_ke&l19*P<`{o>mbTYmk)F_gvB2$GD5bq0_;|1rF#NJ6VVEO-H9PG_9Rp@A>!p
z0RIdNdnmdkiHw9x^29if6(`};ATd2QeHRxYl$~RoB$%^0_Z~2aM}0iQP@X)eTqoK3
zD~SAVxz9ma`@-IOvA+9sLvVoO1E4^cwe0(dt(KtM{>_O8$W(z7pghn3A~P*Xd2C`<
zz*mX>W%)zlMhpMg^-rG7G0P9RwY&>oc2G89uuVU)K6*dI+3s|-_2w-hEBxM(EyU}1
z=6Ao|Q_o&NT`lof7MhF`)?~{np)vZZS=>#-$m>tQ1IQisq7Lx_DWg?QciB3@(YpOq
zIF(y!Q~-XbfOHq0kb64{PFkatXm>7yY+lV-eqDzj@}OS}EBy3v(okxz9u+pT%zZjp
zCpf?AMQ2y?_T#G3Lsb0F{|#N$QjdvU@}ow#(uxE%kxdrSDIxBkd*&}k*^SO>a!kjp
z$y8WW&Wwee(7}dvw{Crim)9ayM#!2w%Jbx088jw@UGbGAT~-@E+m{cMBzcDaj;#%;
z^6@#n3KCsh$U6waAKs2ljmos}&(tB^**llSLng*ZR2!|&)`-t)cntFu1Z8i%7<zV)
z|AivOXqC5Z_{T#)7D<$pHcx~aD6vT3@G@I}9RA;-yaY$*RC`n|t)MoZu;0Wt`8*n*
zjv1AXFJs^pp(P>+S9{dTfxAp);5!T2-Li#1vHmSk!$uKfA0%i3CF_9#;cYSqDU<0*
z5lQhj3_B%{u@&7lbZsC6xC>gA0WGusFZpzQ8tH|^b4&dX))b)rTxx;zV>Wm?_Yp7!
zm2qgGliYBv;F-Kv1PF5XKuX`M@^baXvjHg$MZES0Fv-t?*dL_Y_i^#jWf8_YV#%I8
z6S~;O5IPVzo<`L{>+*OTKQS?d3aa-XK@^@~QGJMFoTO{*#_FS?`v9|~+pe!yWR410
zCXZY{ke&fW;ocEB#LH)E(^%sf9=2s@fhDkE66Q@_zhC?k>j^&9&!$ZkO#5q@M&^Ho
zQxl;4!Hu1S=)iL*lAMWCR~pXNVd~B09m>Ay#uPN8vp`TW52gNcjnw7O(P`07K<-z(
z(<pB#fHGYz{#Vr#(e7(3S?G7PX+|#+?7(O`^w3Nqm8vPh{!oop-^HcJk=Xy6N@?^%
zp=`w&U4d-(wG<lq4+5;L^YbWEWYGsv3^MV4K=QlXSgb1bcpj`6!(ov>vUDG0uY7Y5
zF_I~Lruk!+&9P1#gU^<6&@6Hxl8|H*KOF~tanfBrOKWWay)9BJBIUSNjXbcQ2}*Q2
zP&b^I2>p4wAYU}o87iEB*pU^D5zHjAo;|BP04z_?M9=f1RmxxzL%%-<<}OZ~>Ijq0
zX8KZrV#a$6(lXH*rDR$Uf9c$-famc#*erLVjgA?}01#x^u04kF(W;A(n9eZd!Baj#
z)obMWHy_&v2_*qX)U?gC4}S44l(T5Z@=}+bb3M4(pJ>YLeKFRPKJSAc{sbUcfs-A$
zQI2{t)Xv}hOG6<jJA+F&YZ4v`fT3V{2d9$Abt=gD?;TZoRAo!#NOKN!HYh3WCMT>K
zTOghc$09BwTUV^<WK_Fv=)@ibI;jACHk#*E8Mi{AQW+yf!5KbMk+0Lie;VJvwznrA
zzt}J6`u)@UZhfq9MLxHO$%2F{=S;L>MB!3YbSVL1(QiNSH}-Bgrs%|YG%FoosoP(F
z4qQkW{8U{bs5<T|Hf(^`R^a<RHBj2u`3s`^JlkRkCdJ1F;JdS)l7|0yN7Fql%4cch
zt@?z!xzOjVD(!8?X5uB&@6&PPQcA6=>wFQRF~@Z<&|h%%f`Y}~rb0}YFGCGWRJZw+
zo%kT*@~E7uSjhZqo+MdXWSt?J#MqIxnV=u6mU|5Ddnd3KXd9H5c;YeHFU>MEbQx%J
zKC{17>soOtW?{|^UHcxcvg%@BTEe{B+BJ@_%WX~p|F4!8P>h9^F8k?{+l;d!P+rw`
zMSE7TAI5j(35pFWSIL<%H8Q<jVt*vAD)(_Bjwd&XQCsCmw=eZR?;(kD8J8=4gTbxx
zo)G+AYK12dS@uKG`KKj<zNqVouik|@lj|s@G8ZN_Z6X<M*+}a4Jn|WZ==@sa^W)Rz
zCH3Geb};+HiF~TgEU~j!1tvx$lZ<L(C9cn8zQw|m$NNY{T)`N(uWG1UXG1vNAD)Y}
zfqncX-}nIYA{ZoO{nTjSlqW4-V()k+UoFaGDX2=L-i)0?st!ajqVU=4uEgJVJgW;w
zjsm;{8S~bE1QhUNc&oskT#eVMX1s^iQP?JiKwlvOa>h1<1EUYkatZKXUWdMXyz&TI
zScH~ONbh;AXmB4MXXf@+Gf%|qy7Y_>UM8aqA)E2;(aEvmObyhXp?tz6n0$loPs8kQ
z$}5AJUbVTuRp!I}-Qv#NT#<p$x9kj@W>yBl4kaU(7kPu85QW2`B`*D)r`hKLx!XUI
zlV645cJ(_uKaO@rX6*+|qRGN*zb&nU0a8(W%mu^ZPIqLKK{|BVO&MF?01xA?13HBV
zFfDaHo6Q;#!OLI1Caogxk)zP=0(#Fe7Bt+o#X*wn6FMBl>I>?F#K66jQl9K%>mZu$
z-wzc|l&J>982MoKZ3tcRUUGwa#e}Qz`*%=S1_-y5Ob(z!+o3Wl%hBgIfzinD8tv3K
z1;vZMrf+X_l`<ZEhswb2C}S`m<1D|Schn;!i^-23{Xw#Lk&pQaZRtc#_(43UbQV{m
z;Umm(4`!Jd|F*Tv=Ufl<2@{mTT&zLRAf0+;bbn9LX?r5~-9F_|92W-LEHy+)X4^=~
z7@{U(g$Mp|15|<1OWz?2h0jQ;pEzOYM0y^IicSlu;{rn*I`Or=CEfj!IVu%#xqRu+
zc$Zj1eU{%5FYer}bs-4Z)mo5XyqUI?@|I6k-#O1g6+fAT8eLCCO+$C<<VZ*M-cZ0l
z$Mp(ia28fS9Ar*XJHU)hdAmZgQ;26TvISp{RXwHcL4=Ackdo3T`!nDnCFOp|UARE~
zg@;BUis~T-i0pda&PUEPJG*_3%eRckix@l0l_9D4KrC8WTV=WGdMlXwnaS(z7x2Us
z<kd<3%>}qSJ&?t`2z{xQu9mh%BO4zaJDFfZeLYPvlb*2U+StpWcP!7h?G#40;c)4d
zt(1IaL-rad0+~UoelYAmr0*O&8voinjI5NvMeWA($kOomTgEf!mtQ^U9;ZFg9r~XD
zE^tx+^lZ<(UQ~B%E6AcK!x_H+;t-=HFCk6)g2CEzU%o#Xj96*NvRdxTTD}?hfwS}w
z72MV@Iyn2>Sr2V%osoXs)Q%}Fy|9cw4Kd;sjkpNx$#7N3JFeVG(Ver>QwsMfOiWQc
zRyYj66b-md4?ZAo3VEwdjK?y}YQjb0?wgnKFqI<mU3MugPj)PR^rCjeYu?{gdct99
zcNXJ)0C4Yjy`^zo&5Tv$9=?JagKK}}0S^4cyaf%=7FE&wVyhs>clf(egm0W&8%Q>S
z&;cM$9B%N1S8AH2#AxdV()E0N=!Zo|CrRP);v>t8E5Tl!i3MES>{x5_aF|*AcM+0|
zyX?QrRzSGPqrvo*PLQT*i^p!c57ijrMq)bL@HyK_+aMR)`{A!CCmDlaAc+AqilD2q
zHWRX1Lb_a$G^<JSXflIp^xW6V7_Blfgv81TT4nwwe9{Iz9yQbH4bv*Up44~q(mQu(
zoeh4i)GKqWn+C-s!icJR1yq*>>|3OnyXr^-GXBOpE!5M7n|pKr+snaHJyLB1!f8aZ
z_{9P`{UPbth^Qn?Du%Bx-oL9dluxV+)`QVY_ky#(B4~6WY5qGc{@pzE@Y}2X>^wDT
z9xeO^WSP&(Pl`FO(Pg3-SvM6`W7`rTduBMPo*gEJHz2_TD1aCQfM)Dmm7*CVFXv(=
z^II}<vibxJ9+i!b)FF!<3Tz^ky^EPG8-+%z7ekQD9TkQLr|s)STb9dCkFb$J=j8&;
z=+^ed2>qPg$d6-kKvebyrtH(h_;|yPjtEbwB8r8`V}d8GHUK#joj?={M;>I(=*d}=
z>p;tzLM+1t5QJz5)!{%tQ@Ue}%<r5F^@d~Ng`aV4BZG!_OW-CCNS@T8+xp5++`&U8
zC>B>+gh;c`Vl37&!82^&bbP+b_yLI895G;BkSv}|`h4PCn*X1O^REP_^^N3mNCJgl
zn57V2re8z=ELd7TTIM5K@2NS|p|5`Treubr{JFMjLy>!Gy7dUTdXzSZLF<YNk_wcD
zO}|6k`!qBQ<B!#Q+z?gK!ZrEj2QL^bXz_GL^u5s%pQ*4WzfE|v@V|R=W+7pYi(P%a
zh9aqw8Ho?!d?Z>_$EzoI+JAHHoq{@w>Xzd%9kob}lO$X5V>4eL)#byT3n=uUZ92iO
z*Qpq@d5BE83L9bzvpw|;{}Bgdz4k#7J$5%t4vgY-%+KcZ8PSMV`tR!c&yA$b(qBA~
z#KW=mm!_I_3Us0teS$1(7%@UzB6UE#G~$JY&IBmD?s~9mb|2Ce<64Rl2pYzp$PRA5
z+1dlTvqMbc=81t(^c>;MR9~5Wxr+j3(ZY<d5|Jln?VS3*`TrGsI0}2JEfQAe6EgZw
z5FA?@Qe+z0uNta}zxMEZb(*iRM%O<N#(ESe<5NF*ZEKC!20aRl@VkR0u=088vX;es
z0)?t)n2xa!h4&_?oPR7P8*Sw(<Fz}=#fV&rRKKD3qJmP|5ybxAtpV8gb{sI_{;aS-
z<*07()Alf0f=IH2wJwX~V$%z=6%ArdQfXCI%(&&JRzss9BLwzwgw#UzNG4#8#3lIU
z1;euBcL&1~x_*!kx<8UEDED{}QmXoc(P{~AAW(c;4Q#xGtJSzdPQw3f&^|4C<B%@w
zwMFi?UslC_RHHT!Au*|s0}6@>gRm7ZU$P(Za!zRkKV%+o9gWK(0#VpAs71<6?hp$k
z?S<=H^gkwPa5Opjlt=7PrYzya&X>J%=VY%Cv{odpG`$?FojIIooVBF*A2MDBbh{pR
zOcINcN}A=E`m4g)64i=@GL7=pwbhb>R+$n29b$LO<7$I+Dkf3oRsW9_Ht1uv6D#GP
z$U$v)(E<z&e=7eveeZwa#f}~safKU=><vMuZcnipUryzIs^R<<Itat0g0w*g7oI=G
zl+5?bR}DnXivk1Sb5`nY2YqUJzmhMKjSio=YwV0}e}L0P#masfhAkFK7j_9RKNdo|
zdz$kg@9k8-lXUMkNO|WM&Dbz)w_1EmVez0)(TJ+&LQQM_>bQG)6Qm>l;t7Cl`jnjp
zP~V}`w&xIcNR8D?%#cTZ-H*Vq+EjU_TW_p;{!lCOp0ZRjx4Zs5mTtGh3jfJP!s9<#
zKC@tuualk3aP65RfPv-gQXx_#JZk$)JlVgAv!L(+cG}ziTW?$wuKzTS<RwF14La~{
z`&{PNhbo<_l+o-bJ^gkvYA|JGPTZAD$woW9gkFyTouq6YBV>cX<aX?aapVqgTSrj0
z+h1|in9@LKZY0ZZ+@cLJv7n{doPn#J7v-(B<%|Ns|AJQL7k%RH^XMzX<k72d^ELOt
z)<R+*PjQ8iH18d*jpy`6O{_g4pQdAvetwL$^1d69A*<KmG414%k^V;obRk|RgEFnt
zua@}GBCxJVEoGvk)M0I+Po2+dI$O3J|F7LNeR`%O$Nre58=;!u^I4DL$(Mq0%pCa>
zdCRkk&|J}oj^&u-_#V=)c>2ttj0ckC%O;hzKMtM;mIM3+r!WVL(_>DnYp$Nbs1+gx
zj&}*l5^aeJ*@&yYi2G_aB4^PQVQ#M=7XF~e3o>te?JL|GGKk!ZtI)=K`crjYD%I^-
z%MW5Eh#wU(E1`dai??fbRWyBOT~~oA$J^AT6!o%I2rBw$42XhgjQb8~xmXE}3eiY#
z&b1IN3gaV^VNgAcS?taBTrWX2IIAg^Tx^8=r(Oh<+s_N#jZBVfGH!qCyO#(~KH~L4
zOxT#Xsi}I@WaNxU*9x5rJQZh>#MaRijWPYc@X$P)#pPU2_<1?{!oqFUSw)^%2Kbg6
zJgBB9WnzBoJ5hdSrR6xd$Z?1dNHbi=<Q}|hn9I=?blX9)BD>UU8KX@e(9=D#c%7jr
zG;>f-(dXq{K@kyp9U8&Kvw6WY{1DaUXACWqFIoc`*UnM0e9K5gR^DIIWZj8?0DJ!J
z<;3Ba@RWx&vzIHLt?7JTX3Z9uFsRSd4h6guxJXnqeu-g8i_u|;Lo}NRE!6=XJ<9B)
z#<aoCDygdw?yNQR?1qy~S67#-))mz|nNOCB&9{*ZxunEmRlaw~vlW&L=6CBslHVSy
zQC7DqM;aY%k$US66^9*L<n>q_l_dP6H(ak^VVut*`ssVHe!R%CLo~OA4RIMlb<~k4
z=4=UHBs>~dQxZH~+|nvf*%&Ct9+dTIoi|sM%}ewT62n8iz`SlBFOt{cX&nD1cJFyi
zckgs?>)>qd#sxny4#YKci#n6iPCw(jYLqFJMarqADHEsp=bx$Z7jZnzF57`fI1tN#
zU^qW24x~dg`yQnFu>0I%z3UaTf)P1_?)+N`RqX3B{YJ|Q_4wgB4RR<6TkCDG?u&}$
zXv9o99p^<e#VJALU3{zzws($a9xH&{Z<-e=e+7|334QgXKHFz2B#8tQOc}hZk9hAV
zLC}`PbK4+T3FFS?4b{irilZw~U&vBRcm&ms80Hv6p}%Dp)1t6zu6nchzM%>ukYrdj
z{#Ijx=*~A_gR_?>>7hej48e1z4n)L?Rw#%P3NoMHOoy<4hYq4(W;)+F*M7*K)-dXO
z?FjWPDTY=}B`SL!yQxXGyeg#u+y$ND?PGa*ZKiHKthdgQ#<}vxiz2a?C)Cp$l?JHp
zCIMUS*&FI+*qFmZa)=Qh<Sz<%GrQpbr@Oa|tEvh6M%Sj16i~WklM)gFk{ghg5Tv9H
zQb1DajUe6KY+69Nq(P(`q$QN@?m7#+^}e6y!+C%2`FOqzu=nDcnXA{#^`Aj0W9Gj~
z4DKoe3pqgC%1zbr7o_7p2cE3ZMYb1ay17x=sV&$frLPPVD!{@yE`eBoG#a~2tgZSo
zXX5y~TH++BZ+N8K#QK=PPWp2Ik;{!AT;$Cv_}i?6kIpoxC%H@049Z_87H^I!UCbL1
zg1eeGpop6qHrFLS_UB<;PW0KaLAO2|#m;jwo=j0JB&6ShB^TN4Ef8P+3JKALcI|%O
zXUUQ4xhik+z_QoH1+WlwqIVk48WPmPQrW?ilhq&=o}xUnfJL#~$4No`@2LL@#_LXF
z#O^c4>aO`HXhLMligg_8VK<QvQ%PXD^~vc%`oJCZ=h}EPY<$5<oUU*CS3axny6d9F
zm!k-$8IxSE06hiJ^<2%CUZwbDGPQro<F1T4%kbE}Be`9W)$-0|+z~8rnE-PI@JJwR
z=x4w!K)rKLm@2Prt;p#yVT-nSRd`E;_PW8B$mh^2@e6HdJ*pChU7py49pYEn@AIUr
zlITVAe|Mw@h@S8KXn7ruxf=)@3Z<R2ya-<<;P@~U#<`?X^KdBvtkL)*qN3{;{7|3q
z$5Ic5yq0anG{QA@qw<=CEN2?I-OqlB9Z1IT6<BIxypQdn3eQk(nnm@jY%44Ay?Z?)
zML_fiQ_QP48Ypvkx@H<!9WVCu?m&20)p>FOSXcE|-waW=$;*sPAv`-guQO1HQxcPw
z2)QS2Js$nuWr6<H*yan{RaZh?np!7&`>j`Zr}E8j_@xMhbb77|eUNwsk)_Bv>QII%
zJZ%iCI>1A}E}Q#nFrZvg`U5>Ws1hcT0#Q1&_Ev2qo;-gHG<k1ly`w91)K31+4(FD@
zb!WQI*F}7*A?G)B2KZKAZ^vzm#C!0=?0O7>2S%kP4r8UnqQ_51d%(ExFN2Qlq)qMJ
zvX-}tpndUqj1RLul1MAxdVNP;g7J+L{cb$oJmv-xiCKP}04z*muiabF9tRoBW>bV5
zm+S>&6manJ`ii(IddNB|{H<Esx%Hv1j$w<{ONtZM8}t9Dupfv};7HML%8ty0=9OYP
z<`}vaf20SjihY9<bJBV1F2<2#Is^-oXMU&?*y%!2{NkI93&tz#8&RuUJy74eK&^zp
zi~7W(9^Lc8R)~=DH`8Ap0CW1R_cF-OS(A4Xj+CZ}{_PEu1%6Zz8WwV!Z8IZu_?}}%
z8mk8{8OFq`QZCb%10%+LVnGt|1IPQ!VS+%9!w9{s%AC%^bC_bcP5;FH9f-^Oi-}?_
zg4c$Q^k+jE0%J4DS9fmrvm(>B2}rjWN|pD5>Ez#uF~jkCC0-Ka?aK6adtk-8ezY7&
zhCYy5cpO24S{SKjy%3x=p77f7z*l+o+r!-o52Tj+C!$QLlDmkcYIV<MB;U8Q<K$U)
zpyQhAy8Xm&c~F_@P$Eyi$Jp#DLGB^}Y$rd>XM#$^z%mffZQOt>z%4;{kHz*~QWewJ
zZyUWOad4%|cJ4;=kYU1<sq)+}>E*ICG<g|>Gok5a-G+1g*+VR+uAUmHYj!x9Gg!0z
z(|r2Q3lnh?5i|xs%cm;+M`zjKs=cIKeHbeaQ!>NKXEF1u+ZXp5{$s8gz;wuyEgcV)
zwOgxShvZP{w~*;CN_mi(G3q*EEXWELuS{ij{pyt^f~O+rtD(%YcEfyKeamEWg_s6c
zHP34=JMUYWTdt!OuCHA7)t=-<4WZ9FQS#sbT}TTJaPTUrHZg08q6tbZTlbn8L7axY
zPVK3X#q9H=C=C!OWRc`_9Cfui(4A0Nn=WyGt9G~3Jf6H^CBON(5(wGBK2y`lQF;s!
zqU(D~c>XH#UR<dQ1*Yb6L=d!DBH<$z11iUK2JS_lWQQAv`GK>hre4B5fpBD8X~cXr
zdv?mZgqy|MO=lq+J&9BWuTYu}Ur*By{In<1Ms5VqB@)hy%4`AkC+~cJ@w(2u+#<JH
z<_gQlOU0nsM1YvEsU^%Z*qz2zGGZvggjPX)nU@*U*F%=7m}$1bWaydJG*{!woL2ZE
zX}C<mXT^xe15b>thRaFp-0wD~SL*W!`8~uWk`$AB3Sb7dl{A1yiv_`nG!qPLiSla?
z2UIE9vb3QYz#<v8yTEUZR<_XnKK0$JUkrtDStuSq>yYf?hX>GO=Vf|rhtC~`Otg~Y
zSaumOEFkwE*f1a9(2uW}E_!R&<7SScyq%R`n|9eKKY9pOq>N7c9Y$@5xVD{?daFUP
z-~_2K=0ywrS<dh?OxW$nFf=IrSog~G`l@~_em8ny9eK0Tg%_WmvqFf`^k|NczOw}Q
zi|Yn5<@1x-^3g*R%kL*VIq&#!P`Brhr(UU<t^`|N`WCcsI{Spo-|jqdkC<tR(}~lY
z_+`gk`|+NvKOc8m(*5_#=dB;<upp#QKg>U6Cln$zd$i2bwY+R4AtilPgXuDxqf5Ak
zWpPg-?@=<E@}mP$u3ZTd=9XqCDg;FcnRMhQHv8pkR@Y!FBy&$Iq<6ihOmItE3)7X0
zYyt0M+hZplU8=kcSAqu+tPDhiCqw&V?^II8E&YqG(xLDvIq!QI3|exFfza16D)M>A
zgSd4PD;glWvsA-U1ZzQ)BpMQ$Mbz|l4(%cc+JPi@lKtlCPp#?Xz}|EDCry9TfD1_=
z-<bnOC!Q7uN%LgH6qkK0L`>>?mm)Q^0a~8ge@}q_7Cw(fgib%gS8R5pplrt}1UR6E
zxa@Bqnj(X5%Y<G;`R&n9dk<2>gm@>VWiNPIiX#QK^ouf#=84j*<sB}CTeI20wY=ZB
zF*iE+NZn}f9p-_r3_I`#_(T#g4?5Y-1l4fL_Rlh0HSb3r6B6;*rpcqgJs4`DN)^AT
z9l=ih0NhJM?ci_RWrJ-*%DsEC7_hVUkbzJB(HW+C)=0904*JcBYn#r8!@`AimldsM
z98bX_SsZXF%mA`a4Jo4%;pouC=dqO{z?Y(diBNv5WsnV5F5t-TZi+?mMnj>;P)0DB
z++)TtE#8FEGrdH(W6b&lw4J)}X8Y08X0pIVEWd`F269NSOkpv%F0#hEH0e6c>-=$R
zAUqdw@Im<&OEH1mA_N5}<IX`;7r*$nZ9PZEW!9<CNK0SJV*zFQkn!j~QFw+y?!)&T
zZgtJecwqKDAgfjDXGnuinr?IZt=-oUn^Qa)NpvXNEux+bT)lTk#7XS8W?cXB55&yc
z?q+<&PKGiaJf>E8VbEU48xr=~vG$o>1jMuus`WVj8Dj==yzvkjzIfcDHR=(;z#%Z@
za69~-D-isGwhxk786tRq4#j*F{EQW=Tc^b8*Q%)N?n~o<uWcxQB13-)%3Ce_)Q_~7
z=}6lsuH#zT?)>j&)wQVDhlh_5J*3+;-Cz;AtSSoOXEhCf*+Q!?un^Itxko*y+dEsl
zcz8RZFIkEBH^1rEZCHBopp%Sz+x5uB#BvbYH=5kgyJ6ck5X7CJaci$MKNJPU5ZV$t
ze|u_3<O_LE=z^h&Yv$CV0czOm_ly$b6dzzqpH3Rs`QJyLSM?euP9q(+h3kk)KZPpn
z&bH@>rYatLgnoc8Xhlp8@m~ea?cKha23U@a5XGGRD5WjNg$&7HDm5wP65s4)?2}d<
z35t>%?Ec`BN@M!UF;>X=d0hlfE*<QVN$Y#9t;xphQs7W`6pD$|hZ$-{euR#a(|h%F
zw_$JtcG8YkGCJ+}BI_nyzXl5JGlGZ2Xc9y|L6VMRYNZ=zy)#7iuKs;-V5dYdS*3}5
zNo<hR2WSNP6Y3Wp9cnmHafAIm%?nocxN|&r8~K*JP%0<CsbC{dQ0=KbhH=9^gb*es
zU3nx$;Qp@YdjB<U;HI$pdM0Q*c~ui%HA&-NS@1Hv&JseOYPj+<e;3^TN!~ad${~Ln
zzkiyBOsySLL{^+`Bwt)(jk4901uj8dZ#TIO&1-3jWzO9&)h7stQR1YL0F*5f>OZ#M
zWH+1MgvxJsZUEUS={?%LWSRzqq=OimGySD}R+iO|r?z*ZdC>dfQNz52dge#6Qf0&Q
z49AW?PLAM2UFEzZ$<74r^WcNlFa)+6in`Hv%CdAG_<SO+V(h;8;>>ZD(JgS1=QoxD
z<kvYF2yak$O^|^Bki0`Rqu?_*YZ@7`a<42vv$^Z-DTI)uMOQ;lxaIv&G4|bJ2E&P>
zZj9%%N0g}B&4mj;o)G3oA9z?)x&<lYKjx-;f2-M8_uy@e7WPff38H2|-QD&QSTHV$
zj5o|wP)wHZRE!!_r!u=Zl;inC%lvS@?3T#=EdpD7IoYZ01gzIy^SbWC&IHRVOz^@K
zeXXrM?ZjVUL+(gOci*N-Zui#A&EpN;xXWu;+i3()FH&iGNlJQGsRZ`O!Y$@m;4bUS
z%TXud_##s$1re6gvD^T9!UUDrjmLZxmp>1(lU)vA6kqwLzyKCa?<v^`#Lp`v#$N@o
zrWdSOE`AP5aF8DOubdGF`tssNxA0NK2B|t2jZzb9wcO7f&Su%Pqes10l?haR<tSAi
zidXH~$s#4$KF6HxcjvHTaTud%gS5)vTa;7L^r+U-MD{wf<d%U;+Mp8zVK~bgNMqa`
zon%2g3J|&_9AD;f71fI%jxN;&Hgm@B*=kvhWvkdcHUnQX6_(a&u(vT26~VVd`gCFJ
zwx`gRvXBdM0AX@)SAL|nCjE)Pe-%bz4|;ph4D#Z`1C@I(r6opr^bQk+q;EIIbIQmW
zDG-p=Nk%iKqR)}Eu7E)N%0MNDz0ah|*b-!u>R_E?MEjn~oa2!jfyM9FtACUbJ#@<I
z4igs_2FjBcCy8v@18N){+(Zyo;dD+8g`x!o%?9fg6FV&-Vt!!@l<k8)^2Z(?*m#qD
zm$K^00YF@jC-qA259YsaMwFm2RjmoJY&0y6kfO_2{(PsT7W<+=l8W6b?m2lVG(=fg
zBOie3?`a9m@#~^&U-bE_X2K;99fy7?JQ=369bpu)J35tQ|0*#Z#N`XkeaUCLcMvA_
z^fs!S&mGrfPG<7?5aYJsv54hrO=%OE#QN@2x>qCat4lkH>_}gXRT|gJCgTqGt)Rn5
z_=&O<)|*{Z<`{p~HUwpRwd5RO;=zCxE=4C+MYdRo!T098rFWRlhO168N|Khx;`I>w
z)TsLjCRqB8Tfdi)*oW2sxKhm}i#a(10=Nu@a$k*2?Qs8h;tfk^8|=9LM#=P~5wk_n
zhyzu=)J0mgdc-hwC>bM%fbE#m)P2!&FE8Ro#s+=vGL~Uv84SmR{S>IIRGD?!sG7h<
z6JN~?)n7+6gE6};%FB1_o@5w^GihAp=<diR8`8noqw7jQk5wU)PPuLQSf#2gYXfid
zjQZDI*%x5aE?S8#Q&Pm5lelHOy4)a;AI(5ZH&#-iccjaz0510Z21lL-0hds`PjZJ<
zaF16cjWw&Q+!+%#9h|eQTG5QoLK5n@;VOFpC7-K^*DWr7c?7&Z8I-Q~Q-XqTXU*u^
z?&m)v%oougb0gDAAm1V|yUa#WKhf!?;<)1EHUGDd06=7Vy4Fce^#Qm*)|q482^Xkz
zg$r?o628%4?&8jvk8e`ZLl&nWgxRfd3qGAIT6D7eDQJyJzk^bAb<ps`sdrHJr+4mz
z9d)19qXZZ;ny=7tSF8*M!nXf|&rIOMtN(UIC2rigJLnNUpC;qn*-y(nQj&`Nfgwh<
zJEh7|%QL%-RU!!z|5Q{UJBv$G<P+xborMnlHaemyl<nhhA(NtlZ!c>7$Sp%IR~9TW
zXA^38L-G{MVfQnJS9_si5WOUsJ5?85;S`eDJbEW)c>Y;n*F#q*_-EJEjn9C(33RX(
zIBK#t@nr2HH^%NMw-vJb`K0%iPYfL(5n%7BfnzH(P2feg>oOib`pN-(73~fTy>It2
z<Xt+U9bu#QTW8{ORy?0<pJwuOs#5>w;EwUt&x?)Mv<Y#k%<)Eu@QXV*_|kXJ9*SeV
zoMYiw(^_l+!N*saG>6w!^Pd2xay+oL+w{EO&#U#YQMn@_CMs>7`@VSzSES!U;>sZ6
zb({ljY)lHv>*KvnwnGFBh)5Rc<9f2SK%-_y(0EDSEM0A)`d&hQ%}4%gjPzOQ0OssA
zR)dak>IjFX;{c044}u9s8inVbzLG+y`=aRNl!`qLO#Bqw31LdXU}b9r((-}`*!I`q
zO%wnX4<{1~pbgwK_>mG&t@ePjOeh|XQq<$ZSQRfbTV=oSjoOZA9<NXhCzy}xiZ46T
z`!^e2+hOcZ4_&;?+LT%YPkB01g5?1H^~Ct_H%1r){XDZw$xx4{C(;$e=N?dnNy!w^
z^5tL2i@Kl*p2h%9H?_q?u)-rTY9biL>8v4-QZ!!(2A6s;TBvc5RB~oF!}BVP_5xeM
z^GBpq9opKg!-zz93xljfIg8!%?H9rh*~rxxv!|v*SwUu7Y2?i@QX-bv(rbB4C~uej
za%6ZUASyf=GIL>KDHW>jFo^d{l3?5C+`U&!i%BNvAad^#JjIa`EI36+?O$^RTYMjp
zv6`2&a*l6GrZ(sKZ}~IOiG>lha7ocIGnBrV^3O7OZQ*1_kc*4PswitCyYPuNvR)Wh
zjn=LP{Xu?R0Pn{X-;dn|6#zi>MfS1SGnd-?t9cm_ShH>Szv2#<av>AE1xOU(0Q>_`
zQrq#<sy=mz`5HiYNe@JgU6&ez=Mb(8i(LO0sKL8r;rrNCXIaX`RU7vc5JL_JmsHVe
z+Y(F6Hp;}aMb+@J#6PK&RbV1ySbl_IAjMKzYQJ0a+m2^)-+cw1sZh-Ph_#0qET2Fb
zR^MonA(5q%B!DkMB*Ne?TvFpEXEp|?%C2!mS9f4X*L}|+Ek}ufZGS#q^7umqJO6@6
zXfUy(UWY-(fDCt<Wcyn#tvNH>jk_4823Co->Y)dpBw#H~-V^totI3lPH2{P6B$eiS
zk=l7L@Q~EJkKyz6BOr)Rl{m2b1DLp>2M3=6_I2+jWF&$OHIuW29qI0-xNq6&n^X+q
zh*+7PNckAmue?a9ZZc+LbK==67a1rkbIFqC)?fMF%$8zq`F(BRB!Y4{|Ll6nznhnY
zYOuH}mdYz66YnznWxpr~Tb+E{;{J0<5P$F%Q4{7`&Gyd3qoKd6G|kV*14bH{IBbCv
zsU4IS=UFiSJv2@szT}N$kO`M<fdRr$X3CZGpy{gm2lo`}1BP0@-JON)TEYtV83N=!
zJ%I|%XZskB`rTFw)z2WXD!nv_WJ#I?k8|B{Og*{$T<lOx0iM%E@ajmPJ+@PH)7ysN
zo|W>zcG<I8c}mYqEz^|^0p9G<z~P9)CcY`$jrjOdsZ$9zLU2LR>z*-mLnPr5y&qRD
zRy>xG`_qB|2et3dG6FsnezH7nkGCk!tm3G-8P?n<I*r64VerPu8g@F94?LJr`&MtX
zf8Gv7p`jIng{xc^*Q&8RKUs=BMYSbMuLR11S1+MJj|g~bQUAK<Cf004{CHPr8v_Op
zj;<v!jb?PQAe-EWmrt-?R>WpPIl90?pr&6vln*`?7_9{qABb^a&8}~|-=j{!*y1-l
zOQkM5&dpqaWs~duc&@d0H2D~#y*)|)mNEsNsXqk880YofXmFWIY0_!Zd4t31M17B~
z+QWv&JzIi%rt)0LN8L?%H-M^#qJR0P<>27;R?fQBiJp&oy(+FnPmt3T+cl;nw>g;N
zb#T+R`!&A^Yd~ZPwLSJ}{NSLo{Tsy?ZT+1U&zQ>Ku@W;A-=DMw;YjSgkr7m#eFWY2
zBqSo@?pQcP(2c51D91zy7@m#f*KA&}vb%zcIlnz~k9rGJpK?LTY#ODW3@3_9GNv8h
zqV!^z?CMQP^RL|z-QzRjquPT3;u)8V;o@eUlpzrpLd}FrXTmRM*PyVSgvPo-bk}o+
zqF~pz$7np3o<i(>rb^5fmIa5|tj|jgwSu&tJp<l9g(E$Pij*JXEtuhY?zWR><#g<l
z$54$m+q$h57GiUp%H*yB-j~`%hmLMLb|_g7_shWifOsU|nTs{Oynb-AC;Clh{kltQ
z$i5sk*e;Kw&zwK8#16S1m62wh9^e1VKE1NtI32Eq4Hb(w1bU#UFHD13yW|I^gx$Cm
z?1&q$F%G9qzcCIu^l-j~4rKA<L-Gb9D{qE)yb<hVxA7E#LO;;>!Gey!Z*JEJ4`Nv1
zz8Ezzyufw)B86QA3(4ZyrEL?PzHWebZ-Uy2q0o4yWkkzYKf7CU()Xi+$omh3-MQK9
zh+%8-+mh(o4U;6nPJ_A8z;rYt9!~ssX;;(MZ+@Q*az(a+*lo~nwh%c9ZbGjPaLq}b
z>(Kj?LqtW<=*ac_<mbr9MMK2Xs8=|rWUCTJ8jUYvN_I1zn)Q4+U}Cq2P4?>p)Czb?
zdV(UgMQy{dhG-uJOcI<{XbmX_!@zp7n|{m&xa<v{2|!m&4f^v@)P8bTu*L5aIo|Gj
zr@(P?wK3)RURCXD+fMp3Ic>zpG|gKmXofm4so53}F~!dMB=*P>qHiz4bP=#Hl}_e>
zk{kr$3unk_HDY%a=>76bgdOe=7YL+#z(UjN&zhMTIn8&(kgt%SIyt|{o;b6FM;I=&
zl&mwfXA!@^z<>F@Mtf~}*?0NNV({hOFz_V(A%~0DfG8l>bQ&1htxztST(>PbD{U!f
zrX=&suGBjy?Op;S>xBLuKL#8h2_Ml&Jq#&w0s_a-)$w4yM@{6lqV{%Q>Qw{UW21#9
zuoHzGPbr10m_+f@dYGd{?v2F-VOqSv@UtA42X{Bz;|?yj>hf_Q_~O8NV;bZ^q$ILI
zWVrN+!c;X+NYEyy*`B_1*_sqHB~PdbRc<A~td860_fXjK3wvNF8qTlkO!mpEGjlW~
z8ZTfBf<$8lxLy4Ga)`BAWg&ljrNehV5+fHtO*MMBdw(uL@<{7;=aS9{c4MLQjLVzt
zs0}B3I%t}!%Y;w402#v{8frkW>was^j7gr`c=p<(rFWJ?(^KzvOGhi+*Y&n51~YP)
zU70qln_2oQ1)^MPY)5IQLfTAQfJ7HwE4wcBgVh#2@G6<SU6P}-iYqgGfEJ|@7a#CL
zdi%?|mwCbbYQqiM(74ind$YwYXeq$kU$;Pfp*!5IYLjJtaVPBj<<ZC2)cGp<`1NP1
zNlS)Jz<KC!vHhg?0Y@EcuC;`CK@v|~y?UPYAHt4E+g)S={4ZC(5ydwsImY0r%xtCz
z#up5^Jt+s)=T>yM@0ZU}>L=bmBY&4=^eThjapP|D^HLq4Q-oyG*AF)&7&`<bq}nS~
z9ocw@*z6(hA3i7n$r<V8b6=J03;3-yzPj{1Pw9M@GinMs&UX4P)4PO|o}Vtb9i@E?
zcU)~KJssZNqAh=21=NNOXG!L@1~r5Cd9D(QlU;-vjqF+!T`P+LItH8v@^+^wKsdln
zz&$wIv3GZT-}FF^4G^q4Wy2r8pr90JV+^#zUE`XxO<38-)HQxl4p4M#O_W5?RXi2$
zI}+6&sUTFCrye$=K!IUEijYxY@xc1xTwQ?MUEN;p+9lf+<khI>0PfzaBYN53HB${~
zfs~PoW~8m4Q?}uQM)1tHMs<O9>pHql=k=Iv1`C11^SzRZ3j`e?jsd5n0R~ZKg`E=K
z3HPN29J5XDVXa7A507|caK(knLdm&2z0~pWKqE%u1<BEsnhB)jIl$Y`8QdE-J$E@`
zes?JAfVyNRXX?f_eoyx}<5%USqa7Rdrm(1|NTFU1D(pnV?dvhNlWV1e9drrH>08!&
zE1uZ3l9I^M=fDW`Nha>WyDQ)JPwm_7Xq*EQ`h^Qivwx&*inJ|kJbfVKS?^eB8|}br
zMKW5y(Mkv`>@Dmoaa!ZXN4K5)@``Oro}Bsl{ix>n^IiA>UzYtFn9Q@G?QN-`0+YL2
zBzuMyZDN2IJ|GAKGB9D|o;(ptiWG6-V@X%FK1WIOdh<&R18rq9_H__t#oFYdhO_p#
ziE2@yQ>hKRowt`21>D7&fLqc}-+&sgP+x|iwNZ1Tsl|Bu8Z+U)+M@tOZbc7V3Olv(
zg`}$1t55f2yLw|I&T0F^&^nP3Ol`-?TZA9?9g-r97Z-aC?CJ{KAtjnXu<->ad&NWA
z7^%m44yMhE`-@BNpD+HhQwXAr+iTsbb9-jD&oxmVmuW`0gLuj6eC$l(R2ztt&RLog
zFFkuw$F@;T?)3E1w-?U|3l(9~mc~BtG>Ckd5n5n6be~|P?r=sR$P>Vz1svdj3=#b6
z!jv~OB(?|+>blE<UhiH?uI2^{>TKH`9#grIK+d|cZ_~Pkq+G09Sh_RFxth!X>Y7n`
z(C3*?)}`_oRb11YslDbne+=g-;IPLz-LP5YF_a|XeUN#w<ye?Y1S~A0DKhIm$mlAP
z)#A3=?5bQ7#_K|v9j$S)3k)z{G`TG+bD{XsiM{Z*XT8=Q@Gy3{?C!XS?wm`G!s*T&
zvo1UH^*@}y<#j@o+0WO8mnsYDO(PGRex4v7(}K<-2#8#t4s}zAU<i6aR;iK#lx0-n
z{Ib_8`J+CLNXiwt8)_C58<r%TqNUE<eYPHPc1cS(TOttUB;5)3u%+>@MAec@ewz0H
zQIU=s9Y?L>%!4&uvMe_=MCr>{DK)D?|B<m=3i%iUyxOHbHmgiuM}CXZgRpApOT;kJ
zHXHTIh3&w-BvDj@FyKwE?V+!Qo#gv3hv4dGzM$Z#bN$^M#XqA0mHKpjBq)KOB0pZT
zr6MFVx$<<c_*{OY(+o1!*XD@ac=oy9kwG1!F;(ys*S^f(l*%oq`cnGgJG(dIkKi1_
zlK&|4D@@qW5_9c}z^cKYZ@=+|_>=v*c+qb?O$^CJoi%7V`nWiFW}obT2dRk8IkHsu
z%vZ|LONZe0NRh0L%cLyFsOw*+5)HV7P3BAmS)_qS#q$JUSP6{RR5R|RPm~WMDc3#`
z6o{RKo`(0T7K;Oyg-iTS3T1~ERtMzLJh)ctB>P5SZAv!R9|si;)J~RmJ=|~2qZk+n
zO6S-LTh;nuEVy;27UT1q%pAv6g~h|JE3u^<O<I8CkV3w&>dD#ZRZ*<>)BFjgDZ=)@
z+W!oKD05y?q?{APSr&eNapC49x~Y4Zi;mcK6IdRZ%nnkVtrmrS#)nsyNDE975Zt|e
z=eD}&=Z-JV$b!qHf7DYRIAmKL3pT4R-;8YVW`Hu_A?^e?9{^0*$tt3Fty||0;0=M_
zyKh1GYiSHkC}@k5CU}FaC{qs9ObzB}{&ABfD6{Lk=Gyj8N3W(g1nsZRR$?@4>R+k0
z6xO*RA(*6RCNDdUt3SW3|BMfeMhi`NghmSsmaV%Db*g#<|FOJw0$?HOkw(i*f5PE<
z|M}{jP1LQg9Blz^HG@`wcnbl9|B@nv5z10=G0d0$d6KwM&HInd62ePjvdT;c*Zs$M
zDWfS195P)XZ<A);5z_Ku$9>sdi0}yc(~ArtAbDc8@u}W!dq3apt`I3QVxM-tOEufm
z$q)G^@c<y9On|?N%It2#tSl*jj(H!`=C3B?0vH`6aN7f!8iT)16Ca2Qgz&@K8~?{g
zUJ(8x3fiF$l<5CZg#piu@mvl{4arMEH96}}^2C&#q@MpHPXK9F54V!7r?=O2-rSSd
z*aL8>$;jTr^U`UnL_d0D{Ez+s5ar;4tX?sRGu6vn&Pi@jfd0NS1IIF~eGGCwhd>km
zNBQyLziL)uQOC!zjQ5vP$V$Zakhf)fi`HwqFtzwct5_PXAiJ=?&*Hr&)1Su$`fR4t
z&xjjNUb+F7m7|gCwGReE_rA~KI)?niX&NA5Sx5<Z&oz3McXh!f3amNz^du|DnD<)!
zOhM(w|3|Cfz>hhdbsX3U<eo(LIq+urA*1s>TFOV%eWG|qG=FFErScxI;Ny$O36AYa
zq3rmI*N=9`TUYF{9)ElR%B;0O8#(;3TemMxJT<(}2CYw?;ai9Cqb*!B!+(db^rHdI
z9P%Y;nz19GG4tq<V#FtuR`qljj*F@T`OE&^871lNuHN3b?=kML+tGUP;JX+8Z_sHe
zO)n>#{ssUVFxM5BuGs<@b$2~DCre>~BIWoYu{eW}?nOVp{U1G|0aavwWNLi;q|eQ&
ze-Tvd7;_ZbCil(X*((46#oZ!}o5K6%r8nxVxSy|chdqIf%zIahDn)c*uSv|9JWU_N
z-m2c!2Xe9=U72kUBn$2dds|oC+9%0UjuF;bw@Jd+%aJxTlk^c(F+`#*T;bPIrp_x2
z*b`7ZZ7R^!AGGGzb=%@}q9&@=$+n{?Sadbe1bVB?-m=lojM@pql9qR074hD(OWbGk
zdNXYOg<ERFlTbj>$#t|K!Jc9>^6A-ZWN^gDxE~p#r6$lRC#hJi!zL5Cz53iX3DYfb
zP~|AS>amZZoQ|YGpvq~$^$?eImi9?n7fvNZ40~70d~_MTG{|;w1%^>6H)X4|F&!n^
ze!i)~<E>7CEb}KC_Msg7KsVMc7!ttM*b^ygH3x8H)@Iw9%mMNDIE}d5WPl#Ba>u*J
zMa^Dzm~MYe3{>zAVt5}0DgODtrhy3O0Icx27>@rYx+pMueIS`}{&4occIW~blaX=T
z{_B`9;BxiA56Sk^|MgU?bbD#jU!On&u$&+v+>rh^`EU%(Rg*@lf09)I>3S4(n3$^u
zq5t!<(c5Xwzk>dn2uQ5|!{g-#e<C9s_%x3NLmrg#`S0kpfZhgoQCf>5^ah4MeiR%o
zR^!LIKH=lWMPXXe+>jN>v8k(~hIst8v=W*i!C&+G^c_fi@2kppX&~myXXF4V&M0Wz
zqTL{MML^qA{~8Z$0k@IkXyDr<@#1vy<5BP9SqCF@13}8D2q~GP7UuIh8lS)J0*Dd>
zu4Z!rs#t;=$R>i~DBy#4Ty1Q<Z;PL}RMWN>{|};aOGp+MOIU;>UWfh34Cn|r<4$6|
z7xtoyfm>(*(Z8eg*A-=@_*&<^(@!-vJr7XKH9Pi}J#99AD-wxb1^=C7uTPgOd>OZW
zFzK%6e&GxLqFuZ;E3Qrje|2dq!)IgzDDKe;F$xNX?6tVYpr7dccBivtU3b>WY2Os~
zFR<`B;~L#}pw#}}AOs-qNz{mTyWKv9#4lhrK0AFDDna{CkivlHjx9LF_HKW^;$ERk
zor)cY94F!>W}+E5=({ofFSb!lTLo8N_E9!AU~jZ`<HE)RYWc`*<^3bJFdUfvIxyuN
zY82;>f<DHtG6YTz@vXYo0jAeiI%y)vUyPOd>Hh6m?u^BE%KW=AkSB_{kpN@E2PB!t
z4kp@s?2mh~{w3#FkG*-Gnk0y~y8BX_4hv;AQa*bD6M`=Wtlv14a946BBxUd|2Ltkj
zgciVnA^sK1M*s!wud9280W<jj<yR%Bvo6-waa40eYi|u4C7B0|d0)3UvD$Q2z0h@|
z=(L+?ZmE8@@j=n~%ke#J_9;<}Y9W*L&qQa97kjCFD(d&|VgLXT3xNPYC=n6}ePc!e
z0GJB+Av^;g{6>}*Y{JFC2Ko_fWN!Vh2ffh|%LCnIs0s2dF()-1silv~4o7}gvYp8Z
zHpDqz7!1B~VKkVp&#BKZ{`{3sYKyFr(zMMiHZ>KGT-UY4<<z&BJB5E$<cuiqRqZ$-
zckK6f^*UKgn<3p3pSp;*F2TEAX5%}?_*-oh{n+>3AxjwVxpBRW%VFIyJ)m9-)LA(1
zX*EWY+9Vg9%q4<Lply*2BQ25xdsMtmEb)Dsa^PWTDl^X_m!$FtwBnt(TGn$&pR5Y%
zv#Zver4x{nA8<(Un%Dt%OHkPt+IcF}?cy59NggHWtB$WCW$tHw`^G_JKXCFcOe&^O
zwczacRCY@9?-eSI59V@X)qPd?EC+Nuk1d<tNhs^8L)Q%{$tbKNeqExQrP%RIWybMf
z)BSTY(N<!67U{0t$SmN})B33){USE;IO3}7e5$d3ytZ-CT+ofMX6RxDTb=0C^!o$y
z6ikvTE3+~6j_;hP{>3#?j&#l_c?36;66hePJ6biY=y#6$ZjJz%br100WqXMpIEH9~
zjD6c200F(cSHC5p)pVJf<krK$)LHc{K|(Yj-e`^+LU*KqGwI}Cr)F<Nd$c@O{F2ZW
z*<iX^N~vbtV%ipcjA`pp&L}Kr%Ow+FA{@h|<288;vZu0qh?qrK<EW~81_wK-rv%fv
zeCX;^p@eH&$-jhy-fR0rIF`qezW8kQn<*UfJJWe3stbksGIeNju1ojPX*}zUzM<jZ
z!q6tOlj?gGPa1A0DpX~U6G)DCZ2sV)vshiVk`N`nD&2{Lhn-2)=701<&+F;NS3|lP
zr{vuD7@M>!M}!kj<Y}+Qtwu;|eZfiA=>vYibiK%-IzVC1{aT6w<b}e`ac6Ojyses@
zY=QJvI2W(Zdw(m-&k7Oayu7|YG&S^t-fgva+Y7SPInW-EF&!ioVl?`_vosvSc{E?%
z1d%anD<9&}%KArQCjKchFA}D*Ly9=5lQepTTEpu(jd?`~AK1)bR7OSZyB@@mru|ln
zJd-xvmDL*4rFv~QzuyA;WeZrQ<B^t#c_P2HZLu;dz0hCWLi6sGI%q<PKFoDmoXuw6
z`0#K-`@@sipgxqqAAD&RuHV#8<Kk#@Q5tJ~f;KyGt@NWi$6ijm%iVCbOBDTqfa{?|
z16e6Z^g+PJsN#(J)z{4}E_c2McD(*JHuNv^g3o@uExg^JIbt_jeyv`QCp+^__e|Rh
z2A*NU9q|%EnOs6Sx1`-O^pG>__oDPsczv57v+Kee`;rw+SFbNE4VK?T&XK-x8jjHq
zE{*JW|B;hbQ=rh;?n*Z{vh!_+_`LfMJC(>|+%()AaGjfMrB)v*uL`>Fpc@d@=nF2a
zZ(VZS7_ln<$0?1=Rr#M4F~R|pF>j)oOLvm4rCpS7JjUS8&HZgGHwF2@nqhkO`ONfs
zrI+y9Bf1;OG#n!rg@6~NrzBD&JGK5ZunP54I`MJ-3T?qs4`Gwe-*$TYjd}c$5*z3e
z6rzCGqoRSK8O^MtjVsYi&OS!v`I~tXy91GqZ|rka5}P+wuXc7Tta5c1_wWdfxRYG1
zo8+G+zn(i}k{eL<TVd>u3);Gt#^t3nvuNqO?etpE)WckVAoj9oV4$jFeX-K58bRdt
zz%PGSkKtyhF*Ja5Tk+HJoHr$h=1MG+DjQrW?j)iKgBkuuh6*-UYlVTC#tbF<Q;XeM
zoqtFRj{VsE59J{x?jLyN^UsdlZ360uWpCt7j{#RH5}~8FRZVrMchSvpiQ5vzPU+!P
ze+2lK*jhVH{P4kXl8<h1y)pNohK?WFmjp1-E)izH6PK~&m9Gh13>jI*`P16gzm-zY
z8+7gPhcsaO|KjTZPcgy3LfX=jwCel1EYR432`MTISM&-UH|j$I*cCD0nRT-5V|c+A
zmm2`EiVJ@0gK^Ta{s?^Z7WfAa%9STT^Q@>|lMD<(sh$TdU!w_e^ia^?p)i0I8v?+U
z#Q|0vuviJ+-*o!I6da2e?g+8G1Q@aG{_|ghV9dgx{LB#z&Iq<_;9ATBhBI;WwSY`8
zj^PVv>a*g2wEnM1K>oi4#{Uzy{_jKW{~g6kYHc~IXU|;J*FmWQvXY9A3&agR{6DLW
Br`P}h

literal 49770
zcmd43Wmpw$7dAY5v+0&nT2Q({KqNK-B8r4`gMf5N_W&v=-O?Z-C5?d8MwC(kDM{Ic
zbPCe(4xsmQKhOL9|K1-SIK#{p>soQHbFJZpijq7b9yJ~S0E7y6Wgh|n7W7{X00#s5
zF^Hpoa(1Djs4fRx!eB5s9FBp3fj}TIF)^{Qu&}YQad2>OadGkR@bK~R2?z)X2?>da
zh=_@aNk~XYNl7nVx<p1sMovynK|w)DNl8UTb@}q;D_5>iQ&ZE>(9qJ-($Ue;)6+9B
zFfcMQUcGviiHV7snfcnaYuB$|XJKJsWo2b!V`FD$=iuPr<mBYy;^OA!=HcPt<>lq$
z<KyS&7Z4B-6ciK^61s8YhOn@(h=_=&sOZg`H^s!nZr!>iE-o%1At5O#DJ3O^L?Wf7
zrElN9Eh8f%D=RA}CwJ%09eH{AyLay@C@9>!ckllF`-+N+4<0;FQc_Y@R#s6_QB_ra
z`0$~cnwq-0x`u{^rlzKrmX@}*wvLXDuCA`0o}RwGzJY;(p`oFXk&&^nv5ASvqeqWS
zO-;?r%*@TrA3uI<VPRouX=!C;Wo>P3V`F1$Yinm`XK!!s;Nals=;-9+^yJACXJ=;@
z7Z+DoS2s5|cXxLW4-ZdIPcJVoZ*OlOA0J;|Uq3%Te}8`v1OoyB0s{ksf`XntefsR#
zv**vB2L}g-goM0!@#5vnm!YAdVPRq6;o%Vx5s{IRQBhIR(a|w6F|o0+adC0+@$m@>
z35kh`uU@@MN=iyjPEJWlNli^nOG`^nPtVB6$jr>l%F25E`gL}8c1})CZf<T~Uf!EG
zZ}Ri=3knJf3k!>iii(Si-@biYQc_Y{T3S|CR$g9SQBhG@Sy@$8Rb5^E?%lhZnwr|$
z+Pb>B_wV1=*Vi{RG&D9gHZ?UhH#ehDsFs$N4<A0Xwzhu!`0>-HPi<{&?d|Oy9UYyW
zon2jBpFe;8^5x6duV1^nyL)<idV71nef!qe*Vo_QKQJ&bI5;>oH1z%Z_u=8;k&%(n
z(b2K7vGMWoiHV6HKYmP3PNLE1si~>y>FJr7nc3OdpFe-j&CSiv&o3-2EG{lCEiElC
zFR!eutgf!Et*x!EuWxK@Y;JD;`t@sTYioOZ`}gnPJ3BkOySsaPd;9zQ2L}g-hlfW;
zN5{v<CnqPTr>E0p2ELFD;5y#baRvao;In_Q$e)|$0Kf<+$V#hwq^_h{Aj#!vK%-o$
zo5_TqvWLq)1`?9bk#IVO413FbGE`MHQ!#zrmSCwEx%i3v0dMjvUI`X{7XGi6LaJoH
zi-%RB5A91)<>P3lFBEQ*^VqgYX`WYh=ISQW(qP<yj5tj9Q6n%$oIu8I7$yi^s;L0b
zg#hON`xmr4;Dp*b-e%a!jo)OCUt+@q9IjCM-^0~t$QvXIwik}4!1ENQ)1v5xi<cZ~
z5S0CRM_?~5*ii7H*<kLujs8r<m9f)H2x;rfM3gAv)fU-li-cI(QVa}xYRXgmzFhAv
zIZVf+<@KjVgA}MS(mXs|??T-Gbx$eNTX=x9S&%8_Mgl8%ihS?6X)FcwNKp3YNRJZE
zOk;u9Ck-4`VYAZt8%4=*k6K~m)=ioPwiXOt4b~`*bqia07-NX7V?OX>7f~~Q`*?ca
zygIYn58ul0DY@EV>DT92HP0rgl+!8;#N}}^X9w7)euRgAojV<8+><-*`-**pS~tD{
zFZkt04g_}4$>U{~U9D|YeZRwpOnxTIE2XafJ{8lhP$QEWVf^vk<1Jn04bDd5EjVg#
z3H#CU-<OO@Q9_l2YiPeh+qZizpD^^50g=*#SCpPOXlYQ++-8+=;c+!a0S>fPlFy<l
zz^AJuE<?7y4sJVI9~|60j1H}D?FK*l8JQPzaKdj@|MZqrD`C%_uMF^0(VJcM_S2O8
z5=>hIa}DtiCpjN=G33H)D@ydrP7OwyLzxA7*2qys&)?Pq5jq_|7=ARRYG8}&JrFOX
zSW{XYDQ;x)!(!xEnZDQ8QKK<8|Ge5Bm{XS24UA0M9g>&zG2xH;r6T}083^7#nGw`G
z*sn?55BBh`n}HOUiH4R9R0b0CMoY+0ng(h)WT>yCdG!3pq1?D=Ly^`z$^*A@pCti%
zZa&-Jo)c^!12v8C!235oEk7%Yde>f7?gl4@8@y_0DwYLae~=^=cr0MltZN9I0*b{d
zbQGA3T`{CM*^*zV4-8tKn@4^fy%!;KewdknCZN4?v`#d@y^>CMYF%vuBv00C{l4)P
zCXDB;SObQ^J@7JHUTe{>m)cZ3+eLgVcwh!YJn%nAUqjCnP@%3E(1@=@Cpwb@yUA}j
zGYp=&X{9zw$x45DS2}qGAYm`2S<ehXKw@e3LW>-J5?QlDY$TM`Hvs9C14EiPvb0jn
zB#(Vx?KH#CH_Z>#)Wl(KnonrY6%_+B^RU~qug}fhQMoDK=;7`Az6@FSNuzj<czD6(
z@T+qUCO4?iR`><sw8eznCj-Y#68Cr0G)Yigj+56{L`UI15%IZX=2}@36c+2mnSBl&
z2+`N(xJdP3GNAt1AagpO^q%kj^d~*@tK;|^y3dyKNkT;Z?l2Y~@U8ON<7OG*;b&mW
z6g}E;VgnG-ghb>>EO68OyJ?;`(E;zyn#x%3aXDMVCdPBb3?a9_C(M@5*Sm=VF~gN3
zEN~fZb9`HYvk{2!`T%cDj|39VmcnZ9&S|zzY~Q@_eG+32`I!W)Ez?0;d1$AI_-TBi
z)&W$?S&K!Z@HPeECWFCN&1-=c_ZN*$xDG|nOa==Z5BXhyCIc3V8O5lSY^?wsXI4Ay
z-nU7wjvBp-v0$$6rN|3n!y;%zgfKL8#8llV(;E|deoT_jl>+NFSKuU6@2JtLZsA`r
zMO93)|1;028II7wZiSRtW#olctZ*~7ifFC=4F3Hw3BlLRrP2AKhn^0RfFL2v=A5-k
zA4Oh`+c`dxX%?lTPTC>lbND4YqC*aP?ws&-mTm6dLoo8JL@^-W)JyStS7*_>>pA1+
z-#!XMUjD86UaDZlb&04q#rN+amuXIBw}(lyz~Xj>q(m#zBBw?(Ujz1aGWPofas`Ad
z*dBeg9tOOi5MQ>Nm6(&;nKu8JbUDozfT^?Dx>q&TE0qTG=;TMU0z*k}6B-Vd)?Nmv
zcbqHyz1`qe$Nq!>l24LEP++KN34rfW8=KV;(mjCJj@6pHr2#G4P4*cSs~Eu|0P`mi
z)f?dre3V;%MJXfLJx9nECmBd#p(PDWuN*%}FH{65@HxHgd@xLIuFSl=@%pos{{A40
z0;3@6Iy1=~-&eTCPpv+a;FvN5!;vH-Xe4aBkl1jp^zcOO{xvYt#YlRxZ<4$7oeMtO
zm)__$*=GY-CEqz3GextT;g|88C|gO*wT9Lt3mhOVafk=FsN5FHPk<~7&NWMdAFN$N
z^q86dnQH>h@%0TE$sgs6pv8N19hop<Ef#{{!}19;l`+$~1mcO&L^jl`6+Rod^I==0
z7E^Kn-}vT$NO-b}zU@v#G21PJ`ootduH{V=Z=)JDApLM#dF6Duo*O9DDvs6w)IJYQ
z15Kru?>^hR5>U3(x^i{Hwe`6Tj4$cW;W1-6HeE%v^<A~4$1VqA6px?W8`Zmn<Uvs#
z436Eq{OD?RG$}B7Il*fYvkc!j(ZoQly@4kGY}$H-W-sQ4lWUqCKR93^_%>$L8*4Qc
z@??+o2@sKedNMsgRlg=d&Y99^|27#dtxq4e9xfXp5tz9!g<vA=V-q{@$8*W>c~(H^
z60Wg;$Fc*dK?aP3nJbP8Rfn`ZEiEgokMa0&+nTw#UOH^UFK(k($CBf|m`vFT;Vg}H
zzH2R48e9F?2?^(TC2uFpw_fk+;mcL~s4`7MVEV%V`nd8W^E%MH<Y!R-BzL@wq7=3#
zncvLC?)h=7_pK4_U#V1`YU1Y?iOgtS?LDIWW2x;*I9q|(pt5z#P^x5riuafd6Kn-x
zc|0BIy&egd%0Koqd6vZ}3<!N0+qY#w?mogVc%TT!N!mJ^lSETwJ8(1?W73;4czLUR
zezG3$qOv~^eyza?{mD;7?@VTf<7<@2x8I3RIQ_`9)ra)}!^(6|<;xDQYaCaZsR@)O
z$;dyvaTpiU+I404^*cp2Ks_Xb6sJ-$?C=@lhhT=k-}A)Hm_Vls3k-Vm^k{`g%{^hx
zyQ(?F8}>k8$5p>dW>+vf5UA2?B)}bIsE!-`=xW-ouuj)X_~2*yowm5jE^IZ)djBPA
zW<3i__fI3CV-q6J8Y!@Y$SB^`Ch&i^KlsMOfrC8M&o8)hL@!`Vip|swS!AW8;$d~l
zI3atWd%_rx*o94gi~T^gPTkLh5)s?_#Ps%^+0535JOP-c`wrgj%U9d6J39l&j1daV
zObNdCFSKNvu|v7@cxK|`ywbzl=m)5w`_1`w$h3$V&M!3@lJ(B`qx&$nbV(aqcR6TX
z2CLS&TG{g)MA+OMwx{_-Ehl!^O{UsKY{2SIZmX%LA11y-@!ei5M6z*^5xYU5$i_e8
z$9PZar!Qo$SAYKTh*f}o>I`;qdd#P$T15zT-lc*(GGmCM^ce*0Apig6f{FySIz4@f
zet)A=j*2kSR-ZHZlF|)*pA>oF>#5eSl!Zz#N{Mej$U`in%3aZLINN}w{bq-}bGSSY
z)eUO;)h@fQghX|gJvKt(6?LBqp5K`FG*kNs{iM2SHgj8l$kJZy7M6vxMmGBsH-hfG
zyV-S1Wgk=kCKbR|1#p9aHk<?r8^*!|aq%))KsXYHlL^>jV4@J1Dr87lCV)@@-UYyo
z1L66&nScraX9LENFg|J60Q3ts07eelWdg4<0c+?N<RE~D5>^2eRe&2XasUqrU>eh~
z0W54_GJpqQGJ^0jLUMqN9QgqFe|%adX_)@aUlR;$K&R7QYoyYd2-R4?X)<{w$Nkv~
zX;=snR^~(pJK9y0_EYLkX7G4ABajK~5QPr%tLmN(ahP60h3~Fs#g+KYzn8$5xp@8x
z?5OGG_7#CL2jT-w+(=L#F@let(#A-ebacC>^ob4R)U)bCktRtj9zS#ABEV#XW1`wL
z5RQxXgE>AE6sT><ZEt#59nKRf)SGga_q0#kxVk^Gf%BaxedxD+(q{E^f2X2=H7K{K
z6q2E0JPRLOdXI>z2ioHC7=^!&l#7xD0Vlb~32FgH>Rw;`o@D{DnZR)EOW<Ap6GC|A
z;<iv^h(1OVKKjuc0T_t|_KYF{d@oKh2a6?kVu$meRh7M`$_>pk>(RR{uYx<lCAu@=
zzSA(aymUlr(`IfVKO;*3_XzzWglEj-Iry(hhPHED*pT0^yIejFm6t7Pkp(5XBU)b2
z-k<_gjj;G|&;|mnY~YC5n`68gCNOc}juMW|cZL)pv(EdUG#SF-#zFA1k27Yl2N){c
zS;u%hn;{)cgSU$8P?(tL4l%Zuyf&wUZXB(Yw0`uL5oM7P!g$8C$K!s_oJdiRw>bK~
zXcoyA4c2KIeSHlKX98TC6m+jL5K$G^Ld9E{89%G97r%p1k%KbJ#!gAVT8$d%JM_jD
ziV^&vA>q8edOB^$**#Lq#{<`XE$Lzq*?Zlz!0W{w?<eoQVyoq|&qs8!!M`#LGe$WM
zcFl;xD-?fsThlm7!$hlyPL%knPmfpJ4YXX(R#zY3r|7M$r<AC?E;CtUYNC(eGc?$#
z7<E`84Wv^x&&eqNWl0Xukb`nO;&+!}Rm<fETJ)QI`9E~*FnHL2_Mw)h^fnq~@Txla
zK=eqQ4=;1p?q1BnHPZL7@<AU1g8}=TNOnEX+l|tvx3rG);)5^N=e$*P8I&v4N0Sl-
zS*H&QqrKpcE_A75F(YrsLRd?OIsNq<k33bxD0_`~C*;33ZM-?Y*Y}GQCAE}V5udx%
zcraP2?Hm8}>>!Mts8CXFbOa^I7S)8c=>V+}c8Lge{>*Y_TR7Hx<;H`-B$>Rox0s5<
zv`&jBYqc87^j}XebMY*k7|wNbWU|jjhn93Uwp~rJN&Zaxq|Q$%dn)>?hSH7Vo9)ag
zpT(_fVSCoyg1b)^a!m$x-f|b77(MJlfV(#zx3}yM4Ym+QNE`KS?4vJFZ9GbIb#;t2
z?M=Lqe4n<m=qQDa7}ijrS%e*-RwVe8KlOP&;X1c}QMw_|H$uWd$C)tfr&<*$t)DVA
zEUI^C+ncRuQ8cubGmgxk-iy_u_$|8~7m7^{=*YzmdCk@KeIL@pJ=7(pi}Xf0Ba5()
z3I-qN6Y|cVzD`%?A;klxI!r*K_?0<Eps3sW1wQJ%Skyb9xClcBA6>dmC{nntRyt(e
z9w+w1zE1YjckPAl49+iU8c63C@;vi5l8b07>I&<=Turojzw=C9SW9%UY>qb47VqZ3
zMkLU9V?L?==J-k3Qt^;s>5kFEHfaB%kK3nqf?ZuzKJqVwzQ2=V>PAcdQAZ0~L=@|7
zY^Pkrh{lQHvw0N~DO}yuY{`71ks?ErpC)@GuLva(k2h|j9NQIft!5xCD%m9&emb*+
z`b`f{anz?9s0~K3C>3ERSQZ?p9d|7+z15_DetynnWX5@>2;=j2IDz)TfMq<CldDQU
z!twXl8p5CoND|E3alXjr+k@-cp5k~QAs*P04GfgP^ViO_l@(we2>6VP!aaLOML_pB
z&CjIBD@F6L(yMrbp&{;O$7rw$8z9UEU=_jM4Ql{{5|BmzM4+FFIo^bcW)WE@1690*
z@(jJ!rBaEK4?WNgDvc%DdT~YXyu?}0j06a(19*pbDOGOtK<0M-j=mD0lpjRR3Qn^Q
zE=9-o4Kn{OJaVt{((Y#btxiiAhXKvT43n%X1qwYi$G~M=)TZm1G5F98h{7zZ4}ZxM
z>1>2-Zh6>$-V2CLFFR3&e7-b{Z-KCE@&|E1k?a4wgoLT%Uo2$@@V}Ku6*c212NR{%
zA_udj&jaj>ZA34Qj}RNM7yyOYVLUt#x`W2NM-C#Sp+IYkiHVYZ0^cKMq{L*DR`Rdx
zNskpDz~RC}?<B=0uBu+fLn83N@K79d2Gu+v%<|na2_Ev1k5rbLt!<)Ca4KD}b2%gU
z*?~Jl-0kSs`aacjI0+QEXYk-``~1As(M&+5;Za3~hwO}4hau48CZLErBKKp^_!4SA
zqxo>cIIw@W>ZR=!>f)8g@~7@5Yyc^AB0P*>YmxW!X9ES3B#ioWM(Bsm*qStT_-LE{
z)Qxh36SuPv#0Uy1tI~10FX~q7Ehi6#TYoPw?Kl3#xBdrbr1FBt`wwJp9|d8^0D&=r
z-gk={<LUvdI4B^Eodu*u;6&%aK{&Nk1C=NG<<8ht7XJ%bd0UlI|C@hz;Q~w58+ofH
zXdWUkj-6?8^ooA$EarCT1JBaB5WSy9T13gD0TGdy^8NRmpn1~l^;I>}2bV(9aek7e
z{*8+7fge|W<L%;HF(zP{R(m9a*7Q@g1Oikjk@CmG&iR;H@+2rfu~$nvKN8b9Ro)3y
zo#woKf;<0r8$2MlSy_dF3b(-)akXwRG2`wpoObi5I=~HzmQ|J6PE3J-A#)`GJkxZs
z=5w4~INXwoKhpV+D3z#C5}ny`+|TAF`+18Wc<4sTdtU6o`wJ%Or)*Ic4&37<mjIqI
z$G&B3MV%_Z9mn~v{TC}I-%AEOgw%&LghQ3}e_e@m)W?D*&aVS%eY~vM0)cc^{4jV2
zQ-*JDVUa(U;``@vjcs*&<RA{DmT&OUR|G1x*g*Fn7U^*6_Ky>sZy(>DgV&B!7ay<x
zTrp04!3av2y;cQ+7P4Yqyxj1(eaSTb+lc(XRzRC+rIrFmVD5Pyu(wb8K9v_L3HpMW
z`IVClW!fF_DG3KIs?lpx-gKOitLuN?TS(>#HT^`6^uV5o$A6A7PF&%kE<SuCLQ%VF
z89{`3m0gK7W%M-Qdm3G+;J0S%4W3nr>xn$nDe8_YHH#UZ1*BaauRPZ`EmOJzE)?1Y
zW?3rnpcRO_l5O<fg$6q)mfHWa5hl%6MkTWohZk(zFLcS_uv8%IRPCtWCQ|B@G_fsH
z{6e3*sla!$HKJgAh7pzZWlY8!CF`#egpIpz=_?H-hdOa(&N$eP9Sz%iM~1(=m<<l^
zwf2S`F(2K5-6c&W2Ieu!C?F8aj+==A`VokdgXI4h?>)FB-0)zM4PX!aYv6GRG_lEl
z-+GF0#K?T>2g78PJ44WVkV}b<0!}a)4>12d*lqaU6-KpY5FW*JhOFO0DEsbR;1uIk
zu@Mql2?VcEnh?~!b`AUvqg<v^0UjbCP<_5OTwF9i==#6IY2$Cy{_m`mu$2McOdtm$
zS}~0|A*g=o5|R@2=6_$kjN_>CzbknGJpVSW`d_nV5`tNe|L4}#0e-Iko&cm`BgFr=
zj9d_oS2D9raQ`ikFrZ`&Yr$ksPX)M?v(Xq#Kf&paK~8tzvD*|&#PJ7&j=*bwpA;4V
zZ@)8;K(dI*-kW-=P1kwrUMk%S!Mg43*Qtnm43bk(eHZCYDi^4y{B8Cc$VPFT0r^-7
zd)}}c6)huKqS5#?mtvnQUQWEDdHgeE`k%*7zYUDH056<l@ReQzIqgN|xju_4+NtgR
zW*vd4xp7L@mg|JBx<i@G8Bni-*EaH4@jt*^xc?^Vy9&_9T&V~MLR#G~DtE~HnG}Hx
z?pvo*p}%nB&X|rbAFCuI_@d^nzm<pGjSw_n^|wv6&4BwnXrI4=;0tY70)PEk2`AX<
zBAROeyYq0=LJk)XX?<kR0uJBrj%d?axjJ$O#>J$3GgJJ+(C6{cqI8M{my9WMCDuQ2
z0-c+YI){@<pZ9gi_QTAsVDe=&Ec|QW*BXrp^WFhkl4;C7BDZ08!}Ju);g{DSf6$g8
z)XxSw&5~iq&5Od~%6f6pEY21bz4||+LzRF}v=!__<W3#AvaA68$oQ;!Yw{M9fZCf9
zt}`-Qz+`*P3%0Sdpe%I6D$NgyluAIo=bTpmlCfP9kp45?;z3bRj2=w~c_To{QY!kH
z_47Rp_g8?uB}tAhZPq3oAQpB%V$y4lNTE)jK)3Z~g#V-{C?fjU{gbMzV}Ke;{-NUP
zX^8(#bDf2JF}7LqvO8P9r`Lii=9L(tqgofuYE`VzL}aK{Cd&6)^BiZ(CjZqTx0`k-
z7HnH$Q{vc>dl!PX9dp!gsE1N#qQ9E!+3aP6I|1BeTj`DJ*9OYw#0AaEvz3-<-W^}3
zOae9}+gE7aXpUJ*Lx_n*s!r9EH@m)>xH^VveUy_s#D6mFeT{|ZG)n$f&dlT7fEhu6
zbdLZGk2OMY?mH{vgYgb@gQZh~z@pRPr|3qnNPn$&zDUAU|M@9y<yA=wUG5XR9=#0O
zYOS=)^K4ACj;fdPo;b>MMzxg#5BWsj;G53rB^2ez4y;ejPpDT)Smid3??Rj}-k;R6
z5IAW{A-GILqNqtfRGJOsw1so^3xfJOMi_Bqs22EOW`SVkz^vin<v-e*1ZVRD6?0@@
z?-e!edM5{EVRyq<%^N{U$Fqp84}2fW&r=QHpCzr=CcV(_B)F(7_QPbHC?)LvjHiTb
zzzfwFT^6vjo2V`66+W6qERgzccKbg#MG&(rSTU>W3Q{;t=?A57CUC9I|FHwy{paSD
z;sWS-4S{wx@Wm{?H~Al&q6lkHc&w_6J)Y3-X%<`t`A2>~K?eN0k4{H9On5+xyAi5D
ze@E~R+aK#sf)78aSY!eZ;<_0faA_b%F_W{ejU@7pZpGxMLJ<jkX9E>x1^+ev(LL;M
zn}EBjSlAlX`GQs@Azhz8i+y!KQ)YHJ{4z3fH2Y7)w7^5pVQIMGAqCR1biECRAx}(A
zoPcYbSZV6SG6!eVPZJvRfRzpM^gw=`38GU{59m=@6or;xl}TY>(19yCI8ZCr;BuSq
zrW6LxDLtx2{l=rAqs5?(!sgY7f6$!sB;sVIi3KoN95o^+I-?ux6Potd)zM|;NBS|p
z9N_EC@~13ii<lli{%D3LS7wSU%3^+3uI&bd9QU32Q}@S+Zr#FTRWk%eZ4TX<w&#?0
zyf`8*!j~EB2u7my?l}}jo;5SX?0)xuTL;)1H0XNvEd^aXq6pJJHa=zpb4+s~DfZ3u
zj6WS7&e1p(yG1nr=n`t`jn>2Wo7u~5+<yfB0xsGZkpBosru4BM4TMM0uD4xDhsFAK
zM|>=}j9K;u+`gx*`t9rdnuqKS<OTDQ4%pjXlW`t{<y=>r5i}+TXZLbL3#e2A)7m6U
zu0V-;#?GbZ0eS+jY|H2MphC?V^z0@>QMvHt62d6EnRaPJvLK!QQn3rHIYA04%L|Sz
zYqYt8*nR6O+J9uZh1fvoAh#x368l`NkK97wQJ1yq6L2#lD}$U^yr2*s=2v9F<JleY
zhw>0r29{&l(lALB7TFF`OvoTKP0P<f(7o1Zq<^^)@XIj`hifa*mzn+naL|H$#b9Ar
z3)gVaTCFVP;78FgIA10e3ML2}_n$Yl#kw#QPq;@aSI**xyGWt56N}-eXX|($SAE);
zMi(ad4Ik~0Y(<U||5f}CS(6S(X3ON`F$Q(3+6i=r1w)F~--RgaI)lcoGHBPreJq&S
zL+_oJa(|`yQaEag4ro(aH-AO^Fccm|FXbj&eJTO*RFoEv7O)PuS0&WGAI%e8-*w*K
z5cqQl)1k(vZ<((G6ACTudEAh$m46;4W^3VzWf-~~#v9luuaH6DdT9iei);J`Y5AKj
zX8i31_>#rQ(@cg{=Hfkqgaop<`ME*VV5|9v!J>UiUr|Q-=wuJ?-uFg_$@!^2=SN4t
zLk|D4Ob*3$9@@RdmzAcD!sbCO6lrzZkN&~b9N=r+-Yy?H=%Es8Avt%Y>vpoO(qj+@
zbB<c^&^XAtU73G_3U(Z{8i+f${BH<Bj@W=B)Az;jkT1_j%6EZqV<t@4IQB-=|K9t8
z>GAn|wtui!An+Vm5ix>d@IRm`hY=J7`~PbSg|Aeox2A-s>~U!GxKEas6Ae$I+viJJ
z)S#>*@@S*7u`TG2oRfijP5O!JSOI#(Np><&Xphl4g#5I*&+G~^HS{j<iD8170PkSI
zss69)<ATLC+}2<RT5?df#`=)oVAHXqCUmK<lyTztiRrk^o-E|Cl22YSt@V#L<ASYZ
zA5$t}SxPWleJbb+X9Eg04HsR2G(Q)}K5`m>o!)`#A1F{}gG;w>{bwHAhuoR1u<3<s
zzq}UjKoeay6u$b!$D(~&u?rVHx0HULfUGr)WvBMWAE$f`mYQpQPYpQeYsmSXtEUP%
z+0>CZl*EbzP9CS465#)>rLEv9RkZx7${#C?ML5P0<q8oYcNG@&RJ_<At2#1~WIwuo
z#tA}`$cweH0fnQ+!5{v3&1)c;OK-)et3W0Ds;m2=CS+H8V^B`=@n^r93@O6#u<zNU
z<E~Pdxbov3X5!(jKkk_o7d={axeu37G=qn)9uGq2VKlhr@D3L@zV4j6&U-(-*&7e#
zOdiix3PH?j(yaMEQlN#6d-;_yagj6LZgc^T3ta>uL`5Eli=ub0w;_<5oWH)U{)RO+
zesXz;{<_X{e30Ba(l7ImQj?;_s<eo6MZgMG`Clqdi;p28?ir2HWj3cm`OD2pE<aiN
z+<hGd&XXbQQ=J0pB$=(e-8Vg>Go7Lrz)ua~Xd+iS?uiSmbs0a&&{4L8MtEiDhFLZW
z*?X+PW>Df%^S1K!G1sEz$@?#e4pI%U6sB3h?4-7Ru@Wg^tB#n&mz>IW5Mxh>hdlh6
zdd>A7C)nYqkR~ZKwssjd>;s{#B0Q9|t7Go{B*1H|fFPy(K3%HUdO3ED|HG-@C9V_m
z)ClAuCC)@Z$%?N}E5@ypGUXm^dw4Csnq5BASKDf@D%~`j-!?X9lnj{iInLx>E-Q*p
zEOLc!Y>EUne1_*j@lX&F8-@yjGtXPA@Vp14JA*L~9CnI`<0cJLzY9XaU3%N>_vZpy
zdx$%;D}J53be(#&b8+dx5dDiYQpv-_MWu+IA6S#15|swtWk5{iYq+C2U2O%hqKeJX
zXBbP#t0$_fp>i^1|CJoIyO>R=v7O@e)VN*k%A!TFI5{Tj&JKJpt&Md&egQvIov4Fp
zZZ(@3p%Pv>ilwgx5EWriHDQ`XiuD#R#AiGFw8$5ZJ9SB(girV7X<)I!^r>&{7|2ja
zWC=HvCKUPgcbTdJwr0E_-L?A^6ohE`WR`1cTcl2OjNno#w6++yqcL6Wii~mnKufz>
zV0n-~i-fv;<igc5->GZh;F3c#pZd3lzYUKQw2GVV-G-G#Hv;sGp3G$9u+f-Tb7@9G
zlQRHJeIDYM6Ujl0W%PXCeemh&idta!a4wPIwXN{8%sqZgt%)W-ZHgJ(9UQA_<Df1*
z(P_3S^)jiucD+Vhzt8gJt6N3j9A2b8kEM`^sF|~jbGQsoP50t#M*!0w;<QL3f6CQt
zzd{tqpnR~*I+u8eJ=tr?!$&+LP!LC9Q9ZPyo8N%|iElohFB4;%qu4C5f4(ENiEnuG
z$>Lkn3^5M@r>Q3xYL}$gYUY=AJtboL1|-PvcSf3tYH`rBQ7cSp6oje_{SJqk5{>%Q
z6Y~_(g<S`GBRA<_$)DP5=&E}TT0J1l?yPCk<8-9g4r^%Pujh;?48kA3eezfdYIfVZ
zj$LQok=+`j;i{tjv-e%Q#m~!^)3*XFC#NiaJ$aHK+4Y8=NB!XYCJA!$C`_s<kE2V4
zCU5OTPNG1ci5z%@acde|kpS^Z+~&@QKL58;VL?E#SkZMJV0j750ECjimY4HB`4R!$
zEea~cN6)0X$Q(9jT6C{d?4+M8Z?J#Mn{3`YN)!|x2sp;vC_ccnv!V!Rcby%7^1_~?
zqSyC155K!6>?<aHo#V^7m^r3keY~EfZWAK?X(-95gpwTXXO=%!-`e7%Ga4DMyKC)B
zBnl-ve`ysJ2g}cR+IQ8LxcKPNbP*#?9!>)gq1?s4@V30X)bUG1D-~>c39TD1bIV&f
z@>`u4CE?1Z?7jrmjpQwgnDt4<Yld5LgUgw}=&NZFjvlDM=?acyh{BzGuhsaWYi9E*
zA5D@Jg2yw~%~}T=Rr<i<VDIa%!<8eF+V^J+{J4xU>p)9udAY+duRBed5gl!PN={gs
zXrai+wf9P-5AaIoKf_~$pQyio*`&4l<a4VukWu(~PwoC-#!^jF$q(zFn~=prbQo;Z
z;$#I2xJoKvL{fdaLy2_#H>>Ja9CS+mCayP7>*MYc7Y<rhy64L6;H<jFMczT)VOk>i
zp4-dx@+E1tJlcNAEMY}G<R4ZTYFR0Z|B|~ei59kZ{9rTf>-N7RzYXd?Y(6TO9=D)G
zY+Oouv>+gsBK{>CEwD$3Wzlf=Ku1EecHe{DIjWjC3qstp++e6sd><FcmYi6n)X>i#
z)G~7B{vfJ3gn1bvkc~e^yBbRayW9HqK319ju*sM@d}RF}?zm}k6a%t(Xb}Gel6$d$
zm+ENb%TstTUgpx7G;0CchuAzkK=6RH988}#jWX=<!zSL&O)6A#_F(H1tb#YwmYK9s
zHNZ)K+FBJUQaIOFGeo}CwHPQocft55Qdjg6+)<WJogDQx(2oUXrozL0oBaqXb=R05
zp%j$0rBWFnMz;ae(NyEZ)2r1;X)B@;mLH4SF!g=NPiRqr%uA?2Omv7C9~rW<yvNPo
zNx!JksCrmb(C3#m55v!gj(7OZpmGEy__(mb4+nf=@5gei-aux7h-E6_0OTLCfMmpY
zu>s%YmHZX4X%Kub8@g-$1W;^z^mg(dCE*dU^yxQ?-p59yDiDLfG-d%!i$c7nt8lGm
z(@j%4QJ!%;wD*P_B^OoCvm9DZFh;&};by13N|Cg21~x}iPsABV1q#5%M`c--ns*|v
z6Q_^dQ+RkOi$=+j=TFj54&MvTQx(&9hxOKfj&tB#JEPOckx2kipL82*GsePH%Hg}o
znzhZIlC6Ke6Q=mON)RcQrr@SI9{y*V2AMn>Oui%^v(i5h0$S4Uvn?YSUk1;$9*uvV
zTVdP^gTInn5hkU(2)$6xByLh_24-(a-_6;<m-XkUQiTN+=~T=HDmwLc{t`=BaJCam
zWT@zMr29SjeqPSCb8rYM09t<j_Ic&&7H+JA+ZxKp)`@>tzy`+n79GfYW7cB0q>GAJ
zRX?NRJs;kf1ocx(vp2Uh`uGXELHfweGM#3_r%i3<G0M=%ogW@_wlw=+Wm=aG0FS3N
ztG9P`?m}4=#QHS?O7@J6!I7&}6vHM<(*wd+CmySa74aUUi2kN9N&vQ{zCM26O$4ea
zK;S3MeT8!(kkLo7TU_=zUaP-S)BO=~9wsq`Ng7M1UtKR{<h7uZ;u*GX$3;u`DUZaH
zV1lo0Zm%i)=T$%L<mA%k!{(}X^or(^`Ez+#Hn$<J7TLvn!MK@GdD~iU4zrkO(j@Wc
zCuZyr5P@*_)1LP3r(C9|8SK|GzVP!Xvo_2J4ZH?8K&S}82TRQfI*^kMiYR+*MlNCe
zpI0hzE5pK1^eezGH+Xs&-Cn&UyvU+V;G#0vkJ)H|J^7swP4hm}guZhg5xM8n5NGU#
z?Lny{$3m0X-p{-Dmt5w9jq9d~rz3!mF>76BgQJh`Tm6;B4i}xmgWY5IZzs<e854iA
zqi<1#(f^Ty1i~+kV0{LYL|)fzRk8WqE5tby`C)!$Juqkg;EumZk49qp)hqX1LPH0(
zZ~TKJMPd5-_xpT+*R6pmJpO9pS8||k2#5_{fVu)-u_$-t*6DO^tE~>&39<Wz=*lt7
z_vJus1LvIs=oA#Ih^mOAT1*uoWj$b}3iLsi=Rvj?OCg(S9Jw~u^EGm4u>=tIdAzA)
z>H4SNU{)I`A_=$L?hRfMraxa}TREuI+-TSOuI5D0mi?ypbZXjPMrgs<F&H?|c*k)5
z;7^ukHBuM)Mnr*n1I%!aMV{-}1+zRL3tIuSACCC_i-FEz!9&Bqukow@&Wx;I0|~GO
z_5XbkDk8D|Uk^Tn`IXvA{VOGVNQT)<n%ReSvGxn<yWj;J2V?!-E<{=VLqgwB0cs=|
zD}EuMOzSU`u0TkRvI1-Gk^P%Yh8~oH6BG-<{yoVI!f^{5k1xM3^5S+EZsu%cjnVL?
zls6%Q6>4`mD~j>KXmynJ;Q5;VS_;(Xn|_6l+~OM~#yf5@xDz<^hZ<pn(HbaVwPUy_
zygnR$ajr~|e+*rIA?AHD?$o%TR8R3IQDTJ34xnP56VN(;ZNi{K=@~qVkqr=ptuVI!
zCX?Y=lOY}}`s;Xc(Y(phRx}`&Sh8P>BJ||B_te+>HpWLMo_5EuT<Z_xKy(RZaV@m(
zE?DQnGxDD&X|rSqSEUR}3SErHm;P*s5w^mjHrMm(*{G*bu=v-DQ6Ue~%^>9_eTYwI
zP7XY1`fm8wZ`=hKs@Cp~0z$7{ggxVv@XnRweNX#(<m@8#F}75uOxWtFz%Q`zuR4VR
zN^=}CrsM#3@GSA0Nr701kHNu179E>g`SeK3s44syo;AmUDYDeBX#B|pWe|?07>k9r
z1~k_4(Jq6!JEvz<fC@1BA$~%Bc1ibmq*4CkW4uJTi{FzL1@HJjk|cro$OUsL62Wk+
zv_xV~+LRzZ1agL}!7qZG6Yy3~8a}q<tnXbJX^mNEcjNryEv~{n#Ql7BYK=j!E=^y~
zTkDYXm3;@V^_?m%|K!2v6d?@=`7UN-#%;giw`r%Sj}JZ915<xk+)){~WWIt60r)+n
zJZXPYD+V9ZrVe!=@kc{qABFPib*{<&vvh`ZyLXY8GBPPaUaR3TFXjau2X*b$@Jb3~
z>x2Y1HxnagHeI-mvtsliO_5E9i#8=}EA}j;m6?^=n?HzA09qHKji_8$nA?cf<%CBU
z&E(I$L3t_A-SEDuUV!go52sKu<p7@6*RPrFoSm4ExXjo=o8^uU^UrPDi{wdFJNY~1
z9~bYywlnI4HAYy<jfB6_+<rIq#x@ui-R4PYaRUOPXT1dp@;u|Or(!~RsOd*i=i~4*
zI3z9jlX(+eA0Bvk4D9^c-oe@k#a$oqb<K6v))A765PQq3Udi+23ibv|`%0}3L&fvk
z5z-DX!Y5kzZyiqDqH@X=Q*A02b>1vapGcAA#K@*>-fww5kSZ7I8s;!oYU-k0ls7dI
z%xySeu_;aCY(f0yI+_ngZe)^mc#TTn3{T6^*QNGb@;G<Q31l=YoVB{)-Gq7_xFT(N
zbPA`KJtf41iy5fVi$7IqfTfQ@VfqIK=5_i*Ny&9vc@Pm_G*}WYzydUU=D)?0m^tK_
zX4*^K*4EU;e=FFRl&w%HE`@B4cX0xu;jd;{wbS2k)~3o3wpr|x%dTf}yKmRWe>Kb5
z?889r%q7V({@vwxtcV&=7R}8lywiZW{pfbja2xD5#gA&2uTmZ!!B|B*n#uS-;4w@U
z4VW$t0$-W<KXjbhb%+k;`LBOfSeVd~d+yRj|9T~FY*elwPq1@rbj+V&Z}VE@P$$7f
zSh<CV?pj!&eEFXKNBV83UcAIH>Y}aU`30h_n6zMKY%>J#R}vt-#D4cN=fZOl=h&@a
z_IkXOY>m%((Zo%9yyBOKj|9J0T!5|>R<LOK?!<FHzvn}fBBJ7VZfd&*Pbr_NSpwYS
z71t8YRiW*iD>_2SYgBhIxQ7h0tX!Dt?hlX0aM3E&s0^aYO|eK0aC`nUoK}5}?GJHe
z?LP6E<V&wEmK@#*Jx{2uh@I&^%;$D6HKJE13Bo?>gssw(na7LMIUODoJCtT#eW~V+
z^Y~goW$dDAgn2e_);l)*0HDJ6DtfK!y`<~!uB;phqNA|rIx*byk&tVR&IL+JSq)M?
zMt_WmF4PeUIx$ZrHqPcqj9^x!^S;Jm>fNU=FeY^Ci!<%#I~N9#0_^mEOX|V_bz5A=
z#m6Pa15>zt7e3HEk%mB{4#aw_`_eH{SB0n-vV5y6B>Wx@ALb^&me9>rA5yE*DSwlh
z;bgmTG^=s_=;c2np2)NC-u(sJz_mgDB)fD;a4qThlY7^hiOsSDtMN5`@ozgbJzczb
z4`deOqG%N6#NRV!o4U*uSt#`f<c*m0v9*7P=y4_}?O0&}c{{Hqui{qU_Gy?{Zlk0v
zV}S@u^5_DfYP7$ozG&czFV2LTFnhye{Obw|PwmAEY;Y~#L!JKY?ONjhjL|9-I|tu$
zkH9U>_<fRQe{%V(t~-K<^#0jc9arM<tEb{-K=o+Kq;2zh6KfNRp#^7C%^N4TSJ(Af
zasJuX%kM>xm0%${LN$qlYDQ3S4DrFt98Ay`5~4PT^UXgH7-s~f^)0Y3fJeEJNUsrI
z%q7_Lo1H8Nos@G|rvtEX;b(403RoKYoutBM{n9b?|9H2}2k3Q8IiQNs<vSh2N%BSE
z`93hJmQt`$0!+01l(B?gz_oN!SlT)!pTj)=vh&#~kppVLD9xmJrBlDN`x=ARbq*?&
zSEwT)fT?oj7b*CT6ID<+djWIb0V}tE!hujnA{fN!+=1g@$;F?a^xqn{H3$d+X@zlT
ztIY)T?*X4Er)c3&=A@5M@v)>qKK}DKcpa)_Y|cdb*V-n2*tlADhH|}e(H+d$Si)G0
zJqCrh&lEfm&PIm()})p0w^(_5b+yr5?e;<;v8VrU%KQodPjK#H8AR~fqwU!cl!`wn
zY-UP9a9)Q8A3`0RnhQG$TE6?#t2cd?2Btc<F}o`RaMg8y^#I;^5=cAD7(ppk4w<E=
zlN5+vUp~)o2WQ~`f_Ilt8Jy~^k9m`CiCK;2G~rkFsXwjC*yTR`P<#ow*|sfw=aIDx
z$@v&Lyr9HPe87(ryxc@3VCH~QTZAYIINyF4ob88V#cyV*FQG5K++*&_&x;Ocp4;U8
zLJb>VGb$ld5k4Q`8ZYQu<*6+wH`4~IMlrvhw6q>XW_n$$gdCNTH<ht9Cg-m8E`<3S
zzwlF^PmO#whz&91$31oM#fZ*o%PB=fCFz1RL}a}O?76MzbZVN63`8cVLdCvzc>H=c
zSa%M@P|ycY3JKCc-2`0khVnr5DV^SPEaH>sT~ojY5ry}a*Qz>cBS2G60E5yw0y=x#
zOH{~B;Y)=Zn+3Ve6}IDn1aBtw(k>1*489koD2QY1+2C(Ub$j}CJRZa3MqMPuFK%dW
zRiNwpj7VhD+Wc>}nvDHBAUq%O0^j(1&cfV-z($w1%@&TxP)iMe>tjoj^F!vtL*om4
z6FtS>#8qIN$R>Afh10kxTpO$PiMwBZ2k<LX+uc&T%u_s099xG4jYeRF4>1{uEOg2H
z3#0xunT~RC6}LP;{yjW$6mGILIkM;ROZuGZ*XaW4BXMTtD$DCJx<NPK`ERIDP9YSU
zlF#p!K%HD1GNW*hL;@B&-S6;?%?<pwWYqmu)zy_PmKQjm9N9Tzrry+hqRe7IPD+=;
zMvcXp{3%KYzf(pZzw!?7f{_5?ZO7Cs`-wo2?@&D>{;C~JTyBn4?f~Y2IPmiF2qqla
zaHqI15^zxh_^KNtU^1EmGy55u#6N{+^sK`Cdr10ctMzz&?<s9&kiA#m<5w@ykD+59
z`s58RqtJ`}e%n(m!~uF(?n&*Uh%hI$*zcZ0;kYJTpB>!e6qFtj!>iF^b2II+l8NM)
z%@1)KBA%3w!vQypDj#G~8q7}FM!EX=5Ao06p)^mX4<e9_M(%pRxXR?rRF2Eri(XwE
zECIasYSk~+=JDf2Bma^&PO7i~*JNR}n5+9Po;^39EiXMllLFz@qe1rQ{!tomR|*Gm
z1YK}Z@oF^{<JeIZ&IY;L{n2lVk}t+{$%vHcDamu`$3B>nI@MReQcg+e>#cW6JW*rI
zRqG%sd8Tt6OtFcZdT(U2JyFf-#C^VDTzq5Y1c7S<dhRe4k@xrYr<XOl&-}j%pt~Ai
zzWIXvSlujb?e*Rq)}Z>7s-tE-YR&WgWTh8nDq8ndbx_M9>+4BD7OP)8C7<x2sCQQj
znCwe1aWY_@0G6r=mYP?+;xAkuxFa>{JIr*?Sw8AUVs&$`^z-ga89ljeY7yZy$2H8C
zTd)wwuBO;PJsYz68`pCPk#^?ubIf)fR2TSoo=kp@i(>RBdRAf;yY1=}+t)Daf<w>K
z($xEqcBF`;ygjPlzuA5DY!RA3u$9-Nt&c3#z>@GvX`V$<`k9bA41k&v1HGh%UPdst
zkZ{z%?42s$OjfgcI)&K10xKp(*6D!6cs)UNP^pYr2j(MDpeO4{2W+gqR^NRie6)N9
zEry`Z!KrT)E~;J+A^dzhvZsO*T%0xVj?*!RHDDR0%0LZ~(S<<CTnv_LqwA`_<?X-<
zkr8W|a~NR(8~2Hb`C%nH@~{zF1i9{Z(c%?=PSKP2L*6{Z5%Q=9ubJxJQ{fIlpZjuk
z%nkK<N^UVMz=Qpki(Ox|pWWYqR&v;#*$sxculYq80-g#s)iuJ=od-Mqh1M8}lyYC_
z5F0ntXX!8Gp$_=>o+w05SYtdz2oXglAH;E28bHZqNEoK$`^?Cu2?i4SA>XDlzVwv=
z>&W73Nh6h&*wOL+qqzgOaL^v<L|}se-j7+jv+yE|xq+1=9Pt#`WB2uKedso-u5kvC
zOkqcnioO2dGh($%k45iou@-9L`ZAx4<}i>xnlU+y)BR}y9GyANeBfQI+<uS*EUwFV
zQaaf7^EsgB0@Lz{q{zgusu<!%Qi=s!_PI`eAAg@b{ge=DW>c1D*E1)9@8}6dl4~3w
zYv-e6Op?2I=if1mo?BCgHc%RS6ALy=Xd@Qa7tgCLb@Sm36jjR^4s#?!Et_FIz*70z
zbH(0y66*|-lmn71v0?Dbq^I4Trz^iYYM3<fGWSMSeyEz5{UY<3XYPmkxS3U9M~eMh
z*)W#=PZC-q2S!lUlJp8b8nK!PdA67wIgO;t*<RYLU>biO5Av%M7h@50^z419so$!?
zX*{^nzGsY%&q5hth!4Wis)u>k&Tmnxa>T%paWT|o)nCcFd#7lQy_8-`^><c9@h1+L
zZC~}#2^|*EB6+6XaOW)q6sXyu{`FAv<9h7FWu*d5r|s)0oHS?o4>M@Ac2!mm$2YIo
zu6YMXBbV@=+Rq99K1D!2gDcbJQ@>9;z}-p>JDMYEq1g^nGa93XU)6Y3Gstl!K{jxV
zm-hvZJ^5QQq=9*x#XDt(vvl$aDe7B%;kVF*UrFq<ixrWmx2JZ!1~A)k?|aZ^9;Ntq
zCtHFJqbRUf$U<(wUlFBeq$D(NKh>d7^D&CLVK$^W6Jq%1ql6)My|ZEu^zFx7wcS`)
z6cX$Xx_iTrqr4d?JnInV#6x{^MZbiX$<v9;J{>XmHLtv0*D$`9PNifh^Z>VDURzq(
zgfH``Q-O1(pS=SLbU>wFW`CCXGsM{7Qb4TWK#)I~CiJoN(l={Y`>g77qgMiI{2$m2
zXp_>rYS}q5lf|x8H<qmOfJ<pmr795a=s{PTc~(WUu;j@7+<uL$@8+X#gL?A*dyDEL
z&%in$U6>F{<i(bvF-tKi$RKX;?icM|=<aEBdii|bPM7X)a_O-9gZG6-WBUhFiV4q9
znKe9t{^|VRUI5YB_w^rozN<~G;*#WiQG7#X+uij8o<DciQ_c#se-D6~sd-#C@ag0d
zb7MkY33FqlbaRmC?!C>Gk$43qsxN=Em7^1O_3aK<`|g$90h;FE9+XEP4{HjIF8LWS
z<VofYp5)PA9o|&Gb8v<8+|48aJoa}9e{a5>nOWm;VGGsFBgh#_zeZ7B^WW{;W%uc5
zq4ry2^Uw58quPE4F27SFRAk1Sq;^lnMRR@S!vpiaaSn3C1l)jIx&`YYLkA9um{;My
z=XF=fmIx2x^{cCQ-Jz;L4l_8-%=_5=p~t?5=l=ft#pKh`mnB_QO_=X?9V-7nCe4z0
za@XZ`$#L`N$#&+<=oo^OgsSN(6N@B($4|K6Fw?IvrwZ3oXfQ!Ra4pQ0IH0jfh}(Y^
zQ}EMa^S(NRL+9P!(o@t{rG7J<nnu^NAkTgWIQ;(Djd0s>%Vle8a~r}|u1lPHoQKiv
zw#04!x3nFdU3>U*8cUAX)q5u1q`!Xf#!}$XCH^K;^Gm3oiTDX3L=T_J%Cajodnzm!
zYB1qLXiu5ssr@&9jryi)9*-lFn}ffx2L<m}jc!`E)*pKh;=d(0J9!THu#8?i;^3h6
zu-1Fg;k1^MnvL2$JiB>}KFsue!01`613GRHs<-?`7=!L~KC<;s-C3@cM{6M+P2~s{
zNpq941@sMGAsn5U7h|vdTr3J%p=8Kjd9~;E`<|KcV5speSB4RkeQR(#<`Si_o?^vg
z>*r!dmc`F;YPwu}&OX{`AtYqd+nIL>Cv>1CN3W{ZC-k1w*Bv4}FonL3n3TeA8gNf;
z77wM*T;4wK5^P^g*i2hS!`m}!KDIS|<iDjvm$0Uf{Y-;krTMK4Mb{Aj6ik{>rWLOH
zvAu?nA*O0C%rnMra}oa`&7c-y=tao#D4WlUDh*VL^P?Oh#4Im*X2ljl8AcVbb*Esk
z@~Jz8z`*lcI?eg4@%N#`fIsr%F0KQ;VNHa*o$r>>*jWD<YlHU*s~cxjEU@OG`56hg
zmtZX7qq|*IG+>rWcK5=*-8gLd`axz1vnnEToi;tW_KSFn!_7|odDqB=e1Z)^0(^&@
zFV)pV9ymz72Hz(SJv{nmSx0Ivl{6-#Zx(6Z${UDw{=C}cl>SL4e2y0CZ$!t2ewJ%x
zPH|2>Vfm^vM)ZYwY2%fc{Xr>7wS<gsAUR>Hpnn;^pv`OcinXdHL3h!jYO2CrkvJJv
zPz3s*W;JOlZ7DO#v{wwaf__C;6Cvnc7{sYTv^v|@-CBwD3~CT09nQJGyJpbp?dFdv
zn@>IY@}tpXc6xC_&t-k6&<Xk3u=SF&-7$_9k92ygv{bQS_Fy+@dc5VcdNa;qae@rU
zNwx`sOvwT&S8^RrZ0fwdo?|_|VJY9-xaAZ-ou9>MD-M10p^*U-tsv6>eFdajLOYMz
zSW{Zj{4{7LyqXhQA$Z(Y*1ADpOx&60tiU-@;_(x>cSyNmnZWTV91u>K+&OjMa(+GW
z%xFb~%cmgy?cl@TZw}(W59(bgGBwnn`??3moDHn634T*knpjinufE1QNZ8i>KcA`i
zxvKn<hAD&)+0*LOXwWKz+5cE1odL+OU_SRM7B2NEvVx*N8Q&{bmycrV=5(purhnS@
zdg9HX48eN-s>VF7ggHZRsmkwLn$Mxq2Q&1kMosjKj~I;ks%14bi8VD3k7a+$lUG#&
zh29snN?cd;6L^^g-O+K^L6&OTo!zVV%zfpW+j(e2Xb_GeM6ZS9MxI6ObUC@udWOmb
zoaL#87`)%sVpNwHw(*w-lr5{JirVo+jCsw8Dt1n2&Y%loV`QL&A294f0S>3V&oMIQ
zoI-#6??$^_9qniyERAGYY0tDUBnsccf+{Gb$<l9P^LaL-3sboU3%5qcME2u!A|P;7
zuR;f8<st?HahzUT48#&ySBs=;5$`M4y!%f<^(+C-VFY6^Xw;H%gHAo3vH1jg<w)&_
zyrF)#E(o4~aP|c5c$=;^6<9IAmW$jSGyr7Hcn<SyW2YoQBYgF2s)&^zj5#+JWi;1L
z%{X(H-JGn5zSIQ)P0wGBGb_0`ztEN7vH`uyPELmZs}KtU_L3=N6zaSk7$~)GELKZ`
z7Ot!w4@i*bp9D$3M4O19KQrQ162#0&_VA8ksxZe3Kf6&gFS0$9_3gX!j|=&$&`Cil
z?2G-pFN&unxKVTnM{#9rq~R4zN~{((Kj)T^t6T4=#|oYG+K5xPKMua^ynKehkw9s$
zTlgB&3#cF}^EvwP>HD`={wEeTxvKay`#l2g&9hAScGkXi;mZ8bI94Nh@*}G`!l1GI
z1%aWg)sLq+6zv`qc9~s0z#at3)V44kV^~;){tsDS85iZ&#XB=}H_{-Dq;xYhNOucJ
zH`2`r2!axl0@5knNP|dsOM^60(#?I)bIyBz_kNjgv)A5xt-a#E;!&a{2m8g+lgJ2t
zc?xmn2b}A$wIE0Zv+Rqo8!{!OYRBXeB$IV6!5=ZUw(g-LRUteMz8)TLl?)(PZvpuO
zmZ?ZJcXL>()dNA{IK%Dnc(Xt(!H}2?kSL9w40w<=?RVa1kmK#$7Jy>P%0)fbdk9iU
zbgjZ6ZCfbc@>2U}jZ$v|3xBNDS+x%?C8l4Q@nIIKpkust$B7Td++m+Ajh^GRZ_J#y
zH%<Hbc~;rz!%R=vA7kN1zf{u*^ibyXi(uuD^#f?^Yn%&_x73e!kP38EXaM9gNRhj4
zuez_e+aV8#H<m}=^-jNTiL;hbi35dBbQ;3_MXtV7fm&aZt!@#l7@@V-%WrwmMm96p
zOM`WRqDwLGXKZ>T-hkRCBgh^n4fxV`+j4Y$iYhkM-#;lTjko}y(_@8xei7udG}EDz
z$N~s->SZGG)r|0Tb{~NB&!3|}`^?Paos+_p)k?XBqO5S+?Iu#C(a6Pf`U<0PfP~8f
z*)eDt2*w`~z#@8B;F6M)Zn?6#E{qf=D+kNAAiYiqKtgTjWCUHIKyMa%3oX-pv4*Ia
zUogyZqOP<Bohu3cf`>NB(ihE1(JJe{*cp6zYSp^&7eWPwkx8>}@#%bR^1bw$K`h)~
zJ<Qs6J&f))Vg$X<1er6yV(V@E2DYJZU4Cp<SvARk{iNP;zQhLEkl<((WKkC>9o~l=
zY?1*fFQ7?Gi2z>p7(dd1fXJ{%QVa!ZB%RZH@2e!440OB#oKFLZX15fuX7IgS^geI0
zAdQ50y*<_FM%fSS;W(`54pu*iP*n=dC2=j(|69_Uj|#bbbFOT^ko&S?6g*jKI@4XW
z#H=6-ll*i4d|>>t-3GZpyTI4w1Y0laNf3@WiF5Y~?H=%`e@&{z?r=LvEG!z5B)I$|
z58@z|*pB|`{*_F9(9&7FCsR}0lNT3%#S*}N$HuycMb@^h@(h5TWqhC>6LEV2>d)-P
zh&yo$4;>9&06fCSH0qZikTwIBmdEj=3Yb{c7-y-jbC)acMirDO4(?R~9mWEd1{|+c
z=ZN2K%<YbBGxxSe?GnK~gN3%Ap3(&is4#!*+W;Dd*tUgqXEm4aZ@jL2uca3q&aYAb
zFtPz<g_X0~jq06_LVZf(8%G8>By0;Qqi7iK+G)5Y4jRpC?Y~L!)Mtj^5+xEQMaa&6
z-qG0wmZI11s#(!LlI4QeUXXVPR-w08^i_m>ZGKBn=jos&RyahaksaL-_BVg0fdFn+
zl)+VjoezD3qe;cQt6uo;-)%L^F8&kg884y}9a+gyv$s2W8vW52(7E-g(@yxIK;!*m
z8G#md`|~*kgcV62?)?Rx-y6sO-NW2tgSgQ^s~h(0%epapj<54p_oBY8T*;Y9->r|a
zqp_5MVDpbu1MPzsUw-|9$9^eBZ+Rt|9f<zt8fH;e`n6Ny1zxGD_ylHY``TP|K1O);
zxWD~Cfy+wc^orye5(gq(VvxV74`u}Cy)@s@nnV0ew3T`4*ONtyKO0Ka{;)+EK!p#N
zb8cJ$FN9|9f7Cv!R!U-HrRh6jXnUfBd~(<skNqNJMf7RKG$I(^a&YX<&yK(C(W=;H
zg382zL>rCMg<wZIXn$^et&2eYuRv!2xcg7Mu&Cu{ez6{r$R#t`(q3zgIILBCx9XZ~
zN_}F+2UoksP58zDZh_SL2e<h08c;Tt@HRh2XP9ZZGYP$-lyk(i6@&k1C7^cx>u)M~
zC3L4{?(ee6qd96h6_;bo#%sL$ejB=Uu%D4T4j;ZxhEn~(943Lf1JQ@0(a;We|CU8K
zQUu5)of6H67LbE|RB?%ruP*1u3~noa<HDj6S;%&?EQ3t}i`w$$X(Eo7n%^IBTdYmZ
z&riSl0J?IL5p3vFooN-4p;F-q`!+Dq1DE-<Pu-#96DEQQF+S`TUbV`JInT``?6b{!
z8KLMsc<FvA0#^7bq-52c`GHN05QwMB;iNvOAbk3iHbrr3%=8B<S5m@l|6TiA$zMcd
zv3YU9j=1$%jcU|xL)|NhKT=wS*IsJHw39I=maR8OJ6=<kTAxu6*7S|9%0KQocPA#?
zgKj$sI9K9CLEnMiFB5MXU3c&?Zcinlh1uE37gRK6y?IH|y}8VK$+m%gtlAaN2pagK
zxazI@{az$}=Tkl3(H<Gm@Z8zkSlgJ&^^a3?L!701;nz~}DlEd*EE1&OF1R=%OJ$5q
z-5+@p&&>#XQ78`i@%(EZCkG?PsDkx_T|?7N*t2=Y8-0>i)IYbe@TZtIFMfQZs4zc#
zs3o}hmLUD$zSYsR6pwz4?D&KCcy=_OTH|hmS+gW?F#%K`G%uL8n*x@w;sk)FXQJT>
zf25R>-5T9ikjeoDeQomgK|qv|@!V^@R`1IFbA)wznA5gyr{ZDb;R_{TQl#ImSL2P>
z4X<r%#XB}eQs|ve<@v&$W<u=rdwjod0`PY5sw(WTHEiY*KlG1(W$ih{%L?k8)+qYs
z+E=XB*4O*;tM5!P>4I%nIpKKrADx^ET4I<ugM5;Sk2Kq5_7_?3Zx)8Y1$S~XrJ0ja
z9BiKy+jIn@^{#ZqZ(1=J?CKbi0~j84$;@5z{ytNU4N`lm6a3?iJq7xD1hle_etS+V
z-+2OK0AHURp`$A)E_y7e+y1MMkjzs|Kh>KqLP+@>wP$(;MrN2t`Gjqr5AADY2A<e?
zg5{O>CqCAUOfh`Und_~`ivmCp$k5w*+_{eV#*7h3ygw*+PAKw9T9*<{3iECg%0yce
zvmZe^#r+V{pkM5BH<W)As0k&x88$RP-~5TR*DJ%Bvi<ls?sJ0%oH>;KoiW)u9-?_n
zPa_eD)GbHx3?Dhv1lAYLoEnDJ71S2!OFI1}WQn|T8*w0_UvKvlL3Kc7^<bm)=#b+i
z_2JXhUpE4%UH(e#9p0*EJncDqw}zL`o-?q86`@w6NE^~h?RM2rtIcElW7&9ikl*_U
z(KX%-c8L$37d+Ig^gNUFvgRCd<qivgalQ~kE@LriOPjpWR08jQU0-r*V*Yrk7X~!i
z04{<H9_(CXF0iaj%Q!~HxV7?a?#Rn@F(=i9?dApFm%_Qr4-DibPUkn$Zt9p^sIWsx
zP%*r-Yb2-Pg<a%5&N6M;hMgKHL5&Okh&OpsMBfXvw{JAx$(p+s(ckp!U&|a}oSp0B
zBPKz4yLF6UI;gU!^Ptqz9~yigs?4!)spg?pbfy5eBW6fb;laUwx9#TJO{+xMmZs|e
zs3Jo_FKzK=1}WQFpA}^p9~tQ#el8OR%DsTiSK$GTukASqZdVJlSMiQl*I`?S%a?;Q
ze1n^Cmk}hJ9|oye<88iu6Q1(-lYO_sU)Bb6oc@?Z29ssrj#7=2QdN-K7+jROd*1T>
zt8?8JgkOtkeL>2{zW%(bVM`BNE?n(aadgpLR9rkvZrSnlQ>gzJuAcpN88!9jgV7_e
zjJ`Ya5yz<oH%rn;#4E6#-?IF5n}Eb_Q<^!;G|$0_#<tM4I~N##anTiLXynDy-)BJv
zUy#n}PW^}98qk0^RNPd(C<cK(+*A@O7#0#YvdzjG%F2wc8oNu@TfV@qzgx@$MU859
z4jwhMXm5*&Y-&gx<lSwWW9L=9O!JY4J4U6rN%P#%B5b!?=@NK^Ad{Z7cbyffjvUcr
zG=m_ih@NdHca*HCFoZ`X?v<p-<Mz8C2jjwX5W)GzVy;oIrSVO~_Ao1=$Y9=E<`Dpf
zHY&t~jj9LPio^fO!5JEXtY&zyP|wOK_b^lqX3IJ+jZ69v|7b22gd>Qa2|l#W^fb&R
z?pu;yJnoVd4#`xRlDXU5@~uMK-+-s{&&Hkrqp*DEWch4x3rli@Sig1pMrQ`<#$o(<
zW63Bh=i!uCx2`wRoNbz*>xV!Au(jBT1LczPHYLdnV<s#R-4nb0w=g-#+XBD&FWRa@
z*q)rUl_v{~yoL8hp@B{B;3ve(jOhQfrZ>t;fs#aV+p0Kc^5#e~_fsKH{6%(qc6M0(
zT-ais-MtW>MyPka|N5{IOOaec2KWdG@TA-kobPwWzKdnng@vUSIO`E*T9ZyC2wQ6b
zgB>1i0Ua#5(6;?kC2>(h&b7q_os_)Kf;2mtkQBrkhfqaMD36-|N3r-8!Dl46+83bN
zW-Blo77)u0phBzb%rt|qmmJfVzRf)7zlkJZC9Pwu*u}&~g#jL~GK60uZfo0ke^;bX
z>*OerT(A!L=}AOQ*V`HCdf4@h3T;|~cw{`r?Yvc?wo5uQg)uxHPy_}by>@8Xo%rtl
z1TR~^-L}=XG{gBr64qMl*#5p57u<HP?gr3y$yZFA$m~>ECcMUVr$9%7yo;8AZ;Bui
zcU57VGV5G>Tum`HU&w49A3EUqtOs8VtdN3HGjR<yK~^X7&J`KaspSQ}4vQCoZ>bq`
zU;Zc*D;@kR1Rvm?eJ#rU)~c74aqcv<q!L8^ZF<tf`)ruP1&W8l{D!r7ce2s7c`>zl
zY0*w%r+-IwJ_GyV=#`OTFCznK!Ic!qSNFi1#^?Qw_xZES{*WL@Jab8a=wHV}2J%&L
zs{1^nPG>lOz%@xjuvBYZE{%Y&Xl*9d|K$T2(IVql<qC-}_OUWHfj)6|WYaa9zyL(l
z%m_Jskg<{j4laOb{dPP0tp$(RUCv%3`1e*gaiKU$&rvenLrm2{bM@jg_KW4spC|Kf
z24h=!8}vG3=s&7O&Ydcgj;Cq!j{8hi=JQZu#au0ww<Ue!0pUX}zUG4N<s7iLH`v|o
zzw95Z{o4FRKo)-l)ZUB#2`6a4)_g@DhVH92)do({PULAgHP~T$Aq7f(Z1<N}e9`P0
zh`cuA)I}fFi7B3+lFe`cmJr=svn|u$Ez7VCcha902}radzqG5o>qkkpw^qQMtl2-+
z`x#5R&v{?&tP$d_&<Nk2oom5Qp@ln6SQ=?nHMBp@&iW+O@MDIA&Gk}2al6&JrSo|c
zwd_=f7ElTm8R`Gl%)^67sDpE+h-@ykfw;j3vDThPuPtUO$Ml=u;vt;?ns157^>Iss
zH&*kFrlVu@!oqF1l8_YhF5f=(lc|!SlnE{7ZQHyz22WwQDqaQGJZ>k{Xhw0qZyKsm
zt%px~H(1K8zpiJ$50&M8aw&Ck(dnAXd+t%Z4@~_nwR^Dh_V&8Iddq7+BXW3AWbT@-
z!2N_frFt+86A$_{tnGV<GRVnw@SrhgcLlX4(;e~yXuRkr6^k+~b+Im(K9x(W^ZM=P
z9_@|hGi01?ratdu^X;%p0Ig|6f12j09>EN=;Op(LICnmSeTBv&vRT%KZE7lKu{y=t
zqX{qI>Lox);MP@5CBhB^sJ-DXq3T%U<tIe<JO2g$g=S&*{kFDLK*-Axm7bK=);&Z&
zr))+r$7x2OEii!2xnhx6f_zs_nQKPYr5ma>;!B^X-?3)eq#@Bq8q3(|TU(1|m5;7B
z{gf2hWrX<Qvs$t$=?}QRAYz7h?xdvH)P02jm^YyC!wJ)gLaY8|=C_%F0SwUXtw>7?
z3GImAdZ5;A`d^{fhKw#<-!}}UrZ0MG_}Ue`M7Y{{eS!{Az4%2XjPbH0!trC~u$&r}
zJ@KqyiH($0bs>GQwO~v8b7Z@8uwu}s0kZe_0Qj=@0`g;71!r#q5YDx$!1H0a*2vv}
zmGQ+UUq1J_#FE#s7ZjCoUf%=i;`lXYpFR!VYmq!}3=!$i-!rn-LwLbd14YYwOK5#1
zsR1!PRI8RmrU`%Wtg8Mn@^K^1tm}|ewD&kIJ;2B3-v98R#g|?13U1_P|6nIDGKY!<
zcHRNmHiGa+U{SR0G&bFzm2m3*f`2Jix170+orO?X_#=hJt{ANATFosxnLpz|RN(h&
zMW}}w=JZ1GtO?z3q^`>hyghCPe{@$ogXR8XA*j@SuPr>DzutW%=XJtcgi$tKP>Vc>
z@?qnkiNeYCaFCLK+kCIEvB-j1c?zCGVHoRqdVdNkJFC|?H#iegWB%!t`G^b)#A~~Z
z8&8kR>Ld$A9K+8Sg9YU869PIvZ1vwcx~KS<Z!K4oA2_y|I9|;jUM&3g93qg*`=vxo
zm{64Rrmv`<n|to-A77G$TWtUg#k-e1+=l1sDJcqkF)st-GZ~FhEg3}IdHc*unTlbT
zV`P(&;P<;mxFNJkaD|a~PlsT2f*U1O<NLo^Ro@_tiUOyQ0U<eRhYr51wLMUE8(Jmx
zHExyB6XuAYDR1=Hzxvf9wH03M_9O2DdLO+y?56|W95HkXv>{DH6Hb?k=D=}$ms$xD
z{P`ViWdd!Wlb0FI#^I;R{v{<cJ{tVfE`pa2Qhw5}z%^Ke&~lu1*Z$t^A!7p>W|f-z
zT^xj|4FQoiZw3g_4TZk2HNMbIpndVQk4fl_$!h}#0XjOSRiz7ZZq8W8RE*PkzSAGI
zu3L7Eq(U9TI+2VU8)-tue3_qSJP&v?<N5BJTnliznh8BI4wuIH27ci=a8sFbKg7i#
z*0j4G+)o5hIwd1c9StYB^U>##C`*8Eos}4pb+qe>;@BImqyl^Szg~}W#{BxlX4|*L
z&KR78ku+fYeqEfIa2j~rU)2`{#WNR0?b^+1k6&5J=%cvwY1}(n-ye#CZx7i&(XQX>
z8%Ic+NJEnc^&i;fAy82FDskmTB_6o}5>czOx@X0tt^cu9j|s<eA{~WWwwU?WmCa3z
zBpx_YvvoSa_2_4#mYpi0#8ii~nwHJE1VZFz?kvtJ1KQuQ_%ANF?tf0UYUPod>ARNW
zxygWx8dZZ4Vm=r~dy%d)o@7;u;meLSi?n!jziks5%W{`fM>SE=h^yohN%yJ%%~?dJ
zTo~ApPhM4W>g$C>s7n9%mTTTS{?34HA|EA1Rfw7|#z;4OyfaUnNu#0bNMoJ|;b?eM
zkipbewBjn#YkB)d<;-g<l23Ge@dwJ1*YZ2{rGT|a6?DeUI6hJC%y^Y$pEV4ch%3fn
zk^mc{MxmKkLX$U6ge;DM91c6fWzr{a#CdMJ=Nn0z959J!1OyN`u2oPF=yE{kPENwK
zFdWtaXls$8+Gftr6mF(2lCq*xONZsVD}%#;K(V6oG3}5;JljuPl_rZwJ<MN$G7XsP
zkHl568>P*m&@h<Ln=^rqw06(T7lEeGru`XsqEHBR#xA;pfBE~d73G!lM|vWahaCLG
z2%~;*`m?bQGm}PycU_Jfnij<U?6C`Ag~B4I3b0Y>-F1_(cK+3i2(FE5i+<r#RS0#s
zJjkjTg~~Oa9yDgI@92iC3%dTS3%;%35K=i--OUn%1KHuQ#L}^_U)B_lnf#G*GTH*r
zOZ8A0hTtV`32*xA_al=UZ&=S(xtDC;L^&5)ZXP9~P@ao$sGn1FyChVZ9-T`W{`epb
zD*vQ38#v~OG)h>=jl)h#1|furGH8Ns*!S(u1Zt;XQgUjrRl25ytq<i*WEh0d8x!z%
zaTnB;6{G%koA*hIl(`FYjjaP()|H$NAS9cJ+9UO)Q2BlXR??Sx(^kE=O20NSwF5HV
z$-Tm9_lrCAaD5L-3G977#LEv4C?VxcQ0r~ySK;?c9-oxx+fnhs*7Q7PQ>BQsMt_vh
zyWn+v=*hiUS{7CA-sN4E)9u;*nd@{2ESc6?ydptGMB2OQDs_wUzLQI-JZ;%!;z&|P
z8RGLv7hcsw|H&dx<6&36OR!`+WOOS8Wu+%fn>zHTjl2jL6hmP8V&iLVo7ZSewb(zT
zXbGUMCo>}=w?|@a{KvOfTt&ewruo*0g6}2WBm0x0GrkDCy74y_b?uqr6TMY?_3fQB
zsClPsyKKy#l2Z>I6O0DUz@vk~)jJHk@$ciVDALExGa!uDD2Rnk!yi!Vg7$+Eu#x1M
zVHv-Ff5Trgt@lz-OFRuauKE>FjQ+XNYv+T^#<f&kgf6@Zm(iciYti?uU{`=n1=Jv3
z9vbrDBR_wmS;mE<oVjj&F3k&IfrSJCD>6ibmNc*7ZwllOOhyUis+AiYUrtqR$9Cjk
zju3uCeuV=Sn^=F`r3zQGAmJ<s^!FP91phEw0+exgHzU-)_p$>)`aSPe>dMgzyllGw
zQ!p()3yfdR1AKclPI>S!VK;Y*8J&gLP{pZKHHQ#UuZbRalci>tT`hh4?Z+%-)Kfz6
zXD3b@JZ@=(BU+0&6X!>{b!!3zT+ifJGQNEapZRHfawMdSW@GC?un`B1D(g{}V$n>j
zYQl6HjypFuZ$^n4&@mWK!_mRk3K69ol8&Q}Pl)n}L>@ZTz5`bBj=o|OP6>oWS&@1h
zO8X`U;A=q>!qHNfnfXVnTJa2X?uQc(fwo3{8@<yf|E%wVw(?LnlUC(*6*RW}IXZ<>
zf=g9q2&PDig3Lu)GwZ4rzVKCAjsBTY%UKoNt45qX1P4wStWPuf6=yO=Mg=y&R7b-0
zUuYA4!}uSOV)An|0*@)0J|`NzHW;w{hXeAgFp;fE_Q)k{VK^iIBLRzDF9ELaV8qyX
zMj?R)vW={(2u`1QuE4J+^(vlsh@a2jIXru(k~CI2(Q$ecw))frygiWyx!f_HM{{5q
z5?}hJmV5qI^3A$#D)(vQbV?UQj!_r%I%X2$B@<qDTN`lQ+iW^>AY5#^Q79p&5fh?5
zda%+(Hffw*Xivsrnd*_e{eb~*i~#$WeWuf$%SLZ&fxSN!`1vE5NC?nDTYZO~OH3G3
z8W)Hg6t||)Ompbaq6aXf@{u9xXe(vg@lZ4p^I4jNXI2j6{KbQorbKAJKHdcdW8~O+
zPfp_mN64@nw!HEQ#+^Z3=*gO&h8sl!F6scUJ)KA3OYDSAR=A{VATC)Sh}W)1A?%{#
zzY}~#DAkCkbI<y@TV-PY&ZWB>!Z7l@gIghjW!{k2Kt8!GKv#M53+qWY##}lqwF59D
zqe2MIN0esorF&x`So*PT$w4t|@bmA&q8m@&A;hRwfPPYaLC8!4+mljXNEdz=bFuB3
z?8?~}wnN!OBCE2;YMhwD3(h`ULv*rB7I1T<&WDSX1|8OP{CQbB^x$S+4>-x$%wF*R
zWk#kP;T)@|ibr3HnI%=vf14h?25D!Kg*;oRQW;JRGGpcpYNp#xQnx5pq2S50{<#=k
ze~MRV51k>Qlwg5y)c&(H1-RKmrmoYm67=l)EEa|U!wr7wWSWYCTnWNOC4`dE)`HGI
zYqZ(3JrlnZ=E$r0j$t2pmIcwny%R5%bi?{2EO%vg(iWdw3i%Pr#v=;P0^>cE0Pi(r
zHmWr4IoGWRB3qwPu6r~2cOFKUqW(`5H=oqaU?{R~uf0-dR+ERRB-p81<_|WE3*)ab
z%gj@Xjy5X@)M|l#jk$NrVk__lu3+<(Vx_)Ai)>}Zc$9XW_LF{WP5dg`lU5v%GrBZn
zdymijJk3ETv-6)DS%6sIk9eb5)a#sUQ^hmryKvctWmoKBm#Du!OR=LE7xcfdcUqXE
zhUsmqbR&)SbzU9Ld{V3@{$pJhsbRj)Y+cP$c~9eN<W#c-bgrg-GaRf|+5bHP8*G#E
zpbX7Y;03mXfMKS}^AuNGBL~!Fj*Ig{to_$d1nK6`nzL8YC1X1(-Ohc6JDvi1Qab`6
zQ^gZ`rI_hu0bS!ZS4ey$o}Yl7ySD%Le|G&Ha(O-#xPN`AO=pa2LREnH)}*3NF+I!v
zlSW2hr&_~a>ie;ZS%oL*>NAZ#xs`6$LHFfK`AJc2I#z(9VOWLmUk+Q;YIj>aRiY32
z$R7b=(kP$tRrz<<%rK|J#VkexnRVV_^3T%MCCu9xR$1!aQQ7ts&F7Ih7q`39+aa&F
zT4LUD2(rSSDCO5Bg*G7r_BNcb9ZBcIGYkHRHAVVrbCuVPMixnJ#RlkvP>Bx6yI^yI
zcAY0V9dTCZik_H;)s-QuZ`PR@vCjI$nd2SeirH??f24`|?Pke7n8`_85vLKh!_hDO
z5pVel(2`&^#{fOQ9yC?4omH^(O~S)LCec58?t2C^82!!~!ATp>fJLrnv1b2SXtI3p
zSD`Myt5<22uTy>3m>&36)9e|~A9Ib%45JKCn_wWT>0=r-=Nw~8A+ESeh>H04dFMhd
zp~S69YP?yB@kW#EO>%V+_o8q1YOe3))XD-&zgVHHEOjop%2USKM0PxCy*OQPb46=5
z)k<XpG0%fW4L-}w(EYiG(7z|AC1}^MuFGcprONxPIv*X{^;8?xYh4e^(7*c%%WrhD
z6W)2x!K7>i<B#zTL0u^*&GHK`$kqtN4Um0(w&^6uhWM|YL0!SJ4u1w?WrRz3AsV9U
zZhP=<EbpYA8JWqz@v34A6-s1!xyuchELtEOUv%|oiLRUdzP*Q~{RK*)={fuQ|2m}?
zI_qTjd2|eO1g#zcG*BShet?wQtGmjA3#qi_cI$ncpVxfQ*hf8`R!29gO5~Ne9FCtV
z^%#$9OM#F)#oUH_1HNdd&}O3rChOJf^rM`r-Ftt2Rq>Br_yTxx^vnYV%0z~-iA)7w
z7(m2D*qfopyfO^_q*dp?FAmy%-ZBS1^sU}K2s6;?m1=wUFNUu7)2!@d&O0BYW$r|1
zB|9QN&E${BpS*Pf2TiSEvNx@HQIh?47beRMJi#$61K!}J4Q6KK91Z+DWfJ1FW7C20
z-3C+%cjSGqD))<mC!reektWZ77pSx?|9}F@*cq@6D?R0{&XPIHtVf><hc3HEE#pw;
z1m0}gs$5OHZbSI@IXr`HGKA^gdDGE>yEk94Y+!KJF_ChP8xSHdRn@5|b>0s3=L^6k
zBocUCIAZ$TP?Z)#m5rVD9CRxI2-Cpo_JuFm^WULSJwO2t@I+)#R|2t|$V9#d=EuF_
zmvbS5lMB^65kt{KHZmK2|FquhA7@1yyr8p^L>q1auh8CEfH7O)xCGVA&wHrV00QpZ
z!N?*L_d5Z7bze^dv3iYRMjZ^*g@X0y&~BVK(9$q(^6*3eBVRHiQ40kifk;FQu$KB^
z%sf;w`MBLc78tj+-sQ}9^=6@)y-$Kk|KnDxal*lcPbvPsQj+YXWv?`{L)v8F<}I;%
z)t=?$+^ibhhZNVl>WtQ}IlbTJ$YKrOsRjn#m!o6Q5`z88Hv{L>=2Wcu=rz8K-L(TE
zLPQONYdX{y$X5+5p?%HrB|iX4K{16O&_>fXB7kJ}4EcXdoPo+RuGb$sHtjhq2v(b%
zYPI=W+pK>1Wtrt{qlKu5lH(mwW$V?TnOj1x!^I+f+8>Z=81)R|%|}ck#{&7fU_LjQ
zl6`T=kzB^h4Pt>saY%p;4F$z=m5x=$(EC7$WTNMMcrf?Aw;k#m?3D>2|Cn2O1ZTSl
zt*QuE;$CdaOSOE*rd{;B=+An3L`70?!imihqwPZjzcym5wm)!wIm9pTVI2ryf=jqU
zJ>=nz(S_$|=f3TYtCSUc{ulx@(!*LF?oN2DSc;Cn(lI}2*_DKFQ4O;8Ef1=)bSbpP
zl&;Sanf>bm<wA5>DwXS{&{&3CW`RL7gQ=p<yV)IG>JbI}d#9)Zg`goj^n}FT>UbH?
zipn`hwEilivf+NC^X#lQ->E@iJKZ!Yjy!JzTYKh{1}XRs2aQ-&&W*&ZeufH}9kO%-
z{v;aY@<7ePE3dPU6P}8xS3ER#Fs%F*6ssL*W9nsHi`mFJ9r4QovaMbm6`alpW1H@f
ze+q-I38o~qv3`S(LJW=ZAWlH)pJDEw);J-KpeIhOwo2^nRD0=wDClos<U>!g`IAh1
zmd>iNnvLgN2Br%SQj$$`U$A-=P}jNKEk*6xEi1E_C1Rsb5w7@NikHGl{|ybGD)HJt
zqCdF17*M#o3L7Y(v4YR;Qc*y}5fJ>L8#wRk2OQVr76p)z-PG$*IE2wDL4IQRn0FpW
z5|}2$AL#~nllgT<`4p;@-g#V4OK|lmQ0MXoR?H<mIgi$gzplac_Q_f7SNKeo=XCSq
z-H2O=E<84;1OKskoV;VxUGeXaXrfL_fe(dhbL!7isYz2(#0<$AXr~`zlW-jn5(VW`
zUpi)H6j;da*nQT5N^CAYTxbZs1J1^`N9794j3dw1LCJhyDv0Tm4r(++Vt50>9mgDA
zC1N(`e?b5;{UQSgKhLCpZ=8-G$3%5QhOTr%p&ZMJ$gVBZOLtr|6c^E1#mc?9$&PWQ
zDhq^Nu8QfMcp(36p0sI?8<CIhV;b3{G**0kZMXtj)oTo9GI#9Ve>SH9LJHs#TK_Fa
zwqo6l4t~veSB!K=OLvS5XWigDo|5{~>uf77$_-kM3w>hXPIf8`c5A>?hJb0a$@kLC
ze*Z|*4p$xN4#(F_#=%Fj`E?&vGhVZv1P2a#Uo|J7!>=GyZJ8A#tkER)weqOhDcabB
zAzT{79-F_=V-Voi`2G8B3Et0z-p;We9qQ^DpGL-ObF=jhqQ`)gK@{wlwHfG{(#2XR
zOB5!1i;bVj4L8b7qo+>IoPL)%kx~8sUD+7sOC!ydII@>mSg_<%^M1yWx6@4hi0dX`
zN@Sl!N4a^@Yn!$EPWOQ@L03zr{_8#h=JeT#vyq9=Pmb_HvO`o?6TFTYQ#yC6MHvpF
zuG4|p(N0pz6T=_V--$LbY*y0(?@<uwKMQQL9&d*+f=NSU>0i?^MuD+$X~m(bH2iYj
zZ0T6D#lYBRZE_?L6_8O`a2ymAZjHol{|v>V7b1%TtBB7rJ&_S$)@JTUT~E}NS@({5
z+Pm=fY!x-6e;uWoD{PXN8=nVVzs$bvu9__{(<~f_g$!=PIK%f}-L)LN+p~5|C!Qq+
zU@O|`w`PEV?bLwMjRgB?DnfiJuneN^hDbmh+ylr#<|h~AP3uoC+<rDbjGaCMlTi=<
z-Bj4SrbiZ%ee3n8Q1~KiBZ>rv4D}(j*NjI<PA*{X-%}@DbJu%1a6-!zogDeCD?#hc
z&qjxf1+UFdIC*CIDvnZH%}%`q0*W8vI`*5atpz!*c6r@nz%R+3s2K@7ALrEhe)6%a
z$NuiOf{3WbNyI_7^5|&;P970q<kP&uF*nPLnGqRkB-WTL@zZZ&V3d?MbL1x6@=#It
zg_MV1FSwPIGTb*d!Y;ElGUSt@vadv!+QO9Ub~np0t{7SU991Xmx?<L{bazdn1JEfw
zf<d-Ig>Y+S$&>|{t^1QYCkLUSt2Qz9C?M6S#)Mxp>_ZfqViXC-xykk62l&xQ>UFyf
z8g`I3uu`8Bv9{aenO|Y0`fUMsjq+1g59Jz8XB$c6#=N`V?K8#PzbX!0)JH{GPe^|V
z_apJYcQGG+J|AfPjrLPCMf&=N!{@gMF>mJDGqzidYuev&cjo3#4K*HrQBiUGuo$A&
zcj#5xcy)qzCx7<**%f&`X6t11>f!-56v^FWvT$*2x`Qw6#xX!oOJ^GuzL{_?m6)j7
z<EnkaljWmyaXSh(lh0HM*7uV3(b@rqXrfrL{l&m#Z<}1}gE_{WY0;&IdfQ#Eg&4I7
z@dtlx^H**ujXPoF4ac9z0mD0}priC6jJpe(o{hpans}8b7fJ<Yl@IWq%-#|c9K-RT
zK}1>kTH5#odF9d_@Q8rlQF-v*Qi>%z3+tc<lA&ID>==(-xvx`bPevL}4H_-oS6b^@
zosc3G9lZP4qUth311(%<Eop5V$R}fW|E;AW1Nfse?~lD#td52_S5BwDQTAhJ$H*on
zOPeU&`8<O14?i$&6zAvPA>voJ?fAmI5V(}q(D>kc@9HZaV@BmZnW=)z1aZ#h;c>VR
zUA{Ac0%BlHq}t!%k2vK#p>u7Z;sqEK5hW!VKZURzuzN*;&^5@%i@at4sdL;Sp>Z`h
zM&qUH=9=*yG^*q?x=KuzHPql6&`DUw%V9nLX{dPY=&GQ|C5Fa;KzjCF+5Vod`{gP+
zPJ00tjlxTtJH2pW0{>=q7mE+&-D&OD-xpABr_YP3G~fGdJ71==Q|axi7R`B0?R(x`
zz4iR?W%cD0%0lhqbE?4mTI}x3=>}oQCBLmGJ)f+^vn<o8V<}C&N;<?`Qr%;bK-PjM
zC#=V&`0CebGOJN(I#7kIU+KQ5ZNc{J%aDYtCb&eB&vQR*-UP@vuzfJ}_`0z~)F6Ca
zD0Pp$Hc$yz5IqoIcDG6B`bVhSD1&5Wj47SfRCnSp&^`+)C05G|kdZxL)K1_>Hw5`n
z;A1j9oRvGZ^WLuZ^9?m61h^P$8Uz>ZPsg_R)~3c!AiW*eqG6m#)2J6$OUyCUq{Gs=
zNRR&D@J0rC#Zxu|YyZRNe2k7upM^`^a`9JySuF-O#nc<CZe3_}|6Bw1mbb%Uq|bQP
zCx@L7y0!D_(aJNx*k(H9sbKK6!A+yLiMMUIcU28n;@r<ZC9@5zHp9qbdI}*wcrBN@
z2g<*?ZY4$3sUFpJtBS?Hse6_h%6RDUJ1x2k8O##i?3PJzHk;CmaFg1$Ow_<Kodl$N
z@Nz(y3$V|aQ8sCZZ$j}08Z;D&;sfg(dm!7w|L@6<I?hA4QQM126OvL=ouoB@Ol{NR
zGjG<q!QKd-_p;f{ImGK47A1A_KO0qFBkH$#6&}yC1!kLZKtVUMvY5|}{jbMrdUI>n
zXwZ)AGZC?KKonxYS7|4*$?xNYrMX~Knb-%O-4kJTW?;9JvbtUV?bncPA9C6@5biW)
zy0)tMoQ>Hpd<XULdE1YX7yVUxI8sjj(pnhmWljRV58A{OB5I#>*yq0?GkKB<k=Ecq
zF8N&)33vhnx&|v<9rup%-<Um`K3?$NtYJzVZg$O^jL&Hw`Pno%12%X9MLF-$&@w07
zYPEXAqw%0Kzo@PdV!nDD^>66EdgwZtV@z94inj_WVKZ)L;M|_!GxpnX@+2o|c0~~M
zKWUc7qaPtj5Lk_EGShEr;v#aVggya?ZCwIk|Ln~$3DMCzZadE8MQZcR74j?V)GT9N
zP#^S>j(h-^HYS*|?eu38QVP~V>(+6k^s9$fn=isW;ZXt^ygcNDgG(Y0s)$s%>7<%g
z?90h8Dj*h$(S%&CqFBh~EkckZCfGpWFf^C+&r`s(8bT;1>=Ei^fe{8=)&^m_JRiPw
zWA9N-KMK;pL@^nmX;Y1_rq`KU56V`o7rq*TZx$hhk^*x^j9<1sv&XYRjZqDmJE8=~
zA`~B7Ux9#_${d77g&F%2<jL4c=%j|I%d*?h)b$gsacr>^a=8N0<a<Sn>Ww<Z1)Ai5
z7$w38<ST*wNkRB4a#TIMSaFraWm0b@EKe-LhV2d!fOCT$@uw!xON~&6X-CAE#dhkj
zs)6gTX=aCk#(3>ZXro_JYnUW7Z1WYx99HQqN{&N+A4VgLhUQ|o-`^ProQ<`ZaeAj)
z5aFQG0#3{`nl=)q{dD?QU|P6GpmH7ZBamIc<P1UOqUp+Hd9Kxza-Gay)~KbR;R<L3
z<7J%Z{g8mRp~edLW;CuF2?i{j&I&6}zZsfkwu-WWYIWJXjDXSh#FW|v;7qHEE6=j-
z69Ty>^hbP>6Ky4UUPtj4rLdL(x{HPYe(;9BBsGmo4dWDx?@kNx>Tc1uY;ePLiSuIP
z6&`htjHd7_KXhG?!*(CU>+s10(v=r;?o$M&AX<4$)v)~g=hZ=-k#)6!!PNg_TY+Vm
zqz5LVD?@sn^NAO;UBV)Z9l=84sf6d;VBS;NAAKBU_%d!!WP8Bg>*S%d34l~vOO6n<
z_iSOEqY002VUN>FOF%!m4Gd;j;zpDVko2c~;W}gfhd>`QEGa)k<77Z`w}Y(Q{Wqfe
zx3^WD^tso$dqvIJlOL*7dsBWU`iV1Joot>{s<IZ8H!sHQC`BWpWnzeEfsB_Xw`<=M
zM!~U^rh{X+K{Wz`XAJPL1TvSzHnPPW4Ip&=D}Kx{xKxRbXPTrT+}ImI&KBv_j)g;D
zZ~0R~?F0c9e>)?>XqG0ra)MZJh-#&<{ioxqF;gI$l(mJj685lA>oE370&Sfb?974&
z;)dgrPZ?sBeZa>6Gy60r6C`by-iQ9fk#W?OG?d}ven^It8oHdg)+)Yn8>Cl_1+Jt)
z%VHO2(S-T1&Y<-3j6&=ee+5aVF1-94p%0jFkEf9bE$=m-JxR67Mont3fEQO|QGu(T
zb>oJYmWVai)sDA%*tKg3ssP1;A5#(f;9FnV)Zq$MCeY;LEx^R&EgoJy%3_9Nm<7t9
zR#8O2Hz%UfcW7R(?el0eOl=t2=6D2<fl|?{{Q|7w3N-Fqru2MpZZyled*Pan2(bwx
zv;rX^hj8KkV}RPf&!QOADkdZ)&eN%p4aM<mBs(`3un4D@F4m$YMvOqf5tHU(L4vWb
z_YI$zCtk1*)L{lJukb(80rK{%$s0Co%y5$nBQdllgB9-W1+!A0<YP0DBR<lmF8<oB
zzpiKZTQEJu{+CD3V8QjQ)fzRK^nu;fxF5K5S6t_>L+yD}$*Z^-bgczu=o?F7-@d-F
z@bxJq`!eapS;qa-X^ZiZoFu&SigAvgXrH5wLD2NF0*;%)H6I)##RjSd+tf9|F)BL>
zy@@g)A%x-{O<1g13cF_qa}a5P*6!0(cVjg`4_YJ(@G$Hc+MbW*L|M*^H1ozL!7w$I
z>pP}V`X;*i5j!lx1T=f)H;E5wUlxqL2mT^JKY@e`jZO)P4$h_?W)&=N(iQr%&`xv?
znJP0)C^lvlLAy}wE_|)$J;LaT0KfW85XRxUP-P)Rj(;0Sczb=#Y5pcr7&r`{MZVl&
zsQJ{&YBE=AY7Y5qJF0T_j;&+Ogdf>UVDQG?LjV5oNm;7x>0<jLFt;UrW-a64vd8r?
z6EBv|la0?qLKERFx-@cnD&Wf5jIM-hnW8@5GiKCk8ALm^rvgMQg$xLz-!@Ovt_Rw>
zRJUD$M7-F&jnwlIV$Du+sS_Jd9=kO!;cIeVWjCVkAcV}$C}&2z@etm`=*-L*jkxe<
zl5nVFlcJcUPNA{IQoq0;%d=<-)jC4~-O&PChLBA*aV%0&xq6h9z`zc&Fj*%dr{B2}
z@>Q<XlCir_R4DrHEV%&iCJ_<7uPQ(wK3z|p%Ws$XB7OUlFQbfSf;(JNo!`WR(AaAt
zR6-c0Mjra?u!I(fUj1iRjT^UzU_ZLviNr8|6iWQXfa`etqC_>}N1V6!dIWMZ*1fOR
zw&1=&wx1hTo4gC#kb}g+@@*nA^>6&9=(CN^7>I!sb?keF&#(I;Js%FT#BLA9@W@|}
z(4in4K~LxFh!%Q0gE}DA8NpK@y1s$-2`L~|?XeNcvsDez-*wgL&f3L(lp{V(h6UI`
zG2zu4RF&64q^q||e&jQneMrZ_zW~>)hMa+g8wL3~cQWM@bF-7K493MIZFKIY5&`B?
zpz_Na77P7ogL9^Yj%?k$+P_(Po!!Yn2qap)t3W^CluhUv2`k9xRE%5wet&-vrgZ$?
zaxPIj549bM<%5B?yy<utR*^1l8nzo-<H8CL7?5^n3HYP60``m<_9?4c7l>rs<SN=j
zt_Nr4y(~EPr?jWZw!m+rg>Hx(L^M4?JN6x&m@;h<D0Dep)izcQ<m&-%KdqVMqhEgi
z28}IMI2&^+#9=4>7BQJ152~l+5dPK^Id)>dO8y{%X~T6_S+B}gE!Sm22r(=OuuYCn
z+V9ia^f`+J2c)gzZs34#-A{;#tI|rxpd;CgrK$-8EGCqHC;nsh58>Vg=Qd-l2;@cc
zF}~&ozL$)mvh8HLFPf27Y00W8!yiPQl)hj2f^5Kh3J<30NH2)7=+_$bVqz>N;mPgn
z4`Ok^a*r-TF54!0@*>bP<#A&tvl{M{Lc(+P(sSj5R$g3VrB>}<jI>_b-rfcOtF!Vz
z6jC%BC7ItP<zZ2k1y&l|A>1Zp-e=V$!FmDj8TE|Vv@`t${BNcvLrL?Q9<Eof67b%S
zUdpAp<Sl(Leht**gdPCq>j@UBb9Qe9NFCqXx>?NB5}aA8oz%He)^ar5wtrA`TKSJ6
zegR&1Ply%ErM^#59?Yb#{dDY$fGGo=pu;p&USGyIY(zs-!@552*u^*mbt7Cdo;`(9
zQ>H`>-=b63t@<O7_pQk?Ksd_q4(J#*&~4|i?RFwQk;I0F*PmerhF}d2J#bS2If~pr
zPLDb$Pzp^4QyE8Be5KTk{;St>yev_Vlu9pG7yIeVMZG9nSBGKzHzQ^JNu<=<Wt;4t
z=u7zfNK1m*cNWHL@PL(BWPsg0EUC6vp%1(p6Oz#H_B*ce5O|j_muRsMN56P3gY^De
z59YrbnF+Soza9C}4-qOesHeMy%W{RhaRk58^-9y;%%M!Qi`Zo`6P4Lyk&X^-bYq!U
ztUysmUl;blkQn!o6csF^)&D^(dcn8*NS#z^f=q!tc`c6Grk9ft#QMeJtib5z5Ki#@
zH&5U({B`Qo0f<U(US~Llm5nrR>YaKr^kGhzR6a%iQ*7Es2-7*0M9#EhvIs4Yu=Doz
zbG!S-@+-BPLM>21jxi~qI-sPuEHD)c5v%i<-Kqt{@SIwd#mNWC)_EPFqk<u5&9!U6
z!i`H!F_52q!UCfOp=J7urqdMQ6Ta^1v-r0bz>8>@^4U<O!VV)tJ5)QB_#KtNT#8U*
z07yd8%8~fFML<R?T#U<7S<l!vZE%T<CkYYCBVk$?`^&)NkHsSfPrjUf)kPc!=Fm!W
z(soI)-(j~aPC(FX<f=P%e;XW%xz#@*gaO5C@_^%cL-r$G2-HWg(*SPkq8P?}lIP5V
zFXyn!9>k5u!4ek4bUyfVuRC-?yWCiy!LbLfe^MJfC>{L|O)6Nh(?SiE`r?db{V~3h
zMiB3&>>U#HmxwVuh$WLTb-OYzcj#_@SS3mm>3<~s%0B2fnM?;wY$;ivdIvW!)$qc9
z9?kcE9w!VBkDNac`!<K{_}F^fIAnN!tb*-&ZS!fwN4zvZP&>T9eqI~gdM0|I+8V7M
zd*vcAbMg053O(#uO6TpP+^jE-gHq2a0I%!6^Z6Bia1!`ztnGBMDG2{<Qt*)7bUHDB
zPEpnQTAiHRigdARnSX=VqF4PaOp7MIDe}XkkjkK$9O6w(XgIKaL_kN7)8~QlmIq%Q
z36IQGMS)`AI|Xkg>)IwaTPi_rK)MYAsKhM=I^3aOq)4Iac1ZeqXGd*xG}f*ylIirw
zGO<$iQnR*%H7_r`j{^`eNM_%~8=$uk5IxdCS!oMwHra_E3%(}sm6EGet+A%LLy?Q=
zVcAtyj|aw-9BpBK_TMoxzAqtS0T}_l7myk_7w~#*bQ&~FTR)#-(EB{3k8B^r@r61Q
zgaqEfX!;DXv;0)^94%)CMsa;HrI61)8?fINnZ`~C5bQ=aIV4x$vRIOtprrzVY>ydW
zvwn&PGY40%;M2{Y>TsW?JI%5GC!Wx-g7{u~X}wqs;?z_+CJ7}>kkkZZ#X1yGlb=N(
z?X!<eyU*w+*8Y+zX;ca8lRQcyGGss&(XrcGADVnJd)VYKF^dpKEj{U@`flgcopU|v
zpP+g5S-!R?y)!Il{P#p4@_<HpV!ye;00{XY`g2-w+pw^BJt4*rVQDz1ZytVmE<H-`
z<;2)mhp9rB<JbhWp3G`7ZeDC43Gb!^5^K3^&vQl8vQ`i(Hcop6ishY<oj#o!KC<S*
zGnz1JLit3<NI29WE9U$kJrPj>&!7BlpKMu&6X6|<daaLVqA=yPkSx_te^{VNYrt^x
zvOmn3u6|Bjg>sGYAYnv4I><)2?ek9@27g2M1K2QCV?trP!HM@Tt9|I9*2x)Q)u!bn
z`lsvg%8UYMby4(J-yx$2XjCXbFp?1*IGd&Uq`-^I?@ki8JRcb<Ltw)BvBrlJ8uNMC
z;q9DY1w3;|)&6QnLCq2Z$M~ptrY=Z~CIC!!d#ckZDJs9B3k8D;&C0H=&F-Ie74I0=
z3TEB0RK8hR$9UMg%mHwVAcPor;C+lxCcG&q{k%N0C66{Btb}KqUmy>$?uoFSwtC7?
zBpU(WafDB&9%b9m`br}A_2}rA;`$-=dsAy#AR>FG0zwk?->piX67Q|z?h8)=M;#nZ
zh5D%DXHh&%u#B`rW%TZPy$dn|fkB=v>0t<&Xn->dgo|vHj+8apvx4QE<)e=E)g?dy
zB)4p;5^O^^WfOJxMYR#F^f*hGfp^x(hoa}Jwr!HJ>kA!gqsHhM2*9PLJ!$t;c_|7<
zU@ljK3P<Gd%qsgVs4ba0ZZ0UwaPp8kevhCQ@5yo<^XA~&|B=MX^8Q6Xtuqi+0u6mK
zita!^sR8<CUzdBS@GF1BfXiEpH<el(xu)+aTGCm0O(<OK5S!p<y0{}^7$yTSwbihO
zZSh+X@k^0bpqGl0cnvEv8G>JPgga23So8dSz42A?O}cMA*JA<jKQ}v2Yi$}#HY{)2
z75@{GSR272`w?Dzbcd#(OuMec$XDMYlBQEaqT|~Hw2bVd(dSoZ{Vt;i^*Lb}e-BB}
zKC?A*j^ah9*3_010_3o;#}NF%C{H`lJW|=6hOB-|1^@4GNf_411T)#3v)}rHB|k5|
zN+84C!o7r=r)Hz9OcHRtUyL#JimHqVYV(kLz80^g?k#&R9f}F~pw^NM2o+l?KVu=r
zMT`A~VK=OvuBk(o4y{Nj@F{piy;W~cC!FrCFD+_;H10nJpvlk}pHV{2VE?1ViK}j}
zc;B-;3j*)vu1-sq`DPELwUiAW)9|-20{fCKF&F&=EtkCOW{*!IQsIFPM;?!nHUoUc
z#mH7P<q8fzNIFzN_(ujjwmKnNE>myQ@r&!6EZ5;%jpmC()Bjy%4)CF)6p(82#B-Bw
zP}>Nb75@5wRx;rnQ%?64-t{mMKJt%did1RbhM@5duH8GEDq8ufmVm0`&_B+~Qx9-+
z;A&At0u4xdTvq8YGcS<mf{W~sSmdHF>80=Ce3>8rzYiVeVPjiyF>@T~kiOtFt8u~h
zu|+_@w80V@5$v&9;LGJoQgm&l#}brKtxNPJ&}W|suCsRI?Ul6ux>^C_ArbMTMSn%m
zZWTyGe5AWCjMR~Z_Fze(PD`r==1}dYXx*<od@RKL_bzW(6wx*k5|Ye&C40VH-QItc
zM&?npjaPW#U{kgC+d_wBK~UFHAQp$!MdsPBVaDHDb%tCi&&_-4Wn>;*TU~f(ldZFf
zVmR)B(U=V)dcpAwRq1wJczLCJ!mW3G^OV=U?^;RzJKK3fQ8Q*Ar^FgB`r^IPiEn@5
z#EiW`RF#@mh2`;gdA;)>oH#;!TLmdjZQlrx{g*^oIbqiLX#37KuOtl+aYzDRq~2{z
zFnu$X=$dN{#hL30j13>Y>T@RS<8tc`X5ZVfR=ZS&{U?{ADvWr98NBZ2JN*zhN!OG6
zliitT{pbx?&tj|?i;-^R3v9#u<VNwWOsV2_nVGlvb(BMesnk8E8A4^o9Uwm*(kP61
zdHjo;D)G4e9i*6S`xobbQy8adkhnCm3L~^<i9gVO5gnWMjekKpBU;#1S#J}KGv>E6
zX)g7AnMAWwDT7pPkgunB_0x~yT{+L3r>*bb`Xl<qN)`qphjOQAH!9ID|LPf?#DEm;
zXJVoo3FXlQY%h48bT#ncP1`Run<CB~7hRu@q>}2`r{lA*$C6I7?xa$W>uFh-euS{l
zAmhgb0pH=#!klHdx>)wx3>3OhUiP2){dU_CuN^C6XKZpZx2cqiNMrDIwvEp&E`CjA
z>|4A~o&Ig`xgao+FHVZ<1{Y=RfwdGz_f+KbJR+?wnfP&2?wiJ9y+G4mS%G_`h~So<
z=p~}o@#xi-Ke?F-hDxV^5RtT77mu{cGM~|vl-)=b+mU<EwZ>gjp5aO8P>I?sp)#$=
z#?)2J8Y?VY0PV1{Kx-xeBfZjeUizYKrjPG>uR3ZZ?l)Z~ubUsrcF+8bRRumRF|vNQ
ze+qH`H@#Aiq*}`<C3bRwmG&gqX-5)l4#^l25xm|EKe^C+QK^8x(cH8=OX=@%G;7SL
zpmy+MLo&#tqZ1-JDP{o6Hm-3ggIQ-8?L0N!{w($VF*u0=ARN4cbNBGiCOKFbw_r=e
z7wWw;<(9uHlS$B^l>b!lniIhKTYKsUKUc2L>13;52HYMeg3-GB^r{AW6ta0(jb0Je
z<Hx9z<L|SH*CI@ZKgdI|0W&oEq&-Pz=l~0>Fb2!fKNb_WnGoS{bW#)B_+sV>)C3>_
zKJ#Olp*KQ>r%|xIYY5z$qqP8<b->++UEzlB40><@@}~&2Z>)k5Vl^Nfs==zs<=fO|
zNpMlX6GJ2illk3blVk*gA+lVw#N?+cub~NG6ziwlhl5sgd(jA)K$F6svT6WFTS1*o
z)=2sIs&Zq<zhuKV&;j8Eid>Yl{DUH84Tsx{6tTrF&^XX&>*QTnH}=DzUH?%XHbE{g
zXQ~JG(G|qos})bTGpKClzP-pEN_XFFG)TYFv%&)^X#Q7zEJkg$0YmTGT@kog&kKBC
zx6#v2(nH{Tqoakv_}cQ)i`x#l-vUfJ{4tn~$?7OEkeCt?&`AMou!}&h9OJ~g-tud$
zLR9O9OG!A)QSDU2=9W8Yz)d^_-Z#LAN0hE4234(Y9lc+9j!k}T#7p-M&mVC}+cM`v
zfEG`Ji&=ot-8bo+8wGr;Ew48hy>v9`f&rCjkCwRy)nu7@<n!}B_6mFiVv@S^yA+#D
z(H!I@qqglb_IY;1|C`tC(Q6I~cNWDnc1d#or@OC?s<P?gea@j95$P0=E(vLnkPZPs
z8l@ZQ=AeWiARr+n-4fCbiiC7`DbgvO=iCSDeZTLnyVkvTt^3dYXBM3EoSD7%?0WXh
zZ-%As&BNv&+Gt4&>K#lo9b>2!uXrfh^#*2>7!_mQ1cyuE-WnY}{mvTx`Z}wU#aiqX
ze^-^SPee2DvhtiSirt_kyZZC1H7E1Fde6Ka4`*;89L%uMu8Zzq!8O8JGneN%y+jx)
z$7{`4!F7i|G!B_pv34WYg)E<3Ew-)U%h|J#Xu0Um=L*(03USSB`I^B%Uip;?a_NyZ
z%qR<s%9e|K8af^=jx3|YV54=se<8&3pUgBfAG;76qsT>>fBJ@Rs*`kTcd#kEk^6qW
zj7gn;_a4aEqhN3eyO?2+C5KKkF#wgBlY|NloR~<0)~xPDh~W_P2N*qW-^mv>dgX_$
ziA_f*7sCHJhQxtZsWL3{Gk9l#Pl5WAV%g@Zy0c~HTbg>0s?I*VnC|{~a4Rl@A8WU_
zA2;L2!Qh+VkZaqcgPuN-F??sbstKu_Kaz=peRxd&Y9zX0E#vj|_PW2;mwl_%2rB^H
z6Qht3XkCxLIF$dOZP&8;qRGGIs=H^`PkUzRPNm_^)=7zK*_{_;;O(A+Bfmyaeb;<3
z%@eO2ie2=@iHo8y=}SE6Lhq1wvs~BS$1T?Z${O%Z<nM%N0ckj2ZKIyTrjVJsCL8_Y
zUC+PM>b>(guhULD-=1r^m^hipj7S1<zxXO#-JIZ{n2QoV-9NdmI4`cr86rB*c+gG;
z-q4s@Ch=-$Zc4s3j|t{&mm(WfUq76g7r!n9pM!}lXq6Q7#emqXrXibz1OYNVI%*8I
zU4S7c=MlT!G*@|M=5BXIzZ9!hUs{8$jDeR}k~r!YKko<I$KvD^s6?$##2Wp!Cn*xk
z%NvuftkfGO7m^T<ZVa*^(RS89U@HAw*$=)F|H~}znDy>!^xS<Po5;oEhCx8T$%+0-
zt#=?=+GwHq>?`e(V%4KoRNVK^VrA95?DOrvUf%Dn=m-u#!8co|xTnOf@S);;bvoyw
zC8oMLI@5Lu=h0#slnois#iE;NBsU5852pNt@}9iPWV4IyrMG&f#w)*H&ZaBFI-tj7
zjA4Ez%AuHEOV6p98d+}YtyYqj$5gc0&*v8L!jKb&l-6A5J;9fUqZ|N(5N0j=9j&GD
zBk|ly-H$M}lK8<Zzlv>maWYAK*8Y#%v_cuY6z0{LDn}QT22kQnnbAdYRYU;>x#yWA
ziun@H4UKBc1UdtVN(S;txb2$C-P$)Wx><uC41odsqJ*k+d-|vv++$I;8PVUe(wD5Q
zY}2LcIt*@D!x}{(O`GylqO?D2U2$CLg&+0z6JGY0)Cf`c4S2dvaDWBknRK%Gqe2j)
zRkulrq~K{WQ8HIg-m+d|Zb#3U2L#Whfb~HwU_t*SS)t7quX4qGJkH@tQ##Usygu`o
z<qTyf?rb_KT;j2DLIX!u-6GrMk(c-^J%~3M4_%T0_({UMPCAPTq8SEsj2L<VB9}xH
zDGwitPDB18D64dF6dHMJg!JiJ?yjfL`T1rrw&~{mzlHF#Ll;w5c^7aaogj^=w6K&8
zeXu%8I@Rm_=SgvEYPqHF?XQ@Yoht2T1PHtP2bAN8{H#s!o3w$~^Pb=<l@>m8*z;K3
zDS9sIFK~07C;Uk2rrViwI^`QlvZ^kXZ&jjWp>OT(iu!?zzhEg1_w=dk+2GfllH0LM
z{ryoMA#p*OLAS>~XXR4MStX=l4Z<iqXX%%%D_zv7jr<mLHPzJI3ZhvJ-N93ReDgb#
z8{At;!6a(Sk8xad;i}VMwVz~sfDx+j{5FbQu4_Q2<qjhtqh?MP6>C{zn!BQ9NG!`m
zz2$c=Qu``=!hM%N^X&;v;n%rRFr8{>5a&vd#dF*BlX73GnHzf_@tvNqRTa-IX3We?
zSm-Q9R^>1xJfS6>H0&<*^6l7%daK5d<zocmA=!z%Hwl%Kl+Dxai~1z!Vmt4h{BUAh
z#YxB=i%^H#8RIvV-t<`DSyg(^T&>7kyzD_HaWse+jHCXf1WAVzLD4->@4g?orR!p_
zWzZtq3hsmW3Dv?t8MG>4X1$mTjm&5O@d*!Oi=~#IPd{&rp+R8^Mj;}S_+-m3%b3QU
zoH~XchuYJB`jkc)nzWK$W!jDDk{56v+<fGNm@3grNEczv@*)~0#Hj2RbD9dF*Gc!o
zk(kT;Skc%PvLZ&R?fBOScL`b_Syp5+*`W56r<T`2i6pqD+2wANiV{W`hM1QZVnQ3e
z$Dd;cXX?bGDxDW9uaP0WSmT49u6^CsK#RBNf(8=ekhbP`o)0Qfk#1+@`XzmPtU!Lk
zsXJ1!A2w9mZgIJ7?DG-aHu&Nf+ENj?msK^biAT=3yk8;~;jFdll-cv(re@E_mh-ek
zm|<EM)F?9OJ8hq0Oi-2TG_2$v*Gt3Ou&b7Zc*Y_x26?`rqRumV<n*KN*HcQ6=Xb1<
zv{NNwD%*^NH>c6w`7TGyl{OVuOe=ybwEqnGI#VYJh#jknge#79>hL|a4l^2#Prdsl
za7t+e8H4F#>bq60Wc25OL@ironeA(SAd8gH(;uHc__fS2l+kDNq72I7oDS{$Fq1<c
zL-Oy~VM%nE7k%(u|4%STcgIM#>TDfL0?c!2&vB@QtUJz=G+7=SuHEUoqb@3j@*J7Q
z(KII*8k=;*{2bt)>NzAW+2ZGVIZp$lrV?ccAtpg}isCL}goL17=7%6)nywPVF>*sn
zUQ6}fEtc*NPy4}l<X`F8AqRvP@u(oJ-sk8OJY|8i=dN|Wi$Yzwz8zyP{jQ@NwxWj1
z121trR;#(VDBcl%VfL(<@bItM#I34nq7`0%?1!jTKweFj@*F89kV1CY9Qs%w&Hvl_
zYyW)>Z?OK1@+TM!_8<>l_Ud3t=8wPAHGENsiFCikfA4UF2#0{_=&_O8KD~03{itlY
zAfx~ju)LR>gjG6X5W-KT;so)zOzvxI6ZuZUiV`Wb$5@GLEawb+f)bCXrS!IC3D2S0
z<pkXF0xqP3s2Q(hAgSeQ)z^K_#rnZJ{!NPl;&La%)J$xseY7zJZV^p6@*5p(OpQj0
z`-QkP7-WuFes-3~jNq+l%>v@B4+-VphzPi>sahiW(PYyYdur`@Y_2r<xrlwbcKp}S
zu0tV+rHGowyQGOc|7zQZc$Y6MN8{@riA5fv`f|Kmd7oIGc8pxyYKzKCV_9JPw!Ejy
zT|5zGZ@cQKoM}G=y&!%Q4x_31A#M#vi-K;@k@p&SpZv-8LEPIQcw~c&VL@-Uz;?_l
z&|=C510!4uj<;X~9c3|WC!01e_@_D(F=BX=r-rK(hbSL3Ux52D9Qv_eQN3<e$41h%
zKX@b~!2=5rNw>7*T$?D{i;)$#)d1Jxz?h5yCZpOIi<=0Mi*FXzNl7$0Io_TU5)Z#s
zsM!hW3Nq3?7G#nwt1nzlcU)itkgUl@Pz$ieHe>CMh{;IX<yl}z)XgrPdxXhjUmi0R
zF54Rs#7H!dNpk~2iXrbCzfKLzu%i4-oUR#pg<Q6}vptPJn2d0@h%ACWvY6Dl!yvlF
zn$Hr~$s4ac8~kZ}`k_r+qN@`+H0<Pzf+Lb;%<%4-65WyjGfEkvbhR%Z8yXYBX4y2M
zExKY+S&(W~l;uCxNWL(teE+A*xkmEM)4@tP9=yowP2$J4Gni|s2I4$3x)iI+QRy#o
z8mkig_gAt;inOfO{p4jP6(G9!Bw8Gg^WaYv=h`2jI)NwQb?U`l+6Ij09tD8uvN=;&
zOwf&{_GpF#@!2OtZGUH-3Nj0>Bo>+<*yY9Hq>3=b-Z_AoVU$>Y?0v|$-jjCj;&6c1
z&BczeDFQs4_ce;K)5go>FZz63(xL-We4wZ2XOH;curUTop-V;@V+f;8`>3G%wx~Vx
zUq%e+V6Wq7D<1IiM<&+X`m&S|r<n42gpdJ+oS&&;$V6G$`dxjkeOA)WYG&_-V_STj
z#@%!2yhsf#oLP84baaX>bn>LAJ%Yd(P~L~;E&ErDdw-I|vM8FV<Ox>)`5to(w*vv{
zs@lg=PE7P!U$edkD5Ua?_2a`Fy)+(h*iOO?!Q;xM-;DZ*vuQq}SJm`?oWf_mR~dA0
zXPjA;gzSqv2m>uOR5#kFl!`<OPo&XZ@q+QH{>XgHznb`V827J^O`tC}SB^F+1hP1N
z?>3&&jitq+Kmt0s3KA_nUjfI4FYlDhyd1`oM5<CmD73|HV@h_PA@1_WypabGSM8RZ
zZO;hhxr-;@%77X#5YA}P-^6`~@~3yaMwo%fcO7-quo&&HJ;~vhF)~jvz>W1T;1OH<
zjdx%xuQ|2U?CLeBDsKx}tf4P2`6JH0Xk*K>%0AmiUu-SyxDXMsfe^Mut;F9lgguJ#
zFf0+ifU7=b|CRAyE3O`LnZKDpjSPS!QsvCzv`wKrqNW!T7HXUPsNtx|(rCGvv2<GB
zNWfqdQ%y^$pWLr!B)A_s`bi$tM1LLDP;66=Nv}iI@=Sh=sP^GgISV4!kAL+D#72DJ
z6CdSdfw9P5Ex5&wAfpsRO}@(8zs*!Kyl9Ym@d#I6t#lbtd~@tgpi;K}Uhvn96gF^2
zR@ktM;2NP-x$7b$j+L&`v-9U8X`7n=vV+?oBrd=;BeQUNW7u&#+@U2|WeB%M`x||H
z-Z@KmVPOyMYpu&Scm0wxl_ae8VTnbIW2sha=rj>;WPp=Yg34M7++4n^WO+L%x=Pf8
zlq08A%zvC7NOq0y+NGCHd{bnU#iVcAJmFJyXaPZ|BrBU%Y~Pp#B~fd0&rny8a}jiX
zpXm{@GqwTF-R>_mu&Z5>V#58~glqAnX;xv3RsfK8*K~)vW!v=0U)7(#4O`h9EXT=<
z=o)a!o8BKJB_`gR_AMmjlyeEqw0$ph(siX^G?$<5!+a~^HCWi!=ccEokQv5We6vtO
zsvEt7LLb=Zc3SG1E$619;F_T9z1G*C0M^yP7zaYlzzkzD2ivgDE<445tVB<Yz0~M!
z`BWC=N0@waz1_31I6MRA<0)<N8)K~>C;Y@kE%v+D<Y>$a_l;iA$=gFHPEX>430Zrr
z6)l~n$NYJeHFz;dhu&2jP46x-6a4jeHeh^vsjjX4eIRCNi|>WH9TnC9&ydOEJ17xF
zg_}k<yNjf|gLbR7d@!cRV-Rtq-B_q-G4G)QwJm3P)Z%s4hH9d&&!aLRC>ZPrjoU?V
zw&h4|_^F9Lh=l(W+g<}nh}~~Locwuhe*}MYdg4}cZ%p|H>?=wg+u9ESww?6w4i(8&
zU(&e~ugVf|DV3>|oI-XV!+6(5b>44fSmSopt+7PEcNNdvqSbNmn|>tRkV8+m*}Du`
z2)~E;Cxk`=Nq8xb#OQ_@=T@W2%L`de$=mjB$Er)ZVGTIOu(!W9SCTp@65jDV3WjUw
z?jt$^EIM3uvXHQEwx%)q&7O}mZ>}oMI<CUs8ci`ne4GbXp2e+eS-z;`Z{U$vgXLL%
zqS%I;$wy|6Y;btz@E{>(z|yGd?oElyYa~TJP+xW35<Emph<!9u%tlYM$stlas}OC@
z!I0I=FAagJ!D1<nHp%;^5bcyRaX2_%zv_)%wphjPzl1pWU<YFuOJ!GuCB@6$9=RPI
zvGDo4`z^}e$UODJf+1D49!XNB6>8y>(HQW|@`Dji<z2TDM7u#xBdCpsVvx@Zg-hw+
zm8*K2U`unu)iKicX3BkHQ!A#4V`4$9B^;EH2vd0iUHVWIANcc!-K?vb9bZPs+nnG9
z6?gD7OYV8yLC*r~_#IstEYpNjgQJ&UzjZNuXy)e!>t$c-M;JBHq{%s_=tEHnQJ5;u
zjfGO+f?wE!Pz@G>viyG@V3G{f4}&zwVx}i=f-McVYgxCAdv`f`7#rd6cNotY0k+_5
z$`m$`M+T!i`(jf4E+W@XBR1}Zs5*QwvAgZZk&Rm*c>kR3b|6TD7lTuwd8cYe__<Md
zX(YH6GliR>&*BL>!65G<(>wYAoh#Yu-ZLcmBsV#@gx8lKgT97qL2y7^ioxxwc!Ux)
z6W8tYjYa=t-@%<{Gsy7)nVN6St8`K99tiNI-iV5(4}BWU$1#1^@L{Zut5eqvHp6`^
zDfne%J3d5T+;wC0Veumulkb^pOC4P=xSNGgAru}@F=N=&0%Q5jV$KF-(&DB3B_P2j
zB$3Wj<^5MoJ9tCa>B_%gKijS#_$Mk|$U}yCg;wez;m2n&RjSb;Pet71-dIc0$;XU<
z*KEgUnuV342lc4kND`znu<}8?dcxrZKmmYiR9^bNrZ1qOS7fR2J#=kJlsk=q(g+YY
zAbY030j@f?LX3q_p#UbdYyUY9*IIB%-Zg^DD**uar2WJlr4JQfQ;;4F6FS`;FAE};
zj(&t^P1?^{k-bPjD)+XJ7pKx50I;f!(s2Mvf{oW|i6SO5B+F+d&6@*U`T!3d`=-6<
z4V{}ytgjhi)YQo-ox>?fOUx2#>=obz1Uy!`a?n8$V<rN%aB-jMWA}zb46p(^n0JDi
zesFCPwLpE&b&LJ}1)h0MdBOcO&c)1L^XI_?X;6&TL32r8NqlUawM5mq$Z@8khy0TH
zn(vC=OV_v_8Tbc(g}x>6WIR9hG5rCeAf~fIlD;>O&z-hd^phtdL6AIgDVq0@Dv#!^
zcyI)RpLo7ICJKM-vU4@yo1v$`c-mO7PL33ImiuC06hq9$`fldi(umO9f^{D7jFRqj
zGW&CV_pFDhW}P-;eG6oJ^0?e1rsppL+DZAIUe>zqr3Tzbl)vAiLVCqx4ZQ-P6U{0i
z$77BkH_=gFK*(vB`SA&u<Zl6ek};aEtuVr}x~b+4T4AKDAr@y&3Wd(7Y?HOkE0+Qr
ziw(HqiC>{4H_v#_Pk+NX!d_T#lIgJ&pq)k`6^Fp5ZN&i~CL1DC9-~p7ovUSdXfG0E
zqGBjyf7nsAu`+;z?G^mxP#Ede<`(bEXA>V2P*Htlbf!{Ki7ZB6j{QV+;BFUNt>b>o
zcn3Jr?-2$`5yb>d`zo`^a}hyw16yw?D+^B$X?w7xFf6Ld<dqs|_-awO?wN(8^*Tb7
z#J9TQI&ELe0w<+Cm>5xaGLqnZ?5rZt7$Dm#7+GH7s&k*e&UDP^8xxtcx=AH)sO#!L
z0JEn&ybW0!zrTvIw?Ex{^`1X64})nrV$CFDbTk(Sk~22iasNU7++}n9#59Xum(9Ur
zW5;jPRj&$5US{Fe>j12`*$>7<Z%jSwJw_7W`0+?%q}lCWV{MhcCyj9e4NaD43t_qd
z)i5CvdOuM=2M6m8kLAX=@Vw1jlhx}TJr`&dpTC0yw>Yn_g#@#B;aU63r^oA~j*M2;
z=d<9S{2-o#d7(&I&pqo#D)xuU4Ra*1t|#V+O8$4t1hO84Q`F%4-Q`*<>h9jTS0*4;
z-%Ox^rmQF1>%gXfr5h9uOlz8<nDbyTKSWL$dBc*E>V+_L%{!?k)cV5zAXenvSG}6Z
zne*7MuEw=MY?sDR=f!R9j#Gs>l||MeH9<WmMeL!w(?-clX-9&^JEM?Q#Buw*`SZ0`
z4$bflIi7nfWq9t;%@J?m0O6lL+a2nzNO`!YVvMx8xG3Yg8F$3Gs%@s#bt7kJGh2R~
zgByw*fN`}VI=RodkDk=v^VBSfr-%is_-%N-K+k<|5)K$SU4kriTo?6IpPYAqnDzty
z;Qh&QnP!;8Ft|XS!e@8nxvdiVYqdV*SDe5~*V{-B!D{Ui@=zyjl6hD?9>g=`lk<^I
z@NtLv6Nr2sVX1e%Yg%&V0Zc{4=>b;*XZ_wAj+XUnryxw1XVr$+Om$Ce+$PEk>Y<T1
ziho?~cB$Tog2_u=Xr1ErW667({nri>fzXy~cT6l%GYYn_XxF`^D$A~|IMzbDs}5#`
zERFpzX4<LM_{RPE1l(eDH@@lYK4?9?x;6`^PySWo1RplB#KFH18rmBei*{zigHh>o
z8;U$o#t)flp2wS%-*JDgiGHJpUxQ@yfjRZ-HhBD_yE`*_MzbqJ(JiygH#ObMd~uJE
zK3!SwI0(_hW(L>UZ{BwxtJy(dd-SB4Gy-BtM2VGdKBo5*ERe2rN%;jW`4crLh-ffl
z+IhAEaC^XX7~IgI^3%JHpbS#F4vwH0+fw=*es<XP-oVZA`<5bqSMH0pt6K7_jq1yz
zGT@!$h%Rzj)w7F<;6^jOaFCf*?OqDu)s8009bex1)Ygfy@P6Jzk1xzGMH19omIZpA
zfcXq~j11$C8W8TW4qb~B0$Z+nTwI^Mv;-;<ZPA+~H{<s&j#=ds(8lcicgwC%t*Bv2
zLNKN`gnRXvoFR^PL^-;xA1<9LJe2;PYyP||KgWE%#594i`V6=Bpc!9AlOyxNSA!?4
zcXBz@4w9^_bc97$J?eF=F9LToa<%f_D;1A&tZ<L8v8E~;VrYIBJ-|ic=E8cil6vyZ
ztJz;*uit09Va3(_3RXvYdh;VkafBVQf}EW6<1E<A7Nv_lFI5{UpcWtC;j#pH%?5$f
z?vHWZ`*7)xoz#@?Mw&IU+ixGO_qk28Akp|IGBTK&=AOi9oqdA<Q(X{&hD*5j&LLg4
z{F^;-Mvl{sa~o7>YY^}j0;SOf+_*LuE5G`htV|&m6vxfo8W?jEK4XRzAxntWcf&U!
zJu_=rM+@l?0e8#QdU&zJ_L+qwRH$11yjH~@F8Pxt2>=;9_g*6_1UH7hpwrIEXv$&m
zB5l9~i=X1dg1D{Bgx_S#Nxlh2-SO3q`^l|yTigfOt!LV+k>R#8Wlvsd^<mW6ahTbs
z9UCG2p-T&h%zJ!WH#l&9xXzheF>I+?|GZ(}Gx!+T@>V|^8T^Jf>Lnwb20_}>aGS_K
z?jyTrNk}pq9wmIYQ+{TX7xzy;nm}Z)&BVs|%f+%NTy9als-79cva96%5J5#-l!{^u
zla4#><r6-JKpRTB?TMPA<x77#4FrH_O-8qxxu-a-nfDAI;6QH0#s{LIYUS@OTsqyJ
zJdD?(P0hx=hgde0f67YXzsV#hFe3Em49D19{J5Yr?@2AN)n!jS#4j*8Fu+;HO-r9&
z5lcG;zeHf71H@2cRCJgeWO9Ij;bOz#!{a2q9jhConY*JuLe=3nE(94iP40H5U5U@t
zcTbFP=>jUtGO}(JmFU)w;qZZoY4<nZrWjqx&>)MFKraBnX9o1LnenP|acj`hUFD?T
zIQHOQJQwe;Z~W|scze?NejniK_Bq*e5X?eZR4zzcR*jcfwP-mFsoEqvaKPGdla>8F
zhtim>uZF`%$Ccv%Y2={lt*4}gY^QBg5FL*0(By4w2?C8J6MGvlT)mHuVRz?q_1kf1
zi_h536#Cw1RP!$>?bXNf3E!cwjc6b4Zh7{7E?%><p;=zftWH3?5WQ9HxaUX~tlvV*
zMBJX6KJ+`rvm~1wZLxYY7!|M^J7!0L48wwOAt6g-13sUJAG_rZE)EXn7EA6nD|f3C
zUD_^FQY&bER$?_mtR5QXkLT$fkgx+q-t?|^l5)n*{)qH+E%uqmG2$hg0{_TbpoNS7
zRm6^;^GsK~yR@`?8wD%ew4l`w;UUb(&%n*?E2J1*ed<Mqiq%^8C4O%&865*1%YK%N
zDcd07($|&hU&7v}B1(?u+iFg36LPI_k$<~{-im%4bOnFrVyb!YGC!Y6)QkP#OnLS(
zL?Creuu+DP76+QdZ#}q0WHq{Zbc+3#zkHx8jl(p?Wnr*cWAf!%2<0Y1db9kgj#xvj
zY|UC%*Z7m6@zolt_vd5<io}onpS%r#sx{iEoJ6%QrgsDe-gB5|aT_~67llRwj<kR|
z40w49IQzCU!EZa|Ust;|>BIa?^7u2o&HTONK0RmDWfd~9vI)wE@s^>YKT;2dS8Lc!
zY=`c|i4zSAK$!ITihV~q?tf%=xSj1v-ZSY{%J)Zl#*ovt54XBQUVJbM*)}rLo8+<K
zH@W$KTS(?GQ9vol$Ph*yukZm0tz8}qH^1JZ4S8`k^P(X*q^44umm&LE&G)xKNB;<r
zAJ`w`M4X+EBWZtM7)TN3$@A~GJ)Sh0nPaWmJzHy3vdVE<ZK{=4sIqvnBlY2oacuSj
zH`LJ#krd1E<jMVvk9KiJ!0jB_^ABE}I)6y4g)p5QEgX<|vw5bA0*UGqRH=H8y^d5O
zEf%H+yc89f{jDu?57Vr7hVdb%M=g&}#v?AtdF>}#RXuT94dTB(d$l3@*JHQ=-I#Aj
z?*$xJ&Wd^RD^}0WbI>oY^l!6^?r7WJVlBSKiM#oxv|Do3^d(@MV$;+jqsHp*-^mI~
zmYy^7RP(BvU!no@(P5-Cz*DqElW|$kn}f6BL5JODZMSc?=?2Ymhz_!6wh`%Y^lfhH
zBXn*H=X_I%z=tipKBls?BKLabk7Ra**VsPHw_hUUhpzPK50`_G+Io7oV3Q!#T@M=}
zMSChjD)Tz+m~7nmI0fMR5wbDKg?xJx87<J@^x(0>HVOWB*)C;>U*D%sGnE6gT8O_U
znt|jP9@j$lwl98qUeMZ6S|UHE$oDG2zJf{hrOv}M#&kbVnF-3Gop5{@JGYg30)FR3
z3m+~n`Y~Sh;fzk_38Db=kMj`%E~ZFUg#qQJChCTJY+T>$&8}KzhOIta&DyE!p4Bbo
z99i#-p+h|A?mXHU?)q_z@XP3f&bXy44rv(N`0ZFTkoHK)K?md3`>kp7rqBb|(B@Ka
z3q}22bsf*7Mm}F$lY_fU`;kR;+H}Z^b17fPj{{FpkWo+hj<FQI%i;bA?LPvSL|gQZ
zU&C&t<0dyHf3?Y~!$wqq5U@&hI(C1RUHcIOPvOk_(Ol6DsN++utyV`#T^E~$MrF-p
z+f`Y*3GzQN!v#=C(5aw=c)U02r_wQGS5D$(5%k3*?{E|D=bYUaFgY+fe>it7>FPe`
z9F;)n4wtaT-2et+5z4=gLI~IwJDvzVK0*uyzFj6ZJ99MDUlHkW3Q=TbjgJ}jpVCBl
z08+?{hc@qVXam@V=Haab1L9DeKY|qo_Jx(aqHxRlb$paO-wnumsyke`4~LC3&*srW
zjT_FFkpgFV7QZsg^m~i_)m;zZ`!0X4h3#0R27l*GQPt+gs1LVx^UvxRlCTK?!N{mm
zPW{8hFKPY~e|$b3?CPD@j{C(`+lo)F|GFn%+Wv0*L#Q#W-#pwC>o+wL0HHs!1QIGn
z>vHgSEr-tgzkiJn1igd^NXi-g$44+o?IjuQ&8B~Z!3nhI+3^`3)9V9%9?L)Af%Lx1
z%YWUoCFB$?zE7%GTQ}%hWC;<7r#9J^=t{LM(pH5^0;gck^E1B)@l;RCz-Y>m>D1@P
zB7(6hJ|!@By-i<*H?uS*@yA9$ulvbMRc-#9ajw2=c^ePROY-VoIIYYpb#1MW=TJ3c
zsR{lGFz*2oMr9WNlN#Rj;=&metkzeeovXb^bbYY{Lc%+JfBRH}2n?VKh<{_4IM*EG
z0ejqdH{x859{SXl_hL3V;7@vx0o_AVEIf_v6gzuF4JTv+Y`oy?t4}2B@5c8x@0rv5
z88#nCxugT|`iTXA(!<_fdD?6=6RAmg2h=~gR3K(+d-s56nwf#d5%x#uo~J#66eIBQ
z6z}FuLw4*xoT@>hCKQ*tk+`1MUABPo7-lp<ZNG8X*vDaINEV{~r?bzPt)Z8b;dSB5
zk+Ja&YzEI!B;e1UMFvHcsLy-PN4)q~>^L}6pG}8oS)C1VfOiYE25F%9)0%GeFg8;B
z)oM0izciVMyZY$`!$k2XlvZ={2VEA1umAQ!1n6)yXzH8HM3BO4v^};IOX8xIW)C*I
z-G(-4xB91T8CbBLr;Zyiv-FuOr$$L~C$-*paPjM9khKvOI0bIIzp0qz8DmmP&3BYx
ziC{2o2y62^DhIFkEBuz777+;#32Q6ZcYZzIncijIkp9iFiW;d8>NutKN*Mj@L-he<
zqrE0AOIIc%><yK2S27E(`}%o>cBjrQ#wH;ludJ{(gIf63*t7fX@<m&=#WyF|0J9vC
zRl90*f-cd9e%pjb>a5C9Kbz*vPh8rz?JVKy=gED!9b5@^31cK3<`>ElPyDpk(P#Rd
z1`c<8%{IR#VNPU!k_y~XdnTYk^`h!{0nnsSsaQZB6G3g6u_b;)QF3fz9Nq~eMKSGm
zel!8}J+ND`_efwKuOYS0CSs_q?jPNrNA1l26(9|O`2)HqNbB23{YU1Y?R`TrMx*`m
z-%knAYW(+8?6jW$^;Fr*sImY0i0eg^cGG`$FX4$2e)2bJp~EB_0qi@Ui&T*~LVoMV
zPr?RY;g^bGms9-qlQ4kkGgfPAV1Ahg-S2?+^QY2Lz^#sdxd;!4rQ~!W*B>VZ{tUFC
zF}$=p*;m^0r5L{z8i1xd-UMc5Hs2YxNX2;@@ztX(_HN8tN`9KeT5Y1l_{SCTA?^7!
zXTL>+o(qo%1H@@jYxt5v>g$QDhPXk0$qgfeO!j^2=uhEsc43<-J?}rnZPl`pwkM!H
z=_CInw&%q3+a3XknkHUrP|&lH#M`WHA2!ll3!MQ4;!e?c;g|TKeX-*nf<*r?T!?0j
zr)aYLpqz7bKCvIdWcotK-a+!5XZ+qkkIbK_2u|4;(q1laJ1Gofc*c5HX%)Eq6kI+V
zN*AnRZcB6E1o@S6Pqug-ST(Jt*X+)FuEx$g59%EOwlU=?khiyQ<u~!9A~{Cm7}p63
zI+G?(ct1&WH=+G0X8p>kM&H|k>$j$C8N3GWJAh-w^UjmPLUK+(qUQ0S;&f&SXvWVE
zV-lo({elKShh=&Jk!2s3HxZ5Zw*sfHC1+IDJQ53-9+`JGQTzw#j@zWnuJ3%W(uCD@
zSC@_Ct%rpHGlw!mUe7~()|usPwST58|A!FFC_(OOxFh#CI-a<&nWqVfCc!th^b)n~
z8*GB^U*oT^24ESPDn0cq^}4dh;9I8xE~;i~d<R0+cmEww8t%Qg*j{`mC018QP6H6z
zvaqY719;c(OrVC~Z_P>EfazX``_>3_hy)<=&x~UtxAOc$82Hs6uLSDu|36>hvjf_S
z3f-e5dD206@d?YU8@;*jr6Xiin)Nfg=Q5OX==3uc#!3oY0u}AcmWTah)s`QanCA$f
z9$$nADe9z>fbb+B9Ss=(B*19|VCnea|MzLZ=D8%;AP0dBIwk}F@Cl(Ha*pZ09<-u@
zeIvdbV{XT`?BMJUQMPdQS(cKb*4c|#+Kd=oXG}F;5`wV5R_^o78CH<3661Aii18~h
z3r?E1)jV4NB;Tf<XMZ{cUskLPb7^AC<<2Pe2@2mY5*P6<W^rir_SHpe<&PWYRdD9q
ztFCK+<yM?=1q_EvlOf9Y`&ihu1>Gy2CBou6JNlXnkr!(_LLP9xsz;S*NsdSgvrreJ
zH$lz`6*fDoiVpKg%1#NN@U|wqdETwv<*0vk<40`dsn#1xYf9LZOT4leUv$ZLJCFr-
z3ulgK4%S#j#0_eHH7%l41go3c@^<W|IQayIrKj!eR?*}-k5J2buTQC^XfY9RWzTP|
ztH5(I^*VhUMd-B5Sp#)0Z!^|~;V@3S1WvV_ZMxg$l6j}4-<kr}#;bEix;+ti%Cl4R
zlbtDP3nAP~uk+fg1@6kplaYKgevX@i`;(*PrK~q&u*~f6A{yx|NPZ6^q`H(kvEeB+
zQ>VPfr?$pwL<;DLJy^fN?w0=y7$3P6Iijz-=$G7pWhs>0e&|n;7qb3Xr<p>$`=lb@
z?r+BlYYk%-E6$wRxw#z;zl3!>BAa4FnL(FggwceG>AMUTd#DuMtv*OaAaNx(uL{g*
zF1x!X*zItnuiFxZi+&flQt_nNpzr%}#b%ux(Suwm!hh+TDrd}Ro*Sw;0j#-KJRG4F
z`s`WU&7mpcM|()dbQo7k2IF>VAQB!G08G{rB1cZvuuo=Y4J1aUG(8emdiD*^PglOk
z-;Ad`xYC~;=rDuoQjx=(%{%>#Ow5z=f619eC9fda-)-*BuO10tbg@(=SD71w0)YHs
z;F1e<>e?^%$@7q;3a;)2Ko!=bhGOqDu3vnqP5njatgr9=gt3&wJB29Mic;7tUqjs$
z{-~i)#O*hs*sW71vLo_IKHsZ4Aadz>lzB!Ye`pnmwIxxyH?4|Lc_UuOVx%jJMm=Wn
zOSFzBaiXW4NjxLcRX-)5{i8&Nxx2zD58=>GlrbRz>h|&`u=Rjy!Zd^o%~*3;)mrDb
z8a&*hQhYP7)qD!)Nj7P7>eP7d^7a?CjlRB;3AKrTsNLvRu{DU2TNhfbkt_|6G&&2L
zfP^c3@!Cn@zZUGsnl>od*p{+}nUx)AQvZYa0nQAUb*ls{s^M>FJ<}!iCf8z}Z(h-A
z@waxer1&KRRYFn4f`c6KUtJdBL2aa$0Iu_Zr9+(hjw^n5;z1Yw{dH~^5WH!mGD6$z
zU#DTGJ;x0iHjaf+-qJ{E-CqKe#{0!}ek?Dn<?eF^8X^TX@89zSRTGq9IDF)HS29cM
z8n<3I6W+3AHJWYC9dSvF*t}Mu6yGMviH79FJL*L(xDutH5%0r2%8kT)t1hvA%I`9Q
z8pdhm6?*mka`3a_Z`pZPbW`8x9$xe7B(_<rsPFA?y=Jk$`Hm1`H1kKsv)`(7?vV7=
z=NyqpuM5W{{BS)RU~hmYs(}_u!uR34ScmFc6vZF<mdPp2c|xfgSLl06hbxiF^)J<Z
zx^;Yixv_`_1k;%j@7MAJAJ8x}?*jD;wwf-U6qH9hi8f9{J4*GVzb%tf0`ODGqH`ix
z7wCK~8tUR4VaKsxDV{w1F>au(yVCz;H#XPw^GNK~d2jCYHSas{tnUUcUhQYoRAe${
zfJ@6%{(7(Jg)%$s5PZOdp<nsu+<1Z&TH+Ea9(ohs=%vjt*PIRh2?yS^KF&&Oz|?cM
zF*bXBz~(TP&As!N{kXsJEdNFeBjXv7I0Du@K&z}u#p<||Ysx9CSts+;hj3oI`Bswu
zGq?Xg#RslxjjY?r{S)2<fTM{dfW577Pf{*^z5;-e--H@l3S-F6UG*Gsq5*3D=&%e?
z*)GziRRFlhg4{Ydj(>>;ft4Eja`S@0#RqNkRRB9Zt_MB|-~)wlab16g0pa4k1QWyS
zFEhZg7lj{#utqQ_EqpwGY6*s|Q`~W4DJhT-_p)XZGB8jzzc^|i1bNz{e9svi492zH
ufA#;r-37}3T~Pd=r1QUup8t;?Jone7U1!~85x)-JfV_-~bcv+#^Zx<(_i~p2

diff --git a/01-exploratorio_files/figure-html/unnamed-chunk-30-1.png b/01-exploratorio_files/figure-html/unnamed-chunk-30-1.png
index 55226ab6fc61d7ba6e87f855d26285bea06b550d..be3452f24a22c24c23cf6b8170b66ad684241d67 100644
GIT binary patch
literal 151685
zcmce8c{rDC+pTCOqC%NUO36$pQ-eyzLgr91lgK=#QY0jqLkJ0(=ea^MC7H)4v&{3f
z*H7<$hwu2lWAA_VzK`QM9*>8g`@XO9I?r{kwa&})n%tE=yASOqA|l#zRZ>EMh-lk6
zA|eu_oy7RgC7P}8@ULC>B-QQ{5ivX^{IkXP_mmzH(P5&i66cldUXOLzt7^1_h)=3`
zbg^GBtr*jxexzm=_+#Z^(77&t>(L*c!&b*->4h7`?ykHYu1N`5<u!A$kDIB<lnrY(
z8nMu9`Jh-CMQ&kQNbEy$OVDdiE0xz4I<p5`)n9yYaeZBR=+bRUFQT%AnV{T-&O{YK
z!Srs&I(v$GLw+jU8R23#m78Js`y~;P4EZjRzh6HfIY&(K_sd?Q7o{YBzZ4>3nf%Ak
zk`w(u`O#AQViKZY-k64WO~#%c^@9buxiz*zsi~<QpPHJQs&-MF_$ht)%9TCaNhp8L
z6Pd)s#L!pf<>l4d9!~8w<UhpRH)WSAS{iye#=^ql=giU!!%lW~ha#<%mVJJrcPUmM
zZmHDerd47Kkcz#%5l0-Xo3o?3k*LNfOMSk{t%D>9Uu3C^_FON$@VQv<o?~gS5?j;I
z(2&!WEu?irLnl^7QY**Zwle;G2c=sMyY2Y<<vk+F-YtK>q$i@X`}-BqjtBquKiVtO
zv*iJau+5mNiV6*-S$n2Ybs)#S%))y=KhkJBotH={Szqq_tpBH0u;<6HRhXc8ccJ}^
zOr4^JhU5C`5|3?OL4l*A<CRO7V&A@%kdUbPbu&0SJ6kbcj<R8)WWzb?n$XNXLQoPt
zh`C5ik<mL)8<wb&LQY2Zx`%uF4#E}w`t`?;9|snbPapEW9OEa+a^uDg{77AW{lesA
za6hf7pf3#}(20oR_Usa=A%0YLO;dA~Vq=m*gwdSw)4RmP0*n491qCaKDaFPuf?vOU
z*|}>M&svp5=j7x)TU%Q^SEYrltSoOxQflhGvVqC_Hc^`&r$c?j?P*&>Tbr7;cBG#l
zPmf!TduwZJQ=*E7`NYHo7aQBi`1ttf=*lmD`T0Lx!V77eue$WMJF^Y9qKk`*mX?-;
z)zb8DKR>@Eas#n#+qNB`rry478|8IHMSn?_&<T&fe`g9Y&G&m?Ap8EH9P^=r2Tz_n
z8B(1iaLAi2z?iegLgCM!Klt4->l-(kR=58Rd9PhhzI|esym~buC8fl5!^PRzSzTS-
z$|`fbk}cpXmf@L1bXnPX=IQx)A5YJ!uU~JkwGMRK_3!cNpP%Sq+PCj0mFN5S?|)g^
z<0kLky^FB4wmxkV*I=lcyfWUAlcHVH-N(Z0_bxg*IwwaZreWvShX+n6o3>|G_4K&5
zM|P=WF(oA>wY9Z1G{Te<t?%FG>*;7{X!!Ai1wpKkAQwL~H8nLeQ~l+Oq>PM}W$3C~
zs9e0T{q!BYZ{JGC<>7cm1qGP`)4Z42n(q!?-nscIY#wf9)W725vMxMP7}GK;zVyrg
z(GH5U!W~v7Sy@>J4jc%fC=9B)gQbd#(?=lE37FE(yvxtemlt3AVaP9c`palbN>E^6
z==z;A#)^uKZX`S6X20E1*tKgH-q7|(Q7I`{wr6I)gg(4`H#9%qF~FflhDE%7{rV1a
z`retPeXo^MS|Y`qL-i&c=f`;JyW?$-any)h*u1+`CYDKV;l~$#R(R8;+$=aAAcbf*
zs(v;a7h8uY{r>$s0`>Ll*Nh@|-@bi&(=wW(RWvd>8WtLAY@uXsZmy;l^7QEyb@kAa
z5?3cD5z60#gM=rH-Mp~KDTUL!Db-$mHa9hWRdvVB&8@Ai?UZuTwZtYoJO>8{eWmkk
z!<8#O^daR`%hf?#23$4!7N0%a9Z0vbw4|r6zuWx+qc3Yv_2}qmtb);%D_898m$tUl
z>Tth4bN@0sbw<637T34=Q)i~!LvKC)8+s~|y|<c^G;%(qrCn$C>nybAJ9bR3&tjOq
z@@BD9h=+)X2wsxGS4%!#!X+0eCMqK0w6;9=>(?DG&YtP%X{0~~VVm57f`Ys}dq>B8
zloLG7`Za#tydmtoyp|j5YjHf|r&K<O*#7B6SUr96<PNXexpSp5TxGh%ZV!sMC#Vb*
z``j=5B;_jX{qyH;dqjA+^X$;|5SioDUX{|mXMBVm=S>|Qg(=6|vrO>MK!It=%YEAK
zLL~l1%>0vs$=oK3PA=!qZ}F{c9GxpKFW)DqpCCqQ)|y)O>OSwSTemc_O^-0UU%0NK
zVqkH&(n8qw55GK@+2{VL2&d_*UI&M9jk#Zl3Le?T+B-7xRSP+O{-X}}J^I1^e*b^~
zJ<7tw#8Y&o;#y>vW2QRu?&DdGMB`}&2L?hX!si1rx&3!;Zhal{!^jx>4<4*ixSgij
z`RyC``RWfJIGHW6J0@fVuFA_V%+B(k=A$`u=ny@<h<7XEC5d8oZcdLYjwi;2IF9pf
zBdJ)V>#r|<OzV?R<wd3*Td(X2?BP4)Jy;)UZfwjP{q`-(Xu4Bl@k{DBh9jG6ShI&L
z>Zw{_P*6*I`&qp6;qS4zxk7=W>i*TgC4Nmzn2#h0433S_)6%xKwHbB&9vvJMEGn&_
z|1Rz|(_Kp_F}Dl-h)uVMxL-IDAa#lPy0mn)U3|XPu!4pLoul){x^t{TmHwV!nH4Fy
zn)!b`=v_Rh+Q9H|o^B<%GUeLz?`rLp_Ld6MitlobnieL+Zf`O&{7>^4jPOOiH)0an
zV|ql8Io;0Xh@a&9#6)3XVLjfqp`mvi@x~@5{Zn>TRaHp9>dDz}-~OWOS$tyu$Vj`2
zNA~E2d+dk22a<ETnWGmM7uQx6>TQJx&&<6IVLKliAK%i_VrymfqopN2KK{z(%X0lJ
zNyVXBDLJio`-!ca$9}Yl#XY59N~bw;<Vc)%&fTtj%ip^zEhIh<&Ro23;qirA=gyt8
zv$HcXId$r=fp=y9f&KfF)H3u94N)`_)6#hO`0h)2b$i<$OP1yOn*xIABzwy^ckkZq
zva!-bNA{$zuMY`IU2yMi_eYN&k&uw|hr8`ripYJn_OPkFUCQ^qPWB~s>Y!Jz_Flhs
z?OV24$8BBR+DM7c0$cM!#c{+^ZRlx<SCNsfh<TrWPR-nVlfCB^E!NxeUpD&e-CW;+
z?ax}>Zf@PW)wDI$&4Y!i${^BNzw6wI{ZtKh{h2@<iAjGvb~&MvnVWM_mRDAWJ?qYR
z=~7o$=kD%~#7Et~Q#F1exb?z<#D04E*8u^SB&a;!rl)`Zp_BdLSLV^Jgun725Ab13
z4?COPi2c;5jEDyejEr&;Sd#qw_T|>)<>jA0Z!zyCCCzYsn{hR<DKzj~bzPl+nF?Eg
z1k2u?g@uLe<UEJGMMXukUppZ&k4@4E-X+_!C(S#<`k2A7f8|ddF5DQZnZ(<c(c~RE
z@o{n6w{G2g9jUQ=w*h{=B}JQ++3j5EWncL!k@*ACvERivmg;>OMVX^fc*E+I4Ay?_
zc7OQ0>bOJEdSAI0yJ&>fNnu;tg~CD+=I_H+rn$nky;AC$@6<eBct}RQ=~)k)AB*~-
zBqbjo9~o!c6Z`XL6~*62-gS#yr1iR$m6f`Bdobxo;WKA`|Nb2==B%EqT(Ysgf~@fK
z=T8yFiLysKtgWq+`r5I4aZ2evH%uCbt!is)&!0b!$|)u$mNBPR<0--HeOY(ezdskh
zw1a}tqx^Q@fv1lj{~8<&W1?bW(!Oy+kDFHRwNRQxf5n?OhqKhzh8kkNewE78{4+B0
zzH#`Kl%BM-w5e&5QEyRJW@ct+Xy}IzAL8QTYHDhNgM*DUt9ZD6j7iw73at%Yxg*va
z9pxsr?6E%HGh|y56CWATJoh9oMu`m-=_HkBtJLo9?rt6)9yK*J+-97B>)NRHM0e5X
z*x2ewVoGR_3UA2RcutRgitB1WQy9f>)R>Jarj1+thg6c)v&?(M8s~j{eB$DcClElX
z_1SP=U*E_`?aP-h-@eUUAWg`2rSw#GsVPYQnORv0O~#`)5o-mu6Pzbc^3al<-n`;&
z{)H}pYZ2caL)a?T#?4)i5>hAdh0FFd-$Y=+>};FDB4hvr1tTNl09}7yUwV2vzzIVA
zDU*v~dtLt&5~%L&+j8M&FG(;tO`eyLk-2s2=hlh4TD5;-gS7f)mK2L#>T11H^O*D&
z9O_v&8+*ef_O{4zcsyr#^5-tGLp3Sc*#g-_Z;iZUj{8^L!Ofv2+`7TxaY8@<(5(Qm
zgjEI#h^rVNIR`xP^5x5#jXR$Nl|N`!W#6M$QBty6nKzpuQa#?&-OYI5fOT)_7RA`h
zgsr3VqV$n7hekF#7uUq<(v0q(+NvtGJWD-2y;w;WBzoRM-iQ&b#b|SKn9i-4sn6x*
zW<AA4`T6PzN)8SVoVb;WpB-NjQ5_w^Y;0_}J7(tFH3CaZOMqno0RajLR>)Fq>3YCd
zXU?2q53j4OH8V7%Wnc)OeFTJTU|@h7l#KoU?AbHx(dLHzDZlJ!j4xBJuT9J@kM1is
z<Nr?KaxNiZR*2&9w%FukZ_yYFOG_zfX;M<s`<9jtNI4fTKGI4lahQ9bl7ceB7IN(8
z?KHc>t;HD`jUsLnLqj@TGL_0oN~vRA_4QnQe2ORbPuq#)X=-U}a|K8VpFR7vx*E{m
zw331hZ>0Buoq(+IHdV=42T_gF-m$ea4CH%<=Q52Cw59*(Y|pwgIQOe<{3l>aU*GK(
zLCb6ZndCFP=_}XP7fOO8Sr8B_^W%K-)l4FG_m*d|A#aI`mjr3Mgv>oFS)1?JL>bcs
zJ!>Blj&%$T51--XO?O(doS8e?<Gj#?El^ro`pNgH&gybYiSmj47osNJX|w?P1+&ns
zT=8Yh5?e~TmY`%hO?sCiHBKR6)?tl5g5gs%Yk-vJm$tTpIXhJSrD6e*2R2@8+3TmC
ztgrWW+`>`k<q@J9Rr@^IHet#eDvG)>;W2I}O-xPiRPVV9)FI-?8Vy{8nAq!gAqurb
zmrm`<GU|kk7tMPmwh)=i66g&jW``P%Hhw;sth_S$^CI&zFE3pUvh5_57Kj#XTk4JX
zdHMP2Bfh)cKZ->jAZ$3dBW^n$97J|j(GA{bI$IDH)&WqIybu>3@95z0vhDCvYpW_V
zFD+SDSJ!~S<lLOBwDeKtm<GdQuLHLH-jRbdO8}2cKPg-nKkt)QEeT&@aUw)8_fzzP
z`eO|-C;9)3W@ylJ%Ub5W_E=q8yLt0w^1@kR;T_wzzf?=Hv$En(6gLm;tfN}=@Yo6H
zFD@=Vb2B_VT;f+>AFr1EPPE5LYz6uGsOt3*q8;DA^LjDu!5`4!IZr7}6wH&c?bGq6
z>Dx#tq<dL|yHD=jiO>G+OSAJ%TdpKax?Hd42{tyVsB5W(;fw~;aj$zMnEjMgRBUZ*
zzTV+&yta9-!Gtz8IPFH_SQ9Pv_H4EStrS5sY&Yz+k>TNnVJnHx`gwo?RM(Z1lyLvr
z+DsiyEiG5{_w|No<lMcPXX(Bt87~5e;6fDl;lqbTL<)hfzt#ni6OtPJ{>_hHb+WWR
z7Jhd+P)sb-xnPatVFKl5fj6~y%9B<$v%bC_$1%{@YcIlBX@M9ueJgtEQKbca2U5Ju
z_`x()O$CJu5=guq*=Ane-lr#Cml)`tm%!&gh~06&cZ*KN!-ro?Yg0W}7b|}a8x0!j
ze0_Z32&K2+t1cbx)I)OvyPWRbOZD}optQEKN}Zb7tV6yYGL5}}PHl_!uCA{0)G`+@
zl2D!z7N()5KF-3TN|tQUHI2VOq=PP)sOE|G>O%#E<3<^G8P<NW-ly6lBWq_TL}`@O
zJ~CoFS~NE|XUfr-n#w&ho}_q_ojOeZfq^QklYs$i*e>(;;`Jv2q?oC8Zq-t5+V}jj
zuDpDGjW{JfyEOVSAygg((cH|JICG|7ptU$$Y(2Pw&60eE6M5_K;RhaHzkZc5Dd+iD
zEl<_2Jv4KZonvBrT<=sx+OA15$F9c4Pw5t3yC#zawXohY-4ZD++J$z047d3Y8BCE7
z2myBTU=j+V3>KEhU#o|^Y=;F)qO!9$x{4i6qy8zWO3K;Y&OL{0_+M+bsmLmJ+Q`U=
z>Uh3cN4DB?YhO>#TNyoPeD2%WkfZY-zb!gBb`CjnrZd8$ufIR>{d=<QpFH=m3f%3a
z-M{~>aw3zcgOZxs$)IY)4@#dxf~k>_r?)qH+^|Qd2(P$zFQ})|F$;^RPfv#@Z&7(A
z6olD{JbM;BIk`kmdOYaGi&V#jKQCWWfs0TiT)OSQ|NPOqb}iLm&e-cBx~dBoFCwL#
zlZc9n8W=7*cX;p4tx0TJN=mQ1yu6m&koJ)Qz;jS}{`y2A&Tlq2WrvJ`8WJxT&Kyl4
zws`pX@wn8~Z{$oggjG3$#6TqM;S)DBv_wuEcbt{g;P+P)jsRvqLqo$ahWyKmi=RJ#
zMxS@L@K~P;Y8sW@%&tipY!PQ?Q8n2I2GcV$sJm>ZPQ@w=Pfkvz>s3d`#Gp2)3c3K}
zT)g-RY|~chMgsZ3eqN-8Z|hUE=W6v4VPRo8ISWotNk~ZxM4g<SIUJm~05M#SIl|n@
zs3VI{n{yE1BQ{Y;c;)drJlw+Z!RZ905v<vtKVm&@r5~e}6V<lpICe7n{2*~fdndm7
zlcHp));y3)TWnk(b>UZ)Kid9_7cZuzr2&!Qwf>%2YBbqSSO#P+LHd(!B~DAcqaq<T
zd<ja+6C3Ljx6^_>a4WbqTy%#HxhywEhT8sWZza?wTsn0^m#+_jl+ms9IV~typW5cS
zvEb@e-uLU*FMMllsor({6Vv)$M#k<RKW5{_*NhI;{{2m&ftm{wZY7q#rLJD}RAp~A
z&f+t*S@_dcQE{1zE5&sw$n_S(FgkX-@vIhdumFJ50P6zW+~4N)iHJyljCUwqKd*3d
zY4MsdTe8NL;B~_LOg+Iq%eV-%)!nTnC@9Fp#Kh<;=&z4%g*iIQq$QN%5d}H9R+e!<
zgJIIUcfksLL=U#bzkU4u`w0$jEhoSAIql)djSvB|Xpf!C{Y<;&q&D+d4{p?TeQ7Y#
zb=`h`Id3omFCx??V`X7t-*tWcxbk_xO2E79md!h>u6LD7DH6CG-Q6M0Rd_CltFg-e
z*T`8WzfGVwraB}#nei?sC&+`Cn3z2~g9Dpe(Re#i>Vv`+&b2XvebyNp<K|XF-{`k*
zLhB6ksD;vc=4|Rn>x1@OTwDZFD@1`$<I5f1qfe6C{go5LbA^c6)zC7{FnsK^cqzH+
zG8<zNPaCp2EdOZRwwTeD`ty>coiFo6b);IK^$;}!N!+ENMJ-lHP?}-b#(qtpn^94n
z(`DCOsjrU@mEY5+Ps?rO<>i@qf4d0?kPxXJ?(A`mH0Y4u;oGxT*Ih<fzXxmVQH*xx
z<`ZsiTe8&6(!#CZhvj#gnd*}w8O)Y;JZdoHrzMj>6#V*?`Izy78%OJS+29Q83CFxI
zxi^c9b){RTEe#&Ib#{&&s()B+L61%{^DTe_Wu*m(N#A31rO&idSeY4ppMtVQ;IGne
zGf`Zt;)&HT+;wd7LexDSol*xfl8!!f;IaDIDA3_9x!ht!TH@T?=Oj2eITan=5=(C>
z5Jfwssi`?29GLgLNzX(&S(#fzrmSrKx&BW){vq!t9)Zo{Te7bMJ{D;m-djT;nN$5{
zoWz*^;+2Sqh-?a%C&w-%D`(iZp+04}II>4oUK%q|;SAC+RXMSrzzc}S5@OQ5jN84!
zp{T2GGs0KYFYI!Xkl0FzlBr>8ZfJNo8XN!QX}r)AA3vFyRShZ{z7xGjhzcTM!q3{+
z2mk@8gwaGJzKl&xt*ouPGL0Hvg>E5Y-!E*_+HSe#pq^b55k>fo2MiohTV3Wxn$Rii
zcJC7`S=w{TEw0J9$6+)X(AJwykkfHBrn!P}U*)7+FA@`FIka8c>i+r5%r+9U*qM_7
zDq1J8;Sv%&Jv{b$N}LV9TmI}o&)Of`_y#f}o%0^vv61Vq)+aW?-?shfn)NoXxq6Q9
z_jPEb-|dL-xNzZu3!xoqP5boeQ-61nBe7d=LVUz>%EmI6m(;*!_?*yk^j`nTblmtw
zm^8TRW;w!6y>~~1U5T8$GV}RsB$kNx54PDjE-o)~mRo;ZpWNXve)^?kab`nTxXUDo
z!<x1Fv{z?vpr1x`&JLNK)nCf&T6rze`x0aTCAPWg>5sgYI%W7M(YwB}rynx{BnjKS
z*iAazcWHU`{>TKYxVY5Ss1#>!V&VZ|Uc&7MgZ0=#L`^9#@33s-uspovv&jYP_k0gU
zrQE!F-~4~nME>6>E&ukZ_aY|nRB>aiX}YMy5{RofU}~9f(;wYQwVkH6Q5LN)k2X4{
za1}51;ad{3uIu$17B*VndP2<1%r;BqgzLr%HeT^3H9NBH@lPqd|NQwiD%`UfQ8GIm
zdbgCibrsqxGW*ro?oUlU=sjDu1AtHE+ym}o<>H)mdgZ&9<6WZ*YzH@&;6cK9s88fI
zI^6qdXz08vmdLkKvn9OBz8mYjGBk4I`{;&kf12*Amlw>7F6G+e-b*bAQcPv@riLI6
z{pKM*zeJc|qc@%IMAPVSym)NnnyK7H0^SMU0u8J8-RkNpcE86E7`3BEkD_s67ZG3F
za#~RS`#`PW@bcUlPeSR_iNiWSh(TN2k$0b!();=ISFc_{klF_h2BJZ6jOy+7_My+7
zRbhCflnvtzL;Mnve%Zqh@5qR_)9}M!bWjVxpT)<<Ub=Fn!)0t}$PfZUxUengb*&Vd
zom;`20(1by_A~H`h^)35L}q1YlarI@=jMWqdK(w_;buW#Wq*jEW52hckWWX6D_HH-
zt^!-|7%M9)Gt<+!KUuT0XU~Gmc#F)%)--H|<_5~h!Gi}qJw4^*0)3T{i8njtm({<G
zv(yjo#YZFmN?f|sHhtClY*&>(D;f$rk?ux$@$i=~Zx`E2`3b!&S@RaEiwit(;78LX
zN=h4Y0vG)Eo#<BA*Sije+oF?WjdZd^YfI<rfKDqxF&=V~tj03#Y-MG1WJCaTn)Rsi
z=tX=-K2j`MiLEWtp{N){Hm<C!x#9T8`Ptb_F+68Qeg$y;rAwD`%zFfSzDy6!&(FVm
z_l{%-`Q(yQ;|FjY(6c~+G*W{F@{|12nU`TR-v036Lr?JptrT(RW&Y~!g|Rl>h$8vZ
zR6BGX^|r?Njg0!0V*8eUB=lbNVb@M{=34Zl+`m;At_>3e3no5%1OZU#%VcnJ>ivyG
zz%fwoST*!krw_@hsU4(L8=N$#dwrcP03Z?>6`X5MPEKsBo}u9ok{YYrYrCm(8jj_k
z(BG=RehuW%2=w<qx_>v4UfVoDHQ@W7RRgGSYP#Tk25JswsgL*dy^M%J8Tr%K#~dI9
zy!_<J6VSL2IsN_pD_7obfAq5^gpUBb03)Fk)GWxIq^gX$CKHvWc{gX1vDM#$zUrPG
zC=j&CuOwMO7RX3Rg<AF6g85TV9+<KNvOBF;0_h8w-gT*38_OcTGM4e<$BzdOh-PB2
zcr8h4{*jUOb&;W=`7UeAd<Jz^wzlF<i>V0-RksZt?CtIC>=yazMf$-K6c@dJ|I7nY
z&3~sW$Qs>;q7WvsQETqPCnnbYL+JGBxtx-<aclts`Mpb|(!!uV!pz*9HTq_OP4dlx
z=IF9l%oZZr>-B=3rRNID2?dbbgx=oeC=n4Q<<!#Z!-vF-zVR)iT$;Ju7X1Jifw}qy
z1{!aRacQrOsMw1{l<DZ0)qv9pNw=-2t~LQ1XZm7kPh!?rY@H&<oJ;d}xoL%bpqfFg
z<E;<z{L<7EQ2SFQ*qQ;)_`(Ca638ec%8`8QQ3XZCwbj*%)!_fsBf{iMBDGk2ScN;9
z1LfrI%vllYgU*%<6e2|I+ktn1pFw0YFfiCobc3Win4;(CSWu>OIGQ!W?xTY_IHr5|
zCO<u;$jffb_}^XtY(91yXp`y-<15R{aj~&Y&CTe>!F5LLf||prUECQh8d)!H`Rj{a
zni|)W%I2zmp%2;F&C#UJ4~s2g8tfng186c!Nl6{LuU@8f^cI*-h+$wlGiDxc4C+p3
zEf33IjSnmQJwLvm-gM&uZkkAcx9B8Ms^qWVztf~F5iGBD)a(en38)xiH=Ua8NxPk0
zT}#U6p|kOHgdNY(m{sv@5xiSx$Pf7wylH!Tdxk-Mc4nq3zoe9urM303Xi)XGJ<d7E
zt_1}eDJ}9&?b1P<5bx7k7AATG4|IoJWs&%7(aE&2SlJjaPjldati#788p`urB_j5G
z1_ll@vMW~zFO-{<+6jX512Yj(QQNUr@C<Jo43P@w=TcmqYxD`!2{PSiN{Lum8`<_p
zf3;DwgIqrakBwO7iLkJI&g_D0QRpwt@)B)oU7cKri~y44d0Nm4^&&Ge78YrmoM@JD
z9b*=YWTK=@Gj4kS>eYU)Bg-<-WPB^7<>c-f8-F*>I(F<>&tiq3KJ6~bgWi|J<*rCT
zM=@zm3i%%GJk!6iCjS0?^z^i6`|5~FrhcvFHUc{5J_;D2Go=$V1Tw+Ar#L($1nJnQ
zOW)3}wIU<UZ{DBRJJHs{f~QBU`|!*^IFo)MX0WXPNt!e{{Bf3%srmWoX-7wghK9yO
zS3$fI+n()@RN1M$${}jXi_R%&YQ8Q%BEp#R=8X=d4)k%T<lf%iAt52nOY=%c;e~*P
zn~M&P(ixI4h;z*$((;1*@8w>zeO^}6zi1tuod^u~U_{tVDj_oBc7fhtXOkCO`c*Jd
zY{E?o;;d4Wy1TwUB|e^&pMTb|XQlg9od`E~ss3Ay7vfrOHBhX-`+Jo4p&@66bORng
zto+V25{|e1{>XZHu*otqcg$Pb+U(KGH8(edZb9%AuTEEkK^z!|+?NLmgZXQHy|}~d
ziAoC^PE8@Z$u~>Ue;U151O;<Ge2`-cm|s|E`SIh9o}TQo-(W~QE80k_$-`??z7?lb
z>?oKH-PAmB#4X2wVr@n|W|81#5ejtgvL`0uzJz6Gw+)bFiFBTC3*H0K;u@`Y1y8*W
z|DiKJx@+GDS|_bd&~QnhEb@2LI^N&2XOG0Mu`x|m)!-K|bQ+}1*Ww^N;WDjqxO`&2
z{+&BcD;+)E83q)fd({0T?^wJtD$DEb)d|!^$bp^-k@z^dv<l`vwWa05&9FDkr61Gp
zGBAjXIW2sc{drN%HmCUgvu-<*HeteT(gT;Iwn7)fXHyY`ngqwrJHXo#{Jgv$164EL
zG#L}}b@5aMU6O6P=A}#S*6PCi{GY#;qhf|EnEO6$^53^}YjgM#nj}$DZrjzxDMCtQ
za#@U;OSpaeb}Om(Wxg`uyKU+2MJ;FoP+j5}Jg-6_r^`%HPE4k$Cu}>S6$1A1+7nsd
zQ{*T}iS}wVsuvampcIMOsY)h|*_-$2ZP!=rn-gFA`Q5U3qsh53+bCaq-%j9Rc+=ya
zeCtGxv@}*y(5cb9g&@1_rkJI}YM*XTbszeC+X5Dk@&p;{0a?At%3pu}L<9w0L0?g~
zn($faC{K>#!rf+F?VC4imERMbH4n~|M&^?9l@foppdYogLKDPAMJ|2i=FP6&-Zw!(
zLFM0vW*6sGi=CE`pJJ}~%5=yk#?9CjzIpQo+5%b~5C#cKpn}o1V(*BrFZTQRkTXL9
zkd>D=7zdf#-qF!E3PQk75_}Z$EOZ2jgQd!0(o#~q+Bb?bGhuhQj2C?AlG@|S{;5NB
zbl#WyiyRkJx7a5rA^WLAv6$!<RZ&rSRQ9HE7+}Q7$*Gm0t{spsLnuH>@5+_RxzQFA
z6BEx95RJi9h%owkK%CdrJxcle*RNoec<o}RJSZ6S^kWkfz8)SrdNc8dJeT4@4}CM_
z2gQ6jCat7ITvRlCTWaaB)!F&Xa(pxJCspd(A*-`K{heK1gTi}N@3ku4WMyRq8)SCp
z&gZ5k>%#P`)H_(d>|9^;^qpqh;z*b95>0^)&(jp+=7t6jhUfr2eWAnL2n;vd<~o-D
zX)hAcd(ilQYc6VT$lp5Nnii9sT<bKSdsS6c6~S`-dIPoso>JQ3Afs<kNJt)BH(nP(
zI7PgAr35Ju57E_SR(uM|5W2#2p7*)Axk*V$@87d|O57IXj;QQMIjs2nId$XSKeeb^
znwp(Y_R!sh9fX?}&<Iy;Jv9|oI+QdGsj@JY49DTvphc&3rK4#=KjUP>uwHH7r;u5$
zN=c<p?XD>Rr}seA=14Tk?+!VEed~a`rvoTjGYsVQ^^ZFyHyCz1jB+ruoj8%|@`8}X
zbtoW#f3Tpp#mYfN%zneqhZ+Ky4pp%u$9%_*9Z1Kv-?f`E?%pF@->RxSy;CBj2_C8j
zNtTz;*Py2k4i5U}<t`BO_|h0l5fEZfkYM*kfAsu8S`Nb*^<pOG<^;NcW&dt)HTUn|
zZ+mA=Dkmogare2GS0X$#LqoKTj5hc0YZN=#L(Gzr`ud6(5Zo94H!10rR4mXY<X|r!
zpSr55TP9y%nGoRTf9~TGD%A&emgCHsrRHP}D$hIj?p;f1A>B$GbuAe51@fA1i!E;3
z#>NH-MTsr>-8)q<V*JcJ_=IeTTqPT86B1T`y$)VCX6Y{gxZ%AXH$>X3ObQ@+9^*uZ
z*TjCPa4JrSxV7qcb&#6bJ+Vk=Tq7@D>}6ttqstE#n}GrSKq&|<Xd%7&IljiWj6&@2
z_yRHR_FPg3Fgd(Ff(T`8=eZj*;~Cu~_`NbJmM5D!FOWa%9Rek1Sy<k}J9hj`;Z4Wh
z%7Sa8TQ|v22A)VOc0jbEg>K!9-#MmjycWj`_d+58dg>-&+P@!+S=26a`^nxi$B80?
z<$a=))YOYPWv}of^H!$CByQWTumw>0Wn}PyWrE$PsIZW`h?xCd(+EM0PlxA^(JojK
zIDy@ff{aY$Uhf0Qv`gK|8aXbY{(`u)n}@9i>%txJ9}Kr|k1Lf|RD94XYIksifB>-*
zC`?<s2W?V10AZf*o>Ch)=1>CJON#7g^x;z?S<AP!v1uNiTU%RGNZ>zwSf@}QKBTw`
z>VnSBP8UeW88kL~Hnpj(H->@)vO<%n+tSe>u+87@DR!oYJO(t6+wnO-F*-ba>-XnC
zALKx|UZ8+88yg#^Yjx7;MC~FaMay_J8i^6ECU`s)_tMf1jgCr7U>WxD>-G<VvNkI6
zs_X~G$8{@-kc4XvvP?=VT-i~y%Ip5rKc01I#lJv!y!{fBi{yNyeA)Ylq+1{E$}g1R
zd&VK<&av&U5<QkRd}~MP1;lw}{u|K43`0@CQyN*u2SoqqGATkW3j4w5Pvx8xTr!@~
z6xU9$R0i%Ri_&&*SrA@s@Rdwl>)}x6fcg={+0dM9H9+Pasb2pYAeuQl>7yNJ{gUZ`
zl)ZjS5*GVn71WfKZQ!5Dw#)MMCBO`2W$Mvf0q}MkOOYE_C4fM(7*jZ@JYfqsFY%O|
z0pNs<lT+ZpLbD;isag3ye0a;H&~;G@ye}W!yHhzyH4Qw-)WC?m{C`^TvS`EqX9Hel
z?Ebv}V!x|ROgXt}6-+O$@c4o<wEVH%GD@HJh!8!vwLNqKn$XHT{$WYEvP1!go96ZF
z$kH)Wmm4xV#ZH_stSN$UJ}AC%My=M(;09lW{Xht;3ee6`-;$N92WMa@TB%4qA<fe?
zU-NJL!ojdS25RcSurQM;*)N7!?MPtobp};88J~819P00{TU2x8kvXdVgKea{R;gx~
zHqMJLRpmOS+>DV88=9%FGQPKBI<>H%lAwhAg7zR&&i5cZAb>Q_^AeMjku5xuzkCJV
z93rEvsF<Nw9f;>?JvLPNLAyjtUEO%ue~rz|*!b+d-g8z~v+Tv!!3hEZl<6Sq_|`rS
z{I~YNV7n#JA}!bDS);(BiMEVN5y!{?QsJ4MW#>hwKJP<k1zTcCNeOzN>umY(-YhLH
z!T|#>hiPYSL;e;zaq*%*w)g<xAFSqle3{M*Nfuaf5)y{N&u6Sx1S_mSlMX*?{Y<Y#
z23r&tAO-45_T4VyZg}igek3RkjGj#A>&d%s0JS9Ui1_hE0feaPmlvg6#{vRo8zbeu
z$Hls9xrE5QPF36qU{8=|t)K>h?J>~Tuc)l-P$mAVq}Ekcb@IrOTX3hMTT)2KjEXuB
z39iTi#INA#)7FQs|M%v^AN7R({r_N2ba0H*tNCk9oHuFx9}NSn@BGcVY3p8}IZEx-
zHdy=R3->s0Nve^N5vnw53<aY|YH=#8elUmnc-0SrukD{5Zj`VBfeaTMG|44$V@uQh
z&U$*kgLjD}nqQH)di84afwAM(b0arkDV&&y1Wijt1&s+-18nv^+drY$`_P}c-LTEf
zaRmjSztT73u~XCg8%RAMIN%Y57JSG%E-C4`hs>Emso1mk2cD}Z+gMn@yGKDmfwss`
zDi*>V8X2^X_$?S$;<zWxum-86N8GffT;}=ua;w7Desj(GN4{v2&sTdD7zi&T=y4B8
zLSJ;7l{Fry5QP&L9LNTPX%km{`c*yIDJXma;L=AFO%LWhfBqazDE=aRGNNbB*dHo7
zTUgl|sp~V2mya9AY6^<pRn4>fjVHR>Q!L8Q&*EQoJh(~xMw~+TK)8;nRN=J~PCwmg
zYJy_E)YXYO%)XO3zR#<?f9>#q(4Vok3{(w?Op_MQ8FGag(gK3xXXfGa=VYU#!mx*4
zecU@lMMZ^{89=nj*xJZw7%rJXVdgNFW5;eND~mZTUOVouR^xuFYXa>H!D8fN(Q$JB
zGY=0BC|=F8-jp;nFdf1I`9_KD7}Zq}3$yJ(0Rdpi7=3eWC$tp*fX#d9?>`e%vKm`b
zvH`3iEi0?nBZ>QB3wY*H3=t5)8rBV2`Cr_O*D%Z9cwgk+WmYvb5pU9BYFuRj*4F0P
z!^(bV(E6@uBasxCO6Xd;x^5~f&#$jcl*~12fDg-cUdgt0RngUrkzhU&jf^01`Euw4
zOl=+MdZd)N`~3W~E0@+`=T;}9*4Z}GrcvMLNHAg0#9dF>{>ZIg2(Te3iQTK51zl|0
zWc$=@Za6dR!gbrez$OcLRJ#zWt?IQaO))=}&vP&JQ?N&Al?I5HB+cI+X(%awj*rXg
z2Bhn1#N90R0gxOV)Ns;8YxSh)g$BpQE!`CYb3_L@iq&<ACov^uLZxK+2BWV5Yz4=-
znZ%qVC9pB5-DeS~Ey)@|p`jy#g9hy$o2k;BLUdC4@<oz^WZ42AlCum>yaMeS8hVzS
z`>@ON8*gEa1vkk;1z+!yM$dBK$tu?9e|S^r?mL({I4pyGgue?7${7EK06Tk4bmeo8
z)^kM6p<OQ<;^a8~w=;FBJ<Rrp2I0Z~&61i=l3{%-p5X}lA%p{L%a9$<pi$=1TCKA!
z=If!9_3@70==0(V4GXI^u&utsJ5P4IeU}rc%2&larlv4e)k(RDA&oYhAHqhYr(cC9
zW`5%|CCWV|^tI-!T}i^FQ=9v268nrd?nl=V|27VvjbnRvwtkUeekm2p>^D>wE{uFf
zUx~tot?jx{(6fh>RM%T=$15aefO6}N!i<djIyyQA2JveS))nlh|C(T#QYdvA?DgP2
zxWN{1Axfvk?4flk=gs^d&CL~B8Q<5EN*X?|epF`?8z(h!e4H9=+Fh8JkZ?6Y=~;On
z?8K?o?bNw%-@YXX5uigX^E#C#!o5{tf2X_8L8kHiOuYU>b27RPJU25u)4Z~oX?yqY
zpPl^qX?@*g-atV?!QLL_`4lJThvej&S;jXY%u7ms3HA@59?x+rJ9qKoul|cC!hm{w
zD0Z#B<cW=~BPJQAu55)Iiz_`N<G1BfC6g=i3vyh1q|0*AgtPByof7uSmD;a1z;Nmm
zGT|I?A5zcguyLAQ-zD=eV@<keP#>Ss|6s#yD@twj*h{?+BnOn2*W#`b9-LZ*xe&l<
zXo8S_sgR05$*~2zkBu$om(Y9CKR9gBvXGQxu)&J<HruoVvI~jE<L^B^8zBY}CpkF-
ztA7_26`AsiN7kdKRK>GONj)Q|w^cOtIc1aFMKk1Ub16#N^7L1n#A4$cb4p~NlKZvA
zL-n}FF4-G|6lH01oX+HAhrxaa7+w=cUcY?Ft0n&m3_1oO8e*=%`i|Q0*Ti6Y=FF)m
ze(&220_lndE5WUqH<)=R)(X%+1e{V4@;OSN*QltN_wOblCjR*86G$vsMJ<j=yF&EK
zr%jqszERds3kc}z>&qE62zPLe%`!i}bo}Vi7#K6-;|-A3X6nV6?%QW?bN7x;dH>${
z?p^&|tG&I4rL7EJmUdK4PNsTu?C~9xDgXSLUd$;M&kk(?^t@E;cUXP$r2j>ec@J8>
zHGlMMcY8sQgTuP5SJ}DKr<D{hnwQ7|(fW>(qQnSM1M)t(CMj(Doi$RX`(=H+^U4_I
zM^p{)kC!E|82hXYF`D$bvNA(Sn=>e5tJaDAP;4+x^W1hnUDKI?bG9Y}(#%}vfRSbw
zRk%8sslC7`c$@&3gI5V2i-Ds$XdDH*6Kwu;R*mQ&Lotauyz0*-rQxK<Sc6icYc}Ut
z=F!KOo>cY&fsD$#F7KlZ2nvFcSMFO(!8t;GxN!saqO|L?Nh<`R6I9U*L+CO1_R!dU
z`*y8yh(!bR6Z|NjKT9w}_Q8%t{{h7gmSZtYT~u?G)aW;}N0>kQD%GAJP_Ys*G4|G*
zE+L8R40tmkW~C^U+4cMBdP1Y(Yx_cy1^B&We)~YRVZB>r-0s&#YbhXTT^JRNjD0{p
zo5wAvwu)tAc=*pB{UTQoMsY8HnEn(sFc?G|^7!#qucgG!#YB=&YM<TY<aJ-aGBbNh
zD5$9!8FOpU*tAoA?Ayo4SQogPiGiUAt)tUZzC28{%)faN`^S`c6~fGbWv91y>xSoU
z_uGXZLqKu$3pH+7Z1Qm=1HSC;7j9j?TtN^QL>=e%`Y9(K@RW$VnYpug&*jU86WtaW
zwlf&?aXtz~A(aHHd)47AxH11s)P(hwlJVJf^l4*3o#@pehS}9VdHQssr$k-?vO-*_
zLK1Tk_^}Tkvc(`YM0m5GK2XHRQ_Cv&QwZ}M7RyYZA~My@z~nmO6`Qr!MHuefN<8YB
z4&tWLDR^ix+qCT?c?nwFZ{PNI2qCrIQbu&c0>dyirKqc`3yKKOy9?}%vUv(*!UUys
z*5{=6Rz)b&&ToM@HFomRHn<n7wb!S;1-o?^dL^v>joY*C4-TfCzc3@@dy}PS`$XYI
z3Pzdkq?kwD&Mn1$2<Fx;REx(3z8yJh@vtH>aQ26y)%YRjynF5Jm|4>IqChzYbLP_q
z`xCzvFn4y{H}v;1qB;o)2}x4-35IdEe|gMw)CKHekWG3qjizajww8EGX6CE8quxf0
zE6-p;iriQ=$hK=5X-cd$c%QN3AV@%d0k?+HIaH`S;vTFwzt(kiiE?rIdRUzeA6_V0
zazD)>TK?SrmMlBxk0gnmUQfX4B^>T7_oAU$-%A$wVPSs0JKwt6kl(@4F^o$$?;c^x
ztuFQZiVMuW=A*9+N2+wTPj3?SFa*SwV4?PcdcrQJr(V}JdzGdtr%Zdp*rb}zVC(>)
z%jwb~U;Tc+muS58JUkHx4m1=wzHPj&bJ(oyLcO@Wz5Tq)44P=DjyBfTRd>>{LXd2q
zy1RpMNw<FTR>25z)(S_ier;$?jZC8(FapYEynLi`ViT);1R76GP<beS;89>ky%Yc=
zyZ)iXs{ScVcY!Np6m=jh9J1oBt;8WRHn7|_H!A~ntj+i}8NgJJHXIP|ui1S*cPDU!
zu7K6QTixf|Fi!#Tua?5o*B5CJ9#r71)?5qSYu8?=C%a&}0V1y^r>(U$G+?RN^w`*A
z0s_taliiqFQ-@{&o)R`Wq*)o6=S%L+8*2-Vun2zoWG~VUH(y0zDu11OX!6cm4U1n@
ziWT5G_OD_TbHYrbLHHcU3G5N%7DObhVwu+}>;2U{x6I21zkW>*Z{*&cm~t~S<8b5;
zLBpx8-~t^2`Q|h&*+kKT77!0W2>foZq<(`iLmN!sLth?Sn*E)kc4bYVSIF2fHOs<`
z<Dhru;f;$}t>-&Xg7?$Qlq_yr>!Xp7Q0J(<CGdM>WO*#3UMt@!;ll^0P|qZoz2sP$
zu_C9z#^tnv(fX$)@Ef?3&Q8;Rvjon}!~Blc4+#)7#rkN_${oN2nVa8^$33%f1)1+*
zh&74h{M>c}B}zj>xVsx-05nHmt`tjH5df@_bdUK-)?E2CZC8j-M;pF5W4VLLY45&$
zimIwYWd;9aA&ZyXld+2u5_nEC&V2N-kJ9J4ph}M!DlIL>743<y0jCm`lczH~h9W{>
z0hE%FNlH%E^PQ(2ShPRR&W<U`)OYU;VNgv=YjC=-*SE5|rUsKN@kY}Bg$3oPsK5lo
zSVxZUv7gUBxzdr5x97L+rZnQ`=e>~4VFlMi>bb*i+S3DtTtDDwhh-Rk1Mo}G<Cy5_
z=?@;nOd=}Dft_0+Jejn7V3>I7A}3de`xD<-ok0tPfR1TE%TBj%->0*!9EVmG7Ym`<
zz+Q+DMfp=(Ky!=PRY`gI9JyW?9J9@PwBTu0VngOI(=5KLoVz)-Mi@_xi@Pi#@x1Lg
zJk5Y%>FEk=0qIS1e_fER6hnBY&?V6DT|MOjaupE$KP>nCq#AIHQhDNMo?njf^{uy~
zu_+`ytwBjl2{_}53e8Dbv^<iC<<-r`d;~V9{^=qp=y=wWZ`3!z4uNj@ji38rY^|o&
zhI&3A9PP=%%9_DyLKt+MaQ(E6`Rv(cj5`nl4Q-j4o*k4TO!OeCN8|+CvduIsEvJzu
zFgF1iV$;F_i!+||$dNLKH*UJFbc@S!x5|_#DDJ>l(|fOiR+0?NJ(&1!l+})AB2o!T
z<P;R3KSPzTJ-X!C#fTz9$`Ka+^6EZ2pMJ5$UXteHyu8!!f#_XgqCVs)G5u70wM>Uw
z=**cSu%~fRT{9PA71G*H`M5ZiN?6{nGL7=~G;KYnrx#bE#CE_j|5i$A>8&(tGgPC0
z8bMEY$JnT*|1dtCxN7X%L2_wKCt-4aUaQ#YOuDV=*5{Y|uP`}(PG3OGVf>4$W-Q!;
zn1rNl*&PFl`Bp~7pWKf1cJyRt{pPE0<s(N`DJb2$Y;Uv4+TPvUAimx&E+=9*^VdOy
zyqHon-=6ezY9P#f#3~;~3&OZ^vphLb228RM$l0|{T4}e-@7IgwlzI_Ek%|S%k5%xL
zfCXsc`0?X)wi2oDiWZsImRne;<-fCY`b4<8$PpuwgJp-@Y)wIuqcfy1RhywEqIUP9
z;Y#0D3T45ix75vD+_WA+PMD3Fi<J(V&2FJ33EpmNyFB^HLPc87$!VzbjQYENWjw?2
z2V1+-KrevK8iW58!{Vj7F)_0XC5=%RXPZ*KM~Kd@O*F<wzAWSJxy3;Rax@_+i6wfI
zS-s%Zv|GPUhkM8>UoFM^CA8f`homInwgXO$?D8j3{rct2wB<9;gO27-BF^&%o%3Fc
zp#sH}?sXsv7Cm;%YLUCEU?J8UurpDmu^}dod>?%Gf6oP&4M?d`RP~Yo8-Q2mdE)Tl
zH(<RV-e~)uEj|M#Pa67%Vak|66_NozgXUW~5f)yUvH%Rj1k9**yHm5a7-uC(<Y{uM
zBsLMd!IXa(n1hYY_@R!FkFKj$fsHH+73OKwcJO)olRW1T4OdqmDq0wtHF}?XEGtw*
z#Bf}_y(Dsdim71%G2lZ=Fu=bQU-#2;FdD0yk2?+&DHnlxph1zOOMTEAHiyijt}TX7
zI8~awru;}>iNeuU?A+mWwE7oHKumNrIlYk1&9B!}TA0LL3gLWRgRld#PJ-FU*x2Du
z-XIc#EC(>BeE8W|<wRKx)oS#CNjD3!A#{J&K@Qjy!JEruhJO+H-@6~PX%uIU{&_2j
z7Hymt31Q4&kDeYB0Evsgv((wx-{uNQNr!z(r6$Z}6O?M$?|YW&Y>%sorbS0TJXQnS
z6jbvIQ7<a{tA0-}FFQczYiPKR@$H+5(?B-BG(A1ya1gVwNF*ym=EIkmP`p1_$8g5-
zv2{SA(tB8DFx!fHF*`d8>WPMiCP7L4#*Neb{1}yhF9Z~?h!!{cZ@?n{TDP2?`tMnH
z1QQd|MXKsyH(MnP*CH;*vM$Hiug$f15~>mgt_8EwoN1HK_(>{3T<hpaY%ttPyoK)a
zC}vLwDYp957`|m%eQ=yZE>5}8*F{3*L(NqguPe^ZYwn~pttMiT;$>g#neouf%vh3-
zq+G!*IhZZ{Cu7yTtMEb;cqO~}HUrsh<H+JLl7mizxFo@RYZRWR9Ml2k<pIHxgG@~3
zXGQ*lFtYbCW2pJ^<qOCS4@~r^rL=_Y;A`?HD9Y^TC3|V&hxf!Bi~c(`u1^2`#}6w^
zcFo+%B+^z;yE-=~kX7gq@TV&#;M6$%x4wqX&ebZ`NLug9n6?BR?_UKuB_EZEhmMYH
zyH7tPR*bH(zi)(V2B$%w-3R=4K8y-kU!SAua~EiUAl`%NRwLdZ4E<PITB<Mo-gt;i
z3u_bxc43d(w__$C#VHnq>~a4z7rPmqI4eus?St4|y9#6AL2#X}Vuf6S3p%B?{qF;W
z!zD~7W}h?&2nc+=la7(6v9U1>>q902xC;!VM)F=BiaC14C%BGc@>t6U*{G?eM9E$=
z6-`Z3{zHIGC3DH<5UCt@yGN&|^YZXm)I{HP65IIVceFp;_|Ba>g*Ra_W4|OgI67Lb
z75wabhUWr(C5%9*Qq6^Fa|>0$WkQ=?k~R3is0|`Ust5+Dz~d~FKF=0tR~m0iuHyD^
zZ&(PV@vyUfurk@hRsn28v1bn;Ej(T;&dm38L`CO~3x6H-u0a1zAQGFJ=1eeI3<}Sj
z|B%<SXXFfLf1Oo3W`;>4g@paPp$wv*0%YjVqMva!8{J`p6oc}AHv9urTTB*$+66cS
ze5kAQ-%z~08?prKaHlZc0{2V?<{KV@@A6>~zAORdN!)I-H`omC>ZrgoR9*u}VY=D&
zFcZRy<~fAAw}*@jvb%ZF95p3Ig;hl))n@~z4R#RSJG^)+b7$=x+Hrq*F(aPm0uaTm
z2TtwZjq?Jsxq9?7+U%+{(r$07?j}q)u?Y(oq^Dcz==6S%l_oeLDn<Tn&?y~zTBBR9
zFm#?86IhXuR4^o$r`g@nQT@JoW@!~W;YOk|PSa3Lkrub$#*;#XNBi!;`ZSiEpZ`aB
zxgL&zC@IlSY*JsXF87dNp6^><ThWb>An3kS%<w|(Atx8$uz_bXRR>hs<2cdh1iCK}
zK_A%HH`ZiV<v0*xP{Z_zFr`>jbdxg(#+D@YEaTXwJCFb22l{=)NlYAx=EQ!C?>>I|
zG$%VdJ?#u1<-UFQ7bkzBQ-O^J^47@!sYhj{f$pa`IGjfk;`?gWSLF8AK#K*bAR&P)
z!_LnBx}xfqXaCF+TsO>q9{ssM@@O2rwo`dN-ZOicIYEiQVD8<?-j1Y|>e|lC%nU0W
z!RAcxPP$BGIMa>?#}4pJFdPqxl!xuD+FaY#wmt3B`W6mG>i%ChH8qRpvr-tuT{<^Q
z{Q9@3ix|04W!F!v1?nb<(C9>*R+oUi^pCNKMMeTU2M0?^SmuE%oB{0(t^}P341vx|
zzk}M?Z1?TBzwqZ~V5yfxG@kPD<Hsk%=^uEYl~QGLw6)E#CN^&vg{a5N$H&nwVmJAq
z$m99*J6r&ee+AQd5Bnn;Un3bX&j;|E8OQUy`9~W$_vxQMXFKm(^?m-uLBYV>sn53$
zGZX&)ir266z|;gcD#5cZIDWg)8Sn{%o5@$o>kDlybaWng;G<<?+HiuwwG&8n?~;>6
zUDpdYnFfd{{@Zo6vimV94RXhH&lx&-@^cb#3O;cno%Ge?pFVzsu`)pFH+E(6zC|3s
z;H5ZaCqifNid{^4DT9PaXfMm8S0grWbOuA*L{Q1w?+PPhuiwd9^1^#fU!L&}?=KlK
za@-$91Tpe#PGuh@9XI>lok_0la7afS45AQmSzSa@0&I_T{By?ZX56|w1aL=gC$SzK
z?i2eVZa%|IS_f`Szp)o36%K_Z@mutjpe?Dc!(axx2F6n{AZVzYPl4rZ{dnsB1|<RH
z(2{LOF*~zD<MgKo<7qWV+#(I)rqYrYjEKitQaMOE2J~q;N`tm;YlxLrx^}L=OUKB_
zYF*^k>Wrxl$?(RY>w2XjKQ#@_6$uGD2L~#@YbOd_H;VBd*#ZVG*3LPs5nKZwc>Qlv
zBReqB((d&t2lI|>GEe}IN~;%Y=={e(9lU?l8$QVVn2P+zW8kbBSmDUX_{SY4H`)qj
z9>o1IwdiG~TcP^MyyYX6%{A)lWz2VTl{Jm$t~RyTS;d9P890qk)MoO9w+v_c@TD{g
zHl~@YkA^kWDM=?btB*XqvHzV1`$OV%$<`Kj_Fa!3cm7(aT(D=AC1!CumpM42Qn+w$
zVwbGj{gR%EB^#?3qU12Yf^EBT1LMPEBO|-X$lmI{RXBOF9T-)Oo%+nXzm^j2CeNF(
z3IVTPAyI_L#3`}mX?}oMZEXC$N|a&C8%PlF0{<N&D=RCjtYpQQ%CTeapMKKrsQx`2
z9gRkdUr1;t5gHpnx}P82-<HP|!x@d@Gd5-lH%CyAEf7e6R5=EUMYM4A2YPKgxM72W
zLze6&x}FI$Ki-?pz}{<cP}HOd4n9uKyM^4b3Rjma=oWf;^hb>{cZN_i04hxu^px#j
zD(&qp@o{wgB=5SLtAvb`Gb5<;({UjoAr6k@++5YDYfiu=1pmte7_`5CS9lgJ26Gkg
z_VUsaL;?8u0Ifn~2GAe@h7x!IGdsK71J#;1j0Q`3;@MftUy`WGIy!H_muPaPBql=e
z-1-2P5vZ|LrgFR}mvCAm<dc;Z_irQ8*P=gi!S!IF%NBWZOCfr<bKDq)*5S4~dlP5;
zxNI>mByu5!qy-;$3sWo&<iE5<q!dzW_AhYxdTh*hm~&|t%WQmXGUUg-;DV(Sy@tS$
zP~*36_u#)?qs|~$=L29_NKs8=K6><%5R>=i`(RYk)kui0?Y}e)#sP$~E$o0zA-<KX
zn=kYvUZ|R?D($e7BY*{fmQPFMFm~WC_cOA=W}@EIQ5N<NAIlz14usw7-=3LC<iG2}
z5q9;hxdAzQk~H(uH;;pX_xf{11vC{wIK;5D;Hq)^(4ehY9+gQ-OYht5j_2C(pk8c6
z=EVN*MLo4N$XR}Vu*6DS#{?y=kaab9Zg5D5gGE6|Rc20(f8YDOyfZvJQ-s5JH0h5#
zWW0FS+`L7*Pj2@jUNW^8n3R!;i5_5^A3rj22+7$ByDh7B`*&ZGlk-tX_;pVR6&@$Y
zeEJC!6HeR#Fu}${d$_o0cXk~g8pL^%m<U+`dAYRoT$VZ}fP0{~LV)mfUDbsvWws{w
z$&>BV8^r!qXi3O+@4n3&0_+O{1>=SoyFmScn|}LA2xWOCB|VtQ=iu<JA3PNR=j7MW
z&_mCnaSTq8>xRp;xL0h5d}+<*3qQS@=n$BPQ^6v|OOldgID;aEEN;&X)NZHCgVPWD
zB;6R2yHP{*r3xAzTBV#he)Y%4fA$hd2R-a&`uc_Xi^MWGV)L`%Kt=cQP$1i2j78;O
zKY0?e9SRVr3<%?3ZSE*IfHcFbMr&*EnfvxKUu$Zf`}&r+tmUh+LpcLN(c}!O>_;Gh
zC8SI8Jk56yqy}#iT7m^o2gr9}_bndx7F72RKZ4Zy{6k~cK8PR4yl8#qY@f68(34+r
zRqcKY7e{)H88AK|)a(>q=*6ex@a_Wmeb?ho$lvk+5<quRScF!@%a@iA12DS+M-@g3
zraLqfn_epPz)=e_c{b)T*eYE6SmnqlDBwA(tgLMGU-g7k*xTC+jkLj>6`WhGzRHao
zRlhCMRAtr!Q%}rzhS*<uy!S%(lWnVGaOm8?fkv3{Q&DLlkXqE#0tR*T5XgDWiZKT!
zlc>uZf`b(BYFQ}?p;*hxdPhb^f?$ERlpNo{{MeL3Sh(Xgx7)eNNVspd??8_?yP?oR
zI1~oN=P;8!W;QDda-Tk}q>BqB2jVmD%5Qo-TO*@yYC5*I(1Y|pgd{gS%(KzAv!S{d
ztTIj#GuqyOV_qUmMrkHt#l&$bv9a!BYQUzL)F8rGFYS+t3fSU^9#q`{y)?vx^M7RI
z<#B`+nylF}fbDc2XE^w_dwoLON=3BfWM;k~UdNWZokmN96PL)26K~zNEg2gh5*%c5
zG_Tu9rw@`<z!gH~@+ki8+ZU&;ctf7QYNXZ^<Kdg|*n8nibLxXgXK)=j6U*xS`7*TX
zs78`3%=b5D-5)<r(keQ8;>4F;p{|i(Z~{9X08MhJrGIX2G^D&>c*sKNHH0rG^jbud
zFkNE*kC@-}{r33_uU)uYOGCr?2x>ouUR2?smoH@)nCR$Ca4OL@!r*v_r+yu=-SiNa
z9w$Z6?*ERODY)~vEvRyVcOhFIl<6eFe8cSbt-i0Ly`88OTv~6sgo46%NO53c(W;OU
zgWK^Y)z42P7p<y`u3^<5V4%FPv+WS|q)iz%585X$Y7Pu3<M6aiuT4*RxjRf@j*jx&
zj}II?m}lI?HZ6f~qDc*)-`rw*LoX9f-23?x95eP103X2}KmPn1$wgN|yXk_LB3o&8
zyU(VR3WPBnI@I3UD*4{z+o+D<FtHtxhnLqWc6O!2Ca{v=2Y{cjFP+39&E6M0Rbtyl
zBo=9I+3@syj(Uj9b|O@k*+Bl~R?}>ppn(I0hHod<Bg}=m7k&O?Wpa*!1xHsB=+gA^
zm~<e!+HEJN{I~|3rs4m00YKpWNpW3ER8Hi;F>vCp#Z|^zj_7W$?yYh#{T3!J53nsL
zAOQIg(lJJB6-qPO#;(#}8Z0LV2PN3tz57AtIL4cpIp|c`mEoX7^x{N5w9HrGBxazc
zB`S&<_VgomdH8f#lQSnPE8y`k-(3aa@)}mXf4uE%vP(OU@OQkD>In^nZI0;N^&d+!
zg9gG!W=juO_in+-PKt?5-%Y-e2?&r<{4d7d1fHvQ-5XzI%Ge;8GF2o~A@fiqDMQjg
z3YE-d9y68XDG8~NDM^w<l%Zru8A3>k%$bsT%<TOxo&EppbKd_s=ly;5=Xv(tk0<N5
z*1higy1vuJA|!E;5bg}PdbOH6n2kOVMSl$}4Z6B+MJxL*A|3}C*F^Q83f-U{J|kb%
zfj3@K;kvS%A!WZsqN3&1=n`qEZx?U9w(bFN%CVS^V%24e(o&Jw*k^td5$iWcf@ZxA
zeVI*%O4O5`bjqz*hObYxwX_`m?aUL}-`kta+YN0`R~KqzoK~#By55vNkn)Q!JNFXG
za(sxrf)TK-p^=2L3EM@2lBW|60ldlF%t5VzQ!is!L(93ixwi$(2!>Zq(kzr2e~PV6
zaq_{=Behnu6YH<$&d+lP4_#$Z@$WiIIPh#a<2b!j-g)1)FXk7ncphIjAK%cy6mLS<
zb047T(hyj0y@l3d^VnYk@KEv{mG|A_(77ad{JfzPBSYbTDEX1*vR3~bBw!JdQ2Rkj
zGxqX)`BVTl6l~~@yEOF8<^3C}3pAo}TfF$pC_>k%S;>5DEJ+G|?>-oNH%oG*-Gm79
z^t@NmJ7&K^V-Tfut566xhCs;^R@J-fEQP9q!l|Q2DG58777zQ)5x04i=V4>l@;kYi
z(<cf2<;2zM{?UN}>x&on&_N=Kh8d{kRDsWi=SNd<$la1KF*1@@Qeq@5`s6=W;}$5d
z%k<a8qTq`6VY_9)Eamz}vo>x`95X?GU;N(Y&83>WH3yu2?NgHo5Ne5+VrPs{A9OMG
z+Q7!nZq-NWfy#-nx^xNmTt*Iit=PoXZ3GdjG(k7O_-FdNDVO$R2?zrYZhM%^r-~?p
zDO=GUJ0A8^uS25)zZ6m2>xAm8y3+aP-#gOD=|Mi67n$qm8Fb>;Ua}1?x|}O0uG`z`
zATQ-yg`#G0(*N!8#I@lU#(MT7)RJD7+vF}VIrn&@qG1-HvVD8<ix;;aJq09!J|BoT
zK!8uZ6OjLWp%O>p#NAIs=g?4rU2_&_Z^W~0QSk=I9bMZWiZBnXv~S-iksMYG&#Bg~
zh|1F-h7lhD0twF%q9%Y}0VNQ3<+%>B4V;0(-1_4S`xEB3*Qx!w;!mxFHP7KbGejly
z{d7SSJ4b4t!_JKx*zVq^IevWK(J!}>uk-AmZ_KNlo7_)UrqtfRU$8yu>#x@UhjxLc
zp>5848HmNfw;lp0OlmAl^Bm6NC!an&sOkYR9ccW9QGJrGhDQ#eST`Rbn+Kl(1H6=|
z$bbj#k*ZC##N|{hpIu#CWgz@Zv}zL*H6f3DW5=<Jy<f&|)=_=U8U_z(elJ;%T|f5f
zn!+A-_AEbuK+#~f`+uc`5T_wK1n~qm(wyRvRaoeE?f@9&I3Ts3P1F!gJN~?Oq`)XC
zp#wn@BG;x(ugyv*yRXWxW*i~RZvVCoK*Zo4z^l0`{Z=Vq8<+XS&3`Q{?mRH;=Fb%T
zU|Bh{25u)MZN20hz!(-9gf79_CUEHM{4^R4s{Zn(p+A2{wsbmCm<5OGeJS)YC`wUF
z<Xd_()_vucFf0Ah*3WJN6(eDvsqJLARCy|*_6s;dtL5bMV0eF+5B!sl8B)bmc;xKq
z`LEU8VxWtfRnkl{TB?!lnU+;oZ`#4dkL5=9_XS`$@gi1wTj3MISIN`_tBq@y+?G-N
zL;h=pC;$FHw~YODd1t{8uc=n~3FF^TkDIfJkUvoOhl6blf47K@;aXh&6LDaU;_(L<
z2{rV=)#m6ikiVqj#oq@aqcqLBxY)mu+QXxGpg~0T(zeT9veMFFacOuL@w;;W{;tE!
zJlBd=ws`yK#Cr9t&vRUJU)0gfjpR<`-C%@)t0rPCx+|QZn_6jk_+y~gm%BXkT*EXA
zyrd?N=Z6DxeCC?*zMMRM+=a806O)N)4BXs&JLzREU0cvlBt#7}AHpt&{7=CMxTH6(
zeBZnJMcBwMBE6BBSEaSFQNsj!k?-GoGPC0+`)iSO7xo>f{sUD%6;EXa1u|QTK&x{q
z)9gykMBKiSkxWY#SYf@d`(^{20K7yvpmOw_>_bp&t;-$!`Rml&2nLU9cDj7)lmJ`O
z(83M_(2OucWsDY-#xf-IVUX6vi^FIn1?lPrTtMc=z4RVRK--Gcgo_u`4<#e#?-Tk~
zOGRv<c;Af3*G>lLrL9uyDP!Zq^t0#)g3P%$MW-W2^d}`^uAHr`tjZ=l`A7&K+6lN5
z0nwFQrA9%U9k+Vd>$FQLdGP$0V`?DJ1`e>{21tL)f=rlg6q-8kCM9)?WE7~GG)W)7
z{BXrWp8NCGP$iVbd*XCqOPaHc_q<+WR_%6m^(#uzzpdLkweHc5L+K|5T@<)JZcUp5
zr0?ZLxmQQ0hU%>OGGZ96jSItC+ed!Nx*Y88K3>}8k&T+6%PF6ztFL!NW#0UmnY?3G
zPrB%o)Mn7CK7aLU3JoS&#-w_O{8n&6uU<)`{12%3-<$IGV?V`eXo+Jxk0ag$?Ny~Y
z&abLogvY=H4Q5k!cheW`=l|qxp$S;e#dR4Z+uHjjK*In0`Lkv(Zau`x)q+-=T@>0N
zdxGA!#F05D8`i}LAw9}s^x!y-3t1Z0t-dc-hQ(XWQ=py*<yBRYlRL;efOC9s5Yns@
z(957x4JiU($qnt%C0AYjyQ^#p&TzEG%nS@MhYB?2&`g4xIh}g!i)O_=xM|;-m0X25
zGc7=<#vHy7gB9H$$KXeYeFMLxJVryWuNH>qXeH1Sq?cw_V_^Fe79D5<^!3qT33O=T
zn<O5|bLdhF2;u`_7U$kh;SYGLCQk8trvSu-5O>h{!>AO#{9d-h8#NFXq}0xx&tAM3
zK@1lb1Y+)V4<DBD&Bh)pCL&^MZLPn96+&>Z>5|c6tDAuOS}g6Mdf?zts%hzP;lsw(
z2cayqszCa1ZXY>95M}rJfjq&dj`ySC%a>X2!lPjLo<C1bP2BMt8fa3>EgAjoqw(EP
zwZw|bO6+0W%tFG#T72sq<hY?<I(13`oDS>;_<$49$dr$U%pM9wVPBz(FcDRhmYzq-
zCF-JC$2F}9s=S*wTh?E=IA8ZTF_H9g1LT-mzZ(6AljG!bLnzD6>Obp7=mt2<I2Yk^
z^h|jDpEueduLve^TxUw}$m+_TVIjYv{=R)zj;5OLV#jfmRP`P2HwboLFc-K@9li?}
z)zq$*&|O4f00kXbgkPQ?@hj|P-adPFNIVizJo3(chMVp6RnTtI2T~H??Gd`C(IL*y
z-v#^@+9c?-r!++AWrQLiQU-94^||%;=wqF5qbc;gf(#yCI+zLklUs^SA?0|?@Xy`m
zRb?i$1*MRpv}4CVmF8|PF2{}^x0Zf@J_TD0GPJbETfQ_l9zSvf4m2odKXs2&QI>dw
zqfN!mCLtn{v|}eS{%#1xNBvA+-4b|Mn?5k}zm?jx>&U57^aPrdFBwThaYKVM-VJzv
zAMMzX6k1}2q=6E%;|_-o(R=boR6Y;Rl}>=4o-mV)rcrK!ix^UujT<&#r~MVcMjaDf
zi42D4Ro-P`_0#yV)MGXdHyMk-S{T1lF<p_bK<5W(HgZJaw1?w@<I8f0c2%x`VOk^9
zCu<2*`p6o{#vMn>1DbBEf0d=F2A~H-SVDrH{@r^W!3)qlQ&W}zHNu~P2rMKE2-1D7
zu6A*Br0_@QT2@vj+|EgEEWh?$!Pe=P^73W)U3Ku;GAbhL=qidm5c*t_3VbDCw;S=W
z{85OdU<`-7waVi^Q76*2bF~e^4n`WgroW3Bv)Wk*0?qk$LFN3_fB^iruD8jO`)(U5
z__e#S=-?fK{GWmkxR#lxH!@OhZW19L{_S+xw%ci^HIg+16veU7>{`B|?t}y@WqFoZ
zy|~qK$}Q;Dr#Jce;ajqAg}-brEK~{T7VN*Gmw7TLdi;LhM(EHxI|%|k9FkzgMMdrG
z?Ro8G@-#{;EYxbZWX>yE+nTc}vtJZM<{xfy$pj^NSrrXTfExm?d!j+zz_B)AUOf!o
z7VCAg8}8qO@kQ7kV}$hH`z;F%HLp>0f4{Qe7Zyotk)c8!k?Xwt$s%=#IaPHN{YNhQ
zg5CSrHSm%i@}n#m&oSO8CDs1LyWrisY@9pJ&KD$&kcmCuRD13vZV=XXe@#q^>L(lh
zOpR!~&S^yvQheg#BRJ2{Bca!Ce-&Dt@MmHIhV1($>6w{F(oet?hn^eFI(F%aId7Sk
z4=kR+S1bS2wY0=2S?(^j$i{mI;ZS9vyFsGrkUg_tL_UBI2+1gl8$W%@4`R+2VO?<_
ziF4-@q+#B2wYpl^B9Ppt&B?N+Vg#^F42L{T?Zb2b(gH|H1*KPL?qJ1PlZ(!{uy7fx
z3EuZdPgl>YK9Hgpo3pk9)>+;;86Z~kp}jC-9xfYT<-qcAdZYWT*f-c$8xUB#@41JS
zWIHR3)3evF8}mx9CjWZP)yH6K8=UW+AugcS`zs2OTTikrS6pAbVE*<ZuxNP`s1#wr
z>#J9&x_X5ko}S<fXlQ7bn^bIA1}DzGG{lRJ55X5vi8CK9F$z$CL3;5@`EWaA(X~D@
zbCyKp6FN0>liYVcLetnh0CA!;ee4hY(YDvGHc=}z4JB4St}qGclda|rtsGyu@hay%
zLM&Y0T$U(&dN$i0t;F3(S~s_45&uV>ft&f1xGmgb_L5Zxe@n@pF}mh0SsAW=ek&ea
z=Wx!aLy>1QER}$&0!sfS<uXqt6Nq*;>M6n0B}gEOQXstIrrBYsm${>lDULqHkrdHo
zE4c{RiDTV5vo`8V=P=+oMG!ejyqiKKObie%3V$Wq@T%SuR1pJ9_%hvXceUeXou%BC
zR_Y9<f2Z+9cuxOOJ$jmvP@w7l`Q!EV-K^BAb8F3l70Ihq7G1r4Ve@e3UVHwJ%)_Ip
z%Y#MBgTQ{OtE#eVz5k<m902TFRs!`w0b)mA3pB-QLm@GS2-#DSbBqjY*YdYyhJT6M
z#=*s<xUZvRylEtHtuTp^p`{)75xyy){JTc9GbFbr`_6vK1fK&Y;=-a+VO10lb@;;A
zQ}m4)z)IKZSjWXVj0?mnPzMH@%}6ql*O!<|R$Xs50)Sy<r6OR%+;m2B{i*aB>~T1y
zh_KjLyM(zglb_HiPPtu@_R8LUWv`h=W^{C&LEZ4MqBW4WJ0?>&F2Lu#d$;9zoT~lX
zAi<uPjK8gJw}^;{RuaX8&o@~)IV*?#X^rjq&gBKN($S0FUZ>m6+KGzDe#c**CI#3!
zJwX-I;A}wTIZ&`k&|D$Dr3YMWF&j8KHXSIGaP9tm>$lXar~Iq~QvnlL*4wg<n{}v{
z&TQf0iZQduh)Y^XLsSoYQ81`X5;RXewmg}z0<TZc<QxD&JUo;*=5Jh`oU+jEECu#l
zmQ+x%@fQBL^sN=>)roTByQGZ}S`7_-BzZz>nIL6(8%MBOq7~ef*47mM*rlMI5xH7Z
z{Q;p2iATa5?G6`3<fza0=(8sKWdLslT#mzFanU<Yw;wOX$cQd&-DK0W>b5Bx=cNGP
zShx(_OtfX@_DZMFJA0P&I%ySXp!P0-)*n_-ga$}OwRhl{>;1He08<Z6nr~bo?sjR{
zt&kAlsUm`c*FT;cFpXvW9+7KN_pQArrM@SInkX+ju(b3uiIdYr(SaySykx`4!A=lf
z=~PSOnS_RX(aeMHuF~t}_YTeWtGZf#zU4ht-}J`y{+W%<8}a5e4Dhse2v>Y#^!3`h
zpNoSdF#&(C^K(L0rLzhTcqzf$h25yHxXxua8y2@*#YyyfT|h}@-uS(?hYVb7!50?A
zaiD`EAX8dUA0ffi)Ko9w+%9$i^~T5$b)<feXcLJ&B_Y+JxjPc@wE!wXBzBb573seL
zgSF98mJo1Fu?PoWt*JDJvKPBGOc&B8vqu!I8G6KNcz=L%!$AcC>j@V^j2z{-=Z8M<
zZdeLH-VrcJRt<RTtK$?SYx^Vm`5-XZYQO@o!K}B+T)ZJGBY|3kw8=m`0%6f6WC5;Q
zS8TTH&D*zNqqMHXg@-2=txW)~d;=NYJ$NNia!ibk{Tdz3NKc2M5nrVz<xzCPfe3px
zB1%`c&*wV@!G8O8kL|#5Vye8<`*zXtp7K1e6nj*<D${o-hKzP9w>_#7SlFk+!!k0;
zYOdHr+qPy&h&F3W$br&xFQ*btARru(b6>}_1sh^RgEahtf9eTv38Wd<ziP0Owt3A-
z0P{>%OlV<AS=jUswxddSc)x^L>B)RvFbH3SZk2b}L`w+)tk72cuDP5GU-zy;)+IX9
zlCENFU64EnU|0oC@cHRMkc4o8lda2XnDdc&vSrJ~G4)T6I<=(u&-?-)G;pZtQT4^3
zW4CZQotJsIxQzAn_kH>SE5TCs<47b}sH?FY8sybJco1YQAbC~|5Igt`bX<jnur)uT
z@PS6jwlsjxz(yv*G*;wMzZ#r4@eQ=gbVeqeqAaS54!~2=(gslo&!?QoFws$9hMI)k
zn^)VHQ>V~q;>?<C$8+c8E?puX9@$Zmt-2H%WOVapr-Wbhwgc&IzRR9Z)7HGR<oHGL
z=*SLf&YFM)Me1f|4rQt3fsl0syzh|9DHu%{a`^CK=}ySr0{DP#NLmr~Dn@jK+&B+`
z1Jncf2)+AZi(JNbVZbBZTt=KIv53ym(IRw7&H`dgIno<Nep()d>_|g{n#g<ehK82b
z*e4Frj$p&25(3B-1;S|k`t>M^S4rH1DGf+Tq6>UKHRaMR_7jC4UWb#0hC5)vm*ntl
z1u=@a)(7Ym2CO;qyyf%qOO|?hA}T&T_&r>fkg&2vOl+)DXsnc1ciL^W-0wE)<Q{Ey
zogF)yH7P|5kx;;Q_n))>D-=L|Id2QcIcfq`*3!EV7rhL)+NgCDC&tH+xn=jyrsi9x
z#d&#nW)P9s$8jANK_l1cH?9uLN}j2><p&>J;AIJ<EO@ssT}H&C`+(f#Z>=I*FRLBs
z4B<V<zb*VA;a_9XUA~3`L;;92Im%E*-03ePpV)X=hScmeD^48U2;e}_y%gK^TV#=r
zAP#Y9aE+<YCLF2Tu9pX>2R-phy3gjE=1y2yMg1No&+}_#LDU6Xa6hBvfPs^MfyJGD
zEDEdRPAn7|r1v<<v$DNUH3&;!@$t!y3yE`d4rpz?VKXc5+&P%RujKtqmXqTpAA4>f
zpww$&VqKgbZ&8^c%?sA!{Q|J2a7f%@tdB4YFhCk8ISj<}Y{Ri4mfV<+F}ji1wo!je
zMpu&oMH{^zg)Kcx3;1ptlu<}IFU%NBU3I_2-j5+vNir?EIik}n&v%RO%4LsH<vPE-
zfllJBTZys#mG&p@-?D!RT)EQVKr(A|iH>Fz_M4HFioj$n9Z)&QfRHg^k5)zK5IHWx
z;U*vpbLX!?3vSsS4XN04<uA{OF0CCH9{xzVTryef_de8*k>>67qpYX%Dd*-uCy%9>
z)*Q8%DX(GjT`-<)MDP!>U*Jd@|2C@oVr%qjt>0=KS-1ay+W91-5|I<yEk3J_)Gfmk
z#Vg2el^`Ccq?CX7&?J~J3hd)QPzF##<MgCBC2P$I<Ow-agc5rZZP{!$nd_b--M)+v
zS3g9ZO(_JU>jCFkd+AWcIaA-Yddk#)WGAH_H<S!!1Vn5snD|uv<U$Acmb23~Tqf|N
zb#5*Kc3yKmB_Zc5C%KyE+q0khuhfV7QW1#upkNzo5LZ!I28M_Jw!7w$&}2Vix3uHI
z)sq^qD4S?Frz`sO;4)7==9GUL?ogZ;H~_i_qd4qXa^LzaxS<b*8obbFp~Ciy&^J(1
zwY8+?7`Ea0k8>#HYU30weHW=l>zNVFjFSMO{;Xidpwdb>ISwC=87>G5q@nGKwk%6;
zx6UY}H!|{94YI60EK$<2EaMsQCenn`+kCYSHhwQWH6!qEt-TvSkTMDYNw{Sikq9$W
z^<B(6n))~tqX}LbdgJV&W?(b-aldRHe08|XZ9k|&Fmb>9Zr{3v@dNm@zHzmMxfKVi
z!L3|BR0R;NU$V8`y240D=dn0mj+<gBD`FnPqC%v)OzaAw!GfDz=(w1$4ti)yOS#Kz
zC!YI{j5wa>@nL1=i7kmGU!p0WYt#kr__BY&cBaiX_PA&ZfjoQ;_e`$;S13cd^)^)R
z)wQ*>{`eqOKtdt|Z!j8m&~tckYGDpQaLa>31xsnEsdE^2a9}heQddXkOtwYMlznrc
zMOm~|-vQt0O}|`<&6*^+9t$aOUTVK)8GoffsT3&r*hJM>8CP+cX)3uUvcwi`OK;;E
zp^-t^b};#9K$ih3a)>(m`@c6fUg)C+h_i{Fny`Zn8^eWE;Pr4oum`##AcJoX=Im(V
zxul80h+;2<CBS4-Podif^EUxeS59s&WV*ro1rm=~xfUoJq#nZ^M}UUFiVLzG3p}08
z&3lF{0-t<XJo6#>=(^yN+S>cg4@c?guLf^U%wyM$zxLKWv50qZ>1^eOrk883UbRq^
zC*N!z3B1{wfBWsfGM57ft}hE`5$#{*CecO`#fah+5SR)D@&O|wl&Uyk2FZ0WZ>n~(
z`x;I^n1tFdk1F4M1YZvmz0=|DVFQ@vkfVr$V$O2^{twtIcs}IJ{Ze$Oa$Lyu*ez<J
z)S|_Q-~Q_|J%fj$nnis_foC>+6-EA$hmx^c%seUdKFOz_z>$5n?)pWPPX=pqf?Db!
z<G5vci^O0A1<}T0A_oxvoc6CL@H<0bixn2mcNy5Ap-*79#CVE(fM$xDzp_xM@IHvC
zxoIwWr8d`5hJ03P6@`Wd$uPuVmWPph5$&Zb<BX_|-y+RKPcKwk5Tj;#_@Y%udwW0T
zj*U*^P=Ehk?e(d+m>66VHXEplKlM-QoJe7#ZYdqMLXRdGiWCtj8o+&bbc_N(b}cvq
zMZJR0ybH!B%%wmd1z3<gKGL)j2eQ5M^QpaUb{Z}?vy+!Yr~L4_aAzo~<8RtMI1)s?
z_~HMr1^&Od+u(kq|IFQ{Gm!Tw@AQkxZj|6XfiPoP7c6120Hz}iRV*2OLv;ymhnv~b
zEnBT&0P4+Qv6Up=X1Pt|Rq&g;54p#W{n;3;VG8Alfq^>xEQW&}Z9SzA-WbMNOrt>|
z3q>|GkUdWF8v3aJ@j;yrXMC`4t<>rcu`x;3RVZZeiL$RCP&0(*;f)CMLmDXD(DIz_
zvA(m9Q%XO_IV{d@JIk<{QiU;28RZ}%-Ka=07C@~p<sFznM!@vDX=%2xXWx8Pc2*W6
z+q#;)pQStv$Sc)Rw56ZrUF@WV(!C*`UH}yl##BI=QYq^7Z(T6%4rwRQW+W=iv!KBQ
zeFxMEEI*Luk(S(AfJ|@9P+$;@jEqr39-6WqQdR@@){D=9Fe|$LW*rD12xw_Sm1m;n
zbe8MZ^o^ZeccZf`#Bt60_nlML#c`B>c@{rBJ=oIR0pbs*@(Cci5BXId9#Peggcv$;
z$I`9z5$+Ag!R^~qE-pNw2;u_OE3$Lv=Xm9`qpAA^^v|7plw307pm+)fH#nxo$H9%y
ze=@j{D;CYTJbw~<Bq!It)NB4kMfEk_B|0`SN+M48q`tmm*L!M7Vf3yrp}WA?*pxAH
zw(mDRPS}YmhuK|L($_6>F9ouwQt>7<S1_)wwHnD-HGB6cQm<l%wE!s@@OBsj4Hf;C
z5!!<N8+9NlEtAt2y^Fv>8;wkCrKD9xBbcpzW9QoZ)4m{Qi72#`0Rg4QeqP)mL>uqN
z14zYn|2E2Sy4&1WQWAK4`{N`ePFv*_?UswI^1(3xAd#{6x<xn+*cFw<A<M9cHQTq>
zwJ@ltgoP0;x$1{L*$D+b`BYy&j@V{?37(M{l9RIfahAe96weoYGaes6z84JxK*7^R
zS|EESO-L~@c<xLJO%dld5x+zLBeLWsK~YHx{Gg0_XgdMK7|$EjBSGe1uy$YUO$zPI
z@86pu<#;G%BLW#eDvnl+svg4pxz^0HrPaztduRA0H~3DcHS1F9sL}-No2mEyWq%Xv
zF3}pLoiaV_-j0sQOv!ibronCFA>hnMi~K*S0=;M@S6>ck<{$$T6R5OFFIuk?E*%{k
zjA9gga&k?14Kc~%GWy4<FaLS)yj@r@+hO2~?cvDZ?2LqiiwiCe?#cHbzW*I71jic`
z$C%}odJOLtQeWVq+Yh_v_3IB7$>dc88>}flEb-3#^LvU2<u{__D4N%&-g#p?t%@}E
zS$~bki&aCP^<$TVLnR|wwpvM@DuzQMw!jjwAIHXx>NDGDZToB2?ETNoGRWn5%fH<6
zFMX@kFZBvLSFD2TGrY!0N$M*e@b{rUPT{fNShTo3>bI+E=RkBsT$|5IW#Gf?$#K6G
z_e*v|A2uUY?k&>w-d#_C909Nk?gS>1$;o+u>BFAda8E_8!shSH7s{mTV?E6$+}v`Y
zU9PHni+C1{z<g<_#3O;UDJ*=!ge<sSR&}QP<JHgzGX#Qm-wZitG=LZYW~-geLBR+}
z&BY@JF|ZAh7wrM6<Ze>)D0)X{d0AQETwe@YD!5-5e@Xn&=njTQ_32ZBz&jbS<$LvA
zn51}l{g&ok@q6XuAdJmBEsl^)1QXo8EyLdQa`gbGT1`~T#ZH!(c-G45pVBTwe+zEk
zUX!5ecj{+K!hXul;!#S@@=_7UMz!O_lQJEeU)xqU8AfILL#DGbKbZ0Qlo)}#009rY
zz?ggzdh6EI%nY2OAqN`1eVc(t{339Oh}%6|uUSMMNRmg^WV13WvE$|jQZ|b2&I>gg
zwypURU^7Vm#X?OCMLhx!kBLd)xsNviBx%`yZ*Pa*+7SQ*FMsY^$Z@VxGoW{PGH{XH
zC9J$qY4gfEQ4wDn8fw-)+3_3BA!KwREc=n{FEyHYpBbZwa$rrBE*H*IP*9YmG}+B~
zjDrG!@PCgKuT;QrzuPqoo*_kZ>Bg3QJ~cNIrk;Eg5m|Fn((`&pn~!C?Qf)~|x~cd2
zFF)uB2ftDG2zYx6=YNY3W-kp9K??$Nx1kUuDqdI)2$DO(>I!=|UX?6sUbPC9|4w!}
zQ8We{!$7Lm)CP9J2;@v*R>uc(Nz5L?ACE2@iyw83kgr9`ksYYB?jG76RElstxTv6J
z6bMHO4>J=JMq+@E!;G}r4=o4qYJ_=;igc153$61|>_X-UdnraHv?CVV?`9y^W?&y1
zDm^TDgvD=PvKr&&Hg8x6^7-Q@;Da<YS=r5iJ|MYr15@jGD<Yx+EFg4Hm{OJ8XhTA@
zFtxO*1H5ePApNWb2;1@cXk>1_EvDuLEM8T;06QSqAP}G_apnybfpB;f3J+7>0>HzE
z6p&(M3%FZEW?$_n_|H@ruin9fa&V6|H5tO&ys9h+JG0w1<YhF1c9)krR?hIP6Fu{0
z3d+TlG<TU8GQ0rDMkdCwa@q&g1>CHP>wKJh?9;c7rOS)$PA4-=wl9>bPE>3aOmjbl
z=uIO&1oyyjBU<z43I-#UX}{*$TcNU>{T5<#*r$TE@{4@<l;ei~O{vJ1M!Tw1q$D7o
z9MDnL#ylPhLNLO5Y|V#;1}Quo_)x2=Y;W4rk}I*nmh)RNQaJ~abcirw>+Mp$quma2
z;7Z9kYuIK;PSJPkHJ%MOZWNnEweS6Go+%9NI^cPNoD`|6Hw%wbSUedEcna7Z?mtzD
z&FlFDSp5-_8&Cq|a1hEj!FRr}O1ZkeCZoBx*}VRN#a&e;f;BHFXmQoz8i-vqc+fv#
zi-T^~<+tQq_xozyGsGC&t^3kXAR(8Wc?J1pdb&<ZgFLpe%a=*r<;)pW03??Q!oIke
zkVjdn(0(w<)t-rekO~8ncDwmNO577TvRoU2>0{(hFa7JFCIT`rgy@2~`D4UVniKnE
z?l;)H2D(jiHcyL+tkn{pOLSt#ggynIiYKNxS=7tksvrxS48PyA@7fc!<f|>Yi5V<P
z%KOsehwVl-((#Eg-3%5AHJ3cwA`@W=PDYdSJAOq}wkkH?L6|=(ANn95kxHa%4`&Uu
zN(ZM45hACR<_<elUv)GD^MkzGdN1J%!3<A&n4C7)V+t!wbBu+LQCz|F7VDKFj<#`T
z$OuCsms3xq;fMoy_3_*|m>lSl1oMl+&Z6`HbcP!x%ErYicA{v{wdZ}^-RHgQ@=}F3
zZjwsSxgyn=8o)XzPlgfD)oB7@7G6o4*9y49*yiD3la{_)+n?Q>_feF*9Vw#A+uQ2u
z!VndOehJ0iZ(fXiJdIus%S7hlkNtve(5zY?-pIcdP!E*hAJ0YeP!Ip|m+iQwr%Ih4
z%RTl_FvFR5oNDUB=e|zm76mq@3GrU*c+F7Oygw6i-GV5OiGo64y`b;!WNgXsPuZ~(
z8D07L90W4r!7nHcT(ED=+jr<R8chfxCq0yC0>I-!hl*e;tgW+W6UVLEQQsot9{s>0
zjbLQ*Amke8Bg7~t(v|?|V(6~kNzrevzd^p&*}MkU1C2RgctFDV_DCGV<V!p@#ERkn
z$0^|K;$mlIl?P2XbfBW5aLziqyC?iKR=3(&@pSg_L5kWfe#*>|Rk)3Dc>&M=i}o{p
z^uP2EV%ogqJC}38h#~<oenv*Aejgyt22>Azz9r8wX_>c>uKB@{9cWYcETz8nSnoFE
zU$SrM{O;GU@t<O|z4j6XOET4$I1-}QBKTl&#C;<YZHh0ChPdAus`v$qj+3ob#Nw#E
zglhZ4*b`~Wo)g=m4YI*8uC-|Y4f_{0!4pcV&wtleo4&HD$6!Aoa?pr`q(Q;KiKJAd
zuzIE2qpcihNWnWt2AbVFR4MnkG49Q)Q^Tlj1{N?n8L!{z0+00s>L6GM(H0(c+9@X&
z_3<23<ROoqf-hgvyGy=;m-ogWvzg+ui9)y!j?EVAlcuJANW{%Qt*d_w3)k6s63ZJA
zC>sE3ZJi@)(hAZ0jk{c?F^}<$&%zASBe32vbPiAU-1P0CkJE#hp4t1Rmn9Uoa4XRj
zEl*eXnSW}rFO9K(@3Ap6QuIpNNg*qD&Atd@X_YdI>`FW1rTkygQ2*f1K;2DQjI<cE
zy4CrD8v3Wwzd?clPX*23?=9Ti71|yVwG8ZxWl5nbHg&1!cyAOuEc)<U_YB%Smm9ht
z+7hZdhlelL?IBfAJ=C6}k{@%g*D><xjZCS&q^SsMh^3JhWdrYGfyE<%!v_W8E1+S3
zM+gM8s(4O3UR>TLh%ZUkyhh}EJ{0BW(B6^eUCVJXH7U*X>tL`JL=wtfI|;iVRKaXD
z2X(v6c6CmF?|NC@`KKtT^)#Q~KiP(36<f!KvNN4i<ejN*oh;M6#rP@w<K#RvKh4>e
z*tRG@Rez3s;XxJr@dxslJS=rog!zDK+5JcrK_zUKPW8^y#L%$GBgMrlnDe=>EnN*(
z#`o4%(R+tbs@>ZBT%T1V-dpE%am0K_eSNquYX!P?k|lKl#yudwzv0G)a_i(9WYHdN
zX+%+r8|_^HR=KqmyZ2k@b?~?_Qn6b6+or89i*o<$j|duY8JlSt>y6COvyyb~e-DZA
zDY2=UtEZTp`--nEby)-9I-2_Rk>s66Pa%#acEw9LkN^Bgnnd2f{*RG4vMC2r8W2bW
z$@Pu>0vm2HDqQ#_Wpq;ee2O^i2;<_qs)A9Sg=N)#<hQx9_lSuJfI&1iY!W7R4ml8#
zd3dlBxCw%Jr0fUH4NQz~h&*yQw1f4|=AIxLvP>Vkc7YmGT>qf{@csGW2KK@@UD;i`
z954(RI#rCtFt@ZM35@nMw!~(~zFYa>o}RlTDaFNes>|%O1;^&td;$?kg-HKX=>Q~T
zoiq9%hM2g@<M0HU)m`kjFhUnTyL*R_*7Rq#J2WQYmZ7LPWJ>THbziaQv)65Uh11Wh
z{uQ-IJExiyA3uL0M5O}ZtRP-Ntkcpqq@Ze@oSdZauj)M>qkiCDS^(X}!_CN8hi3{Z
zi??rMR@!*?O=YoyhYgq_8)@_HRrDSJM>m^zAdt`m(pYGg<Q;$R_w<w}Ffr)e!U&?=
z7cK$of)>awVyBwL(+pWzCl}UJr^*J$FRdU}wer><s8Tfg(y8;@M|nK=g{IfDGafi`
z;sg?hfw&qQx8S+=A(1XKvc&`4<T42CFzY#xbG77-4JHlG1#I7ZF-Nw`#EM7t(ANuF
z2FdS<CL1p9`42ZX9!j1-P{dz<QHdE~C?KH0i@f2{(emTh9Wa^#VPLEpaMN!?^?;#X
z&<epD+t?UxaUY&Qm&wEbdcBHK3B4mZ`l3ar^XI$a-SqnTVeOZNXD?rlBPoe*w;!el
zx3#vy;Njq={6N06l#NP+Ri>)64-12x@j5Us?D?3F@ro<(XljA)k^<o%7YnyxKsIz0
z+qb%+0iGQzWx;F?#6QMk&X<;~y#1w1pR20Y5D3}=0fZl3x_ZUe+-wh~iJ|Kdi+dw_
zVGw%;D*$p#UsXZa(W9pr(K4YF=iuao#ro*2)NR;sU~o%9eCng8Xd2*8S<3cS@q>w{
zBe<heN2l(`Cr+Mk^9!Z#1GD3p@MxvOQ<u*HF($imw0$7c{Xl6C=H5_o{rwf2HtJyn
zNhjMEEuD}Zxc{9_Cr%?kgMxzms%WyT)1VqgktrAgfyd3GsZ{<rU%7G%;0P6vKwXjz
zF&+UP4?SLlNzSup^eDJz;*~E$W6X4Y;nv$;qnodO@Yv74iH>{o{5<*XzCvb3Mmdwb
z==iv!gF~l-94KwbxV;gAER^1U1NWvs{xXDQS8Dqam4;h0J2!{33~v`XZ$#V;CiMf-
zoJc$Q=l1n~=qjpR+Fw<kuNPQH-&N~J^V%yXLLD!wcD55$yCG>XIQO$8fEFO5UaZ_W
zEaKBMGn9Mr79$f2$(Fqm<PgYQd%r(z?kR(}>7q03IHKfV$}4dT+Krm8AQ}V{R5LI>
z!nBA7nl7amx`3?tlbq2CM2r}dJsk{ZfohJ3JjH|NL^OhCPSj7Qd{jC#Dd<=bV;DVp
zL8;+OYL!BcukEiAu*>^v_cU1c?URYQC^_-^y9w9mZAc3Qi>L_5A9yi9e!S4(OAMv|
zUx}5_yZ5CB1g@kTCuY`-c{ob}zX87o-{G32u+0!Hm)gvyl!=)X%px9u93G6|6jF2c
z=G_T|gOOwN^cIHK&EEzt<ZkJ@Ms@Z|h`~#Fu<8>*Z_&Y`5k+uWNKh(RK8yp>jaNec
zC)}6kLzQI)zKuhWj+%>xczsHzFHFz>Rv3h&Ty|^-J4fp9-e;|;)rDFI4~~tGQ}~m#
zNHTZLqVAb*FQ<*w>1sHWH^|uS(tp=*=<wO$Q?ny03uHT|9F?=)V+{_%RrYk!Z@J$O
zd4(R>PBGAd83H;EevuF5rOve_D~q+;&USU}T4w8D3B#&`vj0q;!<9!*aYcU}!B`e}
z&0!iv6^{WXM;o&ZpQ8e=Gw2n%9~ET?um>SoXU@QD?cM#v{S|<L|7X7`T8(6Thf@l!
zM&;j7t4P~2bNk203f)M3aou!@etEH<Lkr(#on&M>y~hJ!acEFvFSW|Sg^BXpxPzJB
z<IkC=X#pZ9%a|ZNktli<$<*^i6tTT%Cm~P6AY5cjqfx?#W;fp7h9^4Shuxkvj9UCb
zzQ(QZHRAup&^hHu6A+G^T2^|<ahVBBqKF18oNXD}J8I8~%e16v=Nv{9jk(^HmD$jh
zsOjV7_ztVkE@Y|Tk*Ys!T9?HiNLd}rGFw6m4)97wM#x6;SQ9l%qx3~0L4JNS?1e~%
z=SIa}`T28BJr>SpO-(sJ9)$UMtoZC2@y@a1;8v#hYGdQ21ea5*{O2^#SJcE%8-fnr
zvIWynky}CsVGYVuXLUIr=e}wqRFk+p2tC4I0cw;7lXx1SEohkc3veKu9)Sl47}wF+
zUdw>;3uplj62Yb_^zi>3H9nPk+U=;+*f+RL#3X84mkxckk#lE`_xe<}mMo*$%naXM
z&n_rfgkS-{J2^&16|qwAgf5Gmd=VyIgYq8bVaCcZMp1L#(LmCIG$&K@$!e$$+S+_t
z^PULG`EtK74Qj_M;vmmSv%wQUF#9%WWMA=Cf*v#^X1GO+`$__0>dv{iuo67aLunsa
z&@vZG_*q{XlklY|T8I<N-p4}R*iU+VxOl${oQD`4h&}q~oQ?ZtkkohQV|UK1hoKZD
zqE(l_8t8@3=oig?l$c_e#pVj*Sy<b%e?XZ5-~%#(QI!r~l#|y#(Cz9Pn;*=Mqf%pG
z#|wVa#1rjNx(J`SI@{}R{T+7qY02Z9DVSjvz>?UhS5IM(oxO{Ix1pEi2d8;BIp3h-
z|En&_KfU}}aTGRi)$>y9pwa9xWYwd;fTNI7O~zbHc3ss~vNUJ`;riW5T2E=5DzzQs
zwU#lM_*cd)Wz5WK?iW^8Ucy9e^GnV^>LAg8HLl%nWhr5KDrMuVY!bgIQ%AL^4<{F?
ztQY^aZ>G)FuYuzhS>?uGpk9RL90wcGiAjSs@YLA`WFusa8kodfy(VwHNCZ+pfz*Tb
zIi&v;G58Y-y4DA*l;YyTVi;UqA0B%jE1&5iENV<6)5OQ5%>6H^I3{9H5&xf9{Eh1l
z<Sw*(_HQAYFj&U1+&BCnIgeybaj-<;!9r+R3Xnc*W<&iJ3AO1|7bK(MVs^?$&<mDd
zV#hGOe90!_A#{hB2Tmds8*r$D$>{C9hXw>#vbp5+Ru*nK*wvvHMVt)w9aK8VJ%LJk
z+yf9D1WBln%R26_ePKV7%clHw!v=V4c8J6%Nea@iN#yvDZ($g<Xc`lAzy<`~9^dK4
zlS~*`_8Yq@0V(=F@bs!eUiS(QFQaJr4Ps1fngyo9evbwvtZ*+cyVqG2Ix7iKw)<?B
zxfGlHDs0Kd%1RFJ9t^NP`yp#h{(`tVpfPysv=_GEHODUUqT}a&R9_!|qV*9s%;ZC%
z`9%aGbfB=Ug+9hBe5tQMxZ&l-eb?$fAq+eiECq0wo&;8n@P!j6PeRjg>=zWMm9f%v
zKGbObMSbZ&oBP?`L!}A1G*Q~KqboE+Z2{=oJ_FJOki<YB2RCHl4|eYe)gdx6eK8O(
zqN(4KtF3L?YfpixF&o=}#<ax5k`Jku|Cmi$cnpeRQY(iZLvsd-{VWFM|IrcN;Gr3)
z79J6ibZn<Qx9-uSclPq3xIljUt5+`2(!+Xy51_iA!5HNhaO^ueH>hARxtEbfT`&Tn
z*@w2>xqtsYQbz$*a(OrYZwmjE@?oKOzFJz9cm-0%TdwjP<KP%;zntkVaVvK{O8_>3
zEzgZ*v#rS;DFUz2=*S2S5pnsDDPsEz4BvUZ9yiLmC<$bc<LJ(_-3J;P6%!juB>dL$
z^JfCqc@j7jc_Qr{cvS+?J;+uMh%bx>NFE#Ojnfhf5N2*P4A!l)Nyq|7JVF#hC<M>+
zuWkSo7*&7)CJ5S(d*v_jZ@s;JYL->~BfyLZ(2zm;)aEg$1tA-J>o)x8@(+kivYLl5
z90EaKY{C82(dYZ7;RJ^MbfrIVvwv;SyUSh6J-7YFCj(ShII^-Ww~FDbS#7AURG$h{
z?SxDvL=WgN2LXbJsjqe;c0K}c(jIw%sdWe;?R+cctJ6}%d;Pjy0#lY)`iaB(Us)xb
z5I*tJkdXjbP7_><aufp;C<t7Z`TMUfK%+^7l4Je)M8DR5$0&b_9AO2B3wacq_-U(5
z{$~LJYdpp6ocuP@1iU<3N)xbU0Mt+gL$tQV^0Zbqac#NwzF1uoo!<hNT?aG6W*f}K
zgNceo_1fh>nO>0lnCFUeU13?fmIR0-y-uee*@0OxYY7||(sO}5wK@^^RMk360W2t(
zpLlG@C@(+O&D$cE7@d4{-<d7{d@Sx&(`Hn9%oPga6}2HyupIIu0~t4RX@E1QNgR|}
zAQ8q97Fq%w3^L*Hjje5Mr8vyCWQhTh!Ayi*dC$kA^0h?SY*&_-y5ky6tL^!A?F!>L
zpMoIKJDcAu7(7fbRGBw%cD%y7NMMR7!LAPh8&hOe+%PHLNW6hJ4H9Zdys#Z}a<4#J
zfwPq0lagxa*?cEDgz`%9vu*3W4z|5x7#{AU>F-rmqV_iCTI`c`nCK~bvPn+i{roiP
z@_qH9HD1i-^&4Qk;JUE+!BafhW@3aR%s{xaGAbLPUch?Cjbvhm2M4i8Y@DQ9Z2mi_
zFuiU2zW>ypV@6_m@YMMDZWXx!U8g?NRq*Q<#kuAV6P(vjSSX@0ZLu!`0;$Kl@+&E{
zf+1q-7%l@=Jc87}2|vDBr=F5tbG0+*N#KV_)Au87507~FKRiTj`MLO{X$kKNo$Atn
z=mnYWD1(5(;Y;9y%?r4WTp2)w9ju^+bb1dw_D?X21q+6U^~Jd=CGfuhb(3&Y{wI@m
zP5$kFHRae=dtQz{ZWWDwaQ5T$Dk06ocYo6uCSmB|1XM+33J!+MBtr8^jFqb{?^adi
zB}mL6lW!vdJF0=6GKj#4N^h{VDZ&@8@zAqPvtO2TlY&nq=OUN;AHkbf6P6g|xG^X)
zTBqs|kZ}ZIK-eM>{uDFvET1L4PD(nDw*JTt#?=t%*s?OGZvHdPdp6pKUyz4{6xKT|
z0C4Xp2aw!w^AT8Ed@^97gOU<Q%Z&^N#2eTb&Rtjf0s05-`SNbe#h|0Q3b&i2zU@vs
zAJS<Jiyn#P^iRU|oq!SGeoxWtF)*Mt6!0s$lPys&cGcGbT@@x)K>^+fSt&7#pOS;Q
zeBcvTV$hXTrY&XLS$~LzMn5`I9IqTP^p25O_*44h^*SDvt@1|5<A`V+!Z#~p+D8YV
z8NwkzmWf9yFy1+C^y2!zJ@T$yyG)2^C5+y?__=ZFk81*J@W-22FgI>c5YgJ-pyojx
zVrr_Yq~tx(QwfL=U`Obm*I2aXJMSkXm?FLw==bu}0`5k1LfP}vyv;?J=hU!r7|^c!
zmmEbT=1T|*^JmBcwzkheQp~^Xky&rTKU`Tu`_*O|J#&rqhTUTB`=OOmK07|~F}K6l
z7)1my5@dZL=2@i>7yzo3JInp#=pKwFn@@l;5WKwmlPD22mY9F&{<-<UibdXM@Oz6K
zSIfe$t-l7Mm9*2Z2hesczvsS;ER+s6Uh7Kh>bADHV?%gpA=&f_cJslQzWn@X?e+Z*
zax|9E>~lOYDn6Wi)cJkIBqcgHuV+f6!odUDHSiVKcaD>Gg&l~CD-eN`ih_u#!O*$T
z@#h{9BFv)a1_zU95C2~_P4Za}oO2{`JLniyUl_jpiOOscsm;(q0sul(Bw96|h;e8t
zR55S|W7Mtc6%Jf0lrZt-h`)SkA~S4VQ%=oHb5qkm15{jd+qqR3Wzw$H)OU19Y`Af7
z>znZ$Ui6v>^A>bnc)ajbXY>C6$T8}26fuAU!6D&%i5wFPdKw{4OG4S##+F;dP_M6j
zu{UQuNbIsAhQIx`ymj@@>G}ZR4F6#qrTf!IJ}V+C>xu*LDAcJ`6!;g(G?&u%@6qD(
zzy;rSW}VmU(pZqgPpZ@J6!f9j<n^WNX!|jL<=9EfOW5%@tgRj@i#U~jLLfYhdKgPe
zP?qQXE_RfOkF5}qmn<t4qlJ?foB}EiOi#jE1@5iV*xan8a0hB>uM`(7POzx2PhnvW
z!}PhNF$;v3%U<mY1^g%YVld`_H&aQ-@VnXW;QQ0(0AN-0SD2%DTKs#0HDCycP4?5<
zMd(mfo0CtE`rn+cJ!~{o%~AeTT(M3rF5bQOPF66)JW&W<U8b-U_$Pb?&+M>evs%RY
z?5-ZIro8nxgLP=-<cbm5^y9Tu%>9e(F7XLbr&ZQaBbsb@0Ml)Uv2hBPhaq(e!x{GS
z@ogQp$T7~<MNo8QO&6VbB-i@&CFYV~CVc>&B_H*~9B9maxw036cnGLbt7<$8hf30m
zS}vf3j*00asH@>&Wv?kKFE542Dr}$VCthS^V0cE~-PivEdM7S~4Rqc(f*sq<PBF3M
z9luwy-TTk##2b|s>m8)aqS>%_#)oOA;?L_jHfM^HbDRf^?+v(|YO#)s{xMMfN;kzm
zrDbZ^rEDZ!*EZetmBNX4tj2@4cU%>{Dw;(@r`XKLPm|^A*P2kf3A(@cpXZh~O|~l4
zZh3G$wqNCsJep|K(a5|#v4a)5C9pvlq}zHftkN7_5Y%B?>CTJF!+7S0k##cCaqX}P
zVmHANZ0Wl`zNZ$^WnbCb3^9VVJ61|=`6QO2v}?$5B6M9AYtWjLFNzmNOKExi_Z{m0
z`Ee}6Z=cO?sv5gjbF+J$vo*F_OYh=h`hn11Sf^kc0}lV=hk|+b?hUUZn1)feWtWL8
zUAcQ!YCD8^0s@o-a5oI$bzb$BBa#6hwUFpNwpjD`_kFAb5&-=eauiUWj25k!tkl8A
zTVr|`Ax5XO%qy{X;}yplj`^m``bcGh`ZePBAB)dn4<2Az;kI>jzO!Alzdboa?t{i*
zV{?P1oRD)H!q~p3RAr-_gcabZh%w_XnX5mxFR305NCtZPYP*2Zp`mTBd5~M?NT{B-
zmd1p;^ANkbZC4`OK;|_b*90kOX=evbyLuzLzHEm!Kb3WH(X3fA&vcXLgNlkwX@5Ok
z8le>Zf~=#dcDA-j^z}^5=RzvY;d>6dFoSn(iL)=F4ple=lQ@(-mEu<?Sp$QE7{mNZ
zn;E>$EvPB5>v$-|<3QhekCa#PdiI24zQAE5$zT$sC;ZXAzJ+`F5LZoFhet;2WE8RP
z6cm2kv0sGT4QeJldt}dvM`A83)IOgRR1<o7Qc_Y7JrDVgl-S5cB|ju;9ZHro*vYHI
zg8PU#8!VY8S--TF;nM#TrrBsSjk1PG(B?gYhE~yQzjAX)Q1AJtFTqsgsG&t4S}w9<
z2OxNMc6Rbaot!jAX1Q;)*5pYzBa5JvJbV~-A_r&ZU`PrJJa!8UF%pQjmG;<#O$qu5
zT;UE@^n8%5Vqg(wI6#RJ@Ye3}v1RB%(cIv&P2psQr8}je*-j=JY6qF=M|c!EtigAL
z);~Cgtmoq!xr&1Iwx7&+ug+p5J&|2|pp`zpKp7_QwLgT6BW5ZQ+JbTdd@IL<moQnZ
zM}kk1+;}Nb_~xr{Fc4%wB^+Q|?;XMv(Jssc=?=%x6IcUI=hzt?JN6ZN>rP)Tc_-sf
z+HyWnxB;T!klh2P64|2rYU{oZht>*Y1x@$x&=6@Z_9}{a8Z820_9WCrG$&z%HG-}H
z#7k!hVUpGuui*L{Pgs4~7y3>7{EHPo<>($LKjRY;>|0qQum>PV^6N9Vlm-kUp(1Dk
zcK>V%w&FxW8px&Y`>8=sAn4k6lRorg%lV@cyZNs@xp$$;tl>$QZGCMFZLN}2mI$Pb
zWZKPjnk#$jnw#Z&qIO^WqOu*bpPz9O(jnoZ^SHU-Z@P*2ueoaotP*p{+rlV-@*TP;
z3B;3-V)Bj-)6`>l{SbGfwe1ecsi4@UiB`mU4Pzl>4`O09wY7SQM*#Yu!?u0>gH%lZ
zcwM#FuZN==(rHEl7mT(UiL}rh0W8DRv7(iwjPY531smR7_V9p!;bpTOX9$K`VXXKy
zwX>j7O-$k)P3Li!wYF_w4vL769~m4>JF)}d3;qO%b09guMDz>3?hvFUI$ozxQnTXr
znPeLCIC8?m$v=TNR%z|G=eIFeIxEqqNpSe_y0&V@AL}3J=ba|sv6Ckw4{rNn${w`P
zJ1EC}8@<7JR<<y-9>`B6HNm+TTMqQzhhzpeSQAJ=@B=vi#Kd$~QXs*E!yNv~eT^4x
z{qv8U<IfL3Pe9ZoLCQeLG^GIzW%TLYii$C(7BG8N726H6CdWfrH*E?ZhZ?8ObPN=K
z%UkyhXh&>pB$+$G`NO3QEGktCu`H78x4@x9Yh%iQ1tx}f*#jV~lPe~8_@!^Rw(?1^
zKtRd7g|a(#=AKAbF~!&<*NpVBh#7?PW+l9Dx+ZO{NvA^tjsTe1<BSq0DW+J$nZ|G)
zKidc#<InUoMe&0@wv2AqEzQlLYpycS!obP^BK|-UIhFij8kQrakq~Z=1?4P&7B_eI
zHDu7{v{WuyQ<NeY3S<{{5=Z;1Q|VASBi3=>8@yW>Y=hj~$1VvJ*liI|v#SzLaeNVy
zY+*+nQ{CgM6iN?R!TOQ0?{{i|D1(2{l?tddDFNCmdOFX+Jdzxv6RNsQN06cF+cJ*J
zFy-eJmjpmJQ`}YRFeyk!RcujO;F)=WA!=mKTr0wjv<lL#UEAd#2W>n})g>^v0p|7O
zWM0_*RsA>L*uQOOj*LG>gHsU@y<j5tX<KA|ne#p`Nkos^{VVe0xp3OP;C119+`zkn
z+YV5FoMe9HOx$vWi-HfTohP3yeoJj60Rl}WE%X)LEb5!3g<IcUj{Fh4gOwhVmGS#4
zcU?w#q^13j(6zU(+eH1u_<_2&(ueXGRSxSfpZh1ZT3<j0QT|KT8f6!pm;xF@;Fj&Y
zh3OK7L#lc)^#d&TVNo^%+rDjn?WLswQ-YbX2e(lY23!u>ZOrTwGLHg{JwBY<O?Y^D
zG3QgaAwhx9UlucXA&c6+;l`02J>wqRM$HbYTTpbWkjYoCcKK^YsG;mBQQhsIaPC=J
zm%U#=Ece52o`#};Y(c^E@2Z<<(C!l)#x-mMi~jK=My3@(5D1#Or!u^^*7o(Otm8qa
zB6GrCcTa~C?O)#=G_Yi&qk`WA;mB})Rn+R8FT1e@NT6W0K0_yL$Li{(nNabZNasu!
zd1S3F9*MDk6nimJ5>zToq!_A)Rz%3lWYtLntcl&*XlO7Xqws2yX4l(Mvr-x<Rpq4z
zm;@xqUHM1?F&Ho{P<Ug2X6h-+i-cZI<Qo6iahoq1mCNw@vgC^*jL&cUr-sclY%GYS
z0@$)^S7uJm{kwUf)1(_i-bZ!<aRPlGPS_ddMbmSTCJ0kGBpo2o0JjjY9hhy9R{Qms
zjLcSiXKU=WdG1wqj+b|p^8yrHO6hfyUwwdpACY%A9{RbOykGV2d!10%fw<OUQ#SIo
zd}-**cI?wvOMCm#zCMgp)!OoQyu>0K{}Rl~VVWV9WCS=09OCa30+{9UmP0=Cr+p)z
z)J;34&X@-ecyQ*RI?|73dJs(}(~#kQnwt0{B0lBfglrHd?c=Wi*XBM^{PDH8zm!3O
zQR+bY_2HJpLKB+3-<sgW*KxIfzj}QeFkg9SNHnqv*VjxSP*3qH3!6-g7%q)M={yD$
zJb(HWF>bTH0RIOclM*?!u{eZN8aVWt@-7I@HB_4IB9YJRpG=2VuksD9vcr2)XTv-D
zAFtpAh+oUh%p;l3z;A#c6f$~`4}r(+?J7!2ttBOAp-O-D?02{(&&wIx6WGeJ4spQK
z2LjZ^fnQp>6&`v)!LqN42I7$&PznLb$jGP#rwaQo&P&wQ>Mv&wJO4`yVDe*PvIxEE
zp=2ij(?}%|5(=vAhxHx1PL~7J@u(Y#y~#)0@JC_EAX%RqSwX%~7F)f0H9oeyuxpoD
zUtV?NJzn+-28Qm@=B>NNnKr~=ot0gUi#r2>8R{PJ@bEl>ZZ=F%)q^$xSR*FOVA3W;
zx`VtJSOdZsqc1>mx%4p+1d{V{PKH;S!v&6SGarQg-;r98jO{}mR#yL3HwAN!c|Ya@
zN`0_5hh_|g7?Fv&NGLAOO7{g=#9}~c*2{A^YY8+TsDh!xK<zQ-U0CvGn-t_paK#*S
zK109O9R>qvxltipiVbMP9qOrAKk<&1z)%mc<M`l^#S|l;taeIhdE}syy7%bOF(iU3
zLhlE2GNgL&&BC{RkQwnS`KPfVL0^xgBRfCq?w${84K(KspO5C)Z$7FK#e1tAm0iGS
zcJgO1ZP9RPgNBA47JeO9g*XJ)KHPI*k1(ijQE_oJ7iid__ext%`G!iNQV1)1_Gt|l
zR32TB0K*-CH14U%$)sn^$SVc)(7nl3CZ?^aX$J<#OGc-qiA)!qN-spMv9{BZYb(mR
zH8Rv=@O--*@If(lz*@zsjDNZx4qY_86;jaRP^IBrxKBHVBDE()Nl6LWU`=*1AqOyo
z5>j)u#7yMCB2R{I7mYs#J>l#}mx?VNH7Qzu6gn7hfCOC3Mn(?&{ZmXl&2t&2^B*<Z
zU;qHrq2zE$WD8*M*oicxVS_6`s5^o)WZg$j)6xdPQUHg0`iU46&#|%3SEg;vBY9Zf
zy{l`s{%v;NBkxE?;0B!(fQyF($O`}Y<y-4GX;F@Yl7?Cg6|#SLU+o%yWICb7fJg6p
zLqj%HX|NfC+`f0{_{o!+Io}W`1Q_7<&uYt>2l%e8kX>VfLE}kN^h;<S{@MQs1rTN|
zqIug8Vc>2L_#EUL${{EQ5WE6DeiOY0h$ILI5l1eG%Kj*b0*KPJbM<3)qEZ?+Hy|X)
zR(5-omB7uqOe)HKND3hhbNZg0aC7ve9l(Sf6bPc#EJD++&2}>oVoqV=RdVuh+uM`d
zHh=);(9B2n2QVA-7(F{lqkDnJI8;d3gD|B4B1SEf;(uy08f_xQj1MT2esesD<vB|t
z>`*e0d<Qr~<)HxDpVENM0Sz3AF<(M8>u_k$LNtT=O^{f(f(<02OT2v@4j(>pc|kOq
zahEb#{$N*wIs*2c%Ab=J6a6>fIHn139BGw@&jecgC3WCDzCb25l{eaew?^_Ul8sh%
z)sbE*5WvxQrheIpt_Z&Ys0Zbhy{f`@4kRU~q+swB0{Pepd{@L9-H(e?eq%r;R$j9E
zp#URWrQ{EwNt#%Y7KuNK3>+m1XrG=w-BD*NAuIZ&yS%1GY|oy?haz&TWe_z;)N_jg
zkVqVx(uLjPNJlBuKsMcT=v_8F+W7UmCMuEFf`U+^sPlwfFO@OzQu4SQFwi1`g>nJ9
zjs{b|cN6aepT2oxG%~{e4*L}EaYKkJ<zdCVZ-AVBrO9e36x86tO3X41pF?O0-61LG
zwUfClEof_nU4<lz&!0PoZ2=f1ET+(%X8JC9<)3B-ajdg4s%Bx)hZ=T0509>n1g7>v
zj{}*)?VC3b>FGfXi<l+sW9N4tCo7)PC!0J#3qbn^DsLWo!XL<E==&gGi?RR!ubIKQ
z)2F%Rf!AHAe4(Zg2QX3GKob#emcNg9dA%8@R|V!78VXhi1tN9@=xjO1=R&F!jxMl+
ztkc=@rQ?a*&%>cWnhqWgxJnsJBVi<2mc5Uz#tP*!`@|e-JlS@S^A+mYb-yM;F*o5~
zViv{8LXeMps$MfFt{V7-&K86;(DcjtX+ReSRRz0}7NrMT$|Vy=VjYs6M*O*qNy&O+
znpo^(Aj;9gfA?m#mo3?rdLTHAjGDo|lZu_DrY8Mpwg+p03GEkHuavi_`^&N(`Glup
zBA@=JqRTwsXPCC@GX_RRQIfN|vHvIBU_U3Z_i`zzUo4V-(P~loS-VO?SvTJM%NJs~
zQmEWb5bG)grd|wC;~*Gq`Y8N$Q??%e6|$m4Ue63pT43EX;6Y$;NCW8-jaC^QB;Ir!
zP}+e4@Cn}!Ik~@q=ANn_><&O~ndB{isAF{2SFfuSfF^PBefa`SAtJuFtDf5#QHPRJ
z31#?K*~jZ+I~ddGg+oS?H5&}{O8OH@12#<rWx9zyKeb03TYlm|ad|la0$KuVD7&YG
z@d?PaEay-6#g<U$J=tNrW)`gE@0SOXDp5gIRY^(ltnt|Y%SB<QzidWU)|WekimK`g
zj-1EY%yGJCHZEj0L+k>m5C`RB=Gfk+J#Nob>VCy{?A+)v(AC^hnre_~6&~$0RowgD
zy(6R>q>$AoXLt9|+qW^e6wm7!8LFO-O%xy{pla`W%<_A(GX<v52lBXAjvb1e94ZMC
z_YP6;0a6kcuJpkDT<8kog<Kkm&B#Wj9N@>M6{iaG74}Dru*aMEqDU+gtnF78%#~oC
z;{mDMLlIQ^vUf2jrueY&PHBt~;8WxhH<eM^pHu>Q#KH`P8%er}Ttxj-d}R#|B|I8<
z#BMJ01GB&(Vce8VjYiGooGl?C)uKtyZ-6?kA^=_Evo!@*z+2!(Vtl))slAGRW<dcr
z0j(iCN{)MVzr6_hT3^2&VLXdU*0tv?Ej68nG`)RWbUKsMJ4aa}#Y>V8J524k8yRqr
zD%(=sW~W^nebNth4jmz&vV*^V@r9atbWy_*G}%o~Du(}b!UGJ6di$)pl>+o?5FnPL
zLq%pMJU`lk(1F7vBq&?R^yxdf;`qlJcy$PHo6j=jff8N27R6cC>x5dFc!&q;P3T?P
z22z46C=Vy|{%&=s%@1m6?Kb>?&HS;{xq}DkQWIo<PPSZ{l4xsZ_D1mDUWZ){a!Hbv
zEB13vfT^VQ{;{|P%Nvvv4>-`XjI_T~yfnGH4tOIFM@viDo+XuQWy~Q5;Gx1N;nxlF
z*@~9!oI{XdXnl0OG#n`h=Mx<BPfKz5!%(x$x$wn{OAZco)oZoZm|(9BBj<7BMAdwB
zda2!x^`+q=3?Ks;=P^wH&bA0y$Qo7cy5}y++C-k%*%hJZUlr(280M|J6H(H_#U^pR
zA*KA>4?3g3)@M>l@7~4Py@_52oVOiXW5f&#+E8NUuT%T*G`$!Yz~Vy3N}*%-0o1gH
zqTk>#h|^7LEL@%+EShi35Vh_Nv0#4YjV$Zf=(AW_IBNj8f|bPF7*w`T4jj<X0B{1z
zHGr1I!rUCb4-wXz*lF?5kWsGi1#@d~tDv;RX1J6fwsR-`7x*GXPhrL3-`u9#>6HJ~
zCPu73xBl1ZH|b0=cK%<!(8hnY=_+Gt%A|lg4Z#p4^^UA5Az^{geL}c9FtDCJkW`#q
z50Bj<>A`Io1n7)kI{%g~vGNB%OnPqnQ4C$F9dvo)4$(=Qdq*2USAc%XJHM+L#GOii
zk(x?D=HG+gnhXHbKKB!0zSe^NaGPoC=-?vZh*Dh~TQjW`a^NtAnlmscde816xE0+<
z{Q`gx?&<MyBuwJ7N@}Lh3nBfBto%dzc{e3FSl^(!FA?-cA>zB76BR2xAlqWEgI29z
zPQh_y;PdR+Cv7(6$^?w}<xAjMeH>B^JE+EhMvXufxp@LQHc1(onO7!9U!axN7Q77R
z%3QJ;gWk*+dh3p9_Li1^{K(Y85NBSGnN{HWFw5P_>z*7meUg%h`bT^6sn-c!bcA?9
zxfdahn>K(9V}O!H<8RR<0GELh1*nR%bM|R*#LYlL2q9sVqWg>CeqOO06R+>$&h?6p
zzgfMfH^G_x!&%dcU78bfU@*O_xLRfwCf#R)(}q-Iz><^dn_5~}7H&u@zxY3VeRn*T
z{onqjfn<wp*(FMml~EEZBPCH;kt8Ehlvyf-tTIDJO2{UqB4vbZk*tvHJu`mC>Hgl&
z^Zdr^cm45weeaU%y3X@+em?K_@jiwhF}zCgv*<1B04;M*I2?)43w#9K)SLS@C8@%R
z!NCFrAlOzhve(PLeS4jd@X2-8{(GYFda1CMBP&MJ2mx`z3yH!>XL`0RVcSyaC_P~t
zl}V}H$t}uK<vM&=vPYfyHYmJ=2|GBoq)d28^5*=3K<sgN_`vAsd$c7O@Z9kgnb{vX
zf+G+^yNnDv63jPIc%M3b8YA^ePw4gFFXnLN`Kdz8VUdmiCJOQjF~vo5gqixlKmnu^
zr(8CWz^i}g=SQ<A7hF?V>z!V+hJ$!o8X7?Km{6c19}$or8o<w=S(_G|!DAIGdaYPh
z8)%JzTZUe0`7a^SsG)nxyr{CMU0rsVDaY3#f~Y0UfJuQtvJjmlQN@@|&Sl=%?2d#-
z>`PxRVuQ1-aQKB~j{FF0BN^~%7*JX}_Tf1wU6qxgEJw#lbuHutwue!X1S6YfdeN1n
z|JHF=;jznK0XaDnI%PQJNUMss?K>bC6$V{!5%Q}@K`6r!2TgnSDZJ!20JlLlDw;)A
z)d!QL5GA|SRXcc;63yBT4P_$DEdfml?ww_2h+vr#_9YOl4!z%v5dtihC^zAkK_0V}
zhKB3rZ#a&@ZV7E<y+?@34`_7JEP&W693!fFvm-cWong=zkxu67?yg9ZM_FPKb!T5>
zHTZQ!GYu!myOimb&Az?*cD}j!2>HZg=q?B^sn$BDspvk<DhrSVjgir5PSPG~CnJSd
z9XL}0MDWozugshr?6h>0l=werA9wHFi-<f#1wbNOc2pAcI+)$aaB87)!zxgY-?@7?
z&VmL@Ng{nU?`k+629%DdJ5amhg+65#M&Mh<E=Q6bI((RH@!jv9p5VuiKU?Hrms=+u
z0I^2fkEa>kJ4eY48(AfSK9;->wFF48By}w<DJ4-65h~KqgAN}*dUiuPsW^_ZYc_9D
zs6spm?U-zH!AMj_I5|^Ct~EKh%4nuRj6k$gY0n=XKBrUb2(EOVRk}7DE8(O@;^g82
z%%RbtskJWZmv`v%xv+;Igxl9{h0hguX|Owb<pxeCVn=rF-HY^u4R=fc|L6X#AbCBW
zg;IiKlG8?@wc0_@WBm`bHdCzdwb%e*w^Ju+F+3+L6uO0EUK-CPuy*s6$Do?!P$AD2
z{m<o>TX_MKp)2K&t5w0&^;yBot5nUo)=!gf#DRWG@c2RIhn;`m^mAdolPBp@E+2NE
zUaF9=m>PiEABNu~QA8;0*e<$bho|*Ykh%blB&mhN|A4D?uEB44#8~`lY7G+WE%VNS
zf6&yV>osblArxx<DfEX6|GP1tTDZJJhF;x5Qry3vk+jG7gSERpZV@TsSdJQHz?S!?
z&X+!`H>u~7C<iME!z5BS>dd8s0WVlkOtAQECc)&&;(?B4RoRtu^RP8UxF1e|NWD9O
z=Jj6n9@~S}n>spZEH*kjc(nrdsdz<(aWX2kCUH0H7C8Z9*MbWuerA+fk;iCtp-?3R
zM1)s&qQxpt*2G(W#s6Li<=igRqz`MWE-Nke<PO)b2Xe}8AR$*FYtn&|cR0;(D~qjx
z$AT#(wd8dIVM5vg=^7ez>_LHFN6L7Zcd}>9oSKOa3+Sr0ke_OY-m1ZPNyzCGKmzr~
z&%D8tAat7FV?(}El9kuw#n~X{akZzrk{bu%JQ7!n)<nU5`AwvBYsKrUi;Dy?Mlzr+
zYR3|&&8~2m#Z(!_4y|BcVq@M5(Dk$$qO+kYi@QApkvKCw{h_aGJMNVQ`fY`z4UejL
zX@rLJ+KQ>R6drGA9>8e9|H47*#8=1DFhfSDA&xdU`wup6qo)swjn&Fq*x-o?=Ld{y
zN=#rYw0miA7&=lce%?T1_r*&gl!uUOHQ(id0mGNFGElwRLL$+%gJbFF<OBsf?~z@M
z?8vR4yVdmSEYK)?)U8?i)fPD()6re5X{lW})4_&Z=OW>V+vo|3?cdRY5*6kpY?_c_
zLlTvu3(Vx-(cV;0)okgUZnQNlUr^BCTjOX=5wa=$U0b?B>eO%%2CjmsL+@kf73l__
zEt~GbdIA3(td4gVU1UgoeemuyAWkHJ<Q(uskc6V!(hLA#LB>j?txO<zX;Y{1=ksr_
z@dlRjw*^YFkD{(dX`cUWJFj^<*TH8=o90F(WQ)JiepqG{y&8(4!s6mUTx#<W+CIhk
zpVwK&7(q}(*g>XymOs7&Uf4eO8f5pO`WMnht~o;Hz!nCnZG}sEsj}8p(OBszV&Me(
zTc;f7b*`MV=-HiJP|yV@2QXzmUqpoASyc7M0uaoy0KP3@w15;So7O9+RS4AP$6rmd
zW-Ivr4@@s&55i6^v2$7KZmbp4BLOwccksLflRji*1nQ>e>|?&G8!K_9q>}}K0|3GV
z)^uP#pOLW_TGg2uxQHMDSy1qg(DXmct>0hSHyNW(5U9b<alzU~q&1>6<UOW;YfkMD
z6B|X}eO$UdmWo&M0WJt!MQWr=oV(KEct$$58W^y|`qo%rqJyr|VCOcYkM12IRIR|B
z4qof_RE{TkqNRnV2?T6#{XkhqLx#qYjO6Q_vZE*^rBavF)zFZ@^9{aNp%g&a@}<Rc
zmYK;80VcMx8Ujli3r-g4I3mo;*dyQ=TWM|!PHOASAiC2a<zYPaz=5Xc=g*2fL8pfU
z5@v!>V6_)s7rr5J$y}jXXd_f_n4^MS;=ocdb<pn%Kt+k^({O0UIa5&Zy@?beZ+9)Q
z^d<zX=6iiHe8yq{yf8a{8DcYysejgz2zv~B6`*1skL5`yyWG+6fGY;=CW>ejZ<s=3
z>;ztb*~+bZLW}|A9;*`8?c`nVz$_2B9294zXCWVMDap6#9<gq&WDz^3GUSfP0XKBJ
zrDaDk?m4}48&)E2{)=<BVAD<5tgkKBdcZ~<-*=x~r#$k4;A=$d&ClrjGV|#UvLdL&
z@o2#W2Xi)$Q_lxQ%}+G$Pc>qYCF&o5h6JwtLR$%t4RrlD6M&@k)n%p}6Br^VK|~M5
zu~f8XEMf+~)zrLq=s#{}XLsewE`zdyq8IEjS_R9}Qi^y2QT}5ZQJEmTCI*&5&Dv(u
zoHkr2fzqGO(sxu7c@GS!DG;s@6c`i-$Hdq?W!COLuJPkP7cD!_C*}h#4<Q6Z#RR8^
zLQL8K*#XqAf&AWN=22Y|dd+uErB8OBJWJ8s49k^T#j7?mv?ND4h26XTR1-n&28KKB
zg|hQlg59}uQHS|$S^V;BJLI4>uD5(?Zjv1Y{fes1JUEaTypaxNJHuR~vO$`JA`y~_
zl9ijRVL8vQMCN5sk|H#ILrhPC>mAljGzPD5MwE*|QicFDC#`J+<NCr1D%O+r2hl_w
zHln;yW%gO2+1g>8TlZ{f`uU0OQRb^{ZEZS&<d^odRSvLAdBl#igGUc!T1xSpo}N&6
z6&X^0i^B@pLMzSJ_5GZNXmQY$9dmcbwnV9UG?5L1Nedr1n1WS7g5N3M@V(@w@87S&
z0ERW2S<l2|$KZ>jf|VesZuC@5jF;*SIs(uwSM~d#<C1rUTv8tTO>85hyAzDg!qu5X
z<aXeTz|*$U{G78#PF5BJ3B}1|705I=&2BC9c&8lUA4`!^78IQLlh{4D#ok`rL-F3d
zL+X@8ljIfV{0Dg!Yk<G?4G;qkOe7EG)G&}dkKcTAb2O@C*b00Khkhy&EL=_pw5U~^
z)JOl+0K-3<!Z__*I)ZUHw`6*c9=%+qAGcmUo@OAQpo|ZO6hv^#W&(nh5KF22RR~Z&
zg|8=R=WVRz#mknKJKO%85Ze_Rgpd>;%tNiO57Ae~VdPCmMU{PXmEU&#8_L7U0Tm?T
zy?##N7e2p#L!puxIIy?>9H<~@kQ!Fl@8P>y9EhP^os(AAynp&dx4YIR(xDSumq+cy
zFVjAL640H7BDc>ChUg@gKA_<-F<dFMFHi3L0B29o_#dM_4iAr0aG8s?`!F~@PT@%y
z(CA;QZxSROO3~Fjb7sq?4Jo=*B)|ZqXZ4n#^63+2X$O>#7xeWPa6W8jVF~H_s3Da4
z`ZbcvI7^O}FD(-tW7I>{%XA<KB-=YU-0lljQ%{-3o&Z$Tw8B&%c4U0q1I2&!Q_aMB
ztX4T0nIOxj<-Kea6C`>@Mu;e0+Dw>JGf$uYQkHxv>?7Y$q(P3wB?}ApwUv3oK2xcW
zs&F4Ll&?7kZz?^oP}AnhIAC2U6iC3ArTDr!I(jk9-@fe;5=zd^eGJ)`zK0^%nixid
z!U??!=1X`r#S_2x(j~Z?mcP2f`povvGFtS_o=`%F8Wj<3?tue&kkIa+Uk0ktWb!Yw
zN%i;^?tk0u_8)$aX8$==n0u)3mG5EXS+Qpr6$lGcbX?_}*@>Vc$2W0wU7ek;fe{`T
zZ0pCRq6YIFyG$RvW`*@XlEsRDODM>^(ne{-aK8`b!l4~?c84#o3g~X%w>9Hx=bstf
z-8(%CI@vA!RW`I2c+JXo>RuuPoKtmuH#D<r%hR<lKz_b`8%~DUF~H!6idy=4y|#_@
zJ>e462TyibGYvYvMNu)m@|6Lj1SocpPa^|o8gW+W?f##jw8hwq2<eu8%J0WsyNSgO
z#>f(?vVWQd#H?SIFD@+wj?;mBK1?dJCY~;}ccT~4ONG`;z_&}l6Vc#(^u9rr3k<9V
zjoA!Ed<jq(J4T?AhFlkqFRBQhxfj6AkA1(rb>pi2P|DozXqKbD3v5dQm6R=h2T?eo
z#qXO2l`c7(12g7o0Y!*uQQ^`m?e<h!oeV6S9omALoOf@Q^W;U5(?^VTb#)*6ib5&^
z?&14$Ymun_qN74mhR@yjXDf1fvmub4T~=GW7D6f#E;b4dG?<J#j|(uKZq9d<i-FDe
zUOFk4Icq@LMD&r+;xrC%bAHJ@nh&s?=&QXh!|C;DB#d9Ps@wf{(S6M-)rp6sLy7fp
zTRNUtzwq~O=nm{hS=HHwzxp<xN}VUikq+DP4}_4`LwqpBLf+Z&_N&>sPhb4PXcg@*
z_H8HHG<s#aPoSQdhoX}YwVXiTS9UP9(rOMzp1QNO34jp7JQPpsEljvq1}YP7BcyVb
z-1)BPC!l3PNp4q%Ylk^-nS<XafM*C81X0B}p9#B@volDG3uV0eJ9z&H`nj^d+rWnf
z5U-|2ab~lL{OwoG-%C$AQ9PJ4iSZydY+m!N)JV^gG8Jcq8yN}uCFrcE9|2Pf)}O?E
zljqi63xZJulwK(&$lQJD;d8Ga^mBoiLLA*&rrrSC%dKZ`BX|Bv85I`L&5SAq7!O%7
z;A%`V@#NzPL5-#yzW`g12Qhu9bnvgqy5S2>?mL5q5)j6-f<WSv9eb`voi94QzR6M=
zY?#M6V$F<12p%3K^CQme0r?m-U|4c4{v@ZI<4*9Oa1tSpLzE>vEG)!v41=qNgE5RO
zEUGs@f@neR72nkqkX9VZZVhFRQa%q^L+%L7@54}q%`xj#PGjiJ&+jEa2(Sif7*uYk
zG=Vwq1=X*b8J`jD`)!mksDJ_F5TnASrB}MI5Iz;MJU%cG9u;+*JT%q{{@%O_&BSIv
zeh)KuEc4K==m3T0?vgeA(uqm7#dUxFdv#u_VuXSYKMvP2RXDgna1Q<kn&n-g(WosN
z!;gZ_Ghe|U*woy-Yv%8ruI*@IVq)Beno_{Uz(5lAhC}oK1CJj^#`BYsKp8nX^bd%J
zBZElz<0^#@XfyBL9o*Hx0*UggZISm?TXL&kodq5R?fcQh`j!mij5THKc9`d@Fb4G;
z`<DyAR}W7J$T$EyT>AYGARRS%FKm__3JKxlk~aubEASy0{{E6@IM#}uX(K1Cj|vNo
zQEo-@Rbnku@+u@UosE;bVTIC|{Gc*M+h{h|R>nQny4F^@h)nQTnv%*De6EVi%M!Wy
z8%Ucrmkc$VR7w4=N2E#-GJsIa(cO|nelno|ggz7z0LPdCfZ@y<!ENPX+MSCrs(>dF
zmhk12zkY1XCW~d<zWprXJoKc*awdrTiA7Fcxxsw61hqbf%zKTB)bn{Gf<r<WSy?0E
z;=15n0fkIV%=@Augg}E7tA1I=jc9g#e0GY8g6#u`t{08KQ_+P$GikZ)+@pQ0nKyPr
zmNCIU=-B$Uxut6T`^NvZOs3j(ehdis&%CSKZ;kx=1#=QMkcLgf(P{u)PMUW*5sL&n
zP#UF-w(^orrZ$_Cwfq;XI2RSw4U%hLB?Y@vz2$_TF%hT7QvOt8(=8V$O*pm>1j{NY
z&~K+u?h!8qGfnZY^#vhwBH#pQA*r~4Vt>mzsHfILyFKcrdI|%%*Bv`LbA-FXi7rAr
z@rXnag0(JrmH!$wy~^+opNZH%2?`!jcB5j3zAY9UI}9-}vc)a|0tIi%NyLv3V@m%|
zk3s2!?6pZ(#{N_0XucLj`J=<fskUv?sH?7{{yj>wwz5}7E{Y?^e9Q8%aqoeDE(%*o
z3uzUtwN7fYhR`%-#7I5}K~c1-g9fQ<ZFL?6vg!Uv^xx2^=KzAJ5e*2)H1#U1DceAn
zW=yBr@r9oKI_#12u8Ny4DjYlrpVV)0;0V~qu@IYPoukE7lE+|!AjugK<t&k}b%A@_
z1=pCJ1R1(XZ$EvMfzj<P{&J}tjFjDh9-QHAJ7%6e-u5!>-uMSEHJ<ye`V;fskGDAw
z3chjUm>b(%EesO&J$iPVe3mT!Gs#FoGjVUl_N277e0@JFuQ>xZokr6=37C^0OX3<$
zH3PdFH97j8_I%A5g0qZ_sZHn4c8jImQG3v4Bu*F4*7qV#lq;pNr0}|#>_3j4_v${Q
zPlsNf0>W{-C>~YvO~0skmBVtZF1uWx*4Nzl&jTbW+sa-H+5{sB9(WuQ`&>;;d%hcc
z(2#cPrDft$!OHITR^lDu#}$)Kwu}I)UkFo_k%9BjlD5*y<J{Kg=zfVfP}sd8CJ~`a
zYom4^SBa}~IL&H;n{mHh<9g%?N_1L$VN6$lVitq`S7aNN$KUoYjJ7=je0>2nkpb!s
z14$7@eMm4N{=m~>UBsy@Q8CmU_9&^|r1@3WQ>uSHjpeU4_Tr7j5Wd3)e-_wqA@)EN
zxyZA6_Uw?yA&kXb#@`oU$7|2HUq{WzKo%K!ZEj?OU)uIwcXw7CkzDiS$+bW9Ki`mb
zbaW3<+1pRFnq|5x)LAkqND!Ngrgm=qD{3$!F_w6TvH)EvxVI#X=ke0dG$C>V*!*?g
zoc5##e1FWs8ygwCv{bY&i#C5ha*q9{wiz3HG|d+*E;*y?Q6m-YUSBRk;;ub&p{}o?
zt4n+CR=}tBBk`VHE3|Xt6zMG!Cmr^4%5B=WPb!8d;2j3z5c&cUsb@MI`sg3d@P9vR
z)!=j_2;UXjeqWgvt&xdnCwC%CaiACC0X!i1LO-1a*nE<Zt00pXu#)=o1`OU2(F?-o
zQ>_<)gW{6B7AHw`Bq#uqHHFNar8u-=?h2kLbJ7ZBC!eTz3=$wSm3|gN-!y4Wvbv`(
zsUP33_-SBze!Pp01l0+|_dkP_+@NzJXrU2G9;^6B4>+e$w1ZlXj3aF&E(}3eu11i+
z5*U9!&rLGNm@O$uW5$u%loY5@LgJ3RLnhDO5J5h1`id!mM35hy+Nm5rfe{v{v>Qk)
zemEG3myyvbCg?*H&WQ)9<0mKTj$9jN8ToNQWzWh38bY(6->RaqHJ)wc{`P8OFz}s%
zL+a2O<~N0GoRjnOSQKYdbO|Dqmls58WlA-@lNDaT@OZVA#Ki&Q6ZE{L#YL15gg3F}
z!|Xh&l?6h%wz~2cV7UfOD;NdP!C(-NKOwklfP43YotT{+lAFPiXi6CnTA^N1d1vYN
zLLcVff(xZQ-HqDu@nl*tpZ;*gPwMFbvN*V$nJ)#dZ10Of$J=N{Nsv0v&E4!J193$t
zG`JjmC#RPq(fFefh3(t6nO~MZu=;2naMRT<*r0th69x4>hzVCESeEd_*fcLd(-n*O
z*OZhin4UI8V6(+R4M-Q&XehG-Mnrg!R|*GW-{G7C9T}&OsvE0vuNdEkn`2M^9NWP!
z$MyK}A$LVbr;UD_?d>yNWlcmnB<Ds8j*PfSADFW9kl16n9f#gJ9kJAZtb{G??X+`|
zK%(w4K)9(L&d1MWI9|KvjP+LO<r_<E6Ov~Pwihd5JMTLTAL}jMDVIgfZAdr(f7aLc
zpfT_6M%d1SpIIP#(viRnTv{9j{0Fm!bnp;%?j)@GUebZ@xiBc~ym;y5YxnMBbac-B
z%($q<zhW92W{1Nb^-bTYl^&x%ESG-Qi1T&e=Q&8+(YPY-7n}(hPdj9m$}~SlsSFHU
z=r#+nfj3FDogR+`x^!ed@`s03%WnUVzk)8mZ1%6`dO#mF*y+x(UxFsli4>&XUF-26
z?`r%<55DC4W%^D9%wF%_`p5CNBh?nsL*1`)^-^t!6qU?O@obse-bv~*uq8rVfbuBu
za&%+Ndq$+WHO4rxe&>^~yL3(AkV4O{-6BRGEccW+1uMN)k4?}(G`W!k8wZ+L;o6hY
z)M`!<f)jtsS?kxn{MyULw!>}<e4X;g12R4&tjuQoCcp58)YkTv$DOrQ_E~Ut|LEbI
z<i{!BzHIj21529itgP_AQWuk3u`$3n*8HPBq~^LJVweN%+y){jj_xMb_>*YLyHUr(
zr5fy2B686gOdGJ_F3SfH+DkayynDAc=JYYKLwgY86PSr~PdekOl8|)94h~gF$S;T<
zCwe{LOZ4P;KR_PD7ovCr!%^!6lC{^<<$Rk$@ucD3zC_~P=%}DKph=*}Gqs6&sa=Hs
zK}B_SQ--p3jFPD62p)G#!0}9?8>+DYQx0~7Cyixi@(K#D{NXN-$sn4ph=>Td4+E`2
zMMf&#y0^$=+36?ag7oG{Vl$-rARR<DBxbsJmNu$B_TcHXt=+Pmk4hu}i|WOgiUzT-
zlnqY{oB{soVz#fE5t-R*9{C+uhIo(<u+xwbS!&(!_RLyP<S$CXP#7Fbk_mViP(X_&
zKNO-%(UrR?T)^a~VB~&l!CqMVD=oLKg~g)|jk=1~skv)3bWY-0?^qw&wutWYd(PzR
zPY1|8RCf`>-xU$ZB)#2UC9qCH1B`a|MI#jcLr%|902I@qcLEjZr=1Q+&exzVTsavB
zj7gwhk&jPVlqFOcgeFvxFS}Zg{yT&O^TdnWu8$7$4JCNBok|hV7J2d&-xJRdPCg;n
zfub{nHt+}_fIZ=uhbqR!g@vicc)R*<^t`ZV`{2sCibeG9;Cx+e;0UA!`Fq11a6%lw
z`6N%D0ud1iZr;2fgU^>=IPK6pVeaW8-`L)Mby1qvEg^o@g?AGPaV}llmg-R4gqj{@
z($qqrmrD3=gJv6BIqQ0XLRWL`4~k&)lAdiS!n0LI6Ih=3VGf!Pa711Cx5Q8O;l!=U
zrmUOv+bixFXYts;fcEk?{r#u%F5dt4na50|dTe9_lKrM>frdG7O&bS6^+X>8*G6!J
z)tXR5lM83%<RDfWGy>!qq2}|a@Y@TNFO*UiSz_`>I%+_y_IvbbZgLVoh*YmvDJhtC
zfn*A}3$jb{>&SR(vdTkQkN?82$5R6qJlrLq-@&Z><3}YB#bDE4X%5yflAzD0HiuyX
z`sIs(bK)8nKAhk>IusNYJ7HCdh8bKnxb5PfV?h&sPALbufaHG9Fs2~EBof*uh?LMq
zLCqYj?g(<@iPuD5rfJx+dBWpB+<vP&1)aC6$|P8?-?)LtRdCn+r3x$AyfDSLrJ^M!
z;@38J-o&pbUn599yzlY-`^6q(%bE4O)BGiQZ;aN9ddokKOBF3$qCM9JL+<mB`aK-!
zci^&c)u26>9+uYjV`{F>`bQF`=YuzHl=|o7a4a{*X}C4Pyl5XQHKMZ7p#>m$4xk%}
zo|97}mU_z;2FgwAO7$=#W0{&$dzI|4SG=`<oxu+h56)8{sKds7uJ>-=x>Xaa5WUQ(
zp5F2g(xtPz9ejisfPa`?<I*^xD#SGy>bs|x3i%9P!bCa*1`&Wj*-10=^N3msI%iEs
zO%0MF`X3M~_<~t4%L5E%TeHAANB@K(FXb#dH}@(41`NK=?g@wJID|2yaS&=dGIZi@
z@GpYe&KwWh6gDM1J(Wso6SCl}g6CY4s!9od^#M|yr6j>WfWzynj6}-^nRC0n&)D$}
zuNJ)rT&WizEyY46A6<iu>%ccd-db!8{$IpTn8fCf!twsea$8FatlWXRXCg$tfujwB
zfT{B$O(H>fH%OpZHCZ)f)|1DFfBjM(x)-pe>Z++_UfKu6Ub#L?Wqi@lB>VGS+-iZJ
zetj`f<R42fxDYP7@47rd+Vb9Du3mvsV4ROow)y$-$LSz>9k^g>GXSG(0V4i1#2MBU
zGuMN0lwzI2!g#^e-N9l1Y!Q-4j}1bB-uD#fhd{6*Kff8svP-8OoSed|`e9IoT^tLt
zDO+jrH*^dT9Wt!gv4ZX^;0YNw%KSQ|#~~rgpRD29>ed;ulad6M>5S67%Ec>^KM*BJ
zeY1P7er7lB<<daP>(@g4&U)$RP*5f*Iitx$p}B97>rwah4<A1or|b0^Q<(b~^<#m4
zPzc%=e-h&x^IKCAQM}uzHdj7Hi^n-z;C-XOPhRucSorCR?+nM0J`Y(STsAKbyLu}t
zYNw>TZOi5fTzV++wzO^mDu2GNG6Lh6KkKWaOhv;dX`1#mEi3b~Oq4PsWA{&4agn*X
zwYIjBgX6`i8Bt0UgQORYgD)){lB#?|LzxUjD2Tu_hr~Npo(-LyT8!_g<%&bik%SIY
zi8)R(s8|Gvbs?gM86XMgf=Oom%IfYDJ3KyAio2n0CLi9|G~|RMHRVXyi}*#%iuXQM
z?x1+_?Fb{)1C1$gv(dz7PGGEkuWs8VL(*Q>uQ+QejcxaY{{;;34g!{+4iHDldaN#E
z07m9n-iy(*8Ne7ezwd{}O2VH!VUssDGNL8H&&q5cbwD6$I3oBHdsCB^&Z&V)P;hOP
zMf^;JlLkjhZYW*1l-y08xXXO<;>*_$l$Lw$h<BEdwPbiF@U_0@cu*BCF-ETa^#>j#
zox+<47^P0!;B1v?PvJbZ1%`x&o(E&ttB3&ehc7Efuh4p%xeml!2-f}bLZm^*zaabc
zw4|0aM+xnHUcWMqIPYpA!L7pj$V$j6@%?hmhqby9wS)N~Ej9Hr;YSW<jG46-Yy;^8
zd-CfOl=+>Yx<<vs)bu;drq23@;UGW#LW_C(c7%WvIK1%lQ&-^AiBR;nj)VHtM|z(@
z;o*87-qkqxF^U5jYsZrSU?dNg?=HYh4o|CatTP`rAFxYM55T|@@6+@_(+ub?l3sDl
zsV0KPc?E#kk{X^xI1CcZq@94mu_a&vjb{Uz4sv6jF1LBZ|H*gaa8y1AX{)_(5^gjo
zd)$G-EPl*X_fZPHcrgGuZ@|Z>^9Lgto^l$PHjBJ*???!~az$L1LR6CLnhYl16CP_6
zv$J*0o5#lJxoKJKhIl)=&xkO$2?~T(5{!?ax$(!~{+)`|<O{}b5)`OH?qzzx2C|x`
zO$lH$@wRv4ttw?_U8;gftr|jv^)|G%_Uut?iKp!D=H6ywiDbQ@^ocZffsjCrww<%@
z@uE|NE>|B6eHi3g4rwo-A<w?}1!SzrK#%3pvMQP@c>1w;FtCOCR44@nmKW|y?D6Mc
zD=MC`(5O1GvQuC#W4<|hQ!^Gwp9Vw)uV{02VO#<4E2v$YYK0CY>!pITp`1YYX$-54
zgWCl6OA1;}V4t|CFb~CG7!3<TJ%Qvvpw2l}ZIu_bOnF8BkQ^^&J|4gWu<0!ELqq^B
zVJX^D0NvQNF`l5O1ZUI*nFObHoQul~%^K>futMmZ{({cL->$$(-8z-TMqVNEFfp%t
zo2pQhgJg;>mQ;uFMzdAr)Irw%TRO~x6(q)DicOnX&sw5LAm&FWVFafrdyJgPs|0pc
zd9D}xY<J`tyJhpnT+}F$JB0<DtN_pi#|?=Dy!rQ<6$Z|80E;or&z*EgCRkV=OO0_b
zs)9kf`lTI^dkqZCry8$VjXa3YQVmhU;hQ43Yfm^~20?n6kpXG6#H<(aS`3=6-@J)l
zOfkAfs4FA@;|T~>Lf*)MkhpK3&YC4?I)!20-f9>sM%Q4v3gK*88bC7?7z6$N*#4@!
zp91s^Q5QfRt)S2W`y426pFBb9i{Hg0tTpl@&u=*npn=ECA02pn+1bSf1s%Himg@c|
zkelG|56V99>h0U5vn%yo9Lx_G8#@nh!L`NoLHfDPW~Uc6YSIy-y73cEPSU=$ZPxjR
zQ*<$3#!Z2WrnQw7ZQfFR#MeaLxF9rA>FHvy{TY!0^4qXCuGAt2c?9ZQb-bEWclTn#
z|Ngz9RYIT{6#6o?gjGBHWk{k>F_5~Ef8)-YvipQp&!f%w>p;b&KNCHrSX!uIpk{!b
zF}C`BFP}?ebC2Mw2_?Y{_|aR1ZR<u3PmW)L=8&<GITk|s4I4KWShxLx-p9%xLjd<>
zbR!C+G^6)F-o&0p77r=2Z3Nu8KOQTZ{~Mv}f5>lFwqw>zO`cd(^faj=L_`yuJoC$|
zd)+6uW?QOOM@x?bqJTpQD7{Pj^3Q~3&GlIdAfd|A=Dx>jW1r{dt=$#DrqarZk^T=$
zv5r${Cdx`bm6l@20tDfu>R9py=OLyk9E_j?-utz*I8AhTQ2)V49uoXLU$jFWm@raI
zaMHfO1>3o8$WgAhNNN8q0%KU`&|JUHM7aqaFRFMc3IE*CeVrSvp8&0il>KUv1E!Je
z$T8w+J!rnjH8xPoL`}|*T`)WQRQ$=@ygVJ_%QQd`FjJ54?F4;^@N9QVz+jPeJP2q%
zY=-_LW0~K$E@Kfqe=49>Be5Q_fhm7tz|s>sO8@2JzAfFA@3%nU;1yMgo~J(HD0qV&
z-Y?c4OUpjR1=6Y0a5a4p7Df^4<ZySp@jV2}QFM`0vOUW$!SDkFBXN#m6mLNvldzds
zb=rC^-79jL-!s_NgIhNt%ZAPB)T#Y4GWr)T6oD<~C4prfoG+ccXCGl8rXF?}Y%C28
z|5aWr;<IO^b(pZ#Ax959Nz`(X6eC1@8S}s>tFDID)&)!=y{Ypj9V~x?s(~u>@CH{x
ztb#~26A=RJoQI{;4U2V0Z8+4!?imBS-hKZ{e-CBrYa0Ft(Na&k5DM3HF|h;s%hE>8
z1vmHW#c;NnPg}I+ZgIE{>h+&7a+a;XtMBAp?I{tpy!SvQzXTS+4virD(iG-u4K-kz
zzrPwTInEp9?;}`u-MS5M|CBZyc$7@Q4M+XIo$5X|29RjJ+dW6I=uSkT!cZYD*CPWq
zAJ7oVq4$I075NQAH<PeGS60?v{I2>h(u!Kw18cxnB^yR28!2e3<CMBgigEVF!Vd40
z)EpAbvT-3EfFpyj&x!7-gNOTlV2&Rpa&o{_%5@1iT*{d!kfULzDtBRUSA(AxUoho;
zoN{Lk4Q;HF?#Bc{JP%#|sg&x<%3TQG3<#jvvSo61_Q3S}pHr2s4%kLqTz~-TrA7R?
z^=y}~wU=-p!4UZQ5+<^JaNVk}KfK}-$po@M$Qid&BQ-mq$$+Q}zU#LjYs8U+^xu&E
zY?AC~u||QGCf)2HB*UZSU(yT^?39hb^*x9Ig)j!PIT$bptLs9w)7;!YK8|5FEs4B9
z46Rc9mC}}vb?777F_MM1FQ7rVcpj75Z?MPG+8U_QGpg_DVEv#6kkV$Yq10Vn#t&k1
zm+Fha!4627H<~wcTtF9+yu|v7<Etq1K$+ul8Oh>RqxzTUmTlY=5_!ILi*IR<HDalS
zh1En^P_G7M{03uY=gyr{Qa4dylr64>X<KX(hA5wM#nQE6`4%5UvY$w}zhGsC{QPc~
zrN6Xw<S}jvp3J989!81vNLdfcyM6awIs6c_#xD71Ai6$CMI=QRc*6;WxW|uGHcynb
z{{>Ake$aHAZPv%N)xW1wW-x@Ux5_&b7KM6NTX7Kxzq{4nygCgWhc8K;ML_#NK<pi;
z$>4%^AI>VMv_beAm<HVxoMq~B(qxAa(17FZn6l9St=)O`R19trgxxlVi>T><Y{O=!
z-V(nSFHF%06zAjN;vbdzHyrNOsl?}(c<F4Eb;FRHus}QXqS3n}2H~nFHZ0&Fjj|e-
zLxBri{GT%F$#L{ua1(%yJHigCbA96E(OnPs_urTuIs<<pSXLJmfr`$=#8gRR%pPEI
za?d85BT-VSyq<95AsoDL|5oQc*5=XsnTK){H9F1{P}Sc<oRQ=vtsK@py_j_LBnZJU
zTAYMKFp%hDPmVX&)rp()UYbUaax~uwep}d{L5N<(7!iQ=Q}?fzPM>bX-6tz<Z+Af;
z628xp^ky8Ls&ft0b%QhLRZn@t01wk{h`O3uTHsa#)d^v%-ZvL$y`LqPlUMQ%I&KIR
z!cD#J-@8{dS;oN<dm2H?6lIIkOER3_FjW_m?ZF#sa_v8`8~G-#J|!m9vYpk7;I(KN
zekw1&?57j0dg@MXO^r>1Eg3N#v6i&*=mA!SM}n4w6Bufyk`i7~QEIz8i41TK-a!W}
z?LqafGpWt^3xV4p*$p@p>czwjrGZ5-f{%(T53!iO{u@|gTT$;~ul-l_ZIcnI5+#06
zM&@MOh_zI7jfJp~5HW^Rbh~}+S`d2A<qcPb&6+D2`Y|v?SxV*!SJy91mx+WShd#=}
zD~#;yOK@G)sM6>F0S<m;8;i%?=MQqc^L{o7n?G`58jy5nSP~HV^~I@6B;=~3me}r!
z(NY_Vrf&biLfqN6BdRO@0ANszqgZ2opU8la3xUN(wE5L~+OPWW+^?KVmY=CRA|UYL
z_N<7(oUr-{FjAd*Jq=zo1}lW+%UtU&M#;N?(I+cU!k7VuPX5-!zHI=-&MH_ng7q#y
zW8m7n=C3`2e4g@wX-L`2dd^wCf_7W@T|ZI}v<>l5F$us6=j%Qhh#ZfyTb*+`OW?Fo
z${3V`taocP?>qPHKfEZ9AP4#YddZLPMlR`DS@yZF?<T(|5{tLP=ix_Nq^rk+d-sB<
zD{)nDQy7eGA!>kO<0cZal^UA1WDh|atv;avj0%tl|2~OkCfiNw;qPGL0^)?%5m3N|
z=w7bgYSAit70SOA>=}5iga|Iddk!tn-AeOEwa-y(5U#<f3Ib{&eUV&u^=j?thzSkd
zNTMb;drLk6s-*epPH<7c-`~99&V@a8NY_F$hKhhhtn!fUJ;a%6y-#k}3<2ME-5wUR
z>f$>3qVfGS40ti!!cs*{&8SWxOH}7eW~3g<NR&d)29U<gED?gC^z^*`R4eEej~&|*
zSDS&-_4skVt4+9(7~%r?>MrrZ6dHRV)qQYp-a&u^3{&wYmN7UB>m~vvKm34D0((Yo
zV*`05PR_PA1EEl|hXp_W<pMNg4S)Z>5gb+V=C7UW`;PgRQ&UmaV-wTUL<kAxY9hZR
zu?bJ3u{pJd#FMN{{xaQM<kaAJbi8>J&Mx;QU%wzQUGS-^YmA9X>lMd+(kety&#fj@
zR8=wc!Mcx<!=zj!w36`TjntTf)o+`-tYiSJ$Fn~TVx-Q=Y3+VZMP<{mF_^aI+NU4;
z7}a6tKFuK)!_CGP`L7w(%=LPBNnLrYev^{tzsXv=aT@LcX&D3t5GkP-e)C3u4b4Kv
zs?*8!$1GS7_>Ee~c3e)l06hBZRc8)b`GpXDNke3pz{0E-e-9`Jg_uyNEI4IAl;6J4
z|3Hfle8yt9s}TD$)*}t!we?@hqL&{I07eJ4Fxzw*)9yvg(_ztV@H6s3^|PrS2;wzv
zC?#O{7c_SQ$2_`*u+HBAj<rslfRde_>ORm>7~*g6g!etPjo3=iK(A94py&mmP5Uc9
zWYz07Q?D2bE}Z-Sx>a@epFhZ!s`aRvD85Whc}bAPfl1arrNsexHlSFHcNe-g2cRwt
zf!igbjbVA3n`U?!oJ<67nF(s)zDdy6gt--wkR(!zyvD#U7tyln2$olTKRHy5;#yG=
z)cW5jftC=PW)m&04E9ip@^I@`8jMfq5ae!izmPYz4t75woQ9xb!3VKH<6QRf=TLFJ
zb(w$tXD!RA%U;Lu)>`q$QtBh#rj~qMWXyaUj_Zi(`vkOJWIP*Cc~!cGHMCs5()O!u
zNX_p_d+Ey8yKH}yv4;9w`O0o4nziHjmJJOmhu-eE^dO#E!}RNT;Iz4Wj)eGK3CYEx
zHTH3Z`A!LEw%;zb^R9vV)v*yo8k~RttJ-Op-JTM80>gGt6On?T3%fJrcnASus)cDP
zXpBwF?(i9kIJ{vqgBa2sQBI;Bz)+MgfymaXT}XfZTJc&pi9unpFO;0i(#B?`Kis%=
zsLNHc3&=FgC$SGLDZRe)s&4m_ub5S4Lup&ugA8@|GJiIRXX89pWf@R1edtn&0K=&H
z`*&oRUXHzG(G+(QadCPdCZ%lY-t-=OerhiI;18}l`_u#&85jb~OGnAd(A*>@@}HV6
z_S6tE+Z|0?K59Bf>1tsi#GrsfR*34s^x`t)A0tRQIPjWnD@mN-Rc8M||D%ZAT_^z>
zgD|KiK?NDg&s1*1!UGY;xX3wcl2_!3)!OPxnu!bp#yyy6;Dt&C-SMW)n+=va=oxS=
z20;va`y5lYGpI!I0?!)oAwW?LPULB#q%fGl`7y*MVL`8R=#c08_ih-?!4u7%gdL}?
zuI`YEghe&$>>n-?iPTMNVoNdxY@h-S`~D99^r9LT%grR#zbJp&K4Ci29q-M(KX|yl
zz5SiXn!DbY>&WSWXb_(r>@nmE17C%i{tmug)>VA0b%S*=M3}X=pBD;6gxB-jT-|g%
z9z|DYZ2#yHV5Nf`vP&SJLyd)s82F$|Lv*5~E_P8QCkyepxcsfCP)8vQWwDWwk&X^;
zusWepc@cZt>e*r`Xw}0RA1!zm@QV;UXwi<LX|i@)Y<Rd61j2}N@iHAn6B73DOM)A|
z9!Wg3kW4<Vc6@NKp2unEEc)lj$@feo69Ig-qh}kVafY?GM_algNi<TUtmhg`1)iN(
zFvS`nT(R9|&s@Da44E$<KR<>8vw=V-FiZrQq=h*b&PV-|&OjOQH)!DSVV@djk^g?o
zt(&5Y{fCzi)~}bUSPU;Nqtb7cyn`_dx;-2$y=KT4bLw&{4-<!&coAH@*?CAifP!RM
zZ9@g$*!6|wNrdE;29<~>FlHf%Z>UaJ;e!9V3OIcSItclxk4944Mv^}RnD+pzAjh@-
z76NySc)O37+u+vu0;Kw8GI|7)aF?+{@6w)^%~ma{Jl>@S1bNND@vU_-hU|Vt+hc%A
z<p`$}5O`dRmI_>}Ocpmo?odX7BOhfBuvvzB=tR)1QL|_IZ{)jZ;|N&^n0mf{v<!nd
z@$pRhA~}k?Kr~T7(jM>1^ImGilfOt#W~>Jzwi>eV;%iWyMW`ljp`-*-<_(t=1~gAI
z96uR#K!!M%E|h?7fOUKpmUUB83C1}=Gcdo+G+e^$x^=5Wl^?s=)jg0{qRwO;`Lzz$
z1N+OG+yhHeAVE?=z&8~kW9XGQZ(yJqBlpAyy70EZyQXUPGMv5aF2pjp=2A`5jamba
zYbR8ofgC5MD6Oq8KN58@^v_V2xUFx(d!viDki^5T<edX|0N)ZNBktaV(jMRp7(W8I
z{Rsm;COrSWhnjs2=A7Z0UUJjR|F*@#Z?-IV98~izz95;v{|nL=R%^^Dp|K@%F+6jo
z8S7-cvTf-ZLeSUtQEcr7B?NGq*aSK835%CyC&0|W-89TLM<&CqVQhA&QqURaU&VnB
zga*yJ%IBVINJ*<L%O;V^vMWP{5tANSJHVu4Vq${J1<eZRUijnj!U1ltGSTmbB2PZd
z{kSUh4JKAVb=^M|`9{sLU|!bx)$c9C>fu)}U&5XpV?e-DRQG>DyeT^TwfOO_)7CaN
z{SJ%hA(J&Cl~wT0SmF0z_=K0=ey2^bJ~(*^17ovt+BvsyF?JOd;EAHQF}SI!u7;7N
z@8b6jBvxB&@J+^9*AcRA-5`^iXImE`HYLltG}zCtfo%Hcy>Bjz1=03hlK~9^X+0D3
z^TG+geqGe_>4d!V(usfz$SV_EgG7%BJ8!h2+uOwz3gRP_{bX=1Yiyh*gxzSpz`;V(
zd<z=@Hdaudpk4h2A`Xl(UMkU%dN;eUq?p;*yoY5qRI)i+deBNHxjT;!4FUE0)7v`@
z1pkq<HpS<b+{P@(+Hc;3*zRrrLtfyo=gWqM($D1$fvN?;D1^8kzkG%Lc{U-RbUQQi
z4JRk)s7h;VGeCex-}h<8Ly!1;CPG!4CN$uH2S(nN5~d9#|IHw1U`N6J3yhE;hd?Ug
z_zruv53-}Mw%p@%t@6%E?X>yEckRd*i=tsjO!1SVyT9sV*Tm@<5Kwc3P9f=IOh77A
zaCrDxcs${51Cqn;54tGLH<0?lOagvrV8G^G3I)Y~Zr2Hp!o814q~7oF&C^dqsu%)h
z4|L37kud4l*=3aQS@|o+J5Se0gD5kLF$K-(a({|u5Cf`AzL&oWMS>>?tZX9<R(WW8
zfoIFdrl`F|p>%P2t`=|%vfVSo%X_mdKc=PmKgocfm)WpCqU}kZ&-E=f8Hqdzd-9~Y
zr6v1Hn2`KL@kW+HK2ovzMc)}BJq4I1HU`jV-@jL2z(%^bwA2ZfI9i;{ygV9O+7_;a
zddo-L3fQ7gr~uEmxLMrYh~Do`fHN+>pn$-ifn01gIU5|7&|0IFy@W~@g)wO01|Opg
zVYrIHAfKLx?G7BHnD$}fgefmlTz`Q1c-RY%DUJ$Xm5}i8@>SWUUGq*-h+RhqhpqhV
zSz%nt{gDug!;S0Ep<+B5cwQ;wt^rj+@r4DuYu8}d+2E>vkoANT0B~UFdj9oad~<Oj
z{T-ejkWm=BP9kJPE~1RGo6O|t)cic+NGra64LSZ?33Y<oe21C}!pPBq6SkZhLVqEj
zg?lP2{5Nkn43+{!(p!_T@%o3m^2mp|Wax@Ow~MyT#@hO6Obp@`@X+JfL}VqH9Dotf
z_&}YE>Td7e%O|=9<34};W(%|#DrcfRLiZ2m6uu8|Z2Szcx8o}4wTO_Ij!+D$fA{yJ
z2nxrP|MTZRO6fZiAg;sVhp|0QRpr%Xk5?HP1nZ*Ycq%qBys#GV%;PwO2zq*Qk}aJZ
zx4rlc>WMYngR$1EQa(d65^FEK){&*Si_sS@M>yi$zZ*I+1f$#q9jmdS0iJW9sm?Zj
z$M@0DsD_DGb>Yi`0%;clIAbJ^9x!y&RB*Ks%cO(ZIpA@4Sj^BJgUDLz2d*ecI{1o&
z`J(e=!s;GS4NxIcTNGbF$AXq=h7Qj(Qf)5<Zp03$-*bBh!P>{nHn{)f#NzU<)5&O-
zTHD%?NtKUc<?LD5DF6>bO_*+wLr$lN8(O^S`o;z6F4s^WU`(;jzkETG#*aKQj~60&
z27hK8oF8p}(QJi054c!YB;QA{89e$%@!3(ar@2R;+&P1R3cM90uY^Qy#~6Sp;vfS;
zp9s1Gk_}sL!l(fijtFAK&`1L+Ptj#$h^r-dc5umAin*Z#nFLNG^d<nR;QR}g6--*-
zS%6hXH<j29*HTdO4<0-(z<3F27a^gc@fbxnxvkxot0V4wQH#Ld1mNI~58zIOalyfr
zI!qRh`=Fa+OF~hF2?ZjMFhVSFljY!uM)=pG_u@GN#A+sIOE#cwPC6C}K_FtSB0)mZ
zf2B=Ih6CE8N3?v|00kTz*hj~TTR+*-VXwxbsUNHI8A4A!I5;^m!TKd~fRKhy_Llqp
zK}PD#%nWC`Ej1!hU{Jxwp4AQkng?WLsAT22?J!TrkY-~Z+=o_H|Bl0?FeZk$jaczF
z-aU|{;I@s1+BiiQ;`ZEw8_Z}^89((-UUTu>7p&g*=a0{U(<La%hvA%^(DO#2noI2R
z=AeL7Ht@F9Vd4;lqUuje0FVJH22K12wd#1{FAO=;_=>TIYO0K@U<$*4@Nbp<k<})5
zHxk>}qrzP?aB0Gw#GmoKCnI^<BcU!UCB^sThq1Fbz)^##01N*7l_eq{QU;t~_my={
zp3HQf{N>Km3cqo}5$3=FSoPjHBz|kE55e5on`zme9G@-Yu&X1T`GaKuqLaR%VGZ|(
zgR-)lDJYQf`0LJAEIem47K9F9@O#5?+=}{&euw+Y^$iUUQ0wp^*@m(^5-tfjpKhRm
zkdBz|+zZAwiYa<HM;jw>!QKl*6fT)-&k_>u3cC2IfJ>YJnM+Gk6Lqo@m}d|zA6`F2
zK<L4&_+plfwDdgCB{aaSKV0zjJ#{YP13=|IGYcXmbmN#NL1dxU%t!Lv@*D|xH7=K-
zP^h8$duI=&bc}q)UIBqVT+k7HuTwX{D7u){3UfpZOwrWsj}(QG63nWAmLRQ>uuEQ<
zN1E*~Kqo4o#U8l+TSL!&4qzva0+A;}XeY3wou{y?h|i(K#DpeJ!9@yA`f&CFKMfXb
z_GoW>t;+C;q^BR^$E0c^_VLTg-pX~^r<Zvy8tUks+i0igJEAsQi*0nn*F#)73hcL2
zQ&?l(bUfHBfr2189BXxBj2@Il07iTO-*Y>6-vMVLUjgMbUQnoE8jdA`)scAq2pQIe
zZRJl=tPe2hP<YMq!9oZx@eWopc~Wr<;RFIJyqp{m$2d->%wlU{go3rBo33Zl86PS|
zH`{C$dAv_{pJf-N#Zh)fS64*^y^{*b7zA4=PE~L}zaJg<$2JM@m<SFIR;Qw(@_$i-
zn^GKcag1}J7Vc=khCOO7_|c*w+Ql6R8YJ?DG*re{*oUh3hkYs&svVxjz;9}5f9s!L
zzdjj7tnR*mbVn6TNLI=Q09^rvcxypC<6)5pH+*y&H{#3A&v%D!J=&mdL=y91m_ahR
zn+ZCi-<LA8K_Rpe`KNFqhL-GEIW;Zk<tm4f^tW$yc47+Lfdf4~I#}&b3+V7G%>oL^
zToGN#@P6XcR*&Ug9<h2J5{dO`wZ_ueK@pLJN3^K*7o2~*3I6m+_1*ai4{XkWpdTN9
zuEGcFO0H^N3`|Du<04MB*}fAN_1sRmY2yL-jX>g*#)dcWzhn^*hOlIe?)lDTUA%vP
z374k8VOl6%|FEy`OrS?yq{QQJwerBm)QnRqn{2wY92PC->3n<~b85hkptx0mxqh1;
z=q_G{rEe>?o`^Y}qFd^;1;09zGG~MHljC1LoZ2x?J9dqo=74l#07sJUqnl?<KBcC_
zzmbu%Plz)VVP2Vz%QfFZBWT0%0M<E_f0<xV9?$t%&09kVWv~DrwnCgQ`kBIM)QYJO
zyY|kh6?KUceJShdYN~GR9XQ>6eA4%9y+*Orzy0l+yMsNiogH-HYHDx)+8#f_F=|_)
z@?5te>VRYKgA}RA$XuIt^&6Druz|-A*7qeiR~X(GSbHGxE>c5P1so69TC<F7+1mZ(
zqCFa#Nu#`THC9dho+z7H|5l61%)>|k3z`VGA|RO5W|l#w;)rK@jo)G(2Oa8iKXGbs
z&1PgcvGK)~4!PsU;=g^<8RgMXk8@T2@}<Cb?#uR>s=cLM9B=L;WPm$RO$8!H4CsZ<
zq3}il!GKk62uSPFyPkmoH5GIen01x8KwE)n)e#j`5@GN6;Opx@Ej&7cyG_hq&Z5f(
zbOSJ_7~u&1-wW&KkqP;wdiMtZAT^%nwTc|q#sifavM)KV%D4S^WD;lc>F-SA;RB6U
zSGp7*`f>d3EEdBFo_3*d3T6=4H?W=H>Cx8EkYqLd2c1C8h4K}3G$GGuXqfb3U}L+5
zsUygU==*^+czAepyaq0c-I5{r(2j@$SHmu4VE-ktT(%{}g@h!yAY%C2!-i;25_g$*
z#)tjP?<HhqXUdft&f5j3dy{<R4yHUa$qcBC<0*?ReukjTnc3MG8GGe)y*Mx;u-pU$
z5bxfIEJUGD7!sU4b0!N>Y~i>CC`I+viobpp0ZZ+`fips(ZiwK-qlqJjlXd``A<&_;
zn;k^)j-HQL_P<y2ZXzmBU5QjipB`p(>(WWFE{`}I*g~4@ziG+n8Fa2s)3Zw-pgFH(
zV)?7L7o2ykr!w_Ff2N(>i4qJinje58q1r@U?Gmvhju2AJ%n4Ctx*h>SIJ%XA=O-xG
z|N7k+d-4h57}$*c7n6wdNjxO8ivB8HLxq=NRGYaJjw7S8nrp)jK0XTX9xLinQE~AJ
zI1PZ8ip~;<FWfA`s`|0tfTW8`kx-gZQL&FHel9PEEgMcv)LSs6MZ1jx2s25{K$%W_
zUf<(BVe*&QLE#yXltya>0mPtzQ;PsA<Fy1e9xm%%;kcytE@E92ZwyD~uAxD)<{IC3
z_&J}`e2Jd6#_@#f0}l$1^!npetLj&Ayxhrc!i|FERLX?tMt`)|#Nqiyp*TWa07)IW
zM~@0DXu|@6*@x><d+AS+8IsepJ7-Ufsk!{koZZ;1H0hjl_GTYT%WJN0uZf#byfah(
z3wz$Vn5d}R=xy<BVTD(Cfg4L2Jn^8QOnC7;F1s`Nxx<YzKtaJfork1q<Qh!Nm2m!b
z+u*2Xp!|M5uJtA<*evpm9IP;&ha16Vw6UwTsp)ZSEL;`imLYIhfifr`-jaS7wwknl
zDW{9yr#EyrcpJlEyQ0H;KUlt^2~>;ocqBVPd?qLEt(zt=25W;qh07?S3-DpqdGVMP
z6NoQtOhrUmUK_d~DmhPm@SeTtlH)_)_~qsE1&;P?%Eu)<k0Mhj>118pn#PKrxc#+f
zwMsAg`WVJrRGJ$BfS=Z+xstby{?M^weIxQe-0<l=4~7+9U$=IIG@2>+<W5((uMmI)
zT&x!+dcdhC$ZGJ$f<FToQz|o%S4x^1@U!lPnEux_+l~=!55Dc;J>xW<+_c_CZcp|Y
zB|SgkTeY&*K00Dq#UipN*=3yT{Wnd4(b1V|iM}zdUAuD4C=^yC(+!Tq#^%~^<mErQ
zNz|y_-^|f<0M<bjqZkK%P(AE$TTBd`q09xQ`x_l4Mx__CoS~7lFrqcNCu*Ra@zLE+
zYQFQt$KSCe|J=gsgELp&-U@AK9GozH>xdPy^lK~oYIB9B&&xFbi)IMbJGN&naFlAZ
z+X8P*Eb{0Rn<VhgZ$j~8#K~TN;NuP%7wj5on=C*+3zVOtWIzG_qq>UB|Ebby@akgi
zsxagA!kLjA$3OH2rd$-PXSlg@O~}&Et6d)mdB0*gZ}G|UY9jOQmv3(ulhe@TUfj$(
zNR4+8t@$~DqL@QP5@schHE48GuTqQKTDPk{c``FqwT(;j(IfM#CrnLq-3}TSD%*Y3
z)IU#Va`E<{;H%VkN28;q2S2X4sNBXgMPl+PWbn!F5;?w6lEd80bKmFo=uT%Z_>kuQ
zB_o*~InOcLrSvk)*hBN3)9FLS6LSRCBE)!It=rAj)d#LwRgMxZ#kIAy_)^%2lEJol
z&aQhRJ?s$=&+ktBZ1HDQi{Pv{z!x1iARmStcNhnuCm=Ev`3Rn2WcWtXn_fWMnB;<O
zk0~H}nE=&ZjH0$t;axEOU2y%d7>>lj?>?4zv|yznv1ZM0f+Yl!8T4_$D|z}~M?@^y
zroFOiT_Q>`rr<*U2_pRp!h7X-eHd0?h`V>K_I8m&p(|t$FxCkKQ7=Eg6X@gmC&)1b
z-iN;-O29pc+4}-WE=3pZABZ_CSj^)^(^{bTS<c&E<mj6?*!W%AdIfLf>01Hd32hkB
z93q5AfN>O~Ds3?4Fn|MD%NLpt*uteHCVJ(;g+v-&bL;pg7g6{E`eYAy-~)GJFli6%
zK=>aRa7Yd03r0h<jP4xg6X7E}`b|ss!-pf5BjZZ{zUhWbxj-GF!@}mgm6qety<+-+
zY8c(##}dMC4d3ME=(*{P&Asph136rp08<D)XpmZD+|G9}uYG<h5(vfP$Br(SNvzDM
z2}N%gWMvs-T|5dE|Ef`d>gkgwuPx17f$Q+{!sM|~p)5uL@rIC4@6>(7eNET5;7S9T
zBPl7#b@I!09Qxq)d_>|l13+24P^u5W{ix5hM2zHl2m36MY?D|&;ao(x)`l;eCyt@b
zfJZUhCBe<ca!zb$_T3{B5w`aQldx@vL#O;hf6i4}+|A-mxGYgs{fdMegwk^6b1h1e
zxILJ)NWTm9A>t!0mOO&;2tL@@($M_u-bRInm0_ee{YoCu83O>VAFB}m`?m{f)uAz|
z=CW^(I^CR-<TepY<-P$KEYXLNS>j^}d=T?mmLWz`@jaBZM^xmw_wL(Q?^uP&2_i=M
zUZti=>=bPPwH^rq@Y<deEe1&l21;;-2Pt?$fo=Ji5rQ)VRC6$YMK2D<4Mv0o>q-=1
zpi@7<oq%k@2lFEJBwZ2Idw}!+!o1Lowz$-+&D@Mf01zNZy@eOXuw!#_J`D;oNmygK
zRdqzlO^!=WWxh)r@vhRDdC&KLO#1W3bcA?}HPkIOPhdmUR+vS(437IpAlhQh!bD4k
zM<6{tzy}HGV6baQgDQj*At#^!zJ^K9pQ8~O%7-6Hmj2W5@B%ceV1dvj3l}(rU{RTs
zKf*b@5`fSWF);)os(bMJ^`UeD>_zm6aB}Qgh1<?I=$C8;kxpM*i$DYRm-ikX?VPeX
zR~WL#FhMv@;63eHT&w<%ci|V!yeEh7X~lISnTxL%{sh_vnx@nsrcSm;3=O$lKs&K_
zxr{PF#(81~1E8dur?QPOQ2qLK=>O;egA1Yp-CTsPSXsTj)&u9DBd)GRCwDgWuc5Qf
z`2+RF0U;sPo!cB_o}!@v`2e2cKuHjIvvNEXNSm3F5jR8Ik00M6_LY{GtBE|plZNaU
zMAUzFe$&Zk*Cr})8CcEZ7)_-Wl>-M<RLID3t=R;H7`-L@zi-B4zoq)8>_i4E%pe88
zAp?V>eE7<qPtq1=pgy3}vV-9}s1Go+hPq!)?roRKg=Eb9F{Lp)x`SRLR{j=1INYu>
zF*RX^?Vdh9t%hnU*eD9@-+)~LX9osAFbcp%t$Bq^6Yc)x0%-GLzJ-^TrY3^GWHBCB
zQMtG3Q}A3XBsdu5H6iQ87O*mE{r)!mG?D&_uQ-$sS5U;4tKlI^PFCGs41)sb6zGEy
zz648s$B#)Mp8$lzP1Hs0?@<*N9CYaR>GcW3gJ#d*+8rJEe&9+QUB5mJ+yO<RRj^`7
z44foT9}*7EsOw;p3(FkAuY@l-<_0VIO0EkNfFSo8f4Fe|JjDM<6;t6$Nlhg@i)lG!
zOeSiKh~^pG={iG*xd1%Cc)j33z>~q0moFE(z36ac;D$XuE_nXt!T&-aGM58M-AaQY
z#qjpG;WWB%p?d!W=I|L?FR_6J1wkA`unM0kL-rSKpu`9)HKl)_+H8fxWMqUBxvM<y
znSh|7+`M?X(MH~B>?tgz5-`vhQo4wE2#ntf>;stniY{S^dfJ!&WLN+g3UTuXAXO;K
zqxPAp!k8Vo50Gh=NK3_p8@snv@BZX0-K1jBt8yVgwGkphAn7pced<$%)ESt(0FeXL
z0mwQMXfW!7Gm4dD6u;V|5iqBwuRa?7BmC>rz4D7^`F{P>8@hp6Btu(k>&!uEMZgev
zN`cK{XI_yKpB4Z`Csastzh!N|13`kV%bU5JE&<?B<@2g5Ye5{a6Zq!Y1$PoPr2p05
z<P$g+zpq_}h36o{iA@tsKTk4UO8KaA0Ww7ZjBIR+HPT(!(pFS1Am^2no<8$Rqfm$3
z`TrXQkxXW^H{Fm^P;hxK!om?|?tL4*!a)WYGfXbJ_P&?ru7wrN>(^QypWE*WvuXVJ
zsF#Y{4HR8~5vKecNmq9f?!UHtQ?s+9^AiZ9IWfzE4{35iLfUg+QZzqAUSm@-)1|9c
zRzT@fx;j<(USr&d8Gs7k;QT>3OenF{mx#2X1+o@5<_pdwt%mu&0perk`-ZjLt0ys4
zzV=tc`M0=5s)_#~yUkpe_*RT|(8IbTQgP8H4l=b%&yHY0!5{JhfP5A|K;9b0QRMC{
z-N5M}%7_~?=o2HhQvZX$A9M+Q2qk0Xuu4c9RJjnTAyu~r;2DTdUIA~_j}&w=<AR7{
zGIe?Hx%21k+Vn-9aLYORaDCaZaQ>nMoDZL{xp7ViY*Z+#sjglI;k0ez!9C%_qoZQm
zr~s0~rykQt1B-wXHu9*$#XZ+BSaTsi=jgWcJ$4on>vwT?TK)kk#!EDrDi`2B32hZ{
zcLM|2lU42~YB;<p)IZ<PDu&r8(rjQX2K4MinCL#kuUZAapW6N|Jh>BK)d-G`KlC>@
z&i_3TZx$T-2P7gqtPg#lt_>G8*njEkwiTH!lYU?`XKB3?o2@9=#m9$-B_gEk+WuKU
z+gWJ4^Um3TJB3~zTmfXs;fDAMelSTW3{Ww!2=aZiq>(hs?zF}Cjrb7E+6dn*bU(4N
zA5gF`iQRb8j3cegi1Pj?s6OY~O))I2Gs1xH7GD8%ca4&-7AP<blg>pbngBkqk2Eun
znmNnIVzL51Fak_hI1d-%Jy$JcZWCR%xRJi-+f{%9V9^eCcN3r~4ywK9^Feg2z)+*)
z?q&ejVQ;DYuX<@bKF}Hw8CluleAU*r#ZnR{It~^9wEuYjv;R3)aJHtM6*tlRGb}5g
zv_dXNL;iBs5TnIWU`fqqY)qI=ZKL`EJU)O90v@Upiaa}~zuXnBH!YMx?5E6au-`x!
zBIjsxf%a=m`oTxX!G_Ga^;KPzd5DA&SRVR#<UnJLQ)!}wfIRaPt{57Ab?(>^VX2=W
zN^>ElN=7G4@RyY&JBT=4i25abH!&so00#+BSpE8(z)I#*=G)^hhhhf1l!FXsfa>)>
zb@9-Idzo7CiHM5I!-3i*200z?*7*7P4Rp%DlF7`bmBVKvsC9fmm#?;nGI3>A9CK{7
z3&Y+RT=eMpb2<B8o3?h(QZ$36QPHUdm%EaIBN!UuO&;VG;A)e4?*rn-sV^vaUyQ4~
z-(egshX9cg6NB+0QRZRqlXaeWhTnzi5D^RZ5U!k<VJk~*SAWcPif_#35I#5;W``JE
zS0WxQ?No{ujy;s`DAbT2+wXvb3Y=WWU>ihnAuMR@5G;7n{lnoE6FWRKYDbU4QxcFU
z2FA$I$XpJFX;g%f<YzwN`Yl_xR@Kyi^ojEWmJ@d{@zW5xdgTfRdROf1pjm^hB!DnH
z8mE&raO3Tf%J=Z$*V<Y*_7Sn^=H^MTfzS|o=e}|qQn&CbOl?7NWJ~_f;{WYl*!Iyw
z$*&Xa3N-ZWjEo8GdEh@IAA_J}M@!j&mA(l6{TOJ1sAnT?9cBHFY9DW+pO9vMGu58h
zIer2LmiXHXYa&b&8SRPS?#%6YTs}c9D~!}exH_UZ1>qT4cy&;~i5V3+W2{f!j&bB$
z*ec@cp)EogR#YV4u?O)|sMg<&y&VoaTTpQ<0-FNtE>SBI&Y#Z(VP=dA392VDQhfh_
zC5`YRWNz(dfCMcY*QcoHf>0<p7Jq2rQH}=z7UGR`-D2(Mako7uKm8=F!uWLuAL5;$
z(E(?HLmGkiAi_l>rV`v>1Ow6+YhT0q65X{6WVI_-uHY(Myp^x<(Kj2N)T>ty1%C~-
zd==sY<rq~qOECxa4?0}vK2WWX^3`L)0cM|L>MaI5c8`DF_-=@NI%^?UJmVTdjh2$c
zgneubH52eka2{5;S%JYZG7`B*aGq{%MyNPyWr(&g)YJ^mD_A%daVAyo{(}d<`}$DZ
zt&hLFG3UpVGZ)Us%Nvi{J0nB;e-ZZPaW(hvzW=HyLqteKl4K@I85&4YL?lV5kP49@
zsiY!PDyfV`ndb~eDVjt{B}L{6%^Hy;s^9a&-e;eE9{Zf%`s03lzqebh^;z%Xx?b07
z*k+{|^fSEp&qZ1=fdz2ISTz){5vFm1?PdOX<Ais=(_a342Ag)Sn%6MHfOi2X!0r(^
zp>s911HB2%P>`;phopN!BmLi+DdV_S@I|L(c^p8e4c^lAaANd=!BPf-X=nfmCBfhW
z%piLQ2b^KKD`L(za!KpWc{KDahYdm0bWfnBzGD&j34=WYfwjp7iM`ENK5^m^VK$fS
zcTZra1oTti`#YoK(X(b(I4?oQ11bzsIxvyOFtg{UCebR!H5bf2<;lqR@-r}a^uG<j
z9GCg_Q=J5rDc@~hNt4>T4jjW&CdvtA%_{4Zj?B)9AMl~y-owM!Dfiq)S_|I+L-8~1
zCXWV=RQW7a6T$@7LUePKy!Fzh>k@{p^YME$ZHDRLiSr(H>(uEu6buw6oH(2Q?Rw!a
zhyr$$Uwlp$CnOpjpgwi=EvG$_*WUGjvqxnMIiqoE{exA{%0!i~2dok1g?W+kGSiF*
zq#nFS(PUZ|9qA4u8g0eRD$m%Yh3PxjO?W4Gi}WljFW>M!HZW(V=4rDl2tG1vYMgdX
zW12ATxclV4znF$R6R=JQ;2|;!URaxul1xY3{==rd4~QZ7*C*Uw*P>^m4O+-2a&W*p
z>`P5ewxVGrfxz0)e?K!g&d@CqZWgEB|3zq-gA17{t(w?u!i-WYO5(%^y9|FTAR@ec
zrbaN`FgErx{0l$`|4Q|^LP|2Kh4KLh6B>no!OM5W1`$|+tAC?VYX~nBwYV-i|Nf^a
zK-A$caL4~G5s`oT_vN!m#0cr{D`gUB^S}RTJ75|vHGf}f-IW9|UwFw&^+@@T9~>Y7
z#o*t+O&5!}FwcDX@*TtnMOmy%c*{tA^j5=vf9T_vCN2h3u^eU#&qeY@^U<EV{E_q#
zi)&F=u8d%Ke(hQ){Du?RT%>fn49u`VLwk6xa613pEfJG&$hYis`fmS5p>fztT3s8J
zBCDcBfmtRm1Yk&82pA+Tsl22_-AYGjzWvp-eBhY8hMvf^N0NaC7III!eE-5bpJ`-m
zU4&`<<vhpz`*SNyPfU1>Q7-8KDD4@rCL~y0Fym{fXVizkBvZ(9g*N=idpoFoLCy{C
za03H66v?*o!+cV?F8|Pc?BAbN1}66`Tq_9c1g-yZ;C=4p5`K`{u_foqtfFE$+V(3#
zH0b(6c*v55qvhZ=2kdv#E`t0V&c{|@NNfkWoC^Y)!L2{{grh@f>z^;?FWl-7Bi_vU
z1o>S!T;#q8-<tGOsuxPo{X&WnCAX^8Og8PKM?HeMl;A8>y@2|2#e8k5SNB18t)k3F
zE*R981=T&qtP*Px@B*w%P{{lH$CyX*<5KPbrswCIDGLK_NSI^ZkF&aS8H)9p`COfN
zYQ6z=ST8(iXofU>jQ$al5EoqZ?p+wDz@bB@r;<*G?lBG)`)E00b9YWdwHSWr&`0P1
z^>_O#D4<MYFiNBuH7Cauci1JfD{V*&dgU!N)mh!}A^G3v304{>b<MzkGwc45-hbIK
zg7X?D3pd-!dcFau*7QV(PRXR`{%Kz)o^}ztMUR%3im8gjA%;X<veV~m6{<gzdjLzL
zmT7snd6s0mui}D0G_=CENnj<ae}W>&IMiM{tdJ=<CN6ST>I@2b{TGXN`fn}(wrfmM
z&jrwE5Wrxlic`}I=L{flkVkI3txF$)f{cEam9_MaW1gUMmSN8D`)9rL+O;d&>5VC2
zdC}e?*6iG+OG23`Tg~?5gt+~Mf}I=JG(ZDVOb@^}epj4PoT$zg(>@`3<chgLX<c0#
z5f_C`?ymJka|QZ};89}NSIjmWfM)eLlVQ2lT|0M160be=m68v?^?!#ph?raTp0|$A
z)VR>h{f_?^&TXrF(t`(o*m?#~AB52_7di+CpOA4JCYZ18&7s#unyjKsy#>by(u&oR
zB|@qo`h*K+LI4Lt`g@z(_%lpaP%G){?_$TXF0pOKvw76ZI1)`q@8`2`8_CwEBzPbg
z##6n*yR*+eKA!fJ+ZV);;?3)QI{}R28O4_@hlS=}<VVj*;Gmnkd*Fs+BHItZMoP-M
zk$TPN|6+`N8F||-rfL1>Z-MT*G0x`R#_6h|{4jPGIh(fv;VSPB3Tg*?;7%psqj$f4
znP#7{1%vtGXEeMw%D_cRJ#ICD4GY%K40XGWy_)AZFH(1|0fmN)cEdEBrrs+89}gTz
zefJLNj(tdgJ&2a~X#-{Kv5FtK$DsUskILt1$yYiA&M<^C4QIOi$LBtLxOXA0mZvy)
zf`eC_)e(2|fjZ{v{rSGCNV-qfWUc*ib4&hXw*RH#nmTf>#|QdII*?`bBqwL+Sy;cz
z8=nCV;1{a{cY$zgI%@6~rL!EASj;QWo;UB6VZFhlitl7U7NOE#ZmlD@ZT;WL0Ma1=
z2(7Lzy5^EMRUIp46bs6@DQD~(DH46S0^fCf{XknIc-9>B`-j8J09D%@|KR+2JAypo
zHjSjw_)GJ-wQ(HmH2LAh>y<{weNZ~2s!TOj*!)1qnEy|5ZY1S&ZdTSc-ZVV=^($7@
zfag-OPIMgIqZ6el`>IhZS^{QT%c&07#&6}i<pRO*#a9w6BYkj!zkk>5JI``6Zj=HE
zU@jQT08@28=U7w0<dGwvrlwNG%~uZA(A1<OlKx>npu3dmbCK$3w!v*nC+(vfcH!m1
zERVLq?p78D&%H#;(W=TE8?B#QcHoaglgu$t6A7hC*nUOzy<PB9AO`sEwk=z#<2nrr
z%J|;Ua2Zf{UUU7nukX(2W^vlc2l1rDY>miw@e?kdQ&S2!CWm^0+e+&SzPb?Z&M78<
zEYTsaUmrLwM|AEX&G?x!sK;lkgm|SnuSzg_PodN7lwP~Uhe%SJ?X?qoU7}&(;7d>6
zdsj>M*R*4{wNf5aQe1i8x9wYIwo3HiG`Yi7E<c%NiO^7qYv2<7mT>bSulgH4%`zx#
zo%vApsH4Rdb7$PNs}`L}CS3BG(&~qJ)Xq|@Z*TuB^YdF4M<XW{D6@(z;SHdoe}jK3
zsV~o%X!uN~Sy?hY5U`mAm^$F>s3Cn(_uQJ(zG^zXg2SOxD^4^)zEXYnA^oQIRD;Do
zdhuLF%cYP@0F*#pHK+iu%(8E^A#8R^UmE$Q-#H@Nvm9(ILe(LeO$R-9rc{8C$ZZIY
z1>sfo;6O(|`z_Md-?IJ}sNB9$E>jT9{S8g_l{|7m%(j6Agr&H}wVWqU_8&NqP-fNu
zS4dqwbpM+dX{pN|78D$r+u#nUDfPk{zW`LI<T>77>^;<d`ENvI7W0yAM!OfGBUD?=
zrG329=w6F4YEHz<0Nd#n1&R^a2YN7dgakpsZ4qh6s^E2E0X9zevQty1YyIh?sQ66(
z13S8x*q>>0G!%~OY<Hzi{n}JsUM`F>t`*?*TD`LTleHJXZ+PNFo-Zm%TH8+>8KT2)
zAe(OOx^+Y^q1M>jP$D7H5#On#t^I(JQdwT*xb&i%od3~f9|$nSSwLz_XSvC9RAi$p
z-a8+_-*>fXpSR<ZIDXNZ8L51U(-}4Crs?qB@U$?Zf__;t_I8EuZlM~S7SvI^SpfBc
z6ceEVkWo#n5B8}@S~%-@^!$@KgwE#DCulHpk_V(rf`0o7afkurrG(*H7@tIaabfns
z%wPKO$|=ufb0jf_=G9H}sa-U#MWnAVM4>|o+Hp;BXW6m0Ulw9WcsoNe2K%>DgU61o
zgAEO7ieE|9nN2i~xwZ8Tvn$BzCB(%|0<Iy7{|_=q;YaU!RnrI)1F%iLjC2{PK3yJj
z;X{m~+_>fWYxAolCIB3_4<2|juxbf>drZt%9b+ccUm%IVnhJdr1cUgjPektxwA)8{
zSLZe({80L<>-pOV3uGb@v46MzC&ImUFf0UiMYZ6~@QtI26M-kSBKMhK__O@on3N9Z
zm$scM%f(iC-00B(jYp4FFHWALA+y)wS}R<`p|YKiWhGL>Z@qBIIi3#WcG<CR^Cmt2
zmQ}Oo#;e2F2{>v{dI2%C<W21f9C~z(^RgxVMSCBmrLDq{#Q>)!?AzV2jV@VINi9Gp
zg5Sz1aLVXtD@+bP>KRjvSvLG(vIX&>x*ke7g#XpOHo~TVn|}H7@TT8wC=;rO<qIEn
zs%y4eZ31c-3)41?wG!@D9{(?n%JUr$g<A0W%#X+x*E6@l`ea628QwF{m^^jc=~aRH
zC2P4<1=1!vpWgGfCB6i8VBfFB6D8&MS&HDmuRSChg!P2$k)ubuYxumXm(B6r1f9qt
z_DkLU`X-e7>j%}99gAOfW9;~hH;3es#6`UT#;GJAmEida2R!n~kv_7rpR20Yki8U|
ziKH9HNK13DYpTJl2>I@^pt=ZMP;VnR6aaQEUoh64(;$yUL`LQ+iX#fE)a?Z}%H;>j
zGBwz!kydUL%0Jq6W*Icj@$qlTalCyy_{Z+ManaG-mrO>EWZEI*a%YGc8Hu_@Fn306
zacy~=U8d~eJl4wde%}T?X1_RUgpmZduk|~lN`ppd6|cbcsCu3aVN{>%-4=SGM%-i?
zlx8Qx)vcel&D8VvB9W){)mDYl{tF~II>%Z+crq3<?3>B!b|14LCXXtQxF&<M)X5+$
z-Kzqb6T$Gewzd`s*Bdt2U$Z)V^l1M(Uz?h!66yJW&NoGVBILp22gEvpC>q;CHSN<z
z9w+VC#o{*jHt>{K5w%~w5L)_reUP7DUdM)(7lr=^NThwVBKKB?*tWv0ujl`UB(PCR
z!AjI#8fxl=B@b%W8zkYtKlj@o<+oaUYmPu}6}*R|(?z`wjrA556JTIvT;Q^?Fm=9c
z^SoX>G;A(V0zeO|98Win2g&WT?0=O(oEJt5+*-<ueD!*W_Nx5<GJIa!3|@J8*REZV
zc3pu0Nri5j&tZl^5{&)E@O~Pr(#gsT5*UuJU$f4MP6L@G8$2dR7dIxgC{-#3QZsOl
zW5P>}{IvpxX%L=aHHS3|+JfW9_sXS^#=4CSFW9SKrGvaPmfVgfhEbO4SvPz-s#$Z2
zEf+ak;Y*z?_GiAj5eE;vas19vok_@Ot35gJq9q$(Q(gT#XNY}_HA|e*cm8m~hj6tS
z1&MxWohXGQu6ahn>&bi>PneM#-a?LYrGf&7?^il`*xax?EIQx7>&mADuA&sHUm-l(
zTcFJ5PrDCmnQGwf^d{lDKiPkCmsY-tu3fs&6wv4SJz+&hM^6|vY8}ip)q^ZE<+D3q
zC5hjQq1TP^m<LzYALphGDYJ9Hk2uzU+&`Ir2)PJDC8o;V_r-u|o|*6X!T#Pc8!`<i
zZ~fA{CkX|JZ|$BQ9?@d+_U#={(Y-WuG_?$v4DQfp`92sNWs!CDdGXBtdYmAYALgx)
z*Ntp{I_l=FPmL5;ya}kKPP`$Bq4^c1rJf*{26qU;+tSiSZC;gqAT2|cg^yN`{{10@
zUp0reH0t;AtQ`$ZKj|}P3K}x`%eDm@$0d_rVI-6V{H*jPKhqyy7W>#;ZB71pKo-ob
zMiObB)OJIt04<+q=W5Zncdlz1bw4_K5o8YiwQH{$C@+D`A;iYQ9IDJ+ADSg7C8{t$
zalrovMGxwut}gC>aGFTDzVp-R-&f$2-%jp*V&=IO9WQXH5%MJYW#hDIqIE}sUA6Ux
zhGquHyex}@Hu|NB5%BH1$4I~v2s_Wq-rRbie+J?2^D$x<IE5aDEZ?JlrcSyhW9+f4
zz+B3zxBfKtvB&0O0<#&5bah9p^{`faV`@LC?YE^KcbM}MD;d@Pg`q<}D5ed=9y%eR
zh0c-!5??bm5|=ZDn$xQQ9J+#na^L*<U>}$9rbZ#4BaV`BQjOOz=~WF4bRB2wl((Zy
zzO)vLBj;O2F`diD=6q)0iw>T{gFHTNhCfxN-eH>S`!BUf_{%vlId<+hPGQ~G{0N&W
zip-rQ(i?yykNz1Yq4V?dzZ^7z%CqLD=IGJcr(HsmsN!f^*?iOnvcajpNuW!wKRkEg
z>tYN|Hhq57Im%*Gk4_dht=E6{RpN^U7fMVN<^WiukbId}o4`tPb=B+{fA1bhhJ3a~
zZQQVg@t%r78BO?bX>KCg<V#QrQVPf|v^Pz;pPK7e#qHzs*Hx&yQsi?_9ug*0gFGtF
zc)cH>=XFx&DHm2h$ERPR4ynre?RR8%Gm}+}ZIkUuvv=v{O=tQx)ua9M77Xw06!o07
zO6jno_Radr>grWNiRtCFD@L4rXH}HkzJ8I}!>Y>X3Gr7}t-=&wosRa*@-NVUaybE5
zw9zp!Z-@CEoTh}=?kT^Z)I4?@nz;0~;@<M^H9?d1rKO_sB}eU8*5kFuY^ULZ^=jk5
z?ApBgk>(P&Qwn1Nn>fR9o|B<Wp?**IjuB31?pCHMVct<+9~TvQ-MPQfflFdEV^23g
zjB|yjmOjt<Ngnlt!uJ%?Tp9!#FmdLBZ(WXTx!JO?(*r;>aBqTdJPI6Y(*BbR(0F;-
zSd0ppco6#5dRUsU&6sh|(cR493d|})xwSPpYaGq7xwqUXh1++S64d!|lGk_cX=q>o
zdB!lxf)ws;Y|Vzo#>1I*2>L+HsQH}U5|JjD4y}HAI8|@c0BP6lZ|HxP%(k6mLMQU>
z9r<3SDzeP(yozz3N~|+X)(TiQ)?tq9x4+sM>1|3++lqfDN(5H5{i&?m9h^re^!Du<
zU~zn~&YmzIx%y|*n+H0hr;WVe5P4z8@5<LU#>$^}{2H3rtgQ2MKx>t$#vhyUXkY|g
z4e|sEBp?hX$c&6o`&fwCjrPu1%|2-l7atPTu>tOez}qxYCHbRYOpxk|J5TH;$fd}j
z!9JmPfi{=}Ul`U<j6zX*rYvXCeH;5*Mz{H+Wm|YRHBa-AeZmh4Y#WTZ6E=~~vQ#PY
z{P{;pK7*v?2KNf46l{9$EVVY)^Vc^bl=_#K9Yg<5UIf!oAb}(T{R&;lemQig$AAIZ
zl4p~Dm50M2e=$_8US-+X*GK`uD$nJy!u$6%aD-}9S7(3IP#GKkeP1V7@%4_rn?a0A
zcg2~CAzYH&g2GP))jn5vEN_V1i_ecq6wAF)WbM(q^=amrbF<bo@9TljFm(%};>jQS
z3iqGSXt~8uY6LG3P|nzwoyF=z)csEK+7MfhZrK3w>`jByqD4n|Ot_ua)per<LowZ@
zTeqCd%$|M*oJlN6blg-dPJ6v+z?jg$sxVkufkoo2goLpD&_@F#n&Gv!dRs8qAGdU4
zNqV=Y&qsrH&x}Bf2#%3q9pGxCDj#*UE;+fqa(P9mu}90AmWH(8FG2%psVEhCXT=9l
zwd=9F_}qkwIc}3mmvh}tEgFzBpo$U_9$n7`j*gY!i`|+y5pMDQzkVIZdnQVQk3uP^
z?)aTSaP;U)#jSPQhTPJu;Eo{&8*@Z6v;2)V)v>4ayuH+hG_cdg9_y8NJW*#u@oSsV
zyoBo3j~gq$t^VRXeva*k#$L6-%F>e8CjU3FP;U=4thg<wS0WWU<c02cGh-{uN@C?~
z8~&J|%e&R`2N=_7V^TuGyVHTVUOPGc<@A3VYoq0feXiM+0l$NSGUn*{?bmH_oS$>Y
zs-?MPMIE4#Nc!BF3(sT9KKMCuhmTHP9Jl(A*U$UK-LB}M4>8c{W<B<pzk#w$;=7OZ
z{<giw!fLvokO1<udcqPRZl=H3%G<|}EJQP?HQW;5YpIzTEha=Vvn#lK&;SSG39Pw3
z?5nx^9{tpd^N1YCvYP;oTUVzzXi)B}9{r3MZu4LA!!SPX)Pa+f9MLmeJvVmFzTiPJ
z)c)Z^hpq;vNvL&__x@jq?Z~lXEr$1AHg@sw-qE4Awsz~rN+!OE+auF%0wirP1!U2t
z7tC6T7#VgJ(b`j34v-s<jkB?@MKiqaX5vf(K!a0S&JlImP$2LYynf?`{fVipI3~1y
zK|y%qYux)>`(sPvh$DxC$G=H;cqZ*2z~{DZ-NyKM6J_<rAH+>Ar8-txmdVKRkByg7
ztEC#8&T)TG=r5f+?OFlJSyfUEt@DIj7wT5S_Q~xh48I}MVpzd$Kkap=`1$jVAjGYU
zjM-wmsBGX5o(c)cca9@l`&dN8hK+6X>tr{o4Il2P=Dh5JS#5E0GSaTrg|HW@!GM7I
zAAeMVoCay?)tGbdXjo31SYmGOf|HVTxz!YQ9lxCJHen3+01DZEE+Rr{bE0)Xudwi{
z*Aqu<sC~T@)LWE3r1w54(VTT_&)(5^nseb`ciB#kO-@#E2h-Ol#XYW?8Bh36<GPo(
zMY<%UW|d(TY7(YV#pVEg;pl#Tca5W?+N91sQTY8oX>Wznu2;-kP;9Xz$it2G#JByf
zk;>DY92$!Rn(q1Y=R+o&9*Oc2G9adm0FA)NtUc_qxG)(Uz>vge#vS0sz@<F9pP#|!
z@^YB%A0f7cW|9+cM(sxw@F3j>0L05LPIGtv1<3FBgz|y8BG@073xgXj8p%<kualWW
zgA8ed7nM*l`7~9pFVye9j4&-~o9#dG7ib;k!Ni{ZMH4veDC$=}_?-6f_hbv{D|Sj>
zx^ngfILzq!aY;WJ<@^ghk^Xf3*-4Z>ZqKf^=2;%~I{7xcyxP55>f0|~ioBTor(>!_
z>y&P!SDN+>>=mqk#`auwLVhD_FrMmV#tdIlnEPyXSPvlcV_zBPy8tKa;!NAJ#|$De
zjm41Lmdh3{JlFLjHSXJrCg3=-*e<Rc0HF%Jfbwa1qDR3bAGlwjU5v={f6%v-qNV(Y
zf}wt6Zya@ArW{OI70mc0+d^RX;68^b7(3(DD_{~s1Dr)E&T9s)2gzPMapd5^pP1;c
z==jHFQu507moia&8(@QeFkk9+N-1{2CEBnC&kLN#;M*4yy?Ju#wz0~}4{awrRQf%6
zFD|S8UAMY+zq-ljiP}#|$nh%2vV^fV=ElNMN}!GUrMJ$j>aUw1^-37e%Acz3cvSt)
ziVkr$RY_weOXsOQzyd_U!(;g56L=8lo!vI`%xSG+$oZXv{&<~|G&Nv$a_rsm#fvlY
zAODbWQwGNnDqDa3GboLRD&twOJR8RW#G5fOTb!J(Pd~9xH!}BI;!s(?4P2z^0^4)X
z9|xt1&!0oU(fakP0*CnEKtIFS!T0ee>VP+0Tfc|Hm%=Y(^~Mm4wQu)jMb#McH|Nah
z+7olS>ers--+uu>`VMM^#l2zOx*-CeRUx|;2gD_J9D7!~PoJL3DI<M5`s|`>lJDJo
zM#yr;FabI7<@GOo;`g^+5tnygHCm>JDE($}Ta(<8J4g_U+g#6ks9(J8hC1p?`JI?P
ztohol=kAY=Kh-F=&)Lbzu20^NVFINURskZs7{*iMlMc0ga8A$53*L{-y_Z)~>J1<Z
zW)ak6FPpUyKBcEU<4x8)R;s@98+^N5#VMc=0t<+#WFT_yC#R(jRaaO0J)sj$J{d6&
zM1w=+#@qFcTbzEi)~`_u(mZ);&F?0}5~7rWmnm2t>50t@SS{$(wth_6DU<>vFuyF0
z2npFt>+en1hJ%N4UvZ!DUuu@Ll!q_X`rWh?N5oI{IhL0cdSJcoB<=~nacr_5@_qXR
zk@f3022pN6lB9Ye9gB=?$6i5L;HRfkB}L{AI}1a22b@oQu0S2HE<G~s#kh$#)sMT^
zx3tWS(bDa#@Y!|Sw)$<imM7OluBp+6^IFr|A9;3-d3u$EZ%xIU30qVXn?FZ|6NJ*O
zMT#SF3WW$TqCs!W3%(WhNJ(?NJ+0?`&T6<7yn5x4Wb|VjBP2p#e*y6olA1#oaz$@#
z2hp~}GI|4?vVwHawxzob>+QRy!O4DN73dxt{0iah{*M_ifgN{jAVtkvTpRV3D_VP~
zxwkDpyngNQIiEKO-$>iF#*!fePJGwyJ}cXdtv_eIE+9qh!~J4~%`ubTd~6?cI|Yb8
zzBtjhzVp-|FWDq9eC7nLqpaQ&uY;3sCeJG@Y^=NAT3@!gsy1#-^{e38UVKkyfq!#c
zAPQ~XxRI2r!$v9;LYl+tB!)agNI-w-Uqn;bEqxEVoUC0<*zn#!RuHGuL-b*dq~48#
zJ>3%5H1Jmt4s=WI6`U%&+Nr;G>lEEwaX0WY#d^JnM5dbp=tDIR+Sxxa{r!6ir<MOq
zX8q9iCt+~cfbS$_hWx;RwWXy725avoU>w!_wFGF}iy_a~!9ymuIkkxG?l>wXW?|uk
z-?kraMW5YPMDdw<`iAbl8OFvgRkAwYvxonXW{8K-TvZt2AsfTlIku;7gWAaKfjirg
zFf3_o%bzo@6{tzG+3FW|2FVn*?cep<zcap*>)o3$uM8o=<;3FM?HNU<GPD-i2TJJ0
z8TU=NSd|o4r61?AV$IZiakE3qH*H_hquO(sd>_w&100I_*{4UA+)MnvV!*6{6+_O-
z&6?ixaO$q7?>de=b74EM@rxaE^M@LT+qbuVeNdFCvL;NUaQFAtQ7WeOJ2o2+uF$PG
zDRsusVWqg!x>b@QwJBY?bo^t~cd~=b+iuWfFy#(uKAB@5u5qYqr%nw`O<4Xj>A`Q}
z{OQ6{-8DE8ajb|bVLQjc2KT?YctvPLtp=&I1alle-bn;gfhjmi5@?+~J%xcRgXCd5
zTXzw-&Mt;b@%h>+%_b`;{y&`r6a9tVlrq1|U6b&76zjaN_Wg7vpRoS&3KHJlPoD1D
zzkg2znOm*<De*$KSJKb?1D)yfp<U-zC^$>4I|nfwyGo%TG@=j%x1u{{DXd(1_Sf9C
zNLNd~8Yt&AG~9xVc-s=?9X*-FfvNx~W`g<2wQK%oFAD&FR@pN*RmZOzHGK4-35uw8
zfVix0Px2wLo3D<oDfmz#{SFk<){j$iaM-w6Tw$Tp@Gd7$Rk-QviKpLBebKR@JPl4L
z$^+UQnE9D9vRnqD6uSukUD;cbbG}CyO#>*Ied%e;Tg(?vQ9vzPw3VU)BWzheW-|jl
zws*(i4$|xqBNa*CRL`==m;K-D%CYBjg@%bc!~~JnG-O{wfw-IU_@1)8`g+KA8tmfl
z=;)|4(8O4jZ8rhq=-HZPwe|IJtw-amZ|d!6aaq3n)YniIFS#yz_$CD<$=7EWxpQF>
zFC2NKQC*n+%OT5;vb??Wz03UrfNFSLhE}$MdTSmmsy)@&%6vd#eE!Rq9YoN$l7Fng
z#OSotklh<5K6}=q-<s-=PmDfvQ}i0{E9LW7{MUUU$*O9TCuf#?6~dG-bW)H}(#u+T
zJNw)=-_++sh<vjnSd6RP{maV{-?!xTIu02lZ+qB4&>@l1u`FYcW8+VlyvomCS+dC}
zc-QO;SFc~!@V<NTqP?f*0FlX!VPp&0?K=N-yPMnd_67M%8s(~=UEB=IM^+tjNqI$u
z9#ndP`E(ahFkhR1opafaT7$5lg<uIQpW+?Cc@gM8^l_ZKhUu&XyMtkJWDB;7&^AwM
z6unQ_x(;?wfWt01g;^4uP-2=ceU#%aGX5H<Jy72<_+3)rtKfUxCW_>bq=p8*9dRJ~
zc=uU9FASsOd}uQA$Qvh3Q9wYzA)|<k1#6RAo0iI^DAaq(2~n(EqUG&MFQ%TIHQF(K
z>&%&wYZ|tr!7VZHKz(FxZg=6>fy_D7(xv`JYt~%JwBE!>lIU_V60i)N8xCTEd!m>B
z<I?1hbOuvyYvY;iv?>4HyS3XxZ~l!7h<U3A#tl7$#t<eU<s4$aIr*tZ59Y$Hm37eJ
z=@9yjQ#KinaI&$!&3Gd2^J}KMMxwPh#;KdPte{sbe}CoO#9nvaOzNM4LRIAR_(!)@
zc0RwJC3HHo^wCZU&8Tbl?~&@hMK64iRLOoFrfE)dHax_P@{jtuwjl=k`r;x2H(@f}
zOG@(MdPNJCQIg1B(^d@fOoV~NaX#`$7m?9}s|33Ym~2LlDx^oqEWlk9V|UJ<?;xW7
zF*=kyAHJuN6$}L|NT2Z4!Mnn>r3IlC;M4Pg&4CMlKhYYAz`NqR;9~t7ZZNuik%&+L
zrauBGM2$O5Y50f{B%ib!C-o_-s0jbiV>(8d0ePps1T1`h!bJ5~J8?zyiSITKAV}qP
zHRqKO7+<&W1++t>zDr9^&COgc0TwS9m!*Qt10$d!R<8=5p)N0N<MnH(v8UjkHE-TQ
zU-jw!f#+<Nk)b0jMOFk8ACIes;gmnX^qoY^T}wVz_0ab>H<5nOGV;&2V=tt)|5(&~
zCT7ZngXv?+b9*scvy0KX)jK64BZH}G^x3mS0VxbiU)cLdXegr+s6Ub-<|~+1(P#l)
z&~cGJTTtaf<EI*DGeciraL4=@wGJD+Zjxz$Cct>{QPW049%tltzK{H|yDxrDg1^#L
zWXap!uwkrdG<GaJd?B>B$^}A|hHl6Bxrf5SrpP$s?EXBcZ|9G!#ktC-Sb5(<rYWht
zHF~cs>AT7H%S7q?ORwaw1iYyJL5rq3rxJ+%)R#%6-lh}AM_;aMHhI3X#8K3qt)AOk
zPL8|?rw{i8TbpOOxmx|^TU#fA=n`=U5ep*5v(y)Zcgx+Zi15u@K<<~?>^E&%2X+;=
zZfNP$?y6reDPynjq@RCa;Lr<>(voQ_Q=9P5DJdy&*N!ky?l)-A&xVHmetwzHo~f=_
z&u9$u57NE289>`GH8zIVM~}kQ1BHTy?`;ix6j!@v6$dw2Y>fYa*(&|cjT>VFXZAuH
zf^RB5GvM-~)5D*-<tgdvLf8Aj1Ku1@z;R((ny=Rg&^dPCorZx53M>QI7kqPk7Z$Pm
zXR*9Qr|G#y9(m|JHo2FIKX+``4&tZgIXGu&`KF$pKOB2d*hWSw6LR+v5<hm-DCjVB
zOgus^J5Fmja<6)22y?)MB_>OkhA$f%8Jkgbhvn66^{wVb?-g$*Hob%ToLK@%nzzb0
zk%2t7Bv+!@N->e(=L$`&>|=ZnaS;*L+w!eba&8=2-FMu>RNscQac%8O;$^W9&EC+E
z)A8Ha1`N}kwGVC(gFi%A6JAEQqdnC6?9rdz^8HhW#EFH&dzae3(l_leW{Rv|2Xfw3
zy|gz7mY|44B3<9$9BVvNr}lAQn-1p!)WR%BBLPK}p%Q-n_<)S}@87#6F8r%`>Lw%e
zB3Dg5fEfd$Sn_-7RV769$10z?i-sRK8zX@knt5V+ZCK~%V?<jeZl*S#`=$2)t*&>y
zlUr+@fg!!&8Umg){PO&Ma6h}syE$!;n#ZilSTY+4@r;U^zoc_oT8e$zfj$k9WlK9P
z8Qh~2A{c3rj85Yr<9+77sng`-0)C4N8Gl<yR(jsC;nhS9jpH2}<Or<5fihSlag~}p
z&)-S&eQ~j#vhrER=nw6_O8BIGs(ByB7>~FH1dcfO5TizyUsPmmW){sH#>0O}f5QcM
z)Gkf(jK_h8%&D6<ZDO5HxF3y7<ePh1qfze;vcvZL#T5${HGOz*&K;T*h7pJ-rq<mn
z;i8JXnkBAZ`Kou{ebc9goxl>{%$dO=NIO|sS?XSXxfP+YQySZJSYyuz&@?J9UjRoO
zmZ`mNimlp7ZFjSCd1QJ8&RnGvca4b!;}=nnzX7y^6cpB-lrHAN_<9LnUuLc=S1OAb
zL_*fa03D(SrwJfgU2$@;O5d&EVdt0Dvc)hq)Zu>MP!ar7C#8AY^I<ZsZ!I1=v`*!<
z{<VUu1#A5W9>KYM@F>gCuh0<!u?db=9y>N~7IX=4TILT|NfbfCx&3PZ2Iuh>c(5p9
zcr&+PohE4pckeD@gY40-);b1QBS!5&rF=+oYkKz2v*)r5tIXX6!|E7yb#=z-pd#nY
zt{5mUz&`V<X`Y`53>P9I1Q77TjFQ4l83WZLI$JrF+x;GZ>vp~kx=o_w3=9nu0jjrb
z60G&(0=SND^3;KCpB@=ym*yIVQ^U&2&>#XqUtN4%Z9G*AEav+dTgOeCHrZt+V8rbq
z8b`#Y&*PORGP23Lb-K0k`mI|VU0gJ1ywPT>j?3HD%V^cHD<9k|PT_C^6}4|)JRo6O
z<~pf5-ac`uei~$esXJO4(Pr!EUBZpIv7`(%3h^?E!((Zmh~Aa-{Zv(z`Dr6bNYT#4
zy2-JcZhtoIKOpYlU>V(k5Qezqj(;BR#L&2j@Q2`T1zht{A0|=?tDgAIMq>E)RE!Gq
zZw1nP;y)`8O$99>ct_HQxdNwc=nQmfp;8NLKh3=tAHV+1EioY&EAvF;TTW~BuA>^J
zoFbR12pm+-&tPFUCi#*glv5d|`*=D>XDu-2$s5}%T{dqh4Q=?euf6E%{A(R`2Q_=T
zRgY=LpCWTuh4)p~=LQ{@-wE6pqtZAq_NA>sowW8YJsKs>!ax%vBW736t6cCQ`-*Y}
zjV<JUk|BdK6ciOzr{+Vpg6r|-jU@*gy>Mu^FVqSlCr<Q}lPmx9iMUhD-pE7sfiogo
zx->W<!iyV>4uN?R2Jyicup3_a^X`l-*?A)Co3FWNS}8nluWrCsSt#T>oq2!QqU+BO
z62&_leiit2cKb3yUUBcB*olU`bNiW4*E)!c7!?H^I1sHnY5e%$(Tg-SG%Pz~3RFUQ
zumoFHIe)lvhI*z8%;(MPzAtU??-UGAcBOrSgN(ZeJ|aK|i~a`wF|z4DXbcW0>ANPB
zY;XmC>g}6)Yn$qe2=N^CSrW}~_H9%=f({<EzV8}kIa+sn(`b=)OXWH65*XR=BQIVQ
z%pM^fg{>W=FBn?)M{<8SZ)&ZKx6jl%pS$ZSR_O{)e1VIu=j_>C3>Ul-9JC+}tkVc%
z6ozT%1D*SRaMsc@GRUy-Gr&8!?$alXzCur(n%gv!nFxu51&$h{<@^KwpEu^bspFL~
zdKxvV8gVS~Lj03@C@j_F`C0wRmY;5+42I_wVj`X2{}TvdyU+r&*%0z{#i=|^Jy-iX
zHtnt6Z57@q$A7#%F)<+#(RG6QrOfhvoq~h*p;Y=&xa@gQfTeU^p4#~Ftxsy4NL~6o
zs>_~TjlWZaq$Uu1_T*T}o`_CXE-JFjEKZpGaZ}m?)PcVY9Jn*XcFFa<U^ZD-x4US|
z*RLDz_zkSO-%M}*BCs16KqnE~aESN$;a|v&!eEAfx{U8Eq?QKOZ6Ge^&HcV@)ycLM
z^TN+=6Mg)8I%@qvhrQI^gYT#f==O2DhlhxAAi_&^#E8x!CaIMU)Tj}-Z0A@1tzUX1
z{0Iz6OWM&xp=++TM18W1ji*?9{D}`5BS)4y+4CLs(C?a=*WZofB}8!=+y*gG{=ne$
zvdv;2F*52NAsBp*nKyFJkLoiQ?aOEN@ON<5{QI~3_n#Zpz<gNr-2U)QtWCa@mvfBX
zw76C^EYr)k{Jp#~Kf2vsEyY)6O7BDoQMk^W7z$mN_v_a0k&sdv(fR0TF)b}4BcJv*
z=N_HpJgh34CQcRm^C@i~Ba8Ndd6{dVs^EhMHKtKbmta<=2>;$r6!zOSSx4iBt8dY~
zxQjEOiQf|0)*&;qp*&ovM@p&6Wpkz858pdW9P66gu2cD=tB#?|)Xf?`-oP?qr!QO>
z%4(tLQ#Iv|5l?Z3fGAvPi2Xe=VG^_D_x|TED%>_uM}O@|m0`K*<~;MlqNe(NeXT`-
z=?~{vjd}2`i*{S{gWS%8_enH^EVegZ((5Qf5JPQ$Li-nF^YQWl%+w%5z}h5uCR!Y@
z%uDLuxiwL@d$D(Y<#g*ETmR!zxYoRB{CQZf<w@-e(~8xaom#`sp*$I&<I%|AfCRR)
z!w+bOhJ<{B7$Wj9yHfi7mtAq8;A+k7?fS<>*l%-NTTj23?6TYDk9e$e<b}7C6$#6$
zTsAhG9wSj`v%Q=7)%b|+`su@cQ`MGcFrednY8(V2>ViQFB_ne{_GA&0#F(whF5##k
zXjz4@)h}gbcb~tQ?%(y78->B7sZ+-_E<1E;%GAMAJof5k*{0Nc%e2?Gr^>{I<c#Ru
zW%G=;NkOybb~~xlH7e?Kfxe|YP0+g6;=be4+|NFq4L#<rm7M?G=e?%;A2d)F6RC#!
z4GMa1WZe1s@^vMaKa$(ZcKIK!)Pvy>y)Ya?EK>Oz#SeMwO9T4Nsuvvs2#t)?^Ut8J
zbs5#Qv)Hz6$B*Ayb1U>k`&OBiF)Mz%HW%Df8oK&u#fyCL-#>j9%&-vow71$A&D*=J
zXy?@WlVz0kJ*-qlj&vGSZN8+le9Ad@jjkgad4A7FZ``!=eqHp|hC%I}cI;3(vdg^7
zo;@*x5^e5FPM&-*Z))C~PTLAQa$lPJkFoquE<lz8!RdF5j8tNSF4Xi3I8<Wq&=G>F
zBJc!eTNq*#;|SUM2lf@kIy=5OCbBhPXXkzV!t1{aZhma6s@nIfIVIF|S91HZx##XH
z)Ft+vIrHLx(*+L~nu_g{I4YY1?F##i86rf1UU>eEQT!+f|B+buP}eg3TRPx7iy$Ot
znIg9r7~t>X^po6W`??Rr)15ClxuxLdag%j6Dy5r$DL6k^p(g+20$_(Hk`yspN<?_?
z$Nj5vo{lJA5z*<4Vb1H<O}8btOAaZ$x5RR1ePyN5+_^g_75_QEuKSZu71$RqRSq%1
zau{2^uUmMg2}cSyNojx2ZcHok+TNXzwC2M4X`>_T1LjF__E<&z8pTZee7cvTqtQge
zEklBh3=}CFmJ9%+86oE{mh$-VPz{Zj<yUf%R&vcQMwxqONI&VBBua`%4`$u$xHU~l
zDazH%UdK>;@b2<?Gxyf6>Q+^?MXTrT?8*HFcFWvwP3*iRd_dnCy_*KTyS-|bxp{BV
zGH6SbH3gv_hQf6<()ocHmQQ8Zq$MG7YyWA$y49|O?(gi2!E@)%U*^=NL0;Sb{UTAK
zwex11@E(F6%}H&{YtN6m<VS@q+d{`QWpa*t;!>E;0@RN`)SC&%v&urG?1C&u0Sr8m
zkp9?HUr%*gQRkgIW2H5kn=5Zs^bNB)=rwE-x?#T}8Aa_$8X9*P)|D*wW)clV*GgPo
z2OqW@jM=GfQ+TIwlT>?k>H6t~>0r%QjTg9DLu4mywB0al{xa$Qfl<C=<T53m$1$6~
zpJOv|$<|AYO9J!!Kar4i7EcF^2D%CDOxTs1N6fCENf-Gr>(UJGCSpZ^E@7*q7pd#e
z)gor|0r%s>^c@#$*|f8-^{v$(AC2=8Q&ildGjMkRPc?GVCUFrJpm6Y5j^0f>0vqfh
zyw<KFAdS1eLh0zTl%U7m`j;>J1nKUYvF~GK`|ej3_guSi6@Op2v`c*LKohA3xzK#$
zN60egX+FGj+|YDZTb_E6mGiUty^n;2bs@wz;lhR0S65JxseC_Z?R9DRQ!Y?J8{+DI
z^2oyMH0_@H{)I10({}uvt8SkqSGLeJC(Efvix#k0$iQukSe6bF+^^MUUl5W-E>2%?
zYh}kjH4Y6~HV-3o<m5^#_Aej*zPE4Hgl5PrA)QUZ2+og?#)^I8B$1M=pGmsBv}DNk
z`kESTnT`$WwsYhXS!^=)hxDA)<dLIim)my*kRdU4>{P&D1;f;@B@K^yE$lvgxX;2%
zqWdoA+$$XWdVFiYCX@^-?1IlECl?D40-asfjj0_MN~^AJz*>6qW;5mB4jno?ANQlN
z@z68Z0o`Y*hQj)ySf+tQe(Hfo+M<z1L_PZTTNAy{f95$nIULun#iaNzSg;3U`qI)4
zRwG|yYtJA>MV3AmG*Ql<gsR9>r`A8XY6FC|WQnmNy}aPmFS-R>z!Y~>{Zwkc;$lo3
z0N@bpFwLDXEy^M^dlzR0s9prfwV?DozIB^-)rDh?9<8*}I3i+^WNe%XMGCtWG;WpS
z!K8}4+jlrPAVMPpSnvtX$w81INFqcvG;F+{j<Ck9|E^UcZb#ksZE5?1nY0H$xUJe7
zCs!yY_<;u$KQxybEEo{<?=M73D-%jX1IuH;(1_aN`|~K#U=55YKQ^r3DJQq}N&4MA
zJwr_Np>Rc-Wb@;3wzXCwpZEs_`wtl&lx(0&X5<en@~ZEr%wxW%Pb>YsYdhsc6<kZr
z>gwe<K`U_m`->h!r%VYu{}fHJf>cAPF(@osJu}nzz*%X@8Gsv`M0$(F{&2+IUPaNG
zUg<yy&kOS*uuK|I2km=wkq|6OXNoAg@pVL@jkA92T%9cno8mvv{XuR<eeN+m;?=u%
zv!9k2M<)gf7TAEQNosAaKa0^+?cA_us<t+hb)2o^kfh!(zh1DmYuBzI$4>@roiZ?9
zm60Aecl({p?jDJ4jNPD*q??8qp}2sWE96(+TQgQaF6_gyCpjH`#rEz^vGqBvbo%n1
zh+YX%Y}<;83ZdY^cnT0xV@<-k8}yWv?BYK7Mos#Ax;So#k5Bb(*TP7rame@YE8a$5
zvDT<t6_@QO-}xgp$MVvE=CerL!Hy9DJZ(|cc)6}0q5Z;(F`gDdGu5kC4C*E385r)t
zhWm?AtpcY0FNhmt3pfY~jGuuST}o&s7&oE|Bm3VxOVSht-jJEu=koHVYMb`{qTvcj
z5x9^Mo6}K*^O(EH^g#LW1IpW<y**1_4FOo#*P{5yv43GR5^68S{8XVYGBGuk>D;-W
zq9PM$yK7boGV_RXU1MkW2va0%3RP4Lu9&0V0H`5{59|Q3>&uoPycA9u)U)1T?yZV>
z+uM+>L>`clN(enHea>CG1>KozoEe^S#OTf>5NN&c+3Gw|x<lrag1g@LoTdfC$BiqF
zbGGsJPNq_!c&ZQ})lO$7BYjUfp+4WndA+oM$NHpoWn0Z8Z{|4$DCD8=j82$$wOC9^
zQ!^H4kNfwRk^!5L5Hc)No+nkZ`WFup$?Afa5TmCpKEi05i=OY%g^h!XgqBq$blLFU
zA+S{8$_yU7Q`gAK>K!@`qHIu$W=_96=#W_Ku$?5>^LBBWg5%NOR8izIW#C(7^4xd9
zZ51Z>$X=mu9mRc5Z;hJ#Uu23X&C{oSo~7qzW=`Jp*?>5OK&fF|cO-5=4IS(e3oHYR
zEjHHT%zV(A&<(#9azH7mzrDZhCRFQIPrRG}5YWa0wT<_oO4GPBV9YD%^X)IF(nV1E
zsfkRfD4@!52{vaS-z4oQ=jl^ODU0UL?Ise^%kB;A$}=1!W@<Wor1+@#;<mPp<<jyU
z)5|LhKSKX}^r(Y7tkKZYoor&!>DSen2iJ*^rL$2*m|yOCmVv%-Xx=I;=Z#G#4H^`Q
zdBnor7hY67kr$CwD-vSR0nkHJ|Dor&^4|TVZ{mAC15sgu&A>dO^CnO3>{^~%{iB5f
zOM$}k`OB9YlO}Z%DQG&Krntf4@@?#Tq(>FK4IzpYG+RrEM%yHv)Ml!C_gf4pT{mAd
z#TKkqCwUAD<v0`b2@dXgzucF_@ir^3kDI8u*4Z0|Q44t&NtZ6|%dH64IxiRwr44Fp
z=v^1>vq6H&V#VUceMLlGloN|}XTmq=>eP@TQT~K48D*FV4H!6(Zuu=ESMA7_SN9$Y
znOQU=Vjrt6MaI6_9^i`cBZ4JzM+mri_PRz+dW|}CS*xY5)YuooOyc<2oHwKQ2kic~
z*{yb>+wa=;$_0<?P$)Nixu#}VaD2R(Bo`p6#U~RcRaKYQ)y>n|GdQ!B9I2?N_4G<2
zCTVjHqy%>T$oPETB?+I%Kc&OUhaF(k<18vKuarDcgp9ef$Rjc7>eZ*EE!m%CH?3X!
zg%NHyNr}i@t)y9cz64ox;_&O<UDG|}`t9xR?!o@60v@kK3%lfP#<R~OvfqDw9_SXO
zb%7v0Tt|+7S=6G~O=yxY*<V`z1<8nV?=J1%jwL(2yE8&l^Qcp&L2_~eDVWb3ONseQ
zPv*?he0WrkSANyc{cN#!v3$<tapT;6ERyfP%Q<;ZT=8Bmmw6GpQ>Qmq=&koFO3mNU
za9%_{(en}2CaR&*ObjncDfCStELb>gG_3b%&aCSIsv{CYOV<~`C?FGb8B4#gFsfkf
z`t235SR9-kX1Xm4G@g=1)T2)y!@41Fo~$%WX=qRrK|fl$%t@=RyJJ@~8{3CRcX@A-
z^aUEkdll6El(xIz;w~vAjD~6^f`N4yv1!<%I;$v)C=05++*heRTk`U%Tmb3ye!tDN
zZz||9x9d%{uwiqKrlben+EgO#_42$loW{`m^RVgdsi1&60NV!a6QOzMkHji&64Ulu
zMFh!@09iuup5|eU!~?_lT7z)3*RN-_+))27XnbAB72Q?Dzhi`d=|$%byRx%QM3tbu
zWj|fZa?^S*%@5JiT{O7$&iyxGx2pH<GDIQ7@fj1tWH&_fx6oR7hTRpl>2-f<7h>qP
zdxHbxd1l$A3S+{$y6tgx_*ao?yWe`g<%Ddaf+sl`);EV`)r+Qmxv=K8o|Ec0S-+#>
zZ03nKQ>m1+8)sy;rD-aiXc9q4f=e;1cNaijNSS4u<L6x>S&!Ga5hLDG{WJgeb?zYo
znFJNR_M)oUQ}@?ul^+d^WSDKZ4t)AqOR9SMH|_fST*Y3@Kl2|mYe!Le>#dWwcRQs4
zH+h1Xn6TKksb$Y}Y@&|AHfWFjCO93;Y*uPMy`DBIVEBQ}o)fn!9-82G*Ggy7BqMP*
z&G7XhqZE7e_{Y$LLW8Evc;eknZQ3GIx4=I4?%AXB^--_b$Wy7dIF$1eXKiexs+y(*
z$_Z_hc!$!q*54Qg5AgJb9K+EozH^$2>|7HQYLHOzW>5u?8!4IYB9869YK0}9Tb-P;
zzb<@p`M;s<rhTPlo0`&uS&)5tk|?6?)D|>l(FF${Z<u1?N%8C&IIA+_NkPHZ!i23*
z=fjR4pG0^j<2^Z#nQ{FeJUlr63b^7tW;zE4VN0xjBHP`%<K1Jawphj3Ty`vW({f7?
z6?s|`gR*ri+{UwuAI}xXno~~%n{2$iv|QFqX-SEy)R}EljAm@5yPL8o0ucRe4dWte
z2e{O<c)>;*qwc3+1F~}(M>;xD{TVYduhiy5MjnWav|&*gDBo^QuZZb6bN%`;AZ^*Z
zc8Pxv>b2yk@VUyvc8F=O374KZVtu1}r@_K<rI0{@jGZI#)herAO<}N&^T5lODo1AD
zwHo99xQmPo*WCCGi-j=0;P)%g3s7}U)O!M@X$i@!3tkGr%n2rUz(LR#|JCzSH}k@s
zfg1JramVqJ1j9f&Wgv-VB_PUx?iqIdRcmyThLpE-5hBOMcQQd}E7wXMZ|#7S2Q~nj
z?tf^$6ug57Ecb5v9@$ND2Ha@@3HI`TajSK0Xk8upAf~ohG|>%Cke3E|#eW*%*RU~9
zVILxDX!dNYtQ_vn-0!{J{b;HYIET-O;%;wirBH<EDNM!y94J-dmu;}Ozxd1M&g=`A
zEYpIxxmn_PZ#kOvjfG37s9?8|N&+4aziMt<UZ>2bKs=;$lq}4D&A^%ecQX*{DYEan
zT+(LIBCMQJ*NXK156VSyx!@W!cW!x9;i8cRu=?Ws{5pF$3+X(J7QviQ10I*YIBc4$
z$ChQ-9Wbl0iW15=;}4mcexrX+9ZQ>x!QI$n1cQeUScr#E!LFj|r8f!SJ29(X()SF3
zeMR;H66k7gBTA;<_~uh2cEn3^u-!_OPJ;Z7U>k&|Lio?BRr{fkL!9eZv4;>6UQ8bV
z$EigRYM%%Tn=S4ZUAlU652lm})6lYqRxZE!Iue_m8#kW0$@{hL)?eY`(uiKvq<+%e
zbRi>-ByTT!JgZs7{5PLB$0cS&_SxdGhv_ae0gG#xMUL-2v1)z7lT{`24|GTu*+zeF
zx*q<vfZhQz4`zMJ-uPQUbK^Asm-ql%K&8KwC~&5_u5M95!B9E>F&idZSXp_HsM8X}
z<D}2w1qYn08_gSzm34Kxt~K(7kta>a?)Xm<P3(HP;*Zc2NRBrtvLdF1ckT51R2ZH}
za(()oaF$j*GFx!tEII7tKgj2;{;~fckhoRsYP_H&FKuUu?_6fZWcn(E)8s8+=;I1m
zG^s0_Gxy}=8M5_NM|<!z*&S>NJ^=q^`aYh5s*G~BM^&y<KUhwaM~t{a^$G2w@g&po
zQ&A6((h@M<rd@zvN5Pg~SOZ6sjp=z~I|l;)F?AA)cA3ZNg0;AmOcQsjR>%_?>4P!j
zeDtt9yOXQ5yOU-(pw1We#2l3nv^(#je}2QQ9Ai5_S2*NM6Dey6r)OPGoIH7nl~os!
zN(w~=rQX1P{d#ulfWMh)1`63Bo2($exR*0JP$}Y1m&ZEw3(vVYd@X`=EIXBhU0htS
zA?}yk04W$YLI)A+aO^P!h{yAXvm9t3F`QbR^mJ|gSBJ+ZtyfIx#Pt(7(g!F(qx;9H
zQ>Hxejqm=^!Syb+;Mv#JXxT6X(bm@f$AG_MH6|zn_r;J*JIZ<#E4bIOeQ^xC`I@k|
zF*Zpb&YO5RIn^*36ru4Wsm2Up2!k`-NW4fW%!D3)mZJOa0Q2F!5#PrDRU|xn@<d4?
zf=2@v=Y)v`?jOB+t%`mkBci}GyMob=5)^6;4KM_F`x7-azx-%H*7pS_O>E87=dWKc
zwzB&2=@an*9MV-rTgg2Ka#S9b`$|s~v;S~vXa?W4+>oV6GZEh8k5bJ3Rw3qd_A>a}
z4!TYlFKjk0*v>LC4!14g<5AV9glZ2Pwy&rZAx|0C*rgX3zN#=qMFN+WHe>8DI0<Lw
zcM@SVnI3*4)Mjn;0n&S@GNvpl+2)TI7ZeJLf?wauy`$8D?g3O|nGerak^N?kT7#hy
zOPq`lOG|mblMbzIr{lTB`Sj#W3l8{yqs_U*5IuZzd+_PgiV$BoUU*WmPy23h7H~nm
zwQnAEKQsBx+5!THJZ@Wh5)3`|*t(M<rv)&e%&st`f$x1fS?qu4$LvPIedrzCH^D%8
z{M!%5b1Ni#aM&p97<OkCw;UJC1$9Re&%7Goj9Y;H;_?I|%*_6$@LB<8sakWBR5PX>
z2-sZ0*M%+5h&0@?Z4cK9h1q$t5d)9*Qt<(^h&+D0l6B?f_h&q2O-)fzG|68FifxhG
z9+X?NmTrZ+AwL&Su>1G!d7>~AA*c}=ZI2&oznR(DsraTsKcaHvUQiX&9&rX{$%okW
z3Vva%NpcDWn=L5mWDx?-zS_4(fAHIRH4vPbkDd-Xbo6KwT?ps~r0DNP5K3km8rBvi
zhIik|el{}RPpLfnf?V7;%>@<~ay@(A<|q!$$wygA&hGj@eZt42=iPfuJUN`r+4yJS
z3khimaxcvJYjpQFZ(g!FJA5q$Sqgzo3I0ejv$F1c{J1*r(lrc(Am8blBm3o^46J{Y
zcdZcwx}%5yQC$+znttN^z@B=Nam3|WOsm5sFvYcM@=WK4ATeF2j-Gp_R6|Vui_`=r
z7#xZg=H^&A;Fyw^`#*D;w&)BJ0N7!_ubMGe-Ab@w$;%hG^C{|hxWqc_c|K-Ws;v`t
zXlI9$Q3kTJqk;Rlb2eqK$ysFS|EuAwJawRQyxj}c{-^XJPoAv#jH%my>kW$}L}WCD
zglbQips#0qbgh^zjL7uC;|&*}CO>98rB3>wo_~2`m(V?p$Xyev_vxR(?^iUf%T*NE
zRizsd5rjn6z6-kszx~O`@2c<BlisPUNNw&nWrtu5uVQs>>Y?C{i+aHnb7`DMFo%%(
zR@yIGS4LF2YtrP&C(f(zL=c3G95W{8e7~vF7rf+QnaHVCEt6HyGxA6j2HC24_px|%
zbQJkuQt{@S*zqyvr6aUs&BUiUt7h1$nRNNM2J1iu$rw*qL$ZT=dhOaeMB1bqBj(>S
z!Mh}t^Hv}P5;689I@#ZU4q<5|UjaWf)fKJ_>C|J;Ae=2dxD;*GFmtL|uT&2=+9qG<
z(oHW4!Gx=VbdIw6S)LdQ_wVn4;^zYjkkX@Kh6`f6X8NNy7Zjfi5Ae%T4Xr9E2@MXu
z4WEHS38g#o)X__#cW2r~TV8iRE-j4QLwAg7fCzUPdnNbp<@mBaQ5V`Nk<Yy22Axh|
zee(3l6D;I6P<awJo&S9akjs(v{^3(GLL&baqmSoZV>t-z{O(=3k~yhmEn<sb-y7rQ
zrPFsxNL`#VZp5jHbcEeR6on5A<U}xPx|bAO>Ng<rRdF$a^I$3i*KBgwyeF!(?%j5w
zVsO`vn0E<@qH5Pmr9)XosCBvF6r}fpSx3iyC@3(j^T^)X_9x(`D~g27trxeP2TZ%@
zk-2iNGmJm?^a=5!Ici}l0TyT4g;?ACYrF5|4iH%yXLOR0x%53HP-eqw-E!)$fbVIY
z+2^_<?&)&9m2G^l(vl_Hl7AQ8Nuyg~gFRle55keaQ|$Z9vNt1A{4?H8kQH%M)LL!T
zx(&Ah2TsdDX?7FFBxUrqc3bhTgmoDnS2$Sg0jY6?6Kt<89H8yC@#+x35Fr{GHS)Cr
zE4@x4lsWR3LgnAzS<|aZX+NMy&cNRkUwehTVfxFZS9|o-7u{NswwJRW^n@~QEX22~
zMf2>|yk;zLwIDntC55Rl!`l|YOrHY`po7$%S<6(si!w4Cu8u9Hr#>^EaswY8c-e2>
zIR3dxxs`MDl9hzOq-b%yG+8$aCkW!NpgCt{`Dc|~WZ<}Oz2=0ZhZYu-Y7-x_u1!?o
z!{O~BFV%qC6dUQh)bv@iq<(3deygC7#r$=B<gXW|dg~Lv|6+h$b!V-=(}Q;++oX%L
z+toKje0OfOX$IK$&z)spK)QQB>An0IIDV{1%m%PU0Ix?I%Kbr6Co8)$vGDZcDV?^r
zXWYIQYidys#qZM9O=|6ZW^|pW)Z@@WKY=`!X>&5wR%v-<&I<}(sIW(Ni2b2Lfa3d}
zAnsa#-U(xSmTr#MR7|~RFLUPoZZPZRoG3V#@i4KQ5R9+n_667FV2ac-5<WLJB`T>f
z&v<4^JzzuaT+rqzP;0V>$oW$%Krp;uHsVw--2x5U13fy4*rVXX?)@GFW?UlWBRXv4
zF(_fe%Xil<IOMoD80Y=C6L7PECK)hfboZE_T|ym~smSK7ZgAoBr}*c|x%AWSH)EyO
zeN{d6ChPUpHQj<PB}nLd+Z&1MQP@M@gnG+8!8f~k7$+`q>Dc0KieDDmd)vdN?&21f
z+MPNJh_tnPS)WmQ>;8<|E4Pfa_V^2Cs6oD6gNmD3eC1b?hk!vuJxv-H*8V%WJ*(s^
zh$fatMw^CAzuKjz{=!T)`jso#q47`t_pN*T*WX@#!qBB*0Q8w2J*H`vDT`roPn~Xi
zSkLtEKW<*B8xRTQk7hVTcRX8&^l@6&I42`=VaKiKKfJVyjU5z8gt3rahUvx>ZN&it
zv}vj_Eo9HKTq#qHx+RaP9EB<NtdZt98QCAZU>A>}05itTi!zh9m);zQfjzI}<j4mv
zi;HDk<x+CrNZ)AsbNYKj)%Gr=RTd}hzyX#`3T+LaYnKF&1z!AYTnUcb++Jc_AMK{_
z*iCC{b#}e!nLQ6z-aa~Z`8(!8%a&b#T}{)<`kFO!CY3BjqNszUWd0b1ze_4&hnT08
zRsYCbN^YzIRW7W=npDguW01G!Jo|mm$JjABZU??~nX;VXU1+X!QxFTY>qn2S@f#L#
ztMwLQFfcs!2t5nuwu-kd;L;?S`2U7Y2u#2MMQG&1Onhr*%^mkR{JEos>Yi7w1B)%O
zO2c=8y9d$Ql`Dfk-L}5z3@Hy^n~@_4a5S1ThcY4Vf>A^g9$=5<{LLLU?lHCjvou|B
z*!reTw4E_#6Mvo$qnHjoX7ps%&%xnSxgcX4U0hy%R$SHSkltJ}%OfN-w)B70jKpbu
zTA7Xp8M-~IT9~f8J6avI&*It<B!l$ZWlxpgX5(9?Br-ObZr&%sQ27%&`}}!>gn{_o
zx2L}Jla>9(Ud!`V5&I)h?ZQL#D=kVH)sY#yY#9V6M(FmsK998744grqmCv)UCs(8w
zTNsOOzOG(sWyN^J$Z&XJbogZ^{;(yM%%4B4<v_n7UE?*l<q#}3HO4jztJ$Q#-`lO8
zTNt6CKH>#I6)XUp&8~iY;LJ3fnLU&Et&H%n`1>(?aSQl|j5pv<xrF`3J$%3im_kih
z!qK+oJ>}bUucKpqFU3?+fWC2x{h?JoN$O_Pb-W#7EWem)Hj0Zj4>Xe6{(psX>@LKV
z)F}6B<5pX^^<hjiJM-kVYt?wez(vKIiy9TG556r%vs$@Ea!m8O2(%`nNiD?@?sF`6
zKR4%+L4rqNPxbA-FP`Z&PBX`ug-#e(MdKnkhXI=LaRL+uN-<L9Mg~EiI1vvFa`6PV
zrQa!y3#)EhmVb>%wLR@S){0gR848YG^yJQ~9lU;~7L}L|Ra4stpBktDcR-%;`n7A7
z-&Qg4&%i<O0gn0Kvj4u7FJBmk@oi}?T}<N0(L48s{rZ!>^O|wLKg)hb4zsUbJM}X^
zg{g|_f*Tn5+_)j>!whnB6vHxr1hliLN>^>};4!t4*4O>G+VqxuCbaN(!{aub_#)vW
zCvo5`2WLgCwT$o}9499{*+}>cDwy1-<~<V1#-GTh<q`8i$7#A14;-buZwP_L$od7?
z<X8U?63+dq0i{&+j2`}gWhbjU2Cd_HE%_zgX?O?U9yIq_0mFefl~Q%<&}A|9-TZR!
zWLzI&lw0!EB=%Hf>encefyna~43D`SwvjrKAFL=Mo+Nr)3uz(DL2`WERYQ$6qr)DL
zGI{uoLpl0>#HM!fZ};Njjy=fuHNq#d@S2V~cz_U$Ze2^WGRgg$%7`e;wg$R~bjYUd
zIX~TJ{I?|?MgB|sJnPwg-;9ec<SE?udd!gn-M+Jl_6$dB^4f?j4;}5;i`mM%J5|0i
zy2SGc-r-;wXdnF__g7TBarG*mXQD%)p{tT!TOlEVmWf*{B+Qtmt)fGR4$WV%;LM9E
zU6Or4{-q=(nXqFnrx>KAt!?rOWxP8plVgQS6>yRJfGe1oZ?*#^D#a0|1`NOE54Tg3
zXTFIEh`Mf4;A=G7f!E`a-iKI?3pRQ(2a1~DbBskrj(;g1oP#c>enNJ0mo1Y9n6=(`
zT`fTE{q>fV4qJc~*GX63#3kcrEq*^1AQdcOz<>cJCMGDt08fCTN{eRW*$rhm^A;cG
zFGqhH>0m&lln?CH_+?{zbb8`25qn0@xTg3S3(N^cB3!$swU5AgfYnsf^&?UbiEp-O
zZS4GlywYa|=`<Z*UeYt+hhX~+9YjuHB|8ra@=4ttzzWxsl8W5k8y+|-)NvRmeUAip
z#0Yh;|4FDC<$fQ)oCRam%a>un01Qviwf{>moebV2I{BXjQ&9?k9Wo~<JQ}i~jC0<*
zQ*DLfoTEQo?2q(8=e}5>LQUj_x(dmgfof=Q#ifxUHq87$P@I;3!oX?wo;|?_MX-d@
zJH~M`K3*9-8}?D<uPT4d*NflFA(PVH`o-MTG|2eeZd@lzmdiqaWh##f2l0+zd^BoQ
zpcFEpZ13N<nc#1)i?4u#h2F2pW<1>-+@^;D_^p||{=KGb%KP`;iV|+SX&aI0Y()R&
ze>45jBkdnA7sAN|*0u9rr5Qf3ckjb2vDHx%u;cI0z*!_W<SY&K|43a`E<!@~?wwWO
z%o+9dId&7E#YS=<0WqRYhtLLG!U+bme#oRr$8WSmeEsU6s-|}6@k$mHzlsv#%^RcG
zQR-tuDmKRhgm5NJ-QINlvG2diAlRtj!Iq^QVoXW17mI0+*IWGl0P+QFJ0SN08yTvG
z;)lptXlG}K!Y13!4%nDIL+<AYZvb5$YL2}Fe+K8z!Zjdfv76i9`~aLpge_<_oB%F7
z(S`&OpghYp_10%d%4g~m1mDZ@<Re`_3z`Za#)}o@<zpL{wdL~1lc!8Mpfe8eWXj>?
z7eX43MeW=v7GmdE|N3V9jckKB2|?}=7ZKO(@^fW_gkXp+VK3De_Qv2bA~-m;Z1ehJ
zW5v1EIH6+tF5$zxu1$X2b!?J!zH{wvRG4h1e`fw61(U8iJr}Z&PHiiSC*fk0FH$0#
z_Ljv{=605F8|5VP(fEg4Xy&rrgJkY}gh}#`G$;r}#*)V*sP)P}m#ir6qP^T_w(}7K
z*X^u?U$gpVhI%%06<sm!A_7Hbu7=6%%3y8SJ^vqH?;Vfz|Mv|aMpi<SjLamJ>=D^A
zvsWmxl99;BNcKobl8j15vQk-<O-g1qQ5hklfs7*6{XFXWUDxlruj{`3aXudB_dLJ8
zbsV41`~7-7rzXM#Kaqfj66UDggNuCc@UDg=s~ho%M(Y@LfP#!4f4W)TbqKxi3ycW*
zyIQf{?AyCnw>k;`{vo};D+BOlEj~t|>;+Lk*m}cf&R=kPIO{D7;Ess~LJvGbs{yqj
z<ZgA+p;Mu)Z9~*ir86A21I~300ukUs5kVv*n`QLX-EcZ`VBNtZ(T+A?4vy*Nb??L)
z9sZoyDafRdblD+osDUT^AR@wneu4$s;0=U{nt=Khi+3?(zn)>l$^c8?AWp1D!Vl;X
zqT?8$jd;UP=Fc6Xz=?<-@H|kl-KMV6wIL)hR1E+nz=@1~?d)jH@I6gyL1y6IOE{(+
z6zhRH%pDiuV-z?LRwZj7&@jdo^lb+;{EtVr@kk^HT*1Rvg|*(^e$s>_FgRFot{uWd
zISrDUf6Nh|w2}|tkUXyo7p$*cNcAcPijJnMM?++U*q2ul5&#||=k-7`Cfcjm;}R0?
z%3mCWO&<{AR{7yAs3$<A*k_1GP%}VrChQ#7+Lp!O_4ZS{<;H}A*N;&qgfOT`l^XGZ
zKZLx<&@i!rcfD21Z(@tzPq)<Dn+Y?jD&x~B6FnL;pDqb9V{*b)LI~Nmg&{o{vnxu~
zxOVTpeIPbtstXs0Worp#$p5#|xAoB)Qs(XreFEfg=h`)H1}gmJfp6n*VPIl{#@3MV
z3|lM){|@~n!~`KU27C*`wJwazj;yWMpxXi={Wy9c1BXurKrQUWjW&;mKmX%ox;fJ0
zs3V_WxpD=H6$qA)#)RCXq@*1zEKZlqw&EU(J-PrU75F~TUy&fwM?)CSpf$1udJEkE
zzHuC_TO|k7!1Jxb+Ov<IIY5Z2lNxLHx1ZxwM8V>6g3xtz(G+0{f|<8ixp-%5LR`ar
zU60AA$fnirP4)?m@$8EU!41XFzKJIzX$(J+8Cz7I7J#_~><z!{n>Wt@HZGH3*hlko
z;hx#m4Mdv()@D`I;ysgPfJgTis^q|nFLR;y99iJoYZDECW?(}FfRB6}e*)3fO~g$A
zhX<@El&!-Ij5Osc@p`>Hcuv7CXJy%f&&6)u{Hoq>$qwh+uMUu*fIi^e1Z(*~=d3O2
z=Ya3ctp@NXRbn22AR&pwVc>mx`zH)lvUlJLJBtr{tJuKHnCyJD>wIC|yBnDTBD&r9
zB#1xWupD6@hc45Yxiv5E;@5Z1G68u=N;s+x*V?j#Pek{Zc!wVN7h*}|=G-X>cyEn;
z9#?}%i@6Eh7AB}>$Flubh@(nD0iuBM%MiJWl~!`S0N)@aKJ@enAmIKU{1a1C#5zD=
zZ3yFoU8;1yy{al1^(xlrRbPp+_5W`<(|5!%aUS*vv3`n9dZHSb1ocGm?bD!_A;!)!
zcr!e_K5+chp%I;I99JbJq69dsBmGzOj%jLMj*Ek^2r-{1a4)MEIqPowyx6KGr2B)z
zDMEs-?sk=@5X$U>n`FY{{(<ZDfpbRQ9v+|^jO=q_)y1bnf3NKGo8`HDB|Xm{$CEog
z0i+As*eD%HPCrX~^yU~=2iUCU2Ap?*A|*69mSHS$tr0%|-x{G#vX1~wLA0Zk<B&z^
z3qO7gM*U6_u>AAK$Y3fWAlOUJEq$ZoBMv!6M7?yY()aMV1_sI{wIV78d23+J1O%uE
zFYHiMEm_GAQtZC)5Q4ac5OI&}n!dRQ`tDhaesTmGu`~#NN)@g{sr!%;!wEvln}y-+
z0fBVoD6gk*12CYb33M0tZVJB~eIAjvQw(r7UOAA+s8Jknu@E!^FO+N<g)s)KD?%UB
zbUp(8#zJ@db`7F@HqPOgK?})Ve>oW$V(=Mh1GWOVpFteMd_rsNoppDug#gzj9S159
ziE-%Ia9g({9srv)ErF==0yKL5yy$wzt*#dzPN=q=$Dzi&qX-12#ucIKTzH=Ft``;U
z`4zNE5q`O8`}xB&Qtlt6B+ecV9yKj~vbda-wD|ky;<xr?dH0Bqk$QPYyl1b074q@<
zHSj3trGw?q*PuSe{3f@BH{!G1L{91Jw|wpgfl2|sMI<>M5@F%4AA5}m?t6o+^ZM`(
zVSEE>UiIq?yB@Se$cRDMcDC}4fADZ~&k+I+0>G<~*<s#cr9}K!d3`+>L2$6St1VFA
z%1D(@QES*ThB55!BsPA20Ge>;@NkU7_Fx-W7M`6CPfk&7?VS#<lEhXAyAl=x_FjNY
z59qD&xW07jg}Vj&3f_5T#XTTRjpp#E6MI3E^yzFb+CHakhNldo`?z=7+&Fac^S>0E
zo{M<vF!ID~t|oe6CB_dnJ0kDA;Otgic|=TQ**U{#7ApbVdx(1@jx~xzS^ryDb3v#)
zv?v2CnD?->bYjD2`ubi%rl(Kw$+~x^cv~xj-hb22P)s0^t1^Ymp|l&7#(+{_c~5Qx
z^UtT3ckf=u5c#egnfJO^9D2r{oexK1M6@lw4i6882Na6{TtT`S*ShTfoc13LR4;Bg
zNUN~9LpmgT=c@HM&#^=%wX{bkUVjMGynL<$ZOw9-TUWI>6r!6-QX_+2armNPtZjpK
z*U>X)_W3g~Fo=}xfhm3Wkqfj0jHFmAiTK9t+r;#FS7_Y60F=WB)8YQ;jdiNc!Vh4A
zIAC*)s0kqLrUvz)_%t;gz>JN~0RqdbMSNmNs73g`?|1xKY$Jbzq2BfOB8!q3*Xsq{
zfeybHh7;5?;O4{+YM*{0-g_Hiu<wVPTxspDKoo`G|2nbc_oR!QhqBX{5RB0ScLU}w
zdkh+u<pNhA8-qJj3UOf~3UxyAwR;sf2`Uc+`@MZS$2J+#eNPIdm~01*@rguEYa;5c
zLde}u@9#m}<NofMDDptC+07hCip34~b3y8MAHEQpNIM4yq;lXzEe&x`ga+0`9KEM$
zpkiicFT=g9f1RKpD~r(Zw`<a;hXMuH-blx!phfPa5M@Zgxtybi27C7fqy-eVpSuJ0
zQFI=F1uU~#YuiXdu$SQn`#FA2ReB=v{Uu+)VDs|`);QrqhoIqRQX|$#t|-!-6p}u2
z<R>6v4iLCd-vL%zcisDxg%Hx|xG-E6{W}6(jc5Xx0)%^Sn=RP0RWdP~k*gF>xP~8Y
zr6Vw@0nQm#*bxEqF-eVH-o;PFPZ5wm$7{4qf=49etqgZW5EBdSeZ3Su20m)?q=yd!
z_U-e{^=xbWaZ0i(7s`Uf*E87uK7MI)LOoYb4jBR3nnPv{*~3Juz}#GcfdGV~bG&bz
zmuH;+OTOjuK;nmpitxEx_^f=qyl&mT%|u1c7lN^fZ*HvbW7iQet`lr=Np=FMvb1a8
zzS)Ve#6G6Y(37_*F&F=_1RC|*@oOn58X5~nv1(v^hX`6iOw7p2%I*3^cYE8g+DTv|
zT6y89RZBW?5!gBkW&U71@R7ArQ+tW1HImHI((CV@zd<8JYds&%Y24d4+KC~b;#6fb
zUt6d;72yD=wI+4|qv5*wW+T=_??w2bBxasBH4d{%un|0r3+}UaOmw`u!C=YFx2p=|
zda>El@^YPus!;^7w`sdCPk;AUDxC$Y2!R;&IE3Gt`@fZw1R_NZaBb@t43bpT)V5{A
zfC3QPPBbZWi%FXriH0`<g@he?2NgM<?JNOw?_|V1&3G^|9(Y!{%fC>^@k%z9-j<;3
zSQ@Jg`Lf-gVe2P&KqF<&em61kJO)|3d-!KbBvQ=$Z;n9TeZ87balKW)V`c~bOH51>
zA|l<Zs%RiP!A?L9J7CT<TBs8+DKN%3IG%XX`95)bdNB0we>R5dHCLc}C6*6^y@tSx
z;D?lcOvvSyZWujub>pGpfb`iLAV=5p8{1W!^eU^HMk-L^svC}Yz{z3m@`h(8g@(0m
zp?P8FT_6zOY?#UX;1^R+IJb1L0^1ZiZVqtR=pBmEEyH_<4-D9-7PGuX5JJAFw{5HV
zW(<%AUVO?B&@)JX2OT%rQd3dk<0y^>QX~e0o5J5KP=SlU$jGRZ-rM;$&<W!ZlNu{6
zMZo^i2&A(cIo5|jyVM#EA;_oTEat@TpCD(4UEw$K%4w;nhz$4wozpSmssBGDnYRyQ
zArQy|H>Z<Ri(eo{o0rCKuiJ|y|2^HidI-M(A`wFVNdk-(h@U|4`g97Om6}fvs|hfx
z!?`3*^z#El;PM+Md37cZ&W*1r_2v6RzOO~qUYrS%qB?XBa6isQEE7yLFg*C%WSSMV
z&J#z1)*<h=->(Mz|M(^voMsN7s~Us`zV7hGaZC-MI50Al-bBk1h|uPlWxNcn$H2xJ
zzc&yWb2tbkbNh69H0u-h^wXzL9lZFc(g(u@fEnN(e6bl!4OlrX>hqL5Z>8xtDl3Ps
zyKN7c7!ra@9EveGYLJ|mEx^b;lB%e<YdH&PHb_Xoh~C#{e;(gX$a>*7J0;(YyYqwL
z)PE&+n27M3=kAtX59+cjfnm3k{@lWcviQl?7$%x6>prPh{_;csuK;z4_zX11&8=?8
z$cm|Kecm&#>lwQOrA1zKGGaBh6Yv*O|E+1*kBDE?qTqkdx-sI3Gkp7O``YKUpf~L9
z4La;w_N?rA4!5aNPS3TgQv3+k)LJdonq0!Jf_Ql18U%I%m>L!%v~(jI9W@)TfQ{wJ
z4&nMInwTJjS>6DG76_mW<r9nnh*ZK$iccA*fx1u%v5Oh+4B~}ydayr2V0#IT+4v_P
z^PHTW*P%XdoLoZp5VW;L1DINn#D#PE(7}UchYyxIn<0)%J2#9=suOm*Ul3|zMLf6q
zp;G~4@%5b1H*W-x6I8wGoWh`GEX^Eynlk|G9{zt=J3usF|MChN#unciKwH-|JuSUN
zx$N(ef&sILo0xl45UUG4hQE@q_K7S50b*|5j(vJMtCf-(mJhJHuy%C|J=wMY+B@c|
z47&??qP`mHPRfU?NcDePNc?`4zL|$TQblnRt1tYp#DCkpc6AW!BZu4`P^r=1$0*Zi
zU*9g-JfPOtE{wgcbKy3FB@3=Tj9nxIg}eG;{`L-AQnLjnS$aonbBO5`|Fx00m6EUw
z4<Esm22}tGW+9_QH&xL9q%)X=({#{h1d}kfZ}1by?m@VMg~c$YyZG;Y8P&Q^93MX#
zh(jbGibd8|R}p2w4sVsR%`Xjjf8Zp63>i0K9XEm&p?Kl5$!N^>#=!#pPD=~ba4|RT
zx@H#Vs~=TQlsW1mGv{xB)CyEG@jq`a`N>ty+LtUXFC!O(71R-cj;^k!qit75y{?YV
zBy>6^#K$KNIL2z;d{N`(2Q&*;JH%RRFLvm^5z8p=kQfyWGJ}V;>qBC)*~SR3gfNWM
zT8t@6;#emM+)Om_clrK|0paK&y0^OjoLP_~yY;NJE?E8gA$DmCp>bY8>B>n`sq9q8
z4m%NYDnZG6ZknZ!e?~;y=u~i*;-S8k5goCOTFac9`Vh%(5t2AE7W?q?kDl4K*J^nh
zpEvYZe0SO`-=g$TrAP4Hsk9^FXC9h<eCPH0Pb&4ngNJ?T>NaRbFYW5*<(ZgRY~A<R
zdyd5-sxg%v4;<D-KsHc>;+Z#{LpS`@pd|~zn=UgVIRP4pK+Ma`A3*D1fm*$EJW|Yu
zEK#M!yMoCT^()gD-wBzhmqAoMvP+e!kyxphYXq!&%F`cw+8o!sbLmbN7L;Jhz>xo$
z|3vdz1snBQsf!4%hqE5C8yltHa&q|>aQi}{asthtSVBw=xVgFlx?-e>6xdM*^6$Ze
z4%7%NK2gR?$jr<P++1L5lgBUr9mA?0rly#-Z%;%Y=gcxOi>m+o33Pu0qcAl^Dlt%C
zuom|Et`v|S7#ka78i%I^qL2G$fMO~HY-eG)Yz@sX;8ni4NBbE=#WWh?=;W#l{@T#`
zH<&SSIi*otvQE`;1VsGsPdhvx5R=Y%ZxFf32YYk;98L5R8sm8O_>4pTgvY{gi5*E%
z@7vl6(biUBM0Y6g$?RNiT%(Q$J`*6lfD=%WX)cHWUcf%6%7pm{&E`e7<4T;&Oz>nA
zwQhuKsj2z*1@P<evZJjEApi)dkBw!oB-x{!P+%W+Qi#|D0S(t3y<j-ZE7}E(1s=QZ
zLZtq}K+)azwfGDXb}szD`u(ekK>R|FPdDejo;JDfTGJ`xsIr5NZ3K$tq*lCUnBiN$
z0FVH>u;}r-rldAUlZi$Rnpv|24q^<>T=pI);S$N9oVjP_!pX*#e)DEj9FH`<S(G<Q
zoq4^R$(hSP>5#E%mI3y1Am`VX?Jtw3)Yk_CR7Us<YG#no)bfQ~86s?`Xl|r>Y^*r{
zQ!}wUQ5V+nN3OKcE|!#PBtr-U0TuLhs8CH#RyzkPAaOV=Dq??17YKe49eIL+wA9o+
zCb}Bn%X4JSaJPNy?64~Vt@rciPjF2Vkz+`tGwAOGei<R1N!mM+q{#<_pf4~5Jtqre
z^<46~u+%`gBFGGlB=moOJ$ObO4@?eVp$3~7D#_+TyU>xo{bHm_Xw@Mb5&IU+l<1!H
zF<dtWhjtJQ1a3fQ0o1hU0Hc9ss)=zqMR?k7RK5Ix*-ZB0+S&8xOO&`lX`-1D+28;U
ztYJ-o<X)qf?hH%I^f#{H%@PE&R99cbww8C%87opx(@|KxESs;%N?2Bwi3e!YSv+H)
zy((-gk`x1ql1WKbahA^#?v^=@E%cVw1s^6Rl=;bAKTJ>OG7s|G|Hkc5b)n$C?I-tA
zG{HwmG*qxTnj{?(S9%zSN?Dd!Dab@UA*;cG46ULBTW4b<4-zEt%0UT)(0O2D2!vP9
zVSO8V`z#ysZyilqLt;MjWc4@!`5S$9eh+rSii%vL0(_w;Yx4sMVzMi%ncZ9V5;ib$
z#Qa==ZgRBl(h1{Uq|Oj!t+PHrQ`SH3E3+h)fUilNb8%6L6rrWCM|3cM{Wbyt`Cd*X
z<2bCblNMIP4h#E^p>PPI^2^V|PUrONE8-$tUN7U7TMtwq8XO9(UNW?*=Da>@Zy)|a
z2n&dd-EO>6PTLI*?qNFh!iLu2-u55pl-n)qoevcZKmSLpk;|ukf7?Y!RSXV7G9OVY
z4onY$E06?Zx+hUmQfd%Ljcbe(hn@t(XU|iq@J3=2xCo2pPTXt-Fr56wrV%Zi`ek2P
zT^;8ZW)&3`py4L2e?<6ze&#!sm6w|Y57|v+6TY60_m3zkD}(b7?-ZW56z$wuLrCL!
zEskJLGTkp`e^Ddxu&nHSNh1W0Zd>PsT&^5*UFGW>bZ})F&dM2xgk(|Zv^!*(R@H}h
z(_Oz~a3?V3KU#o32h8>YG}_+=Pq?$%@l>im*cTjR^4T9%NIrMQpzs*li@f>rpe9ID
zA!Ht?(mVVFBsS3c-23-Y`~*xMERM%v#0F{Q8evDqV5-IHkCMEq3#cReFgdAz`!G+k
zHWZ0$=nJwdp$c>Qdlzbq_+S7WNqVhAP-eh0^9R8vfJ+d8$HK@M7GR22a1`85QF~RM
z^z^}T8UP~;!os0)Q}muSKsz&Rj7ZQ2gw^)V!`Bx%FlM!<h^{$-Lx+F^z(iLVDObqi
z;~2(`>l}LpZgiwmV^@O|5ME(SsA_6k<_+>M1T`IMv{8Spsn~%0@%I3GV|MJrF{X$h
z=cxvw7>wvY1JZS?b8Ts<GWd@|+oQLDTmF8>`%vkT_+2lr4+dmjUS5;$2biv4oxmA`
zbC*~NfNVh2$FaPGsb>XHPUfz{x2>ebP!hx^-QQoO!=VcB=vtvJ2Dm{_<p`4l(@TB?
zq_JPUsnX#goZ}e-)GU=+7aDh*D<_5Ug==gt!@ioRNzS&htiF2H=u$A<&!25dr$)}|
zKH=O?LqkKbLA2@N9#5t;#bAZ-#%JDh*Ksn%97WEW3z?I|2%v$FukRE+k%zh(s=&Q_
z)r3;Olx6$4L?kDhA_{aQZ=l~VAmH3sojg?W_)2l@NdCFaeHp1i{dHF&+NXP+%JGfG
zzWjh>#DcHsvOe<w4^3D1V}<OFg}9qFYJSA2KMfNxdJfo;n~O+wL?sS1f~E!L;<8V=
zjgi(dBJMfOyPg~k+!t_IYJQBz+Eata0RC*8%X_`PAR%V>e06V3BQ1TJ&IK@^$X3hg
z0pWCiKWSX{R~#J3^!wM&iWF@&H8QHO?hSLTAQnk7Gy6y(<5vt}+`tbY-X`+u2M_wF
z4Pg`P4!lnG-=8glZm~Qkj#7G#C*F`A9*BF9KWbPQoPJ0of2+9ka5}jJP7`VAaM5o*
zEM(Xnr#%kCtqNBIrcI=iWEq&BI04xi;+YYbHMbE1?&k@pdc+sOPJ$~Q(>k<!OH0a$
z4&q~52!8(rhDbxky)951+3KKoM;7R?%b~P1NmtOx&^rIk<$42K_sf@&O%omIlE;dT
znjzx>+6IILzIJ34vwll^wJ47Jd;!FmynHQ80O(1jr6q)5ih@WKPoo5fpFQ!>elD<S
zWNB_*^ZTy|K-T5z6+dKZoas!5UbqKm2G%`{UGuwoU)b>TLhGD=o#q$cEE_~&n7-ir
z&f7mOHvwq_wv(b0m(!o&cctzt_BH_!*xu1W9)gM7icW6W&531wIpDslN$O><t7*^7
zEiBf5{zM$}Kd<tB<NKNOj8eCX*UDYaE(nx~j;5d0J&EKQsxYYM(aS|aKwdG2pRo-}
z);Tze(<RPvgW@9=DLdRNiKK2i=<v9-6mXLGNqt}BUlSP$hbgRH19tCW%+UXuFmVPa
zur+)U0q@_wJ;)pjM8V*O^`~`7d<Hr%oP>OeD)*{iG7wFTp)#t-?I9M$^eKwr`j_8@
zQlEl>pFmqt=7-4fnPogMC;|d}W2m3QQ_TQ5E*3O;6NhHRvHO1+k@O^t96@Ucm>^Gk
zfX-mrp#bZi565bnkU($rz~IOqNQh7#1;qiXpz!ViO#}Ig3k)G#A-E?X%Y_fL_0=mZ
z$GPwqK|4n*P}Tu<iJq;(!BN+AS<565M!ZwE@7!4r41(~(Ws>$fi0l5-WXWIV>UaI>
zi-gRm3ch^Eg-i}0&zGK-_BJ{MbojApbrEa;C+GK@4ejM`ylu}ffr|yQeVWJ*_>)2A
z*0N);qeb$G?~P#=cN`trMOzCCl?0=l`$C#Rmwt%~0y9DWP>Uj-lgQ;uB1tTG$V=Jx
zQiy+yJO7Z4Li{{J`IQp_aD=D^6?fqo?F%;&U;;3UTeOfpQ5BE(OUJUiejlhStY7RQ
z7FHqsX|9f!9Igs()SFt>_??HcV=xDk7WhFZyW1T~(NLAir0usz{;sFT5D|e;K!C6N
zwHDXPm(R$|d;tn5EZtx_h~RE5Uc&s!5<|7{s{5Z{utcQi-mSZVYya!+1+@}a>Yv>u
z<M`qEgoi=+S3IZXq?b3+3bu(8|JvPS4p8iObm%_$tr{R(^Hb<#J)Oh+@BV_te&fZB
zPnpLD+;1~rIe^1)(9A}Jr6cG0<=sE<p9dGE`zsD{No16C&{)eM``_UV6|LiS!{f)J
zt?^ybF3`fhudH_?C&wPaE=jF?R4M{Y$mfHzpECgs5_Q;ka$aLEJOZU9xGh{fBn`mx
zwQ_n|sLXMv%|AGnW$?VI2{6=tCT+0YwzgULR*}kp&Krq7Zb(@0B*G^qj`5H`slZGy
zWQ6~Sj0`lV@cp2h2VD)8mJqvwf2_Xv{oJqZQpP%Ah7nOw*{@|`+)1aa8DzHD=#eBA
zKjzz3fJp^9gx_i)Ab98ymL=76=_v7@ncd>&F<HVRjcRP1N}6*>-Pu)PZM<dms>|84
zHGt%hWPvBX@|Y{}owjd|!<DeKzKIzxz|SxL!YQNxpz@b4&M4`SYPHsWrp0FV#idvC
zZ{LoLH_?M$M#fg0{#_|DT*n}VvF5-)YF<qj-e}X&D#Gyt0#V3th_5-OOw2zn7jYjE
zOR8}vuL7uzxqSJ@Y)-;rub|^M<+betW8XFD^JPG}y(_VgM!{gEJVrKruy9i0-)`P8
zd_mTI7j14U)0hq+xFTu5WC05GtQ5Oj0;v1AIJxs5BFqKrpauXsXSpINFMsEy9pqiq
zl2?s#&`SVmGX9EtW1Y}b?kr)6mkJLkYI+=(9gOwE4531R`~p{o^8u!oY0rM%jF)z-
zqx9;ZydkVH89}@?3c2(gCt=P3r+}@J{q*wDyIvg;{DOkWMl{GW$lUZV9${!+yDG-+
z=u~?8q;$d1sc3zLp30|{tRE&P*4(<^&Uor%_Bc%?P*I3~z1Gen&(YhLRk8E)7ajpB
zvxnpl+(Od?X=nD$BT*cfi=d?fSLx(2lp5exKnHd3?84sgMxDbgf-4iwRfyz7DVn=G
z7mnQdqa?UIa<LvOad&@KlVikb+v7!SVi*Ze0Va##880f10iYTKrN4Im&!vtk%)HVw
zH=Vp0bwe=7lPC!=&HJlM|KW2&R#JIEffj!f&=m-}lHP(?Gu*0^^KGU=Z>aX-ceu-l
z(Qwv?OLlgKQ}qY5J1lIE$U?M4XHsLJngvX<O<NLcTl)Gk6(-%ZZSuNLX1zD?a;v{_
zXuo_8y336k-=n_LXUFJ#7f6{4F|_mTr(XLU`tv*2h-O~XL#xNW!cGpf@h6Dp_fx0n
z3B(#hSRykLi8^%j0V+pVjUGl*<T_&Q2qJhRK|5Wokl8_>!|?+LlH9MAzFD7t8l)^4
zM_7+>jADn`L?#~&ayNqC%pG4G(;30P*SDv~HpX#jCb;^Mj+-3t!J1gLRydfSz&v{<
z1fFlRNirp*SzvnPW7I+~)(8Q5x#D>1mfG~Myp`p76=ySEY$jzULrMXieXfxO>Qd$#
z==TJ3dh6`EQi3vpE#&0+U%+2LF@-lz-OBFo`3CtBHi+(~Y=n-Dh+RAI_vx{`br#AE
z3~dAVHYRM3N*DN!NzTT}$apOA=2Io&cih>W<XZe%ow<JG%^3aNAey0vjnnNSJy#T9
zQQ67HCi^)|w_~d5Fr_tbV&>QFXwW6bHei6RpxZ;mi|LV@d%SWc0|{bp5YjY;Rz_qm
zoqh9I*_4MhJgxiMY2^Ie(JXQ16G+yEuVHw&b!HhBLiFjVpUsp-*B{g&#;zO$P}eZp
z;0kkGF%6C<80j9j_lsq7OeHUdC;-$Zo{^v57l5EM$-#<Dr0B<|r%&IgDjl3tNrHXx
zL51vlAf*D7L5v-MBFgMu(vTf7VI4g(k3i1evg7(Q2duX5J9g}te4{`o{|We`U4MSJ
zmFN`9zjZ^2AbpNjG-?WwR=ajBevf`})NjqA!Q~B0oEcJl*1ir%b1QOj2Je1&Dth~k
z)>N*_lNq;fZz45s;${g;#Ci8I*bxhJbGHys6h>w>TFnMx2n!mrr^3tye`BQ~2H{Vn
z#x%w$YEJu4-P%SyU+aLaabZFK9H_mP)>hCBs&4T*cC%#H?gTLYMv@pliH~1qAKhT}
z|HSR37bXOiC)iK`P0)`p4k90SSIkixhu?CusX`J(Rob@E57dn8Jv(G5mNkt{w|G`&
z!DX2LD0(B{$1mSEPwDA6!X#aFoOe@*lWeDIU=W@hJajma<-%u?TIC!qV54Y0y-D$k
zJI-qy!UZDL$DT}E59={8{Qk+4HKhBzqLXv%RdTd^f&oLuoR1N~4~!gYvWt-MVx1M(
ziNU3+vJ$vGQel{5O$`l0Z!~vM)6ls#XR2q<NbpTMFaq!qwypsTjjy4H-T{9TBI?1t
zXF8``FC}Q-a7XduSFN}0MG_=<dqn6z&qBW+jmV2L8LXz^BB|7`=@pfZrzr2K|6xZe
zvj~(6fE{n*EV6#kpp$N;D3ZY4ET1w~AIwGg*mc((hw=|Ya^(kDS;6=M9&WaXZj7@B
z#SAHHRdqFxK+qa1$A%N-XIk32@POfMBhp;}#Gq(0_1TO|-YZQG4>vBP+@CYs_I4df
zGrT{5S92@&TWIC5+NY<p0f9W-#$-b1ax7a-e#}Oj)Or@c%-md0Rmx{48;a(x+X;uF
z={Q`@ol}2=fk4K=Z30INbR^xMaQA4ascl=R0!qdexJW`kwt|akPQ$E5fyvF}<hku#
zw?yKOhMvj_*>x(SMYCbzFm2S-x6`T`H*c!g@9_&r?@c!h*UJ;OD4oKg%NWClR<3QY
z>w0313Q%FPN8e=k49iRdMsGq0&V-T_<?BPFp$%`yL{pb^fTBUnaZ*VC*fC<9{Yo65
zX$|3)Xgxj_J?vNRV-MHW>~HW_hW^7bZ-^CFQtV4<PgJj!;qY75597oWX^_^0;~Vr>
zkM=97)qE{{w$Sas=4n*W|G-vN+?}&|;C8|eyzI<EC-cuB3i9Yy*;LUPXefcN5FTS_
z9a#IeBVpXe;{t_3aB4I?Z3DQL<bydbJ1E2xot;GNJgnXyJwHyujf*iea~yCNRuh<P
zfJ{Yk9f30S&YepOqw*LglUm{N$;-nZ(Z|&A+Y$e7#OEts{UX}ROomprkTxPHA;uaE
zw#I6j@`2<au#FhC(cOiMe@~rSj+^@D_tnR<43Hm;bqg>cfM^WU2$Ki+Adv}$$XqfE
zVFe~_2#daR2P&=NqOJFNQXUo;Bll8(@Bjlra~x~>KIJcDc)+lrI`R&aPQqZCkvH+{
zn%0!33?W}KrP+pG7Q#KvMdCfLKU<2tj4wXk-hB09$U7!Mo`;jHAz1H%0_XnX(<SDF
z?y<4ENl7`CPct(KtC8O%CiT<T_bJn(;lA}JYL|K4QPde{c7wikH*|-*wsNMBO~e^J
zX6Ko7dHsqS7x@O|Y->hCY7#UPQ?=nS$1Zk|B-4y+{9sXbIZ$36cU_XD<Haihedp|g
z#^RHejF;vxk{cGiyKjaUf^PTWYp6Wy?&eO^glPlH1K_a4nrh<mlL|E`p~ayTyW>EU
z;dkE7p;`U@Z;mXs?+r^V;&By1(T4V&7SnPygr@yo`12b}s;eC5BLMq<%MB*6g21Mu
zrKiWamRnm>W55$WnEVS%38L7E`H@n0W;4&z$yo*&{9IroZL5_&<3JxSO7?$ZgW@3;
z=H`CjkXK|8F#F(wItMHGz(iFw-lSFpMa69j{QA`pcN*W0X6+`fE|*g;u$4nGNeMBn
zC@Po?Yx09S6sZFMpDIm%#UC+0pNbH8>5@j#8$n)PQlGcly1J%5Ti_PN=+nvio%6Vt
z!pF>+EO?LWMwJ8@XSjAIB<khi;lg7fU$||nXq3UJQ^6C@x2yy-hHcri@Lg)HX8oh)
zj=6<*tn#in%dTEcofn|xq{Pd$Wh*lQwX5|$pKkI*BhQPt{jk+$W@ZAmMr!M%*UamR
z(o&>jV%p%lW7JCxgQux{_<R|!1<e&SB)Po0N)z?=fwq^|G9V1}4B}LSjx)sa&y}Ou
z_zr-`ZQU9RZ8{2I-Jm@Y5~6#V1xUMp@awjEiGr)BkSH$jFflR;<IIkmS2`=$EPoYf
z!za9J#L);n3@$$mX(CB~&8i<g%5sJS<0dr~)n(k)e>=3n(ZPiV(DkpQ8YNC3(ynKJ
zMGN3+Q^$k{l67F6K|UR`pgH1M4<Ku6WCVv_gi0153V7f#IHObx&(a+|uF3)?n!O4N
zzhA`cKw2N7CBVKz2hM#ND9Gk!V0k9{u3cwdKf8%HKoQ3~5E%Tqrn{h`j~h$_>&Lib
z2bl6*3U`V1Q)=q!dbgx>xfvi<gq#?g2zS>ot+V!|+v``%@G-|6?L}J^W~+U%oW-Sx
zALDUuN>pNaaqm&5_!bN;a1f}RGRCHd=m$>MSaq^9XAXN&+!W@j%d|T2HlHn`QOALi
zar}c?xVrk$hRfA|((l5VyK~nrUvKYsRXzs=1<$rTo(1&_FQPI7g2eXSJ%ck7HG;pu
z+Tv(|xIMG`D%fc}R7%|3?>)&q$Mba3h_yVs{Z{h82bH*atG7E2<UkW;Avkzn_m+W~
zV0K|4%_`%_V_Px=XFu&q6(wr2a1#RDEKKlMw~{@4_^`CJF90okfi4B<noHPQoI!&N
zlnpSk><ZP;HHPsKQ|g(f1X;gDZkiEcsPFUJZAEs30VzlEDW}#Q`1$}zVz~vf3pp;O
z|AGXY8_ezi{Eo}FA>AKAhy<h|!nE$SINpdy7NSpJ+R#iq#4p%(1Yzwkd#JbfgVR72
z>Vafzq2NIf3Jh{C<<1=x0m3Ce2+=Y#vy|nKScGUAxjf9>*RK2IJiLy%6K$9YC(O-Z
zD!!MS3sd_6NlEN(u#;RdIV!Hr<?R>uREFyWRhZJJS%Kz<I0>2WEjaZKcuOKT7J7l4
zB$_NP6dB@^&f3QVI0e(TUWQ}K@<(NL3a6rM398x;1z6@^?$B>A@VxhDVmd|l>=iA<
zvO(fPuH51vR?IqH2pM8lh8Vb3mmVrg%4pLfv+d-_Pj9Ug7q|vOT{G&28t-;SVb#?s
zC_r)NewklbaCdXtLKvuP+4IW$mexPYtMEG=p+$uzGW+HpI6XwLv$Kjj0<5R|*^7H@
z*3-peg*22Pd2+~-UGWObE*JQW`%bvm60I1|w+cf@!`d6ah_f@_=XP1yGTJF5<qnFb
zp4W475{Pb6PG0=KVUW(IV$gZzf-ZwX>uwh+JTSWBzlIYE!6;+|p%i?&cvSLFjg6Y)
znfx*{6U`@&%j1T2F0e%e8UO=V|E@wI#*cZ2CK3V|&+lKY2<Tc;)mq5W1o{^n+aAv@
z4eT4|B49R9q3{LbFGo!vSB~4*St0iDV_*kVt62T@pa{C#b^JuX2KN|JVoX0-Gor~5
zP1!`F-*3n}@Sg-s0usT-Ce!7XQku>hoL!+&G{5RD{lv{9<u)<}kNoxPMrfNH8$*RC
zyh;#ZgXJ;&63!NN@R!a>eDR2ijdR%zOFIC!j>oT0n>lubYzAs-hn1a~7?sa|B&{-h
z?uLn!XliRnt$g76-m{~$8z~BZ7AfCU4V;vU7lWh}Cl1{Ic+_yX1ZbB)<8L7O15dfc
zcX;vfNCREGt_3?;9n5o(pJru|%!OxL0hho~0;CNoVvqp~+^{-zDna{?x^(a;o&sXn
z1JX|ghIO+HrlzM0Rs&Hn*h3|Z{}mPDK<g1Zf$<<sA?Eo4KDwLar9v3QA%UaceR%PW
zeFBv9pxe+GSchQSDo!S--%u@4i|E~)irc?_eoxSr=BITGwEK&h#@pSBf(F`T$3+v1
zy1+`y?%x3L%-B`YD=bZ_1PhmrPTa#-Ss9tEv*|qk5a+FYzC`4CzLX-AOI#<Vz-Eds
zH63>kt&B|7SG7OhJUDlrIQQ{a#0!)=xx4ER3eQ%(xOaQG<R!RJM4*DQfAE0Tl%+6q
zaA-AR41j=<5o50|w{k|Hsp-yJok1UB_R$Jnz1Hf0%bo`b=A#UNR56dklWT0uiSETC
zL7UH9=!xCU`1z~DdIK2>i|5OwE=<Y>l)Gi`)Zu7twFptlYN+yge2#;lbb2rOmQbE5
zme9~Qp|5N>?IzWo&PB;z^bQKMv!9sK%4A6ik$ZSKApuwUHBCNTQcK^z59RiUhk?0&
zQZ@L2s0RTV^!9DfUAz^*j`@=y*D$3F0<?o~6Wax@069*2c^N=4xIcRactr;t83)W|
zwU9Cz5U_!(3px-yF$M+<w>t4o^;rtY#vQ}E0K?kK>MDNQJyI(A-NbM;d}{oHsiZlv
zRzx-g<_Ub4S6+TVp2I8-Z@^^uFCuyce-I{`X&l=i_>c)F0|OasC<F%bCA<tceqWNW
zA=0>mJQG7>>)o@c{4BQ70SCA<fQj%bmGy?&-Q?ulfzW-cU$ANUQ2!bIdvfEi0$dHZ
zGdC4z3Il_NnPq=DYuM#K2GpAcL0kqY5)>OSWPkjKm?hBM=xD;7jbVluNI|ij_>L|>
z0qCE~Tp*rFETGs{1oG?|e=Gw0&3C1URPjeQU*6TD-`uhrCK?8dp@~#Ycvn5NGH4^R
zBmR9g4=<^85GxH0=g2vC`bV`CkLqY{-n4fZo)tlJI9M<|3tE4Jk>=-Z<FpT}@g_d>
z35+-?w6L%ML3a!LBiO0i$F(Rq2_W$x8n&QFXy(2feB+$!VX34O<%1KzJaNhGM|RTR
zl9`<3^9Pb;aRp(B0z^D;<EQP(5hy(%o&XDq4^L(h=XL(U#|6tXwk&jV1UZPaz`SFE
zhoSbmCLglNYBL%kcSap6C<cy)kqBHCpGX70Wcln6?x6U1)Ay#9mM*^;UE>a5M8IKy
zLl~I0E;Ci+7{tpbt*lxbp0VW%Ty1su`ldv+f{ORQF-Qv8MLb@>{>CsUtUE!Hub<Nc
z^DW2$gfMB+U+~jb+z|`3+-TJo)NGZfIzP|P@2Y0ZvkOA&9paB|!z*c@+we2R?otg>
zU}91VkS%`_c84P(qM<i;(IbB4#a`6!ARG17BzSwE(|-Q`&R0cGD#U-TM%mkOqyQr!
zR78Y8@6yuKI(+s5fFv}djzqTAL<@#szEmq+!+93v&C4FuNoDnk3Sw8t4<Gz>JRxP|
zE86zjr&|i4uO<etJr#*NhDZjwXmqh5RizHtOBgp`!$eWd6R+;kdQCWhwgnRa$KG$W
zT1dWS+ao70f68wn5<ZIfhYCC}lRIVAaSB<5rL$=V*H^@DO-!YlEYRHrDJvBf+D@LG
zpNvw;!cLGOb{b`LXYk7c0%Qr#s;Wx&SsuTo)r8RDFP}eOTdN!SMspx!sO$-EN;?uG
zYilnd?kqhb^goEE(pHHm%y`f)54`vTA2JpPDuPNDgUq68+n3!6fh?_Gh+!Xob4%(N
zS8$MBU%DlJ=4wWIrq<=0JtwOvT!+G<%=&`gRBToF^R7k;Q&aq9s??hIk1%_Jzy!E~
zX%2t2EuAEI@vvq<0g3dN+o3Z@Z5>P~64>}~?0`%q5CRko1(=+6dD^wsoVV&97|8N|
zjH=!unZ@+?Ls4gzuW>8e-P00eqJae+L^L?d7j{cXxhq9d(UBw7wx=-YJ;O%>pthtW
z@ugi>oj0OeXam%nx&i^&;N6E!@dz?oA(I&H?#5e>#L2R<GBD=2*>Ufv_ZZMpg!ky#
z@A>Qz^`87cT7Y4)JpDcI*=Rzx5jGe8Jpb@=5wXjG{xdy9%o8U#7WYbGr`M+;Q(qV=
zM}~(%(4Iks7l?fO(-V8Pwq@6YYk&{FkW83?3R|SoBQjjrI7^CMk%}nV#t68xQ}Dd!
z_S;OkTu*wtj%EQ?223r`KZ$G#BGJN}#z9Zt(h<=VJ@M%gkK!Li_Brp%vEkCAA<-#m
zL=GnEZ#VE~j2{?6q3f5R<c`K)4iBM=+uJOY13m6E7HN_=KkrdBO)gzKx79_8mf*lD
zm!?M48<|>~?lkf+S<bumF1s*6iuTrO$OcK@oVS>12F$H#9k8il`dyys`i?uJzn`0@
zVZv&ptBYQ8tk!pdQ<=f3WR!M&S%i@?BLb|zoyVS6S!f;YlJ3$UG*fadx@k4^sOh$h
z#|QEU)@~Qd%8?%C`}vI_UJY||0VW!};y5aB{Xo|R(wk_3CgeYO@PckL6$Y0;fW>fC
zWf^EH+`z>H9a7UYHAXVz0DS(89u2V20abm@YOc-IADdo0P7dNHhaVd~%*IHSCg^|x
zZ;IL$C0C@Y;w15=gzUm+LMS0Uz2Q0Oo6lNFWO3kuTR@{=wE!T3)kISK2hbZ5CEmn$
z)xP>(+YsIn46uwb2K;_u+aQQp-k6I(>--D|2?htlrH=-WPAeZ*EN6PC4ZCJQ)`3U`
zxm!Nxp=X6o7z{FOpaox44qPH2mQDKlu|6jW912E8o`7ih+A$f+R~4^%i>8l|;XJiB
zvpm5gYFPdK>NQQ9KK^GuTR9~xi6cFde}+E)#BvsIcEP8$)lMfK{4qW=O+L`~x1eVj
zISk-a3!e?BVc?|TDbiV;xGA(B(4Uw)6Uf8RFwx{O$GGUnof4k1OYqb~zxe`j>2sNR
z@_29Xmp-BA@v^|Mp_Q6efC~vu2}}}fheSn}ux^T_J*cX(bQ6wlT3T5_Jm^1aYv)`k
zX^2V|LPq7TgckXbe_vj1@9J^~p10j^U_L-t_z`x~-X*CQx_59vC^G<7!UBfXXo~(e
zNY3_lG6MLJfz_pn;RiGG_~tQP;v#h^T3n~^Md#Zfl`#0clgQc|9!^66*l&r)a~)KT
zbaD`YJ>IPULn-V1@^`EWP4xxh?$RE^w(r{f{%1oU%|lCP@j~P4NhBM?r)Baf%6l9-
z2M7{-=8vG;*~`mbEDf8VAahZ6Ht&h!tjkm^c;60x82b~Iaw_WK!$0Q=9z`Rz$#=m_
z{U{+MGV&#6y~j~>IFDzaqYYS2Zg<SRd}W3gkjvKp`F#-!)ntNVR73=nA!01FSq7;f
zbYAqr_-EO369iyk&~X8Lnz{oEYRW<V@sa8|&Ze)8n?^^2@M!=j#^3<TPBNwav{)K$
zV2H<CxU9>a5?$We`AL1UgMtgBif83!^+H{2T-*veXiG}w&>D}AxUi6v0O3pN#m&dU
zh_C^=D|-0w<-xC=Ns9DIt)Y*KijDxt!RQ2d!GJykLEQ=IC}6mB39*rr7`L$50MOaP
z%Zr`!QKh-vI3#1@j@}oZ_vBE-*IZ)%IvP1#rhB$~aeK_s*_TtF&&w<0y1|EslT6uS
zMPT#m_;`reCvQW<?q9!7N-U<pfrmYy(DPZkVWMFpleVJc=`n>LKgT4c$0j@%@PQIY
z7O};mb2at77>?RBP59g&0LpP~Kaau?=+D6V`}q0-i`WzIO|Hy<F^$;D*H2dh$bE4U
z5-FJO=;;UUisK8h5izipxz)-4>Fp|V>>*G%WnlraE4nI8eNnXF+V+gL$<Yd7%`k+F
zR{0!*o@Z(*^Xff5aH(+rnG|=ijpjn<2*$FNl42m6EJ;jj_YZK+mk_jpLk}50jW)Q@
ziG}cc_Dmvx(<E6QCk`>p%EcLbE2w1RkGc7er}e+K#W1i*(uOz2?Hzg;I#Jnn{9)V~
zN=}XOS7exov1!5-ce*Es^aoN!iUXwh1XhTe9bn3pzr&t?%XvL};8*IcgU82%{~Y)7
zO1^ult@QNlYUzy|4Vm0jv=Wa_6UVJaeEBg635vnLYX)mE8VlF`u=^W*vGck)o@*dm
zcvKt~Ew&-KcdEg4Bl~Y0Jc0ILh(@<lIU*u};bIhDT;D_>84e400U;PIykC&e?Pvw*
zT4Y3ogQFv|vb=qL+a1L45j%+C`M}L65iB3=JBL5|R6}%R<Xo@~iyD788L~(I+90uH
z2BNApQe`_$Bx1+ZtgT1UC;8|2wDJTPFbq%IsHy!?g$$4wR60h<U*NA{UjLn&Vl8%o
zg(95wSe((Dk4rkIo1H@TEKZVOx&9m6V2n+jlXJAYn?Nzu^?hxPlz{Hv%`sxR&`A$(
zR;ygDvTPF+ge)rQ-M#nz<@C%pb|!KXLYb%6%)7d6ot@Z=SeU2M$cZw=b!FvlT>n~2
z`>cS>>_2#LYb|&XpZ*`$)<yRHT>AnMsX2DgY_-_t5oX$EKBA)}iWb@6z02Dmrvr+0
z=9$97jIMrkU!Z!`e%#-LhrN>Z@`P0V>fnhDO5#m}dEovt3F@@pHct?W&Jygp-0b!F
zjcDb`I;3+$OO)+Imed-E-I5s6jIy-0#hds>cJ0C&T8BbG&AFlX?|*@v%Qh+i7OOG6
zEsDMg3q#ELlLY$1q>AU1UVC5$Lmn{1Q)Oj)cAoPN<@*HUIdPx<{{VJ(`{hQ}%c@*^
zM@KmM%p(ixC!Zma8$lh~)|@wPIGqzp7qxHbWh4Z;x|nUQu@jW6INm+8wVrI=swyQ6
zDU*OIxw!J&*RQCX&^WKx<ZCOEdY-CHJ5|apy)e73AdQwz?mDl*3BeNJ<gu|>?IAr>
zKOrVAPGW=4q|kX0&)_y18bGt)bwpi=U#-1F`u(^On=)0|Mf??>AeoylKfG=}y5lY-
zfi9YifWr-S93U!j{K;_&BV-d<rQk<YXyyTd0K>y=o^yKN^|w4!q^)0oJMbSoh{O8>
zge{cxoKR(KvB%%<%9Eg%1O_9^`8UIJRwt?_R-9=pd^7DA?@|7uTt)bQ#K>n`egWVZ
z#EBa|m-vy$Mw@1KOrrw+ZMP?UQ_1(Qz&KZ3&0b#7r8Z5|QcD?vrIReg@wyh~6l9@}
zxc(p>kj;V?q&IJhh}d+GN09G0A<o;v6tFv#>x2N4Htb8*)|^+;*n5=_<_tg(kb8Ab
z&LoJasS9D;&rUuBhJ-{Q60}D3{WHsP9WvH1oO|_P)=76^aAZUARA5$}X^zk0y=e6u
zh^v66E8KmwZlW5dvf}iZwHOZr3oSwE)G01y2Hcu(Y!l$Uc}(WM8wU);Z}6$ZzvWVS
zH9zs}+wgsaa;5C`*vn6@C10VS6OS;|zx9IKWFNEsx188o`K1o=#nmCzgt&Oq*S8BS
zE*uHs-m~~uKz9g{cEnsCvMwT(&r#4)xBU7DfnnD!4JjEeD@;It>#QJcm`1Y!t(*(S
zBm7~YZ0;DnpRsTa>eK$agU!uRFYTN;w6tIFm_0C`n3?efR)*{h=2+nIh^j(}ngJQ-
ze=xTe8xa$FlOr)g7gtWEbUG4??r}M~cpd>br~YYZa_JLF9WJ{tQ&KLEJ%6JIVm$UJ
zFP>AO{rfHOm_M`?2p;X~A}+Gt^(mZNIszARd=3rJS4>okOtNnNVe_S>ZO~WTUjLiM
z)Rdbs^cK6I<vj!vQPHriS7H)Di`voOGOmMyNR+Mk-f&IvdpgU*Q0WN{Kf>bb>LDQ^
z#LfF-rf-eM^_gFxk1t>(j5AU4^pxT21BEOSfvG$w<=OO?+E08~Q<JZxc^Fs+#3bI<
z={a#Qzr18F=+C7zB^#-tD3Z1)NN@%2YAQNB8)X3G@ABsu{AePbCfDvBEipg)W3NGq
zCuVluBJZBQKGnjJe5>?Eo6jQ^TWi&=MXB%T?ei_*m`2<{X5w^cGf0WplRZi&$}=3G
zFt_ZA>-MoCU|7V@P{7>z%hM}!YZUZfSHI0m|BR0t=+82*Iw^FfY#2x$UO1TGEj4i3
zXZrHTzEq*BG32V!A`7S!n|T_o!_Qvq$}?~fI&^;(d3=b>iJA`m`r^&~<_@Qaf3J6$
zm{v?iQ9Qet@?*t#uB_|ISkT1?ngWM+TL_T#L4}+_{{YSljz?Ru+kqM;+m=d8w+RMt
z)Ne8#I+O0Xl^;L+@Or=x6Aas)rpMLKAH<0T9fN~|F3~wxG``LCHP~W_=zw&{hfDG6
zW9J>yp#f!V+9%$QF4U*QH@g(g83-`V#;Ps5YAAy4R%MpCRh&ajVNVraQld=}K0D9m
zd9Sz4#b(Y(JoO$&T3p-Yb}K7wN=lhqZhE3fGx-DlFI1^}c$gR*D3e)L#YQw+mw(;s
z4E=loE6@mV2!O_bzyZ&|b(`L~1^P#b>_FauEk_2-KP}{$j#NMcS^kbW6zs0l1S;|{
z@J(Rs@R32%0GerLW)~x4_Q-Y1_^7BT5FYrAFacrhfjS-<d1N-y0g=PWL=mF6IYAHa
zO?vbnh*AlUk+3l^+SEK+pu>MqNC-(3Xz4}PG=2f`prxljE<b~fPn&c0FgzDrhdl@c
zOpM}QA9dubqt4J%EIZlxuNM9?)^`>zxzS!s&yj6ry1HADuBzTB3QPS9$Sj^d9eVjv
zZI$et6fSfw;B=_hCe{|>h(flnI=i#P$iFNGB_T)XP9PyqwEV*-kgpf@Zrg50D(C<Y
zle*K~(B4a9hOc8_miMEfLo;npJiD2Dy&K}g<lcFmVCE9O^ykvvXz|xS4e+q9sM!*g
zdm(p~>Sl(|KH*Utae2k&8Dop1%z!ox39A0D&B%&{@#}V2<VKwo(l}Nn!M07<Ep8h#
z^Dp80vyvv0Tm%-sCAD<cotH?Igi=tT)xLIN;}zElzP;P&zX{)<qYE!O{=I%J-7fzo
z`5uy59_la@f1s~!XUtKo8kUmr&|8t9h4~tnL`7GHN|vUUmSIwX#&0hyPguRHd_I{3
z%VJi6n(nxm)pi12R0DyU;6K0PSjGF9!s6M^KG;hPFGEJHa+EdvMp!xUgCs?e5|TrL
zxD62KTG${WOLXGn*R`~1TY2R^8G8Oa?44`>9a7Mn`r6t?U)0elO-|Yf%>aB3(4QY;
zqh;eRTKUWtCC^fZd0sQusN(lek&n^6-8Mfrmp8Ut7kqT;>CeTcx~F5DYo=&nf1_Hk
z2mK&79G%ABHswF<cJLS{883Ooxs{$CJw3$*1^AOAX>#eqOF7O3umfvre_DqyE#k;d
z-!J5}b-l=DTtt;yIG)wYK@7bKy9iL#vsBy*OS^seFb>VeGj|)^>q!#wSXi!!q?ski
zt516Hzql2CQV1Cc_+jysYoj0~-~&}{F0#XY{h22!hiyTOVZ}==uTS?(@&PoK$^T}1
zW)>z&9sHC1T@7_X!A5gOhVIzYgmWHZ6Zz<GyL^xD76cv9T!hPC&$V*}J(lyLpyP^)
zXV>-inIe{5QiN}3WV{X`zOixH*<rB?TikO1sR#(+)!K5T;t{dbWJ(-JAwsR8xC!Hi
z+!YF*yIt;Q&qA`x`%^CVXf|j|*be}30slimG7f%)yAWf&@9P6hhoDdfMn<u?#%4=W
z0<^k7atT0VnV8HGWDZmbh{R{~=V)u=8{Rc{HfL`_Vc~&LB^ZH$b4}tZO4f!H0E0{S
zNF&USJtCfIHZb3%k4fDNKrZ2-k)jKESb5ZQ&>V+4XV+Ten4`o(+f3nQ%+=t0U0pxH
zm<gB#D-;a%K=`rL5X?~~fr%4-QT%`)L_Dmdw=lW#?C|N&9rhj6ce_v^E+x0lYHw*B
zK+XAezHsM8WB5vnHL0+0A1iONw&BN`{iB>oFG48^Qn!BPDgM&(Id<WK6yaK$+Ux}5
zuWtomF0Cz_lw{GE^_E^j<-5{H$h_9NQB^hTc%Di_J6Q2#y-?e0_2A(P7oLx2$|M2u
z`YWb+_VfMW>?h0*Plj#;a8so!ApF*5W2NNDoUM}toRArQq}Q*P)0<Z@>D?hMQ`~0!
z=wwOI{Qhh06$>{tZQ7Y1mq%2%>beoR19$=u)~Cd6aB*Jg_hY<OMMbV<S9^I@7LiO0
zq$q2npJm4mjLl;>Tj3z=bwhlh%2EsrNSR0m0(H0>#fnf+Ja`bdLi_jS6ifRIg_g$T
zQ@6Lx4q`oP*mLgZ=33PU)fN>U#MTTW;Hu*m(oo3o!Hgl+YXo*~Ut~POdW@uQ_^CBI
zPSLygw#_V)gsA5rLCdMO53~fLOTxIZFzcVA-@5e{oGVCt=^q_M92!Kvw7_9MkT}J;
zbm(>o*`MJI_dlAbIj$_U#<23esla$~#FbyF#YBgKvJTqM1Ia`#{tJ&7HW5P0!UoP;
zG+6+ge$m`qNC&Q*oPdWSCklO^9gBtsCdQRfDD33ED>}L<U6NqqwRyN^nGWjg{MqyA
zp}3s@)eiufqXwjN1fD<KE1fIvajw95gU!$5@NH<bu*~A>z|D&@GpQ96iabG6N2lK;
z3t&PU7JCE|A|SM)Vs&Y0!n4w{FMK<cANTG>K>5GE35C7MJp37~uyXb>`50MkdsIU@
z)1_FR$UVOFY9dKrhwY3A+ldp0FEYQr6W*l|{`Tn~<4qL?5m)BTu0IVQJ8gdd@jRX~
zaFs*0);B_hL|<`RNr{Ny-EN|t;@p`tLvP=nFf+TqCJ@HfGFNa%*@qW@RtSS!M7eOZ
z;<GTZ(m#i^qb&qz^=D@CWZyJ4+QN2#qggVgJ0I^YTxH>PVc=yo7dT368pCj%!>PrY
zkoU1+2YCrF_}+^MaX!klb0>)s46BQ~{c^sR7Zq(iWQzQoeE%MBP6!y14gB%=q)^xE
z*Mm2zBqHU3tsV<fU=fneNVL^9)@jH23j7Y6<Wn>a%w$}=`p(?g*b8tT7V6Tk0?0Il
zl?Mf7FfMF9)Sx!RJMB=<OnkjqIz(7^N4ULhQM22gR=b~B_1M<UIem|1{@n_+jy=9O
zV04IA!lTf6DQtPWgsmTZl)E+s3Todz_Ut7b<3Kh5X0;v?h(iE=IimE6i3D(WlQi|W
zDYV(YIlHz)UoJFRxcu5ppz_a~O4kVl?Y*1dw!V8tvP+1Na)kNjTnM+#ds*|Rbh1aX
zN)Rz^Q({Bw^{#I7dUs1rpNg+^;BT3lKXqDeJXC1rp2ax^nxOEma;Q>>?Bx9U@dA^1
z{4<{~1E24O^K#Kd@(K&<yUpoJR$gKWkFOxs9TCO>Cx6T#g?kM|H_Ge9hhO1*HEvJR
zxz6>3F454}($qBafYnS#T8{q#=2$?|{h0XzOqI_XkAyeMRh2(w7kySS;BX=JcI$Wp
z8blU}qINL0@VsJdNY=&`GYt-$U<2FTRF1WilM}3kh#Mi6+TuVpF6hSqMX&G!2Z9Hj
zK@MAPgf?Mqa!ma^TAco2Z_r|~ovPU==Mf6yyyqHRX(QKgW+MiMj1Z}U!i(>aR~QJy
z$H!wv7PvEjM+_n<7Y#U_#%lcgnwuSQMu0iYJwE&ix^E;m6NuozzaDwU7|=nOXz_65
zvFg=B-~Vm2g1Hm6`;`^%&4xlf)1412$O7)YUk@~|Wh*Uvb$mDH#4J_w<@VGo;%DMm
zob5p9fv8*<tpy@*fQhEnL&8G;D>|&k{r`8wTOj1og)AsRK+z$G9Crq;5CmC&{CMB=
zN-yzM2ul~ipUrt^YJY`eR>kA<Q;f+jzG#A;o|q7I8<8juPn;&QzTQfp<%q0ryyu~n
z5&QlQNKF~kbF+0WTOQ-kiPEpb#)_#OnQUuVnuBo!Yv(c+gdm#h+UJ#6KV)#6y7;96
ze-Mb<aDagtprxh777h5luZblJ5S_Z$ccH-5<avsv@YL(YF1NaIFzm|+c`PD-27N~N
zt|apVQQX*T1W(X^`3|PW!8qUtp$hORLn#D7Q@&aa==Fc9pQz6lhkAMv@1j~9h&QrZ
zc(^KNI&b;nk-#8Osoq>_s8*Kx=kdj-3JV-PJtU9nT%692=R_G%(;v+JlvBz`LDA6L
zdxHY!|E}I_fDZpx5M8PBybq$?pgp;dFeQYsy?YloKaY0+!!ZJvG0B@KepKHk)y4<|
zbGdW;K$%41?_5LQD<(+@1bZ8xRt!AO#U*d<bE6HXRakf^cSA$}tr$UeQP$ml1an+Q
z_;_(M`N?H~%G<V0S5vcF=eM_)*IRH7019vlz@(3eGGcF_+sM&ES#~1AXcPA_R=7XN
za)a|(!IQ#~Ax{|yH+IY3o6&(16hBpKkfqJ~j$=v2Ktw=IrI{P@1^#!Rf|%4lp8`);
zED>t@k_r4KhfQD)l^sM)|KRw-$@oL*F9`$_)5e#v6!DwWyPVFF>+A1emmrz@)Hmie
z%X33}_KT#?6UDY{9G60e4yn-v`ua+HzWraQyGFwQTLhi>OGz^61rq*2#(}hiZXI?d
zAU%+d;EBX61D_3?&V})PxX7RFb%@ol-|C<<L~pdJheZ-mAP<l1EP7ZwujbY2>9<BJ
zv&_=5=gKYijytYS=YKXCj>f#?BdTK|AUSZrrNtJxRuG}&{QKM{d-qms6Q|kgT2p(;
z)n#{oSU%exja??!wDS5SC1(iep8SC6jxpw5U7fXCNQQ}#Bvw5_!>d;suIaz<C3MN%
zetPz7k01qMoTxcIknAyiC2cR`K8nFX8+Uqnd5UdM$|Iup4(c4m*#jRY805BWjR{{!
zAYe0=*Fb>k{d@NaA((GYp1cj_-Q^}0s*1s!d=nR-M>aOx1o(TA7>~Oe=KK=#O$>-w
z{*P7I7lnq$m%hLw_g7y+Irk=wwKOG<ZR*aqzi6(>tB3x1|C3%V)YiG*kRQB*V)^p7
zwYLmhiYt$Ek10pDbw${GJG@AQl1I14D5(k@y!SxxXwGgXLT05!@bO9jo46in-!k7=
zo5KTNaELWkJqLN*pkkfXYdE$QnSWha&=}CkkX5O(RvoaXw~&Y>A><c{+leHcIUvq2
zyZn*0%0bGVGgWZD=n5z*9HZuf>J}^c1O)*B&EO7JpAHz^;RP7F5vXc-&eY%o;zRd&
zxgh-jMK5u29qsK$f`4N-4+by?<5dYR6-2SDt*me-E9k;`3sIF$ULP?_TK=h`oWb_$
z1L?l=K;oySB-}^kI*+K$sbO!&koGAUf^lA%T3hKJ5tommsT%3I$85i!`?2YD$^O1x
zWK(qK>1$dIL_(=Cjy8?9w~yxtDzG0vjm{S^uVE=?mSsB_BX^Sew5k$ajnKf}@*;GY
z<(GW0vFX|GpD7eQh9*7jT$hhG*sSYfMcIt&a`>eXQ;|>RPD*KnZEUZAfB<FyjPW3h
zrpsv46cfxkp~k-7h4cd?P4r7~%6Q%@Dk8FG5dFARBhIZm7K6Vnt<)eR+^&;LmM2bt
z=k7QAjwtJM=Y;NdarxhT<`{7*%K+O$k(<Q4o}pgEQQk`5g^Z7i`TxaILf;$c=g58<
z>wo@o>pmM>60YC}zxE|Gg`NuDQoB(-$@R_vG6Ea*271Mi1bX0^kLu~Tf~bQ3pkX>B
zZ{P02nbRpvuU}uDLP3=A*3)y8TVf@AzR~u}^GDCourQKmna$D>0M7w_?PzO5ZHlP?
z6GlfK9v)&d#S=GV9G9P$Uc78`<8h6(MTADW;%`HRi)-CYj-902=-a%x-O+d;16HoU
z1YG%?`v!raSz1?Cwqw1E4|6-1wH7xsw}(+}wh~+R6A1eaND2OiL~NxMub`pdIx(!^
z9N^#ogcZ!!U$TN4EYk$7$GjD4P37eh_b9Tj82Yxs-k8w?+bB$^CX@%6u7JSK`F0n%
z2NlN7_jzVVPTwGocXW}Y`0KN>KK7he?O{4~3N4~pnFLa@SFfM-kZ6pHl9e`3{6`Cb
zIl^2ZdGSN+VyCp_g{Aeyl=_C%PaYp4s&-^tBPjlI=w6RzFA`<=A0CMlwY&Frxqj(&
zpu)R}@P=gVP8J0_4-bRl2bcV&s}p3S>h-Cf_q!KT(<-n0S`7C&xjQ#Yd4`3a9>Ur<
z6ASkLBf^D%NghVa%9BB#-;_Wo4Z(+YZZFPFJZo772t|@X;@ICw`s1fh$MX7!iVkaZ
ziMFF`xf*|SK!fb;x3DkZh`|C&Sp4>lS5i2R2xwpo!xu=5!Mu3!Co(v&Tfcwzj)IaB
z=dR&;91yR^$Cs$WY|4gVu>x8T=|h^dJ11f&smQS~<0PNGrR`HQatAsLn7@zZ_QH|h
zd1Nn@9RxdIeQHe({K6EWs-gm|3zC)$1n|DA7Lh?1U2#g@l9S`VLJHRCM@~7vKTRA+
zkhAyy(DvT(Scm=l@FfySL=l;hhP{(5J48uFW>y-C$R43mlo28$DJo<|$tEi+WXs4(
z$W}@`$EEN6{f+1OJ<ngy^~e3X?>k)A_4%Bi^L)RL_c5fLIi#4B_rFLd4X2Qi;fcJ(
z>pBE`KwHD&#rv+a38T|S-T5a>HP*9ygGOp^H(RHk8qV@%eHr9oDH4mb?2z^srIa>Y
zgp`S9OX)edax^!;21zd`M{mjx+=Beuw&(BJ3KIzH>)B4v6_u8zx=yCn#^|X{j(0we
z`;L38pm{g?3*uSv#F&v3>LJDnQ5)l9W63A)a#D|CfC$bq2G2Nd^;Py$-W+l&1ioMU
zfPI@EP&7#3FAbJ{!AL+tVg(+HPKAh4!%J=X(!WV0H}Bg})=Li$5X#4#fVaUR3sJYw
zPV1kPLv;*Z$-Z>GEKTGsNe+4;&jkz?okW?J?|JbaW5d4hdlOzIXwK-6lp|ONgDgmm
zA<qH6`$}q+OcCPyzr!@Xy3(W`^BBgFf(~_}-qzpv29r2W&cJ7R^!M3<8e+gTdZ=lU
z!@)_C=;ie6X>vgU{d?1Rfo#$TsNrf(8&N`;Ej!)kI9z`aJ=eyS;9!6G4W{+wEY6)?
zt2TeNhtzgLqTkenK{P@+QPbS-9YDV4=Y1M6WC@&bK}KH<Aq<oeO^ep{VjjW;_6&>c
zavLlnBDd_f&Q9-O_q`+#z;i_N?AdotTVI<7&>tN7VM@EbqIK8P6w47+!-7j%TH7fp
z>xtX}EMOFy`<&`PS_0snmzxWLjUit_PR{(wN}{Z%#KI0F5CVH08sd2Ay9)z0U`MgB
zJ!mGsd;!lI2Oy3cSPD~6=-yl1C*Ey%f<T;K@iFj<ibmFsBp*k-3T)3X-cbGa1R9V3
zvzNdrF@wwcvuGku;EC_+1C7bP<TPP!WyJ?RWe|>`A<+FzhwK3GwPPAOLPBAlgK2{p
z4hIBZ0w2QQ=J*d0;t@qz)0&d)Df!~=cyvrvM>^D*UcV1x3fXNWp`+KrTxn4wFA^J~
zZ(Xf@TFBwa+uJ~=ZS~#7R;SE|8w9DDSAUJnH3-sNMvIwgGCYyw!nW{%es0!x>f>kd
z^I(_5xA&#e2CO%#s;;pi;~lQT@JR>g1pY_*Z*m;CaE+i1Rt60P1(P+~pMOGq%tK*C
z^~}}h9|4&}!OCiXk1L{#)?mT|VLPgKl;4)GGE(epY)n0#FsV97LoXwYn0L+RrAuum
zUa?<>O9cvZY*2*kygW|~f_D^O6j$EAtIb_xwXbRo-RZ@y?@Pl83BzmdpA1tgUhJm}
z4m&N@dCQ_v5$Roqe6Fw=;ePet$lL`nW*<)rvEjv4?09hz0(+0IPRI{8NiI{v5*%qZ
zjW$g$;Hsj>@nN?k4GU<nLVHJqi&s~oc%SIYeZKLra*IyF%^<4bgc}OlG8cNq+<uz=
zYEM#a()j|}9mw|RH`JfEAt(+NG!Vj@93U1$)f?`L!Xw^4!o~QSlM`G{!C)C#TqVjr
zfd_-j{Or^P)Nv?gK{0W4HJD+xxzsv+sIl^yW$XUBW+JnMijl2XhZgf_uLa8JtSSqA
zSl3jM5O&y37hb1aU;oKj12jijrQ+2YTVZ=EKUOzDg~D{t>Ap`<v+~g?s%?AsUJgKu
z#e@z%@6gXdPD#dN(*E^<k1({Zi~qQY0SMY{AP?AFpmdj%jOj<890jXiX69p+Co&9k
zl=v9V>gxkf-mRpj`|>lIEx3BWJT_N#D6V>qQC|M2FPnn{%35r)HwhB^=;%TsUlOn1
zjvARmQQ3AapI#|w9w=Veo?&RIL*069CvgXEBwh(rY{h!FLFBjuFCm`XTCEVx`CSP+
zc@4_%?n$}&YFGqPd=R04U19?OwuVR;SVN4A1n9@9`S>u07J$f%<hQYj34ClE8gRsb
z8AuBlDo|<QvO0|1))TwuPQlLpYcZ*|mjYL&!s6D~AEt+RMeVK6?7^g;#(^OGj+{+&
zzFd56<nH`<`Qw-51B2{4rX6h0lgi5aE&JD!L>7{pZS#;z1lWkm{E~z;k}DNl-f*@)
zG=8Ko#OxL#aFtp{mF69}&o4Z82z&W9<hxEk;8yjpGeiV`9EGAem`U41_d3CF?UWAk
z0XmkbRU1t9o`@E_O}cFdbKxbXl>y<AFqCn4;lPK30Tj!a`=ge7kgtbrC&1lW_hs_F
zxK4w}9^v`T`jcrF+#d0A*a!J#7%4Q98rlmQ)y)Z*&T>Rj%Pu^R(Q<5hr+6qZMQ!-w
zo)=l$Zi4osuI@L}2XG>TU-Z$q?NQequg!wSZJ)H3cQMk@TTU4~8fv}_gTfwjW&*q)
z9qxdApQU>Umi8DnpmfTiA@1vYMWJwiRT0TLU5;ZmWS)$?aw@Cp&n%7bbgZ34#g@Fi
zVmp8Rkak<{xP2@(Dyri732aK!gES@uAqh6OlJzTV*{h^y{gf95Lk}F7o64G7>6@E*
z8x|(yS`&&o$~-e75HLN?BRH;t-UUSK9D3f>%fvBuV~YCK?3d&xU9Val#0}kXuf%hB
zGHsdKFn@dL?~ms?L&P1+s9R*yN@%2#XiVN#9!z@lg@?->-v57>Zx`Q)Naev#-e(mp
z(&nk{=y3T1r@!Wr-giG#{5MqT{_t=!ocAZmVLETHFO0F+ve@eO`ztf=3;vk>2z~tg
zZRQ`RzT(opV!1)jKlPW*2esvzF2}vOHp5-xYIX2)%@wP;8e4kT03+7GZbtrTC63Sm
z_OwKuH=IHoBxSSa3yrOfu~mD=Z;ZzeO1Zaz1xWzrgIj#rl*^jx%NN}<TTdpIe}0&m
zev{T%!gXfLjT`Bj@81i2ms^s$?FgIx0%3%Tz(oDiC-{h~|9Ao3e6aG@TH|Y=T7QRm
zgWi#-E&fA9Zj8DNe&{N4g9+V5Hyj-tj^qKpMph(@x<8IB{7VtQvjp@KkF#?~YRW}1
z=7<2(O|NsgBs^o0P~(csyzgPz881uqSh7mchvdr_^S%SL<z;JBYv0rwY*zl5uWzwy
z);c|ICm#?<)?MhVctp(E_ZMmL&&20jj<Loc+j{L<SXj2Mi!q0FeL3dwTNI#9!)fah
zdPys*3q}NmCaZ>pFG54g9et_Ty$Gi!2MgTX$}0lq%GMc`)%xDpQY71ye_=KA-s7{{
zFnMDASM9wE0t0JR1U61<QVi5VdPCS%V>Dnpyl`Ryrbhf(SSavVVoW4>yAL(T0A+*L
z`e*@q0m%Ny{LuA5otzt<I$5?hCWtC;+PNNZ%ejHJHli};>jz4Ki>$4iPvNC}<EBf1
zt?>d{2;9Zn)^>d9?84gO)wQGjMcbY~2QN~A5F8v9p3|MC)#9gP*KzE%81=@8uY!xX
z?&q8O7kO0Gh8fEql|E7>mH)$pXvMX&_Q{#Z-Q?haEn`|?p;1aJJT7v68n<^zWa4mo
z|NhuQQtP?5z0hmm5OJ3oxR_N^GxGQEs&caK+K(G$TuIQtpK-N^bE)s&DRYFsVp0s*
zSmU?rA5cHp#~~EDn-~fRaRh30T$zxN9Z*)k<+r$T<Ol~Z)iLR^f@{`f0_Xf1Vus(A
z1+lMxQWZZ=!M|@~(N_r^_&r8PxWWyWB|W4naHZdTLSP9tRkg}-ybrAIK>kADIj~R2
zyHKeSEUbQQHqxVDR4;(O1(v@J-8CSFp&=nqX4%@<=$$-Yi6M2ww}h67ER(mW0d#ji
z*FS<42@yU(L=;@W3e|^EW<eE~-r)d&7`&?hO#=TYJZENLAR!}j4ip)HD)jWCf0vIQ
ztAq>1#jgH>wx@df)-!iK9<V8%*~5LH$vMbYK$3CPDP40M>{`Y<%8TbX_aV}gC?cP{
z4)>So;e!S%`9=!u*AZ)v^SXPVwre^dTST=u<o80q;mV4-ttY7~+g1ts{K$^0U>4hX
z93zF{8o1-4W+p#-Ov>3hx~lH83dvmZ$TFak|Gn$m@>~i#|D}9drG^l2+sk&U-7n8H
zNu_$0R#nSR`nv6>%ieOXKn7!&^%`fU9{LV3gjOI6%F(zuq7NJH5hnl0JOyPt_ag8u
z?0cxvP)>Oc2QeOHfe9s-(GM`Az~F3RDTGFd7{s_f%)VYxRTa>a7!k4Xqi7bmFCK5Q
zIqaC|DhP9#j^4Z?A`ypK7#V+}e8W(;qpQngPr0CTv~)A|Zie3T1joMzzAfwOlVwPj
zoT$0GhqrE<D?vV`RKl4}&P%<Lc-%gF#!dCbL5mhs%BHPzs3Td+-{$9s(=^`5lxIbs
ze*<3$Hxfk|1sQDgox8WV-cB9v7S3yx9_cu1IiXWN=ueUCUOj>2p6xw7b4eSNQBl5w
ziH|7c*M(W+OV8PIHItFgQ6z5_$I!ZR8johcTsP7>pyq)BxMT7<$`Q{b9=>bWughE=
zD!<L{)7{>_I8+e;NhR=IEdQ{uFn~_uQ?SfWy#O{Syr(C!Zghf11=a*EN|a{xB@iyQ
zf;Cnp^INo|`^|+&XktHzpUB7|>%3;VC1$P>{@<%Y&6*>U?WGctbsIPQ6|R2&e0arI
zYvhVJCP>M;FLVz*%8utCDMpyxuV3%9c+Su6-+(r0+2fdO+wb-qAo;J)6@t(fIa&9p
ziXmGX{vYDv?TqaX!{!@%1We;LwsQtXOYYoupZ)eoZF1JE<6$YK;}wI0ogKO6-Ov4L
zjdgL~jwS3QKyYkPEP3cq-QXa5>3TCY0oWaAkAxxA)j(Wh8pg0UVS8h#j?--#b2^zM
z_qU3DU8q7oF6Uax^jB*he(m<CX*P2Lp&L&X+Hu$3%QfOfmCTbuvVW%Og<U8|UcYYI
zwaa)Ur*p-fvZWDtxR=E#C9S?LRlVJl7xS(frke1>jMw@tGa<XEL*3CBGq2C>*S}#^
zK_pD5d$xoOxYS#)7rS1iqO9Eb<NA)cRE%3K=c1%8ka%7nyPJQ&+1b*{N)9KP5^Os-
z!u^zQwiem^BFBF6q%IFYzw}~F3!D_padF|RS+N573KZ^u@^D)T3W(|SoGTVCF#{P6
z(F&VhwHM*<qUaB(aW?}4uzTdgq9Rj;QxHV4WMtWScFB|c{x0XVJHG#&NIFXqG%qeb
zSXUogFh^Iu6eP@^F?n%nW=8zmMNx(K9gQ>MKMXL&=jG?e3WJ7@n_W(dn3Cw_wTAj}
z$-^6Qh_Q>%l<KsLW!JQ{EM4Cbu0D$4VTVJ{zKLgBr}az|@(<M3C>*H{S)$;yK07D1
z2VV_z7DCAV`|E9AD>;uPG;{1LfXWoZEp%>Vg#F?65PR^H5nVepH9?2N;q<UtJTypY
z7_NQ5>@lLk0|g+QaJ8@>3Rx8xv)0tS=S$LBXtA0#yXso-T|D9T?F0d?n5eBdO+Ja`
zDbfgiounl2hlhLkmC-CxY~RwE5&B#izS=a6Cf|&H<{JrJerpVM>x0^MhkRf!(*&Q6
zS7*R129k3m3ij5aWOLSQ#Q_0TVC&f4ICiUkpnTPyq)2FH`bI%Tg?suoq-OP$6AywX
zkd9~YX$*MhEXyXO%OH3`Ow4*s-xR4ANl7QoK3*&AM{FXP4I?A;yLaP0TFbPZ;bSO+
z(h^G>ed5VP8BBq(4}+I+Bx<0{SDu&m9#h=CMmmb2q;*Q4#m6}07v%#U_u-HnudEs_
z8VsV5x0cW)@OpUm1>CGLJ*Ph~u*LS%@>G^!$E9Oh(JnB2TCnW&zFY%kE4F8l;Sr14
zpQ-wE9`KAUvkZMXsRDebG@o0gRz!+$k0*VZ+-sI<us0EpsGyI}r&i0$mMnF5X_Z{b
zkqP(d=EBO#kMG|R8wJOi%8=6Tt}ZnvxzoGF;X;Lbhb}Lh3Vj|%<|KZ)hqT~E7!qPF
z-zhJg_n9fC0cmV#SJO!puniN<9BSa(3^fmg-)lqL<mVo>vdoT6ZmhAtbKZ932wzlB
zF_X_qUeK#gjU}r*AZ(vC7v86AYC5B+NcePs?k~Sz%g37&NpH_=oiLegAN6X~3~g!a
zs*z~Fl8;zMU~b^eqmoOt0?CX32FfWAhQR!D1x<u-am(7Yj@qO6qoQYV>PuXSANS=w
zKc-1RoiNeM5qN15=C99^t|PYtkosHQ1sFM)`>Y-y82)HKX93YI$hqVMEI!}gX}-B@
zjbIJ0PnE5oNbhhu8jopdE5K)+@1Hh}^r#6$<{2a~@ht5k#4IT)%|pAl_yJm+TAtp0
zeMgDe$!M$bY%`bd*Wbe~2B<w<Fv_ee{qKlC0(9AViyh&G^Pm{P`baeKqfQ>VIz7%l
z;yC!jjwnWuocuY2g>!Oq7cgOA4o!t}9f8m~0$zumulbj)?<rl?KEx+xYVq^RyX%yR
zsTz4|TS(B^waC_5Pnh*|sK{Nvr<~*Ba&_rYylj!ac^Rn!<Tf&)ZuQZ4I&y@+I6$xm
z$__jb+BXXU5<)CuWOr+~qqRaeMo{RCPPq^_z28#wu>4fJk<dYYiw8RKes2fKQ})bi
z$A~SPFmt|H0yQAhv?zKl*HjteZQ#jhSuuVmdVH`RvrfK>;-%gNb_C-`Jl+N$_``RT
zHuogf1(_clJi+)IDPayBmn>UtLIGEI`e+K=Gt$zKzeLQ?#=5hXnZadOWN<nN);@ul
zR>VZw_-Sg{^yG+e4XWG$P{U!7%6v9Jt8}X$Wn=lJ1Lo%MJQ>%2soe?v+q?2X`Pba>
zckc9!jr$9@Dk^l1w{NT_(e5H4JWZ3->iXUh7D>t|SlQMT{He87(MbVu%Ob}sdb}_5
zls)g7x)3MvgZcdV_Z`3vnfB}vzl>u6+K9+KESNRdc_yPwYwJ-<tswP7l`=9Bq{R8h
zrIt^}t|I=8pRA_NtrWF1!SPwgt`u7&D`=**!;?DhkBeOk0^p~Hh6n@{R=Htk^pOmy
zbqESSJX@ykem9YBuB9WW+=YZ_32;If=<WSDpwhks_W||PZ!n6q^~Bj>X~{_-=F`F2
z<>?DdaN(0{kWP%m5UH+nUK<*EQc-?3)8NnQL{cOn@o!8p-!Hx>&3c7%1%R*@IKfot
z5p*vcavEme*u^G$^R>Vsfm;!D-)4e7Y996RYAt@}>U5KRV{hE!<jkf&g1*;Wu9Nv)
zVP`kBP0hS0@c{B13|Lm<BO)p<C7GH(CTYv8nOLlU6?%!MsDonRl<ASE+=J!xqX&@J
z#|6#n+G2~-G3Rld9-oV5Z@wIp*l2E|mixP#@AVSSe=D%)Fw@JtD16>a&KJX-(M2F^
z+AE5PQiD(;-V5jdPpgl{Rv-7fzd+jHp4}=l2$6YuDBt*Rs~S5o3KIPfn4YM?k&x63
zY64M>_DjAg{=<$$#_d7GY4<*4bHM?CjPQV6p69qPi63`r=j3%_b{?C&Pd|jc*fx!g
ze<Sjs^v#=xe9p2+Z80}LNp}Tx`~8O-?MdqoA3rw1PJ<+AJ-utFU5R9O1!QkiF0?jg
zoTJ<)S$T{4-qF8jt_np(^K9QqPRKNn2J3xldlO^fxpg=cMl6M)MWD9Pc1u(-e=OXV
ziS0$*zDcMnQ#CBR7<Pn+sY%rFJ6JaXA>4v#0Oeesbb^#~+ZNZ4RixXu57q5z_dJ_*
z>*@aPch@YZ0%>!ZunyX0V`F0MEUUGP@-3^#MYxHc^p*PeA0Vf3p@0XmZpa*>GA%2-
z<xJe7_=SY#ppyI-mW&P%uSS8pPf=yL&JHwPI*cUinVIK@zN5;VCdt8db6@+Fqm&Hk
ztSZQ25KLm)3B@|v*;6kYEi=Dp>pesPe6O0jx(;Lq*sKswI(XhkM}u$ktbGzE6J9-0
z2r)V8G)iHYQ+!NyQGuHdH;kYFF0a+>#|xJ)+q7&;zCa9Th5GVHRK%_!U~{;-cD_MN
z`6V0cf1YLWgCvCD(3T0_DD+>NZ7(Z+TogSlB9aRRP+B|DmL8uK83_okM|%tg4z}Ue
zRzQyt*NHujaF+51E8*C`sA(S5o=+XCYL}d|rsKkYuU*NL+q}jr4OOJj4r9$XC*A&P
z^=)RWpx~Wgf2c>l=?O)3&~U!d$5or@&Kbc$htQC>SKy=nG1n#37ETG!0D<A{+c~w2
zRt4RmWfmFI(A<fZei&?!`%WeXnPnmOTa;}spD@QLrc_4`J7IIAr=a_R?MRg7Rsu}5
z1_#aehi7vgu$}WlX;Be*>_IJTafun~NE^g%ZEMRzm>w)M=VOp}n<J8SyuA8>6Cu1$
zTs(@5fGK=xs;z^=T_axsG<-Zfn4AY@oIvFU>v#nY1oxmH6p1}g)5H~bE0In2NJZ^7
zgN$tV6+V%h%mmEP>)b-C({qWhw*CPGw_;miOwm{u{eDq2c+voJAB%~{#MQd;{@uGf
z%1_ZV5&(ZsCrWQR-D;43L|Tl<>^##+yNbNGekGCW3j1?h4$vwUFz(xDVrr@_LIx)+
z44byzh1(n61^a)Hk|Nu4@RX?Osu5WB5-Q}{k&>0(i_?SO!CAXe_J?(K?R$Dw?uaEG
zcHrm6SzTa_u?tIDeS3*QF`WAY?geDQU*0;h=cPlf{!9zDHUZEP{xN}*2!?DBeOmLM
z!4NOU5bU-~@n3&{?=7ayF73*P^@9Njs#J_bP@mB)mMNvBrXt7%i804FlU%H*BgkYo
zcVuTrP%PSCi+r7)jt2^mn0O_*5q+c2{5|tWeV8vfctvA80tgRO$$*2NgzilnC(n?R
zmQr?yjVVk`|M``E8r*oxpKefbqm)$s?_s{WyL>$G$w|w9+|xg)QUQ4iA{W3j7ZA8F
zTi@>&<$aj}H(VY<eAu`P!~i<d?#omSg#RUB;2m*(Y#p6iLH5f|Dq(BOz!s!vAHt7s
zK!Z|ZkX~|aq~@MrslAi8_W6I95jTMc=|y;f`UY++&=N%n#3ueX0w5B7*CH+9qNAfZ
z;<EoFS+_;j53bbUBmx|RJBKH^E7L^n7I*@$fNE^<gHxI)eWd-BE6w%wmM57&KZiff
z)2CJ#sX^h*({f5r&r-WUzAc2?0}FPDMRnGX1aYyP`-(jFl>U?zGITB~VO<}QANu<f
zWXPclPH|h&+^CwZz<pXvN9S1YqKKO<{OItK9Z^Id!`0~02TP`i-Fs@ZT+>O)(dHGz
zt`=gBV#GEemC{}UW9m^-PYo*4&bI`d6lPoeEf6h$hcdij3~OQBhcYh|p{mk1bTSP{
z6(HL~$_`AZ5G@Ld9FA5Hpb24n6Dcbp<0vGU@kHj3y5;s#7_ydQP6b*C@4<t&QzyY{
zUH`eUKIor+y&uD~Zr|G2>OLtJ9Z(WLNV#1#2`kB$FIye*;n0eZFt`|q9##2=7~l+n
zW1a$sP|SBjY+O*tJljXiKy^+$vK4hWo|VNOySW+1fWn}6qfXs<+kW)toRQ-XR~G|F
z(0(0tcLn!O{Fr@k&zlRUFJGPk6bi{FDBDKb?eN#%l5ec7_j$PLHq=6OHJI_E8^MTr
z>E}<(^lGcCS3za<^rQ+>3R0>F93!I~0;wIhwtiv%j)jvV8EqebALQmnU<9BJsHp(S
z;latheh&QhqKkg1lQ7aJT)lQp3VV5bGJ-(xasy8uwLY>W5GIWnjg@({m*9HcXOnf1
znGEsfa9*wY4;Nq#m0#HBG%SwA@Nc^}zpq5wAkD^1S#0P4WhnW57XNYE$9z=+bis8W
z@ywX+El;s|4R$TUtc2diJUxUP<*xwRc{aRcF#8DDQueze3*{z%Namxn2tXP~;RAaB
zsjnkO@Te)Cuh~bSWnyy2;J=48FvP3}B_m4yiR>#aR_n$)dPD_c9>>PIL%Ee{0PVt`
zl3DK-rD3!caLph<;b8gnM&<vG$o%}V)eqZokWy<;kCeE$%D5Cz34kpG2vUeZ)zlOY
zA#RCIdofV7EZ+3iy;+$Z+_){jrkK_awPQm^N7T99o|-a*FrDoL^%Ez~(`+O7m6w-;
zodNUwSNcG6Utxfq_p?n5P61|Atwm;88%ymc>%XS#leh#onk2ok%s|TMT}l)^b-<6S
zs@fT+rUQQtJl^Pil>H{Hf=lf7lE>dqpTZDHa`-UOorjNGTbZIK4`3w_145W4F=>%O
zv*67eu(%1`*e>S>P|J!?B`|TrN{gHTJc)1D$7x`w`~6>=W!cT{zxT{g$|rFc{0IZ&
zyRz|^5}@4_*{iA0JOu=bHi;giH0}jEn%RTM9t@g3=uxtcrUj!Y0xb$>-$}=L>dgm#
zb(H?%3A4g2^1oZX465H?(BpKJB`#IiPmQO}dahON>MYy!8v(bsDJz#tmXfr{ZZl_h
zDqO-uK789XiCkrKn2!8|T%qAHczrI|eVx`g1RE4_P+nV0j99=XOxMKBCD?+Ae$(@<
z%ZrNuKW2R65+%gM9@dT^X$pk79K#(axkhn*z=+1((=%QF75>kN4}l>`@I1>7v+amS
zK?&M?gzH13SEi+)Dj_B%Y1_U<nj?B6pn)v@O)&ncbSJ(i=%x#W02H`SC0rQG^xxd$
zO!n?0efhb!GGnZWAp&h(R+gi=`5-?0+TR1J_a5kM+xX*Kweh#?N$Z;T`3r}=mUd;B
z5@!L6LX)jNWLvjBfA;KAKT7`BXm2{Zx+3S(J2N_N)8<N|`Q_f%kH;8DhI5OoUG84W
z033<@f-<PL#F#@7kQ`hCY&rJAoECwGfHMNI%INGqhCT94UB>qV3V0gTI{^l}6>diR
zPdQUHgu_leR!jA6j!|OhAQA%g^zrmS*rS?qSU_L^pF)e*k(x#60j*LUgW-v0Ek`%E
zxR(oy<Z7*4$B(gkO_|KWnKA!cg5KF}7cN|#9Z>4}C<i15uI6FeeqexUa&{Z1*a@r=
zYbQ5U`ifViY93WA2#<${ugqPS@j@X8PVVR^H`7<0Q;lKmXyn68M;Yt14wj2PIy+U<
zS|pCR5klC~@}+1``a^dSY#`(HOK36zM-NZ%w+WU~e>Ff+RC00(dO}|`U-^zqQL*$b
z9fjgi<nddiocDjVws)6uCYSruc*;LV7_PMr7dbgBlD}K4fwvH-#Hpbwj@%Pyyr-D?
zPlI8X)-HAQsP1l?srOG*4vUH|V#WmY2K-<M$e&&#0X&QnJzRYW)@p|CS}S61cc0qF
zVdg%+yd0E9BBR}I%^18IYB$x0oH>9+*u3pu6PRS=FVp#uVyEO;9DqI32B7OUF;iV#
zp(!~EDj~nk?{NQqwvX(?D5Epo`Zwc){2hl{yRN%bIQocs3kPj!j`zPY5DPe&i$_JY
zR;$beWCS>Pp>;UO!}H5)9m5aQr49>5?hq70fP)x9dwU83=IzJ|c5=#vfonY%f*>D8
zMTK+Wx9(M21uI*<79wcDhA}EiTucl{c&0(a#3GDE_<gHA06XAG$j?V!GQxoWGaz#a
zTkg&MH?b*1LTLk-IjA?W-C@!M>-@`~in&`hfb9L_<iK)RoS7O{IuvBjf*>H7pSSC^
zvR|E>pVv5X0vBTdM;LO6u8}PG8X&i2*(36=vPX#-GSR2b`j?fJfrpFz7Upk8$^3^7
z-SLWvC2uL7LtDq=*lTx?mg3o|mrBVN3&VMn*N;V{_@t7^v^T)@De2K8Lp?qCxWM(M
z00O9k9k#@HccnU2o%n1Czqensj3N35)`Ox8m8^5iA$z7mw_Tqd(lu@EsvbMujBJ-V
zTgMju*#LwM9%pl+G6bb!qr8T_@!7LCzJ4I4AGy}!*%zUT@wN_9!ZDl!GunU;`eNFM
zZ;W~PP{`yT6^vch16SN8H}StlcNvp1s5tQgn-^yxe6Jt%*!Y^_DRy-&h;w-hMmSzr
zHW}t+a-yQny<DHf5+WlPzqN9$sJ5L0riiB++>;?@8?Da{`52o_{rJJN!JS-(ONl^m
z5TSecW$rN?Z)n142b~#+%a~SNt8K>uA+hlCg5{il$;V@0mJWDKWzNrE3#g{3s;tDx
zgV-*nvF2tyJ0WVH)YXYhm>=llDz@6AlM6ll4X9&e4+%K*d2%Z7aX^9{HX?j+WOCs)
ziFW*&`nQa^(sU}on`wWTX?uBK$KbD;oBDNRJ8#2L`z{pP`NqgQ8K1mNFc#ZRmB}w9
zv!kOUEbdrRevTTtb2cU>5<)Ir6bao9<SL<uK|J-TOaoYgVjpZ5``crelpyCFH_3Go
zg=cd!Ii?%*7$5E3TQ__a`f(V8zcHS{X$m)3bXSkKkU<PsY6}5g3Q$9+>~Dq8=fsJ7
z|Dr8IYPTRI5e4CWvTjV_An*nHg~g1g0gD;baDvd(UZ3mNm%y>an5?1#S03m|T}6eW
zKuk<j6cnXk?cTn1i#g-*<!RJi+#YaUfaZ;kf^4_S;i!-3!!cn6xL|1b4jB{yj$aT&
zh2Y^JrC?<`T3Y03;>9lSlhHmY2MGYkZ=%}5?YMQc|7re;4br_AMuRH`!0tz#fUk_{
z4<t*G9~Q{8+fh4ml{f~QpZw0m#%2b5;>#E9_ephRZu?>F^y<}JrdbL0fR^BQ@7{sx
ziIJkNF2sY~3B=+Gj~?vClQ#IK%F)^qZC&N%eVv>ZK)6GBpBUgFAM~$Wzvgr^o~65(
zDTJ<M{mZK}S64;4mrCBMbMKFJDtiB3$j|(Tho_yco0QEp<mIw3GkdK45_V77hcXS%
z!-y3`FGr95f@Oiu$yy%tF227f(|~=UkU-EJ7<mNDBmol=@EKX8oC4jKk+=Q77nWgr
z!S#Lu+foQihv5YK$Uh4e2o@#p;lq|r_QEhg{OlmEqK6M|Yc=_QuY1FPH|n1_(eUlt
zAp(x8L92;on9q8`wFC_x+$YnhhrlHvC~(jcbERRS^S_K19azdyBA{IG{#i6=V|)}b
zrL`lFj$vqn=QQkjp<Sz$z58}*YV;P8g&Z;}dPt^shbvx(!!qJ$cVi`Cbxtg9Dsc`l
z1>xkV<FBQ?S8cr_t=$d_2frxZz_pVHUyiuRut^srnXv3U2>TQwYd8Nltd^d|gg$+N
z`wr;tls}{+M%?9gCV<=lba{GsfJy=Urkj<eh2mcaLC;D}!(JKArF+K4YIbD?(Hc#n
z{^5Ez37@`vVMT&u-$#ypkM%SqXdbwy-{v2~UNtfCxbK7>?0Zik$u9EcBq<QKzPBwF
zXW>!e-jM;As{9=2sfp`|PuxNN10x4wrjw-P+2_FC&t$@_mGEPr<V3p|!P>?qCoS{T
zZqy6{5aox_`Ch#m7FY|K6}kT#NC*8%w>2lqBEt&c32@Mhrs-bP^_fk>S9jD9Jl);l
zz^Cz2Uq?F8Ag#SsY-Vy2!PSlT2T00WTZP5M5RtT3@Dw%Tp=Som2A>2GyU{H|&rfMQ
zeHmM@*SHy$A%N#HB*Z-r{qHzehPZis)=fk-?DGQ8Q-zx@d}$gf*{s~~4n;VqQv;k_
z8U_4|qtL(Nb9qj6@+at=)zec{O`*k=2X`B#nD<KFrM`k0Yz1Q4&^9d+2`7n*BGN}?
z$T<VUv{jT9+@;Qy?igI-2@LXoE2*qnd83R<(y=l@TH?U~$znHW!8iMhCr<6oFq2I>
zlXDtYYZ!O;J449uq+;r*1%GCS?|Z(jOp6GPRQ~ErO01V$We<}RFs37LFx*2^71Ds@
z?^X+_`>5QjF3J4eLqN}`ejs+FWp{na$#yZLs+(wuFT*Onx6nClj0;5=c9f0LgYv>g
z$6d_`r2E6`;usPy+)Ey71aid1Rn-r#m`^cg9)}?!slK}h^F_S=apMLz)_=l1)I#Li
zj`o~twjv!NR@X&IlEm=1xZjYCA?Y3bKM);2UWNt(^&6o2YdTMF5NNjtS+kU$6Y%{t
zHyqDGKet)?_l;Z~>KnY}=-b!1EJUoG!9l`FR8}_F>7rQR`i*w2%h%|Pi1?mF*S~zi
zI8$maO1Fy19zBYO86LBcSYsyjGhx=N_E*mrYXiFxE6wK-yKNbqZAbo7lNaAa*Ja&V
zQW$c&hV+_7oGf~7xg^I0USMaaPV&|T%WRwCFjdBp5BU|AAc!3pqWk&z89k`$m|DgF
z1Lo2KF%Ssjta;3P!KtxKk*eis^`}$Iz_3kD^I;($K3v~#hQTl%M2OUQPrk>tKGMUk
zjPogeR2L>M=O^lf)DG?bfhQ=Av=orUxGO-5^@W%+vA2@4vfgn2=!~!QAQB^Jm7;Zb
z=OZNpA8M-`r{DCK`>fgSqPzEKo42jC4_M*&eHam8<<tc|86;#dsE>p9USD76h{E`p
zY#VMrNk-F1Q?w(yPs?dP9iExFoSS|7_BNPfo&3&Yj`XYixo&Fob8zD~JZ(v;lspHA
zGL*4!hc7C6JQ_IRR0#d%U39Rap*9A+d6%#hkrs}T#PaKm)w@M|@?>mVttdGE{A+5&
zLh+S>%|DKBzZabVI`gU1{5ED}elcb@jhYh7rde27F%6%ez`qbc?+PXTRcL%Cd{o^o
z!qiCasEQt*Bx}<{H~;qJ8g0w5%J~&_)(rIz+#@^|MqvPcM<*>3N|cvNsI%aP*>2kn
zx03pnmV=@lU?(dm_$hFpYNXu09qNQ5c}~<ir~j7^CmS-*Iv98xZk8-E6bT8bRN!91
zF}<?0{=SgZa}DN|9#A#Fo8pd%(y*SBVG5g_nE`-pmUT96@=R=8oX8_Fm{NY}=r}+Z
zYzAY=3%utoD+_JT2@pkcXv4?}aN&-u2N(wS3Wl7ZJuC#Uua#GM#T|dC?}06Y=xhuy
z2}0-dG-Z-zpDwEY_Xjk##=dtV;v@ILgN|@vRjLPw<O4Mz!NT3IUuLx<%Lk*@3ziWr
zeBZ9W2M-=qIy|KS_+cA_dl@Z-@_s!{&Aau%Hw#pZeNgh@_jybJi^Cs;no!r+C>9%M
zf7nt$LPTT%NA2SEXi(psp_|wrj+y7SXd@~vYti6fT@$LZb#YMJyuE$<>V_4Jnb#W1
z%CE21M@yjm{8tSFz&2u6=AZ@hRpd(lQTlwHG%l&{8U&bEp_90E`?hq`^_NR4D<buM
z6bmX*ni!R7mj(g+9_7t5*_m)%7PKzm<6|iG0Xl1GfyP=4MLvAoE?*`dkT=L8#1d;q
zkp0o-rZ3~pFMl!5UN|u9FGlVB2B^DWmCfLh#aM%V>C-3L+f~9bl(w&B+!sbs%aSLI
zp@zl_vsyT1g6Cm^f~B~Vg!3`H9$GZmx{%P&)6&wSRF9aoEVO>Dt;FTHs?_l#_R+Lg
z1sl^JFmkedwr;FAe(I)>l^y!JB|KaxUA!vh6?O$9U+_;}JbkK^(hBFVeQB0iuV3?|
z6vdVmqwGV?0_hGcLdDjd(KcSn4$7dGl9R&*nhj-f0CAEDCn{!kcDvuIcHcq5L)#CV
z6WmP6?&^%KmkR>|xMRVBMn7wo<-4*+*tU1`(Q2qmt|pELCnoxeZ=+`WjvL1Az0_Kq
z_#8MCXyF>)BLQ89hv&e39|vp14lxw6l2`gI1g(Mr1rA(0nr%xB##K$|-jaQ8(F8fb
z$JVC>4;>ORY(R$W6a>fwcz4$S)*@uPOc^Y^f|{cCNm>zFyPTZXp<m(Q;oJ@-2&w{2
z*OT#IO-WjZs>SZ?227j@e)S!n0U6hAi=28ZOtu9AHvJ=X^`a=VCz5=1w*0blCyTpq
zk27$38yfL@hS2o8xGZOElnO9l>0-AL(}ub3%$FibPEP-lnAA!18DMc?nPCmT2G82M
zmWmM-XbE_Ge(k{%)CEFHb>q0g0QQ=ivI@v=S7CE_=QjV;I8zQ{Hv|;HFIhzox;0cE
z+&4^N<0{I3?$B-?K$ivv;CT|A`{1mHv2&(@y!!&&--8;)e*gNlBpk^~ICQrc*!-W_
zvMhsiM02K~)I{3^*C@1#yu9_lywDOCKY1U8GPtX|`$hZY`cj(w0Un;a%KIfH-{M-~
z(Z|e&m@dJiSSp4814ON32|s{QcIVpMLyv8o2~z-2KakLD9Z?Ah5f&_tFFv2)mN(V3
zN-LzKZf_2`dw8Isz!V&p0?|t7tW3J>OYQS}vD37PHI2;QXP3XyZ`m?@6(3|`c=+1<
zR~3{b2&q{5ruSICgL$`#QC=l=ocN{T>IUlxs+GDY)puGZXi$ZQgn+7nho^^?8;VY7
zqX7!EO<srGXqwaZzpELWm_gvv;EbI<9ci2iEviicvcv+m(R{QP3AhSp-~aQZP_`(J
z7(b<q7En(Qp%p}b2th%+S?bL(-<`3ZWmur&al(ZGkN`-)rb`>uRyBqvmk7V$X9#4V
zKJuWQ9#~3ymFjfRm@^!}?h~x+1ZoDZ31;E=UH<Nu%N~Dm{@<)thvXGoHz6CVl&We5
zLaVcx#s?zn6J^|=w%U)+8TtBW#e4z+kkAZ{_iF~0T)TT;tr%4fFiJ-z2zGM?Fo0Sh
z6stu}K*aHU3+N|C><0mZ%E%ON0vt%?GR*2iH4RH`WQsEh?!s<Yx|lGFE?oc288|G0
zoe7)@R~67B7VXW?wF>t3-%bdY=)+DP+4OC_RAgJA!^G-YU4AEyxCKD)ppT<i1{?2<
zF+U9SDacUMp|t5-6Zlf*>x(N0i`#Pom<)mhMFlytw)O${^#gPo+S*N@KR>*7+VBDc
z|1H|wZNR~+K?y*l3lh-49RXzD^5TK-e^8#N4>dq#t#YvO_8dLf`5!W@A-wR<pC5st
zC1yU5)*vbjz+ZcsoUpKE{I!JE-rnEfV<IdiIG9o{smy8YBP3K2yWh3%PxvasMogTI
zK7qv^F@e-n1}3KZ0jP12UygVjSbM{?05o1ma-g!p@kl*DFnl*)^d?altOG~vt`-l&
z?{Uw_dn~;`2p)DF#LlXwz$0Uezn>UmEVX_4s0%9bCXY`em5;tu1B@nuQbl!$)8Jz`
z!A&<JA=y<KxO}6MWnEc|jlD<H+xss@%$J-PVXV5c!mvFEFg?bI)8gsTG*o5vmz&3#
zsG)U^N`H^^ooct(y;PIjID!hfsDdgHu4ZJS;)iMvz5mDzWSBB;b5G{s1O8uC+|)@+
zFEZySb68%0H<%%Hr;rO=5au;SJM6`^92JQ{43N}{-MG8CUdN+HlI-4CPxlutqB}Fr
zTtOBQ5g3em@E8R>cCzWMIxvn_N(KbiUFz*=Y#dkbct&3z&J@l2EgR8=AndzkE%Ji|
z5OUx+j|$o?Ver}bA&+Np9j?fs?34C!n5nrhecR>9_3}8%eRcIZgbBkCZ@#tBbavXf
zMH6p}BL*n<Xp(jE6F@Q;;G!YSd)%<#f{`K^TYIx)tk2qB&w+;@CTNY(xL_!G)gG`J
z$*)+u-l#nHSDCN>f`p>M`GWpdqVWT?W^50B_Fl}&@qLTb?7_c3Zj3Gxy#;+_U;l5M
zYoWf(=`Nw3RjWcZ+bgYN+UwFp1r3-8Ha3X3X_;cec!E>xG}f(3NXrB&o-tTFK(PMe
zMP1yUF3ei$-w6M|;x;=<fn)$19U32xj*B}&OHrr0LNvASdpnEj5a{U*SJ(B)lfK&#
zzw;FkBd#|fc1_KXfVeswT(En1d&_FKCt<_}<r_cE)AQuntG|CYKQT2VQhcqB{G<5#
zD=6qaJoPf;_3P+{F|Y(d7O0uAUGT8`;?Y^;{igut4Y1Y>=H}o~00edRth~<}ac&0$
z172UYPgxrv3F<g=?%W!mqdtS*GGjxA;L!9m2+q88!JufG7#mxQ0I>459eX^}^M%rR
znO^wDT98`%68dXYkwg-<Sm6H{!8T)po+>2G=?V{bPm))_%GxIOu(37AH(&W9-igVy
zL&uR7hEP?mm<AZ{i8F_4KL0%$dD6^%Y>(}pi)be(dN8eS{aOG62DoNpRiK~67T++^
zb|J2rs00Lw9h7+?nlT)|ctK%;4@(U8M5IFnmkZcny1Ji$-S)KPem*^##4~aTCZ+I{
zhAF4*DdgY$LYB{OG!etYmY{E*zt+PkqRL9BGrpFmYhXYQ7Q@izm#1;YG_D{r=f<En
z(y&lRj%0rvU2cBAoRN0wr9r0g@|Rhkm<z8lt3iSXELt|nPURt?Z?BpX&1>P@@%Q9F
z-hvg(<Lcs$qiaD$+sImqFNMPr$+`(Tc1@GEHx4$|pZric&uFHtO&vUtHe+vb^D00G
z0l%1f>uEwMCNf?Yr59WNGNX9oF`q=Tw;9qbkP!kr50V}<^;G7P)^xUF$`qVr?~oET
z{qEcqkStX6j58%9CA*v~>AsUH7#lPH31)%1nH0z6P6DPWMk!H>?Mq-K+0FCZefk2s
z3zN(x;i>ZQ!&3}GTpe~XPgVbC_~Io2@I`otuY}RI$S&!r&OA01J*;3!djZ>6b6b`D
z4P7eVB`){%#z!$i)nyu&O2Pqp=ofhvz0bJg>PJ#8H2v0j6q$4<Xhi-Vs>v7=F$oFV
zHU0>D5cdCovE+kuSW64mh8HCq2fRSvZYw_Z^5H}7ybAw4(e&JY2YJo+F42ekf_dxV
z?jFvsRaWyREsq@=|KQrVq{f41tOkx!Tujv1OlG`%u!xe~gD`k_%E-bJ9~P#2=FEX8
zO~9&%U~h`zJ~;q@fM-zp05>=GA3L=0B_&^;_^ETpnJV?7kQxbeXsr7CM)bSJD7h-v
zV&;cUruOA4f}U(eB>%pDqB?hOY9XI_jB#e-mmbDPQH&_lNeEUs-J{H8@ENui6tr&6
z>!)m4r_Z^%W>=xZZ+QSSnXV~(Kwv(0xhaaty4ljqh_wZ>P2^~Q-c1>c$8FE{tpucB
zA!Yys^_LwEZ5>EwCT60+r{bnm!k7<rKzk1jJT-T3$I=27*C0J1@{+5lX`O|;fLFFm
z7Bgqyrw~*l(KmtZfy9bZ1RMMQ{rfBXbLU2b;3x#nfY?C8$UKao!MUC1TFy|*e$3M5
z_&1Wt8Iv}Lv5!%C>7>UY^SOWjmHrW6znCMMn>)UKRzH>CGG){<8hG=<<A$O4kaXCZ
z`UhDzRXd2qg@^AVPyq6G59=A*o45X7cUw?MAci=TrZM!Uy;Q?WjfWG|reg`OQ&Nue
zF;ul*;e)9)Fc^GaQ04C5-x<O6zN92!>7DS%#O1au;j`*lK!`z*0bQkW<$vbfL|1Re
zyA^U&YB7pDs;=AyMmwKSs{WA<2l1p9M2w>VvAIQ|@w^TCf6{UsiCU=>N1;l@wv>d?
z&I2FoT-H)dK5>!~%8_WJwNA%#`sty&zmW>t*nf8MO@Gz7OksaCCwdSq@o&Y@60U6f
zN50G~{f7(C(+ZAgqUR5($ZB;82g{?!l3MWCHkMo5s;~9F*!cS%S)}{GJ53(2Q`*F4
zhK9<ENy+@Mp)D*pA-6~O5bjQf)ctyg7VIjLDa)_zu9>=4$KA7GZa>Euv5A4?J29{E
z88;^S?k7ebj73o)@J5!$w^XAYd+x1U3vM!*27R5KH9Ln;OG36YHZy~{1h!3e!pWKz
z1Ox81zLmFreFZl&a#e(=i&_@MBZM@v*nSpbs2*PC2-A*^E;wJibL6;DSy+Wu1CqO}
z?d<Z@`okj7qynlxc={(A!GAmp5QRcWe<de#ic}tukmx-3_A|!h8ic`5pDb9)FXkIZ
zoZ{r>4qWSvpB-b<%AGhDOu-pA&O$zC9}M;uLNgF_L-f^9SAX>7vBpB2QhEpTrKDww
zHaiybdgU#cD77=Nv9gw2pQKx5pfAS=3e5dTMmP^j;sE2I#ah8=8D(!c<@*}j8ztk$
zj&g?YUW11fk>^cm;X^xqh5PPP3`(?hRwRUqq$@lYz1O6k=^sG_0;PWHj6FxI8=6E>
z>O@L7#&B~hKR+ehKBNKaR31(4Ue7pCT{JAFCnjq9J+bV7QgHPFQ$yhnH89-M(O_Xv
z39Oz_e)QH`AI<=#sDrU97-0njz~ca2T^O#^IRUr}?PFsT5ELXfGr_^bSC3*egYE_7
zLd$VQApqWWVqrMdoV7XWg4IuIpFf{6U&#P*Sd?lCiGgk8s@abUZqMeZJ6M0smUx=)
zH#Roj7;snH<lqw5ZOvAwI#J7mE>g%9Har-8x<`;ZS@+x67-pdQ=H^M@cS%LD(yI%#
zcVQ-pvY1fb(jth#5FcL%t1-|BYt^Wnq|Q7qQ&fD5rv6(;{tHWzlzeT8Xxg~&ZfDPf
zd~5R)*giYZRAtQC!)DeI_`&z@aq;nF1d)b>B^ub^qr2FgcjIrD#}Q8QJ@ug`e6kk<
zld8BlUPGmxn~%)z|C`stq*au?c|9FB1Dv+{IeJf-Sd)=K%KtwkP?4Iijq*}K(1L0b
zWu0+}hxng!M3P1}oZcFaOTkbRUae?YMx6>PI%Iv;_7Z^T;Re=66EEqr>RfU%91c<d
z1C5V?>;w)Opan@GAzh#$=diwT(l4+z9=#{Xluxm;8wvz)MSXlqb<&D_H%bd%l8`Mn
zHpx8egwTcXpce6(&*ly7d5$MWO{Xo7ue_gTdQiJKH@BCK?FTRxqbKMsdiC!JZSWTu
z`s@diFe(Wa#VqQ`%t?C+Dk@Tf2zBDqLvg8CTiA_B{IuSr68Y-%OzQ5H&40v^2G5_>
zj|)&+Y{<sK!mWpHY&(4kJtQF85K8H}l7YsRm$$FzhM;FwVbG2`*%bXF)Jms$8zK{T
zLosE|VsRu7Jfr8Q>KpqGa&7(W;Bx#psk9&B37)l7kM_Gs(BT`RzXC25JJa}Ul^A2w
z$dRXGa<tvR%A~lE4mEJo&T(<Y6Cd#-8U!O<U5mB0h|zhV)Yxrm1~UOJ9S?6hup=)R
zESd=ISE;ErmGr*n&vW&Fw9>?Vb^mG$d>XE40b#?S%~qfO&js0rE*tmW#iz%8;=pX=
zIqw$9dZuQ4MaS^hNXwpu(uWOWY+yDXIr7mWqp1EqIT}*B?FG=S%`ZZvX00z3OHM-o
zUL2By0s8dA^V2@myeI7r!qY6t^GBhh*T_D65uh7w$8$g@9$COIaaVF{f3O`>A{XU9
z`2*JExMtMb;lVS5kZYVg+#91;kD{j~z%%?qGlQa8+3DT+t~16XyjD~c6db+=JLs(C
z+!w&Dv5S`Dyg=o<5SjGo(sMg=EGVO--8gXfW6WrL3Zt)j(`OwHp;e}~Zgv<~R|MFi
z;YPDW;@2{frxt2_VLh=Lu}C0})}@yrbU--8w+Bct#-^tiq`0^OV)-?!D5Er0zVhBO
zNdHTO0}XAO{+NL2<(SH6-bF_>1RO-u=Ls4#a_>_(Nny6s$zpwFHZVwj{T!#a7LG~L
z@Ae#m9l&`#x5y~IKWi<*6kBdJC((`dL@d8s>Tnj8Zv351lJh<$EF7NT?$4Jv0p7@!
z{WMHLSK>3zr&2K)8YQ|<!ngkq>>{C#gJ1Nc6XiL82nm3iES#9i+X-^m<rB^ivWR?p
zQdC3?cj!_79e~>Qk=$ugj}jmB^_W*c`GID}xwOAZH1XecI{vNo6g>(DcJ`A4-b~2L
z1Zx!)yncGeL`%}amk$ozMZyabmiE*5LfIjzIjnj!9WV@PMnqHMYa1;dJ<5f9g?~g-
zhz=6`&_e(VLqc{_kOAig8?vx4Hh7Agj?!KK06-QDMXw+GvW?v<+opj5bcuZ9FI`<{
zk$0(o1VbsJBa5mk#;4#8>`--#D{hl=Ea{2uo62=oN*3beOh!(HK+Fbe0!&5dgOxGO
z1nL9&0-VolYtyo_S{4+?i>etKP6m=CL2U}cE_u2$9*x*3;${hpO;B=p+>3iCMQ|<v
zfqbOfZ?UCVf#VmpFZD8V>6oIicK%1`nMR)1HFB0&qXZ-}`i&2z5~)zsf7E1Bf_7vG
zqTB;^i^)m4{t>WVG~Tj8vXt*|2DKcBlT7cF(Sb=xOP|!zLSY^8;}JxDB4c*F^NY%$
zL187DWx*3yOGAUC^B0jn8KCZ28sjN+L3msfjf0ppn%o2Sn8V{Fm>vt-i|dOrz64RM
zI%(-IURb2@qE&5fY>an3`rIEnu?28l<w*RXw#z-J(Hw<w3amoQU<?CL7kh0@iYlbf
znD<Cr>8~5u3o<>Dz#s$`61uwl{fw=&iXKE-d<+0tNz4D6b1!vSe7tzswswISFbg*;
zCR-;{SIlU=@ize~Ipn_wzS7e?=0uE_imXqi9-3IwF&x5{kz<HwKeK%jRSl6kWTO^f
z$j5Ws!ai<%-mTHs-R(Dd;*uG1=mM$*8BUbl63;N2^vibAtFk-DaL*hO6)@{WC3f&+
zA_`{E35ohPO`sb{k*i)x!}t-T1t1P6;xG<-@bm?WVFL1IwHBlr-~j|&w4T|8+Yg5p
zXFXNy{9>SX8-oCNeJFbpccJW|#3tSV>=*Nc`VthdfYb5ZV&lVV@hBO9O(q|MWAB@K
zr??xDsM1}3^uxMIPDJD@YJ~CemBtRq%g{dnqP6Dnf-j13CeP{{99&>SfnZ;M8Xxq3
z=2io$4!U&<UnhbCG)6FIL5ni-9d_>X$g>@O|Hk+-eiaG1A+H+}=iQ2Dy(Xkl;f=~>
z8)v$@xZuBjfldOP@JAin_#Q7xTlzncKZtii1`x2Xa{_NmQ!4(Ao>|BMhRW7#n#6kL
ziX8_rU(u<yx-H9+`hRAYb-x$ztsQd$e5!q_9SI(>TAiE31}sunv>EKs)-{&j#;GIY
zv1E@eBQH-B{HPEgNb|(%e%mUdB?M;)_yc6S@bHyV0y$awavCmiuls9dfCbJw0Yujk
zgX@elJ1ebhl|cZ&btx=SO1_G<(;5F`xywAMpb@86=-~T}wF!C2vt}H$hHSSS5IK7u
zhdYKZA*Fi&4Vu{oUq);(cn_A2IpZFgj<O>Sqw{V&SznZ#fpo{M-3^*TBp~DrVmeV}
z<tQ;HzFU!SAp2{yuM*`E0J*T$v_ghkOaqk8f9QqfN;8p72xas;7z-kDaIlq@jZJ5Z
z+X{#++}zJ#Js-kTPJi#Ic@;Ca7T9T645R&*pt(yWH8jiddL1TDf;P2L*Bvt`S_=40
z<J8JBElO>*$6PYDYXl3)x(;<Isfr#qZ&B391SMU0yBN&7Sm^1GN2gooF40RE-M7ri
zAR#s}uQ!=v_O02af<$<O=PjhSFl2l@Jt%x*j>thbQI*CU@bWe&iD%4i;$q2aww%N)
zYyx<xsQSBuB`88w7duTHF#<cm9T7D7QVMKgBtHOB0bh&SHMrVp0m|W<J#{1V*m>b(
zhmBrcYwGJ)tFXLs5hx3MQACxNqF8TZEFYQQVemOmqgP3~<-f{Y1`i_qV?@Wqq+xKU
z_<tSdG(=&3A)w`rTnQ2apn;6BW`i?lROG^<1eu-<Dn36IKFW%S7EU9|Ots48tkU0i
z<M$(>FyRQzr8g{9Rz{>-a7>@8ketJFNt+w^^aXBoF@!nT&{H1B-35HSV+Ue>FjdaY
zjj+BRqt0D(FSaET<0?G(&_BxmojeJ!3FL#OCYY=|17H04^Q}bVR*kThJ9q9t%JL{K
zZofvaotRJJo#0^5wC@|FA4n5YH{T|qY4Yx&(}DeXV{)Qger%#akf;`D<${1^kp0xD
zAqaFto0&Mt2<Q&65jED=Gchwm;CmprpYlK2rV@*&;5&Cfw+0&!qYz@ZH#Rn10-^&v
z*@Pd3&JGRH?%Ji;sq%ASK}5$+h#yoFT3UsH)uO^eCOg?|jo!-V4|#Wi<{?gGxbYwK
z_MUsNK~Y#}-sd(#zBW|`CTKY*y%sJ84uR2+(OQ60@e+(CjGjjQ=_$zIbU+vUVaVm@
z7Il#LgM&dMAo^=CGq*Y9!}FJe7OD^MoMRi`FtW00Z~2AxCUR@)9<0VKzu=D$Ir;li
zM8VbiQR0zfO}*VtHCg3$y%Lh>K4j1Z(kfn|!2pKd2&0?s+^xNRO+M~b<aG@J?j}^U
zJ)K}ZX<V}Xy`pg93&x``57H(OjcB2W+p&VSHo}ZxqEFo;UU+UN0coX<G}cfP;+!-o
z<2hi=^Vwm!yL&I8v3AB^V;VLO0R~h?M$uq-xXlq7p9$bvxLQ|~VNAGJZ5GSplNFIF
zVDIj>;r-%uVyU64gp8ZeM!q5iKib#VsFWU^`F#v4wa$n5Oc}Ec_zw0)*tIm<$<iuO
z*;!d(@(Sz*Qpqi75r~E-fuEk<+I-{k?!8{JC-V>d{Lyq?8$S#XQq6imItNP~Jdb^y
z+pso0R5z9OJC;^%E>lVu6;0$Fq2TPh6B^o7SC{R=PlUGYZW0}38vN3yc5FY7hbH<i
zoF3q|1nM6S=SI`pc-|>DW#2n~t%Ak#Lf>G|{kw5Q`7i^;gIX#|%A9|HNGjm?H~b+&
z4Ar)sh9*aOFK&wdiI;#6Jtw}JnJ2ZIL5S4CV%2lJBmN=GyOJm;mr+)$D%<!YX7lbG
zT;FA67@iUT5udGT2XTShjHJm2iQ4%+)tYs2ffb|0>_AW^{l6dS-24OaYhM}RT`@UX
zKDO?ik|)X{3&(}4qjzQ;ZsD)s1BeMx$tu~$_PpD*uxM#P_&%kWOz`zpS1*4)viSpH
ze~DN24d3eT%FfJ!*J$d9?=R{{{6haeO0o)JlT5bxzwCDruLtoH;y;YZHYuL}{ucuY
zQLXy_|F0xw^7_##-3v9N85={kRaMo{EX`wyijPqRum4>c<KLT~Dt&T!ZbH5)R+J<1
zuQ!)U#cQJMDlE*RC~ouPCqJjT{fR3u%!P-C$VKw~bvsLCmhhyjiqYIp8P4o|FKqvP
zVWZ72oUvF*1;5&!5@u?d4VnITIevToeb)vm+$c`#pnZ}<DGyQ)A8)_<eRoJGzk1NW
z%by!S<V_k4%dc0xvikC6Zus`%xjmZ~?2Q(2X(+Yp9Z7!np||qr-G7(0iH3L??T?8t
zeY@Vt;Im-;pQRKn`0pP;{ynqz-yKN*_ZKL4#48Rzd9rC!K)frR#P6Rxc}JO+_44K6
z1nJF{kN31+B`%r3jJdWG274eJpY?W3B$`v;dqy87{_(o)gBx<(2ei$n8OzE!Vd+5p
zrNqqx&*19yz<3r4nz#Dns}sZdT@0`OeSP~LqCZYx2K{cb=g*N*c{Rkxd3(^@pFucb
zk*f4O{m(@l>UqBfnlv{GG6*#1eoxKy*y*0##>JIwwe{I^MOz*>x9;~&|K2co^HH}w
zB*U`4nsnKe^54~(9-`em)iyuVt|v<#C7Dy%)~GS#QZJj{J8-A)!r^$wkdUxG(phTj
z9lkCcR7)PR1AFVZ*Ya-im7BBHJ&TodB6@D1^>M$-q9p$SJc?{zYq?KinXUE&#F>D;
z*8BcgU1?Zf6)^hq@eVur9g%%#oXG%7LWLUCg7$F*>#LOf><;<Npbk{_kupKwBg2Nx
zDQvrUC;;TRxGe-;M;d{ci+?NTuRK=6HdpYvvJI{5CS?XhvrDF|*5-9H!*Dkn8WK9V
z4f74~oB?vMu(1IPz~~m96kGf)pfCi)>+H0>FAln@kx)qj2iV_`!56%4mSINTHOU&w
z0GoY1$J=e|+KM&m1ds!WpArBXA|T+VnaO)|`W}j0k~Ob>HX^?)>;g(`G=2b=rbE#~
z;e{YH@LRa)AZ0y&A#i^4^4G0zE-D_3!q1g@#TI9~T)hf+?d6|8Z(O^^v~M4LXMw$e
zi#h%M`}?cIl?d;f0VAsKed)iuH@SKDcH|@1{9D4)L&3_qXcr+kbM?om50nV~P^*|X
zYF11o$+m6ldsWa<S6@E~AA|$!mrNNqZ?-k1?YtsIau$*CSzpiD`UhlcFqALTmXP|v
zO9Ui{viVR)_$PHZz+dq4sSYVcy@xoZaCdd6Ae}&c;7t6=pSoa^vXh2}Q~`)0hFU=p
zV1l6BBS7L0{GQ>3kL|bR->t2BeP9g{-?@o*Q`zFH)_z?@1!+uDQqM8mGw}fv0Z<VE
z@(u3Y+o@M{{vRX~Ac_$s6)9j`zy|?e9~O5cez$MmHsnL~DIwrH-pCY}`hZoMVRfUg
zXwfhA^^)VzNdlz_hqA&hT?LZrAQi=o{1@sb=Z^LEzB0`yh;~RXHMW};=@d9MmF04w
zL&xTo!K)~T^tSi=-wo;PG1|=-rcJVSizU?wj`}~fu>%^B1d=VB39}PfZa?ZvT4W^|
z7f0hg@4x+t?h>S6RNQEk0Ij||2K6J%Ed;=B%F2Scj}-%O#BdJu_ZLIkz2CUlEhiv=
zbQetf)l^jUsPaC66AX$g+AC;8Vzq1`B5-tMr==Bw@HpcGIe|d<)D#BzE9SzbXIyIE
zMm3J`ai2cDF!!WB@~}+2$JMVlz;c9qa6iCah-@$9l@fXG>FJ%?7)4__*t|S?8dEDw
z#^d6S!4m_+?N=QwA3p*>KCfo?Pd1Lw2p+9RLZF~!s=*QuqQg7zPvI?t-7fS7(0puU
zAl3Q$BuL9OaGdXzIjowji%>=zYwH&bxmt44(tQXABytS9a-ZW96FzCE50N~%$Rek8
zB3f6mB56A)bn^1Qf0i9S+O1%jpFH%y4-blTUJ^#>u;t0=2)Mn2=_@9k7cQ)VC;=>A
zQcMilmiO=IJO2W$4c!%8I952;0KQsqum+x9pxkH?F%rSUt{Sp8+4Uhr2#8TOf}tUs
ztT8ton#`O9qFS=HmOdN^6n+`HdfXmcTU!l{dKeyLj2u2!=oDl#i!=|EwCELUN5CEc
ziz&GJE28Z&lfm{;Dw;_~{8BSF$=C?*%cJ?<Tw?m$`uESl$;sW!nty&qHk(#_d6}Ae
ztK<_4&7c6V`1GtSyY``LG;K8Oa$ZZ{T9Imc{=$XH>U)f8A$u#xb71!jec=9F<adf+
zjK*`7jvG#VVVWJK*%B`$di|SjKHDtbA60QvqO6Haob2;c?!T8Vz<yL2UlK$33>j{2
zYutNe5Vz0oer;Cr(G@Mp1xw3{sBPP}f#U;`08(2x5w1fZeAB@VQ5e)oyd>;t$INWs
z%<|qN>ei0Qn(XV9O{8gYxv#De186!W?<zlqD_m|D=z)BHsCVu(cuWre{421#1Of@n
z6FqY1ojV*)F+2o~WLGO1@psb+*IZl@ge<{?R|E3}vfEx!h{^#IC|!<NgZ~T0BeZ&M
zZbxHE=)(&OSJBwPCNIN&tZfb24p@>S<N>nC{8S-H$)9s`hXe)D*IfQV{M>}47ZY;9
z+X<=~0*+jscF#`zu9bg(;K|DH`*#@XA1jXB^qbaEK3@~k+7BS2Gwme%{`bBca=ey;
z0CdM$oRmjrXB};9o*O5PeEr(?*3mp)sk2sqmI4UFX9<WyBrf|r2Ks<U_WvsG%)@fr
z-?je;%QB{ru@+H=RHzKi2%(IrG!s%%N|Q9OER@KQIVo9&N+n66!L+nUDjG#aN;IO9
zQr^$q`tAL`$MNoc>|^hL_VZVG^gQ={f4|@Byw2;q&a3OWM&o9!-UWSHdho#!r4dZL
zX#VtL#~okuD`zu<SFRhEG_B{k3)|94*c{s>>BUF!wjt>OHwL{)^vUaTRWJ}n8u!)+
zaR50EOi*WHWaRGw-rpLwIyuFyEulJ+mzNL`p_`al%J`Q=?b1*JMd8|aaC}$=-&|PG
z@7BVwZlecQuRs$u$_RHM;wwkME~_y0OIr*t9a|QDIc59Z8D~8zSG}Xxq;Pp$G>_%}
zsF#h9MA33hQ}@Qj#&#n9GM{tOLJ3BVjPaV^(h}`bqZKReMMteyD2`fJ09TUKu%y=9
zY?ZMwdj2Hz?ew8OAFx>XKv9~mn7GJ$jm7KWrwbNXN-gT+ayWt!wOC7k`XZGb>_VGb
z<=bNYj+qJR&2(lZxvHor`ad+!x#@15+KG=do3(ED4gx&%8MaomV^&vtP;+lRq9goz
zw}}&)L=h(b=U2oId?kNCWh!q&?Mq3Jy9pn}Fr~s;&VZp;6;CYMK%YFCo_FuQJRa>_
z<GSLgKBb&BYuA>n>fJheckqx(=4bJu0VvjdP0=N1VaO}=XCRRepAE-~d%L4`q-wNJ
zXbGQOBO#yAQ<U~%Yw*pVO|1<#W~=}G=+Peok0G_<J8v!2ae4R0rp(z{HwihbqVn67
zkV&tYbi1Ttfvki?lB$8F(y7EbpWF>9)~Y3#UTdjwG}*dj`SOXfDk_O2(?}n$y0_HM
zIQ-9NNlSW(a%1dY24udBx_oG$;kn|!^4#>$RMvXYV4bn;?YyOnXG_qmjE2!LG1WT5
zPmh*xPxDeAKMpJmld2GbJiT)}kaYgg3G2HahKWcJMvV?>j4u7dzpjP?BawSesS(U#
z<p2-<s7DA@v-?6H?E<aF5@8GN`q_KLJaPY3=H?wpG1-dD3`M*t$lHsHnHFM4Uo!op
z?~RK;^~G37P+W89`Tg+ntzWHl%8tX*&JKjJ^wO%m^YCO<mJCEN3Zanne6W+LWAM$z
zA@W*GZNB#Bg$wJr5<OK0xyNof=AN><=f}OnH<zd1+M;A$nEtu-&0jBGxH7SZBiVS$
z32rtgS4U}i*X5bzJ7G@&5tNN2N{G*gOla+b>0<yg?Wfm~3SoedyclyzOa>iC-XeW8
zp1uB20R8qz9XSpe1kv1B9v;IoaRUj+A|*n5lP~87lY<16j^qjQ@~yQ+$i{?0g?}F%
zukLZ4RYMaM3a-{Mb5RA-fwQ#2{JVWmVU^qwlk{p6-#gERnT1r0&KNu%Hd9TFGCi>#
z{#Dzz{n@!WFWJkXJz>jv&6qwtFQFshA!Ry*`O}jE(=lM<B~Zgy%tJi&>D}Aiet^G3
z=uB+?w;HS%K3KOgqujDbQYa05IAsWg@u@5j7uy~-z5zOF<Hj{qvtbIk<5=YA5%pw`
z)*uBo_SHC?xf5*^pS9B3Is+R+$|VHTetREKSR{TX&aFd>^LpwxdHKrgMyLNx3m^|Z
zgA>A35b-xm2`2B1(L%6^ugSnX<n%AqH-^dg?;k{5?V|N$eS#Mr6OcDZZy4;I0i}VX
zC11z1{Y`tL3uu~?=R{VO&X1;t40vK94K-bBs7<JJJSh-ChCHEiMBjzlwhE-b<rg*y
zkY-owTru&VE}!zFn0qIAlt*%X-9xBKbYakO6t9^jg*cNajj-ukb*S^Fwy|;Pn65y4
z7Cw0JF6~=c?VXD__J-A-!wLiR$fPa#kQ0o|88&S31W9O0w7cl4rFYK#d1a2qY9CF*
z(1km8yi(}ZJ1;)t*oigY_7!N)-#-Xj9xdECXzj%(6eP)HM2FrmQW6qxD=Qx^42=;Y
zn}(+oS4?#oK`IF9{#K>~0T-5QwH$ZwWu6(lTFAuZsG)I|EKJV1{)Z1+>P{aW$6{Kh
zzLz>@pd)2?xJwoP$wI6g*|BkJMfW^4H~Jl5^4NY?fB8*P^4-Ue%*?QoeU;%aG~>RO
zxg|B0yLHccn55TOS*D<EDiMbED#WKo-uk`yQQ#0|>vVKOG^-4f(Y%c-^7&x#IimxS
zJSEm)gt|yi4;QT%CuU&4=p%@QIcFAYZE12m^LH-H+d2e^396>&RwHHvUzk5xZm)L7
z!Ss$`qy48e=HzKK=0dY+ZR^^B1(g%bK3=#i%_&o-UW<+8k>d`}?Sd!eHMNh_uf4bj
zg>^CnaA8(8Scf#JMD>g8(X04JvKCoE1)8tnlKKoBb)4)FgAPtDqbWRdupr;7^r_@$
z#gW=Rx4FcUL_10b<@x#VYidRe9XfRI;B9zE$a3SQwp#u!*l_RIAvbX%kQo^orG7?7
zLCq6HWHI7jynH!^#GiB>xp$)6>G~DB0|Lxn?7&6ep+rOC%<0nyr8R_^6pD(m=)rY#
z${#+wSZZ50g|ZU$YRpwQ^ANJk2wi3{tkxYZF*}A)&?r!avqb;u9qKA)6N5+YOVf(A
zULNkd@W~US6}p!%t8N~AXv|_Qd-=UHspDW7)p%=Dmfet|EjO-Tr;N_)<>zfcS)|lA
zl+@TbZ#IQ7EzH5sxWvbcF7MaMgZN~}8NCf-&vm?HdLL#KwUSeWtElNOtbi$#Cy$dw
zJW)}RyDE&VEwOL%Z=@F{ODbp=EPh&iap9pC&&uYoD-mwem2jKIM5#-rBbnY2qbc>(
z*UV`mulgbk7yoY4Pdog)SL7D)9{wI+R=U@Q<RP*~Qgh`Ft${bz)~r=l7e0H$@Ci-2
zdJ|4%1?tz1%E#*URHn?%gJx}?Fp1cXJr_P*CkG(O#upV7N2#*Ffo&z|aU#)2J7~WY
z3_qAqKRD0aIgR<iZtwq=a9R@!z3%k5g_2_SEUOgPOE9<>F3g=Ylv7$6Wk{x73z>Ex
z?KX~hio_O;(e{}=`d4JC5rv9p!o&TcRufFgnUIE2q)JQk)z!U+mzJSH*G<z(f@buJ
zNmh(|=NSl<b2&MX5&bkV=Z^;;idw^I&rnVKXY(>BpPw<9iQbAY&-yK+gPBH+{CuMc
zk_px8mmZDD>r9RKPy>dwx1TIQzH@N&tYz*)LUVo(KnXE$$Qa=Uxe+YJ%6{VHr~G6z
zC2iYVvE;F5MsbQx4EI@8dzl0VV)@n#snr}*R#tfF&RF-LSJKdVjnB``xu|Y9JPS2p
z>i4G#@i*tJV%YCl!@k-i00l`mZjMrOI9~n05im`+^!V}flaBm#BrL4Hp`nX+Zw_nF
zkdvz-nw{<M;%;w?l9Fi<F}H7@?j&;4GtKJ_-IGj0V`j~u@ZlrQt(esO`Ev?sAseh)
z$Q#dl^knJFr{6<U0^pjo1;-R#HRFJs5-XWshgFXRRgLOy^oPc;cJ>lYWRuy8wjxL;
zdj)r`U%$%K^z`}jWVwPWlHth`;}^P(jb@`7oY$NqwDL`U^5DUGhzY$%m7j~fdQ_w2
zrZ500j#wJvg_J47ud^Nf&(MB?GW?{nby@ph7mq@{{I^w~X#jh0!-CN-$R3aC^BWkT
z1=~T*&(E*x`_~U@Ceg}DO4^<c=diHhhykn3a?7Y46*9;D6TWF5c{G&%$o?Q{NQIeF
zZ=oa@CDCA-;@1)!HoSRoj&?^Kx&tK3ULc5D2Q@Rx%KSJoM(7IJfXzYSZZOo%ng>W1
zj6`=?ql&YHb?eprx}B1ebW<b@?A6yFD@z4#ZQ`bsyLS`6(DcZQG6PWwIKjnvL?oa5
z)~9uO?Bqo|c6{H}qo=S`8$RrtyXP0^f@$N&UnG<xQaEv?Juu+V%89aEdF#H+>)(IU
zn?_aV)6FvO8d5F31->e00~@#pl+-rmoZo71|74hlz3mJ<%KU$JDW^3{7U64c#leCW
z0f2~J`}NcJ(i+M&<Jx@CKWxD7kyM(r*wFA0CEt*cm+T{&N!%@#cjq47qkInal$Dh^
zJc}fYu3`?47X1J87)3}ev57YWz)<+MG<NIsP%YHQ{3KEq4{pX6%cbwc%1@uv*FE?H
zK!#$Jt(ATML;C?A`VbxV^rkeuwlP6Z)?RD(hlSCe_~NBl#I2Y_ZbA=kzEstWf6EB%
z4v&dhIcX@}F3}|#?n`YMT*%_Os#e76FL9oQW^L;3#-<%}bM9AuCUfUCK>FA2bvzIx
z7teA$IAVGvU0#ji0!q%uS(Iam*O4O(bXUH*tibPQm#^%*uiwp9F99`j93nWUcO)Z7
z(hE6y+{@Dsvr}Wg|B*5ZLnkXY#~B`2<S-5=`R1CIIUZ_k&$9}<J?Gov*jDqxLRwo{
z_;68;Izv{&R!rJpdn-_XsuaFMW$C_mNdPabAj`q$YD?FT@9>z{Tw%l{S{mMza++>8
z;c2lmFh#H|GLLbO-ti{Ug_>vUJlUhn|A`N|Kk39%>idCu4_?(n*C`g{tFO-(J!*BW
z*B>)Omwx*yCVcd+xIMjO%qq#esJM+8BX}v&C5*3BiiO{gw$HQ5y!NcInPaC=nFFq8
zkH-`a%Riaor0i6^LwbBFb-{TTY6@Z?5r&(X{=XSw;i-3(n+M(<Ju1;MCfl<jRroAX
zMD5t!eWoZWL1O8t`m{BjR2Tdo5WS3$mR6Js3OaBA^AMtfl7^j14%}DmbelEOquZZ<
zL>I90eqpwgB|)5s_o(bIxu#P}mP-r_Hq8Hwpn}%31+qs;l1W$C#U)1D&Z^H0GBufX
zIht+T{n>8P#3ynI4c~TEiGd@gMVuH^&mzTcDA6c`^joe83<5+4XkGG5_*`bz=gyXB
zS{R(0!txniz+uHd(g#JFk)(Y>I=Onxw>IzpKZQK=ozVSi*g?P1i7L%It^T>};immY
zbaiOX4E0YR=D18_)ebj{36c^km(IUd<vJ(S*fR56B3-`8M_MK$7zn^*8Zck0Mp|4F
z#2!i<?)b&y|0a$tG22z%f7H)7R5G$%>!0<L+n;MA8JId>gOkiaeoNF@qbb?2wzBwA
zoOF<s2tAsPvmWc{tTQGrH1+B|GW5=TD`oS6z7pW7+YeRsDSWo#hm#!7OUl21W{Zt?
z?%e5gXjW?wbFT7zZdpVbNkOVA92+$2a@ogk|2$D$Wlp@ly~J)!J)<WZ7#v*qO5EDg
z5?6z)LU9_Xs?KZvHhq1s(BdCx(;bJYE@Qqp(LO+b($L^BbB+z8_@6krkbVOcjt-p^
z{t6#nTT|51&PYJIM80}yGUf+}W0~%=_Nyr;6T*ZI@ldEGi8alO!c+8Q-#(U~1|Xwp
z$0pDzbZLpptJ2e!i!wg+?!NKfyX`=AoAHLBwTT`TYO>|@;X??+51u$N%v{HyTe*{X
z)8qZ$N5S2%M;jU#AY&)dtgWr3VT_Z96Vpr6ilmAr=TIk-&4Z)&&<^<e%^SEFayM8J
zHIF(|L=gn-qeoq!FBsT+M(T#+TKSNTn>MM>nZuYOCs9T^a$l591W7_L9b?w1UU|$p
zn;=QAGv$AdT$dpkC>d~BadoDjH{H7>+Cu3kPNnC)-t}eP5@Kf(OB(x`tOz(cV_*!l
zf)?kv<%ShL-0W=YA>DcYpY>kTTYwv!jQ$@lu@@^@P551a(V>_&g~7osJFB`l_H}0F
zL}o_fLCV-2s9-s{y`>x0t((r&u5}OhMHPmtyE`xp&uEHoQ$>%^{IDT3zf><<T~$~2
z%nIf+M^DxM#h<k7FFP0GaagD8_KHcGem`y-O%c$RU*{dO-mF`=WbM{s+h`#HL%#^3
zRa8*W*j6Djd0zRMe4bfq?8~KkBqYqPh0k>gU3W8i`rUxCoc`U9QC+;R=8rz%IsVmW
z?07$E;)#%u)sQ<TCU#too!!Kt{Wt*@9U4`zs_5H)mOHHNbJuJm93nYKxQ@Rzy`$`U
zEq7-pr!HD{Fk;*y-^4czkvF^p-XqL`d4jWw-0zWe(o<0q(gb15iGbb|krk6YkSDTa
z%*3^rv5xQ`0wBTknQ@UInbq=E?RbN9d;&I1Bvmcn>(?a3Ourla?^m=qD`4ZU2BX!h
z$DddNtH%-M?z43Jb`|q2hjKpj@AZ4!!9fSbJG_cZO33}PaualOU}j2lb9MFFymkq#
zM6hW;CiOjAEOv(|z~9Lym^!j*S*A6*5q42f%OT4>|H|yC)nsEMwp+A#cK4ms4W$y@
z{gY^yguvZ(zifK3iU0X=L=PCZe_jhkeY+tZG2%aY>$RTZkN0O{_n_JOfUkJ?p|7SW
zzap_m?7hsMX1~Xe$?fMeO*clVQ)llLgnpZg{`HF^lwq3w`T2kP$^VTn`5(S@nu?o9
zBz-D8JOiaQK5&S(MyN}PYVv^Em={q4U|_hGrY2BHiwev9KD|X+wASX4hoP~ej04?K
zTUn_>N&t&KKa+oc&Vj|Oc3~EjkgYjw-Mm@w9K}*EIwppMv8J(6XM42)a3EN^e_zcJ
z3TF&%vBfT5mY?58Or&*|o&GLJnbDU9dSbgH($j^;LzWJ@+JXK1tGKy|a_8g7`jK&}
z@asHt?B#ipN==^pk;;*by4%$Y7klnHFNe;P6seX4k&3$pkz5unAwn-+_{95TgVS4V
zZlB=2<C<-*p)jW)G(ftysM$?OBta@DdQm#whASc!Ps6cs$t71TOlStm&Fi7Ne8N$E
z(vYdAvLK@poy*e+6ttuYR!LBi7$R<VbX0ZQCCXjR$kqDSDKIpNm-_u<4K!Natp>F?
zRAd?!9qd2a%^QjnF4`{A6BA$fk?TeK!&EU9x3Qw!izt<3w7h{os`O0fM$8pTyp^;9
zZza*=qlo@9gwGj5aS#YL5Rl2U19ZyDl2IYk;Mk=+Lk=@iT~{<Ml?g*Uiga4qDL*O}
zK*Bx;eZs_Rw^j6P6)hvJFvcnk>kGg)7+}zt;XI^pbc^vBU1pPp3SA2Of_sa+Z}LzJ
z3k!eg-8)Jvq!p_9NBPgra`5HAVIxG%8$if<@cfs}HVb{YpO)x>7xwk-FY-PH*o3UZ
z|76P+`ju`LuCwe-Cca8qv7FY$ar5_q*0AGrtGMm;i_F0s0GpiT6B>C*T`$=7*d@at
zj57h8e)&&9I{Fi4=;?`UdbKxyoUey92i7YcR*-1LtFM_&j1t@RQ?0+$RMYD=J$80N
z&jC!g6rwvG7fE82^aax@?);6|`n|nm4jM3Mf*V5>X5pbxd*8;3a@7l0GGOO4*)yOM
z{)50l+DncqX??#(gk0t0LmED9_{f)CHVkEfRn^p(k|qxsGXAJOmk5AfS0n0iZ#FYt
zha9c`H0;_R16%EtCp6g@P)me}d<#KOOY2Cs+cdGd4^8mAf)q8>^MGMw!sfilzC+|k
zsH%%<F0}b{(WrTgbdNg@qjypvoZ=DWK*0;swIY!ky14*VyU^r>gmIB1LWAmt<lNhn
zcO<LI@y%`4F-Ai3_xBVn+$C2p(2fyv_i#(;7)*E0==311faSuY`u?74qTFK)`+<NB
z4Zv(0m$Kq-jlgG`sYk1;il&Wn{A&yCHc75P>-I|XB|9mVU6h~-5f;`-N`QOoXyH-@
zUAyRbZ&we*#t>b^2arCz^MHW^De08m^^Ctjx77JblPBBbf6MNL-3ty3EFrWAQPb!9
z7h*f5#vkOn`vwgb9qidrACJ;9_FPMvt+ZF0Le`Tf!l-iMkhRy_-t#lXL<<pl{~jQ}
z{JbNTG_x*9pdk{4OhJmpIowbWB96uNuHQrowsz0Qq*!!uMvPM8dgL7+za}e4ay4G>
z>|DOv_uWXZX@h`AkOvA1TZB>Zq4dg!@2DOr`--w1_iccLw5WLnEFQ@G?AHRae~Fbi
zj(-~ZQY6w7YAj}EOK|F)>ktax4`zbg-mfDqis(<PsU9hpLzQIUA3}BO&2C@W-F+%;
zC~_i;H!jZ29eY$C-=mT%8GwR)zYY|6$C2D3Hq-WQ){jlScI{vQr^5dD3Q@#kA(4N?
z<f*@_T$2oee0o@0?G-*Ii2h?rB3c+Gy}s=dUFA3v6S%`H2lns&l@}LA4YIXWvb=K>
z*uXAhBIe_<D|_gQ50{W|VHtqQuAKLN`sjVe@<Sjq$)lt)Z|C*gCG!-aNOpE?L_`I?
zUv(~{2bpI`J?s?eE+VJq2ti-Ig_4>4Thg+0m&$!9DLa_V)u%PkuA+yiLU{}2r1|>>
zJju=$HeksyS0X(8;FV=z<c`~`pH1&RyAYTJ1t5|ZO13y}>v35bovo#?iX6-{h!*qv
z{3?394xSiWHp##ZU#qhnS1hN+<&JwTg?a%N;yB5rY7w-XOm1y>a!g>t`NABJuG_ZJ
z@6LYk(e}p^*tflTx)4Q%2xS-{4|1!?HK9i<-oG=8BI8skOaur{Py`+A?W2S<bCh7-
zPsc+}xK8-7V=prNf0A_~dXBfw;9Z`M`zaQSD&n7E9L1t0BGOW5M*m$<P!_9(&>1@?
zQCcLjD@)x0ClDAEq`*vR1%>UHku)`Vb$u7Ohyl{0L?edKGRi6IO<;fl;=9qCa9FFg
zm|p*qRZY_OC{wi$l=v5VXdaaFKLRWNArAO&LLi%Fl*_}FENbAh=zs(wXlap+@HeNA
zAKy@u8?nGO<(vSBQU{KnIYRy9`FGIzRiC|h>@OOf@atZ?y0JK6kAiAsD{YDBv48Vx
zTPLpBz`L@~x?uO^5PI<HUNymZ1+4*aa|WSM^hrQxBN7k;K@c%<1=9qIOWl2!Ct0mb
z$ds0*IjjwlD@Tr-zYQUW)1F)T#((p)RqXA?jGtZ90XQ)RNcXRKPvMoxpNbNS=C_`W
zNNhIt`SCe(Rg@c!1N#Q`66uDWxd0m6HAvm{{ogp+P7HYx@b}}#@|v1Hj{+q4UdR<t
z1kLHj`C6V07Fem^0C;E4WWl%Uzya1BPr3Ga<?o-(?zq`g5FzgzDE(`Vo_q+tTT@Ov
zm0Nn!ktTuMw>4+>?_acG!+XIR3D$vr@{#IeC`HZ#nc5sDX<6Iq<lt}zBaCn5{%CTq
z8e{4V$S}R@5o;hHI^A|5i84M^7b1Iaegm$@9c>!<2tw3a-w)_C4(+AYs-WZfPdko&
zC46J+lgn9+i5F}?N;Xm*N-p=iiv=cCESgV+XwmeW08MRtC`MK1gy-D&h7=yU&scG1
zH@ik1VCKCfO#K1n`vQjQPPF##RFj}%BwoRTT!v^2A@m!hUR?3dpJOkrDq*Y8W5j)C
zF~k7})Lq^^LWqJdgb+0Tb)h~~r3DWJ;n2>}{A!?}R?Zj|ZmDD*sS_FIoyqA=Cg$b}
zr?T!g55}~Cm9SL&!f+TWf>>Djaw0V5z^kI_e#E^p!jPsWcO3Caz@t{Ht?l}R!GwYa
zrtzruxEXquX3yrQ_R1nT(JmP=F#fKixUoCV*b03J4!*fo2fzhk4}xGe>dV*OTo!l=
z`<ry4Cu0$ciiEkQ8C~r$^Dg9GqvDkW44`yLOC0&UxU_mR6Ljuk59AQW=v_7gw@p?u
z=r6}n4oS}USe#2wP8~U811jJ{x?@Mo*zor;xuqa3H^<#LhVRR)9Qd2|xd0A8f*pn-
zaE4TRSiPS0uP&*zc}OP{EAVL4);=FV+h7H{jdkl>xMa%*TTDWtS@r7_Ju2??$d2V3
zDwLX`Wf<14mzd09Pb4-_9aMD+@k4m=V%6BeG~Xg=2r7t*i}M2)`}+7m!c$-*YHkzC
zvtipiqGDrxQT_?q3|?f$9GgYf{#rUZ)?4m{r5-n+X$gT5Po9+5{6Msd-NX}1)a))e
zhcPRhqd)WD$d}MZ45Rw+AsKiB+;ND#mgRdP2&U8E%#2IiQz<DaG>2ULrdq|4xUnr>
z7HJl4Z=6GBIzNuPYYOFj?Zk-!qza`wJwzf){k|g6!Ya<jI`KAP2;G^^zc<{Nv%UIe
zbH;oeY_@Mtbm-(9C@wAC68i(O%C-V2C5x+T-_l|FZVGa&Kt0)Pb%64!*un#B3&}h{
zh<1C;_}I@hVZ%d2rNvlgs;Q`q7d%U@+fE1kzqAbjuP1)N2YvhA*EiNMUooh<3#CkP
z;g26aak-Bl?+bt!S81J>O>f!}6h#dUaVY}SiPo@lE>Ct2&&bHwv{5_UpEKsUs=Y$7
z1SitYY(wukcflgYw@Eb$OT+1Mr~9?w^~A%O=In5@jna)lg{*RWrH2dkz*Mk5Zriqv
zjl}uWeYU=qjLEm4vQ_<)zbP-Ton>oceT_9ZBo9>T=C}%8b00dadw=_Uc`~tX^alzV
z=*Vb*NazHy`GckKWK@TjCX%ww)_w<gr99<ib8~f-l_jVmZz8Y3EHEA%oDvk&KL);X
zT|rp9S<-=kF=NK$58|IPZnQHXfo)q(TE#g*p9mp*?RUugR|n7e9>z3o!$VJa31Nr8
zODc2VVrE)z$Ce74qm(sLxbFmAp`?Mp4Ve~)wYTG{PWdL3u6n9=&(E<+EJ*t(jMWvy
z3}0$%8Var-tF7Gyp+;`Fv0i@gubfu-t(6}?F4hBbd@suAa;Jhq0lur=+26Jg!M4UA
z<klYpUyhBWLbb)IAgd%vVYjlPVwGzu)dz;V2`(<b{cPJG7ACDAusYlP?&^s|;?|4x
z<{G6utxr=nyFuC5xV)^a3r98jOV-}?P~6#uQsbCfl!NRWVa>kwuxvhQcH1&J@J2ys
z!f2(!osDUW^j2S4R&YtHhv-fZG%lj?58YiTy}NJH>G{imv5|0_q$V*ZP1dYQP|CJD
zzhj@m`^Pd2Jh*C%+s?Z3H*`_Po5iW^hNB_Bo2Fn=`&=lS)fRqjX<1Y2|MJgQ65n_X
zv((gFLJ>493?B1jCiPy=w%b$frm3h<N{zX8%~4Bc+1;~78GSV+NY0{+jeS$s1Eiuu
z@mf|i?n3RdrNIuW+u^fLM_r=h>SfCY4NtWa1U;`A3@3s3Ywx@ZDQlGNisE|tJvk;<
z_ky_%Q6~}2(R?`g=%q_?MP1yYn|F46y-eqg1b<6)bUTw=&-H$IRsKbNzg5<mTN@WR
zN9`?j51nM5HuCP<^$DR2AHCWY)<ZOHtbj}w?hCC6%AS7t@+G?yfs^P)jbUysVyNm*
zpQ2CxnC3X639i6Jt}f^Ximy#4lN<`7AsS*C?;d+ZRnn{B!p5ZgrO);a99>+<I!i7S
zLqUL}70wK|)TqL_7c&mrt^2l2pPsVg>!#ZK+g_Fa*=|^omX=m<ql$Hr_!Oasg9$nN
zcE^`M>Vj5V2Zys~&O~`wI7v0oW@#K5zZ9u=$>0Xs94%^%)6Ru8^!33gIID_2rVAsV
z5k`y6g}FO7j<uEzH_JXZC#1d={OAyh_#!@Dx%8>ri~528+>D}E59-qlk1LiRe6u})
zros|Gk`A?GF_D{&!TLTTn+?T9Yh@u$YMYjk2U4e|f*tXy;PGSyh1CilOOur)7>9vp
zPs~=c#7)_tgPa+yg$ZXOaU*71#-*q+Z*=39J2_oRSdGs-ZieQg=xdJ7a(~VBxo0Zr
z=Jcq^y`cWf%QE?bcznZK6MOF;Jz|WM)a99LW`BEjz50)J0MHSqQd~?=O_rGa^Yn46
z!>5kLUa6?5DPAJ&8)g~GuaE~AZ7#X7@vJ%YGgbu)rhK?`s^cAzvzrwKg~SLN5R%Kr
z4M*D5<E)gt_d5^C4o%ppBo+GWuCBHSG0})I|LhCj3;B1ZS_sx~_9Ij_8Ps;rx5;V+
zS+?Oyzv=loyr)(s^&>^p3>~M5KMxbb5_7bPvg={}?#gC(Hp)mqf|%C2zD9CU*WVM3
zO0dR&e*IpUJ%65@o_^kaGuxtWcT=!^;^LMV#1;;6j&7<^f0myORdvZ3muzLt6<Usu
z{Tk`g?C!A@8Hv0sf-e3nA<^M3G}YIIhh->p)|mEZChq3grbk1TyUoz|l{RVorkwO)
zW?t&ZyC2q53n<LVITXyoo~klmP?pux+;`u~BF%eEVq*T|$JN=h%8e8Z^!0P^#6U^G
zn1#W{Dk#K|H=o#p9}Y*LqQ&iL6DNKEHwt#zvPdXlq&{Fy^4D-{-0!!lxSkYk4pTYj
z;^`?2lhx|6+1})}XD2!>GKoRI6=i1|X6nzrGs`~N-eg{u_vF$)GnYQ`+wY*Y+j)WA
z3wOi1R89*s>#|fB;Ix__zW0Om6@QxLli@7;C91sgQ&psAMfdg8ucz21DBlI4@_t=$
z0W6`9*{IA2x~NL?n{lF_e5JPU-?t)N($YfphmXy3jDG-#c-6$GwVvJ7iS&&!fXwV|
zHek=G*`$rMF<YR1IWlg(P};Yh7VA^GX?Y*4?c7{VUf=dExH3R}n+2zr^!Rn`QmqxW
zDMuPkxMv){rg7@f-kGywFPf=N9@?*=N8_@x;n*doZH}5wT>|W)KJrhi;8Jhj7vCd2
z>;82G30B^*m4gtg{nowIR9xTd&x~wisx4*Qt%d$g6>|jw;Vp2H)X_4<LYqMpD4jT&
zaTNXmuMkcQrA7LQV!uoKay|%$q!8|S3-yomj)uF~M#)mt^J79VLUHrM8({17@}H{i
ztlR&~%)utZ73&YwT5jn1pvBK3R=x1g`WN2e4xTS;LvCnBXNxcYX7kV-*5PIKn0c>_
zhUTYqJb3vs<kBNWcMGHmQQ01O31c<<1wf=#CeK+hwBLe_8|6M~&h9OWh!IrF(`G=Y
z!0F7CfHCa)8vfHUv5N#aB_(sV<2-w?ILsd>lj|FogPSrfp($t}cM$tnRh49O|99KW
z;H@P$_69JGY*vToQd6;Q%m|D1bwl|EtCs)3w%ClpADa~chvk8PQw<*2v#`S@$-z0~
z*0Qh1B=R_GH6$XmKPD{RX(X0YD3g_)UGS<)P{B~xo26y-p;6HD+&b>du!q8aSnGxj
z;QTAdT2NHE^@Y~(rHmKTq9pn+9R7K2s@L?6FKM75h|Z+u&JzKs&OVbJ+MgUQB_(+j
zPO7@u2C`2}I@$pql#(CKq0!ZG@`h(0SzY&wHol%xg)T0o_n9@FNVec6<YlwxB$$7p
zwZq=8G6#Jq#=;%z#Pw@CZ<{40FLo~poq}xfq(Nif=Hfr=&U`{CthK**9I#^oW9TGf
zJr6M`Jvn)niV6<rY6^K-zV~GgEY{bz(b0_h4q?rTOAn4Y6%?0Q`X}eww<nHOZ~aBP
z8vYQ_R~sNS^U<U0b<GUy*zqaqit1ObKs|x}Q#8H4%c$`6yX((yvTKv(wvow@7OV?%
zCTuE*4;U~Z61FEhDi%JexPSJ?Yiqvg>cwLqT&vnawI<(nA6zM1&b|Q9Taw?y!YRti
z&IIS^MI-<_sCYq-1k_MEtdFd8DBzTTN;^^PUrBa3G6m)G$Cp%LAAj@tMw`LjjxPk*
zxDv4k(x+plrIlyw&aYcKI+dYkC<~)Q>DOlNIhD4Dm7j&2u%Cv9_0{h+7Y`T95`nWT
zjiv1bZkScZ*%2?C9ko?6*WLNBK0yWKm3yv0Ty$@&P<7e0vY)tkIb0?{{1tAJM~+m~
z*K3gw!r?P#9APzP&g?9FCSbZYP}jrrF)wmKv~xKAHt;%_T~6cepkIvhnCFKo!>6;Q
z2tWWIjY~7U8)pu2>$L9JfP)9o#BO6C#MPA-i)9VS%n2ug&x+rV*+F*bF<y6-%)dg3
z3mnX>S*buhk|K5zA1mb5tbW`yfA8JjG`tKbe9O7~`KANyJa4ld+e{FER59ucM(*<R
zaMmDFr5k9-3T<bibU!)rhE4NMi&zdWGrTC*OEQ0e%nSE9lG(1#M*rWY*$Bgxc(BXP
z&8EZp#qOl+yT8kXy74Y5AD1szbMC+Hb!guM96tfzB*I)Qa|&{(96vifor)hCLu4)V
zYmbW**KV_cYrE#Rv$20HYJK8VGP@c;Lt%zIB@4Qa$|}M^{B0&+rq$e>$`pkCd-981
zT3qv5TtnTiNvnG#pdFew#S%BHeMQyJenSvwOIY%Q@KEgBD@>3uFd!n6i|P%Pl#FaI
zBL%elDojCS85&7PYx%aj8UK9aS-#a?>=WfsW+E1Z#SC=l#2DkYP_%f_KFoT8q_n!5
z{;9+=gxou`Sasma>NqRE?;DZHx5jx$pDH+b;J}M|b+<nkrwnQ(*@|g+5BA~NnU$Uy
zU2u-7RDqpVD-PkS`d_|v>*N}KQ_L%QCz%^~07hkZPCBvVWPb^PG%(82@Sl7J>C50e
zbii>bi<Rv%*PtoUFp0g?YSJyMKD)=b_>#qq4S{A=9&d&#QZb~~8(l3kb?Q_GWa&`Z
z%3iqH<cPOsT*L6ROHxja*5?b!N7GHw!(P%UpUj=2ejvJc8=VG#I3X_<Z6brPQK|Yc
zMh#R#o4u*@-gyb=doX!jS)ayIJF9AcEHrd<cn&SX{w!Wf|8(DBP5(7}7Po_i1v!<{
z&UkU5Y%L}xhVcX4!gTxa=v}}-vAcyKZKMwsU%$h#V{owYjT^$m<TCFpjM%A+^2*G#
zgYkrEP(MH8)ICOtB3w{hd*cz2CFcWzZkRFwB{wg8bGS62msQDm`K@xhF>k|fbT73(
zCi}+NB4?+jAr(}jvr}IzzOMPwhSs{0`HBV;osRUfiZWtJG6%&K){4t!GB9=^M0Xli
z@<;jKZ3r?J(kP^Y9<HvG;OW{sqv5Hf5m~5rCvEPh&$rdBnvKbo{CB(pG&QVGXqk$Z
z2}?gSM{Ul|zgFMUY=Q686>kPCO|?or9e|@|%Fu7|(UkmBQyGtX%KUX#{37`fge(<;
zJ7h^o-AMag=PO|Og=+RmC?~9vggQafZ>Jv-{UbiO!n=O`+R@*xQE!8jQ#xG*vaiy1
zCR1E$i^o3zp6Ysd{~jXmU%Q!78yh$T?5InUwwyAMS8<yv($j;+e2NQdL&}F&=O|J{
zX#CSP)U)>Gl^pggQe0sR2$daKAxH1q>vu|QqtpkBcIhDG1`6&Iysy}o^t55q$s60O
zzY*OmQ(UV6=brBFYJ85!U+AC`9Xts7{}b$Fcy<Gzv&K}{8Q4TT6toE#{c~+UYbnN(
zdr&r#6^f`4gA#N9y^h`dzs3@Cp(1K?|5Z)B?Wdtpxz5spge>9xix(`@i<@Wj`~LwX
Ct3Rm#

literal 152363
zcmcG$cR1F4|39uth>TQ7B_R=+*(FJ`QrVmAkyTby60$DIO1Mf!$<E#tvXiU`W$!K9
z_i@&JfA0JH8NcKE`{Q#S$8}t;tDNWge!X9>=ku|iZ$D+lOM6M_Nr{Mv_Fk5iRv{wV
zb&iN=hdIf1{GW>lb|&Lrdv3{UI1mvX_96VUEqHMDCK1t5qRZ0fuRA}V`0l3tbT~?A
z#`0hzGx>+h1LWJ8?NyIfUbwDIliZ-mKtnTXBBT&t(iLniX<oQyBV?P~5Z!c}(mIq?
zz*@&Bwe4%Nr7!VUcH+Gx_WPuT&we;`Q?Z=9{=RUrXLp>KXw)`FiXG(ZrJmvvP2x@3
z^UG%&7rTGm{Vo<-BaLrHc=418H#q(Ik%(x<S+;N6pZ{PaqIUkr|0ce-lj+Zo<=c+l
zC;s!}Eu#OOzuM?<n5g_!lss#y>J@2siQdY}n`s~JuXulFN=r*qRCGSKM~_cL#DeFb
zFG(J;I>X_^0`!44HZ~52N$N%E10VT1i!s?yW+XP7X=`gQyKia}Z=d(qO>ZAxlr|3T
z+EJ!oK6{XScbHs)%Il7;*F-}dlNEFPL=Kv%;dh<-mWV7hQ&m${o1AEQj$}^n?PlU(
z4U-cP5fN#>T3(Egtt}Oquygz0S%b4pPDFj>6h|bwJ3B8+GYHvqzcX*y$<-7m={erU
z#q2tsnrFf$=rCP8AEcP%Hdn(I=RQ|6`aCHyaj3h%!HCKJ&K<o%M{{FiV-*z@XXj<d
z9&dAVb77D5vs8Dt)_M<+dbda(`+e`uim=S^?5M4)Q%l!9yl)>*o&h-p;Uyw0?EUCb
z<D8ZyKYgIg@etWJZ{Db;>&C>zP5=0DrtVCK4Fxyt?{|<tMC}~8ZMWBp<m3_CttDF*
zu^z@pSGBa<HddYP-1#Nc>)w)P`!qb9j*+o`quGp~f|Bx%latf!+w_4|RaG)k@~Z0U
zXKfvpHHwsfzaNj+U*#X_M@B}HQ&O&6xe`5o&g6o;eEnE!O6o7X;>6u@^YZf2)2HW~
zBzto#+b61b|NcF){f|2;99LFXU0qzLng{D5F3HFYi99c$J$O)EUH#y}gR;_rfq~2+
zW@*O>2a@QX_l`Yc@)6O|UAfj>zGlY8uk!Pq1$)04^S-|RrqjK;8jHxy&aU5e<HijK
z2J+vZ!O3z^ijwc>$&;_s(xxiBh;bb!PoBJb^=fXJ>g(&_5fT2<FDoj}Gize;9UT|G
z<ydM+<`(4jdUTobPjzOqiHK;jP~-h-2Iq>4izg-~UcP)eHfGT#AK#pz`KzFyV7x6g
z`>dZV%Ns1_<YcQE|MvU(D|3U`0Soi<(Q~GGbWBW4jEvpw?d@N_yll1j-rZdvjP(!_
z6eMip>u*{{i%UxG-o4xM<%?>X_UOREGO>k7%v49FQA|vX7?bUAW6b5tm%EQ?rM==h
zVDXIV&r-x3z3&qh5*N2sL40InakWWu%e}YsZeye1cTFxXF4Lx1fg{e`p*6<5os(U8
zLe}5jd2Fnn(Yt}0{Nw$u&>H^Eqv<+Dswyh)H$n!JvC3}0reeNc@Tq<r96UWf&UcU;
z-zX#`#44%TOw4U*B1ZI1y;jyiO)ahC$B+M-`d&9Y&vrIdH(k>1=gzvGI};rSu_a2A
za+;AI4}UMazNBoQfeDHC`@IZ;?z5E#R9?R;F76KB-(HZ=81wu&_PErA3;3b1sAznA
ze9?8IHAU^NyZe*SP%hopt}ZQ2&9NUpLY_SNIXryez=2EBD=RBXk{ga#h|8D#ii*}4
zY>UrJ)18=Qpr<$H)i`(VoQcWM++ba+g_DAU!nJFm_|^_X^|;@r84owJZq{`?a@nqT
z^=gH^k)B@f!h+)q<xl;y1NJUDIx*KdtZi+%Po0YHD?CL0D3nd<ai!ssOT+lCZyqPu
zN|jpz_Wi!13DP}c^^wn?KY#RSx1Y3%u5Rqpr~I?|j`7XTD|0uQ=VoR)O5B7Z1``tq
zub8uThB(F8PJVuVbULLsZ{B?I;)Pb84G!2H2M4V}$CFgErSHE52!GLH4ZSEWJ<*x%
z!_ZP{SoydmNvXHM!PLx5mNj%`ZB6MHwfn`R`!e><pSBY`q03HNI#+vYYHF%`v%bFm
zf)535ai+SMuB#P4y;6)w>$h)m(p`DBdf6s{E62Lh2R<IK5c$WJDKU^aRew+-PET36
zynlAV_3qWH%^w3b-Y7|rx2F%=E3)$QJGLY$NWW5Mx3IM2(lz6!&$I3_iPin)Tiq|d
zvCu-t_vPBPYn_jzUPMmpVe1<ks}=leA<)HhkX$CdxvR@<@>>pFMNwg4*0%NjN`2la
z#W;z((>>zMwG9mo*qM?dgQv_5{_$<fInQmU<U7gC{2<|@VP<}5smFtG?04ybp&^r=
zsQSwa3X2O11J9z!_Uze1O8QL73d`5Vy)Zj#z;G?{s^qpb|LCs1`4wVHru83`p35I;
zl+^R$^jE|~YaFM#3s9w)HFGS&oHR;L>Rl$g7WC&Hnh-zw{DHi#uFlfNhUw6u0>@Dm
zD=W@Arqr52d*-?M`F@d<24PWA;+;Dk9Uay0pfas0a%G(HS!n7jzC-9T)&W8)_qgsQ
zDYL&)c`eIgVPQc(eLdo|j)l3oK^G;Hm@-G0RU46Mt(v`I?j@4n`z9MTtDZ+msVP+}
zFYED01h?L1TVAPv319eadxy%@Jv$!+1yQusGq#&bXk7F7L7U;-?fdEHinlZ~p<<1U
z{8eKdA0MCl{7piFcXj^>R#rhlL5rMkKY#wT^<3*3+MDQpqMF~Eq$1Od|Jj}1yAi(l
zlckAHzV;u4k1uB-)QN~^&otH4+IxC*)zt7aJr&g%@6e1@3)hk!!?HlKuNv9cOxGfr
zB-}p<KJfA?h!L{wDYhSc+-%0eED?8vk&%;@bYo-V!~6HoqN6po30sOWrT_i+(c{OB
z%*>Lrv$J2kI3_NBSH-_4@Xm=;1um)I*ByCSwy)weDJiMM`pgBd3gzq9-`Vy`jPI6u
zaSJbfef?s-jAOkt5m|A0c-PSs#>h!3|4O{1QU3F9)`dq#9-L>RvbMJFkFh|(KXmAj
z^wYRF&yml^aO5`ESMIJY*;THXj>#zs|ML^s4T*36n4T{4*l=ZjBFlm%oMCg6mekLz
z*zVhIulqMt?|j@@nv#-|m-l;_@?ZXX`zkIw+rhxVz}>yL^6d7uQE}GX&*D_OcI`^l
zDGK*xp}u_c=81Ow#I%8@$9EF0>t5&molFxqzZW`jnsah-<;31QOh+dpt(lrzSom!<
z2kqs<hY!-U^z;^6vZhZ|QwGC7*VWe#ES(CIW2PqAsjI6?$9*PBUR+!}_qjVx+r(gd
zy513bdP7lMNu%t)s%M4n^OzViQc_>p^mZ-Eaz#bOrC-0;UViza;=jMq)zvjkr$|P6
z`~7=?OmR<zCr*XSC3vp3#4(AxN=v7xr3=QrZd|R|Pr9ovRM@p-^Jk?$r?|(2%54-5
z7Z(@lS1qF*x9&Wv3VWBY%tw8UDn#o1beWRqZDo7jgv2k6F$>L?_{G+j5B%;m<qvs?
znI_mdI5<w6xLLpJ2g>%`+??X`voBOrBsZ7)zkdCik-@_h-2VN$j+Ry(N1AnK)=iZY
z<LWtrF^YH4EXvEz2?+_=+D=EUat=u|hsebL*hF#h@$o4tDylR*Q}@`*%WHmqzCn-r
zzyY-zH*WGA3}O;5H2ClcznP=CG144gTPs(fJvBClZuwO1CaRFUJh{+JkEWblE&7|N
zs3-#iLrY5wPD;|P6V=7ad^A5ojxWR?30b-4zv?RD@Mv>TZzQiQAx1RGvBY4BJ*;+a
z&OSu;qpFJfe>DCbJ9ZQZdal&zP4$*dPfV;2McY2>RpX7CDEZZ6)Frt&8{FVypWHg;
zxt8b27bTCvRCH&0XbI5ff9T~kQ(zRp=swsnmcR*U%s9EL%I@y#*ObI>Xlq;X9JE2#
zUR$!B`m3nO?fm%)U?#RjUG2*JvdzVG>8H`rMK<3H%|A+{J^THd4Ps?wrK6>Fc6N@O
z4hRg?DY|Wa<3^Y?^U6YtQo)886E~JZL}Yb&wqJSy7*qPvr2}+yiZfF~trNO#lL8WP
zlAWi7hrFNNC+YQw#+p=7tp2$0EyusdEtrBavM;`(n^4Ng&e^*e^8&>e;kzX&M62H*
zV}9)CSKzr-s+&GNH5C*PFhCzoNi;e!p^YlLa-?%6aQ`XQ3~`zGSv0acQ{87M@7<hB
zO-%)a0J0e55cs{5ya)?X^W+4!AW#j%yqT(M+vEMGI8L3q$lUz-GfLgu(o#@RP-9I^
zl)l0I?8mCAp`jsFH8q?bRL~Pw!sSKXwE()fbPD--d3l9}g~i0&>ji-hGBY#7!opNu
zJ7Q_@2>^H+8XBB2jg5_I>FGxfAHF+5Oi4;wTVMY+Ir%TSg#G*X<CpnsP6_vTKU00U
zH7~i)rbSa}&D%`rajvFj;RX}Qyxf&5WNtG30|QT=JfWti9~&6>iXE<}2Rf3jTQb()
zANu^cyvgN!bxtPk`_m&MJW73CU0t2Ke|>xWm^8d(ZFMy<DT&NmiYb^aqHcU_?3IeS
zo12@4MrWv!gn2b30iX8Wa~4#v*cGX?;3}>q5SY;5aF}A>=;G7A&gF6n<eBX4bP;#C
z?Y6wUoS&cHu^7nsR3_f#*Y`Vq?59uLG{uS&(AoOT2g(!Fem*|cQ{RgkTIO)VY1kCW
zDJk*!rJJkcZ`IbZe}M$4sHnJ07O2~wOfD=eFdjOjk!uyTvKn&a?!r}DTibKz%I~K?
z(qCU~FH>cwm3lD~rlxCm=gzR^m}4POhDMybO!9RO!IfR0ZP#93cP;TCZ3;4|2$f5a
z9+;R2=-To2a)JssH~0M3Cn8_jRMl+WFH8C2JT?1>5>wySCr-rpzDargqN1_6{FMMd
zf5zZ$ls((n5}K=KW+T{PzOpZrzqYky#k@UUQdAVJRjiz#0%XHn+t|qZ@=`F99gXQb
z&B+fwlz!}jg5i^A{vC2`r4#^jj0KZ=zn6&Hk5w(df9yW-J@ZpsIci3c`>hs=z!;th
zSiTKTOiV=OeA0Pz1<i$-_aM2Bj?R$j%<L>s<uPWgk%#a8I|Efs^DFCK#FU#eK9cS~
z{n+02Ho8+w5KeOBh3%9?ElVTKEcCw?JGm$#nG3WPr9C`6KyKSE0lw_mzWvi%H8Bp3
z$E|V0ta;fxCB($KtE;O4{@0Z+T_UC`$j=W@V|s7Qdom1{)`&xWElE*8K%k-T(hhup
zx~%d?3ZldYtBQ>33!M*Ed``^J$WPOf>^yw<@LW0_iE(GXEK8PRpC3CfugVMMjN+Km
z+JS=SaRj!Pl7f?O%zH?P_6Xs-vJz;2BG<7B-9Q5m{gf<#Y?~$=HfiRoN=m_sNynK5
z>%;-Se3_3Qx3sjxZs~YYS}MUCP1Mv~>@r^)cFM|3Jg$k7f`X3JdvkLWJrXeP#_cpm
z0<P)%`>*mH<?jZELsL$TItJU0PAU`>EK!oMUSg*O7tN|XT=j8x)7r|4om+iU#k{jv
zeDlbU9~Cm+vU#0l;zJH&?X}%xIOzcHX|C$)_l0w585$baiw<{LarzQIAY)mdLX1dW
zOaEw8>LJ09BU_PT;**Sf<)10Dv%H*rl*K^4TQ$B%>^=@PY6k1erKQ_~9ul0K0oDCf
zR6=`8<Kw*o<9^N08`)NTZEvRu`B#_!?ApdfNm!va!X?NrS8}tT96e)gZ$H8&85L%x
zZ~y-Ndtcd<R+JsBBdIAVMwU^J9zA+CZ5lIXnAzdJUF?I%53F>DymQ*7Sp0PkKiQ-E
zoHcGJ@%--UEnB~(88T|<FOhq6bsa~kZ{7^-Iw}+)u@+FpVNc<3T3EQVw>Qog<)t9!
zXJl+_tYv!|Rmi!cpFVx!k^J~&4lpg(kP-JGV=^|DDLhb1{+?;cEUEW{n{$jMx}LK?
zFD_Ao!Fe4afLeWnaz=d8v5Jc8sxn9JN|&wd&3U&zz1f;FdP(x-dnx<(G#^}QYidq%
zb5DQtqp`?Y?GKhzR8Y9t{{zhzH(q11kV`99J-{hAD99wcH&7-XOLglO&mG^z^6Ud5
zqN35mOVYIeKl*Y2P8+B8Zym~qum??o{K*cVJx5IOm%6-;&d2_%#w_064NmGa&a<<#
zO5%&x?d&ou^z+ea(8&ufpnd^;0m9@uqo`eym6g47M~KS%;ls$JU)%4W0~31VjIMI}
z^ywk)|K|?)?MFYM0i9)GnIY0AzIv5*zd-x>@&V(UH&fL!UuI>guu$*ayB7@)AO$)G
zSl(>7uU@9%t}A^FY~_6TJODdmfo-uhO-;dpfvIobK9x%V(7O`O>G$|?ZEdaL_vOyc
z&c($=AP^c(1;YNyAerT9+Q%5An{F!Z294`V#OavvTlyXbc|O6x!57oH@5v>6G2@qh
zh}!x3I!B3au&Io`0!J7%_9`DlFlbjL;LP8@4~y)J)C18vMPwD+^&!NyuWb3JfA7ZB
zowAh%TO1t0K@<AGA?O8YY8NhCICJLA>({S~ZcqN4U%6)aM`<DA5oJ7izs!AwcU&yW
zp6~Vb)%2~^bi>R@Uw3!+sh$$_=<gsqahnTCLx&3l1qrnXuSkKQtF*4h#sf1yKHQn^
zi80B>Z*5WgSAx{~P#)QA!7mi`Zk*=ie4Cj$8|=B>Py6g20%ZNOm7X1*gSBBle;Q`0
zziqWh>YUizSi>p<0&gt`Z+#N^Lf}H#aT`{EK<#NdZa=&aY~Q}!%62yqkJ9&o3;Xx8
zggwa<I>B~Lk-}h=u%hJ;czWFkJr->gG74l9C|8cpz8AZM`nJ8*Fp?mbU}tAPcC5mf
z_r~?>@^20il^0*rmOpv&=@Y5+Qn{@Kr3K2Za|}W%&Sy6^e?4?fAY5aRZRxM_vh}$-
z?y_~>E)QrGK93$Hc27V%kbW8)Ym+G{E?$zE8FDzM1n;*#Hy_v1<RK{YvZw1xda+al
zcT1+xz?jH^5W;nr*GEo^&+sJ^_{I9h2G>$HaK3n`1+FOAeSvCYUQ+36!)ANgm(7jr
zgTE<heKD3v(yK$$z3t`zZZ|>IGFEA@Iu=?Ze=yfVNi+5K&z~e2W_u|p8l6b8t1mOp
zZZrp1#Tb5E=|4hDMk>>u8twhpP{BnfxAIpR`^&by9bI(&BYZ+!8WJImOtLB#XU)-G
zXCAu`7<7V}l9JMY14T#r<KTV5jgDmNmfVaq?NTBR?%izbsUWP$y$#NnOv?#ZA}Gko
zlUptPH{M>08Mn{NjN$?(Y2IN=$r@Ut&$BNi6t5O$SCU=1`gz20O>uuu3~ukgz4#`O
zOQ){(`QF|34%wDs^Wl&X=>w(TC1X>E;O8iLcg{{CRX`r9S!c0sCVhqU_WogZ_2>zu
zpv&K-{MqkA?F1NaGdkG81i9w2K(=>qaMRZh-m{~NtM~73W45=mi~4$ezdCux<d4C@
z!Jj{Q+cWFK8x^&*EOH+FFd!w(GUcC-_;bA8<>O-m@$>j1l)1DW_g#DfpAXNk<cpU-
zyUqcT;oaRCK2B<joM#PnOEWW(vt341;-?MuitG!z4a=ln9A%c2EW0d#h2kcEbeX?x
ze3I_BhThT6EZ2t;;Fp}1)@bN?_pUhA#fuje2%FBGiq6FHibR~0)Wjizw?Nd<jK9mr
z$LE&}l32X&`{&Qf7cO|Eq>NVuGO)Fh6X}<B6t=d7kGC~HIk7~(i|{KGW>hC;Pz8g6
zgVoq+2`62B4}Yra+Emdr0J2R-hCz**q=mH=ww7M;9BDz}x36=H?URiE{5%fBOVW~%
zMtYt51GBQSAPIqP`Aaj$DCW7Z%th$Bnv*sV5e<1q9$Xz&8r|CLt>$+dJGr`)XL{l&
z;qx>6FMi#jBwbNaL0}?zHs7JgHMg~~?jwnH8H(CkZ`yj7Y;0r1M|5xHr>SIo*+!Qi
zy|HP_Z2!ZY7Q*i{9-<@*^ZM{3JjB7l;ls848Ra&VM~=i*<|}qgdt8!SJ9L>vuWNKl
zQF3`($s!%s>P4IUn*(&CliO(C^V;(;<pw<e_bQkXz^L9A9+F)>nVJ9S^9~|T<F%!$
z2FG-YW%WERZ)^;VU%7mFA=NgdY)kvn?~2Dqv`37neEZMQ%c6I{XB`gzzGwBtl>)^7
zG(*k<Boe+U5z+sa8S=klvHZ91+9%eI0wmw~<AV<vpr2WPRbaHB^|QYRSLTOpzZYJV
zwk=s~``QB%sx{mc$M$k--g7fJd%k${z19EiGB_M=EH>sjEADP)e~0ZzTmc1PH|leI
z-a$ze^rpC2n>nN&_R4gWJm}@q=<&g6Zh5&m|3C<B&o?)xM>jVLf(XP$-yY!VUOZqP
z4h3tdIpN&Nb0joGoBy&YbPZcr-TLmb%yZ&IbI~;Q6alZ^V#IZo-*<9!{My>ufRE7I
zySF)1>C57{(UI2i(Im+P&&^?ty?Y4nO;3353v=kUb92lu>ms;=0|M%u&hl_?^btG$
znlk!UJWruhBzUSBf0;=B0Fc-%Su|jXNXK8Il|vvapfF`-4FzMtC;2N~Av-L-x@^Kq
zv(X~iDM^oCx_5{t-JfFzk^Fya9NczveD~(f7}_Q{a(X-2&Ig}Ae}*+sUS2+{%L_(|
z)dz2dtpIrI>A5v9Fd$&rHsdTdZ(>`vk=tz6oo%Y5t=+w`_UnqA-0Ire{M=ktc6Nc_
z<m4ozc$fklVPlRWU}dO&@7}!&JEbNftRbP+YLNqu%N!zmIwGaj;{Lh5>FKzzu>FS)
znb)zkjyhHc9Uh;U7?^byy)z{*C&w!(@>-U$UT;fICSHN-6i0G)==%?ojD)sR?nv1C
z#Nq$hJ`hY>0J(|puMP(6w|B!g!wVP@T&>Jy-ef#t0dCG&JE_;#vn0nBSHXJl!j-}u
z0I$+48vulB6Icq6>F>-oo$f9$Dh(ZM`j-kYzta5zzks*L*S86V5+ut4;Gvamq5$dR
zW8C%hcD(z8o71AVCwQ8A*ckl!GfU1ssF8MZ`n~D(HwpZULY|!#_3wv2cPWTG%D<At
z%7Bp`<N*7DPgLR5Y%DC&7caUvJ1;CQwtoEzSrnQXb5`RIrC2en>;);QpZ)#2wx7rH
zUc1(UQmm!b)f|6G@9yFae~;y!o}MpXzCdXJHr(U=9>P{`*~^y?2J1^pOGigXrC$71
z-48FQuu$an^>0Axs;X6YNyFrv<_0e3Xion8`2_MR9o^8+pUh!$DR16@Ie?hOzkEqe
zLsPx>YPZ*qftpYP?t<D3!>2x2k%QXjm9qTHOl_OIvxNO>LQdGPR!}E^m8kt*6&H)0
zJXup+ZTfQwz5?1*bw6}=|2mJ*Cr?n=H`j+H;j#7R+a3SSD|**&Kum*J3QfdH-^Jd3
zuFz>*F;=XzwRI6F4#K$qi&3}{pmR;)zn-HhtZcq{!@*2V>YZLZ_qeD?G*6O`Z~pU(
z3)tJ7bPnbYmX@cY#kOLm-Pul`4zdI^Uf*1s23Zgk5D1JYX9%nUgU!j!rEbQ}ZwCr%
z)>jER`zn2FuY}CGd*6Q&t|{^ig+wJ45uNydHXn$0?^gf$EGo)8WOa3U*?m5`7uy#~
zZO)nD@$tHw(*Y3?K!!yfN9V^?$-)i=Rwc!{)-t;Z{Tuxs66T{K@PFZT(rk}@pi*kU
zk`@qMPN{~Pn#+()7Z(rD8$FXxBp542cY=NG9UK~xdX>1SeV#^{TV7#f%W^Fzpf?`M
z7ibQQy3*3p5caLDQ`I?39flQ`mqVkpw6yl_-a#ck`;n&Y5_Iqf4<0Z{C>WPQ68@6~
zaPR2sG`C|H740fajt&l1l9Q8DPzaP~Z875?in=8$`w1Qt=eIw)LePQeMMGz+7;s|s
zYuAjLy(P}~=GPdD+?mogGCEF`3gZIua+@_P!=6h}TAZCb)GhUcCcj;j+&D?#+ELM0
z&YuU@##h!aBot<{^SZiBDJ*veYxmDTv896sKm~;j&cYdn42}gT_jGI6^;K3)3O0Eo
z=}hUi-Op=`RXD;RSU`G6QjEdnW|?JJv~sAasWC(6T^)%J2B|45EL?Ei_?T1zpTydl
ztIK1iZ=0I4tw;zD4AHIGBU`gSe2zfTtd4b<^7d&NPnRr>%H}JX588+)tY_r5pwIr>
ziDu3MZNm8yhm8n|+Yl`TeNfnhk070(&ZXw{JnOUC?R7q-wXmST&aLXoa)GqY=?G$C
zVuv-h$j99Ok~TU)&2}0rudImw>UKb1u`eiw_(}lV%kCb~{iLL%R8*c+?z;;Q?!WO$
zS^^##Z2M%Kh4T$7w>me@ZEJe#>W%$jVG3V|)XH841ay-AUQ|tjFt9!3pQee&*5(GZ
z8c0Utf&hGiv4Mf_JUMN&OZ@%)S2kyNNA>>EOQ35IzR0ax8OrP^l$2y-u$gjW-3aPq
zU-|vEwl<W?++4u~6?2}0Ful|{PoF#qFlB0JWW=TWzJC^G391-M23R+Y%`Yu2o$r=L
zrKO}HWun>1#Am*I$;`#o;Hh%s=g`BHQ-D99HPu0Zc3-=qJ=U~bXDc-*$q3uxG2sf{
zUsYC~!5yL}e?)fRbn((R%eKAlckjBnx?-V9f0axtjMW2XTgepeT2ZSDM!^tYVu2Tz
zpC9SFeaDVGt4<UxquAlRoSeWF4vDxjd(gR4YFT;XBP02qTOJM$^D{FuudgNPdTuO&
zEIiBW&9&B&k=ac(I6Qnr#PO3Ef2^dZ2iV<^f1wiB6L>l|jg7D0xB;7Wez*z3w8YZ!
zCUGSt_!efp5~~~23TkOHOcMMB`O5BMw$&7?KAz`hU2r|OI3o$$+^4L}Q&EwE>T<#t
z*z4*-`sTS(FCe=Z#e(r23Y2N%whjtp%4=(BxgZU{q(3lINKH>qNg6lL+uIvz16|2N
zo2>{B;iWH;pvO*&jUv9aK97ELlsQ%P6ty3xZY9^*1GKd7Zcl0wEF|>X7QcttXnv}e
z=UAw}mw9ew9&suCRP*BarAO1f5fbj~<6~dcZZvn<9(N<aNgrt3M08JW-TN6hBTjR1
z4Hi}l1EH`1cu?hje6|Y^zD{p*ZZfaeq@-mCXYFmvH~43&r%&_pekt~J6MGx`{M^xu
zsd}L(`3a4C1;;N1t&gddtwP^VRoZ{}aDyN}eZ$V_eDfC8CEY*P(KCA%MoP+H9IWg~
z+~^Kohu;v_gcOZLT^Wq6liCtapDVjFLN#V)>{NG`I!)>BwnMB(^u%gc&bxQ~{s(FY
z=kUvs!!uSpMN{@P4ZfqHA@wGB!JHf)#ur!p4{&F_>vT6G((+a0r2DnFsPX)93SU!L
z#gS0-j1<A7meKk+Ie9^92lSq|tXqA3efUCbtL0b|BA)nVnh3+)kv;)3$H$!<Z{HS7
zr=jqdwzi(^>-#$*BBJukNI<ihqm$E(l-8ju6?Q3!^UlRUQ{TRQYi@3a;f|QZbq@3@
z)M3fBiEM~{1Z9{YV@+Z^BbCs3+txN+mPI9{6$%K>3P7%@Z6+$({Eb0iEP2*z;ME9>
zT(p9B`QiokOHNKs=Jq+{5()}d;90<D1T;YzN^04)d$+5bTXt7;PCC>-JC3m0rKKfk
zbO00ku1ZUPL_<YPiK@mp3*U?xb&@w-vCOKIH$;~I?Aeaq-Xly*jkQBcg4F&>6=}zc
zBx-7EAZmrKoqB#;Bqm7w7G(9jn;&;_a)`t<n(>R&HjrogiFY{J+rOXO)+Qq(qphP;
zouW<V9}eXxo_5mCt<T0wI9}y7oS?;#=Hmt05<EOG-^T_A_hX~FE{;V;M!GD{ug2Wz
zX&!Y18e)kc#?mM>6gUi5i49Cml{$`oLFIk*>J`rGXfYXphyD7p9rNnuud=2>tpi~Y
z0N>$|7aVmrGcz+V_yP2W@{1Cr{-tilxfp(dbyu!Nu`ztT6xCFYlP5><@nd7WlC(55
z^Ywg99S3h%Sxq490K~&W?YC0y%oH3E69dt@VY3$KEEI(ss%l3^$KrT9?l%MZqu*~}
zF`vDQn0UIUB>I$><hFT>+vz1s^w1HkSm@Zu$;d$JUjH_4xK5w;z@D`(-ah{k45#4y
zUSbMH66bE?Wo+477kA#C=m?38ZOj#*Zz!5By&??((MUp(p!h{<AcsP6Ft7^u?}n->
zP_@l4z6{nO{!%c40yJ1b{89Dhn^-&GCcIEMte!4fL778}ia7-3Dpj{cL{ieT{zU1p
zpK!6Vpk|iwWnJAK=in>I0>M?wK{0xlo6A8<N<dNP&!2}9X!W&l7aLm=(l79^RN;N+
z<w1BUf^($>XGcKb<Vkat`zVvu_)}F?s1rYa`~c_#a0NEY6c~W3uAy2V_OUDL<`qN3
zpEyb<sB3Uobkh%z>_pMSTS!R6E|y5DDl03Sn3!C==zX0dH91+|X_J+~&l~*>If_7;
z7I?F5ZMjI5T-NbfiGTgt9Mpt{pHMSZ<%vu-JsGe@oGit&Cc@i0nvlTK@V<pYk3H4b
zrw=U@Z0Y{}`>;8)9@YXtwyNnJb}&C(qASV6b50tRijK@op#}F8`~X{gli-+6$IUV=
zqoRbfo{@>Sb9T;$-=3Wv+K>fX07Bt&MGD(?rsdhwU)_Hz0!RG4I}NG`5?(_JB?(D}
zdZLX72$0oMLsFba?P56)BJ5bt2Wkk;Y$55<x9Q8|Og6vLrLS_GW53%u!S?&*#mB?s
z{62Md9)dRoLka?8r0zDluNI3lGdxtC{Y&R4#p-k0+El|!knEWFUepc=aeZW{A(|lj
zp*W8_76Ka*{zGu-j0l1Ukw_zHP2(l$wq*5(h7HGgmSUrH2JbDG!Tp6C3|vL+HvmFC
z<W0o*05=TQ==3yHCzvJu0+tpQKk2m?fi>44jMfWI&~p$fp%=%3f3&88>bO;>){Ptc
zyO0^mySSuCXjee|Zy%~FDN&x{$aUVidp8{&-R|AHsjnIt#V=C#63zPn)&gR8E@8d7
zNWV3CZrip^LPA0sjQ9JVXPL9}MvYW`v(C7Bybewg5E!y2>o33dD5jSw8W;?WUiR=9
zm{RP@Ye|ax^0q$7d0~W#3Mg?;VR61oCEczMbUTp#d+=a;ZiDy)QDodpNZHiX1S>?h
z#MLS7Llh%ZaP<SzyAwuZ<Ksr5uBxe;0)s+^CpN6(G6Xx9@Oc@|`Xn1a=ev5p1bawp
zPHs&#Ta11=QRwI=Xqy<yX1-qRDuehYv*D>z0bpqZ486;{siKr<hhfJ9lub?9!1U{%
z4L_|zfB0|-I&hm^!`0W<w>GD@6r~|(XECLpruIWLKu&t?*EA(LIVT6lPdRtw9dnBf
zZEU7cnBVHDBh}Q^mC|gsb9)66VaR}-9n#)7NrvM`K`}KxZfS1bY>`73?|@B!&;TKV
z#{u#T-Uo(<$mJ=uw08+d3lx(83e-(a*@COP3LKJL_5!wP=;}6xGai3=hWqXBB;Es~
zPMG&_3LQmYB<g5tYN9)6U*g{bc^q1NFq1gFEv<OnBtTZR*qioo&&~O`&{jrgRe}D3
z;aLBmAeb07W@eu)a@-{VMzU<l{j+F8t(yXZf>dwZ_+E5Ji8VB{rVGg@>^QpKhXJZ7
za2{y=r0ol|;jmsx1}<##Kg_!<h_#Q5`>ULr7irFg(6|7%1%Ga{ewo#|D_QiN5?<@;
z>$u37nOiO{1x~K@`9($0gq-Rx{7DMl^v~iYnEj#OOYgcNl7i!$cjdEQ0dBzM%X@xZ
zT1t7BnQ4bA4O5+OSq3-?*+8U6)6;|3ysi1^VMMzPHy)l3=X?Z=i$()GK~Y;4%E-U7
zdDDUSijq>d&SJ~<E>{Vq>6Ix;AK=zq@A*|ae9lh;Nus>GUtr_l<d60;yk*Fkoxe8j
zf#mO{AjZ+j$(x@12g#}Zpc(KB3YNkEY3n^0YO(c4t;n8uHcB|I)9I<HI!OQEXg4q*
zvU*t3quc$)#oI&hH&QjSA2G~*sK~>?C@=Qj>2c;Y!ath}XRMz+8(C~09l5%8F(f>^
zt-1NUv|5_B%S_*SXd2qu&jU~VonJeK1BwgBtpJ(%{{1^324XcxYv6Lw->>;{NJ?&v
zCdC2K<Ao0I*&it6`{|{JE`(lrRx%QjPxbZMrS1iqscERf78bnpfu@bmg`H>m$lJR5
z>G=4R`c83hbfg>n&-ui^6-z7h&$`ULQs*>@6XN0u6gBoRxc1K^WmRzojb5_33<(7X
z8c%7d?h9pt0MgcGVr*RRB%Ny*KXU5SscQK?jk}_MvT?hLsta8g-@xmipRa3dM159G
zX+>d!(<5+@JV_}|!g=QJ%20}~F6e8F!uAhO3?TD@O@Kfvb^uxsePH6dcTauu8Fdp>
zV#HijQdNuc^Is(<KJxa4(Jh<s1rcN{8G3zka`MNI(x=JY2@R*-RJQLxsOP3BFa43;
zi6+m)xb=ti9!(8zpFX9{RM#dzb^<%16Gfu=;Y0L9ya3>ZMT75ZA%cvGO%sB=1BxMN
zZ3nMCPAy(+iq8)+F_mK1qG8gyIyvhZcrs>_UHy-ErCWE#e~MQs@T7QV<>y$`i&k_z
zb$+3|Tze`#GF>(#D#`^hGaLyz|LV2#|97%d=*X0%hXK~8L7e;G{Ch?B7l|~-&YnJS
zAOZouSTUC;zG>P8PyYI=Z*)|+{d}iI>zf-O5tK(Hp4J%CIT)z_DvFEK)7L-Re#S=B
zTZ)Q_Y4bf9_pRQ82(W;sU|%5>%p9_F*Di(ByXNMvHB*OXoe5^1{WS2~+qZ8ew^o|a
zdQjJ76%|c-Wg)L|ggy2xosXIRSXpUj8}aV~jUC&c{Vq1lv*WNq;p*C`Kev^IIsX^7
zxDM`pOu@f|Jm-ZV7XoapjqMFY5rf8b#B+TXFv@LVq#1vt{_73`8?r3a%y5x$|IbJw
z+*Szs!K}hg3lP2Jzm9#Mdyuo7bSOJ7w&It)lhaLx?qoHu3VnY1l9Cb#IF^~QuToM@
zG3biwlCy@6E{wKtGSvJd6k4l80EJrezA<0DV-z3ujZjllgAn@>G9*71f|wzvj($nt
zS3N13Yg#)9<gM%BQCcaqTS)A@P0n`Tz)6CJWb-^KDr$L=EtODR)`x^##9}M-`{eW`
z+?W6Ig=ED^I=Z&J3>}t*jSWc$d(l~I>_eWW7sMXT$?0rf6>U`l0!!rHzU=72!on{4
z`sdd7=k5*z+#-~m5j2t}i8aoM0h+652~<cRB_t(9|GL5lr&SO6=1-qwO-++re$A1S
z?%kU+a^^U5F8O|{2CEvgv}^wV$f+`>mn`o35h<VOFx)60C`jG>=0?ifaf(|9Zr6{S
zta-=SKg{5P6{^EY2fBqLgOsuj{U@<%ndV+JHiUbIvWsr=9@649s@>xbMd?-P87r&a
zZ{KR=?n@AKDO>1{h=wi9{J5|*{md5-&=(B6?dw<db(-8MqUzs;Pt0ehM|0$LJs)|K
zytAb;aMMGgq*=n`kiSCJ#;8Pq448%(l+-KSCAiVG^w|{bl9G}ThQ@n6u|WYk0PB~=
zTMuP_T_<mJ6H3~!h$8ZDJQV~<>GgG_6m;_Tv{e*BGD3WcAQ|0Pd2PHA$T&}CxiNQA
zPrd8-L+^^FCT@s~DAVAe>nrofB|VARbgIw4k)n#mG!Be!U+_HTSEJ0^<<-@o0=h?B
zRnM>SRS0BemdLXnXGYG*=4GHa7$WpO#3HjL(=>D5ynQR5)PlYXp2h+V7(xUwaj>w4
zxq%16L2`b6>7Jg4&!8ssEu&G=hiVRU9xb=A3#jWQyr6krJ00N(l7WneK96?(tsic^
zYfP5q9vVTvqQef4jf`Y4MT%U@Ps`9K-ye}1d4==ebg7S5X6?}Zu{RN70bWy00b|)*
z$}6+((Kq3qpk;oz%Lu?5g05Twg#PTz%x3o}uN6*a=J$1VR}xfEO6@LOPZaiuEgf5p
zon4yHdnuP-SeF}Et<RHjk$q!*J=Sfi0OC)9c28GV9qT-VL{ONA4<EjI#X>N2L%IL5
z_4;A3+P7o1PA@qvC_v`t{7HeZSkKs{X}0WOI5K_egsV0;ADECAq*KXl+ULkThOT-e
z1~PKhxF3iM<T+u-k-x5UOpcBoVq((GGWO$QGd=M%QJM3bOm=6?r^Hn15SQs*aFKNF
z0z-a!Ki{Td&_pEM;BjPRWPowl)TZoW3RY8A{wpNp3_t%@;1<OCVq+y<U!1lo%edyh
zkB(eOD6hM_o6odyUo%7_D?NQfo6M^VP!v&9Z|+&|>pObIF)6gXt7c{<Bamyao<j8F
zpdg7`edl0Az)RH9idK0IEMPUIa`|!%Uz6B}u>&3N4NCWNaG1599vfZh72mz%d#OU6
zZHwD<b5h)s{<mQ`>-rRc!pCU8<XSJR3<ulR47W^uzy0`SYU(6>grhHgeSP<cywlOv
zhB>7EhC8B8V>>r3DXtFHrOItTL#y!6xjU9aSD2eFpl&U+rt9Wj`Y-9FOXay>(Rbc}
zEN6<SLr?ReIBfVK`qyMvb?aE%YckP65`$w<q=5iNM@E!iUkBGiDRDR)q5v$76=-d5
z$8pb|x&x;aWUR@_Ec`zNyZq+Mxd^-io@y7^pP+&V0UsDJ7|B}ffH)DSvBh>hQ$9<&
zPkHBWp0Mtb@4s7f$MN%{z1KA#GeQtzq<Zt2n^kg4xFBDww6k|hJc>(zKRe&P8bzNb
z&sbpclNGsebhJp1vb<MC`D3=O$nrW9%h)Yi$TFEdlJgYuxihi<N>Q!f=wBw=X-{_P
zzeJ@^PdvpDe^}jb-CG*7*y6D=7%_3Cd;I_vRRt4m*4G$?drGphkv=WUon_MAQfG93
z$Rl{iPha3^2b_%8P%x*rl!2z^H%tqa8C6p-f3e764)GP04%)Ql;gu@79T;5_o!4wO
zaS%@Ly}L*6>&dHJ8&xZFm5&Y}M?57<j|8-JBBC24&Pn9EbHgJapCe+<=cLgav=vUO
zhUliWwguc^DN9QW3gjGo++&x&GQb$$-0}G{&tk9vA{V*I$-dU6CMNc4oiuyzpVJq=
z+GfnV7#U1^xEE2yjykYi34Z>Mx9t!74t<}X(N-rzikg!MZaF+6>5sMQup<TS?Cs@b
zWskm0Ok@SGiREehFgq{(ardr(fPjo7aysN0xsf);s8O`jJ(}i_<Ci=ZQ}phxjwZDQ
zB1P2sXej0S?PKl??laq*?nstuxQX6f2s!)#dpU!WWyZgLbHj0gJy4BfG|F_upL=Bd
zMvF<?mxL>wKQEAa|Ac>IwDf^;?_La|JZOy~*hB3*!k<5{f5Q*`1@c`~*{h_aVMsR8
zEc;0EZ-z#2YP=^siNTZVnZ*+eurJglzeEmYq^om;ZH_8=8v3h22KnZmDB||l3uX5F
z%sm0R>1EEd%xmY@I}#?4Qi6x+E1T3}S2kThhVh+RF7e6$7Xx9s0~RvO54{cQ!nv#_
z>cN#d-2Yk$P8bI;z?1>9Ac#c?=vCQo`uh7Dt<0Oo-K|=D&i6-}oTv*;ojNaVB`%q%
zZZ(^8S-RVPs+&!Rq<6RHihnAXQtZ1cSPi13hr%=dO+6*=oK9^F2{|rt=H!GZUfQ<p
zojFa(I`80M!}~SNGZerm$j-jT8hTj5{Va4>RHv4f&{RG&Kd6eGwiN*Z<ZZ##XN84f
zDPq;6QdKdyb{mp6tTG^8fPSEzx5;0SxXRCbhNv|!Z&pIWAgn_$6mS44lAXwkH^zvf
zK!`Cltqvb&_?6N!igXpj!Gj}^{wF7`gZ+<E`|+E^3kAybgN{K6Bw%^*vuDprO7^aw
zu(Pv66@W<yTEie{MM6$~J*5>i60PlN-5~Tykr>6-*I~R>e*7q}EX;Z0#2`TMTUlo?
z6?oDs-W$vP!3f90p>h`MaXt`URiMnb9~%?jKU&Dv#OU&A-D0ydyIegU2L${{0l!X8
z{(D6Pjpx~`nQF<+tEs9-L~nB<L&nd4{ph|=gLCjRm&|w(1JqKHIDWhz87_Q3Ox#>R
z!vhL=a^lKUxj{JDSVTW|<VjDgEG(vL_^b=^(7!J2A>b-w-sa+UEgaT^P3~J4cipC<
zl`dP}wej<Sw6rEygAK>v*w`vCuy&#2>)c$NG$1a2MV2<!<HtV~uHX3gu>7B-n$Ew4
z<r}&n)-#GYn*A1<#<}&T!)xm5`V0RS{<jQ$UFu1JY<OH%&q>Nz{^aM~@4kqK9Eo8>
zFzw4M)bq_5{}17LTMq-<KY9JM*+}3vG665D_uJ9yFg)3X=>uhUo9_HfyZ)+zppi3x
zotWp-z};Y)Y-pODm;X=NYCYS|*?(%Qc5Z+TP|@^sb!~VK!ZGHmszUD}D5EES1Q>ZT
zkW+js>>#Bw=jZ2@%1yP0w8^7{%C|7|^7z+=25VUVXtmIMF@toFkrC-`J13{b*;(vK
z)FjKcx3i3<ES^vruV23oYY<aFatX}Lzr&7ZZ51}n9bH{(Xdjq@qammZ&l?KCo`J-m
zV!d(NeuRM`IyzcQM<?|<2l8yz+NCx!`Tt)b{HLjD2lm5P3#;@vXKHGS^fUZo#f(RM
zkXxTdMQL2UdMr#1bl2YA-k1H<Da*>ogujeIPcySm#=Hu*462I=`*wYh+Y>Vf5>zEC
zL($L|jdGFu2F=soRRTvHLr(pMSHfyL-#gLLT*Xyk7G!<PO-WwfSg9?XQ-ilh@`DT9
z%(0wy-6Few8Ti;K<Y?+%4;~Q07E_LmILUChP#aE$$O0>Z(IZ|;>J3Z}s^7L_FXM7o
zd9^7uS)!}+QG&VS!&k*_c+W*Ky*>Lzkop(PHL@_dK^)_A=ZNwi&~u7SEFnrBrPeuN
zEG~Foh&~V_zI!ZbsHkSKyE7hB&NoEynW~?k1IQJJiQ&Ky0FZZ0jd5e?o}aO4XeS2)
zjYe6l+?4`JNK#8?A9U~E|6rJ6epXbpAEV7GufJk$t@zSphS}>@3)Q7oraWSd&oXC0
z+=sJv%#HYxDQ^@cg%4SfQ9i?Mt#3l=9+Ns9TUyu|xAis{?)k=g_OyjJiJLYNvIY~|
z5t6!h9OFos#xQD$CFC)G`g1o9KfVVE5d}!U3$txm>Xh5{he}pCqEdU$q;;sfuH0NB
zgw=UI@DO_h5#*ND#c}C{JD$Jp&?ps%xi8-W6@pgT<gt`}h_e4bT!{gUJqY)+5#KxH
zhzN7;YCA0PrK#_tTRxM(K>+Qt2`boph$ykeIvb`Pf3U0JdyK@!Q$eevVHVwjJw1-A
z{0;4Pr|`vT&QTT{F)%VBb2BkD6+3>8sFYVk4nrL8JSFeWN53;7#>}+ikL@pdogM)#
zC1hniAvl_3WMo|79FY=U6@Rykn&b+S#x;4O-TqNj>2?!Gx|o|IMcv$NbR^I}o<V42
zJZAxF5#uh`zQbBDtH4vy1$y1*8w!GU-Am4oq+mSi|GIIy<WkC8ah0ThW&zY%UR+oR
zw5`9YP9}}P0v8uyD$H@h?hMLP*+G@_KaLBTm+Bw9E3i8B-)V!V^-2k;BB>Xje-2;{
zu@w^~gBCfONB1qSt(o4uc~M3t=z#pSYcyJhIw`HCTbpa}6O4BAd57)@;1URf!bR!%
zHw{~I;-jw?b2EPnSei94trnGC=oLjM`ZPk(JQqB++*~JfGDRVZn$Mev+zz-s`D`(R
z&KG?{7nsZUq12B%WotUz-05ec5yZ$gZR!D7aA&I<K;{246l|$eeKo!Ph_L-Dc=8*~
z`aDn#{G`8hbU?1-mWgU$#{idz$9g_Ajirq6e<gp6yr9TJ$G{hU9lAs8Q8dHvy!bw<
zy_80k*r@6U33j$TJ6FQS8XG-FdaH`TvC1kHxmwOgIp<)|?8guR^7G3`M2l985u{g6
z1Ef9_l#~!86=PC59ns#}3ey-iURD<GxXJ-c&fj3C#oPvpChDU_P9a7u!LZTUpiN+=
z)L3T_(F8<wkZ!1~JOO^j$yr}nd66Rw?E*CJhML-?ix)ABiB;Fr)5F43VQv&77XJFI
zZua*spIf3JA|l&Gm;n$B_4An7@n!r-i+pP0kqxLrQDKpELC99X1JAa`L{`No7iI`$
z%*_7~pxW~=F(?KqCeKRaU)i>&Sn34{C8a0g(fJRS{)<dRb=A@`1=alJwEM6q07Q2k
zLb<)3TN`jrfjs9KyO<;KBSt&CrCzj;BV&wonF~*-xG)Y<IrGDvJy)?{xS(w7K$-bA
zEnBLqf28z$!Q3V<0XAwsfF<~$5c|#@)t}b~8J$?aUmkKpRYfO8fh%1%Ay-f{^L^ts
zE=CZzjrI8^1y)$XMMy)`7-OL!Dz8Nb?%nOxY__npgsqf%(a9CVsL9FfQSyn(>~Hlj
ziH$TAS9pBgr%%c0>1<%RjnvEkk>o*=2-zTLH=hRW(XsYCz(saKB&Z2x^YaZ16{0jz
zVIw&2>d(P&<*!8(Uy+97bn4{EOy|o+M*T?X#an4;9H2|$E{&A#Pf-gg?@CKfMxVjd
zwHWHK=kkXmw4@hr-Xx?OsyWB1hUd{m&-T2A6<Z}rOHKV=#nA!W!OH8DToY>#0#BIQ
zL7p#CD0iy$2wx;vIA-7ApK5TENMirut8gmH_8q?A;j!7GB&ma0cNLYNaL1|`Z1+(G
zR~ubQ5~R$1@!~vd=&qeRjZI8&wjt-YfBniYAb^tvX3Qw&{FkU?dHcdbN6PF=ly#V=
zy3Rj#VcBf&j%z`n%S2uXgdJ#1=ZM<nF2WNKIJ~9MRWCAAMSDuHdU|?}$FyzrsDW3i
zswTFKK7RZd2ngC5#(m&&#>5=i(%W-D9<WK41zNPUacq^+@Ib~{Jmmp%z>&V_jk&-&
zXS$vXJ3QLUQd3*&I`%rL%4#0Z8$h^)rWwa&*Y^AWm`6+g`|t9f8ebmdMM{bi-HR`$
zJbtNijdU%(@~`FB4PIfthtKeJf5?1EQ_uW~GlmGzA$s~JHO5ApwZwm?2e-x6RQdNn
zX48Nf2Koj$4~PJ45DjQ`NT^JB=n7_WAx>Yq<Rj0D=Z54A>Qtqe+d@g2M!f`<cv3eA
zPVv@)FtbES35qEzCzqug;qenHWMGZ4VXx7(v%;MA$BrF?>jHiS4OKI>&5(lRIJI9#
z6^Kcdm>>46EQ|C2Mi#^gxmW$t^V`da6ik^Wa38H_8Q28`!kmUq(URf;9Jw9pZUVXs
z&}D?@JuAWDQNQd9)&a+@&K6!=BUH@-d+$YCC^fKBUnq}EPamP9Ys14Im?jl2;Ykzs
zcMv1D09@7e?Hh(kD)jgHS0a4#>eWD$X{?{X+S+06Nc8m>VF#I{79?iy<P8YGC!aIk
zYZ~_P_RidP7*Grcn{F?PEF`ur%sxW4Ji1T)9XO~#s5Bn1z?MuwXw(FxcrQ_mQV^Uc
z+$Do3d~Uv`#xJ0%I?wS=QmWSEk<af9&M5X&Rt{?QuC{@QBDaR@z|wTDZEO#|Rqi+K
ze~oBW#~}~YAS>%GL747<9Kk|RrJ9@Vkdgw)LEk|humppLR$)bcUYKDw5r2uB_}JgS
zKbGN}vpo?jFYlAZOyElmwi&vTDdrG%aq;gj6)@$~MY4DAO$OM?gjw6pN8*ODO?307
z#6-7fojZzWAjc=DpkM=0JB)pOy=SN*NW+3@zURm6`}ZT22dpg$(EJx=NfKF^nVC<W
z^7loM>a=#gzK3taB6c;r=YZhgyhw4IcY}NC9>QA1OMRP*%zsV+)di}@o;^6#C6l0n
zEvT?GKt!(A5K{M{A(H`5NO&s#>yQKOT{X1>T_j{=Mhswjc%A@Wm|Ukg)@13wlD&iI
zEa^fyxB9h6;&HEf(!_~~vR6;2$3`NP6%3gX!{)5a{u?9G2{}18SVJNCzzWUSv|u)j
z_r`%*e11E$<}2Ui+<mahLwFEMreUSOYLjJtqf(#@U#EU4Txyv5M%6)`9`;0!qIdVU
zwpO-Y(+`j_BFq<FymSdRXIpzaqW=DZ-Kn2g%SA`d@j;8{If(R+$>jaDiL&EJRQ;n?
z`fvFejBVcmmzIP5@7XHzq*tE{93IX#@!jA13hB5_+t3<N23jVjsnOBP(y)p!7DMWd
z0Pr(P0eEvjj3$1yjw(;FQ}>dS`&RyhiHiL`e*s+C%4@vp-HC&cjs<Bp-+PmViU<>Z
zQ1EIu<A;EE^yocNjMJvJTKKVJV0kP1&aj=rca8pKCAuZ9hXUueKRo>O;`I!Ym&vV<
zucQWDPo0!EEewchTz;2snwWi8G^WjNXmlaUT(rs2PUKru^OrY9T)A{9zRI+nJ9lZY
zOhk9y+bJbwGq=`1Kd<;~r|kCoW=(79&eJs|!$bS;J4_EdNT3M7Zs7=fQ&uL48ygfN
zmtg$a2Td+hT{D%A2>MJ)dMW9UNM=gw80ODW8n74~`+Yvl{`%E~S-qmS$B7W2B8VuU
zE(oq;b3#c5cgBXEn78(px~i)S!y0{rs2n;kJMC$NKX`aPv(NQac)>gS`|oq?v>1Ve
zI&%ST2l9Xs4Aj(FNG+X?uw0-c<BJ>nW_|X!6v;8sitQfvn`yMcHx@z$?7~m9d>Jxj
z;*_fT`Sb5WdXHbC&clZf9$IO-Eh)J?8w{jS{{8NcCq=y^J0IXFAvmOP*n!vxffph$
zO~iEAAFpC?*umZ&xcCM~D~#8hHxr<(w`NJbFte~gyT~#KvV^YS;J|nBDw?q(<ovyR
z5vfpSr-j0SycE&C-%qxYf53Ga&;R)A0MhG72W74Iku;MhRWQ)a7{`Z((jd+g8HwHJ
zT2fk4@;&2T5^)cLTusg|VWeUgA8MTYUHFy}q<|FE(9|UKcx!b$om)HqoRSW3ArLoS
zIEnCZ1)gQMef#FRx?ui+Re)gJhr!->i+B!lxBN#^RoxH(&(hsNq)dBpuE1q}2)!T3
z)Vk4~BRr1k*u%dz^tY9jk6ax`1B69?%^UCVMX2YUSK7%BuagqJ)J!EQ>0X>%SuTHZ
zVW#%ftwj#U+uK9;NFC$UqsV`nclfPFc4BI(!fy+jm4$^<^Bp@Ys|i9;5N!-ImjM?Y
zS-dJ>^wN@ggOTsJ7c9A{&Q3ekmDlu*uj=DRV^dRzZ`WvV5YhrV*7*^CqTEyc)&^+7
zkF0D#e`{;;Afb!_tl^01Csg=H;$yPg2#GSoOffMrJe>oA^tY}q+_;g0uG`nI2Y)t)
zr;w3>d1)9|35khwEADJ;Y<NP=jT?CMm8Oo);X{X_o<5bX7UJiBF6qfc1g#DT5ivOd
zdIdbK3FuM#^cJL_*=~oX=3i|<^EIBttf5g)o=C0!__efzVq%%|4M8i6p~F67NN~gx
zgzzwi;o*p=D8x08EC4Y=l;Y%Lh|pir*FJ3>u;hz-<ZUWVa?Dt5S#5E<e?S19Cm<37
z3CjlSbxlCM6xyMSK@Er8QY^9mXY%JU$JmI;H|1oO&h7r<Ia5IlcOCj~u&?jT!b0~u
zb3E@s9vf7e`49sGQm4Kc`1thcsMi-hK0Z7@2p7y%u-{A-tQA-bW-&(c4;{khVWjY>
z8^qLiRb}NNg%6oLhfGa}XJ_s3?SKB9L!?s6`PM^{59TuqGC7(Gv1u)PX%Xo{U%<X6
zR|{Y}1J@t^^&`q;1=O>U`}XidF>{)bhYu{<Sn9$$f>}Erl7C}E)MsCS9h%@1k(87a
z5gs1?^eJZW+uPcJ?ih(2MGl`_#K_`pq_}(KWU=g5u84Z9FL!jPgCVM@V5E{){_^Pv
zjFN!3e~zG`q4{8vjVs$vifdu5t5pwMS>O~Sl!Bu7C<O&YMcf)$uL=t_Unnz5dboo$
z%am<>wQvI2jLHGclJn?gIOCdZGwqT=a?I@%J9`F8K-2J6+19VE06{<r`pqaCvkwSb
z2wL%UW2SM-{f!r<D$A_WM)hyqngNXAy4>{M80$c=v9_H<(!}0AYmoR733+10m4q)?
z#4nUDQ&RlbP53$c(Egl6bfENTYcmohw~k?%h=}5vgx&-yU~q9eQD_Y!ErQ)scw|CY
z*on{@Om2<_wzjq!Gp%6)E|Aq{cwpeK-eu&`2M(+TG6;P%JVS)T!mxKIdaM@0M^~?g
zXrD$JlkeGIP5L~7*58cz53q`h#{|5GcQDYjCxtgI_8;3#y_~xkvL3h|0uAs6%jO$J
z0Vom6L&V3p!`;nIG>W#`n3wQqqXp#{ay3NsF?55N8*m~13SmF)lX8H!g`12sBd@54
z6mx9-@TE`o#`MJqc}-595DNJ8*ba&m9!T{lNJ&u<&+`i1UW!0WcU`XWcCnQ~2)}rS
zQXuJQcei%DzCG%)Wc;&3L<h;MaKR_2_iS_(7ZouOp(O`UTtQ+&x7gVZkDw_k>S8|%
zl=e<Pad@yfZ#1>m-IXquP`Y#sww28w85-yO(+>d%y`tL+DXovCnOky#kV$V()q=iY
z@-AkobsdlWASy>sf>C7t`}XtawWaI9C>A+$rn#ddGCbTwbLlAm-Vd#Pm$xZ-1fre0
z4(*C=2=etcv$Tv?^GXu!XgIr#i3r+QU2QEty#gX1DEA1BOH1zre~Gr70nylYuev`f
z=9j_OX&KhgZTI5hj(q>xwlBJYf`nvbY>fHtEpmt+(N)p0&2Q?HOn)4rk|OEcPd7tR
z(c0<+#R%7so|YC4@)0F}Is$yrp+FmhmgZG|qfoxZvv&fj9gs$wnK{}uDNadu?BUQ~
zNlB(i4oB@-AB@mLPi;K=>d>)g&HT>1M;OSfnrQxfOLT?P5wQ55Y>#a)VOUCtdyKdR
zZa#L9JnQ)MG+RuE250!Wqu>YX#m<@0(F}fHM;)<*`ZH#(uCMOJ4Yi-~`G9;Yv%w?z
zuI_r?`9m7cG7tAj-S*y9yyM5Mvj_p^Eb08Q;Sxrd3Z4<y4ca#}dR-)WfspP^3x21@
zM(RL3+JNvN8<0Ck#gW+aN?v%VTFWR58^L(*2Gp2OWFfMcE><L6#)J`B+(k0)_t4OR
ztP#fUF1@SuLRp!%73jXX`KqR7p>>zwmz&`)*?BiS(rlzc3ha#oUAxpHg{4SdxNWDK
zAvyMtuCebZat|}Xo*7kDOw^>3?h|K-*jFa9UwzX|Pm0^^P7%y`A+f@e{~5yOD&<&g
z$QB<W{|{$x0?pOi?u~C#DPtv>GM5q(GDIq6iYQ5@q>?EjWF8}m%#~0iQz}uWh>(P2
zN|R)YOqpky{-5o4&RNfS-{-vVyZ+zxto3_Vll|TM-uHc7pXu`TCD2mir{HzDaf9~L
z+4oW7lTxf9n|a6&6Cn}%4DwA<?vQ#B3SR^WkftFVVmR34=msvx2#CeVOGqej|A<#!
z)4-X@X(y^7kytm5hh4{M{9+SdKqF<L1C7qft_LorHuOXQ{|Qe=N#gN?`(mtHD$<Ub
z8mtS5=&~N0qIPW-pI|l-+!*=wis==K132px6l`97=8CDsA?1NQBG?fq3D}>!C$#r9
zfY^%N{~6s!yT@n=l)M~kuqVOo(`*|2sM?8$r*nNOp-RiwGneg)|Ku{q#djqIb;c{#
zzjy4rzg>vpq1d+6f)^X!Om4kbS$%G5_)uCo)j>FZl#+a{=;lxxpxz@)b6V2c&eiRM
z7y>#mp^7)zQh$$NXYQd%Y*_0H-01mpv1HQ{6Tm;s9Ua=?twJP{1c&cbCn?P5<fJ0#
zChuSD#pB`P`R6IUJ*n+ilYK5RGIFgj^q_3Fa+vm~;81S&@u6@yJj)@$z&#@7QEw{#
zQPD7c;Zt-r>M}Z^GEf0j2US(8%F2A3p2K=JaZQjQm=wl{0eY92A%kSUDZK2cPff-7
z4#>&HI=PTe|NUaV{eS)d2caMsgQDSNS5L_K$c3$>yF!j!GQDRE*l*v7ZOeICT+Fvk
zxxR~j%=N03l?LP1YZeI<#!uGQ1!(#UWkE`Vx6|qtRQ};r0uR2I$E_PfgNYK5MVX}1
zpdrMDRZ(#`>9`cA{k%MigRZW!JMU1Gp-6=I3DELuC|>ZKa4Np<9^Wn`gc2h@E{@<8
z6eJbpg7<37o*Na6AhQUV8<;QP@1BZ@<)(m(zJB=<*^WcaP5Rki^TYpgCwtpsB~K1t
z`DFbK7EA9wd1CC#kGyfalHVq0nKCja?{O6gvFml4SFLUf(kx>r%2Azz*<(=#)Wyl*
z(zmLr;vfEX^W#TN>3A=c#(>rk3RThh@ZrO)^3H5CkHbSkT3=ncaOKJ#<r9Da@tG0>
z1OqjtatN!zy}ltya*cJat}&_6%N+^k7H>0+nhNYX!os*+9Cy{#wSE5)OrsFDp@c)>
zo(Qi6GPFSD(G&V2cTa7q1!Ec&6$Nzd-TCqA>Kn)kvzIj%lfjRH(e9jtni+IYo{DfE
zJ_7A-WAm}1LV%Z-S6bQ`zn(_ph%FE)btsC|1R<CG@@E={7nI?kQ{uM$8orIqflu{)
z<DY4|2UJR5Q45(>ZpFPao?pYxcNSJ`o=;8d<sOG}cKs~0nh9B2>TsJe2(B?3oYua6
zoo1zTZO(yooeaUcqOm`Inmg|t)oBrP;;-(C$fvQ9<}Br|zI}v$915kYSI_hxiMRlj
zZVFAt{x{sRmh17(7t^ha_4e^S5_Q609kFjsx8rrn(Y*rLL}@Ahc@N6s_DTjmD{`vQ
zNFsbknwnrsT<9ZzpRcF?KW=CK<?9b04jeX?ip$#~7V!=3HKc#bR(#~2l0(fwZZY!O
z+H*^z7PpR$j576&6Q+Pw03pq2(~*CwI_VrmIJ1%^pTe^2;uLHht{#$F4}N4{q_b-A
zf>rL*QCupsQoj)$oh$xJ*VD*1x6r}ITlVq!T5jK`r_cR9${k){R<n_KxML6^k$a=`
z_?dW@*FC4gzdPB5SS9j6$(kMQ;4#g&-aA&g<*crrUeq_Ed>cMx5dtMEm`Pd!;3fbN
z>jKZd+D+PJ0^wz)D-7GVHFhx7*E4g{#mgMNEXHR146-1VWGjfxt*mZ_gcx<+U;SaM
zw76j{XFxuHJ9_3`3#x_3wUa{rw?C^fwjD^w($?V_!=E2l;K_O~X3J3P+J%{8n>KQC
zdQA2nBV=JJME@QiT3Srr-6`9E!~GkSEZ@|<+F=N#DXb8Sf{xmNmw_6rAqtI6PMmnr
zIKWHZWKnw08-5XO<MWF_-0szu-DvA!0wmD&An=B5cyHUa5hfdHqo-fM1E}&xf{4)G
zjCSnETFznQGq@p0qlWHv+8{0cZs{F6`u&*{319eCdwR?mgK`>MR-a>?44&i5*Eofm
z*EOJ$PVq&-|BA<4LZFtKni_Bogr|`EYDoqs+)4z62xv(9N2RH_!v6ik@u?N!#rl2+
zBkte-3FQVvd&`USFd~cDT*gDf^I-~V6bNWsJSWZu)uR0@`)dQhi4KKGM&7ikZ?~lC
zqi=r@Bs@UM>7%c#MG1)hbuwWK`G^}SD}=^q+k$okKiUD+<IQr@?*iO)Dmpr!peZ5a
zK{%APtrrkZm*WA`?<z`A)cEgF0tk!5_lKP-CmtT(fy`d$?5eA230K%NufGLo0<RHA
zAhMvoo%PvDz{c{zvWY+0|HhF=Oa~U332#8tpX@w;-#P8HuML3fg$ow|7@~K8w>k%Q
zAZ%U9A&9W}SW_eCv#dbeN8t_h@Z1Z1^BiUJE)?y?I}p(wcmMvgCr^Nx-$k=Vc11!W
zA|jHK&5z9}eSv&hOT|I;pLAT`*cc=eL!MzBfJo!ZVE3`UUKx+r{K~9bJ7lbBc;_a=
zxbtS%n}>(@@7wp|^XJ>q(HA%lzHcyEm@9#t1Q|5YqPFw10yG?l92QFcxWq&|M{xrF
zm;^;HcTdmz?J!2e0SS8&)YDk>nMbb*W6OiY7nx#l_W)PmfP(~pH3TSygrsC&&*#<W
zIZKVT;(*+Y!)UDqwB>K&y8?Yl#~~~)?@<@GH_Z;496+<AEu85)f^{!KUh?y`iyhKy
zaQz>uo>%rv#ooL7PTDidSy`p&w4Pp^Xs5c5JfWtkDK%^=vl;q+ef`He8Q|@8PM)0i
ze1q-|A)HU1V0X=gR_N9(nCf5hu@Qi6!?((>AvCv3ZripEuwsoF%BTOOO>UUD07*K|
z;g<+8f)o*a-Q>J^g}%tw!%X{+aCgVtzAd<o48p6M_>b?mp8S4Cy1Tpa9r*T=<u?3k
zfQPZn@cO@d_YR6Min82^V*9RZ#$t%JB6B6jU2$e&Q^4|NQGS5_yUdRY=xx61(}@?C
z_Q+k>zVCt9-o01vCZ?2tt%3jmO%ZNC+L4MgJ>&l&MfS40sDc9D?9&|7D_CO#!;rHe
zsbB^#E2yodJ%*c~H4jurT*di*=CyOv1%9|S@CoBNpeQY^vVOoI%pB5z?#X$G!SIeY
zt0IaGXaNL)d4^u&zbdg5>U)d-d5p}fUlSb+;+N^y00G+Sv@_3j{1;SFK1+Yj;-St1
z{=vNu?Nvi%?RPNSqJi7p?*;6)((z>a-tisp&R>DT&~{w!_l45)pQ1FOh}#H+vb$R{
zUi(u!HKiX4)KwHqg2P8^dzq2EH%z(A#63-Oh-A8)A+CG&Oe9v56yz9<{Y$%9XZgRh
zoAJNTuiy9JGKg16z0r5*+~65OSBNY>cyl4LF?X~kcgiFW8s{X{>Mk_=DEcYNa0W44
zI!!<f&{FIM?DgS};CF22#o3U^Az{dk&Feuiq7L&bjp7@orq7|T&dS<d()7fXPO|Vy
zcc5)-1830|FUUx{oHKOyNl8(E-7;pDx%YvV>eua3WF#b_ve4oFU<H!HK%i?i|1<gd
zvo3;#_t{7T^#{5oN<T9`?$+tIw)zj54E@(?@}HZ};7J)CQDKEPZ^wE<@0tF&q8Mm6
zX=u=;$KBNg)S=Swox<0f)z+{_pUuuzF>pHR#1{l^Uz&1$dQJ4BnW<@6{>Q$P5ZL1a
zAen$3xlJPCRw4vtzsFs_a>Jg2U~Cp>a1j*)4epU6NWGA5{mA6@^jeJ<7nU7Bg60w%
zX?b4Whd>XWdB|`Ex>$?6fVCL;Qy>puo+|QL@lf~L@8smGN0*e8eHpkLRj=6E)x$a=
z(5Hj9$38`8CjrrugDh7L1v+<pW2~D!KQ23S%}MR5Fd#2|X4iJnj{)5BrjtT)Om8bH
zfTi6blS7+9TGXhtzHgvH&-&jw!|1+$?F_pbop8-Ta$nY}IfHkK4I9m7_3Eou0we^3
z%nF276rE2$e3&a}cSv90eY3ppT~$S^-C>h<k(jY6s0RMY1;`<f6K01U`egz=_G+Rj
zy2586h_rgUBN$fIK&sA~W}Di7E>{ef`ws72ueJks#Ht<^?{tbU4?8<wjNM@hCIZ2^
zpjhypqT)u0iIhebWyGc6gQ#|dn}}}T{yx8RAW|#nuFS{8;1(n4W(}5*<kv-)gMz|e
z7vAaZy}bA<WM$~m<-oN(0j0~Ey~q3wwg-}BL60nouIes8TLx_E?}J+78Roxd+{=~T
zq`Dv&3dIyjdehV6s<34s+(1q3hV5Fj7y3}i;6w06*P!b`h%aP+fR+#pOG1h4&tI9J
z6OX9{{0KF1!eENoSn-M7OBrlptg?_2Wy3h7X<n7nprdL2<>f`Iyu9+mhga4;+~GQ7
za3lDd+u4f*`M|5SR<8^ya1i>4w_h&X!_EEUZutISbqxvZ_h{Txa^jM`qEyI7te}9O
zU%kU!f<lpx?_8V#(@8#>{E!g3{ic4{N(z4>O9hr<5UdEbLjN5Vb>0?NIE;|xqp8?&
z@Mu-~Qg;MTJub-h6^iEQeXp_GPm`6%E`B-jA$T(i*`I+7+FbjsZ}Tknt2Q<gIYz%2
zSFiGr=@$0@F2;=OJr47_C3#R1X~7Wf6c*b6nKpFHT{(I>T#ATQKxs%p=<xhtwz1tF
zHlxmQP9ipCix&dRm6G}g1}KcG8X%wO(MFUn=<Q=J>#i$JKvMoT&{Q#z3EeN*T}mxu
zzc;(*KcI9O;I@hpu*H&tWj5-9W8+ml7cX07r0c9B9&Fek(X8cq?AUR`eKJC+)6?ZK
zXW!ODAJ<#M8a{E5T$6G8$&FanjCy)*PS#EE(veKPl?LQG?aC_FF*Y8$H~&5BUanv?
z$^-AE>2LWL6<5>>+dDcCqW+JHK&{+H^d(^g(y7fa5m||<CXK53!jMnKO34@F*gNrh
z8C(#OQ;AM{*{MN)a0Mx6y|6nqy$e7f%fu^h#@n~c+WcQX{xWaTu+J)QF@Za=>4ao?
zUqeHkIz{JCIsF%ubDrO`DV&FV?{m@8`c`rK*VoI*)2vo#pY3T&wuaui)j9ZX=Ah3L
zJiTVkx~uk6Q$v6zlb=F5iL(&}A52zQCdf%{Pi-E+@g}3F$VdS2nBuIlkl3^5++V*P
zcaHwOMQ+C7iHRLhbE&GTLTPSl`mn1UI(H=#ds!|{?0tfJ_Mpo_pb@8{`U2k<5#<ws
zeZTBwaophqMkzxeXqth?80_%G+c6ELf)L4{Y&<+2KYwbT_}o<y1erZTL$G6G69Ui*
z7YPv%7>MT?<@)p*1$Js~S;t-o!uGjOS)M=75137y4VRBSU`p6)>RVuONpdpcw)>Ic
z0-9ucX6A0~Z}i|0MwgpLj3%J*GmNPpDrg$2r1aGQt$aw|({XaQiHGeIZIkA9MMVXA
zV(K{E#PmJkc7<Svh53dv?H&!TuaDBKET%6{4UErrB>v1UFS@|5`lxvTsUIZ^@-!{+
zP1<$=mMBT4qSpk_{?%s0#n*hR03S2tUSfWZ!cG;<Af2m6<A76&4LBZYr{~Vh&t6A6
zU?Uwk1-gVxB?MvuFIgD)-*H#SC_s|qAT?vX;_klBEzJ5xjaeMGb=)QcZo5;8u+E_f
zXHiCpDQ1|9)cMH2Bou8X?ul^_Kp)y_GuP(}M#CtL5D+-TCRke))jfDOClAC(gG9u?
z+hd^&7`3Z?pj0rB69k{6gajFseL@!t?}XX$<Bgac(SMd+kA~_X@GQtUkixk$xCqB?
zm$)lDTvz{mjqIh#MNhEO&9j0B$iv1qH<Z&r@A4Ztz<2M)mGI|usN`&+BQ#Hc%iuq@
z$Jn^q10}&ga#P0+x<bU~YNdUV_gbI&se8QYltuk<0|ULSta6MK)3027(s>v7TsDpl
z9FU7xvdl9?mHbv~SATr2p=d@Iv@qMwGCfV$)v3)dk6{A({VXGpqU$9i#45fn-?>Yw
zs>{8A655wNCBbkgqXu|@f*RE=h_gU}GIsj+km91%WLS9`{S8K)w4g_J6j)!MJTgNI
zgcDK9!CGfxqIE(S-aRj`lCE)tC!(iNkCwaSFIr;;G635NmUBvXTwY?zQbaKHI~-oU
zKx0w_5sU|Ywxo#20G-TYR3lSr=o>k>-ZeEeo<4pI{tkZ=s<etO`@)|fr(eC2O}>;X
z%tk=pNg^Dp@y?hIVv+?q32Cw*6Vf!^A-G|=U-bZf7n)Ov*jtNsVChXty}hihg?Dyh
zjfE<%&FOfbA;!9K<44bTC+&m7T_&ceiBui+k)@`yOd5A$Dva#vn3J1c?VqHVlBrmH
zu6m<47&=MoyxEaa7PA9A1F-{Gz=SDwP~Zn&;#xoSw;oK@i7Y)gG(K(GDRADYfzS3m
zMHzlGJ9@q(6}}9lqptA<!b9engvQ>xH-};r3XSTcX8_G0T@qc>)Xj0Q;K&+4B|)0o
zaxVdQ2$g9kd`StGlfTrZZuCn8$|KP%rp)yTikl{;Q48*Q*7)oLSDh$457N@C7~B%W
zv_n^zmY&-G%0hFY_f{Pp;$Vhu)PZ-QZ@48r`uW6-&Ad5$7C1deJ@mNP<>+<F;@cyX
z)c7OG1@S&>t;DuSHPn(Xd{tGzO2>*`3eQ^0Iz~72udoDuir3<nvz`dURyVN&*oJ8A
zo@w+FNr!_>e^}cyiVh9*ru!!LT{_}eX+se{au~{NE(KY~?|}?X!?qVFvUIN{@>oi-
z(%7}rTh!~joAs#SKt5Sc$$gB9&eiPh89LV?rO|^>M5QqShcJ(~LeE`Lud%!QSEb}T
z`@*rA)|~w(EYefI$fF0WpK)5PLr+#vw)>HdJx}AeZ!*@-k%9q@Ug}A{FC}E%HXc3N
zS^waA{rhrPccl|ePJ2C%GvD7&cj83O{-(KCo6BE`D3Axh(7O2Va@5lC5D=glM*eYi
zRIPcW>XhJj%b1XmTHrNmCg1mdv8xzh9ZPXcMQMwgUqr+PK7`5UB!;>dz2<XYvO04b
z4rVT|6lCjXTW?;RDHtoGqKk_a*g*U65hVnXu9L^aiwg&a=sjl2o@ObjRde01eBvJv
zSYJ-k8GHe0zBI*$X=!Q6eF7<LHa18@|Dt=&Y|#@{nWJM)lh_*nJ&AQEMn{{`Yg%Zp
zS_<wyV}P#|PA1GdVcoaW48eArpb!){+<W^`=P%_QJN-_Z_?=yL@RM*<NJtoK?RdM}
zXI;#0e>G|jhJc`oC#S@N&Mst&-?-_QFCB92bXG#nBtO}O#AZMDU1{xDUPIvhQhB=k
z$N((|pbu#9Y^|;1p7>%b<7Ggs%cf&=uG9nw=Q?wvH^n)5<6&9o-9KlX?D3!^Kt}>c
zq>JD9@<)2&+wSm+w#vm@1Jh$q&-y*xRXq1=G41zvV`I~{ee&EZH9}n$2jb;!)ddPA
zWMAMQAMM|-Ik*bsiZBrvb;QUz<_Jgh0FFoW?I%MY33d13P^%jO2$A*XukgJ~vCs9M
z)(qBx1WMJpBFhCM8Tvpl5zXEDY(uo^mn<z`p_V8uU2$yKuWAyl_yVLG3Vx)~K#w3P
zQjZ}vGxL3kM%)ML_U_{a7H?nx9LVkE$G!1&&>GsJ^A1Zat-5uqELvO3prqNveci+)
zg%4vxd|zS^lj6O?Lv51{hmC(03$8YDc00M0Mt4AM2Zj=L08nadVVYFcb8g933oT$6
zaUA--2O>K=jH&?R|6iJg0=e%Sv{d2mp2wtxsRKy-MhYW7=oFnt+lNO+6l-0pWROq?
z;SvCgi3VR^UsY~5>eI5NLM4-oYN?VzR(WF!9TgobmsbK1(f6vbJXSK>eJ?$|Ci`UL
zVf+3@h2DRzP(Y+}D-F9czNl5cuh7yk!3$b1*S}2{a}my+OFEj0D@OJh&<2#V$d7yS
z#2Q^MFcm~gAxJLJR8@BC9;Aos=ui?6!{I<dnf6bmf%6g)I&v@-c0t8>*D4qA%BYqb
zXAj|z<7~OIZp!1!K=WM{?-};~X<q@6-ebmG#;@0CE3T`oeE-_`43sCGzhnGDOLfTi
z!|d$bw?Ha&O-<OAVM>6?4U*fy*VWxFVESO*TweC3WBUH}YnrYwg(-Lf0Ps`59C~1h
z@<3Odk#SQjDs#PaK=THacSHn8^~A)6^DnA5z*jNfq{;*23-8&DBm7CnEgg#m;^BNA
ze42#=rRD&R4`8AkK?1I$R(G@b#qM0wr}y)l`MAdGq&}PH{5Io4N+GSXOe5Zy!}j9O
zjMu$(KBWI$NhpZy=Iv;3i*Fp{9h;bd#Tl?9m;+ojwd<v-2Ee5MJDp80f8Ss;R;#2Z
z-I>r$D0C4W;W}9-vojb`IeUbwGecc8AvYA#PPpB^^)ecwc}Q<A)E0Sgi^+}5={@t^
zyo+_;I`Z*)X`G`NJy^ZtFIYx+ggIEppz5=<q<yx#vqsnw`Ux0h9y1t0plfq&YnM|l
z0Kf|C`~L*kXTZyd0aVt|T_t6HeJ{_E?0ppwm5`7HSeWlwBi?1*hg5@<L4%IDI@hlu
z`xaU`R--4k!DYTeU*Y`5<MOp}1r1HjV#pMbHKVz`+o8}*fc*ft%i5fYQS#h?emYd;
zBqU*=a3Ug@;pyuh|EQO)M<f4sz!ICRc>x5EKYkdNn7~7cG<=&dmmNn_CrJI|>(_#L
zm;r`*Zr*eGt@-s{&CAbIX*^ZdeE;jnwt6*f$u>JXVRv$$>Oaw1A4Ib__5Y)92S93D
z9%jFfBdB7}>$Y#-Vv5CAp8ywnW<{^RQUhBF0(B;^7zdOa-=8%ayHBWVc+~o=EI4b?
z{Bu{hs<sL(4j%zZ2oa66bQGnxnOOi!0eB>`$_L)xfuIY7C}D8Nb)`cmbg@<7V`Vjf
z>Jsgm4IS3(v2c^m52T*QPtMJ!e6hwHIPKY|jxg0=EFAWTEz-NBW?#NEoxLpUY!jG$
zv6iq46uS@N5wjL3$kG&**}yOvNllPs_?snbXMv@e#YNBLbw?&6COs2a$SZZMze~A{
z0Pf3(Bf6h{?FJqYgycv_mIzKz@SYtX$M*kX*jI#g7Yc7hMQ#GKvLMI9*`uLR3;WZ}
zO+uuA6?W=Yayb#=;9T?F;voauearB#5oc*9s3|m}>a7O|Kw};~k*fXUl>Oe5?H;@v
z%fCE5x#oz<+l-T@pQ<SSd4RKcfGD|P&5`WO;DvYpXGO&Y8=G6V9tmygM3{_?<q2s8
zv?2hlrtF=>0E#*wj0#GLhWm1GO)K-CIC?ZVk9>TFK0c)fdBxk>zkO5Xag!dO*%n=X
zc4fUq{Xrwn$GXfLDc#)0mPRWkeiIMB{XxIh@IYR54NZjiRj8n*OevgrCdPcDHbOJ)
zI=a8nI&}7O2jl1)y5(iMNOtn%W;QgH4QTHYkNpVwHq7?W2;II7LY_PO>f$04u<(px
zH-tsuQf`RJPTSQM92@5_PBJJLWS&@UUat&k6@i=v_PUJ)((PgE>AGSqOkL*=7#-nf
zzS%i;4nmC=-_9PG{2c?p#*8Vb0%0+)Ta1|cy}Jk3`^?iV_6Qx<r@eDWT#9;mIpRLO
zyo-|iUu;3<kdpoel_a=(WHg#Q*)8j_Dck^S0V*DOtk617xt9Rqw6%Q?;0t0pOe271
z>DIpQ(dRc4{Q^ehUX;uLQ!xCC$a8TzhG*~n>)pCcJvstju-J>mCB-X>3QJP`)8@_t
z(~mS_mY4qi34UF$5!E`jc~QHeppmtE_Kek@6?58XIRe+!%dlQAA02YbsFTC)h#0g0
zOko4Y3qqv)>-Z!AlXv@n_PThafFXg#Kvoa=ph1ub$b3Yjt)4vy<;+M+Ge-^`0<0>!
z=;`S(Zo<*bjGCaPJcLkOk?!F~Z@6bhWsFKrYML8pX<2E7C6p5cnlp4|=?zmqPi1{`
z!I74m3-%GrBs7sI3J`{djTZ#xh7IRt&OnU4D!@Z(Q+w1+DsvAfgFma9%k8jB%gB7|
z_2+f!NG9nW^lq)*%Op41|3)f{rr0p^eYwY73@-z+Wp>N~%Rzy_YxO@!_q^BJp2`-u
zAC`~(8k6Hd1K`7P86LEsU|WW|hktD;B^*meg1kJ})2&}K(q2f*$@YDDaq?$M$^g^X
zwzef`Sw1<=0q23M7J0is9Qc`Oxjo0q8yg!7kMi)DNg^>2r`k33!>$f?Pyc7hXpTz@
zZZ`sc)CBom_+ayZVFwN?$U8;lc9Np3e<NYNH~5~1d>dFBJz2(Z#*!Jhh<_VsP7h*8
zUwD)OdILZ%r7yx(Y|{4k{r-(#K(an$zDP67)T`~X2TcT90a9|B{I@<O_mF^tyx&WG
zd_Zhmx`dp_$I;OzF`duadJx@dub}scx}g2Y{g(Y(CMVRU>p%a=`A~cMY|*=0A^y-^
zpo@S=6&Y=Wy|Z%|hUn1J+B-UOqPXVZfRPBsY^*sDiQi#NM18PdjFwO8G-SGXU#ka`
zh=V6?BQz$Y^1q0mlymmD2z(pVV*{f9T`%hKz8U2Mm%K9y&nBbl0kCO#aMGdklPeNg
zUFQsc*3V(!4AntBy}R4WS(MMfpPHRL29@U&<(cYgOmMpB>;Z0-{ke^>px`^FE%9{;
z!iuu8t{7y&+I!(#cK7|TF^Ctl&yhMi5rwR6<B|!VoxglJc*1;k47?xf?iu#2UUMBQ
zVf{nY#H-gcugR3pM6VfkB3mCN(=+H>IIeca-@CU%RJ6o>T3Go6vRZL!Z4a3N4c&a5
zZcC3tAwX<%^X=hP7|H|HiJgs&;L3F-F4==j#~!7ZVSmH628q%3=oU;>1bgzu8p_&E
zP$f(I<^LZTuQ)4eyz>x?L|>&@Bn^BQ;QPjTv0FrB5r#f6a2H;M;6gAQZDWnDhaQ-i
z)*pkC5^vMKF)j4^@E=CN;^W7(#CIEKh_LY1THQkIJ)R*OE2}dylOJf!B0VNBAi&kb
zqhV$vv(gM-Of3y%IU=R{*)5>ju#tYfnq`DFhmmsxkS0Vst|k71iK=MBUh`i}lpfMw
zF_f{+u6;FSbsh0<<(h!v@q8ICik*&r{_NQz24U=0OiYZ9o{TA7y#Bns#byStZoAiP
z8~O}6g{?}n!i*Mi54?4Dw%SjRW_Z+M9o&4zv{FXJbT6~r3jLDVHB9ml-o3j(y2B&R
zFf&2w+hu~U>fJv&#m~C~FJDk+vVa3MP+w$?@6eD<a?Kzt<;eCY33sjCgXg{?^MXJo
zUosCbc4~BQpaD>n=M;0TF2qc^dsJsnpPog<3t$Xn1p8$d2M0rh0)=gr&qp$UU!Tmv
zCO)S`HFb(L6^oP(=Kgg6-O-Lx6L?Y@?r$-!DsJi>Y{iJ$4#oYKD#yT(#Fa%0m@fRw
zzr&%0lEgy68|v8Aq(RWHnA3nh5Lba@C2WPwQUb^wLn}Fq%<+5o0O_WjjA`N7!s<Lo
z=}XQT;0QHG1&><>A{WIs@<vfLBjyEUSMDhI=SAQs&<NwSs(FqUFC#(fhqD!J3+bp(
zxFFNh-F?`ke{}h0|H#N1Z1r1`!<-PpJcvRG?=JL5b!(W-&dEK{I&EZhi6t=P2&qsA
ztC^u(MurTcq3|(l+8qZ~np^wycomNwvflI8@k?=9BfF^VfqZ8-6-S+th2xe)rr22f
z-yc4r3dYP*=LU?NQSUAyABx+kIWoha>A$$9kc;VW_szE#>pk6%`MVkIke*3gn10?o
zfPt2n-C+7E^7d=+Fc_#6zJKOk%b{Xyi^xq<u_HC-7sB^TFKC@Xcla?@5_+}8lgUju
z<iLsDto(&H8s)97gA0EJ`W&c#R}%oFg$-@qNCnvRj*i+DHRWaqxv`A)^;yi69{5i=
zCCp9D_jm2t^Wxv~Yn(CnNVcEr_LtHWXscjzJpif!=D&(zv%EoQnQU5KUZ*dDW;`xO
zYJ)`Az(7skBp}-P^oJ!S5W0^wsXja&!O(v5=7v9C4;=|Otgv@!*zR0gQ7q5f(H5$c
zBK+Ug4S!N-9&l|x4yX&zShYvGa3|Uu5s{((e!}-yyfnUlo$Gfj>Zpydd%FN<G8T&|
zH$~hs*fI^lC?JyP9h%eK5L^Ye1u;UXqnHpUmZ1yLBbqFXyzKyR0Qq=-aAm&M88F0O
z2grf1co}fAqqTuc7)QX%mlJ5aQJmq|0RuQ+xnc$3-s8vE8Sn?7OoIqKr&ub$++SgH
z;)b1_awi+AAlroCzS$*WdU6t4XvnbOJVQs8GQB^f5>pvbCX@dZ5E}_E@|A0exk}@S
z1QHY+3<Yqy+hW`X>Is;-q!%l;@OdB^7k{O<^y*n}bdKfKH@foY($ePAwwsEZTGE<$
zFtFy|YN+*3T%z}$eLZmgS9e#PO!^OXJu@=;_O(>5LAgszOyXeDeWVOxfG+#V9GnqW
zD(xJo-YE&_S3jTBW)J%a&+j~fxOUw+iV9Ejsd0ub2LjGKW-q%xx5INl@-hbphl;9d
zs?JU~Mk~j3u?Je<YBdsNU{-Buna4yMYF|(t@G4@lMVtW%QhEN)-Mcz0du3%U=R#i&
ze`0o^?$Jv9nw=CV_y0gInww0J>nuX~+0%3VgKa47@yg@X!&(P=gp|<tgIncrY-5<o
zWA_!N5^hKd&#t(8J4*EBee1k)+Vly<_g$yorfM|jz82`N$#hqIZ=O?QX;kr|UkZb*
zfQ=v~oPyLYL)I@Uw_W)J5^^!^2e_rm<RH}HHEOOovvd$qLqlQqIb#?6c*PGp&f}Xy
z&02IsXya(uMTFT&4sv8eM>BElCu{8QR6Z%D>Id=_cF#6U{LW3GU<j$ijAJrVi*9%E
z80@NuLWJOsk@$zYOnN1MV;$Mdgq6K_yNvbRj*<k#f@rK~NaJFYW#@~k&XJFOJGUy(
z8s&rOpNwi|VWTIZ#IB&O+0vj#{JcIla`XWtj(t9v?NPAp0=1^ad#!MV+XB&j447X7
zJOu^sCQ^kYIuNXilb&iVqh{W>p$C-wcy7-h6hXkyWW3IOycki68O{R?+xLF=w4$w}
z5{qeVYJ&0_CF?PvXD3*cM?8<6EsOq(A?w%QY=%3w(D*V!IY1B(@bkeFm}Z%J6N@=%
z-r@{J18%fyF{FeDk%ImTvJ1pZW*O#v;OpHi;55W`<A`JsP`)3R#s<=7v@N%Xu$CWK
zrr5WYp21Ir>d+x^TOZdrY{R<3c~5So?65w6o|4e4s`W&+4@&Kxs~#S&m6Eb;HA79Z
zP7B|`g@Bhj!k?_g9{BO=X2<UtwOA`p{F@8brO9ROtliC!<W|-<iEs%x7LZecu>M;!
zB=T<qi7>#oG`Z^k-h#n86}6=&XboA3a0Q?e0eFM~>Ig9&hhnqD1ae`|IWS{EOhEww
zEG#TT$jvZ$%9?cj%HQ17$jVA_MaGelnUC+!lh}t|Nkvr^Od6`mO}J-t->ebx&**mO
zfPEUK7)Y?)g3A&w`jan%h=Kw}_in-rdJ|kGEPrm?mgko52GpsiiwWp47-e^TVvbvj
zeF(ZzM8QJ#0R)7k8e!@?P2y`@jCq^W$Ny6cVDb0t2dIWYStCIWh+DDUGVP=E)UkO_
z`4k8WY56z}PM4v<BNMKXHb(}}q8K0VuELEX9s&ae$AD=*GC&y6U|&eo-SG0DMo4Pp
zrcI44a;L&ws?hr!QYnT6SG2%JtPDq64c*@sj?i+`p1wYOVF>7)U0h%`yMWzU<x=6X
z)JKngLA=IaF<~IW8eRpo2kHwbPLZBJ<tKS@xN{{#AJ0xY{|txHr0@S(rMp6o9BAH6
z8}mQ+fmY@a0%*yU!W9${A&`BA5v<;K@+Cw(2JF*b5P4UH^9Yq>fulF^p1-(pa<+7(
zt+`neY51z&lIJC*&6TrttO^u{#P;orD>ud1a|jB-H?hAy@mTi>HXg#>!67&}SmMV*
zr$1Aa&{cpu?iiPe6eu);%LuCe_^R3sNDW9df_lxG-rv7bt+2DwPl-~+Un%?vn~2=N
z=Zm5qXE~H|F!HVdNiOgL2R}lKQ@!N$_RE*O2M#2T)XUxh0f*=bti9kzsH*T`n4NGf
zhBnG;m$!dG!Qs<eF4<^q;i3oI;3SQ&+vsGz5F)pIJH`zBJp~O44i<7+FGv1OMYjnv
zkZ!mWusf|FgXNEf@dQUwOU~fA6KXfwCbH!jgAV_AvWq1cNgnLoyG@EAQ6o(81k*&b
zVCAa`fm|6vLJD7f5tOs{$FtEx5M|~$NF75RWij=tA=fA&PwnY&Mqri^`7~!xYMAdv
zlur=qHmq5kgV<UhphR0-gzI~TyiNKgoA5ZviR%sy65*s1!^PPd!@wZP+`VImE1cjz
zx0*#oKCsLitCwfqvbL!yBBn3@V#w^oZ(f$Si7WJ|wpPK^Sc~pWt|!S%)SWw#hE@cf
z4;C~zH5fhwf{l&M^$!L>xc|vSejLP`Z^qwqI$l)?_^6+^UWDsA*JhgLd>02#7!ZJ=
z;PA)p1x#tGa9<`SvMD?OkQINWZ2}ksCh#Js9n83>VQ#h4+hV?#?^s~eoSKsQk^1cq
z{2(I5cQZcXy>0!z@Tsw+_RE($+?UL&--g;qx7zeHAelF#z8`w98BwCMuWz>TFw?Q=
z_w3s1$w)8E&Pp7oo#nhqoo^TPQlfi!>(;U8#In+58VnR^mT?2r-ImqsZTS=XSUbW&
zP&9+w7=#ZerS%e+dc+qmkKRvYOtJ$cwDYu{%tYJh54p{+5em!e^9tTN?8ay+YNqx8
zVLfu?$~a&-h&rNcJm<QCER6TH!!d&_9^zq>Wi{J`RfG__zMqyaEa7*=APt@0_2yH@
z&<?}JN`Qfg71^npQ+zVUcF4<$Hi;%#r#$VL`4bnzQ=1j$cs(aq6tU6t(Dn|kE8f5w
zatAOH;=p`XW_?Iq5*`hJI7#y_i)H~f5x1$e4=~_&Bm<O{%Rg&)b{r;Cbs)6GPYr$~
zhD(M4913Cx6agyUIf|;{dwaXq6L;2-JUCX|A1_XUHo&`6guGz{a6wAloaT}X+z|t3
zf*)Zncm=eU_}loaNSiV+*cn}OA0tUP5Oea61r<8t7c-Q7Xf^)h+c=K#8o!o4{IM<`
z*c8^8!tT|x`tHlq6xxo?&g~e)frbDr1HQ(E(PlnsyJCvEI{Wcow^4B7XbXGk`|}xw
zPyn{WYP)Q2e>_tUO9vm0N(9{_^I+_mK}LOcd>eLP`BO5l@uK79g1-~RI{=<ft?cwF
z=H`g4P;ee%BRm(!t57!~66)*O4lbv`mvsup99Mv}y?j_k-qYQ4x4Zun(?uEt(%H83
z(4xB3S;rc-*0XQx-0$7co7b;?2fggm6z~qCUhC@X+rq}CmXDZbgj;<daH3rb3Ze5H
z2Oge27E|hL)t7VXX=n8Y@&0^Q9_Wi*kzoU2A}W8J(3{sI8!P~mNf39o>T=4qKO}n?
zy*J8j<hnmO!e8h<ZH>9hx{=r&G1NH&*<xyfP<^7v#s-MSvB&!m66L+TmQXcHT`FwY
z0d{}izM|DBnW14}gVXLfnUap*yK`r6d_}z_w-^JTR7Y@#=$Wh|$$kw7f~t559LbY9
zVl$g`L)SA12p&IHTdm{w{Pf22LA_&K>>BB0rG?>F(^)=Cm@Q<%7^RHaE097MoJP<}
zC8CZ#yqa;Ab*2{<MoC><Y&w4a`<tx*FH4U!De<lv7@Ex&!qz30H6X)VZqFVHVz%)c
zRb7ey+O937*-~>B2mCvh>JwTXdiWmUry!)`LkliN?!0;A7g~hD!J@{Syj72MGOBy7
zczEu&aii>LZk~juFM8(0frXX^db&9rLg?0_zC5{s!Vo(<*1%TejJ@BRx*<;E#0e0h
zqD~uW+caR|Gw~c(o;aTr<{RN<@Fh$&BiD$V{Q9EL#5xh7QQ|pIR%S&S$xRAa@+t1n
zxTBvqjED|(^_xmbu8>zjOpy43ogRdY_!|FnMK<jUJQREjFVi@0TrDU6sU=hf*eFza
zyl(4`0@d)uS;=7WK4IV0;AfM!@7$RNUL5hF4qG`=CsC`AQ_Q^=yLq&m>p4)&N`yQD
z)`^!5rL$7swUH|w_h)5d{_9^(JmB-DsEC8`#{4C60JQWE(nw7J1RX71QQR*nhpcuq
zv*>qOlp)bw`Ma6XO2tyx%j>u<bzR*<okPc&dwR;AoH7lf`PArjFzNX1n3#|w<iryK
z92%lUaSSFe1a1TogadA^FFMkD`G1hC%i}%AM=WB@vY;JA8>@_Pd^s3lb)G}e!{}tl
za)n{!3yd!BWQ>%w9I1)^p&`8DX^7;7=_T|%Rz36sI2r&YL#(30t>Ahb@>3QT+Usvo
zy+NS@p&t;z-wp_hM#Lum&#?onwc+7!N}{6A4d^_6o**Zo$gTG*+vlGJ`|K$LRC)pg
zGc7flN!uhAs^U05agnITun{#i@t?q7^3kfHC4k(!LCh#%<fMZf`C87F`s*!(4mUN8
zz30q!o1{#=&ZYZ5V~_O>-~XYKE0(vfCS9W)xdc}a9-bsOCXhhcQ-mQU_5vBX(4HX9
zg4<(gPqX4q!^{~}#(2s-i7d*Ili2PY5fK+yyG!4}^HLT^TS2=k)z@cxmfbD|)C8<7
zoh@tMkJZC<3*9V9R^Z0mn;9)}_`QC<^)gV(G&?BT5hG)RJh)6Zey5eek-)2e9_D@d
z_`J2?+#fyH=~Hv_4SoAXdmcjbUbAuP|EzfLVuw8J|4u<${q;94alfPc4d*20=XN<1
z(yuwZJsLq`XZR!0Am6=n=XY-}N68;VM&O`ZjS7yg>c$8dAPLy4by(ms&%h?q*ziSE
z?oIO5N>#e;@|^9Sn`TCXGFX3|6Yq2>^=IJhl@Uc7XKDU*^B9V4!qoZL0Gj-$EhMMi
zrg8V!drEG;3&NXi+HKyr3r)FW4m7U<*UV6CRc?g;r~d_J{Pu5P#-OL(s9Hfl5Z?<O
zEo^K=O7_di&10}n!u|U;5L~?Q&H9&a{(1UK%-wY9@Irrt`GNOX$FM2zGGp>g(1;b!
zf3tftVt(o3oPJsI3be4D<EAU#ujJo!AE2b_YVZlXcI~<vX>W@2yXr_;(_xQuE$ua8
zP6)D6QR(hmCmqi~Kz@IhYeP{uCTn^+HfFi4Wi0D|yp(guSqki!xcEc8rxuP8csp)X
zo+p>KbetzQ7B<cWBBJ)US?*_Jb}lYzVtX|E)~&X$+N!FAkQ^^3*8*JUufTB@2q__t
zh!Hh1&*gZ6DV>EoZJe)cTI2r$q2S4z2Oy2Fp^Md&uI)~?LmluilR<Hga2^S`SnFhE
z6&){A@p|;_aI1a^P*?~3ulxPrJ%Z<HK&Ft47lfW;4argMSx}kSlHT%xC&_Mi33t19
z>m#)nY@pcI9>{i7n$*Y)>@{Z&@vHEgbA840jT;tj$h!gYppJ#MjYS#O<I#@Nj$XY%
zvs5OO!8uIPh{Z(ljPV2QZEgF5wBu`D+W*6+_h+Xye1zjgA2Vg{S!Z>bc?+P8o04>n
z3lC><ODdILZt2BWk#})Hz*yJo4jSJpsSs~F$lkxcdUD5w;q?L|4oVeeeP})zBE70t
zh>5Y72{p49NNgzBsg9kZ;PN5u@YM3k%2{oSjy48N>V{G}@f?kqSoReGn0Lj~qYJhK
z7-o(XJuNLgi_o7DNClF11l#`kMcEUZlbDzYyUc)zbo^Fk=0@})1={ft^ZYxDcFsIW
z5$a?hG!(pv;70gsj%|h=#jfbe`&OR2JUk@8(cgZ)E%7Ysa<oSCFD_J(lWgGea>uPJ
z)i$~3MR^uSp0P1Kf!8ik`3Y^COVR60+BbHFYMy_04z<x8{VgN?<cs}ZDbE@9#qsd5
zZl2x8&GSJ+0RF{uz|#ObsiG_oQ9mCuEv!J+j%{=<baPhR8*%TTf^oN*zYnzLUjaEY
zX6)Ag4Up@Rg%=4j#-~p;gzrRoypf#9H8&1qsDkqn_ND@))AxjXPbD8%8qgNLV-2Py
z(fh9z45$Py$wvg{SBNY0*y?Jii91K9$#hhfN;3%_x1LD3qnt?#*G}2K9|`|*t3EMk
zt9+Vq$<0TdLI)^>gyy=K`ea0nAc#l(J~K51Ll#)|l0xUNNG*YC27}0VuuQlr#H-dD
zg*jdPrd6tVcN&Bn^0=^J1BAKU@h0>5jvc|s4aW!kO&dm|X=;9)QpF^MIq-SanGE=<
z-Q%x}pQCV3Y``nPmz--}(f-djbD+kyrNBL^df;yp&+=Fh%n?9?02ogQiL{XT0LL24
zKf$Uip8BK#ESeZA8TGPjdZg@lM@DIu%8_oLruq)lA(P4(0bUE{`X1~ABbP69|BoNL
z)zs`mWC)s8l#A)<Rk@tB7^1w5nRd?e%|YIKlaC(M{)X8JW)|(u2iG(#%+DJ*NL5^I
zv%ttYOwPoI;^6jZ6d8L#!sO=S9w9CF7ect;XN?;y0k&brrDtkt3LI8f_jZ?f0Qx7R
zg!k9Dc-l6>uD$=S&u`o&3!B%I|KgWSI<1dVTh^(cnE+VP`cvcAf6}$zq7*kcI*<<h
zj*WWZJ)XmP-$40s+7&~F4#PqJFmlD*hF(<l`z)h0`SIFi_fi8Mc`*{Gaz1WOS~%t;
z_3ph^LE!zQP|o4|p-Uwziu8>2Txm@Rd9imnUw`f74i>D~yDY<)?Flms;#GFDuR?8y
z0&idf1^X}@0Z#OJ*}lk3y&Dj@fcI5w>bTp#7cYF!RYio|T9lh5kef807&o`z(yzor
zoxKl<0g^wDZGv8bgdFYWNe;uYa*zxvSixe&p*W4&>KlfNW$JNauolU$Av(sU7<xt^
zAxn^k80@{XA;~Uei-~V~VuGBA0+;hpwvV*4^Wg2&$VBh;qa|x+c<dE7>*S*R!Z;nB
zjq#Y0yP1|c<#OVct$BrS{|}(sEoZ3d*=#XE3<!O8PYM}7#m$Y3ap_`F4cbvgdWd1~
z_0pO#oL~&P@HgaO);DqpTog`Go;F7gPR{h*ooKUFuGnhfYt8a<dhGn&M}D+7F>l#2
z3tgf#1Zd#iVHVHN;dNSBXrJwzyYkA(s)&u1uH9?Po5G^${TYQVpRmyjZoJoVUFf%i
zoZBamP>NsJt^(f$LC#EzTByJBWA90b+(31snL%-l(J=>kAG!edh;$Z2;)}^nK+_cn
z3QKO1l9EEe-b?y@JE7gyx@z}U4dEuM8h?1{iSR1PpUS&s4=nKu7>gB@lwj&4<ogwo
zGpMsLl^~@|y7fqz(u|^FQBZb^4ccu?xw(1dyz^$NYtN1yycX(=At#VThV)!NbMj1?
zB(xeuE-mf!ZtlP#D&34}>!6I>BxE{i{>q`!(>F&Ff=GqA&c(EgJSq<W$Q}Q9m;x%P
znhP_iE}JbrQ=iF&QxH*$t-E8f=f<Dm3IorEhEQ|6Bl>?}W2ZB8Rl~VKmO$M{za|+i
z9(;+n`2p2T(c7=_M}gjim=7r_r=IFa$XywOFc=F{+^`?9(Fbn3Xp1<P{~^vPrLtzF
zs@<zf=IVp+ZXYQTx7Mh&?H(U|Z58a<0>$5Ev-1tmaj3LPMdYggh)A<lYWDo}tT#Gk
zn?L<0JZ?VbuR)Q_IeXabY!t?B&whTnt@Y(zAD{a?v%7Ld3{1W^EbLD5a2knBR+}yC
zyWh7C31yiY=Sep_^bAOxmbK?aWENUju^8mS0ZqMGdEOxe0g1>igc2v5H}FOF3WBJi
z1d#MGlcIrr7Jek;1H>f;2>t03;OBore_7;{W$qOdN!evGc4(@T<0-+-XSS1x>6HUb
zy9X0$kg2;)V*13&vP?>fW@EKVd^nSowBhF{6{;MHt{YZ)hJFVR>X0c(VC&fOlbQ$6
z5S!UzBHrWV<Zk_2@0isq%#Xs;jSv*#h^i{Ub0`fpGWC#jjL!+%q2XcB7So@uqj`pU
ztDv+LK<aNCrr-AGVwFw`ZA0|p7BOSjRg3Kzee(0z<1>mX_zMLWN@BTzmS6-ZHDx()
zn2bIvDLgkr0E5tK6P?IK0BM{iZPwDh0!0HVT4cUh*sk<bC?z)m2Lu#WXX8KG-_JY1
z?WT}$jyPb8nBY7^TMG*r#%T$_%b{#*#%PaG+DI=hjgEcNupij?`LnO?dT;M+*F8UL
zgB6rh(v3LEgtnFUU$4XHJ3#IaA2jrF9wV(xkSePUu11y6#cSV2=9$WoBR#)=kI#Fe
zEdOGSvmS29YU<Y}7m$z8m*4?QZhS)@G_u!~C<qKRiB#oQS5wOt^j+%<_7G`Ia8n?g
zti9c2RlQUH+X3P*%o`I$y^kmaq&wBtf@DLd^G^a30wlIwnm`c_y9VsTu^VTTKb6Mn
z&Rc$fEuprqZWzM|^MVteCBk4J?XH`ugP1)dQ+^>&0GHn0it}f~E6=;Ry91-ez>JEN
zC>AjOu-0O4k(CAIj4Z1H5Jqj9(Y+1OqdKES0DFp*sKG&wO`CE|%PFI@*a%W8SK@*Q
z9eu>QA>vCZ3D^&otGz5R9!M43WqznVSp*jfN(@vaBFm5z3+<FxjK-Q8be<(U@v3a9
zd^8<RdrTNU>7yZ`G2Kj2PfvGs1&LbE52Oo-(P%4j<9Gpkk`gAvlN<ec998|yN#0st
z8c6pbi33_U;yq{e2vI|Vt$5~=n=U7<W!Wi!jkW-m5Rf_uLV-UfB|V)pA7jSDkM981
zaz$77a}2o3*$>r`@+ZyT4v_JbKpeB5Q$UvX@?ywkO|3hT&DMFaGc;e)PH!f)v+h|@
zk7Fg<L6o5pbGJZeW|CuG2a=_0nJ`Iix(~B4{8TC{u-fr~xBIfhlSn@S?!BR=W{=1V
z1P<n7fUNbY!seYvcd(G6S+4P|DpJ=Wb%M;m%3+<K2tth2O}bvpxpvYkC|MSZn9y@d
z(<g>_&)7hcK;4G08nDJtE-NN>V7ju@?jn?xu*#mAu;oy>g4^WUN_WiGY_k_C4a8ua
zBeW-4DacLzqp8Jvz<aA*5e_hP>&SsRy@LgbQ>cW&)R5DSMhg2LPi?|z-B>k2K<ki9
zgB*^o{SEL5PY(JB2z`LGNR_&$78U&kf3SwYb<fSY;wXZX22<uR5V3;k09$@TUEQ`O
z`Jf6UG%&{0rZ!ow#&{mt!2(#(94S^`Q%ju&qL7y{HiqP{$?54kwWN>=n}&7XUqtGx
z{**%y<TlDyV9$~Ydo#ag&U44kJuPJoOj-%lA>GypzQkOzqi>K71iM;=v<Pc$RaFr6
zFmOHh)8A>6GJ>kQv2DUYgGi{hn^#`Nys5_-2z<o;M+?iHqgN5|NG6oQewg(2uF$q^
z0JMf-UqRP|?+b*kC<~1O)|$4$IQ5;lJ^cK?aAEM&A`_86vRX626Q>`55Qq1eZF0h~
z6rvCuM41TW@>D-`h{B4T)6upr5nK$W2n8Q`gc_!TFj*aAW!qjH%Yo6W2u~Cd6-5**
zrJw2hTkNdMcgQ`7syITKHdNsB>FWBxjumc3Zt999`P!I@<)zL(i#JTym2jZ_@C<K~
zlR_vFGxNd0+_h`hYFGUJMSIKwUmlAfCAS>w^D0v$1M205CyY=(s3!%Z(i#}1*VfjC
zPzi&#V54)VPzY5)mRO$2ISdt}me3Z9RQpIhhRkAY9+bW?*Z+C`)BJ@#l4r?$-S#5)
zJj2W;OGIvA#M;!|OE2}GA&e(d?TF6oKvI(>a8h@Fz$e0@qBO({{aX}pr5qDxD@(39
zj}ij{C*JaVa8<yzV39TsH4|al_4YdIf~_qBP*`HZmMY<jd74*_IvdN$SNKtlr5`9q
zLHdW4mb&cs+o6_ht4;Tw&!=$aA`b=|d4)8H3WQ4v3I<ekVYJb&K&gN<(U-xBP|`no
zq$J}DG)+)AU)nc|tUl3F>(DEW8Y7O8f5CK`5jn$bDy5*Adi>@EkC=tEA=4V!fTPSi
zg^{M3ynxg~U>NBc?YH!P@-8kdsVb;ptLkRzW|cDdS>h2BcN95v*yLTmt&~fO6IZYD
zvRo^WIil<rd#!bRoSDb&O%7USh^}<fOENR-EE`7v9gmJ`->pJPh)JqZ?-<j6Ziwr3
zC5LxGWt6TjlGgjN=C2#Ltq+^@wFEa(5c5%c+{Kc!J{gs0>+7c+6Je$D1!vhL+K;!}
zouZgKQx905P^4Ob5?^A2&>zNI5=HBYT{U=XQm#`d#EBqWV|w`}#iD=8%Z<z9OtK!V
zkKMRsWB}`Q=6&N8Z=ARcc_lD%^})HYFvjHQ1K3-z;8S|yGyYfwG)iq*jg$Wg=GMTV
z5Ig=vPxaRZjF!~-8xR+hACWhFHIe`m^W{kb4do#m#qjJRyew|hyNIs06u4W_Estr&
zK%OGu@O!G>F1=@LnUaq(QcwjXB~cKlTLGbA?IJ2qwOD~Wv|#re=%y&9#8^E%x}IhM
z&Yv5R*Dt95_!PGcLP)h5nUiv<0GA(s9>*PE%M}RtDDbqOZvA=$2B{Gpt*shw&OKx3
zteJ5jvYoBMMLTqRE9{b+CxGx^tYS(aW^_VbjTaoH0(w35V0m05Q_%_9PUwKJH7h;J
zI@e_`Y^JNMOjFp4I5cplSj>0S1kp}{2i~+Fd68ZVz|th$bhrCg@dWYs1pKZY^)XaI
zl_VoWM@^Z$`wPPlI7U=mKGP=c1c8is#4uz;oSHKG+Xi#0-7C#lckrZmhagd@xgetm
zxhRYSy25+iKido9bcEK8=o}u#fca0T%G;*IRC2xo7jkEA;V45TCSvf?3^O5gRs7R3
zp&lZT%!+c@+1Xie-qy`+UUz*b!dUv9{_5^hX#YtM#zvF#&Uyp~{V-%F*8M<)N0V7K
zQPyRT+J*e*-_^pCk_n6h0}|mqc^RBw;~O3ty0!eBY)mc63PFm5B%EZ}{$I2Von0C{
zozM7}8<VxMu#=7hNoH7U<Ju9ucv&L$Bep^I=fU4J3_#rXmtVX@tRUSxqTdSu1lnD=
zM5K$J(eX*5;KpO8Kxos~`uaX9alil60*EOn+^;i7j2Yy#3KP`=UrnrFG`JfRgK1j$
zSva-NoRQ<w0@bgnc`Jj-0>&sP^Y9$u9UdMUg743wApxAN&D7_+vn%!TqdQoFQF%bY
zc7MM|eqo`((=RXeMR4s<Gr;;u92dR=S?sgNkKr=s3+k<Zz`13MjG>+F$C6oYdd4_8
zu`c^K|7&qnRDWDXNA_HysliRVJ_Cn`QT{FdfC#aD54brwjZm=~7_h_D{OJ?r*Q`V#
zh>M}=gU<;@DMXWAWwOfq0g5*1I2${=+`ev%e8KjF;|(e>XzY>T(FQ%}XL0`rn%k)e
zlx7wEm{vf}%=+~mhruykBdWRLVg*9K)O!gfu<8W`%YD#Q3VG}@uZyE0^HyN;Ma{7~
zwA)}*r-mNz{NU%EI%i{)-nQYOh=_<l)Ck9jDW*bT^q+Vk#y8wK`ejujs&nQ{$+?O(
zVE;K6Fhdz;oD{tb+_irSg!R<dXy{ly7lsRf0&wXF-$ChVrcS!KY<(_ucfDJ8Qq}$E
zPez;8?P=?`|2Z;>A?$!f=eXMrNZ=IX$ILSn>h0|b(b35$hmDOXGu5A}$D+yt9&Rjl
z04FJU$*XpIQBH+;hC&4y{^Jz|(%7FPycUvS48%Y-hH(4?%--{udDeAnUrcRgR#prd
zj|e*Kx;qAJo_S^OM?Ll`6lsuYL_DQyrk;jyGH@oZ+wc9;1fqP}Yif2Pj7FiNdt5Es
z5QAjg2-LV5CY4?Gy&q#Memhu6#UWyN=gtl5uj4#~Jk#B84}rO!N47@;FU5@k9)y+>
z$vBut4af};yTZ~`^Jj&{>c6E9zk^kh5lj`2<poVQP*!K>{G+MZbzghTKGzj~fZS3T
zzM=C)d#a@srjlHPuJ-8BPbHL`Eg4nD+Drhw_v!O+FRwguA&OMGF%$`&Fx&zbmH_1+
z8<e^c2loy!;uwu7BeO-N$eWwM@3V>;Iw98yB|7Pa0X>TSIYSpQQxTcdI7)zCCd)c_
zcq||`2rN1>l}Hi^f2^uXRd6HNZO9g|?ML6cUv!ldQx1+F2lfqhG~5a!!^5I6wXchc
zQ2ZfpG^7$<FT@>UREUTpXJrzac@Ax9nlY?{_17siSvDwFZ>02ln3J=K0JHS?YA2fQ
z9_!?nFCE~nVtvjew9^^=0tC5E-4*|>4BCNPDue4p?Ed|W(CyH%ipV<jpb*qR6bj;6
z4Yr8?S_c1zrIyhL#}pERAOxW&VElr}17u|nCVVsdm%qY14q(GWL!TKsRh<X-xY)H^
zJMrSwsXGb@QI+d2Ta*8@xpZN6mdZR_6(tqk*iD;G8XIeFPgKqKPQVYmAQ21gG^XU1
zLW?|-#4c&B03Am18+vQ34?rG><=4rWg8J_Bq+{>LBuZD|tfMI0FbnL^A-{vXjb~WY
zlbi4^L7~OCbssf>3Xp-04ochBe#Xr>hhM$0>JSO3<TuL1s@#9H`%p9EspSX32nVe(
zQ&Cp-({d|37b$T78Nj$5?m4evOOldeA#QEg>b95FozlXfo0Bo;p)t;T;e(#(dO!75
znt`saL#Kqco7dl0=Z=W0kWO&_=;q-+;IS)=ipq$I+tZ=rO=^so^5-uv3XN~IV-<ye
z5dPKE1E$MN<2006*o=`P=HlpxFOT{M4-Jw&ZxH5%`{=?{ny_NM7g{?IIVy`hD`+O?
z=dpoY9{!Ps=T00<ZgTQr0c@23;RX&6Ygec~$+6Id4i^Ev8XCs=`%^{Nyjy$-z~Nwy
z)tZFL^R_=<jsM9iUG7f~m$qxSxhF%DbS{#B(WQS_<5ozBKoE{^krw@GQ*q~!pH_nq
zPxxjWMmZfQxrpn7(i-(R(!}lU8{6Bx-z{;PJE!DJAjv*%U&`x~SKg&_JK9L6q$yo~
zt{l2WjH|neHy|yKOe^$KR8Uw1yDH;e-Ov!e53qtNHzegJhbhQvI~=TLwtq|BrI?Jv
z#$u<ld?qcXkWQ3q72O^!=nKr{OXlXa1QJwxW(tx(dm~Zl)7wMX1LD{>LP5CB?9#^9
z=q<^iPPcEvkTwHDANG)~cg`hQJpynhxUqbxc{E@|_%DFtOEKFqHU$GmAR}j4Kkup7
zl5BZ9<LFh0!|~+^-bghJHB|js3X-H<@-w4m#jQVgZR&FuKX-S1E6tbkU760Rau$TQ
z<2q)$oO$EPPE0xUQuyjU-Dc3w;1f@t6mI!z#^84FK{A?&+i&uMKO)Nkd?>yGB=l*&
zB4j!|o8O}V3otz3Lk=kVmc?l+DJ!dg_N<-t(L><}d3V4Jj)53Ak6^Qlin(W~S>0t{
zP4z-w^Ne8VLm$a!A5x`?(z*A-+M4LKW;VS_=KQoZ!uXBXpo}`{hZG!W=gIhYi>{?z
z_wZ1w^LHBMPm#sTK@=~T9;g%k#TpG6KKZdTY-Iw$dQVd+T_>?)imoGTFpOeT3;rX{
zMvsP4Eane}ZX)gf%$Z{oWd>n}KebY6Iog&pCH&I410QU?zmI0JFJ4`hNrBsQwoNt+
ziV<#CY>KFcP?F-6`C=)B?gweH{E?r(evP6Ng+xQxc0HcmK*d*rd$uHb!==lY?;cGh
zm^W=ACyZ3A!5fdr&GpwGJ$Z68Ls!Qy_r(8RkUj7N&Yxxiw)(=<1$b*}sL)y%{Viyt
zPWZu}gpgXIHNX6K?cX3vDYDS6-Cb^)LnuK?gp3n;w?*(2?rv_eb}LveXip=nNL%aQ
zRKG^umtfHmcncpF<};CVB#MG$x#~%stw_@~u(rZ4_WH)eW(r)7U<Z`aa|oFbrzlLc
zu^l1ZTPf*rX6D{fz3kjv7NmOhVh5t9a*%Mb)kSr5@StzkWQr2g11+ITLs(_Q+s8iN
zotO>Mi0S;aP1b?>sKz0|YwVJ|+i0Zi0#AtEx|R0i!-o$D^o+H7h%ABj68X#BP0i?>
zQDLVyhkBppl~Va`GlQKSW|xzOhQh2N7bRjrZDCt4{F52;1_?loQhpuFmgg;{IHJ7Z
zZ`+BJOji;$!HE+T1ip7Km8-+<8juX)?*PdOa>xjyewLFnu;V@qPV3hV04Rb!?Kc(*
z+IA!sWirIY`_%<@nvRXJWgem=M9!?&hzT<fvit$&NSvn(^3F&a`Axszqb0Y>f&w3i
zaD}YtE!`hASz`adXVZRWvLUI|7kvZM(}mBHaUAFd)R%Bc7UmmF;aM_}Lujoe!11)h
zI{v3I;5c(Vg!niegibuaY%3kCFR~0<j#J+Y$Wf7Xr=nN^rvAkXpVrX&jLqxkdNy^A
z^YRkCb^a-WV>5v?be&TJFW4lNSow2q;A>(^m8PUaDN{NZ=>nIk3;j_^j&+B~E)l>J
zid+DLp(-epaO_>mDw%=m<bBk>c1y#gPh`$#tQAm`E$r-Q4l;Bypsz%!Yx{)Uad8eA
zW|E^|-QA0DtQw1r4i4gd+7wcm!jGErDY8|Wm@INyin2rzIXN_R>*mcXj*hUQVaS|-
zqn84AxZ$2l8Y|Kq*H0GYzb(3Y+}N`tnuFnI@wz`>%aO8h)(1go$+G4FA^<n{Vp2nE
z>m~1vfE6Lxd5j>yswR>+r9CE=ys2!33{8j08?x*6NxVxS{=NGixWGCM=+Q3pLt!J!
zL+Fc6y+U}ay?yrld;E0GHz-Ns*L6Zr1yJGtg7xBH6;K86ib;+LTO+YxLqjKde8ID&
z-X)<+;p>gGMI*~Tr5-<dVS+k{$={f<Dx9kKuu?Z_ugdrDF^eUxWlaZQqs?_3A%saL
zhPzv%RDUWgulTBu)bTg^Ut*!0b7^<$vqQ{?8y6rAq=p(=oOllTnC5m56vG!KTA(>H
zTNr3$IhYxQJY3G14|kqDf`$QX3&a`aQ%9~qo`cuY#3bSPj-k;}IFm2}n?&C-ekI(!
zt7mBVL|6DKT`59^@BxR3*}o7}q5YDMA4VR0M1=UIuOh7rJfhMM-21c~o8+$yvUpB!
zM$W>ZfU2cF@(G~+1onVM)!f{iO;LEg4?4}M9m?CI+g@GqZ|DQ$OX(XDqJicF=3y)g
zliJx1AJ9cMKXOqJi|pEkbzx&^iJ&C=LNa3<?|P)yF2Wwg*xR>C6RyB(R%J;L`ibgi
z`wl`|`l7hlcn1mbURUbE66>;uTK~fN^HVcxOAsf|ns7Si_Y$^xCCd-VEw3&WJb4ma
z)s5Z<R4;0`#jYTpHN?iK$OQatc;?)CKeF$8V4|mIuE@l$bQg1f(Up@DUGxdt-DRUW
z6PY3-P1V!>`VyAbag@<xJ>6aZA6@SO&t=>Hjen(R7)2?R9VH>NvPu#q$_kYcq0CBV
zp=2eyjD$oaWJUH!8A&KxR`$pqk@37w-S_kS{;%i%d|$uc^}4U?>iUlJJdfk}eBSGW
z=)|INw&&JyPhKFE3TquW)!}DnF{L#KNzN^|Yonw9J|kKY2OhTD{d*2^URs}vb1+Ez
z)b2=S7f46JHEbrKP{&@?*MOO@|FGGZv#l2)1B22R9$i4WQL>vRfp)^cK}6^+Q9vmL
zPubZFeL61J4}ArXvQ{`R&(6&O=TFZe_hGT&Iz@?+$DdM&Yv)$rvGC>aII@}E4y^7c
zN+v-DR@%!Z-T1Z$X2HONZ_zSVBB41z!k{n!dyGuaRY?BsL-QtvTzez;V4lN|JIP*V
z?V6rpKU_0e2`?Iw<jC|vW%qGM9)wcZ5CA?Y!2640|ARNjf!F^T&!5<E0wl}r+tWv1
z+!A6-8UphSE{2-ip0{`(@ZV=FS@H26=3L-#up2@lV{^Q4Jy{*Z03oymG+|Bt4MYSc
zCn22^bsSK$EBpcn4??MUA*m4~CIN@>L@I-e&;l!q<%bX{V=JpQSOhnXcf?V+BG#Yo
z>Ybx6j@=o+L&fBaM3-}C&pwhhfRNJGb{@t28Ef6kON2_HZl|#s3I0NI67<yX!_VsT
zF_^(jBvAC{RpO0}kLtA2-`dz-e_c<Cx(y<T0FjxOl%0bMM*fI=!eq4RX-qdc%@J6!
z(nww9LL~L#XfM+?Bh{IT;0}{Wl$4&xuvC?i2Md80MzxHi#7DnoP^9E7hxY2d!;CTS
zoZ<fk!M=A=ATQiMip)+@N&juTv6dJo7b%2=;tb>nd;|TNU`cjgLG~=+O^7-8wW&V?
z1p+DH=~JU3ZJZ&b2CDoWyV*tZ##1H5pY)8FBLEjryWhpEf+_Wj7Y&L^KWXlZhG6Mo
zrR|=Kj;p&J%D>Jh0mgZ70<-aU^Ec!_+Y3au>4^eY5X$t?!T|YLNOOmFS-y1s2Cc#3
z5o}nPlRGl<JSgbZhvR;quLVAv<XY&g{3h{hb^T;<Z#_<qKoE3zisfv=l}neE-uu+q
z=Zilx*})oZ5P^AWC|8kR=yy+n9@*B|i86TGmE=X<j^reYwi`aGF)oIDqOa6G%I;&J
z#w@CTfbvGg$Kjqq?OS3_QgPO6G0(0)Pl>`GhBTHvdrU1Y{k*;Fr2ucrly$po$4Ne|
z-@>ecC{viB;{BM$%RkH)aE`A_Y}CS1geM`YC0PSIpQ*Wh=MGuPS0SfIbzyZ?&!niB
zs0V7r@4P7MLFYqW0v9dI&+gLdY9G0?13W#6Z+m-1lATONo<sGMnK>i!wd9QKEPCpW
zM5}4mhh;qh54Xt&&V!@`88CSuxQZ)6nu;gMx3iRS%GWqW^H6XV77})HoZrSdPfN;Y
zbUI}Sm1Ddh7AULJSUU9n(08ulneIo^=bN32vkUYNnLavNu_zjLqsYdj(rSz(>AB3W
zz2U(FW53MSF>p(JWt{J!OGnT+3hP+0%3C6gZ~Zu90RUk?0D6g~n8;yK_Ac7fiL4-e
ztPXW5FaeKqzpAJYGH==VkZ}0Ka|bFQ4D+yqx5nVUlWT;S=lMr->TbB7TUDeN<Fc~m
z(kob;kS9aTy@OPC<PF@I_m^g|pJHkc`ecPw-A8qQZrOT#=zN$E%@WV(vuUDG+}s2l
z6;f9c!4n-k=<V;{oMI=+2cz?a3)TzcR{`JO6a^X{_xkm4)6nYb-wA9;s32jL`)agW
zYz2II7P|d_3PSy9xy%__-37%8<QUA%#0c4Q@<KWq6^9+pK*<Wb#%`a_2;#+5%*f7y
zCA3HWdyj1(0ds$j(~ks43XBcgl#F5~#>c<a)(*}rLXh=8Xvo+fb0ND=M{`vcKEoP;
zdxfq9M_EwvU9l$C{o#Y<*Tm(NmA}COE6PMo`oG%*)w=To>#Jb41hq1b-Q&1AP?+zh
ztD<B@egGxw#?FXOsJB=Em$&b}A<_ob&9S|HDZQsRYnxTj9N>K@P%P5ws_ZFHM5_K0
zhoyH9$~A-;qN%BsoOxmyW+4wY7wof5@O2xYb?^=q-FG5rj1WF{74Y&%N5Qes)TG(5
zPuwV2cuq_F1)2<QCy&E=<1E{^caMzlK0E8naM;bMI=hC_m!XJLLLwgtd%wGi_4pX{
zFxO>qx?lY-P<SvdJ4kqdAky>PHxHy!;A&o6?xmfWY6O)LUdSH6$h6Y}ck&G`+CcnE
z=MA6Fn^!}@+x;sbAj84S_Cc~Tawq;MfHdLMU$=7sOyiu5X9}3x>FC1I4vT@&_3LD=
zA3iMLDuEokyJq($DJi0F{U9GYX*9}WPh>$ODBE#$aj<H&EF^?sdKqwl!fX@(`QYG8
zn_<=VTKDZt-oWndx;C9T!|A!a+`YC+>pa7}%?TgLH3XhyT|ToPsQcfA37&LIWVs+T
zgjjFzCcuOb88@0yHARs(Y%iXEcsWEdiN%rU2?foZt^}XZk8S02QDrL{6&5aMNZa@B
zeGwG2fdo1hLb8`p#fW}Z;x<gz!aK0quFd=-IpGW0yKf)Jaj1mTVfG9dVfn{#{)p7&
zK3xw*(~ylLB72FUl_{4IJpIW?2$eU^KW>ZDwd-&=PtHQL$4^CXUtiI+1t@Y!`ZwKk
zbyB#GjhvBk6(A@s?%-Tt-a|vCg%R2fDIrLNA@&j$x$JKx^w8+w22KH)CC2FjhsEXz
zlm`hGxGw;paeiJ~b$G;0=S$(TatL!gx}359BgjQCd^-b$SyM^K;qE>o4HFDR<#Er)
z2<;<KfEx6^!mk8{!Q?wn==QR5Oc3&o8U-fdVl)|mvk9m>A%O=zpUKJX-g4?%$^j%|
z5+TMQ3b^vD#i{<#t_*qGt69i-+Y~G}xL-UF$fE2l!u&X8EA*87{M?=%SD(kJCc@6L
zk#vRM4mBW`?j~1!YvRzy>|l^kIXPiUy_a=&MwaSpt?p)KwK*;&CYBAy5)Lyc&wtXe
z*sfql)CVUFiV&VX*i+-|-s3%ri&;q<Hf)f$doL7DMNHnJUU-iOAUiu-@!w|5A{Q%2
zuCj%idSzw8r5FRG(3?X5M})w_?Wgs@a-aOf=JQ)uSJTXPBeP0W^n&&lliH7%T*X{S
zPI4oLjm6cQV$niKJixp#I~x$02Rc_h`5bJ^DcYGH^08nKZ;A_qD_B`ryf#+D{w&9_
zgM<x`46dEfEBrj5P&`m#^ynU=s%}-i`?Jk%BP1JUxY(pRT+GoX)SY)x!eSSsqm$(x
ztU1KT$CbH)eeTsOI}ReuddEmeg_qB(b|ZZf?;&;>*74#(p$&ZkUH=mguNlzXNE*?t
z0upm~w;AaK2BrhHA(cQY%KKjWnqzQhU$MI!2YaB{>DQ`}7`U{trL5_duwnAw^%}~}
zRD|3fefAElh=2eR36+RZOzJ^YbvLt^NK^^(WNd<h0Xsj%|9S_Yj)Wqv!FccO%9<Kj
z16W`yNf~FSQ`5dY;j42x^^(%JptrqEMVuq|F_D{+1PtC0zc1!O<+XIaf?{nwJz5fu
zRS&CR`PdYlteb#g8)6K6UcrRTL}S(W_;GjrnbIHhzkcbDU)j9nLBMwVmAg4I1MjaW
zC~RhAKPY{CY{`Ly3DaD;u9&Pg-);pn652gg9UVtN2v$Xl@u{h_Btj5tP(-1IvC$Y5
zh>7UiHxsGW0f(ue3P6fGU=MUX0NJ4K!l|gOtIN`=#vNRR(-n72-uip`zZ!$~RlT;X
z|MusWAp&~3SFcJ53Qpo^f{^SuQ-?n73GP}P%I7r}r|0K^9fKF)UZcX4O`&kiJxotO
z!pBE7?*1H1I>5)hG#KN6RK}ndkL2hNnVH}kGi;Y7Sp;BuRNt2npLIC#xaDt&R|H>$
z3y;XopVL}J_j27Gq^cOvXZU(#I4&^p9emu`&|svsv*V`D;esg&xKLh-Z~57<aqSP~
zy&3A|uC<f053>rt56?nR2zRB>2XlkfMQjwH?D3lu8%O@Rlt%&z3-22n1DY02+2_qa
zGK0)C8Bz%%Z5a5pwqvGUP>||{3vAmrBZXdBG``*xAS!5Cm<%_#Vj_@-yF2!)nzmrS
zyX)13<8}j*|8}M=n4$wy@1wZm*h_%`j6M<LoHD{B3}CjF@#0xJU_Cz>XDW((8=OfL
z?NKMu5+QkX-jCWSRN4Z<y4t4NweJ3$wv&@Gw6ek&pQ)9FE^=3py!DK?Pa#sV@6reB
z;596U%fIKEdwj)}SS&0(_Mc-D-~H-U^6#S<?83DfL~F)!(^nj`5U!!VR|rRifuukt
z1_m*g)qBD&)Y~x68?z#lG1iYe_}KB|$2;2XgJ7q@uMQ36*|`;{47fk^Zzws>ZCHK(
z{yh?v;ez@__T)dmwp*Y0wJ`{{LZlP=ZTuDg8+(137?(@f{>nTX5}ae*Gm9X5!EcZ%
z`R%?tVBiBx!+6Xq*@GH7+NdW#c_sPzBkiej=^kN;!HF93=_)cUc*!k$9BP;&Uk&Bv
zP7Xwry$@8-AJwd^*+g&FjXuI)b`gtEP5cZh%&$Miu{#r8lwj-zlB>)bhE&?x%G2h;
za3u8^q}v;2sXXG2EB%ZEu1-LKcca&VYWDEssT%+3ve$Jnw;|1dRvqV(G&H}$wUUM{
zu4O&&V1T`X&#L32^xY4o#56Z~%qr>b@83k?Ius!v%kHW0NU6=gr1CBCTXtYk0Xe2b
z{v}*|-(DQcL_uPpciqI~%-g2XrWfLdFCM=tp7L1z;t)GM(^J99&RG_JXC3y5aT4_2
zCu`aV@^!tn-teFsq@D-0!pj4J3QDi5o3Iu6%8M5-JYf_DeM|Nd9@XS%$h5iIi>t)!
z(cItU7#&0`)+}A-WlJNCiZhdbB*-75kKa9ukSEYo+P|RlMr8K6a>Y(IQK!^FoY#ae
z6VOspyS=&3tA*K;&pshy(zmYl@et=Pl(xjBF$xBG8DKgIoZCYpsmgpF;?^zxb6Dtw
zYpZ{)3Z~?dVKhEtq*{ov({|*NC%JpT*j6gro6XdYa%fbLy@a`sJ6K!m?meT411QpB
zW8IV!;KdL_oOeMiE_!&Mg1CRf#Ze@s2p#wy-+0sK@nejkOyncQ9Xa^evZZlo=jqgx
zO@?-qPx57I*K33A8Ji$DWRE}w5n>8SfIWaEJD!Ps^-3Yx9{a6ybgGjAbYw^|*4HNy
zc?b~kq`_Gdxv?=Xf3&KX(34O)88LSs<V55D&<^ue!y_Z9Nnp^zju=w-%3}bAk>DN#
zN+(1+GuZ~fS_q_#<O;Fi@Zq>(Ogp)7U}NO{jlCClUggbtm45xu8qqaA3fON=D=Ay4
z3CRlJ0r_nof0(|<SwzH~<Gr9e2j&FO0Bs9|TsZsr29Y(&nf2o!wilxX3^qq-d*11e
zN-D{!(sCmqU*m6?dlQdK3kvExG~P1n+L2}7iBU~B7$t`>C&guT8FkDhb@i`NQdnAT
zUm5p-)_n9!IqJV!fd41O`FZUh#YtRZ=RcDMsyn6F*&mpnFtxBKy#H5hF~3!7=f3eZ
ze^Dyd=)map$6<4aRl^2*Ox-{n0&0J_)bN5F0U#YHIE1tin_|4rX{crvDmneZR3ZbR
z>=b@3;6{nNBkBMbfV+a_zw0F*i{sJc^W$BP0=IyXEi9md*hE4v)*-Ncy&bR%iaSg|
z-Tb$<`6iF#XskavnVH0~c{N{i!51P#{BC$Ba7Q*#QUXzrIA*A$!xUU4iG2=|2H+Mp
zv^8Nzkd@8aTys&T>2y6iH-)6+39gJcQ_G8Dk?Vt;HV^NK%vmB5Mr?n7Z{N`dVhjY`
z+VY^x+TUX6$ekv_dFg$(XDHkeUS1Y{|DJ?YzS1wx^sB`N>IeGzAikack)XmXg|L(*
zge8t+(c<{x+6In2@`#~4;2(c)#x{h`(v>i4Za^jt`*4?eD@lb>yYbdG;&wV?n{f2m
zRN3T)p(vWX&oHbp(EOAj=en6n5MmN6^rssarl--9oK;rle%52-d*h;wzL3F;@t;4v
zJ(q>{q&Z}!B%J!UV(XOlt7iZ7WOdHDGpIb>(%O0q6eEOPRYt^s?uG`1iQgfT4%2;D
z%@&foWl02#ypb=W22nTRLp@IlJt*vl_aAkbF8X*?glYNL?#b1^`nSFwJ0IZhiSSF`
z;F-!7<Cb1N<T2J^#>Hz=_m+lps46-B0f8h}R$3!C6jWZT`hMKUFVY{N&Z)#rFT+w9
zaNz)vPE-7Z=KQJKxi8|VS$zCGQxwM#NwbPeO->4}UTH2~#ZbbFSFYTHJP?O01N1(&
zw$8vBxlYX?5Umk{+O4PLO?7f_k*lh9f7-H=(XMlNuhiD6s(=WIlbol1gnbZSo+Lkd
zG|Vj1ntl0iSFh;$tf`nArA?$&^!8vOAQuJ`O~7M-J;pBr-Xg|<BgY?OjlfiN*mY%1
z*=#;9FgKTG8JSyCzJV<H>NbkT&)b>q`0Os58`T$CVTq)Xusxy4kk;BOy&mQkllT?y
zkG_vI30K88@yNuX7)-T7jsd9~>Or6#F~i7QEPa`gC;fng-+POZT6W=_vd8po$g=Eb
zFM9a`giuiS*sTXUi@}6$$=sv$n;Bcv#AVA>-9QA=6k*!SjP&0z;qQCcjdn=0D)l}3
zVK=tt{reK@O<#8~-B;p9>vK6<Y4#`pHQ+FtfWsXA`ryp-)+j;Rm>Ycw(R|Qse_Wg~
zJhqdd82}xA$`o~7SXiKR1t}eFCC{|+7Ztl~NajrwC&OMXYV!4CiCyGa3sZ@J9>&ZT
zEe<dXAGxQi+lhHn)FkvWnYdK!z19xf{+$+Tktt-$1lZAIjjF#1s3wN%s9p;d0;~$c
z=bd&Yf*Iim97m?|2jVJoYVNY9-NKk;uzs7^A8}J&4sOBFDi~8Hh&-~b?gZf7g`N!6
z9OOd8A(w}QA{7<d8;gKf&YV8oYJdDueBiS~hb<IiOCRt0|Nb~iS2P4T07qa0H^cb=
z#U?6IOUrkbZYT+1m$U=MlpVb-<^~tX1|gyDpSeAk-|5komKw^iJ*r#zQr**iGRJ`)
zVBP0UzkVT{NB_6Z%X)4{BjWBne{LH4j1kO-3qnA!Sn>XNk&s%8*Ay}yEIRU@Rqb_`
z*tB^E^w`N`nD_`qDZSK-boI;I5)-Yh7=3PC;or$1YdU1_dn1FBl@|B{jz@2nZO^X|
zmMrPii+}7Z`{7_NDZ2l%eu(w#*@teF?A+z?tMxe;-wQV!juG&2BEy^AWeG?1pQ#fb
zk7Q%;Xcrc)$kA>*&wo3&fM^rTfAt=@Nw$162ip}A+o!N4B42Y43r-73`$>IV5_Mi=
zdSdvBJyHh{_a7#9ivAkYrb&`@|5o!BeIbv*+dT~jwQ{WQRfr^_*i~Z8BI)vbUaw~3
zsW2iJ7tguJIPmp|<oBOH3t&wmrV}{~gqu7mT7u9S2LnkEy)|lZMvIwNljys4cH|{k
z(a_xLuQ-K~z#&W&{pBaYO@jfBDznzJ#3z)JSi&EWNebOmBL72q<JvNolB44E0gQ>S
z5$^)!LLzeduqomN-3X}q)8Zw}9mbm>fMWGnzSolbqf%$-?r3+{nnzli_Z|k>A1#@+
z&waoLeAwNT;`Vr}_1#8^U+^tLGOErCc@jDw;ITbP5EVFF0G0n9;1RSbBs8qmwY63o
zC)S6?&>py)S?BK$y#XZg5ZHy?Ohadl2Og(6{t2S}>~Mikqhp8u=AfRqe9d+yp1?rs
zdA^3Gk37*OJ^~-3S2sG%pSTL*-MwBywl^hiW6_H3ALEo10Bas9!h4<|wt_(5Q9gT6
zBd`gPx&H352L9=d;`m2Kb623*iCooR_FHF()V0LPRZgctT>S_aa$J~U>9>`OnZ{rE
zk&Ulk2Qr09E&QgMh6emD1F*Q8HS-Idf+=C4%`8=F=BqID7}#4fLqnV2h@UboI4xIl
zr}q3TTL4)K!{f(nJd}9+YpBtW4y%^Dwoz;O(8bicVEMLRX4N39jvpt>*K`>59x?uO
z`w?*$IOH9wkY69Nvx6v0Y1QY?ibbq^KPM1KNiW5KU1#Xz^{b>c<c|eXqTJ?kO^8Yl
z$RfK3S_6!JM2&G~D->x%ek(6t>?J8ED>q&3k~S=Vf=mxIGNASVF+Fh<ezg5zwDZGI
z-G^Vk{Eh0fMkIL8=iOw%cFKz4mRfBYxs1;vs1L~noK1R3?>Wkmejf>;R#vGXa8pyo
zsk-6no6U1%!_|f0-kXJ(_cE&=F?Mbu3LCcDLRg!N^1vY~fm@Zr4d&%TVy8Brx3enT
z#K^c*k6~_Hlg-5+M;hyz#dnm+8YU-IKQ51q^r#KAm?YJG|NhGEk`7#9qM~FcFi|A@
z1ull81ra*oghx!Aote;+K0VvtGMB2Egd`2}rv|&Rk1(k;o!=fP`GTNHs@#ect=&<B
z(2!sSb*v3Gd2`+)w*4!e{mKz}o3*q*XXG6#&&Vb<`u#TPO-eO&S<BC**v(&(ZPC_S
zQDIGG8fU&}Uze<S{q0>_3OD-t;aNOM($Oln>xV6?Z^8un_cQuB49L52Bk>Bqm6_Sh
z!a_nDHjmsZGz(%x2n3#EDDy}qvHsG8cfZVlV4*>2VWV-uyg<{1klAXZ3>1IUi_0+B
zkgRXr+OloiIb~%yS(}^|(SfPB%ccX(hB_358SqszVirk%PtOtPLV{Fu6V)N4$wahL
zLW1XB_#%9Le17~!$Xp>No|6Is17jlGqef&9{fFg_jvdMVR1|L2{Y#6B0L#=u_vQfs
z1l55K13@)ONoKIU!cC{a+Xz<_SvNKp(dB&~Hm=6A2~YMuP%5pZ_<rg>7hiGTOUAKW
z;m>&_RL`d-7?E<8<hW-W*O}s;Wl8O~;(~&L3^jc03_eq9ChdTK(5Ygc0yzl=7JS=r
zcHycWvj8t3wPzp^p_l^$VhpEF2YagX?#ur+eC^2K>rf9XyXm;G60+0s<t@G&eV-g=
zr4605&hE+BX83-7{+hUWih_}SC8|`uXRz*nuaP}-8z_eol#$%QkiNawf#^8ba?|sX
zk(<*c)z#;qdMcz7#frwxSx-H;C0t29<mVYVIoRr_jSLNYf!*KTmX9E`Km~{wo<4v6
z_EN9!asaM)w4{S-%!L3Rq(}phfS(H7K3a6;$m1C5QOORwM^_g{2jTVN`SY`Lr;5tS
z5CEnZH)F!S0+8<>T9r64mK%I%$~T;Z5*pC%?Ad;hUC`Z|j_>EhNEV!TFqvSCq@Y7Q
zT)xwX*spRJt`%<I07a}jia>RM9!6pHa=~gYE;8QyHJNYQeyX>;Qf!(Ilv>OKpiJhv
zbyEHVBa=9W@#c*|><%Nyq&%H$ulM~)wwjF1bv-k+M_i+??F<he6E@+)<C?8x_NyXn
zYbM$pdr6i#N$>Y7%gb;c;tBypJ;4U<tjYHc{_X5>P_yubgKY!&jicAWW!fFSX*zb`
zRzW6j^1}hKqE~{)ML7WyPRt(zE(FdMA<zT_eFSs>)T}@-@OLN>F&O|%KDr(lG+`EM
zZr%+S@~&MN(EwN-&;WcKID(+t<SVFm0ic<vJ~BSe!pb@Ul?U9Ci1~)g2>&^_4MO(a
z5b1|ShI|B)D~2vXfE^lo34IY_CKb4jyQ~)MA0(<NmMcx0CSAmfjT$x(n4YiH9>f(v
zugQLH>)pWC+c){|+eBzTtTOsVRRD*{$m}Y*!Sk|?m?!A4Wj}@6Nh$qnw&{uIk{V&4
zqi{Qtu_%5nP2g%q+Uu()A6#`QX1H`ui3W($*VeY_sS8#uFISt3_lSy?y^sCjxjMaD
zO-~ffp%S;+YgLHMR4tJGjDR}QMoP+H9#HsrEO2I`RZZ85W4uvjB&7I4CF;rJ$8SM!
zC#vI!!qv|k<j{+Ow?NqVE1S#4APjQtlol2aMmdUI9IzV^1qDzQlilg*Pu*1p6A7T{
z{d*B&dIb#5_g9wgk}zSKpqNBkL5)@#34w@5kc%A0kt<Vj2|uXUgdq3Exr`dFK?23g
z8p=wr%P_JCJUd$z1POCQXecEKI~TK<IjnU1eK95rG(vMri^A%$hZ7F#DxtYf(e);X
z<SHNAQ$fxMO_SzTzgIUtww<F%l9p{+DInZ_npxr^3}liR=Qcci<VwD56hc1@>%Y?F
zD7V`}nb7d77McB^n!ruK9v;wE2^cD*OEt#$0k9hI2^FR#rXN9PAp;)5+#nuvC9SA!
zr($o7{ucdu%8y%4Pmfhw#B#Kvw{nM88tH4Qm($GoQ1@*$dug|B{WY^_RB=?h2hnb)
zoX2{9{i?Lvxz0=B3<Gwk<PD*J;OPRuta)aE_&7`hfbkd28h-AZ!P6W&w4X+Wg()n-
zAhmznFk~+1b`mAH?Ccg`#31Zza5ajET==ot_XQiRH1OBFOM4KF%H_HMY%SY7d^`ps
ze>!6%g^v6Ukvj>y=+K+L`<0dw>P_Kg{L%hk4b^T(q0^-g(TD*L@*}!>;@kKF3R^>^
z8C=}6xo8mF0452o9%nn%$628-h~1eAZar*&)w^Y4Z^-h820xp@{Lt9dskCw1z=SOg
zE4R1i?v43WUT)cju?F$yv;CU)Oy56i@jEtP#lB(xHyqBrz<VBE@!ikG6)<|qTTYga
znnT1OSx4K@pvHQZq^RIJGg!*j+P>zm?D?U~<C5-?F3LnwLgUTEYq#URc*n+ahVxRo
z;)uK*`+(Q=dsCA<gYB?}E00eP*-NNX+mHzafgPdLeWO*Fo6Crx>FiGV1*ZwV2g~<@
zjq|`M&J9q$X`KKuqvpw{=Yfn}mBL$F@3Oua@#H%om{-|_I1mnwj~{;!%@pl{(GB(w
zj-XoHU;GpC3&dfGpo850I`9<Fo*^Mzl?2cfvA-=~$}4_J1vpJ&mvyYgZ1W8#FrH0I
z1i$#-h;L0B8<8xvr+p3?0h&pX0be<^w%p*&cj>I%%T)8Ye26SVpG;j7g{F<(o?7Fw
zwk(J#ktJ(A+E0~K8vNcCU{iyR1P-3(7s(_WThca2E_y%^9~Tl>R!EXYQTO;w)$bdS
zgOo{3^gZD^B>gPV?^@2YNT!6I+nGDYE(kss=}co4az;lLn?!wAZHcHv)q+wa9>7Hc
z@(h9ee~^Caf+VTD;w)?f2nhR6;OyTohf>YkvwbUihmA)W4}(ua?l`i3_cAeE%WUma
zOjOs=)GWW24spH+LrvXL!c%f7392b9DeNoeVi87Sz(K?uAN+wU2Kxq2lsyN`-f!QI
zKxlxqhL1{^w_t+eo|2Zu+|m;0JvM+aQL~V7i~=u5t|H9-JGY{{OiDWBS%wKRQ1aF1
z-@6Ba-OjC_<IhJxLx2KhmY1E#`%9zyj@t`@B)S%qJY;G32;?h)Z1D&kXreLR7|D4k
zS&Iy{XM>RFt|uBL5b%0Z-kT&_o>-H$IBaA4Fkj-4r`zWUM+H%Pj2E-rQf0KRQi)rZ
z*bk#loC!`p1Pk2Tk7}UId<MUfYxjK(ja|^!H8!4JYDhVJW`n-)p$GU}R6SIoG2use
z7r7M@Lk$8STwLIs?fxc8T>PjM#@*6?y4|n~5oQy>Ch^W6349mIv#S1SsMfm5e=v$d
ze2-(zTaGZ(2ETSgeTCNRoob?zib?`EHE|?U;+Z$1h}hVu#2uU*j=4JOEXHa$;r#X<
z`(9rklmEN{5zEJ6fbs4GiJ8P)2rDf;kMdYoaWPm++$aD*F&7Y$g-Q)D4$D#N64P27
z!v&9a^07S&fXP}0#t#n<ijEM3Yaykp2lhy4AYQ9Nkhyd7M##H5id|$Bx2IIMm1;E}
z%V@K5*~Z|!3z;j1k~>A}_fTOv)n^}e_R{I0eWX29@{o&=0My?QVG>%Qp%^WA^TZ~O
zY)dy}98|*ZkL-E!=FI^T1eP3KFN`+P2Uj6eB<r4jpCwzC-u(>rXV;WUB~=N;L=uPz
z;_FEyv?c>CqR_t*(}mGRwb>t6LkinxNLd76Y(c6gmv8{u(3zRPHCE2Twl#HHoIWz(
zOx_$k$l`B@GJymMF$}-Lb04v}0-#0GM@^&z29g9N8Jz+Hhny|0ci$>2p$KS@f{^Ci
zxBs_~24A3QbOz`Wo%bgozlRUkz@@?;3Z@-pA0{slm;PsN@N)EbnEjEghBP=6lhA;G
z6SjFsYQpGkEGV=iz$anK$|%)pKbG-4RiI!+4AH29HJA^SuDJXl-UIN;^xyAz;C|iV
z4^0Z?ZG@Mwu~CEf8Bkv2-Uc`xh;OP7uu=$DYHDj^LV}<zGcLq%d@TrlP&Zt(Z!QK)
z_G}BSwlFgbe*UJnrM+F^TklC+*z)icvU|ixO3Yb(ZY4kcCFn|yVQF*o0?}jG3|#;{
ziTp6YrPw#H_-1bp`&tT%#4Aj3k>J0NwWK{QCzqzh2Ojg%r4MyB9)n;S5zz}n+A%<V
zH)HEfA)nSnGC2uKeUYz&>e%ZDvK=zGywGCoFW7K*dCQ@y_aG~f<JhuwYa0e9XRUDv
z2_dK()m_m346bu(NH%T_U-QvheyH?(2aZKVGyAI|zc9JMH$8>vNmgWR;GbUBnAe+D
z=LCYOSsXXhu(H^4!bSe(XaO>9+elU|m_)J+H6=<~-aboS*MuT=)FVjdQ#kN0AC4pV
z6EP@KfeQvOkHkM!uh_#rkrhB92-D()1E_2~MprAlK_U=xIKmC8!L}W2J<{fDl!R54
z%dKhCG-y~PO$QEyR^Nxg+~c>AO+-xksras*&Ls!bg23RR0Iu$zhG=Hwwx+*)tajGT
zXW@hmF6!v5hw3~!ILv!!9R8>E_@)dpqOrea>~BmnVe$<L3_N*yLQYkcPVwOZ1|LO|
z%J_L?pBKKU0&E5a1$MwZ3CWawPc(2j{&tfqC_ccf-{;TAojU7x=-QJZoqT+44_lmP
z8-?DLD}2FKkh#emJ*wcW7_P*vQE-)Zj@kiADb?aKKr0hYLV|*6Z{EO!`3AEufSG<D
zjfsnU^tABE-N#!HI&emlQT@`Ticg=0Sc1;JhQfkgR1jO8={-P_u$hC%U#*Kdg^b^a
ziPMLn8D4|TLD7&|#WS9kuH??qy@i`5!M5~WKy+6k3FBqEklm?X`_GINN6+H_0Z!a9
zvHytc9NChz&PTQdomJqv3X<U<KYvriviOU}54M7ucm2#WZgHBokec6>)<o1bVh*|B
z9qwdZ>V?#Y@|6cNsH18JgCi{A+Z$>)UUd$0FcgJBWt34*?u1YufewKR*gz{~7^yrg
z-C8qE5R-iBz)plxcVOo2ho_GJN3U@I`>HMv7Z>7s?t&G;_JcaS;kUwTRfqX8ZE8&m
zq>qR+!U96*RPgihJOfp&f%pn!ZBjFvjLaHfbWug-zd)j~Z;70L|KWp=97m4JQsm1f
zFRbN=vll9kFiWrCbZ+1~k5Qe-e_`aG+nIR!zxL<U<YbY??;p=9fSi|*xEynVP{Ls9
zCISb*$AIzqdN;S<E&cT!BT)$nt!%20!x2NbUQ*JvQmC+?$Ybhw(+_lYbQA#?%7vel
zJRwmu(v0SE`f*u?mSl7MK4e*tkz&GZm-FAlioqvi>)~QWA9enA^I-^RzdK*W-T{jX
z2AwG#*y#)(JA|k3-9R>4P*9MW`J=0=I3<OeG;4X8z>U>BQi|&!>1joy3k(2~JD`D;
zH;sg@0>dLbiu>PP2DgJLDdsGK&l%p#&7drW2Q;r<42K^&M0p?f@Ki2!RM*JtR$9ek
zK!7O*Ef~Ll{a+&w$KTV38FSsA1z^;Wm_?fn?nS!RPy9aq8br3jLI)id++I}8YnP?N
zcG&oBf4)7hKBs<eM$X<o$DOPIfn^ZL=B5q~48%*6Dtr6+>1N#o92cUK!q3agac6Gh
z7J`Gg(b2v~4;me=ZHaNJ!w@#T865l?ernUCNBdu5Ur03AX_;QU(CGvxU4&H;rfA5k
z)c`KkL$DJ_>-3c?rpN1UW~q~`bb<B}qiSac2K`J$bf9O(^b&Pm3rkD>Lx&=SP5g9+
zxk;5(Rk`+49NGa@`hqY`<z9K0kYEDb5;bWRuZa&1bCc-%P!naZE3TBegV83Vb;S$n
zr4+qlT{}0|EVuNxcOKQr#blSdog_PU71axlNK%Quphrg}T)pQu{J{en&O31&Fa}y`
zIB=z-(Vt-t0(b*kE5MtYr(V&LFyW%^xH~t}2zB~v)#3jBXTU7bw^;sx$G^~q9VbcM
z(<58SB!jDnL3}W{0d1O*!J)wwF-vrBi%UyzJ^jgQ*5&Na-;p7Q{E~w^cK+9p1QipY
zXhf5}d-u(h<N6Hd)1S<?Gd40B>&U+pTfYny7@<oX530uRL%b1ba(*5uCP;hB89)C3
zS0J!ZREmm%lbhS4&w$S7GaejVWB8-c{^6t{5<e>2Rs*TaKYrUGOAkB1GEi9NY;DLM
z2=#|SG-xkL!7DPUL|Q$5Wr$4<fLJw8CEv%NuQ9S*6AG7aj~GV9*Vu6I-o1Nbj8AWH
zG5$e2$Mc(kgvY#6K6ZW^!Umy$I6?}jG6L6tLHAJDSUrf{K{!xoRn^pd-yMW#5)~kq
zV`@rD|CGJvoQj5K1b9gJEMcNJ!B$yaO|;aF9$daALgAogg|F<ga*E<oeH30}td`W8
zi@_Sz=arQB`1z$F`K+tkPUInMOD1-F&M$=ZK|^B`33l0x#0ma&CfMNi4=xdrS#O@G
zu%JVCktV-x-z307^JJEX<$8*WTSyq?6|mu*ni4J;YCST2km-*3uK|r#SqYinxQw4S
zRC=0qjs>0tF!gBa1Rc+C5{cBmRDje8UzSP?qGaID7Vn4oft!5_k=#}9Wlo+wo|~V~
zOz{=NfydI{f5s*#g9zN-3V&vnSL51LG&FwZ#MaG+J{3hKC0!vB64s_ei}ywk6~f5@
z=SNo7-M!!3S!@VY{X@ax@B(G%y7vwQ^aA9Mvoyf>10Fta>TDz|Ei7!Y3?)d+C|nvz
z0`!#@8oJx9l&u-MjDf~WdT>;;rewZ6m|9d6)o>c)sBQA9YE7Wp(q=~t2JBEGzkWcW
zxlX0++`36SUO7Pf40_gwCXZ20yR1rjhbKO_<~;HCCxIH-*bME<0}bzRqOmq}{RHB}
zpg)x;NyYdf6t?h;Zej2dl}>=8VIKdFy5OjE!YSHUxc5*VKp%+;hKR*^VZAZj^S@ev
zw`2wR`4X*9?eP3z2}3On1OUJV>ZTV*Z*3q!9>L>4CrvzU46&JyR--|KW1ms{ogZtO
z)dro&-$=Vo$Y+`_TNR4(4$7)Ai^gIqnn^rka1}21yVO+Jr||@5UwK<zQ2{?Ez(`~Y
z)rJdue!U2?4haq@Sb;BtyN8($l?uKnb_Dn$M_Q_kGGLX1v_~w$;<qBO3v_f1H$@M+
zK@309*N21``~wX0BV$9=$9vuCw|?@-UhRi*1>6x5kOCoeLg?Yz&}LuUw{c=MH^{&O
zdN|o@imeSpvhZ*K8ieyyc5AcRKQmy>&*k;IY7@J?q!`8<b(*O46vP(r`D$nIC>ja1
zwp$e+4z40nd)rY}2x2&^@li<LfH~HP`+0e>`AEPaX7R%8dyVk|*nhuLeZ;54$HfsT
z!5cSpWDSWh`R+W+CGiM@C!xTCmr%^C7WoCa(1)!D2#=uG1iIJ}i@e&r=9f)0<!`?Q
z?uW5Og+TzT#=IkM*8*ism3dRLwvj=-`e)-zYyh`#*h7<r8Rf`5+PfFGO|er=Ne}Zy
zO>bt-*}X$6j0bX)D~{p+?HGD%N$pk&bOHMJJ_s*hPnwzivBv6T$I?6-{+o9B(W5jg
z1Nzp$(=i+`Pwd>iGv(DAQ;X3zS@-AKFfb>z@JBemu!U8EK&o9^c)Cxd?Qxy_CvE2w
z8=GG9yJqYpYn3};dotuv+Qy84{)LJ=JjF^IHglU#JFM+0c9>pa5j^p2!6E8S)W>fk
zdmdU!-IJ61q~GCh1NExb-`7CA!<CHBTHR4msRfA!@+VN?DQ+Hm7&#p5a7VyYg9;vP
z1~32+ZxGfbO9ovyz{krw0yoNN$T`4OU|pdsfnx|7ETr`#p^t659Q>T_?u_>qIjN21
zcsyf>st3znuCJ#j{r+EDway#ls_=niwSE6iw<ciSj6@8a@Yw88XpC32-KD^j0vFB&
zZ+a3bp3U#zUwiMN#Du7&zg=qxp|SbhCVI8%$ovs?JDga@aDd~M60Ub-y$T-Ihsj<&
zPj+<3or0pJy{xM0M%=Ceie%a7sHksELwUKmfO!m0@tuR8rnb8JkbnSc>oGr=3XrNL
z@+}@WC{~>pr`ul0An-R3Zn?{r_TR?%G6)!8)p~Vk*%>p=+&1l$BSa`Lc))86zn!NS
zDiouF-VgEd1JG*nyfTS$C}(rUF%B@m{qEYp%pz)eoYSCZ8rN*8HQm9<044}UgS3~%
z1TG<L9q&?7R^YGoGR5DL^FKNUju@>Aj3RhPnQ^-V5R)G2vfph*n=`2RT6I2HqVqw?
zBThPRFRzN;p5J8UU?m`s4;gqs7Oqr&Ji@QMy{v)p4~|t1LKfk1U=I|K#3*1GFiLZb
z`A8*?kc;v;!|uU}NqF070t>+)vPj>{K`^V`qg`M4`S`B9H?4;}^NAjE)p5dn+Uh|c
zsFPv|<lM>%8ge25RbT&QZZQ+bD`Gp7anWsdBn1B0cSS_{1_mtAMq$)}SMyu-x14AE
z(#Gy1^$@?5n*)V9>Tj$+AVa+yivVCjapPqS6Wn;K1c^i{kt%7}Sd^U|;)P@riCMuW
zhX_&AO{B0tASJa7H-JWji_yj_iv=PW2t==GNJ2}#3uYeV<{<|g9eD1QDD+iw<;PI$
z4DP4%#z$N@Fgrbs_7eyS;*Zsm8c{>E7dm}#cmE8!MKe`55(FU3cor5bNVg1`oj>5K
zj^PccalI;=nviz+9f5{`s(}F@QG+W8iWVXzmDzkPO($g)^-+t!GJtkO31;>(FfgD;
z_)=S246hbq-b&xw-nen_^5f4L_wn;cf`N%!B%^C5Fy1%Xj_JawQ{>^3x%4D`KE!yP
zzw{%bx*tFe&5@&=oCwmsb@T-q@8LWiLU1N6O&k#>|GF3C55kf+s|SvW*p4kb-X1b#
zrK9`UJB8K<Q8f_sK_Gu@?IT5)DsgAfy+wI#+)#==(PmpOcC_m3EztWw_4DcTXU1a9
z*m@!`CE307tws_=RS9e6_V)I0>m|t_xig@&lAxvLO`yW=S6hi6VOY{|wgF;VIUCR{
zF`h^O++M%EdW?+3?FA?g7>Q|?PPU06A46t?-3jdA&^^mx2#ZPG-#V;JD88UiPSMOf
zvV;|n8$i`oKddQy7pWrsX7#z4j+p-gW9E{JQzE?>1a!YP?QUpnWGsdd98F4cLBR_z
z=%bO^M@^Boi&E}W`49#Q!gWSpU)UC~7vD~QEc|(i48}Ie*<kwcO4BV0<XFSdBP)j#
z32t2)lK%Thkan^4e~^#dmW3Iu?p`1{DU~d}0bD`9Nnejg$2S^1F6K(cxunQ`hflR8
z>ib|{-zSzZE!~Daqq##6`R?6oK8evJipt8m==G=6(CZ><&71?e(0jNr<t8u|VAd|@
zD>#DYQ}kPs5Q2ZpSBkiTITG>Qlb8&Yj9}^=QuwZTq^=={YBdvYf}lxav=1I0{M+`D
zEqG{<X$g`K4wkmoRx|?$p+JoQ0tf7}Bqn9jV*^;xZ}e^*nS#{Pt(W4^NO*m_$){g<
z?wKa(u>Szqg4ylJQD>;IaRkm2LCk1CN`f<2Un0JIO^os!f#HEHteM<Id_&h^r*6a(
zDZe-x2;*I#0=&hjM;t1en?t3t^h+PYM+4-8<CpbOVj14iCXwh{u~)+1@?~(w`v4mo
z8}v8Oj9?oB0fhocbkEnOJVY8|@{ys}9WZ6?vN6T_&R);wjZrgd4W90xS`$Q@hD2$!
zE;w=BWpAPN%O3es0u^pDD(u^fP#GG34Ld1^NmV#|9?i`V-_-c}2$kdxu&=~L!xA5x
z;WuNFlS603=LDZKQngW!3$5soNU;{Eg_rh803QD)=9{TAO5vs*T-#Hprz>5K>c?zf
zMgQ$Hv$8avD2Or;fe!x>2OMYy?nBhDhBn>oL#l!K6lOha*)e5bK(tycGQ>9!jQ>_5
z44=V7;X)_6V!}Fu@(HCs&H;?`g@7G^q#d@vH)zU(H|6h6G`w*m*Lh_LIl$@EJwz+y
z>^Sl3*Y}2ozTdxBKGCZ@afoxfCiU9}G!pnv$Ymy4j(jSCqFVcEA$A&KS_ul_3RV|L
z0uy|#Ha_z7M7sh$8N+I;o0~r*CqMCm_+YO167l`?IcB&fmGt$y38i&8JqCVZPVp)Z
zb}-8Ty4Mk6YGxZ1D7onZF);FG7I|kPFfw!1wx`_IoS$tAM(H<i#Gf?Rc$yAC5XU2|
z*w~2G*N#S9;^9Ck0oe#BZUC_OD-iKLFW6By)(#$e_5~X~(rW`1HjfUH;k$gb7?Zw$
zZ=j1Rg71<5FK-?ZVe<!%jyJ@NwJ*JyVlN!%bpHDru^%(A!o=<jn~KW={duTmE`5Xx
zAU2jIg$gX%*!a-BvRJ*joQE{N>wo9rG;8SSu$biYuP|(Nh9KyWo{6Dh+0v5@_*~cA
zFcS)0-e<@xXl`sww+K0957fTq_Gyx9`X`3(#682(L)Qun2)9jcClk|1Z|@y&ZUD_o
zk;Qx9fWiVDQmUYrL?8+W(x$|Yklec7p-M6@&(F!xI<Rvj^#(W^lTCfIaG!U>D<%{^
z2C*v?e~qcep^=e?i8H$~n)wMlk(e+9y91P9IQsJOthte4l|sBl=gkF^#>$<z%i|Lk
zn3IGTB=to_`08JbgphiNz6yF<;03O(uBEy;Q2o7DB?6tp!t5)OD*LAq+enAm+h3qy
z{JL8jzg%gvu|l^D$O`0!YV$s$F%hJ*Cxyk~^1<YcNE3BlDU4x2&Y^-FN)k9mQBk4i
zMFDdiv!PKUA@%^wA;KZg#bM3fs$S?ZG@HPH2n22-{tC4OxINFZEt@yLfji&ALJBam
ztn5SR;*e3r&;Rbt8~v1tSrsJPW*Sumq25RC1>upx5VQH|Y1EX!zzUpKj_)`hEe$u8
z?T`ZYs9<>I0S-Z8j26}9!$U7wSy?OwNbB+1GCH_+_0%aMrjzP8(vFCih`2b&UPPRy
zAn>NW-Kl_MBW}84X~V|Ve|lq~a|f_T{X|)VwJFDOQCqteC=wj#(j3zW&?&m_q@$yQ
z=|`C4vO0a2#uI`3nYa7Ts_~EX_kU&~lsW$`{njrp*4|IZpj%q(i9;q7(?&nq*B4zl
z&^KidUm7B*Ff1lxRSjhYhVw9edz9-envYFjr1;bK?}S9Sg9hf2Ywn{DybvO(5_$Y{
z(}l^!Wh6P}<mcmE;nT+qJEUX5d<u-jjgt1AUNMU}aX<Ap<LrQZS!{>4mKG7~g^gx@
zYKpO1_B@Ve@56C2_LIoQS^?*T9la%c68JqnSr-mCq#HC^VJYKVA#(ssFSo7;7nc!0
zLJ%-u+tZ<@&csMT>^ac@S)lgEHjj)&Fllo4q+vjq@PGW6$+~$34;MfzD2pJ6K+H7`
z3NUp*_=usiU-dfh{=kB<yKn>&W4YPcPp-9MyG-BcjN;l<`XffDuD}E&_h~&Xn1TO_
zY2CmH{!mCuaK@r*MF7~;;xcY;=UhKX0b-g>8YQ<p4Sqd+K#x#$;Bzqx9Smhbfm{p)
z<ekqO1x1h!*V?Kmr>Ut4V!L-AjS}D3?;ddJH*Os3OvjX3Otb6M1=Misr^70AvqaY_
zH;V~rD0H4&+T(_JRDeca<zv4gOd^J^BG&?-HJqMspp?pElnEg=#H5<?a$-C#vgYy<
z<lj@Wj6YZB1q_4;%HDmOm5wH#-Rxgc)=vHeiAm_XXO%XQBZ}<9hfpv1-roiz=En}X
znqR5eOd^#Bxw(~rYA^p_`4dgPmnrTBy`U58?oj@#*LIP?86_w;uPlc~3wgwHXvnF=
zj^gK&hQjh5bz+A{rNSK?zMqMNrHq>dUTaJL=;~`!yTsm)MpdizVGJN%co*ejF6?Ar
zfD6gIgn1soKJE)dW*}EvTOF2Whv5M>pr*mZ7b5MXo=s!O$<dKtujq=JTA5{$#L=Ue
zu!RtkUmn{~of{d9P%YyW9M*gN3{T=)WJMr$2-_JpPB~;81e6G)i!yQ_1)*$|HgW$1
zc1OYs`U>X_ggL!r2((;AUx+fi^_msLR9CJPxU8+B=I!~UrmemBp?KvmiU#CR!5aST
zY`WG9%rt{K`MR;O-<TOWR`jiF_y&hk3#QR@-=D66*fS7y6yIA0GLq6kM9!FZNeF`y
z*KDD+QT$WqGdKskf*$5P7dE|odKGJ1iw~0%Q5HmCFpm;9497V@r{!ahIWPag(e1hV
z_p-LO0~{J08EI)@NN`9x>Vz`@QHZckt}Z*^Y5wVeb1dwS()n&I6L6S_DojItgC&Yf
z{I#pg#p`A{QQCQ5+DLNszJVv6h!CAPz`a$|EVy<ME_{O`x#mpQIiWfp1|J}r7)t!=
zkO1r6z4yji(jjd^+<&^(7*ua9H(>5v??Mm)12m3i^Je`D%*(|K^(?W=fip$1G?fxV
zJQ{XJH#exrq0>`5cW&Db3^>AzMrzL%Nm5B6q7%x?%3_m}z%*{VoJiERI|6-uYbN#3
zGhkeVYNAnUBc|)38M3I`3!MBd>@r?lbXD88+`~x?#pn+3^9rGb!XO85+)6o2$H{RZ
zltrXdj253w|2Wa9n<ymK7NggWr0?E`b4L}N1{<3`n)=F4N|Eo8*Uh&uw@aqhf*67@
zru!vkJUa=_>h*w~=0M-TBq)R3B_k`Vk(#*C-e1wgB3Oz(@b&9{WD8~EkhHRzU~j6$
zzg60SbMabwi#`AS&2X{ZeQRfJEogtXxS`;i-~<f-y)<Det#imR1!xHpg*mmIl6cE<
z^p+5@c<DM0zb2IK177&Y9z2J9idI4%`y^%tOG+J-juNwqudAD)*|zGBctk_PSeX44
zxNwq{IVfJf%<Ke3j_JMZe7{HgC$Z|?s4?LlV4O)Au7W;?@Qv&OQ1ZOb=q|z=82*cl
z<SLPSKO!jjAp3*WrAx<MqNZntD_Do;nh$b3jJhAvH$CM?>*ZyWUjr`1V=)MOc?Ko)
zJQOOvH-mR)+bF}`(YYUQUBahywp)XD*I$JPckQhGf0({`?QYq@@%nbbsRsvBvqOIU
z8k!v@i&9M-Dx*bQf_#QKv%DfM%3pdg?1_8JQ)$ALu!)_P0|yJ4n4lq7GL>}FV;Q#j
zGxtg~mhcIJF=0<jfZHJ}hvQnub+Yl3#mlm^Qiv|->pxvoWMrkX(4}(YM(;ibC5^ND
zQ}k3mSW=5}AM&vs-<#yBe*xD56^vX1<Qh0nB>et4+&Nz%EA-sCbHef;>Fhz{bah!R
zC+nmbi;e`A5clz3Jj08>Kp_yh4?m&Hcxxs)!-)Odu3Ju~#CPN?9*DClVo#i1qnTg$
zk$dO80r~5=3zwR6$|=>p^F&`ztcvu!KecIPx_amgxm`e5SY_{&7x4L{ygWy*Kqy*~
z10o0gW^h=TX(vIucq_zQ0NvO4rxhy}17YX_QQtVEVrwHAo7~)ykKH{K4+lboTtJ1v
z4>1Ex0`o#{{X&JzYW=4}8=eI6M7>mz**Dv&mSEa{mKY$PQX$A4Hchj4Pm^Fu2n|bI
zPR>O?igDPX?*sWo01Z@npx|`hn_fg!R63S-g&$!waAv?{hy@C@%$|#*e|nIVUx121
zf$K|UCD@SpnVBC_jSUSp$P?wbZEG7PVfXHTf6ookh*5DuaCzzD@3drFB$a4@9bagb
z2#z=hmk@G|!ZvO^(a;Duu591E;qM3JH<U5Z(}T)$S(-&Bc^~u1P{qSS3eAz3nHf|D
zXt%Cjt<lWBllv34rMinlocaa^>DN9#26}_YS7L|d<^6)*udD|hMp9Z@K;t{4<XrP1
zdPO|-bE#PX47Goe!cM%`ar~WS;Je&SzEop{Q|%(nW>+sS9UXbWoLq;7U#@{b-O#8=
z-e0mgcg{yP2Db|@;XOz_&$8?|E+CM4@i9BXo-lz6<^RpB<0yv@H=CdVg!3BZTG7Qk
z5qxMEfRuYGD*o%A`A+xs^XCS>>Ia5CZE>8g`uCAzWAbK#Pcl#@Cqy~EMZb$Z25@~s
z!WsThZ1OPULgx%`{2@U>>-UCU=(TBRzSY#EBqk1zkK^Xxh{r+%AbRROu}!#caYxQs
zg#?9bAgoQvtU)Mz0ZlIoO!~Xj(W_9|O$yi6t*#9ClZ_(?`Y7y{AN&>`y;Kd1Xtz${
z;pYDS*(Y{y9T(<!U>eGfjtDDSLErV#At#i8@L*%Hmn@Jf2~*(80AC`1n(L1peA&Ca
z_W6GBOw<^{a&)sh$8zFK#hcG$lN0um-w#RLrc6wXW!#FZL3%=-sgZm!sS&eYV7)!N
zZjZ^j!Ve1^5(%Kb0zh)+#@;ZcpFiaVTgnU@evt&Hr>kvd4|JrPB!uq%-ExGRjA+kF
zT8^mbMuvpULpX-nf1|9>bSbN-tbyTPC5DTtVAw6(Y&EwJyI$PQD)X>(s9}>Kwz5t~
zA6;jMH%BB{?$0yM6HA-))9GtK$pk(aqTI3c!ovGVrao~3ss2f5-;<MJV8@hOrSs>d
zoR{r+Lu@|A37Ro{{&eAFY2=yX$9(k_g<Z9m$L-=S)VZvQQ)`ue`sOG7_X$qYnY8$~
zonQ;SZ<;A4?RRlGtHgDtV#;p|K5|?;e*6J4QOkV<iWKBFJm@CI!9rdR|2t$W^H|o9
z8zXst48wRZH`dE`tfl@*w!EPB^03OMo}Ugatg9!Km19|{U0E=E#wC(pFTTIpyLx4<
z!ou9NU5xi&3?<8+;_mn$KN`nXhue2zpYCkS@^M}<(<zRUlr$YwzG=oh(Ty+C-HEf9
zsF#z}d8MOW&@&R?>&wFhIC4?6uE7A0VbSSY1$Gm<^a7NE@<BmEHR2P30}m*N^vqO9
z<!ui)*^KY)C^|npjG5k7b7eVpV59SO78hdU;1x({(Q=%S3aVcFog;ga^}gK900JwH
zo<>SDX+cPh{trqn`ubjfnJ8qkwS)l&ooGpS^8%o1@~fXz{%yiVh1L1!r2IAhP+j{6
zek}X9&E9go)jzK6v-^h!^?QA?);2#o5jA-yiTylk=WAY)Gcg}L7M?alh4=qj(=gi$
zGqaos?65JOXmn4A@6rF>k{kT??NH;TgU6qb{jO?zE_-UY)nvF$bZI7fOW`h3eEcQd
zn589&CpZgNM~ln3P`8r8WR+@8yeHSZ$Yj(aNJ*~5rmH*KbQ%BGyN{BjMD_c(%}`yM
zyz|_y-yP>xS7VupIzwqcuoe{7^cRuMRae$RLgXecAWp9+<!J6_H?=skbe5(numj@{
z$1|vzbYDu=x~QQs2TvT4u$<;T3r+$ZF54@RqY?JUNUly%5C-7P0O4h6iN*~9oH*<^
zJSc&v3NHWJ!3bw>fL3k<>{w*oBm1LPi~_===GNAOJGWj+(GmR?LzO<)p>=l-D-)lp
z7MLhJNu&-Tu}L-t!5<w+@nKtAa4fi;dv9%Z1t*Wjft@&pA(RlBx{X_wjZ=TTt9Y#?
z-4GH~%&)`rqGw;g6$12*c3EA7QL?7Gnt5VRLbic&vGB@DECYTqN$E7M1w<e$lJ?d}
zD@WEW&@X}uM#e59^0?$s<bmb~0Bha{ZfzeE)6HM~`gEw-lW*xm;fmlmk44@egRTGQ
z4>Y`ZO^j2KdSAbM*^P=Gnsu)ik&zIg<F80n*W$Chc{50X3tK&8q(p%X`p=$HLSymI
z6aBR4uK)gi#xJG`n347R^~$(iaT1cs$Hu+_d#$RfXWb574w}P;2K67Qn@HBFH%(3T
zTl1ei)4&)g7%NsYVdjTC35g0YaHGe5f#&(p$qj}UpC8vV5zn7B&%m%-7O=CYPoDxb
z<v4U`6o54$On8Mx4KF<HGLThEiu~L_+VkS<>m{5SB23i&72w|^ZdIv>jgEJBTSL2-
zr4tz*o{I+W)|R_(w{3l;k%S6L!XE`a2j>|XHhfl??AMS>I1&+Gg|TET!p&RANb!ES
zf7o7k)T0OXFm2fZoM>?HlAfN+%)oj0#$W>TIMNxenFN3mk)Es*C*)i8R%I`}yP#5g
z!O+}%sBod2cm2(IQP5$;91(gUhZX~Zdm%LiJU+_MjS%=EQXXQ?e2bF0%n`W#BD6y%
z4Fb``LU7yM(2(XthAtP!6BM8T9?SfV{_r>eqCmg!FAmV^XAb^!XH!%5wa;4-L(i-R
zXakX_NXg|0L?AU@w{Afi6LsuPy!&U8hB_X9JZ)$rk%Ec6`oaaci1%_hXWpyg6f<Mk
zCv_H;U|}ISZCZA`KylpA(8VXjvmVbCOlSXZP!K9rtSkb{9(cTkq<<ENC<r5-S8!v)
z*AYWmOWdRahw=)X_DDE_=hEg3vf;G&kf?<QZTQt`9}oJfj!3{?Yk2l_QZ(0I%EzLW
zIRtT-Or_8wU>ls}MC3<x|A;?fKmh(=a|VGo*xn!e_)s2-xfUX@$e*Hn{?fZT;o1P*
zk~JQ&kok#(mX?bJ$MJ2IG`ROqUf~y)lq^7ZW8p}-rzr4SCgay$V8tj02xtA7Gqgwj
zD^LT$ev_7TR6;@}TKYa>W5Au4b5@wJ(4v+>0&6Qdkc51RQ`fKeLp|<Y`SIg5v^i*v
zP|;z*BScu$qWI~^f3*PXfvgDjdR*|sc`RQ-5#xU3<vk!&f@}zJ+$n3Av}4ekkLXM2
zqffA5Rv21KEM|IdIXr48;FnT3;3#?)9jzjVZ2?RgDnvv8LllEGj&Wk*t4GAEL(PMZ
z-5L6_>q}KkwDYSr9S*Q<I=Mk~bY5kQpGN_Lmh3bhVfoXn>tq2ghX{<riGUBPp`-uV
zE`PjMtXmTO19~02C-xzV8T;dC*P&HzUJ2g625iz_a#{hr-)iO5iwbhKYae{QDlz|w
zU-w<TDc<Rf97Jn`OUbNHQ4Vp&c%d4BRi@B>>KZs`SDAy{QCz~pKOmh24v#;)RH0q$
z2o@5>1ypIHeiffT-v_ddtcL7t&Cn8*LAx0k=mL|HM)6SF*tH^J9-kBX?AdG0q?*<J
z5uO(m94`|aJxRXtrCMm&YzaAS0Zv_j#2|-(DduTkd+5}jsFt8L4{~oGGcz>Zd6?h@
z7833J@}DW)SgF@bGxq3kF;=XboK{j0aRwnHX%3RTN!VZ@odEWps+~Ci#Z3!XPkZ~L
zJ78CW>}51{?)r>9SV-J){8DTRSD|ymzU3ep1vmj}!BWIJW&Y@pMD<?e6AB4&+O9u7
z42QqJ&UoVQUki<~w*p6C8dd8U542_294T7!byAv4MNoHQRWG$l?7&biTT0*GeI}4v
zy{R|#Iv^^l;o?i>S(b|*>1&o8)_;!p^<g$jkq>0R(?R}}lv<RBkX^_gA#4P41DzJr
zy=y-l(9N7L;842B8EF#xU|<U13Y<_zMtx{>4kIaVaLNX2vAB4RxX{=0P}d@BQ1i+a
zpn8WHd_Xmt%GUIj?kvB3iR0JL#_yy5ZP~4@FfX+MbU2nXM9&m^^?xZsA(3!C!dNh>
zx_`X0&?(JqIMvnw?k#Vxl|@08VsJ+g!GGhcCs28%o4_)WKlObL8pHwDA2G)jx)X9@
z23w&$A>3Xv&R!0ZXdl7Ec+0(#@lSC12t5+BNGoX0Yd?nJ$k{(>_o70=VVWzV;uI4p
zwxX#iL@Gdm3otY5a5cJFuSyhp7=9*G-8+R}11C4-vf_~yt}^X;C`TcKN9pY?2ULur
zIt$J#xZQUw{`o`ohj>QroF*DW!9?(hGxo-#T(oMX3WKd4Q5C^eMs8C5n4dX3e8U=?
zrngkx;g||Z7Nw1OUuO6O3S;s3Awn8+WTEu#Nb=Bfp1|xBJw2J(s);48Jk~>oI2{1F
zLdFCk0wADMde13}cjNSeFLf?-@c-P{^TnQjlqmGO`3G95@KgpWG=OQr+%gRf(UwZG
zb2ZGVDu^Ri+6dNS>?FA1eUzZsU<Amxl)iD*Piu=EaC+G%Zc#>Q?Kx0vpraMOd<zpK
z3JVJ0b%t*#HPs4ZW7?SvtvyO@w_6|Gi-xdw1EbxJ!}#eXaUP3z_DI+rsV%3v!=lo)
zJ@a8Xy8#Prc_?s3fFs!XVfqJK=D~kGVf0u+y959B&FHkSCsR6Y$tM@>zIu3H<_o!m
z>}=E+nS>MsY@&!^In|0}*UOnVK0ez#%Ro1|t-PFDSYS5C5rNDP5-Hdn9S=1qPIM21
zlLh}R_QAX_N+gjn2QfD{{1TVH?OkAPFYD+)^@C>_FfMeVh=BrILXqFpv5Q>T!$EQn
zGnSHEoMzIT6Dy6vCGGp57RE3Ef$dePr4{1TG>3ZpwiFEW)E|U06th2w$!lq8f^OQ5
zZZgz}dA8ulL8y?ciP?=C^%Hvy=1qx1k9%9@{>kFy>1r@MK)&NuqKi!HUm<q}TwhyD
z3m<fme$i(P9_!DTi~Nz->va$l^9RKVv5WIul3pSHLlvpH2B4^TfMBcCvbES6LZOeV
zdKIImM?XPkU?g}lWdeU-OdG7;)z;IR7iIs0TkOX^W9U;$*`9Z;Ja{OrMhLJA&VqG9
z3{jkJZewN;;WG+)A~yER8~OF;S38XT0g@|~;s5pbUqc~G_4x6@XJ3HsfBN(ZY~r2;
z>W7H~c*5jG{OxuN`cu+FAfkaxJ=_~p5tvG9Fj(5z2@*4mCwL;In|{rGoI(>C+tIi1
z5SNY!jlhO|6-g>Yu?TEjLXKp7%$w$K6-q_E;AV8oU_^|!6s*$z4cdRMB{VoB#Ms(R
zKwNwQKQWeR{Xqk;%wq2=k4NdFPQ|U<pD3BFSC<!2;h<1P;jr8vsz<{rovfPZj1vMW
zK}%?(#Ls*c^hO3Kv;~<CV!c&IdL6Q32B*N|C_MpDjr_>yn%<gdnXYc_uf3&OTj8BP
zdv*zh102jC7a&lhV`Adnz8PKtR7V&*$DfH-87NlXZn!u=>HLCs1RP#wtu#WCFn<*T
z0*TX|kI#tt1O&4lmQ2?|f`=Y?cntLSL$8a*?j8_X6lB2sv$7z_`+)eRbruwr8>}g{
zVMofKyTkChe*-hQ=>7v?cXoE-b5O|QM8Cq{3={%t$)~RfbR~>WX~p`^P#t?nAxfwT
zd?UyX^Aa@QU0K8zE5dMG0Q=Bvt^D#47{{58SfFtG(_sxEeA>vVds}Y`g*NnhALEC^
z$Oc>CCWYcSqT)vuhS38G>i@DUq=XrvS_(J}&=H;HxQNOGC-1vGM~Q`A24^@L2!b93
zeFAsOg$w$Et`)t9@u8&;H&IkzT4YdgFi6ZK&7_(2*IgCu%a;@4;(ijD9|d{Z${$h5
z!dZxGgo1c<!K2UmsBDZNqCO{k%H(E=!0t_(EV_y!y#PKFqeGFP7hHw4eOCZ>7gWZ`
z5@=TGu3`SqgM!8y<A%2fg2?rz=Ye?a2_+6XlqVqBp)*}X>=SW_pt7Sr>Td^8(2Ezu
zjB0GgpmbIo3IfaB93;a`2`C8+V>W%!>(@vihoK3!DBQ7^FJA((L^KGQ-X2UTMM_~r
z#ARgOcXS{J8`YBA!-uFtJowG5=7SM<{{OF-8xwH<uVSt_GNhsxR^(z{0RfD#1D+C6
zWd!sZ(mEpkO3dQ<yZO}v|4y2N)=J@pTpWld0W<(74Pk;hMddX0d3Tq?;QwOmz2mX`
z-~Zv$2$jr;%n%LACbJ}?L_<P~WF;cmTUwGNEm=vCNJb@_w~%Zil9g=PJL7(y_4$s^
z_jiAP_kI2Ge!L$o*L7a!dAyF}c{~RXe))YoM*<j51%M3%V9wr-z{k-Km~r|d9nf>q
zmksx|!iyg9)!^~8OdL<t9n89ICC{x3TlXVJ>X5|L%G|Eh0cHLDgO=dmnVUZ=PKGHR
zjF*%JzGYY*f;eQ<9rH(IxfDmJM_vWA5YQwmAmPB+_>P;XcshO|+eBC0-OFDB3fp5z
zj>EL%s!QO<ggg0bVOkcTh<#UIn;;J^+H5z;MSJdX;GHqLh@gQx>u<LbN&2>*d%HTs
z8}P(>C}EY2#UN06Aab0wJ?(3%ddLz>6n(UApeOM??3S5HQ83bghL{8aCH<bP+>ISL
z9BbQWdI=8M(u3sAZ;`8qf=+5&T)wKPgh`x=F|e3kiJk(#@bqYOfLu~Y6!Kk?jpyL;
zm)ZUg$wTb)TUBO$eJ20g{mMT-5WR>|!fb#w>loNd7;CxH!V!w(311aVrl@PCo`YNe
z01f236PJlFKJ^<TA&x;Wj9p~tGlUznBu5_p8d$sHnuF);>$?KJFm|MNc6Lzo{TLYd
z)XXvfE=kIsfBn!8f`Dt#t+9q-^{OL$!N$fNPRTUfu!qEB6%q=_VF^v8a2fey@p7z{
zk_i9>Pv#Iwbs}Hl)c6<By%ybn1eck|9Dje6DAG5>gMte2WGlHgC0`s@NF4F+|C`bC
z2ap&7UXQ8o{=w_;zdDp!UH|$6!W2T|e|{UMF#*EBpO?Qjfo;;CSCNoEHSgd5HDlRJ
zNn$g2iLV{6a(PCsBjp^x^o+YfZsSd`2>bJTtN=vN+R_pw3d9Ci`M|Nq-9QS9w7}-T
z3q9T9q9+!A;T50}z%2*)g(xrJt(Gb7CI4|Md{X>5Xe@!eWU7-OqhM2A9T|#+vu`;N
zlsCMv+r&JFh6vO|1Pk-Np~FYQ`s8pI7$t`v=3_LTKsyf?w5+tE0#6{j68B<b{g6Hu
zqOl)qZa<{#K*x^x1l>KLO7w<s*o=$oMqxn>550Yxd(0jGWY;MALWoRQj@sG|q1IX7
zA+M6z2Yd{Y0WSa-VW6+D$Nf3Aut@dO38}T7h|_}Sq3Z}oY;7zI=jGRs2Xunu`S&Kt
zO$H}VvevWeDV<ID^Y0jbG@qyl7YPE%ldK-3oNON!!had$-?1z{#Yf)GV9$r{Pa%IW
z1BP25GY4PXy9p9b)kg9wp|6BDFih2hY*4ImZtTB(8~ehbpe+IsI<8pL#OZ;Nizb+}
zhg%sD!%6zjRCeJu;1L#!gP{r<3k0;GAKJC6#zca=igjr2aDd(XR9t)(6gM|+<e@hF
zVapyhH2eToIJ!sIC0GW1v603J0qi*+_9Xb=XidR*fnCxlna6}j48RnIe=vC!6>;0b
z6$Kt96mPKcxR0^e)O0t<);v65`GUNyhSfBz&9B`gV|xRCui!fdgDVv{{m_uP>Cr!m
zpp=P)xqm0|8qKm@$~FnuD-LA^tj>{9%<RpI&mhc&y9kzT^KfRxATkOp6U}%v8#{Tq
zx^V=jIJN_exP7vQ3lx?}RX>%;w{H-S+OakkcL_+qJ*^}75=w3`W`22c9-sHZ1=#5R
zi!n=I1`<lFqT8KP4RxFq=u@$h#%#zVo`%Z?%LPz}+L-}*um|I;#Hv3BVZ}X<K=J?;
z4E;i;->aq}Tz0&0Y~wkBL088_F5b4dc7*ZAQPNurMFoTbyzMN%)fJM2wpG;M>#@l$
zKM?dF`e!Cln7Qwl-F#yHI~9mifY5PA6FpM=Z%c9h0EdegA&yPU%nS|<MVcO{xg+l2
z8p15xLXwmD?|!5Xr#B!^<O?`uD<S^u=Je%x+~!bEF_Fweu(ojw;~c?fG(L5z5Z+}y
zcj0c!$fu&Cj`<ieXpqg))6&pP8yr0fss-u20n+)U-X%cQa9=_D6?bHyw|8rJ1)R)K
z&&$@${OZW(+HeWa*%?b)O46Veg`e!Yfk}|H7KWo>xM9*P>vp8!Uj5g8PRA8DmZ+Bq
zj9R{Y`!+N<xTjzXMbYy{k{W_Xdp@{~{2RQm9S=;Dl#;>|j^@%1OI|;GuK%Rmk|NkZ
zpe%RLFT<O#6+zOlGlhisz0jHDrZb4|BgwS5B4~|T-Q$kTqlxLv^JwgP238)*HMw&n
zU2bWqsndG?AR+$-g-}1LZ1Rg!7nqouYQr||!2>o$v=slHygPA-4cm+p7>*IJmQ#)`
z7y`+qq%Pl_K!Jv+d&5dl?Gr-$m|b=Dz5-=w-q7;?DFC!9o-P{k^+F;dTv4wxGo^O#
z4!sFJfpe>356lyh2>^})?k{{-;Kg8#up34wTc5#0V<5v#0L~ZqS)ee+Ts*CeRa*lT
zcj_cwVLA3MyaIxl?O|HzqBpQ}f|`uU0e{a5+I(n})xXr&2bjLX0rf^wCZN2)&7T>_
z531)@5|#x64yt(E(J<1^mB-?&$pJ>|vqe=$DT{<8T0FCg)86&c1W7Y{czF%?^}7J-
z10R!_Y~APeaYv$i2Brhf{0Lp-Ijs7@s+Hov3KKLJ6GRL7-OaE2{MokX@LtH9Vdx4y
zBo0z>4&;&AM?HfcJe_viL6pNeaXk8P1<CwKA%5UKvOz;@@E1T)!3SUodY+V20ZPU5
zL~vR^1k#9Gw&F-CHSfuu-h#wXvg-llN~4$k;R8n~rgn97I&V~?Q)@x1K=b+fb;xoi
zx_z?t(tN*y1LWe}qRa{4hE(j`5Gdgn0IQX)U!k=-quAiZiE0y0a>0q)T7USY3rnc8
zKsw%cWxjpuJij*Y4}#1y_vQ3!GVTYz<84F3u?ZpSFp@s_;+Q}f{bOIQM$2!2i^{B}
zcxzeglJpm0(m;R77<#`W+rXP@aqkIdbX_a!o0dy8l7zEWR;TmBDkV;1j@`JA%5*`e
zM;W<k1HsE*S^#{|Vda3uImBnH{t@F$VoxwDL4kC@^TdFZq@>}Rq%BKO!}d?CtTL|H
zWSOSI)<H}RdJ#NuN%|-OIUhcN2M@y;R?BA;I?PN=MeA%4{)RHRp?_<Rq!T7O8oj|)
z<2u|PJXYd%FMQAfGMyX!p&65}esTud52ET}5duiCc}AcZ61(v--t`Ewc+6Q4*<+o@
zI9~zo(RML0OQ<VrV-Bl>Xd)__0yzuNI;5SD9GtHV&Cgml#((oQim!)7I@_jA!-Y|n
zlbEmIkP0C4=03~w1V;gu;0!mg*TusHCERGsx--Jz`1s7sxnA;7hwv|^ZBT*P+egPT
z@OH4%z}H9-+cYNhh70P!($9m8H^pKQy2n$~gc_YdIrXy%c3)s89fZHsN->g~Kusi7
zKv<pntxhMZg1p@!`jzr=UW+pvaHB6S?u0AFix0Sg_6v;%zO#`P2S^TGR!<LZ_)ed>
zXh0XB0`ykJxBKo95xL=W9=j<p!YBh^m<$l#1hN4k7RbQBkP=2bN*9POo@iH*iU8ts
zi^Aq*s}ZUk$8O3C{Roi+?RZbrp|mO^?26ufqj*FIHh}BNu;ng<XUfXR0Aiyk&%83V
zh$#^kNrs}3%g8YVwet!6=><Li|L=2cZC?#OqI&!sIs_mzH@680s6FbXp{31C)rBBf
zPgj@ME@;q9?8$}}vhn-2nH$O8$SnJ6`wk}`npyte`pPuF9vRJT|1WsBh6xO3kQ#$a
zf52c0?<j%Hc=+RjgmV@Iv;9@cz=nJK`&s`moBT$B%;>74#N$XPlcR+T97aAediu}W
zuC(kgV;Z84D!97)iT@TibELt^x3uI9>iNvji5|5%5^cFKkLa%JoZ@JIlpQR%8VqZ9
z_#hd=9aCA57C+MxXe(&)UH9!PMlcdNkE6cgX#$dz?$wWs$vUyzHIJ`2|LSaA^tfK$
zbM5-|duP0|DK@k`^N@dQ{0p5U#NZ!HMqxbqe5bNc`D~^7$8Wp2q8!s^D@OuSoA3Xh
zsPOxSjt?2q#O<58xxWEI1z{@VK}s$#<+9v5@Z>R26fV!63eh+R76<bAx>tqKidj}n
z0iH*GTm0&MK-8ChP`CNB0{&-!mqB{$95#(cep|@L5p8FP*H<`~XLwF>a&qRJan`{A
z3?@v_Cn&w<xJb}hLzap>8jY}~K?+n8kH_lr9Oy~0wW^nXsFOq&ccAV4v3@Fl?$@KD
z>|^b@bKu8I?f*iAGac%J<d2@c2o`H;iRc<=BKLBo9hh=7!O{BQbq?Td0AtSOnz_r8
zI4i}fBxIkJog~Dsg5``a8sIJp1WEd~w#F#v<?e2H>J*&vaMdAid)V}c#iH+8?)Vr2
z6HLMk-17Y46$R)Wu#Cb?kFtlV39XPTE6izzd`0lr0a%a{7}(PCt<GyYJ5Pg~`O*9m
zo@TIAva;NcA7>4|8j_WkhUCc7FhNb|6(QJWXJhLf9<J<myx5*|!qD)O?N=;L;%;N&
z@~^qMx((i{dhzvnzejUNV#!FpXC!M_-s@`w+yp>KP+1z^|K4Jg**Ywqt#}^HBzAHx
zpO^-wz?qiXChn!qxo!9ELhl91TpR3(R*s+h?u#&J)P7W3mLOYOTTp^kp6#Y4t~0>^
zZE%nzb0X=ax#?L-N|hO`&b|T;#m%VU>BPqui@KTk;srWDc*~_}JjNCQN`E6%*p+#_
zKU`4QRy>1ZRayD8o071w#-pNP2RBAQX1cmCX$NNudphgkB~DkxE_BUY=c{lN%%z2G
z9fy3Os&i#F3sQO5`1!$BQ~jf-Ij6&EsWf2yxa00_puzn=+>i4EFB)!S@nCMIjp8G}
z@;Xql;oFQdH$82Vbuqd)^IW4ZHjvA>uW?|eXf=(S(h-6zOibgjvOIxC(ZB!>S}3cq
zyT&RJ%yb07p$#5%yUCICZ&c^@FaJq(E_$#J?Hrh&m{*R0rwnaPZmt)mayUEvCV7wa
z@0&iBm*Zx_KnK<}LULfG2tz-hK{$A}_-dlO>l_w5)HOHO2|w~?_#m0Dt}KJ)42YV~
z>iACCsyv=FK*Kn#4|Ux@Se}9zrD_A)^L5vL4AjTt-1_!y5fn~dtJ3p630bB!pFfj8
zWNRwiWrorBGSQ)L06}f%jva<UiiG$k5YW-xz%jOAQ^D-F3vS%!?Cqnmk^+2Qby*jj
za;(s3XpIgi-~#lR+Mv&{lUn7;pZvTNn~8w|(yobKN)7j+w(ZA51Diw*k6=IpY(!GI
zfi~hC#E?wB_aEf{s7o7;?ezCL1wC4Fzcl)FKwv!-DPN+FCFOxm<NxcvhC1|Ds=g*d
zENK0p!&yg7*&~5NDf|C~_CrVPPN@*z&msQhQujY=1iqoiEza5U+Mx1PU1zDJFK(Q=
z?8tw_!6=7+z`4dj@MEF8TAXgw%L1FbMQst+0>Wjc`pMqP!$E9j_Zog}Cao)7{N<Qf
zSdJMOB<SzJzM5vFSH&C4Q50kTXQ)n$25Gi$+jaw|W@g5^zrZJV;+0pA`4_^G5ilDD
z6E0@70lPtCT}?vXi|eVnkDU&9GsnNTpLDgfOeR^O=sFN$hA%iDg`6zjoo&L8Dv>ii
z-nqKmiLbGt!3l?g6`jW<SPdUj1L~jtPK7ihsPmRie0|!LV?g%MqXn)(fMV@+Xs|hI
zkHCVeXVXw~*4)g@U<_71*7=WsY`|<ns-?H(ZJt^jqF=te^KT#<n&8uGC<8|6N7~>F
z5K_DS!NJx*;I6ZxZAxQZU3FkjiemtQvq<JxmCe#zx;(hrqlU(e37gxa1@n|O^e_w`
zK1>Dzft3&523T`wiI7OqlFErk2mINf1+U0_Wnn8jab|S2dCF#caX?B&*yY9h2OXFT
zJNMXu5h9a7;p+Y3?_`4CycsJFq?NO4;b#dV^Du9yE}aG1lJeRae~Yxx3>gKpMETjM
z%p9Q-8%pgfjkVV`wSDB&N5y_L_6Vv}ORYU%#eIXqK49NrJunQ#M{=^DK;WC;%HGX>
zFYF?7Rx;5GVMzei9rhW3e7SyX+6T-MAP$~EOnt~3?7=wUHH4W9^mTP?=V3N+F71>{
zk(Tr2d2DKwcQVnnlu-c7P^EfvED4f7e2i9A&7Q*0d}Sy{TsrxPgu=qu_Su}QPTr^-
zg^sz-Q}-V_DJZ^6O&w}6oXK%9cAHDjUWBh1^y6qskP}tn8$JLIfI~~#*_)<;J;^Zc
zjIY;OD%>tU+roE=;YMH<EeCdiCc8GZ?<)3(BzVS=r>5;((JuW!iIQ9~AV*?izmO^>
zm%5(+Kezy=F>JulaiVhs8XipOXeyuhzOE?6jG4FpgB;+mhuAhM0f3%h=x1>LosYis
z`W?8zkS_9nqh2H>kF9Q3iFH^qZl}+~G=+-?RO!->t0uep9p&*ffj}^lYf*$<sE<E%
zAH6uQemX6HFfU{FyB8?`@R+%!6i&;BWns7pmbJ(R?FT+M+_La>VwIB)VGqzqdwYOs
zzXk@(?5Ca|JMZJO3?<%mzw%4+&{*RNeLO}PkMEQO#?+<lzrDp;#<NHQbOdWICVXQU
z@M*cHQ1mjcea52%n`+K??m;*Ijcp_kt`j<gM>mnGL-KfD1+=e8(*NAi#ya)uEX}d)
z`Qwy4N?NOnVL}2Ok4zgXcQ-ot?)RIg3R%(A!6m{^#K7?ZdAi`lCnlDIzgtogc!G%d
z6nm+)?c9&m!vX2#Ebi;quP3RMI>KO#JW0zEKAQ)J%ZEjwK&O@b4x%Fkp>`5W1q(~d
z_B@YpOVJeWQD{K<Jbs1+Ey;12O~2IN@t&t0brYk-KruE%Df>KO>${Iqm0NVbumCbY
zfO*l&+Ce>~7I#-8d)WzOM;tFO<3pr9TD(NXnr=tvt$@OJgD^4xn-Yd~>o(_5_vL?t
zjHsa@1IaV5UmuP;Lhs)HEFs~#^2^jz-f<ylQci4xlK{ET&bJW>L!HMiWBvNz`hsYQ
zxS-OvcARXR7{~QZ`SKTpd;1!KSfv+_--M2fl+q)b<JWrf;M!+ZnT$+(`0PLi9~v=y
zm6-E~EK+s@FEjI{a^)<1iK4m5sj1jZ0QQXM>3I3x^h~DJVSEU<m1_wNfy;yxA$ocs
z0oIC3pgCb~Mx%ItXsW8$38er|B+{D?L%@3JKnXf{%-FMrLM&JvZ1(d58r;&sjYHbY
zooD_glqleCR^T*z@#xIvF~hPTgF}qU(vGvKm;30K7r(;9@EjXC1wQ7ZjeL8yf#ll0
zzw(D`TjG9$igo=TZtdYJnFe&GDYeb$d7;lYAGbro-e6;*%PQGKT<R>Io_01!6yLtR
z;%VKS4M#nTCyt-be8xsbA-!{}_)8_)U8ofZ&3d->_v{BRQ=bDGK3{2zy6c*frH@D*
zGCMSo_(kjh2f_{2F<vr)tt-qD1eSXGRbG1BJhMSI(jB~y^C<`p#gsA;W{5j0=Z)U`
zcEISiJz}#Xn-4Ih;v``Ph8u_%!prZ?i*nY^az4fjj|~p_t^VG2gXBNN+gKbLFj90Q
zmoMv2btWrH>a0%itx7zbKy$^?E<0aVrh$&4FkKCY3yl|sXBaM^fRKEJb8s<6SIUgl
zjBdw<alp5)4KQr!j-R<0zE$_wj_p#HKLpms@>Rp3jWBZ#5e?`}KS%3z7b{j4_x7-}
zY5?tzM_lB}L-`6)AN8!xM-yARLm106c*f%tg-HkmmdnC^=IM`{_JffwH6VWsr^x?8
zBbzYrfN6p)AEr`VAmE@_romMs&UyxS2bL&*xn<A;|Fu;3F*Njw3xOCNp2K32uheRr
zyD|ChA;X?qGtoq^yLJWIB5F!kmVmux-r1;<z<U5;MCkMl49Mfw2I&Ou0_28Fefs-l
z(rE&!?n!iwcKASkJt1;8Z2)kS>7>!5t_-aNE}GqNNOW{Nsu;=>WZLwp?rxXv6h35M
zJgIj}k;`WA53Q1Y8v`TJ*Zf`Q{Q3KjG^3+`j_6GF)nDHreLgX`dKW*xzt{r@JG(D>
zU9I~7%YcB|%m8sZ;QL{b@tkXOi-$MhS-=u{qHvS=xd?BxT6^~v35F^?i~S`hD+{NE
z$L(}xn2hm)8V8cONB8bYnzj7pTFJ@#oj*T_Kh46yAs@Bd1e2YD^ULh)h~?vu%nzex
zljaItLTK7>mLel*B=+Bc=9+iTKwkfYU<V=waAuT0K+Otf%mK`$d$!zu$sBK#D~sE&
zh=a?Bvi7q(Q`kjzwH*(h>=7Uy=WSfKxI8wvJf~f_W;t&EwCU;rc$_Uyglh3>IgEMY
zuJ#R4cvu7#=H+Y?RAATww4>94-~yZP90{t6{{LTGwj35`SZaZ&^w$s&l>`VMJ{Pot
zApifJXt5lH@{L9UtSMM<gKLK(kEa_m-jyp?u!MmNCOB9y{PU6zt|C2n!?fVu1a&GL
zaCC%0ti^*K@n_T00#%pG&NbMMq9X2BM{y$2Xy{J=M2X{sld%c=06#3yig26Yg|=aJ
zHmh)}z-e(*os_e2aEU-I1$U^))&<`;Tu%NAzpgCnF5h$XLoTFDE-o7`z(sDGCKci6
zaIdm=EN|-0!50Ep=4SQBcwM|%OqEK9fUH8dLc}JE446xZ$6|O|JhYWl5ns|8Dk9v-
zVfR)M#=K1@<DO=9CXCoE4vJ)C@lAaaW?!}+&k4j6QJp&MSxKhspbY{HxX9DfN{VZi
z&o98(M}UQnnhad^ebI=k6{@;|EyLI5<{_5(SFd)s_RJth6yXF-8}8h`z0?)Rh^-K;
z*)B#EA;KM{91Hycfwc?_SWb~i-EMAKP|97toZGVretQ4%m{8P)!xG$(;RFNXlY;|f
z96~tzD%Z^b9K(D${8?rFwSV3et}NmxU+n+6wElOZuPrRQ{~J5daqnItSxpr~E@(`o
z+cr`>;k-wF+mX!IIs2*2&+fa>j6<?yv(l$^SdVcSqt6^N{rv!gA|#(ze_tk!mrzG)
z5dIaz6Xjz^w1+1xJ(niZZ(02?Z|e%uLg>sxS7^R6dw(phRU+AX;3K(00}0ns<}k!V
z@9Ni2Zi0}rzFtCBw)v^v+E~oB`EUvuH6wL%5Y<wz{?<tWy$PCQ3ybHVabK#!i3&B3
z!D@j&>~C)=*Qu^D{mZ;USJG|NWcDebUuE3C|B0x}+5ZL9OJbLIZQqWIC4!|AJLkwR
zB*WqW+#Z|<n)B$7C_~@hr#9EPmDPvE__$8TQxG3$A3S`>Frnu!aV~}$<C+g5Ov!T_
z>g^;yrL9|gVcn?U*yV5hTDI~UVRj|QF#w=bO-aGRb>9__>G+5NLlrQaT)e&+mCNI1
zZvW-|{zLTilM0kSz<rBz8$f||!y+A4j{lHG7rUPq&PAqYWr+xcXdDN<V_Y8zqhRJ>
z$9B;h*6Msc9yE|W&iWs4(L@JSvS;nTWAI8NY+Zon)zLh?FxAx)flvs}5OO2C@ZtKZ
z;vSh6Or_0risWbD%M{nah7y}@^^M%zRTm!#2Hco1zfN6{nmP(h!3$@<DnfUfJGE~7
zI{7{QoTj<Cz=cq1VlE+6W%KB#v$81#iJWeM)?8g6VK^rjdnoa8B^>H<3B2)=4P-vd
zLf{~R|4@K<0~84^dSq;4jj8TD`Lm)BewCON@zlIhY!a92P(9#zs`|$HBFot_k+a9Y
z!=;Bx_kAv-;;KdeB^MV<(>D}Ys_0tgd7C}Hq$>|!E<CnRYn`&+I`G+ds$7Ry=DQ+*
zs}{9vINrMY0a4`;QX$TjeAzL*?wUFJKse&7|3oU{c)FyF45^MPC|JQE1ijY>&zYgC
z$;(uPmBf;VOH~n<eW0~svbMvrk`w|s@UUV(h<avXVn|urGche!B=tYpf+pYKt^)u-
zL@#CpYDAz%@0#H#2+ces?J(1+Hom&Ku39znA7A%g#uYxVIi9b59%2kjbU|kmn2zgg
zku~4kc7!XG+M(x5^#@m;^1(!HAE_wWi;XMBBGrd8pUf^MoDWZkl@B`CFYFrb-c+oe
zN5N%IPji)7i;mhnkY>L>lui@h^G*nD5&D?nQ{Y$85PvfC%+-32*r}-fln&rdMusY=
zusU&133;!Fwf&9<#)%&l<!jfJS8sp}(lD1D{-sr?Sn6!jJY^6q`LnmW^CyhyW?kXw
zPmGU`Baj4W4Zb`A*|U4?)6{HgK5Q=HQ*ytnK87^j)e^pyA~WwXf6{Wx0bD?kntv5S
zSpwcDBd^jtEHpmM-=d3KW9`k&-L?j)%hJ^G@opa56<=|Aqqf!C2vv>|h^D`77f4G4
zso#TjCFMN6$>7!VC~x4~Y8fDFR{=38>4w;wV}~aYcJuxRDL90Hdv+e*5K9n_c<^Y8
z&20WIf&cm<fl^a?yw7R5b!E*>&b)Ox3^JCp#r9Ui1}!;kSJhr72X>#OW4Mke9|g~X
z#A{4yS?O_cTP8*1Wn`LQU#_Ck(b#xAz8-#NQx_!}Y|%K{*zmU-qPHt7^n+a_u!VyM
zZ_RYd<5@K_!a5F?(U4P0M&M0itM2MFQ3Wyuc?_T+X~ehttuCWaZj4!LEN`-*3m5A$
z(E{PZg*W+Dmx*XyKY-Xk2f!_#0m06k2(9D!)Xww$dsptm_Tx|Mi`so2ia2xew|BiW
zwnw<c+Abr}hX*LK!D_{k5f&DP2c6-@e%ZIDmZ~paxS$AQJpx-A4F9mCi@Y(%F4MLh
z`44F4wr*WR;H_dmz3v*851?w!PxOG{gylwLx7>(0docO&B_++L>d(?fzMz0nE~YHZ
zB^5Mxe|kjbQeE`gzWwe!$<FY-ba9I6z^S(Tt5KTK1cpUL=`5y!=b^B&3EfZ(WZfM~
z+Uxv3B2E5rNjI5r+81&ZeGrEFR|d&RUKViv7Dh*EoME9|uYdZ$wC6H#d#!0-Y+E8U
z$7*V*#kQsXiE<{OWngUKgJ;rvFsZ_d@q<Hh6+eC~aw}~}WMU0$@9O^D-j%drytxGJ
zByNJOENkEGxX4@-E~-Kkn2zkM;qMOdO()n`>O4I?k0niF>tAXPw<O35&{1Gh3#bV;
zU5`3q)}yV-$<2lK0!{as1re_e+{NbU7_mRaP`V{w(4Ywn;2x@>Ck>0MxsSc6!IZIh
z2r@m5XQt_fb{*kO1*mu7)?f?Z)_q)R4&Ek`(odcgc~prm^Uaz3gYSIMCXzN|@O1(7
ziF~;2_B#Q@zIp}K!>i<EL>;8G6&$0dDFLjlv2F|*-CIROtiMU^TOX3#%KSWb{C*@$
z#fv1MQJYK|w!GbIk05I;yZsx8PWwPw|0hp2KRCE1oOYZLmym$B###dQoj3`+36EXj
z6>sl!92O3D@red&q7Y;*sAAK6#cU~|H){0MlR|hZC4pO`HP-SX4&3bQ*5aPKXM7+9
z%E|e*SdRrD@@<JUc`_LB?jB^bD5OogEA2>ELWZKgmi_tj=>-NVoEb?WrRK~1ek=X(
z$?;G+gU1s)8F1Y`HCeDYu;O<3XvC;IT?dLgHvh+uA4e9`0><I6F~(*W+wB_h6SyVK
z(>Vy)i{EY9LrRk(o?a_!B4VnosQjznX!d;~hRc+24}2`}VR)VF<&K=NOx4IXqG>~n
zw;230W#=-Fx(rsNfhG{VVO*4bJ%NRnwvb-^@C=AY8`gGJE5=^Caf1qh$&oW`CteA9
zPUl^W0Ny3c<B4z0CPwQ`;~;b_zr6J_+_xdV<4gGNojWrJ*uFwi4W1^MIcDfbUIHHd
zT=}Vu&PX0%?|`9RVDx*$_SD-<&&@$dvwcBI6FIH?2OO)o;_nvz#>vLMd>oWCn6QRL
zM4V6mbqD^Zx-hq7i3O4w85=wHL#Ixy$9A0$@~%OQaYx`J6tlJ&W8Pre@cB$L6Di=y
zl|I@}CwAo0rSZk#@nqA(5+g{(&Hcyjrd^;&R;!)t`9l|rk+)NA6>dCXmK<OLJm<s-
z6i1{chC(FAws|w2K1>aO>_Jojykw9J+YB7X&j%oQN`(y{2R*{>_o2Yy7Jw@I;0u;@
z*M1)J0VjIxWIqHHC!ui!gAwZ42n!g+s>Q*|8k_#$+qZ>IZmpW@3){;4+5jjZSRr`d
zU>QNS0G15t;R>Sb&re|(AW^vT&%m3y^UD`?+pn-lFg9T#ZPTm52&8`22{d$1xc}S|
zcjWPtCknln2AmfQD!z(qMQ~jfO0Thz#$v!wG_FniqFjUy0G2JRg26+nwbWwLslJ!U
z1U3+qzr&gX3pdzz=BB09neRPtptMm>6|7c({4<rND-C}xqe<b?R{9Oa$#1T-dX5#=
z+l((w9oL+9W1X`3ZndJs%*W>-;;*NpV**QDTtppVDZSeukVDLFvTF|81Y#Vk11!1P
zbKM{;290?V*d0w^9G(^dzgyK-P)_)CT&P7@Q1}~uK|w)mm%l8-0W5*ieailq&?b5U
z26pQS!(3Ofh0ph*qVh1rfrrCIe;u^FCxz76I(K7YRA?P&;R0;`3WDjlBSmwMI@jJi
zfG(@T@{;wTf(V@ioQd!ihlg#LR?1X2CQhVp!`2J!GO8>5arbaDL{`7|YhM5%=F6E(
z92E}^%2<ZvZ@0d}zWF2-3<^__BWH;%3haHsk%EIRpk4f{V0(ae78e&VWHt?AYt6Y6
zLRafd+!rhj->sV+qYZAhJoM-vzEwJdP4lBp`f~1Rw*1z+V2_8{pV*t&Xn(P#15{R!
zThWl3?Be6%nwpphf)Q9{rQlgmGF4)0zYaR<Yfi%l4Tj-=*v8jy+?erH29%p_h`}0S
zoXf2eH5im`!`uTzZ&=_&>uzFU$pG68Y+@`A2t4q8Mt)eK;hn!uO*KH&7D9JaRKD!T
zF$~_{bK2VR@w5sHu*Si<H^rVUntcLy8q?$6T5PGqRQEOmCAe@QIq=1Zg`|;-%RBuY
zTV)J>-KK|SK3d;Pm(q_VjUbNW(d3}Br;d&ekaP_VniIF1P2ja0q9NEy0j?4H_wc68
zZr;K+ad8oLsSoxHf*+2?p7Q)%8!_aE&lRrbN9%968uje!aMHnVQ!OLwS{oZEo0^{L
zQwFa&^ss0y%b|M5<OCfZUD__CJEMfD-?9t%4*)ErR}yaDxpQyP8##x~(8X1hrYywJ
z89V-T+CzwQ&@Z`vx7QJFG+mi|1i9}M1o)E5tLib2!55E;L<Nxp@$v)wu9I5S5*x8_
zGZHm$baZrfeyb$}a?d2Vms*Ch9Q2?73b8<t{diL@7-FsO?YWxb5&(7xeXqB93nd2@
zK_JG3E$$9gS$sJoP%LeACF+?$vi56;X-tFFs8S;%^8N680~m;Lezybe0bT%R^4DS9
z_c&fK#H;`vI&5G`vBlZ?pw{Ecr#$W2AV>oRr|3etdv9|I<xtCt|AAB}k~*u&@0uka
z+hE8zT0H$LEEIf~KIkU0uDd3C>fsX{^8cgB5h+0+UyC<;w5}dhWZK#k#EWBC+W?Bt
z5(1-K5xz1&Da0OT$lA5+y_I>fHKUi@f^QBn^_giR*`6`os`)r+&XwJoyDK*O^H|84
zH=setu9qZJY+PaNhS+HWOPXbzYSnz0@t^OyXl(p#NrwSX=V?-k{j&_FR>C_^uySz$
zNPgC2i?poe!o%Jwd<g6)1OD+^vCw$VFfQY8lD-Gof}{hD!|en>1bGfZEJ39VmOH``
z17`<g2w)x%FJxpm0X!qyMjN=TG6xo?-RuAnDAad49_@jv>>2Wh^8{8{AVQLAFFrQF
za2;U~=+R8(KzRZN0R_Lb^v2*?e2OP0G~&bVU&d3I-rE1*sb}8N++LCWEM2G}E>(88
z_&x@g;gI1w{5q?uvhmo<<h2&2Ef(<j=AAS~<JiYV5fzO$()aR<BLdO@19hUS&Gy)X
z>yz~cW{I)YRw!Ly`!2y%Y;|mPWeg0Gr*Geag9=5?MZLo=ZJD&@=Zf2T*u}0)JI}X|
z5rvL$vwyM&{IutB=o6kq_lY7E#-auk`wP;W6=_ii2yuCNn}P>C1R6wm(e3}oYU-QC
z^A{H|QiJ$a8iz@vBBa98ofQ~AHda)`YlA~heTj0cZ|kb|uFJ>C2QN{BW5?8Fq~G&m
zeMS_)^kxQ!DDJ~481-dZ>-qR7$JU<Q@nFXT+xlk*(CltsyB6M4H`5(D3s12P&G0k%
z?<M?xb1l`b7P0D84Sz)Y*S_T?n+tEKwKWGEhMlrOyL$6ASp6BqJQW(kTekviMD78=
zA9swP_!n*8<zTS@R#%e#y*ti-*GGSb3@NxpOTYXLAP6@D>_owF!7u|lL~tp=1S|O~
z-iq+cxY&``ii%uUmj(@IhV$5nPLzt9&BN@?MO-0eppKDA+jr<Y000K>h9mO8MZ`DZ
zuN7py(|dE{nSi`o%`9tKd3S}#zND*&O|ZI-3OK6onSW;6MoNE3&+P0{mWx3<v`B(A
zN3kvu`}pw()L6{GFe#4V0C@uLXgFfF3zB-@|JNfA-Ql+-8#`GqC_P+{QElsM8i|1g
zl0x+8h)#m{TIq7kJtRdXY#_R%8{@HtaCkJCEa<z~80+coJ}|Qgem)=~Fn>@;Y%)~H
zIN-XCIGGv-q9=cMmQ41hSyh`<k%uAAt<N|5C)^P|vC=g^?+)&DwC>k$-_Y6Ezqb~F
z$_5_#h8W8pRbmqbV|kD-A*$B8@Bo}5grqkPzJwR#wQEYeZ=O9<6Anj?TDZX@QTClc
z0Qc<i_#DKHs3=7J=0dq@vI}z3MR2AQBv{`*{qT*pfYi~qW(u*EbYhjs>)sU*j`3g7
z?II%C$ks3h9S8|>-~m`JBuzH0tU^XFw$Spn2KaQ1=}2-q%8RS0%zVmI`rzB%C-|@5
zsJBRCRLa=D>rt&nyn}OYWRuExDto%%Yf3)9rAsXNR7Hq#AfCc>bqO&YJo1kImA%Lq
zGW$Gvo4iV^gCkenIs%ko{j@^$-ob5~C$0bIgEscjy%A*fqN`@aIgfo60jAdTQSHNr
z*Z5<jFvJU-*A#Kr&zqzF-ZnLIvNU?C8zv^ktE7;DJzl!WvG*3M-kVoX-u>$z5;-|n
z1fKZaY2u)X2{&<f=;r;~x9JHSG0*<~6AXz*ijcf@Qf^uBlZBKV7y0Ak)_iPYo4<d5
zBVi*`c@oU`@ySV0VjdbKzZex-rg`1QA%H)&Y9(2`pD}wQ<hsE!`!mrFI)A-z9)Yv`
zGHmQ(l)HCp9lR2|%6;{pg&6i6uWB8_Z@g<CDCvHh&X+-s!Oj|#)2G+hQd1Ii8#cI1
z^x`MCSF9}P^=I_jf3AoKN<yv4$&Jt%|Ge|=LmC8}s%_EH@z<a-mq0L(g+<Z&-Tw9R
zoyXPP1aY;HWlcw5Rq{ye)a-1VLpUjJC%UjhZZ9e#2(<V>SBHtaQBfm5Dshc-!t;Np
zCTnWx-FW*p;_*6s=G?Z9{P7>bPa9j^wcgzLdM2=(FLr9rmY+^UAvX9RH<c6o8`qJU
z>*0!|SC~SyFEAH)p28}-+ws%7B^+0$y~*P0mXa<pKYQvV<K*J7lmr|3D_QZHi3JHu
z^_pj5eHBD9_YHL7>7wXD)8UXep@CK%jmg}&dmiAJ4Fo1lC@p>5_~4!Zic#&qir>($
zHG#!11eJb^+O;eEz1~@p9A>M&nlzpe(mbuNnssK9)Rc<Qk#fPm2aT=vG8xH4iphtD
z4r*z&;J9Oi&qaji7UxHe!!u|vh%Ur6_V>$mHtt)$e!-oZni^El+BuWE;m?nCIVLQa
zDDVJq0zTB0d#K`1H?zjFUxnAj1xdkg-};|n3=2!U$dh73<-91}AH@C&(dhYSYAq!D
zy##EEVVFT?Y>f9@+O%Eln#=`m?quYrD>;U-_{18&7LNV~-VXG%R<m8H@57|5dGo3E
z;yK1JMbFEJha~!THI0J<2cMfwT4dYm35K(K^8bSiz-XRsHf%6I^CqA?vU@>VuB0R1
z%-}8?(WU2i<*l@PKhKB$7*1dEyI8M&;z7^3_E*UiRPNF@tR($4@%$y!mQXdnxzU)&
zvv~=*)#pM368r6BLU!)H@%q$R$GWopAt7nL%6Y$cEsa<=GrBrEb1OL05&x8w#3Iok
zvRuVwXk%<Gtv@JYu$gY<o`NLqo?04tctF(c+*w91#X%3N6fC#Ul9Ou-PI|9S0sS3K
z$SF848SyEJG#$mJjd+S&A($j&O5}5T1nqe&qGl5pUdh9E=(3f#T54o~@;LGI>#!=!
z+GCe~ir}UTT3E&sliYMppat7(^1b;cd3cNun<`3+@)I%clr1Bzw>lIRyERoac=B02
zm6aBROmZxip;8p8sO96}sME=i00;Um7mjdQTC~SlIg+>`)Gdxqv$|ZoyrVGW9~ye%
zB!tfF&><=UcP4xAFGlut*DNA=9m>K5Vc>7DhDJf9`Dc~cRTDWX3UBzzczP}(Da#Yn
zA_&}5H?Slroc|F_8tp;}fe6f*uLulv7CBYG9phm&bk?D8LveAYd`N#RZG_z|2^uP;
zDf)pKcDKgCY2dPi_~sCHKUSp4CLBxpWDg4Y4o|~@(NW<z2lzZhwpE$fX2pX)g<K`}
zM_nS}@GZZ1cSX__a@sI1jv~RU7o}QT?}W-&T2^<cGiCJIuFKbmZ~bs7UQ3A5A4!Te
zbHj`PrvR;Jzi7Z5vOJ)v`YJUwCUCa^?v9J+917M73OfG0`Ac)qd46p5x+ZT#=~{Y*
ze^gfo>pmq2Z`&U5?t<88pQF4*NPcE!cxA7#l5P4B<~i7I+eutErVkAXadGdsEQCD?
zh!20D8KcC;!v+jTUK>CoO0`#AojSObl7Im)c3cc8JNl}g{^uQ#+s{k4;aWhzcG@Qh
zhr4#oP-av0){`ZJ5Fy+f@WI$eMMVVynw;M>j&|%#mVxOK1eOCx5_JAsMY}oZ<-Ivn
zmN)+B5#uv1;68ZhY-d4h^j__?1I~9C-fz2veeIWbc9GhHLhs=nC~#QX&Z-_Mv&P)E
z-5AhN6Nslxwoti?aAA9YTSM@%SRp_{%nms1TZfq>1b0i&PN~z0IrH+OX5OLa_M0^t
zpNRN1$Zt{8M65t#OzRI;N+B3{vcau&gVSW5yQb<QGvN&r|6K*#WWoKXCYqmp{|Pql
zX=ZMDV)b`nT?1urZ<={&O5wCy@zh(|<856y;l(@=UZRl^cN@ucr5-`*<`QKVl40My
zu-)9#La1)IV^P2UnwrzO%?neIqhgEhdHM3yVl!k6NwsKf+Tu*>Ikf%5>dN*qyN(p!
z$%;J@8`M^DODMN?!*A=qaLS`?-Q>}?9niYP+A1GT$HeB-_8WxTy0)@ZeEbNC5In)+
zc+v9+Jop9H@uyF8=s}a40GR;K{lbC4#UPo2AIUo_%h#tAqLFX$Dk(`D5yuI=Sk(f9
zKv&t^h&WVC3UHIbtkq;I9L@=y?4w841+S1qhE43@i-0fkrW+0sX*hJiv{iI?Qf&o#
zWo55Zx?xI_Etc=1G_Z-kXT-$DcHmu2^Y9E5o;wUrxcTA2qUvL12r)PqGuVcM)rn&p
zonVp#OsJ8J@*2iz$xUDaqjiKR$uoZi^c!k_u&DsKwRr@=)dD$eqcFXy?2Q4}7DE!e
zjC2z-4O37cu<FgMdbLH@mFXGQGI4SHz_sgjdTjt=8Y;_U(3bAqyAnywoD*MX<2b@G
z(!PGI(7~%{w0*q8OFZ%<7js4hU_lc>MhwlWw}sYWr<^nIMD{}?8M^ImzG(}x?o&j}
zcR1ki^6)52NFa3!I}ptJ9)xNu%{<;W92yuH8IRA=|9O3BZYDt-sBmTYNK;lg3OjA_
zhX*qzv81H)sH~`#x_O$XK%d1QeAYM!(~5WeFk?JlKj=*Cz4wd98QUN%n})AVz!`WJ
zqHUoqd;r`n@fR^V;`2v8$vpE$)qCC%4i^iF#b*5d&*S3G0s!n&6^^`(Q?@Y?-Wb(m
zLScD^q9_TZM51)frmJ<os!OOCc(1nScJKBJxZH1MM^6;BaW%gVw2sADNDw%W=FkMR
z+PR{90>)TA;O-QFVX}%uy0%=k`ugfbS3hkh+pXAtgu=w%eQs}UJ%1+t4)r*k8Jrc`
z%gaSz$Ae-5!8=BlFC39vT`&kGrDK$n0Lrn-#@-ZGjR4<J=Aqi5AlORa9ql{YY7L1g
zwgz~%kTBieetOp<sDeXo-zIbHN=w74W=WNLM_x9Wb^P{AW#zD3;lc)@&gSj*PB#>K
zI5rS_`I!u9$A!A@U__0%d<Xfb+js87q*_r~c@BZg_f(0S>dz3~K|xL{1_tbr)do8r
zKz#8w`_g#VM{tjD!(kbTd$ZN<z42?vVZb@23G9Z0k?QZ(+7VTB=fgstnBh(G+$Dlb
zpViy^iN(_0^$4HY6v=ZN7Suw~#%<iVcOS)L9dU6&nDy$l0IF4O0ckH>S8zATpd|q3
zsxx1Xe~CS{{(pk9xW<njZCYCDAOCjYc&Ml-bu#57bt?^~4TWmi3XHQEIl2!LfT$vL
z0(KRciuJ26OP1kD@-`}f1K(?dMw_?rTe$n}1d=l#tDl2K1Oq)i?EhfOa!&^jtj=T)
zujc^b0)>Y7u|y@Q4Z4XJ?CgYS<9u$=J1VVJ+O}<*q+|xTrY~NUdQTE&;v8DR^Ed*r
z38ZCo$LbuXG2A0Wjo;qPX+&IHfj1{%DB28gBFPTyf+g(n5bRi59LNgq8R%O`A0e5x
zj~~@(0*m{RIR8K`j_z$+RW=-~t1RF)fT8=#GDamY2-aZR0&XwOI9+?Xjdalit+(PI
za!iLKtYZ*O4sIH`adjTNH{%b*B5nw*iab3$mJv<$_mGJ*nrrL4()220t1xSak6Q+9
z(#3IXaJ{{wz5~b6=KbJ7FcP>soK46c0x{YIfr(gOWfpN=LA+?wt&XyS6l4Yr0CS<4
z_+1I2?(0`t0=PS1K=wfR&yldZ>M<84?a;5oGy3@o&tAW_H#et@s*07{90Dr0WM~J)
zQ`mMC7V;5zR5cHb5nbKgNA>l8j>KHv9P-|>l0CT35e%oFWp#CTqx(10)5O;gAdClO
z_Njqr9pPQO7|3Nq^b<yOC(;{IQ%RLI$py!%)B&q50KK8*mGG&7<@)qA_HfwOf0+vc
ztRb~?XJzkNGEg|P>g%Rk!DlPkZEVt%`Q<ruW2iGr?Jrnx<yBaar$uGg^VG<!EV*}a
zw}rBXopYdpg^G|M+nLCsYB`cd5?D+Ss`JH+DW=<?1cM?AY>W8#p4`FTl>YTLC}ru}
zN@N2MeI$GJ+E{?-Fxgug`z<}$Pr;AUrlj=!hi*pMD6%MZ@9crQ_1(K>iJ{TekJ36m
zYqCwGo8FJnO&p8N^V{MQLPO$8_Sn-A<A{RV1^Vi>O=r;W0Uv8VY)MH)1&Up#YS3x+
zy3<`~uYM{{105a%q@*ZVh(s~?^AN=L-`-vx0G;5kj^`IFa{2PzKGsp)V5RiosU=9k
zL2C-1E*Z!JCxRO{%2$A4*t$+#zrW5j(<(%8UmpxqxE%5*!tbw(rI3&~a+JbvNic7D
zcFuOA1V`FA+PoD7os0{&%{w;4P>)+sA=3QCv1U$snj0m|!KFa1UtNh_pmD#ZMnQlA
z)t6k?huy{R#l=qz$^<lPr}I5=%CT?S1jD!I2?<W}+(_%nY#V-yLl`7{FwKFfzcW6}
zUW*QpYzg5D<iu0FPd1$L{=uzefP9;Y9Zvg8JN!j{uNRfi=<eKn-dRpw9+{?FS+vic
z>yc)-f?nZpN!5}VnH~-H{R(CP=v()RvLAYJEdLB2CnvH6J|Oal<S=Pu%rNDIKyoIg
ztnBkSqs->pIDNcPe?KkAtc9z}_a6ip(t9eaYnCX_`5RSAD2l!>hW}-R4vp2qlsY6V
zhN1@BHsbjPyGzL5TAZVgt4o_3wcB>p_C8Lewt7@jSaKdUF-g(ifmU3r-9;f9vbAtf
zNW;R;h9#-kW0?(K1#G^yZS4E`lh~`NN!#i>roW=3(@=9gZ=yoBL9dmf{L1l78|^?Z
zgDwuh7kU@T-{kSk9hn|HVG*p@+MD`y7%O#9k92sr<JyoUwr$t0l?pvy$^2d(>8(m7
zHg`mgskvl<e$=Rh4$&Z4++KQf#iSx}ce&yI{MhDtZJ~SLznicWA!se)&BM^Jjspz*
z5wisBoBluP0hwc;06BhxMr#92AO>U#rB=qVzP^V~o(z47v$LXYVZ7J>VayYoBWN15
z*{BqvYpgQ=8-D&wmY#$yGHr!32UmQK(44cHnr6!3!3Y2pYw`tqHH}h<o7CTy{Lc5s
zW7#^l>;eHPV<U#Eo^Hs#{XDXJ(1ZXX$Mgi;b0k5wHkQn0FM_-g-o)EaMMuOp4x+ij
zi+HNP158;UgF))>wg9`kGuQ16wD7?Wb%@f!ws+e`WFk#WTmo4kg96bsAboytcO=Zx
z4ROZchsg5r@bbDYZ{<dx;|dRT5+)H76$O`AZZ)RIPT3ou#^zh!^pK32TM)*Yd&wS<
zjfh^9<{GCL#F7Jj30|Y4DxH@j?t93&+fwDth-c7&^Y?OxGdqR%kt0k567h5mY^oty
zM~f{_PXH6Z>-l*ltkD5*`Y!eGb%KUXu5Di4CN_l$k88xIqvXPw|BI8Gte>ys&k!_$
zkq%LI+K;*Z3pz(LC}yzz<=@?aY=3h_(qa}Qr<Uo4;@SsH6u~q)l~^X|tT;J6okt3w
zqU0r1RD9uFgt@GG?)Ip(`}Nat^)}dJSP6|{SBw1HJ5mwRmXZ;ms+)AKDsA{8JO7`{
za|}P&S)j2d@dm6vs%X{K)rlfNji6S$oOcV+m_p*L3(lrjJ7H4ajaD4$Bt%CU(I*J+
zRmjf<)vThVWPfrKn6nu|qkVlE50_7GVrD*SY6?*VM7pICni?8Wugv-dCVS3FMl|yw
zSpcCa%GOdO?)B*ME`Jfr@M61Dn_U>yKAC_T0KPGn9{4W;R7+D&{&rS44U8FYM=h>y
zcm}IUFe_D6JM&ByU{wF9Hx0`CySwMuv0pYaH3fbf<1=gY>dSHThR8G{^DcWKXYA}$
z6&Du=sch@kQY(64_F&KuK#_oe6;KL@Or_GI@?PF2t7bV@pfQBm2Pb_QSU#twnq;r3
zO^~oyTdne3)It{Gm6u6@zJ=tAgar7zw)bYquPLvUO&Tpgh6Kc0Kwz=Z39o?j81u%B
z_d)YH_yLC(LTTU(AuX-Wi%7vtn?1gWUL)rZd3kq`N})T$4$LW=n+$2hZ3N*O?x7UF
zIxrFdrGuzSW~$<5LoI`Crzm@b7?>O+1EZD~k>@vW8s>Y-7=>=d<^VL;bp|H}$S{@o
z`VVL%(L==@F@!t}Y0ThT0~ofwEy(B3xz9dOB%~E=#cH?1Lk*ES!jhuM9WxJ*!y^c@
zD0m>k{Qz-Jl7;Kcq&Gufy?^q^%U?nu$>0s5Bw~`WlWki%f?MdP#@iEY%Y$NqXvg5g
zI=QekJL`JvGFGI%3Aef|5LkZyK3ZwAAr%m~^2#dl5AC4esXemP9HApnxxdoiG4+}j
zV--0__-A$A#JY#+33JVwap*Ys<p3Coj_#u~B7rzaYLmYXqa*tb&Xv3&$1|7+VnyBO
zrAUkA3k?m{B;Z+ipiR4{hy{o?fz`VOPtov<XG29$N%!4>_P?Z^f`Sfjj-cg_<`Oqb
zAiK~6`D*5Fy-gnW4)CXQsfAS&*2DnXs(UY5hpGx%&Cp{^z`%@sKGe^=&PTOte!gFD
z+P3yK1wZ2I-?_<N8;rJx&B<S5mlbd#aT8)YZ7FF4UI-*HTG9-JLi9!KAmZa;{m?ox
z0OJX`qF4^a{*fD@UH*m32{RyR#*V)XBHNfsL{ws53q35n%+OKV6%=FhbpPQ)<Ggoh
zBw$baabl+oOeK(g^ZriK#&x9tkL(Iwf9s<)OSo%4{!<o9an4rAvoc%lj_u@4LG$CU
z5pNGu8kj<ZnF@GpZD=|C#;}&er~oO5#XZDxA3t0J*HBF7SS<Z-OPdQF&e0~WTapJ*
z7l7D>v-xIihg%2S4?{HWsXAP0FK0vFjFAvU91r5pd~6qYZoGxJjrIey6_!1<^Q&Ub
zSa@UT2NMYIJkPf26)ycdV=_D&Rtx$&v>C1(d|_*6hw^Ru$%T^yyGNtecfl`GwpW>M
z%au~|*T(OLz;uLxFK7vO-o6Ki6oaRDkSj<5pcx?=<O^ish-t;yVPHT*2o=Hn9)mE{
zjvLpJ`w{l?^Dw5kWDJs0c0V~kbwGOZzsyff-q%r&^F5}733m?<OA8BRf<oMTPdF?Q
z7R6^QU1sN{%nfSA*{)*3Jj`B%wq}Yb$j^`d{`eQ+>uLQPYgKgj<vkG4FV4@uegFOp
z+O9WWNtvmP42DSy<T+bdxO;la%E_J7*XIZ+OFsDq!hYlX3P2ePoO;(0DAMrU!2J#+
zU>)@p3{ep2z?}F8l1sJ<k_uI@THhJ}8Xb*^js^`2|ARvKZ`7;6e0XGR-*^jr1K1nE
ze*`Hw=z1Xkh_81a`jP}S`JM*1r{JeU+l!eQ)*hwiN0X~@T;OvGwD!!Z5>lm0CFi6N
z^cNbM#acwQ7a%U?D;U@UZ1Ki;f}e1d5qovC6%OLb;pDCjt6JcO*chV2hje;se|qZ(
zh}@B@?8lR*0d=h*QcmZ-fBLj&Vh(a!><G1zypQdvo#@xyuB_Z{AsM~rGA_=?rHmSI
zxRVH9h9Q0=jlkCC?Vo&_X!%Q5a$+^&p~=VY4xk5R5hUR?y-u)zR&Ja6vugRrc!I8)
zG$BE-z#-vq9A;vIf+b26Ehx2;=x*cdc|i->kye4mq_XliM54IRsM>IBK>4MB7ajfz
z8sLgvCnv@O{7i)ZjT^L06+v(1K`ua0cbudRVb+>qNjjT=Gj!oXyqXFqWRrGevpM=*
z`><E(EjDa$Kv;NceBx-7!V3M9d?{`Y4vvi*pC8%wGA)Xq0{Fxnni~BbGJZFgq!7$L
zG5dv_U2gfyL@t*kxV;Bpg&@P1zoFHG%_FcYS^|v^8iFy%&p++QK<liK4-CRg9KlXT
zm<86K0MY5_QD&nBY;HxjZJQ@gF}#|ap*YfdzJonj`_MRcLEDh!p`ZqDRK>A8v}9lu
zyr#myz|6&!dl`1~oKm2`*{wxaj4q5l)-@6woXtEu`RB?Rg0J`EpvL{duyU)K&X!Bb
zQ0A`GCO$qtI2#a88wc6By0(Mp52~mI2=<1eyxiQ>1bZ+zQ4eEd9SWj!q12jm&qH=$
zgpRC|l8);-a56!m!q68S%>O@#FE#9U-7a5-%?YL`!vE@3>Z#hFxABG1tzC=rYj9wo
zZqOM`CnWpT6%|`Hnb=5wyIuI~FyI+&csQh_+^_6KZwdMg-VyjSKPNT+gQue-e3?Nw
z5Zi&&UK4^e(#P<P-s-5S9wKlki7%e()jn}Y>mo=_!FxXK%$waPyQ`6sINB*Q@gy0k
z%fRmgcRo_d?iQOtW=!TRd2cGB@(1P2bXWXjKbTAS&d837BSZtVR)9Lj-i<TZB?Jc_
z-?mYRm;ci-F~Sc0oRUyNy$z0e5v~i>j6X&utm{%@y76JJsns_%y=WZ##L_u1@KP;~
z@CTv+qwcx6Ias%_2tXT`5zv}6`>k*1Erutvt-@>9_swCz1L<trwom4zGRn%)RUtwL
zPJxE=^qJc~<>mblYb=33U?t80z)y|R-AN5Cr06jgn4y8k7l(BDzVm=>fK8tg+5)Xr
z`13%Ju{57uK$n4_oZtZ;p!IQa6a;J_e|(}f(*SmjN6+{3S#E}YDDR#)JqDxh?AbV%
zhLFQ(BH;H@C@4sQj;Sc=gxmVRUfr3?5;z6=yu85@PE_&*>2BtsvXe(;XJ_m;6{*%k
zBb|kQ8G<17g(re4G&JL>NK4_5250Kf(LRtUwRX_`8B{)J6Dx_T%Y(Qd@P0{|)93<b
zElO-Sq_d#{!A%9GD}pOkVq+gXkZIKd;AHULUe@g<TDSG<(fh}<+ks+eBmnz)$<@C0
zh6c6C{-m$KL9z3X9C3NwmIkRICGn{8-LA`P2wZu&TK?#=iM8PsfC9msb61Mlx%^VB
zCM|k7t0smVP@sMcEUtqql)$mcBYXGmH85xfBj~X3acgUl1+~ul`hBUMRTx&LYh(#{
zR~nD_JpIuREV6<(!|2gYG`yZ)CH9a4Uv$Qv`7y<2ch~u0qDV+6%d6ogTC~e=-n#W{
z)+v(jdyRJroHUGgAB4@-Bn3)Xrths8TB2Gp?ugbaF(qJo>_K9};X(RwSuY#Fp`dGT
z#c^7(`3ty?mz|urTd&<@qV$I%9qugPNTG$?DlDvP5E~)-3#QAu4+A(iZ~g|eQ*rUV
z?Ik@E&?KOfM&thSX7JBi7*OdL^tM?@hTu{%FnsQH3Q})qX|Y*(sS&?rL%48d4;Yy5
ze!C{hs~iwzUmZ4GrSu0A5XBm03FzzKFH0HK64#9T(aI%wH0f#fzxfK>GLm*2B@GYO
zbDDjKhFaUDy@IfFFD$isr2LRMJaSS6A$0ujWWEX0*2)qH_dLEDa;kr+ZIw?2U2Weo
z9%5WOtT~WSBqW4X$6tVFkambG3@1MsnFl4^X{~m@dui-P1082~>%w(>CR#`W3{$TR
zWrpTLa~oe!X~X$6{~IsrC#`M!Nk(!f9=K;y5yBv+v8G(;!NwcMaD<2i4k~$f>Zu$>
zy7sN;>!JEZ9L^vvDM({W<2O{_?z!!3hT{1yR{MfD2O<Ns#f2#Veu0JxpphzbX#rgA
z;K!Q6;c%fq8v&WS%atg*&End0)C8330JA;0@&d~f$@w=s8-r+keG5FV8XA5K4o2WG
z$Jq~plzx$w?{&O8i-`lBKVI+gy{;{djoVB6aed&7kQq5T-u*_SS9bSt=R|~FWTthS
zigig(MXLXiSWcEAIHmq-NsmX`WRgCRyt_M@L9fTygo}woQ=l~3G4F~$k1V6wk9>SE
zY~ibGE?4`N)-9o#1glXX?B(<4A8n)&%>bt@<lVsE>X;;mTG9Y?3KD~G_f_-s7uc)}
z>iNA&NkKz4GCV98$+b2knQ4o31RhkdZOC;;jsnY5<Ir@Y)GFZ}?s!JmuU)(4aJCk)
zn>`p_pP)FhLjHm$U1(2-vqE&W6&5eB0RtWkIj!q;R{f<tMn^OcY8X{DPxKrI`v~5B
zAY?Z}>cLVdv-vO{s?(==w98yyJ>+n?jsk#w+`i1AQN9F_DxA7<b0ecpOs?`SOaKQ%
z?Eoo#k|}9@*B6@k^O0q(F<kn*DlUfuDtlFsNcJBbLuk11$OE<nYG!PlgLL>|DOth6
z)3UqcAG>8{)-En?e6pMLy@Eg+lfoO+-EVwH040ij9wY%$H>G~c`Kl_NfW3~O6r)Dj
z=6l2Jgct|Ix%Tr2ZaXs85ceC{k+8dhYt+@5|H4#amq)Dc@WBce_A6R18vep1>3SI!
z|G@>gr^?61W*Kwi`~LxwFggG)5>5ECMPhxo(P3X=qmkWP*P;17tnx8wkPaP~@Kj<u
zTUxXz3sKEM!NLo@B#u@jylZ|oKW_a2-@H{^()lQqKX-z^tqZeey$yT|k3!Vv{dKs~
zM*7|jtxuX|wXe>{{^G<vVUfAQJrg0ZIv-YG{`}ozL<1ar0Uv517Ss$LRbOvR%BlK$
z9S@>>0{SsDY(e6T3T;Sv-R<%Jjo-F?J7~x;kykg7kWaVc;%>gk>XqKze%j07)Z=6#
zfsXFFIK`vTQKBt}Su-XrG$%6z%M<{fCpyBB!FS10IWB2>Nq{1+-Ue>EeG~4_U$Hi=
zL_;M{9^$TJ9fyZ*MF_6X-iSR6uw+lqHg7@f0obp7j8Z6o$)en!eV?5DH&7oDeWxz4
zg%PR?yfIiMY()<ONZi-&{181dSg=fM_AHXFaWpi1jKLN&63X57Hg+DK!!d{K0~0d_
z&_x=GK3tX?R@hWKdmH<s^z5cs>0TPbiC1fl7vf-doQMJA(h%Q66lwr06XB47<!@p}
zPn7`Y3Vgc4PyCa)u=9E!PlyIAyPjm$t_6ty_HMT4;J2LD>(y)|vEhX4^j}}LX0|>7
z0Rh4cXe#XLazw$(T0qrSP*I6N*Vd7;4b>FDJ2-#%?LSD6BnN%)!I9_Nct{z>oQ^On
zoAXh)(_)fV7XY6o113s7GsUevdS#WB!QO}YkP-vCj-;e5J99uO$I=0E3{a<6uAGAU
z9DnOrlIZT;u}M<ge0+KbzlX<o_XQ-E%Fef@s_IpJX>~1yAw0PhVeW-G8$MkGnSk9t
zG~QhToB~fTw7D7bl-t+Q5ok>>i=Vyn<pl;2rXcm^7pH`O6QD||(TQ=V?vmsN)PO)7
zDM_}gqwZLEynIOqDrUaRtINlZu>?JY6XYS=wNJpXAes0<!=R3{>j`yx!i{rnEUa_I
zYy-o3>(CRi{66|G2B(_JluQMdzu_$-kf5^%O#^5Hie*k&!;26p;Y@_ypbKAsVg#hU
zmsuE3C6I<kIA!PI(aCAs6Ije&{nC1zK0GZiG4Y1;SE)CqdoO|m7(OR2$JJ%MWT#gG
zu^MpQ>XW~LO@t!Y<-I{WHdH@#S;R4YhEY=VeX?dF?B4H)+6pL4XAtO+)6hwJoLOo%
z0Y%m4=&F-&0^*ptx5hr;{|X9##YnN&*SEkW*p4^U4VXp#8Pwk5SKd5?(EptU>f(i@
z)Rq=_Tf;Jn6POtMR*xTtClkr4#i@^b9MgV7g8+Lli02?}qa}VDIyOIe3#;>+H~CNh
znD_R@Rc^Hfh>k}TZj|uefUyQFSCUGQ-VMIpa58<PN(nyVh%Yu==ppk4rHy`y9NigM
z{5Yey<M!kH%wO{0J)q5;otp#MT0Ei(^mzag-Q8yi_|qa&UAP1m;QtHWAry;GU^sm7
zSLO|kc$n-&$zS@X)l`DKcjDIkgLV^8ijW*R-8FxF>BT(_mCqg--Z;ftfC*rO*t37U
zbL-YO&;_vdh7`r2J;$=cHB2L3>`b0Jl9ZZjAtE);Y_$k$#P_HWrYIqC3DjteQ!r^q
z#Cu<PRc&p*Q#P#e*N_Uts6R;;T!buwM`dNfH2Go+68+Yhjo{XSm99-f4!)bq-PW~k
zEb2k_Mib0cUtV(q@`^sFzJ3Dkp}<1PWvk3M24Nfz5eg-+-n<2A3mXNF$Lh=T5Uqt*
z2r1_8;ja1q<v`@G%wIP#sK>&yD>=FH*%396m^~h09rY1g?K8QqSVW@r#VeB0Dzo*n
zsTxKcaKx5K-a8Or?o~E&InQLwk6ts$b_$q~kiXme6(wv@5r2_#Dr*$>M6!HvUGhT6
zDxUvRbBQ2*Y#|H{4RPB=zdOhqs`=+QjAX8dqKbu;70Mo1&$YBX4h_`<3FOQfUL!qt
zhC#!r5PfH84DX0U;pHH7Y;H#H0^=Fj5#e^nJHmHxJhvdxXK4_YsCaR4bCZC+wssPX
zf;eZu7v^VzBoZDXxVd&cBA32+N}K!kC0x4B%EF@c-1c;KNJ;mXrlw*Odw-F>KD#z%
zR@NzC1*W;Kud}l~fKdSCMr~A7Br{Xl1<|rVA_gr#SnIsJV4vXfmYTy)4qrsAR=K=2
zt_V(S)m)^LA1^egs}2TjiR`%A^i-Fn&j*|>2gl^_X{UTx2&edM?JqsFJD-{WmLj%o
z8&)n$u~{9S@F1mIe}@0+MNR3%_Xm%ezDK{VbuTW{w9v}SS~!PmU<R>-Gf^^6)z*XS
z0id*JY5^GMb56yr^7619!mBN(f$Pb8Y%9GH;6kMNt@<8I7@d_=P&jDYhu{M<u00lK
zot^KSm^~Q_@(%SN`~ZuSVyzmp@QP`8CIb(<y2MK5t{Anh3&JS~H*Q~>o!GlR074?5
z+z5Oc3wnC(z-t3Ay(o(8yp^I(1&FL(D|S0^pYRv;!_MqGBm&P@zeCWEXVGJIPqv9K
zE+9s%%koj^E-)jsNpz{xZMn6VKaH;l)$n+|BA!CU2bU?72A|nMsc79t&hL%g;BhY2
zg7Q4Iu!M!jz4)`zn<2W#=^ObI)dN6%-&T7rkQzgZ%|OZy2%yL|0Vbi;ng$TecAUG0
zp?Vi8FcTeI`LNu=eO7M+&=5?0;B9EoOD;6Z9tP=XqKu39f9QG-aIXJ<fA}M^QVEq6
zLP$dPDkD*{XOw8kiXxerQC1SNvLZ7nBdM$;J0p^jNGKz_BIAC%>-Rg?eeV1Jcb)58
z=bT^X`_<?3e!pJN=VR&`8Wy8#d?|U0ia1?f8_$g&>{wRtOG63B#>t78V-VoHJ~n61
z`V-IR3VNmB^}(bx(G-chS2Xa(4Zxdd$LzDX0sedgfoLRNl4+aDMi(s(?ZR(ky@ZR5
z@R2P{m41Fq<j-|HoR=&Qv_uh!o7=~hmhj@>+W6F$WO$>}<mnSzZX<24)PW#VAp5tm
zq9oz!hx~@fssMfnkkQJD$U<PMyW5oZoOyiW%vC1;fFUYHtbQcdm%DF_(uS)Y*U~EY
zUPTDU@Od&XM@NypjdAzxNgXG)6d*$<xdM;C2v}`-898?WQMtCZwy+aI_ejF=3MmC&
z6fp4Zv`298Vi?EbjjIYCIUKWNo`v(7;FEFBK_x?!Fu$Ytv<FASI_#8nwcUTZ3cN?Y
zY@=7HSB9+<S1k$w@{JK@0&W3|<(w|&9DFAL7RydK07Tc&piR;)^;&yF@#;4>M9!L;
zUen%EY^Sephw5I|opIBx%1XtXHDtcBKURb2(H($gY<hY+MJP1+YwOLRpx%b3RKD0V
zP%{MaKi2yHZGWzvUIZCpFTv{iqdz`6nwN<dAs9df0db-*>KeM;ND#{X0YaE>@dfNZ
zAXiPk`JI)&YopsshUeX3S^yf007mHFP#1^e|6#i|2#3&G+;IeVW0c%*|AY{vZ^*d-
z|2>1|52rG)?!eW5UM!)9HV?c7T$!lR0%kWFfiwwhA=o29s}OH0MsPfo^O9W~>)*eh
zFB?ndkskFFrf$Io;WpPaNJ|)bogLx%_&Z*eVZwmht#vHwp&-}{06qQvuh1w1(GuP;
zAL$S__8Z}BLvJ^;-JF;Ttm6n^lOE(;_my{@tuw>@tD({8)w&_=CG;ZE|7KJ?AAlVn
zQQcWt!at+oHV7l2z(`eh^yUw~kCXf##bSS|j^erdyw#N}2fB_3K;wz+`rlkB=gSEJ
z?c^i{UJJTE?SB?V)4L}KNtV&5G&3DxppKH1keKW(K{wmGmEV}0k;9t$14sn4f;j~R
zpHa1!6uh4T3=*zO!0XkP#2<!A>nzFC<=6Z?X1lm*OCksX-@@pH+)a$(YA3mJxT#wx
zHPh6oKOON82<V1c1ApM%J8f(>ZEav~NlN*~h}nkJEfHWMLVxA=g_d1^IR&4_fG0g8
z`JZ2{?Va0HTo2qm0<OYF+?RiGio~^l|D{6OYk0!=+np>e2LV)@3hjV>UN8n>nE2@b
zWlbWG<f)r6o?-;Y-~z+#g{R^6ZA73DFC9K5#DGFT_^tjUmQ<Bh;&Yj{_Lwh&TLl@y
zh}{EV32|6n7yhAuhAl<!E4VwshuQgJ6|WhVG#V1P7XY<*FfcD+BM^*9mUFT7%i0E&
z0**!UDiBJK(p@#+oCS~qgb(UIYYFj^-pR?IH(s8RCs)}Nf<yI#<PqvmWY!X9c-AyE
zn-SE3iDt>m8`&j27X_jkhzH`h#siFGPR`;EpXPph-&~q@AKC8q`^>HIq?b8342>Zi
z#O#Zm2!VzV==e>&w_@Pgy!oMV)_!C)laWGedy6hcxEp){f(o~qCpzBq^IPOBD?Ao$
z5sixPjOcei6ovV1oKON8hd#QGca?;^jTAAh>mKuvgN6kMyZ;`#nPVw=-~hgV)Sbp?
zAm07;>(?YDz9`H9cby9XVUJ(UN6!v;MgRsGf{p%0|GeYp0LuSV2V`^ShFFeaLUi=`
zwL2OQUtF6?z=hGh0r)E3U6=y$eI(vF=)?_NqfF9<_z(R7!0Z<YD`yzG#Ea6;;2XzB
z#1M=$!8bKEJh)g9tPJnd3oCTb|DTkq<)_!j^9o+`8hy)B5F-KCgx`fk0tZ1viiPrk
z=!HO@nUoAQE~0$^$3?a#E^69WNZvU0^%Tibyp$HwYXb-=5K34#3<9aMyL*1g3n~~A
zpYQE&-|%0~6WtdjBpxaH3>c(8om~m0aa+Icw%ApcEUi-1YAMPta~Tgh^e=16-IZ3N
zlh@CnLYj!mX*f^2b*s;pVE$42yRI>a=tM2r-(em+j>i1WdP+C>rFmQc))LUI17hl&
z_#-I;@&|V-6Ya)e^8FyMp!7$|0SQ^krKldjqz^ilqY5k_Fci|0GHGwM(ZlKOuri%}
z+KJ|yrm=Z4C3we3Ygov?CQKp>p;_b_B$Qb5h?T}&SQA9YR}lnmVs0x^bc0w`sxJdu
zHhL31AXIn(^RNI@2mvTNw;tS)!ufi3isn5tu1+69=DODpL<@eh^YhISREUSo?h0Hw
zb6^T0Il0~QLFM5Gpe8QLvJQ2tpSQ!>?#HJyd4PdOd;~5rV$gCpy>0gm_Oa<{3%rPk
z*n9uJrXbf=K!kJtCnYKWW?f5dXu!xEr`MnSsf)>q5o8K4cp#iDY(qN|rU`^qCsiZ1
zk5_x%?NwRApo&D^^^BuCR*`awjx={trmg@!&uRyfPS*QhQ+?7s>7!?o1A%5Er#n<0
zHz?Lk_#UCIsJ2WscCDY24_7NX{q%nI63X9BA@8V6`BEqUpBV@nT2|&U?qX_xy+$_{
z3CFKAGKMl;Z-7rE*fC#J4dMF4L5}~y#AGL&$`gx-?xKa#xT5~c@1GPG>mcOhJmhsE
z-<Vtp6<uYd|IMY29)N;k71?MI>GsLv21Hh5URQ|f1N<ARtE-V=x(?U4NI*5g-Qyg^
z#fcxdSG?yFKa+e{CA_I3CMRhuU_SdR8m3Ws9G6b2c=c)pD>GI!!j-V2kXQsWG8TV*
z{(SdNRcoWIwzd?Rs^;Chw4@nxQ_mYe6mZU+-}&_KNEpTB@O*;&ch3(V<QiRr&H@z}
z2%){OH!2Eh>w2rAT?)4`zTl2#X9o?0N<X2&J|={p^=N{U1h$E!G{AoiNxZ%Yw}foz
zcIDoz(Q`#_D~@Djv|A%Kf~0wkWzV%U)1D9R<?zcd{RHyLkS=UiYz=lG!-blJuu%K}
zPe35!Bpzc0j#lo157oqDXn`zQ<2)OQ-hi!#j;`*++5G?0zj6wfy&;i=c3`$Q?;KOW
zJ6~M9fPYE`^AJ8l>(9H%GauOv{*FwN8zRB|h`XCTg+@g<!H+axDY`P{=(BGxVp@2k
zHUs1EEoX;kbt6Ov5`5Cq{4kp;91M4G2<;p{Mm7A|=IU5Tc=(=}RvhU0O)+}dB7rs$
z?DEI{KOt5c5Go*#6(H)lb37&G*BV#cLp1q_)JBbmw1Kc$%g%*#et8xMcn|)%;9wdM
z3>zRe_rCl4;DxnU`{u7?6hGI`;g#mw<coA|gd6yOqSV>Lh2hXff%vw+Qa!~vBIv;n
z<q_Ps$v8If$N<9$?f0ZA<(HB7wzj_2DSrY@G_@NmFf)Tu=8(Huk5$vi2=SjZB*;`6
z_JrYxF<tQ~gmM6sGR`OX;=@Hfi8q(sS<)AP0O+{Y<%xn_EU%E8PNa$4urfDfdj2v%
zn~L<Tv?6iXY3!^9bqm$qP3C%fY5xoyC(RL3c>eHr^NN*2R7ZDQM=k=Pom-ii5$(C5
zaDil@r*~!c%aOS#utN4yP_k^Kq5|_Cim2?frinsHaoSl>8+CPeBcTi|6u2||7Ll_*
zqa;t=kKoCc)wx%L0ADCupv%f!Whe^iO+Hx8K8fj;1T?d_O!^RtYM^Ssqlia!2RVCx
z^5CQ!P~gWgD}z*uPm*;o6=FGtv!u@TyFG+;Bzk}=kgy(z1CrBQd`vi2r{GBc<Sv#p
z*-=~4I;vbkRi^>?!&n4Z0oG2^%=?tJPJnh$AD~&Ws|)3w-jHQ(RhX%w(`3P?6jLe<
zkJ?UQ840)dnWo8`RptqO$JLG<)D%&*V>rV4<ElvmV0*&Ddsf5{yoRTd?>JSQ(WW$W
zVBTgxFt9Z=Jm7Dd#y*71wQI?^i2pyDyohF`NM=e$A#4#}URT%lUAy+vQ<JzLEqfsF
z>3+*iBoiJb91Z`$qI-I*6sIvoHG~6g`=e-_SpO2yrgKO?@P?`r`ZT1|Pq6ykw%3V!
z`(LeV)E*|}1pQoG+)PIo7jdzuN6Ml%QZuTdc_4CI!a2x#(a*E{KS);OejVZyfTbMz
z)C#OWSZ6f<hK{eV{lXCe>3Z8D3(t+VM_5zXX$MzV&!7M`1s_h(|FWFqAKn0q_J<NY
z-9i<q@&l6Nh&+;F4Tru*Q$qtR4`MQ4NrF(KskN1=DA@>jCvbdx+dy70IzZx%x#1Bv
zCZ7K#A4{C{_iy=WhEqkZ>hde|dr7yA-z7W$IeYFL^b&ZQj_x?vLrO_Vuw75s&|nKj
z_oeIfvkD=V{Y!FtAXNGb(n4y2S&xBUE>2OAZ9WYQ0<aDoM)1xik^?q))E%{ROYYpt
z`JG%s=t_e68(NaMxT6P~w3jjK0GR(k<$7(lr+W5lAd%G|QhOeopjtX|`~c|sH~ezY
zOl-Hl|B#0VW{pqf#~$B31UH^x_1(>EQY0mJciAQWa5eeMgP>3k;=kuOKo`8}Q_$uo
zcHNyd173`oC-*R?^?}EvBw<g|fA>~}7&pu9_1s=z<w$ZJK;f+yV)em%8RzZ7b)J*6
zghrk#2oY#FO%!yeKn4ig{NmdAc+H<6xA9Cccp#|?R|<}q_0|FRI%%Wn_4Vs_uVRhg
z`RTjHuN(J{=7i}B2Zs0x99AtjttZm&d8B-N(c9<rsZ)y{RE;s4PCxghy|rO`h*FNy
zNi%^@o*6u!_8n6<R)0Gk(K^z9rz?Z!($M@$fV#|Bl=A?8i_G+*2iinU1(<MZeZwz>
z0T`+S)D4%EEFwps(tSR2bT8VSksua_EH#j-kg=oHXlx!PT<!ZZX?gZ+;Z3N2WM^UV
z^BgY3Sh9xp2MB{EuFqz2AfVB<6Up|0D!cEV<dk;5rTXdH)fbuXk$MG@GoK|5`o#k3
z5RTHeIT@WlbGz#<$W7{PT=WJeX0tfd;Mv&|41v9e65qo`1O{khf<o3dPzmyg2;L7U
z9JsMxg0|-5cHO{DJw^vgtTaDI9$<bi-b1IbY-PAwp;oyMkj~c~C_4O16tDmJG<|4&
zWggBEow+b(2v2D%mT7~6248QS@}${tXodc?TXnJ28BuX1H@DxIbx5hXxwNwLxv6xb
z%JjL1GM*Zw(<50L3obH&fkvJcWws@+YCAuT+-iWtsCU4_jo&U_6M_*7$hnaBfO~%%
zVPc2gG$l4ROIHB8$sIx`Qss`FTzAWdVjAaKaCmn@m1-UhBP%OfHb3_C#N$t5-MMtB
zWXo}bz-mhgDJh`BkoI-$1ZIK3Z?kMO<53OZ0+@~PpQPJt#3D}Vq5l5-)&}eV03LyA
zKvIq2)Np$fwEPX35k0}!vas4=Im1S#uYcEjue{SJyX;g%wI#5{2lhqAy7@O6D3j%7
zWN>l=vv~!hXV4?{*YUZG!&wyK;zfHE3JoKQ_fv|mYJTujcR7`N60;SVZ_QA)gR0^Z
zd@xZ8yD)9XQ6E1p<HJ$0!f@y}sk4JiPj9B@;uEJ4I@6%)I40adXg0$P3vfF!a+^`R
zP1&j2-ajyL8S>v`k(3O9a;&bY$qon>{&opgn3b_~!KXsEirlC7{rxLMtPP~9#b;|n
z)H$y+P0juM2|Fn&Pq4d!2wm8^7||`~K9`}&9Y5o?{~W)W7A1)FOPA2=jBX}!%FfoB
za$|Mib`a0tAJAD_dqgk>Q0CXMu}u^k3~5A|Xz|cO9lkk*1#_3Rl@-44pj=V9?vxZh
z&!ebe!zT&dq`SfZ62t@fywNa>We>S;5A~Do--p^D?BJsYh-Hugg8>e|7#5`G$uWrT
z_@)eg-vi4q4DFFmDl-i!ayiKu<z~LA;$Md%iwQPlKs~VByxa+35;=J0K1pzdLS=`G
zRN>EL(OWaIKy|gL1?FbW1Ey4mWMrPB5`3|8H^Nx4#UZHAFmXh8vM4Xl?rv6%rPY3X
zA}|P{80XqGFx!DDLnu_X6b1B+e6|wX2If#JgAtg7p=+cg0jRKbDnt<}lD3*gG<X@L
z649?9;ZDvMz+rF^AIjvwW|2xdQ&U}28glZ%FB@oREN8R?Dd#QeIXDK|+tE=9$QFXH
z`YD8@1I)YZv-%3}%u(2p5YcBmy$k6pl;0olz|#>)j<GHKgDsk2@#Y7H?Uo+rEQq;t
z=kio<$zGNM$xpfh(SYP&;QRWMnY&^;7NVUyO|ZzID*flAsgt}iJ;c4Oty{PAAvo@B
zYpY}3?Ol{+{>zV#?LJue(=&5Y?*hgMoMew5YYWDVj*Uh2<rt2`G>J>!s*F4(F%hX-
zU>C7{OBlW1232czOjmWapVx6ucXw>wSW_YQvZbwzy?5`yn?&R#OPuzlJ$h4zI}!-5
zb%hBygP+r%BODCT(Yl#HGENDX{&RcP)v3E6sz^vs5I7#?HQVe1PkHRDCb!dWOepiM
zZf1%Kr%(6&3{G87o{@yVCnQGTJnRG+bdM$)8kAB7xQimli_pKh<8g)D>`;m#zA<1$
zNJ50pJU+h6XML?f^G!UMxD`k`V9*0Uz|O{2Q*t_?eq@eNuz`y9S({I(v)P!@eq(la
zdNbXmk>L6fdU}>%rre)DJ{`9;){-@FqGkLT?0ys|g?f(of}Qq*67PXKIeM2JKgI~+
zt2xi|gb(zut<5E|vFVA4JiAz1mS=0TW;o9TuBCQ~roXj|Cn2H`cSdU8EBIQWLtFau
z2LXHV6eH#x7PlmMS19ljK=}69<5*%1x0;;1;$(eFRrQA6esY{wxC2o6M%;*|LLUHS
zq4^JHFm&&R#w!F%{8tMgQjyJiBOwWI!6S>xWG>wJS0Givt6@j`2fz*fEn5ykKgg%J
z;rylJ*RDP5O)N`KCoidi$ZO-K9h!eGAn`l2R;Ah>q&7y2oF5>MA#qe<Btt4z+gsqY
z={jGabL+f6hJPs921sbbG?5S8{raZbuAaAVDIcDO!m^837AgW=0o{F3(Uxy4>CP0g
z{3hJ{;Bs-%Kwm;c0{9SnBLQ6zmmyR%Z)$7Ho<B!D%OrZbNQgDetU6C;bq}7N`^m|-
zD+<9WT3T`?X{*0JipdJcz}crnZxM9T=PP3|H4Bu|BR-Oov!HkFrm}8MC&b6mndmcE
zrrL=x;|HEs6-HU-FQ2uuYdSB6JqeLR$Xj|r16StJ4<H-VjY66~=IHT4?Brqp<mpqr
zK9v`t356YI2OF-`*&Se}TX>lNtm)&dnL)kwE@xr%aJblb;6a|~#dmgR@;)ELy=qH3
zci|Q3rr*>+dYZhvwD;*4CY*K`FAy^3S&%fgFE3U=yAMs)cez1(E)0ecIU-UJmleQc
z42ghPndt;$<YZ*959l=3L+af{jb04Unh0HpHWWPQz2b&2EciB+>7g>mQ7K@F(<ei&
z)X{7=^;!5Ok#>QeM53|u#0lnv6T9iDf#pZ?H(M1oFvp`n51$znDjk4q;R68_guN7o
zTFmT*6GQMF@7$T?QBD$-lf%qEf^!8*zutH0A$sb3)cy_UQ6PfsW^BYlkdJC-fdK*J
z0#T1dmhRN=pp*X9ns*z#J)qk!!`~6vmsk?K(EwfV=Q9#xLVW2)hK3lzF{6T;F+5ue
zBxbKDOF+C9<kQ^KpI*P_#@`74ZkGa;G<-H?cvo=j63iLKL|lh^6<#7$8_qdDj%)}1
zmqo_;W|GJGGQWwrPzK++rS$qZ7M0~)ZqZg0lU3(9&O-5k3uMH}9hX{T3@i#Es@!dD
zDblvK(c?Q`->7_qZ?$e$0oc24ggP(Hnr+>VS)L?{f7z)EvLiyKg*UXbh{+EDoT?A-
zX_Rb~V7&x{5wC}F9|zVnNZunOcgHqCcHbtl3neee>@}Jam;VIaH*DlQz>qIreuB`R
zlC}F>x4=D1REOMXvBDM!@&B#bFW4n>emJ?fd@8ny%GC(ObLlaIa@HERBDQcjWNnRu
z@deU^7F00KMmq_%_x$<nUHt;bMdKPoHx88|%^X&;UC6f;Vx`%%>Cd<Hhym|EmvRg>
z?)8<=yxvr#@G-cWIaHHR9$ic-X#m6V;&ykEshPUsABwLC{fbbSU~bS%Bco8ZE-&I$
z3`>yBi&u%w5r-grYexo}XSxM{1^`Ay;6DjhsNVhn8vktuP9dhu64lM-xTk;_-gtB9
z4tYeztA-paUNx<7{ndg7eg7Atvu^oRzL;Ww#*3;|Y&pjSibH5e9{352VrXwhh!qnp
zPCWe0@KoRfGOwJ&%n0DB9=DaPErOZ#1!GXP2%(%nl>L`H5(yCu$p71e*QI7Cm3t?|
zI^*MwRz)RJ?}Sc0C#+@AxHpa99>+Tlg)5Y0sDdr=j<2e!0>_O*0-y&1!LK-a(kT`q
z)gM}S=p~`Cq1aH3aCcH%Vq$?=i69=s_kPkJ80f5CYgR`*jF0a@ogp$v9n+SwZxP@j
zdTgQ9fPx8rencYXdvDjGvl2#kdPr|}+qB=~3qm^(q6%a45rZ1{caYxVq%byN*SahT
zcuniV_piIDxp843`x|d`Ma5~8Yvqq0Z`r6?zmbnxO|xOI)tNR5ao4BC#b7JZ$8*Wn
z_U=($h|sdd0vpcfe#&B%mux&ACVzU<^^`=-B+no4!D`aKfe+5ZLk?zjiODJ*PM~*?
zU*;YerbLG{;_6us8?*K$i$1+ZiD@5-+nPNtP|TsH$Fjb0&+&WNBQK!tN~YziiLk$L
zp~SXJaNz_7%i6Q}9HSrdM?@ryxtQWDr!rufGIl;5wj6y{)Oj1P{qgj66p=0zz`~@o
zdkkG-CHEACV^PiB0BN_v0A?601OWkuK|DaB^*rDfmJ%d}QVVwcz?wi%EIU)n2^vgG
zYmYRz1-uqeCZ@eKLTDPW>WmX<*H)mgvq;m0oEmAZI9IGLT!6VBF>*K~NguJ{;}*sE
z3xFSvV0g<2B`YD^1E+~_(;VCB^2{8AMQ+U?K#&N%CqdEOF;W%97{ITqSinDUU{7;d
zDvgykBRp>>Pv+?gfCR;(9?Ad6G}LagQy}WV54Zh=#|EN}S0nt$1}`fK1uf0{)oyrS
zA)}^u`t%yYIwRN3pI_d6d;3`qH`Gu!#78qXH;dJb*fKZwUjU;YTyP^WF#Jza1)fH&
z0;4ot2)ScoWAhwTwS{ihoDqu$G!fP-^QFN6JJ7v*L8=zpkp0y<hC+?~78MHcyi63q
zHw3jzNyvT_8o;NPCaCdT`rI?PrIEV_&=PK(<*E3{oqlgHSvEA#Bpr8|+tYw&nl&m|
z)wNB{yqbD-b_;4Y>mUIESd*cj0}&6t5nRLgMo`8kP5M~wDSfx8!zF}@%i2Hw<&zUa
z5~(k-;03O?DTtR`c&i;Q#<E;zhl4gvL@g@3dzn*K7HY=J!<2Wr{ycs>)9*a<;bf}G
zM@K38ZQBsCtnl6#0{P~{Z!*|_<Dx;Nht_Nu<U;7_!Y@K(`+Pk?O*cx0F4T<&FJHTs
z{Baw`xd3((Dyo|}4&d0OzoEn>uzx>dK2Hft;{b#wLv0sCR#!k7K;nj^MoM4T0e2{R
zB9A@6MJ917RwYgK$Px5r>kCAK4Tc{C-~lpOFmc%#tKcxPx_Ghg<3}P?0MdTRO)D$&
zt(YiqOkrAXY0(k{;(X88$Ova6c$%r6BK1mXnO!3OdVo~XO?D1|DgYLAQ(kgBe7FXh
zDC^_m<H{zlh<XK}y*%6DqEyDLi_}3>=U7FzILi^AWn6qb1OiB+K)-bKBP^t)&z>2d
z49Zk1L88Mew{|FgTU#UgAG+8^H^ABS%4K2)-1qO^{hXQ6O{)0N-X58lxXD1`bki+a
zecFwn$I)Rj-skBt7^O2fQLuae=CFdKWCx6y=tzXC9|XgbYkwEbI2ySbjDNsVdVuhw
zj0OWx*H4T(`1G-J{T_xeW+#R!<TrfzCNx?dz|NiPvk=MAcG(Si)F9gc^*e#$UBPYi
z+soV8x#(%dzAIZ7EO3%8x1y;Iel3zE4h+zpkXb&fBlJWNO}c0Z?fCrs3XhPGW{W#w
z1W=zr^wlKndI-NnB}5B&YqX@=DJUsBtc&q2rRjd`?#_}`bbnz%kxoq8&R1Y|lGx>z
zf9~;O`n~}@6BAd$g%T(2c=|8bk*LIl_ZVcma!5&#cy|C_T-kAaN1$R=(v%kG&HiYF
zC9z%*(-b}3^i5VD32G_VGo2kT^7Hc7O!cL@Y6g9W$M&lTHKjOPWYcbbDkrDZmu!=h
zT)Y$?9&Z$K^mEsZt1sxp)obB4w2?`f)AieN#L9ppUqdD9cTWC*>Soz`!G&O&7f{Hf
z-RR;)bc1G%Ki}XBg1(?Y1UwA8x&jN1qqcl!B)s>Q$NsC7!G;d|15|LGjrqVJ{<HT{
z`xEy_0>xW&RAVeAA!OuRnU?fwXet3p+`I`);AZBV)W2<Aq`{Gq3%?~y@@Kc8*s+)f
ziJhQV>wnQDwj6RE)r4tQPpWkhN7|1sV`CT3oeQ`+ipU+SRfX3WbNLRpy6s096hvdz
zYO`UlpUdi-nl8U`eKW{4`H{?Zdc%jZDoZ-v{hJ5e$XVH@`>fB9y>zYlGaaDT-u7q&
zHcbMiegX;`q|}7z628Vy<(@kYGZw96anK`^lhPNr$s_deX12t8GJpE!21zDRf9~!*
zf`S?o*1rH7;U*(0Z6*&8s;n)$<Rp%rJP>EP(Ue#Az^bTV%nP{lphMRx+RkYBfZTw3
znFa7soBd5nUl19Fw4wKMz98q_rEyc4$$N9&!=7e+qsKJYxve+G)M3KcFW3S`1-C1m
z+J{!Va91p&SS?6OPELikZ|d-$Kuz{B+1Xr~aY4yBq(pF-Qj28yf_Q!aAMGys0hHU;
zako)hXEq*WNIgzsr9uy4S=kYF^-9DO5ycvGjlv*{zWp;+6RqU<&H?VEr=L4@svqA4
zb{IbQ%M$7#+F1rUhM2alj2^u{<EKouU+Bj1SUwwO%K0ZVS&G$f!^a2YNnv4C7VEUf
z?<qM>YsN|L#hMVv066#cbGlcm+{)=Z;GilHK!Fd@)?AHV^&ejuDpKJFbSCz+hYhee
z$N_yIx-;5%Pc%+;bI(%G=g$$UiF^V2Ik4d4d_#OBzza0)bWr9VrWLZ&C?P{KiSxB<
z4p0nS9a#89%>{P_-!)eU|DDSFFsK2QgOe1Yny4=6KjgRU-nMO8N%k5pWv#6k|6vjV
zN6qFEa*R^z>Xf7@F|I(+6z*4)Lzlw3h(w6-p1g}zF^{hjA5F4m^oLn9<c>OTSfO}(
zfkjR|r1fT^be%<z+xXtg*Mh2)j%p!`gCrCfUTr<{c|GXY=kC(f3<`w0IrRG;yV@AX
zQFaRMR3eslC|I2N#udlUX!tA4%iBZ(D1FX+;3X^MAsADM|LX~%Lr!1o5$pOVI4-!l
z`Owi3_f;Al1_|#5*ycn<@ATuOF#(F$(c#a>rWKm@OX=G9-rsOE10PC=O|LYtntLCh
zxx{Uaem$by4C-$wDIibL3dmgrFi0949zNG8X&dgEYSPuvaLld{`9wVeRBL3^36`Q;
z<=MB?j|2(d(?~1XZgz5LEalASLr5h1P>`ZMj4y{eWowWdOQ}P3d-@aSu?04X80c-_
z1wLzYneMd=6^Bi+SbQ_S)rOgiQ~hQ0Q96fg930T+lJb;1!rhW<dzj~DfjVg5P<^P@
z%DwA1=qDHq3`2At>f_XBrVGN;2%8qX5VnR=d+JL&o0YZHI&4l^HH6|;#+xA&+XO*b
z`_}UamqG#v&%~Ir`YUqTnHQ|z>$WrM4gcqF!QD=g^0AvrM}f@Z`o>m8#fOw<M4j&!
zR0jEx1sHp;EwMWdf_(*?tIFM8V%Lo)UV_TUl0zlzrHK0c9im?i>a^cQIM*OVM+gcA
z{qUkK0r9mmfNA>rU?f5N73j6^v4uPr-$mATVQ#@gM*^aCH#+(&3LnS$+45|ysDVO5
zQLOZ<l}#|&RU%C@@4E$sIhY_c;(#B6fMYim$6}3vKr}tiVoJ$d`33m~zT}sOj{Eb8
zHvvLwkoAIIj@6a<D-JB^4cdpXpVq#8t3CM&030Gq50DU)Vr!IM^$j*1c*0)1e5qq+
z+0LY?cEcjg*W|Qc9COuo8M>gajoS~K7#k^j%DZ`JXGt1Rc62LXu~OxR%k)G<p5wv;
zV$7fYI?rCjKJ2}YrGmu8#Z~q3Jy(MriWest55)CoLO%k{-YMZ$SguxP>lPmg?cJ-q
z^=P{lZ-;hzJJjoGOGnPVVLlUZMwA&Jh`?|zOTgixH`T>kUJ-w7gaE)FkK+>v%`sjE
ze2<+ZL{8R4rip>O0Y-K}LL$ROMD}hQMRRg$!zAL}Js>jl*n#wc1@Ee6p?_jz9N{3R
z`<`7Azx3`a_2!NI{DZT6zZcr|-2Pm8?%|?w>9ysk*riKqLJl_Z`?e?$bTXb$K)jfZ
zQ2e~R%#@skjcuQR063eW`6aRVFzqa8PYQeNaR)lJ;L^lz28JBPkog|RtH$OLh`4Ft
zcewO`Zn;jB#zW+t<ml(n(a`h&&dLyJw@%y!BS(cw+DD{$ht>k#YH4glfuNSQHk8QL
z)u0Zr0|MAn;=(%)(^Nx)4U}*%D9^AujbqQjCQM!h$qr;Cl3V@XQhn^}Bl7GM&Tk4k
z{wGx;a%edED3L*)y*c)+<)|FiqHK#7h_Q+oGdOuN^d~TvI~FV|Y4jkfUStj<$pG9-
zX1|%l9WOW{pFGiZcC@kC%er~Q=5>n1`$}pLrQviSw$R-|Zy{(T`po<EP646U!?VSL
z&GGu?#*)?y@B!a@D{-Oi{7+pYqlZ3z$%C8To)?H|#G;dWfW<>Fc$Cgi5GBYNx9`N$
zn424%-%Kg~>p-JHf}QPUs0NNyc0<C^Veis;1R6;{aw;q~Pz&-hV4}Ci+Ii*5EIL-(
ztxw*7D3Vlvz7SrXuB^P<d-U1#I(vIVeq7|c^C5{Lu}eTua2ba&ELdL!d%3}vAj?$|
z9nkPogB!uy3!)I~4z8?wi0>aVKwfUD8fAXuDp-8DD~Z^Zq(kk>!|ZNM-nxG#SU2RB
zOJA_s8gPIN%QRtYrG~rb{-Nv37=IJiAcX5dq(|E@48MgZShdb0a7tZUC>Zr>@Vw8N
zFE$0U9Y-7|gCa&-<3HTU+kAdyn34piuM=xLjVPuasvV($3riUN+G54vk(C_nJ<)c5
zLz?cb8?%Pc3W{k|JKSx~O>GDZ_;C|~2<{$?X-E4UHEf?D9cF+C79DUq>@q<1mL9jL
z(dpA$&=Jz5#In-!PJUYR*`2j_NjXK70rA7%>T|TKznGE&s@=er6OM3Tb4_=$Ak`w4
zD~3uKp(O1-D9pZL1mN?mxEMj#IOH=7G~|ZxUgoOnmYeXNRNi(<*kr2Az<)?8uLB``
z-#hkz;>yU#0OZM_wFRd%r2I*+fQyTR4C%w)6(BB>0(CiF;Hlbw&GC0m!)POjkrv7Z
zxJ$1N+jf$8&K>x1?K}Vm5b}z!qFJmjo7AK{)D=MA2Q?8R_SW=fb+ZS~TL4ffCAlEh
z1Ahu~9)MH@RHGab4tGGU8QJoX7OslIMTfoX|HkAvJI^Dw7U3f1=2JLvrDi?P0&a)$
z1?UjMfe3_2T^(@$Hi~v=z4Z0{H@rY_d*-B&B=J>&@50A~;axa3bj=nr8-s^CP!4|`
zVoOXAxRoI7F1&4nh+i20qum{GLVf&r%b)L=U`;l5PTU%r>=N-;?k6Q7XA-0v^cl&5
z9!zaC)YM9i)KPw1x2P(l9pwmPkz+C>j#rHo-yOJre{j+ig4-|t*A5;uu%)LIk3DwG
z?|3Yo8OyHCQORp|9)Ke-wcx?VBY;}x0}f^q^725N7$}~iwBpGV(PCtnoW=QpRLDzs
znH(I1HKri)!^JeW<khiXor`J!3Kke2WoN!sqHN`D?u>R!OKxl9|Ew{&AxAFDu=5^S
z`5L8zdBe#QXguChyibIj#ems!9IhUS2C(%4?JUvUs_&Jk=u?5pzg+vIEn!Vnun%6k
zGzDD4@psV{4uxBHf94tIi;Z7pZoY4c-4+3FDoGvq4W2)j5f`5aod?;nvNsnLuRvz-
zzZ#|dBYdbxAfz!aA9R};L?w~1MMeMT&#hKPaH`$7hops3WUV9$8o+Ue&kw}GDPg$Y
zVZV_F&Z(^I2M1GeaX;Mcf;`eJ;&a^J{2o0}-}oZ;^JLVUzkud;5-=PhA{>4K4ARJ3
zm_Tqdo-c48*6}pUVQsC<v<ZfWh8VFgm-PdJ!2~9zsL0vHm6}S#FI;64AEu}9`|{0W
zbPTi8?(20n;hA}WP4QrHCLay)#&wG_#)#A5U>Fp2zxdG5+)3Qj`$Ax|N`V`gv&*y1
z;-bmy-+kPtN^GPqnBf7(4}~(ad-Jys<^mt%I@SyMQb{5U=#)AR&oC}(gyuGWc%X{o
zK|Wr8N3iPAS9FU}0Iwk+D7M3i{_-Uo$#9HYhvVEOK4c&Oh|k3o&b0|y8fM|X<>lA^
zmH3UfS)9|ELR*Y>mhlm2{@bfu^2=upZe7WF2F&yo;;At(;;-wZq~yE%JvEvUZ>!Zo
z=6>*3zcXVILqPlpvQmmp2zAmR3v7Rk25PLZiDNti^$i%ocy}@009@yw-ax0IlL;E^
ziHVnxSYqNNSr+vmPzb!y^5RtZ1w5_TOOAG&v$j5(W7yl%17-OhUS7?qV7l{arzy;T
zFD><a{Fr(R@q<aZ%kHXfZZfuCcCoTzmE<l(usl-hQ#(ub$Pe*R(k3u5F{P%`lqGRd
zOC?T?jk(T!Jw`%3Y8dnjxCk=)l3DnLgeH+lf!Ig|4xRnk5lR*1IKV!X^Mb-FEWBF>
zXzvvc4v-swYLMuRm$z##Ife_Es3<mXUq<u@+~X*&sBLW(;o$+}c^0k`jfXH6!bt!Y
z>47(+B%R4}{7Xb-;;g|R-T+Hz6*O`edmrTRi7{wsaP8mxE-4^^)%ud8N{>T`s+?#X
z6`8y|CJEG{E!;fhGrP%c*~Xwpq88?9;7Ld*k^nFq6Z74s_RD3NY_^q1b&mFal<16%
zk0a_+Yexh2Zp7j=jm+U{1Ge}))*HrjgaE%lc)os)y`9}*SzjoV4P|9Y7U$pgYGTb)
ze&@!mT<ZWeGYr)GCtuq46dp@XP)S4MEs&xQU0ui!5&PrcF-@VY8RLIfILvtv_XAP>
z4txa_m)IepMj)EuUpO#f?j?dbA=`X#;vHt<EY)T(AW1--kv*E`RX7C5t?M<15eU=#
zd&NX%S}!Z|qsq3CKt|)oC5do<Rc+}2)nw8-0OU)(TNq!0WX;uhTD1!5_W)(U1ibY8
zB}_t|Ii^MJsaZ5-;e>^nyr;KUPD*NYa*`cW8WQr7k~Wj@TQ9Mch~;5ezKx(5K;j_J
zTwEl_MGeVoyB9M)|5Zt_+P3m3CoB99ROwf;9p5#amS4T(CD<PP4+v@7Sc69f6$6I5
z&a?q>Lu3zB6v&J@TT6TRolH|?!+BzD>Hao3iT)ry#@163+n8wW`>ENa5zEni_$W1V
zhcP32*e*A^(L+*>LGQHPhBqgrq`py=xz0K|*L@Vy5Oeb+Rqir~%Hi*Stfm;PkRt7v
z{5ZvF=gwFNsBudz5J}Cf;SgXwKlM1590{g4<fjJ*+i)<BIJJLb%}tTUnTp2g*T=;_
zSTS^+k~;=A8C}qrt)NW6bO?8IRKrzg=l8t=a()IFlrZ-qdl&7s85^j7vAaPlaq*)1
z4k1k{HR+eA8TLm0FqRC4&{|Bk-*FK<bNB@IqVc>KPUpQ$8hDHGbK$}WzPFo#qBQ=$
zXKJ>Nt-RSuM>$WArig?q>1;3jw=ECvx<z<PdJ2zR;%Qv!>*<NC-1kCzc5tJX6}#aP
z<mVN1`N5w6*2RTpeW1*l%6%G09xfRS|6J65O1frdjl;9gPdR97-?n;Xu}=}*eL&mN
z(|HnG;7Y*cjUT|bAjMIzgrJ+<pG<g51Ac79fu}2L0R<rjOQ<7hX<A%3IgB2q?oa?*
z0^A{i@3myjlM^!w?`IafhYP&cSHFXWkGGrbA#ZO-hODlc$LDME@^155zgMTL`@MpM
z5?R4ppu`I=3DRlcS<18vI$%2lG~$H=IlWUEChIq^UvF?9#uR;?P2qbL%4{1Zr>FV*
z2HRP`&g-;a(0XI#6#C(i7a(gKw;5Skc$+{aF|lN!0D52mZVreMbNBpWlqrFA873=8
zt3mcT4Pp~%(@p#SIOQgd46;5uj*CFoNsLk;j~Rjy)6V0u17YiU3mrZ_IrCr#L`u!~
zvz|TMaqRLz{3+W?3G|5Nw53zXyj{SaiQ^e_4CMHdrd-k9t`PoQxw4ya*fup&<{v_8
zv{n4jp*47hybe{u_^d=#QCa!9ukQ~aRX#^8gaOz{5Y@uE0tiNMF6TEpbl_o1*%6{m
zPktS7p*EMe-9&pTIp^k;*Dr^RrQGCCk%&>ouXGm4WPANO^y?e^emFcS349`YqKx6k
zbY${N;<USW&#9~*urVH5^sTl?N>23e_^%cqs3|Mbxf{7JSVAF2(`{lH->pt{Q7j%5
zD>xtPowoBFrDPQdoM`9;g~%cnY`;N@=W6XX+~43`YlkjFBZ&w2hDzEgVRd8NL;27Y
zyt+YiKLMc7iAP|5v5jCprl&x@w_liYuiX1a3UYoZ@JBAoz~Z|4B~(Nri+i`i!Gjif
zIUpa`$v?E4uo(ap5w-}*<ZG<I6;mW0nLt})quGs0D!maSg!lj_9fyTVFe5T^-=5p?
z4v>gqXNkO;(rZB_Sv8T9f(SBjMDN}S3kz%SSA>=S$HWBu2JN4_>Hi_9y;XsbGZ2x_
zzDit5TUMSs$>aAq;v|n-o-*0vnVx}x^ZcmRf*Z6iTW?Li@{^INTAX-thTPg(ay1r-
z)zd!G`C81)r+DO8Ir_hwVAw#mJTyuOpe4Yi(G9rqpzsAFja>Kks&fw{#B2l9$?xVP
zHX$jAjC7rR9}{iyCfTRwKRp+Qltx!J4wp;<d4cTpOJ}7bHg$-r+*RA)mx?zL?m1|0
zkfDOp00K|7VK=^O#NY_2DB$pcfgd9yfukMh4P|5s)p|8o-PmaR&;17x5f&)4XW|~l
zQ-G9?mX`f^Ap``hX*bt`bD1hGtzmPliWfgx8nnTZoV5PS=ZTjrth9`=uF3k}u6h7N
zG&lbQQ8UhWSoaa6jjGO@HTWlyk+2lNw(?NF%(B`)Ug~EWxl#M)a?etvb9MQ_`G^xf
zwc8IbA<mEX5)+izUI>O6H}9?GWie{)kPe*_-!DFH>*_kv_>P~KS0RiO29)}yfOPdC
zi*@WQ>kh?Fd^30nq(QOyMztk}6gX&Dfr$&MyE`^9F|Ujy+YHE3<KbN{>Xwao7q8dd
zO)^D1+R<%6O(Wc&c1YMWlj^113PT{Ny3(I3pdV{8pLM7IL2iT;S>mGaTN?B+-+4Iq
z*Vl;qmucrti|3cmVDQVyp`jtup6tXZWj%OBOKY<-y}$`a|7G{MFmjzQXxm%3ALIPS
zQq(!|F2ji;$`>z7CSP13=WwUcPV#q|&~&)B0}(|8o{e7y12OoZ!!EJq<p=M6hlkCb
z`Ff1068Z)4v_l=TaRZsVf&cgzKr*^%2VKD!O!JkMDRskOH^0YgXKl~lKAj-V$Y{JA
z&qy8k`hAv8x=IUpYHW}3&#Z$|+bzemKbA56_WJgXBra|##u>+INk=Ju4B<@rdU^=B
zE-EULw2KB&Nv8BqarHL!pW+H3ryuQ~<&K8~8@T&iUHpQP<alTA;O6!e*9z&bt%0wN
zl%rfpN=CHT&q_^v!7UC2Gd>ZB15Ek?4VOuT9aSs~u~i`JA?C@<q->Z2pQ|-xeb8L<
z;ZG8wUyT0Ya8#@7M>!8K!^gsn4xju#o?uLPxC^`jghuz!s2&r8a+@?=Yg~F?zA(U(
z>$A4<*RHa?LOx|TdXBIO^D_ivT>@0-FLnE*qE5$uVcFMX^CiZVtkkTY;=;t|V?MKV
zc|qV*1>Gs(`jNq5@;sA>sgJXt;HQ{eK^@wxW#c=xz;qtwwnHI!o&@1@gL(o{NJwo2
z7t(urM=0Ljo`C_lkdJ`Q>bS6x1a%bhVx!4T5~(@p>2>w>vAze3&E$SsgqjPdWXABN
zTZwX5lKi=lh>sRYo(Kh%wW`^r6-Yry(Ek~PN9Z(L9lQuX3c8ekjF0PwzK-qRXNRLP
z;cE_ZD}Z-m+$6*yKAYL)#E(58hl+Gm2vtjdTg;!X?~Zx>;}fyz*gnvtQ>Mo;nh<bu
zVmilY$TU%WW9|bHfNQzADqB_t2f6g~*}DB&VW@9zR&93`i~;2J7G}Zw_ksQ5pr*ZA
zazMwQX5VTv_u4$C+ms|T^Y731J{Vuneit}5_jAya^yJyI0SH|W@0Q2!Aj<d}DrG@I
z+Te_AW%@T*CXtH99ivgXG@6Lz7<NXvB;kM3p%W-GHn!`9I`;6VX-JjT(sP!}1sWnU
zpF(4h-{@`&CGJ7A<>KT9mbwn>*++Y+fNCNUUqC=<Z3bo+!@5DOW-D`{6HijXcnZfN
zVtXC7W=t7ST5Y2ZgshN#3l#|uME&<!#zQWc>L9lOuNv|xG1HragO{oTS38^zfJ1Qd
zSW~td=p20LE)9NX1qBWmV=h`-*CQNOPR;|?+JtHb^4K$?aV+VToVPiptw*N%F8>l1
zGLvYb)MD@8tl6?c#{dm6WRhK7fw|xGmOl*)a2S2~{-CT(mWPL&1bh~-3M5`eBbOzM
zxD8%k6<CdnzW!%0!l$$Ueu=&5k;|eaT{6PSm!EzeiY>~+glPrTb3RWUem^vaMgl2?
zhzs|kM8`G?Wh8;Eec4&5>luRO@8(S^)`9OJua=$FKsa2LvlYUveBS4!Hxk;ahi4%q
zKf<jCgn0Zhilwz!Vrf1;3dNy-Qw+KYaMBQCqJpxk97*x8#~J6(Bz(+iX>P`l(ZEL8
zl^I|&LUv_P#G+!Q%fteiqkx4Fq_E)V<m{}t^p(-5=P*aaFG%TABZ=HeOoI5q?T%t3
z&O6j;dQF&wG@VxEm`~?<Eec*6Jg^MK9>z-~k7tK)NLJmval^W+Xt#3CRwT?LFR*2U
zTlBpYr%dDTb)BcmX>rRWB<3AH?09Hdz~j}=FGsA7QY#oX&J#j%gj`7Bc7dX>?L{}F
zWHBbEacN;%IC>W5{pT-U06Z33dZY&Ho714=bv?c)Q!_K#Ajv8o<o@B=#g&Iw7kc8s
z!BikzMMd)<N)4Mj`E81X@r~hA!J7xn2juq!F_t84M1^Bjf*m8FGMQ1{1rxo~AU2_e
z&!U>UxB8P~6M#3FjwDjPpbzIc(h;dCQhe6j{{gG(IwNWr1lB$A`GF{!D5(5_24*bX
zNCh)8Hs+9Z`ijq2@)`?f;A3I>A4i?Cn!<9F^4RyieFOXqpcsB1yqU;{#HJ~l>P>Xl
zp*p*JVE~5K9$}dh&HRptV|s!^;nIW2_YWQWwX!rOZLf4)__)KTkp03zr33!qY}HSA
zTgj_HVaQKQQb6?^%H-2g`~M~}-ff8A%2))X8-LS(0&x4xOVyT{fk^EexN`wtoJR$j
z(Ej~E9O+ERyXND~h<>=miOmYP1;27hlOO0GB^Jqbe`4vtWhSSsrE*U(?vXbK$;!fh
zzzy~V05wnup^(Ka`V!^}qQwEqX=P=yy)~c$PwMF%z`cz<lxpuPfSgKTB8GyCK;pB@
zt7k4SLlbpi6YJojlGUJ5Vzp={m23?-Vw@Hl4?DoE89r~{c=r)Z{OZXH|G|iz24h&)
z#(9;U<Vly28G&H8E^|iR<pX<bYjsU?%F2WUZNt>aeFuAcGbXypNM>e(1MI(VhF`h4
z6AXvpZvXgn2=y^nJy6IRdWol5h{9-3bBdyck)F>BcY1a=-Xp0ygbqkc>u6{|TL!hF
z@<}{5Bun6}xff<|tz(MPcsK{_Lv4SJrKotA-bhK&%N;vo+gcgoH2Ff(QVW)^Odx)@
z_AHt2DPhGcU!XPtM!z+-sin1beXdD_pqO|${z9tgCKofh#$C3{pX*NlJ6O8$pTYEr
z%<1bhgVh*wMb*C7WmqSg(2<6=Z{M=IO1613TWbn`z1z*(wObXwiDYBaybX=}d**=n
zE%0dH<~X>FLyx&+6VwM%#AH7kz4!g}3@N$|RSFW0YR+E5wkAqH4wB%9*x+jwc?{Gs
zQe}GWeuEzpuX%KWY;;sq&bIAKbhOPgD}75U?|VvVx-iZI-O+p3e}~KoIr~_Ep{@Cc
zEAA;}q^Pzkg7**U+%zQQCLwL54@je7yj-A$BHxiZYmbO`?>=8&D_b@WZ?E2}fQDMW
zD9E$s+!R_~vNA+JP!H9zCZs5qBoZofY5XBnKdPfcMMAsIF1KOa0#BbkyLXfqVbr!I
zw{e7itt9zw;7zE!Wxc5<$9hogaNJzsF*=pSz*>>Hn$V<Ovn%s61wP@GVh8lfEd8wd
zrY6Z`XHcWj-|y95dKJ1GJk6UG<~Y}Xt7mZkL7Xa?64IY!zk1oSaB!>vuq9dIWdac+
zF=L9Z3tZrU+W3A-EqdL=w)bGE7$+43JZ2_$8Lr&Ijsqa2zuYsnZaDa_tPm1Jz>N~V
zH(RVQ<lt6<E4Yh4cb97HL63Kk9}xQvL}rp@N$8+S`@Lwt;!#{A{vfe-U|=Cuv`8cY
zbDJtR%<j-ZqO3XR@r7*fLtI?A9d}%<jOC!z^qMko5$@Vrg6Fr0W#w0bOW5t#rJorT
z*QnXI+)^DJrEI#lH|6>c6%{_8NsMPmD2XD}XKDCVdOa=sQ#`fkx~mfBLEHWQ^9Q~N
zxY%Do4f^#ha*`x^FS@xgUs5iuPY|0s>VAtUQtWd4Zp7rQxF;J!HkbKCaJ%?e>Z@fz
z#EU*CHPN*>CyppYu|uEF8*7mkN`PV1N>^z|9zFj%a^nWS@s;C<C9B`dM_M|D640QO
z4O)9+H6Fe{A0rIw%26Y>5(6ySn1|Er>i6+Iee#6MHRoUxSjr>mjpX^Jkj$~PE1fLn
zI6w9-$t)tQ&E+%+8JL99&12m3R#|CjQgvwHp+h`;m3QJjzUvC3tw%h*8GmW~5p4VU
z`%rw*CjY*vX_M@X+VyW7%x$YNdn*ruTQE%8lye$jXbbsxPhfn>EmiJx?}P*I!%e-{
zURkUs<mO)M|3K+F&ET_|w-C7Vjpr|I{oOG__y4BZGBT!*k?-3PYScY(cHoseupdg+
z87`R_49Grzrb<Q7yME>j<ZZwU``X%OQr1_fXx>E{5WYQ96=<BP@7cqVul`6}Ca#Ao
zOt)_b@ayzC_NbI)4Lk~R8c-OD<SER#^-P-+bY*=+*Gg}@4(5^PTJQK0t^KQ+U!OSp
z;I1}5CFNW5oZ@13^Rf;KJ{xtubf>a6RlXeqr{A4V(5FKd{*y<K!dl0W7YjF%wJ;DY
z5($`kzUMLGT(BWhY2kDSzlsD#07iJ7C2wxRaf0C-Cj8C7jSyAS(<3V*gY;{_!`2)<
z*|l%YHjsc~fVs%e*WzV>qy_&uUI54!b8crC6WSj<wxy-6KyN^d)Yk417Dla4)Fv>K
z?dj@dg9uDHf)Q@?z0b<5I_cvJm%Q=GGv^8n5rAoEh>kjVH-X#(vV^G%6HeDTe85F>
z$+w3=6#zFQ4Fa<@ePZdCY|dCj72qy&pz>M^e8t73CmKhLZ*Db^y`L7vul&oT@i~4a
zSwyV`sz9IGwe@*}8$J7PT)epY)_%_p<DdfhLDvh@DF><~!QtJlrh^U#5kdMBDHMCU
z8lqqNHPKZ!H`84l%p!-3>h5+>x{@z%xibfcF@1VvQ|4NL-r&pY&6iA_3K!BNqJ=&P
zr5rvIG<C_|zL{=@vHq7rRD(puv6E9}LHEyYz(5>*_!(qM&MEnnAtIOpYzj3t!?&gK
z0<oYsrK~8$iz2&E7qL6eY$b(N%+a>o2&o`(#WtbY-W}b*7Hen0t@r0u>h|_U<e!wD
z8%+$ERb$YkTnOT&evRB1d{EdR%gRjiGOwV`n*^wME+F$~8R}6Wo3IsItoMPIXBH`n
zDEk}N!KX3?ec#c0`!)awK@t{@2be=`X{*)tOMPG<37~DOrDczS2>eg5a^dIID=*}j
zB^=m98>oNcM7E%4*zMM#%S0ycHRExQ{LY^V@$r$K;V<+_gR^c(1Ic|a@S9MnLRb`x
zTz51B*oa#DKI$m4+wjv;^DRbrSb>+vYT`IFPWHWuJ>4O4=e+lD&=h|P-Sj8<p(02k
z57Zf`<M$Uhcy6$-CcCsPV6ck33bI6w?b~Cp1$1_*;vN<H01I{5sUP5nrY6@;7Svga
zf;5pheyVN9oJn=ng&?J5R`C*W;P`Hd%}iV{!{1%>Y3WVJ%D~Fi&ECs2&rj8U%f3<l
z#B1dWYp5ju0DDG4F|Hdp{hm)W=sCkf7xc)a7FgQHEauy`6-`QQtugQ#ESNqFIm)Y2
zT1oj<L-hS?G!6Fj+=<U|Y<X~G$Dg03%%?PE_TXUQ-rq7BV=}nLMml@BZDpdMr8PM)
zUH#1bsEwTb+&QG*Cuwipv<d!TNZQRxY-4&00aE)Zkx0l2%<xa@HGy1|V=f*z0pyz~
z`cb9_Is{2j-7vJF*rcPJlaFw>gEPVHiBTUZEieK?MeC+;8l41y>oHFO&cU2Yv4hQy
zhJ+5TG1T1nmXt_p2*E`Njsg~T=xvbTIyF$9c=efHeL&67%+f#eEzwAvV`%m_*d+Wc
z6+j*3rg~97>}8{+n{znikWW`^GNplCC0vaSVLjQ|Z5|H+qe8>2RF-n*4n5HJeftn8
z@EE3C(7tG=_@^!;5k0^k3M`P%jGqfOHrOv6AK2}tR%sPJXW<^UcWBm)_~>^kp396i
zu?YN=^}@FH%b;?oiN!|Fl<(rIf-!eD-!)kK>88Z)Mzj@IE_W3emV&$@kxWf*I0(0k
zqzGomhCT8T{d>}yKY#IB#D|7N8*lGsS@oKC%g}jPZrP;Y1O|Tdk-C0eRLZLH?5Bow
zLZx%yE_HP?Z9?MExRLdxjs7G@$D_x{9H---efPGOfH5=KowB2&SOj=$#h)#j-qM=o
z9$n*QV@kr^6(_m<F7;LQvrE~>IUL8qikS{*&yrU~uQ}D+O+d_0;L5oaZ%Iq!{09$r
zy?hIT3YGK#4@O6=Ixe9pPe5@NgUEyUTeooTK@J@j9*%(EsN^)g(C`u%SXZG+wqAv%
z0?$Kl1#mc=x&WNOb^^>ozF^*Ip_JBNv$H)2O$rR$*vmjqe;LwgjIx*((HgDZY_ZkW
z1Y7{gS1rA^uFkH!V{+0S&Jg-HnOceUQ$Wt)v{*(f3Q}va<lrWPE%TOVs${OL^ud>5
z*GCS--DC=;_|bJXqNbkO>({PVOQRBO66OljTL8%4#6t}_yI1Qx5{yW9J}&}~hwUAZ
zW^izDZWEzjt-_L*jSYZO{7STjj2r!c=T;BT+D$u@Yz0XtIOQpQmKnWf$g;m(Ng$=^
zy1_t?QD5Tjdq|tW7y{fa^7;icjEMjfR7z;o!{Z2(C|+6YkoHn1`J%2PCx~<>At5(M
z1*a6Eh)}ix6mqO35e#1)EFt%(ctC_;_(h`5tQ$16*<IpD29c3rj@>(+kw$jfXP5A)
z=Z-wes7bH7n4)EuTx~tt65u>GxZ;i)_Q9W1>r$2U6hfz{EC1}W|3sU)zHckti68wr
z?>*H8m}Z(RRL*j1*7HlSLqUK|=<mRve{G*XW&QW>Ny^u`AvMP!Z7a)(j~A~<r@4;9
z<c}YeB-%iHS*b4o?;hH{TbYYK`vU`O$;T7NX*X?ZH_^o|`n{Zo<N887*8DVG6ex7D
zyZLi{cw~CmWBL|+4#0*<P}QWwkvn$Apm8r%_XzU05_W>)T1RB%u7e(i!m&G<n5;F-
zwQn9|Zmv0gjD}fQIyq53DN|Dfn{P^F<dwCRB{1cawL2|%rJ*E40us~@cdn~r+sDC<
zHVr%vUtJFh-V#IN$XE&Ub2^OgeDxoh?LTW%2G{=5_;QJt;oG>s#VT!K(a<+N(clVV
zz(AhlxYy>+Qd30fc`Z3fj`?Naj)(SxK74DS3L5R;+E#Z^wkgJQiYy>IbHeN3P;u-I
zKfb~^%~A1Ig@725;yW=LQKPk5xI`jp>{*#1?|((JQ)ENpLB@OEeYq@$)jmGow)`ip
zeAic}nK^%!M_2xQ*`$&_r@bj4VCGk4jXI}R-hK*YWJq9YCqU~(@p=#h=T9Bnq7b9;
z2^m$7%M+rm>W&^x)5Xh_7dT)T+Q<I{-vHw9$g03<x2SI<`Ex-pfe3i3rxBn<Gw<=6
z@+zL$#hv~*8pb2-ZdA6Y&g)`Jz_EwIs{CC4kMC`^QqIqG^x3`n^5ch1(Yao%o4;{t
z|JB*AE*L<=2zEn12YD=?OC32$k4~1}e$}-Jfq}HNKtccDjkoZel7s_LOl)qpjuYvM
z`KQY6PWp-wr(3n=or9i+$6xXOsxY#z-^;_V?VGpHYx(<CHhl{ye?j1<>Iwivgh-Z&
z_95PrM~~=YioiDm63#uCS5zd9V92*`t)L@7wCeGK!fyO(n4(Elka=)(BNX~DF9U+H
zG3+AC-7fR}G>jOpT=TVx{u(c+K0nE=8LC=dOjYkqsta(LNKLi8HZrtr%NEP9=gn8I
z{Cu=+UsjJFl`q%}Fd5feU11X!+okn0>Lo}fH+J|f8G;drg#1hc14fO(yd(j*Ij&ws
zhIlz7F1)-q8qU8cE3=7tAYuFT>Alq5ngF=WHuAaD3Sk}7KK(#{N(0Fl0R2W=Qwm%T
z6nbGCdT~_=w-#Ur&v2*x3%%z$_eE4Is4^Rci1B;ucbVbxn#0=q%<(f}Vy}c>;^vlJ
z*!hocgm(Sm*9?m{b_L}>aV4P>4k6sgHb<labO`w8t)uAieifmpJdp6xb*M?iNcxuK
zE;kZ!EW}^r;#O&me11we=j_8^h&jB24FM)k^-7YS6c+Zrau+X?!7;<fH`<wF0G}$b
z5O{%YS*DSCfQTJ1nPKw39)mH@$e<v>CtrH0Pi?>!L<v!mCnBQrSnsL5;d`Oc%E_Me
zR!{YX#p-DvM{DQ7njn3eU;7?Tq*rjLse@sa+q?m^U&av&bMx}^4GhIs`j_4B!UP21
z2_CFNDG@=#<F7%FAQu446()+JBF4ooh|$4?V@$qvi@GbK47W`oMl7yNX6F`#vW~5H
zWbZqYdyG5M8qpbch@LX*mA(EGl4g;#G$ndV%kM3x?uBE{b|Za+XAx=<KpOZ{sM*}Q
zRp;2TcA+(J=6#hu>&RUJ@C^bH&IDxVmD=~H<jS#?5T{7k=zMFJx0%|HPpv)KwiZnz
zGqpP}|9+B8Bt7A0yq9g)uKdq_1=-o(CML*z5kXlJgNK^P;ndLqR3x%@uU_6q!@sJX
zpnj2o=U2$X1AuFl+dc^6x@DAS3@@##i%ZU9>1Q@O$VpX0RS%C-=C2G<SvykcwAx5n
zZxK^?<TKiKT_A>wN!ztV@qa_@PJM4(2GP$2VO_g=1k&Tm$^i1hp@{DeW>+R211MS?
z!NRd`$-=vjB47cg#*sN_Q1OFH8pm4R7Hw(Ur$hJd@Nk4*x)WdFeka$&%)$au%P!_Q
zCnSI7?rga;@@A6dS4^EHhkk#*<?PwRiG9<pbMN+Z&{J`cFa_eL&KwXwa>)-G0hFbC
zD<f=S+(&PEdSG&bt!n7wr$qiNyOB<RRx!vPB_zj4Tyk>DK=zPxUs7Up$q#BdNieje
zs@7H=yZBXn2dx7ZrS|skg(yjbgpOG-=JH~0@E(RSN@HWoi&VCpO6GBiHLDDE?^Ur$
z0yc%RHTQw~cUQ~@7!VJ(+B!JAtFI@KAhUpX5EZRmf)fxf=C|TVgH{wjBy1VrI|0-J
z<9c&;7ImyoJbyWK1?N)=OEofJFqfBqiSZ=k+{LqPzooV6_rcZ$C^tL8d|m_7FV<GP
z--a=9ndIqfny`&E4wT=yudr4avrqYEW9$9F>MMizF07vMF!%-O8oOP%;CA8Rz1pOt
z<*Y`~(}zF+yndEisDmZg`BH5a5TTx)GB_~Bho>x0)Xkw-X54D7GjeW`rGV43Khu<s
zq;x!VTWDGTcd>JiqwZK**9+|?%iKSa7lODnY)dDE)QlFg$qw=Is`|WR(Q1ZW2p)sB
zXs^Sj4o12|FEDqR8UQ0?rc;Br5x_D0Re(@Hxnq3}Gq-=V2e3EHe>n>_gh3A}1(=I3
zrd&5}9nqodeDgSDGzRB0k@CXJYkhjJpvd_2G`<(JbvaO;4d1ogZUyNTC{WJp7CG&K
z&cbK?WrgJ9$CnqfbLbLw{AH6wqYm>0+-DaMAo-EtT$0*xPlPHz{PtUo_Y(H*|J4F;
z>lmu^(BC01IPO9-Wou)z&03Y6&R%X$N_Nu_fB@7qCP+REDxE$XW+nGo&pGj8ie#r>
z3MG|3Xly9VAb);>2Oo4%S5eqW#fS%1X!Co25#D)>y0&aYg_o!(Jw5#s#B!CGL#6$#
z8n$g>6A6Y>J+=wGRai^)LVxUyX?*a2n<GMcN$^SwU8PDIz;WF^>&}=53Dm)=sQF4?
zqGDxX>8bRoROLQX)B{s}a$JvEPe8T8q}NZDxcF*|wFj&G5w-rDJ{0B4FN-)8baxVC
zNp-hHKnf?IO~T{lwE=Ca85zD={OmHRBe69fKHU0p()HK(9gQ7ZHFnsHe`U$`)D(Un
zBg#vylHb0Igh>SQOCq@kx45nVvag5(6lGLUu;F4F79QMPEpW!YypVdUq|U)T7RA;p
z7dGwm52BL1&gJ+7*D8)|=<#}c$2pftq@#t9esC_e@<JU2DO>#|<6Vz~de2B~+>7Q|
z-3=sn{d9aV4R{^(JeJ_bw{^gXU+S*FK|pmAm0spRY*O{pG2@k_o<Vc8KDl0P_C6Fx
zX21J}|0hUgaznS%*5X0a{e!afu--I@Ci<&WVvQVhqi1fa_~{(->pykk<<AQHkV?gs
zrB+=t+UjcYi~XIx;xoi3zB^jOZDhO5)uT<q99zBT*>HGjC+WMsHqamJbKESqrs^z7
z`5Nmr6&2Ohyx8<F?)e4@ba=J>xTx`*#Mh`}Psu4M(Rp<J*O|JRUo^^^zY=Y?9Hy_a
z@VOKdt!iWUzU*xc`MczeJVqb*$Wb2f7ukh$m5hA+FJ_M1b);`F@@p$Be@lmkLqDfD
zD+}&#3od$y;IOPh+<^c%94~lRpyNRMCr(#5c|K!VhU65`tV$Z#%zHIMBwskxbKK8#
zo;|n!X<?z+i>u#Z@rLFTybe1f<M8Mx{;;uy#s5Rwo5y3>u5H7Yd5n@sWGEq1GDVqE
zNfMD{PLe4iRHhJ0LdY#sXdp>4Pn9_&M23n?WtNax#JinZ&wAGT{=Vn^zJI>!w|;B2
z?ryH@I?wYsj(y+uZ3yeM#*AE$t9s}E&Bb8$M9M3(pBnI00Kk>3YYT165J4D*ne5J^
zcj2xMFHHeJCatZJS_BaZQ()HZz>S~JjOoCY!4=tM@6iF_;hE@J6v=oU;xUN5id4Qc
zOvPjW?ZeG2<U2OGxhbe`d->%C0E`&AUp=0kCD-mD9DU)^m0e4{tBAkkD!Uh}akF>z
zY7W~N;c2vQ-xJ9zQlVO&fRQE+2#%M-GXxkG;V3+RHG=__O8cd_P=~l`5{Z(!vNNBW
z?xSPpEN^$Iro6_!edny6{&DoC;NUR2nrT3G)@OqU)xc9Ov>#ktD4Rl*cHd$QM@C*q
za4>`%IBZXC3F4>qYqVNkT1rrGXCVPW>mMF21RehWVF39AsxF6~H?iYgvRAWp?vqII
zjUWtQo`O*!5j>$-g!D<Ynj!7a&g_zrnFBitPBR%9KVb>^{58xkr6$&%q?@L7K5Bc(
zN&-#jKL@9ej3Tu+9)|Ja9`9ca`8yxd^!ltRk{~I($Hn&SN;I)6N(qim2KLu}d%p$!
zdkV{&H@v&360cv)dHe|E^U=$fUzIZ{plaT#eO+<<>%P^T%u+vO&j$poIRu3To?Y7p
z53b9Ji=7vL2bODd<7~&44Rj2*kErY&2)RdGX!3Et4;y1Nt$?q<RL0Zu_0VN!8S0X)
zU!{YW_V}zF7#O-MMc-3Lw>r&!ibML0&a0yG_H?H|`ycG5x;E*jLSkat$W9Z#uj=_}
zDGMNK2JlQKVS&4OdU8(x^FL942RL#%2v|Qmd#~==4v>~aMUkC>dePCz>7+m;h7HvI
zo3Ql~0+Sik8QJ$q<#56UB!sODRjJ1vY)Zk^Mnz8F__yof#0B~eI0(=y9i<<d8<NDn
zeP{FSPUD%qrPii4r$IwaZ7CUxu^i?SpG2tscZ-GTUH@gVAmh@H(GZ*~@OhB9NJx-m
zW&0;2$56!#ZLJNfN>qF(9A|9I@cCuM@?;hCjL?j^+;7xvOTQ%%(-EdAuyc#yuV&Ve
zZ6iO=!PPDHbfSj$vN71a{`yM4D8Y)oZd_B-H=6owdGd<VV61hY^4iDKX~t$|-V<H&
z61q3N;n(BjlL%{AFR(4)oCh`(V1wtS2SE@_JC%BP$EE}=@Rar$)~>EDPr{w{??_bh
z=i<M3$-ZrEdd}lqNlj}o4njCF2oxAB*PIM`0?6QOX)1@B-2qcfWF9{B)kq-(m(O{K
zEKm%KfN7Dk^7F651bo9AI~lxHi+kLBVc=-7`(|1D=FzGH@)b839&UPC@oWc6Q1~h7
z)?4P9H{f9#Tn&``mgSrdjZ&yul_PVAfC4uLrtWd~#J3(=no0VWaBWuX%WL-4S;13=
zQWro_|A#jAZXwNw^7C-F`n6TbYkBPSN#V)(0~f5Qd#}?RkFNBbyb<Ttva$r>{Z#td
z6R`=eV3rG11d)#SG*aM7fn9WddU^-#V|0ONGHgGC&cY#Nn7Rem2>=3iE^q)azJ@a&
zV1I0F|6$Cfm#W6`Rr&x`W+g>c&V`dFNtq?o?0b3&6NtWhaa>CDY+%JR80iFWCc)vr
zWl3`7S6KfG_esUup)eVj3OeXQ0rwS&m}56?HE)~y2jw^sj}bS%mr7$-Uct6k2K;`d
zN=}!H3E1VODL#Kilky>*O|uIW{$#&?W!cB>jvD@}Rlw+Uz}X!gEdmdV>({jiGKi$S
zmudi&DSXz`)4^sHx_K9R06Ya3QDR)_6KuxH<RK$-`bI7-QxPSCU6+zfQy@8ctKcbu
z__8&`_9RB`a1KnU&A)Osh+J9KVjnO6)$BWsG1a~l?`qtt<jk_FlC{R=3YG;Nx<nIf
zrv+LE=4LvFJ)XKJ%wI8^wtZ^bQj^{K_PD;qc|q!02J_m0Blgr}vZ^uL#l}^6O-q!$
zn+o@JZnTK-o%zGiyWBr^Z4dd6E>G{_H<B)=c}uug2hW->OjN(%8|#zsJxcqiBR;zN
z=*yQ`p8Ljo6qy5aKrhfoIf<b%<P7iL-8Rcyn4gDjNwUUFR~TRtFi+BeB4!(OwYEOd
zPlr=8wk}!wFE9>8EHh)x*DM>h>tQEp_q;KmJVqqrNG(kqu(8Xrvoz-PS{73<p}6g|
zLBYtwX~UaC7(&7YlA04!H1%&ODC_$xj|0JF_&jA5@u|GF#xKgWQ9<#XC1sXU7KOow
zi+zn*;k-ClhXx0?AmkkrunSl6I5?t>3{>$uC22mF?_|L5{kb|tgEHeH4Y*DlGD`}=
zb7QTGsJZzDiomPEYlAE>C*&APO5PK<EqF2BMdw&@q5a+V0}>ubnQTDu#tY6l&=u|8
zZ3}4&Xx;OPtO$tM;-7u%lGXL=46<P<G!$-#7cOBx8=$<#k=Ge}v~;$mvZ>~rP?Xbe
zX_I2|goJX7gNG@{JE@L_`Gtj*SX<>MaxeB#fKi*s-RH!+f7|rTx6#ydK{SPcyj*>3
zNe_kYNWX=FeX`JVctAZm$))JQdLD108vORF=OgMp*9afu=J5=(4ajFMTWVyi0!l3F
zVDt~2yl>g+mesklWuNO_q|w1KQR{*4zTSXCf#dt6evHgbE%#Ps`fS*|*#}`;v}e8P
zuv?y8C4YUj1H2qKr8?yBe}i=kU;?zbXwQMnz&EXI(FJiUc`O-*xhMUj4-oeM^LrH)
zO$`k@y1HJR6nfW!=(Cd%>a68LUz8LT{{YK{UmKP-fF#Zr#=uXMd{z}mgonO}NN`V2
zcQ;B}2LLhA^$2(du@-<GVk{vw_-Xl1<I9gW>U8^1<msoQqlZ8RkKsUwLm|FNA~&qY
zb1CL0ZYuw&d!L>?uw9v{Y&Cz>)LCBPkM!(pgGE}rslo!+o8_0Irrq6H*Y-S6psRm5
zdmVZX%mlR)733u(nvuPcm?+Ll|5I^y-iU|C!!jP6**nPf)!b<V!vFmILYA&y@#Ray
zPDO?Lvx>9sfg)0yRIh@vzXsQxE$r-w1~i8c8KjDsE<SxagaD0FsCpF}K+4Ik<px3w
z)&{VahMk&PT4tqQ93&V(p^2|kyYsko1+<LP#m6A(L!b{vgq@v+I57mh4h1fc*A(RJ
z=eY-<r*uRR2+n0RqQodXH5Fwr+%mKi_7k(qDtS5iin_?D9?y*|7Dd6e{%b4RLZSg%
zTGChQ_k4RB#8HE(xk<?@q1^1cq&0h44?)-ca#tDcF0x-#4Gzp><Ufme;LdB3cU3Y!
zhZ)Y?aOT*${bqmdK_SW;q?T4bS++$lsKn#qe=14^G}C9<Mde?cKkH$;JEPjzZIP0j
zFYgOLC9Q=0Df;O|ST&!p@cZuW5>ShuKDC0iL0%p!38$GF_0hwJG4c0W=)OkC+rWG%
zxm%c=Z0YI>6UP)BZPgYLp{U4hiFvVj^y^FA`;p4yw+6vmu?hIJNLMbnP|qgmhVl!v
zY`c7DjpRXDMeW<sVF<mozW&{i5STY(heD|22PerU1&ToCs!-08I%LlBQ*Pn+somX=
zU$2`L9dXs4yPP7l80~YT)5vYH%9;24dEn+B6f~dQz`O^3Ox~)&V(9tTJsNR8P9uoa
zd#xr?rcP3747_%Gd+@H=rY*q65sXlm<N5%^uIAgypP3-lM~eu}Q!n*~w-}7Tj!eR^
zoG>q)^?-J(5`$!`IcKCu2nyYSYwJmM^>=@k7oaJ{+z?d~1nVniWPztlH&9VEx6hS?
zK6x_r@u<_P!pb-1sirmi)L6a%{gAa^%6(&}h4vImtMTWIt&K!cV#jH=5D^nQo!$=h
z^xEnl*8r3d52B*xKnufQ8{#+-gww;vWxTWWx+icDprPp1h&X#7NC*&q-UYH&H(1U}
zlJ4Fk{rxjfc>dh(u)xrA25xbXq9HGjsqbrN=R21I!9=$m7_M_axOZ<FYv6>oHt@+t
zE`-nqGr*|hz}1G)$D*c#gP*RprDxJ)Z*6I1cT>$hqetmne0t|`4Valw`cWl@Z+}}`
zixUI}#F#n`{f>Z26M1rs%*?IExEgrDVoXbm>H3pK@;i@#Q2hAuBbM`+P4}lyY#bc!
zh((c&-$1BNp=C{Pzk21$-eDx5)~$GKIKK<2@PebaI5%&W`Wh{D5;G26U5hT`YZGV6
zH_6CU6gWsoT6kpYGbk@}U=Peq5}U}|s45Z?%wY0{mWD~?a`q@$aq-lzT3^%OrOc;B
zHmL?9MDBJ#2gU@{-?GEPntDn+4Mp!`aXy`XRa7+I{hFPG_O9&DyfKmvK;<nl<H0Eu
z5pj?iOiI=xKyY!DIrF-p&j`A6hny6pal{dl3knkc@$V<filJaaAPZ_JgUs^&Am=dk
z`}b`2+%D#Sc8|>E9_UShS@UhLk6*tYcjT{=<iC1CT^)W%3(VeA{xCd-{m1Cs@0J$b
zAwPB(Oxg}r+JXoKxB$25d!1lXSLyWw#fOegTf9^My-!CtN;0gRu@6K<mou1_RV)~-
zIctL2bSBdb8Ns~0s=Yg7qDA_ggy@;JOLCU9nz>U=^qBrbhv!a0oE--U+?cACmu+ov
zma(!^gjuJ?%8PrEhix>{gB-$=kAV6@7j6N8LVoz%++O7z3KbOe5HFoL(FlJP+|~i-
z;^1HgLvHObU@3%f0Upa1Y^pvseuhR|Kq_{KAJM_GrfwV?O`Wx|s(%>vLsIUhe7|$?
z0K<^ud;7=>nmhLOL;MM@SIo6fAu|>*vc;4WK5|vgek(End7%4~pzW(u#*u!VTRbdp
zbub2#KS+td7ary0m~o@x`Y@by$h5nmD$i)K&<C?Mta#c-QH^zUOgy)$0w0y}yxY+A
zOo)~-Zy}G1UFpVAS~209s$?W+VINaaQ&<{~mY4fh%x^z%6*=u#0q{2`Q!z73926Yz
zDDg(ujh+RE0X%5nkS4hmN~&#;wvixu_rU&#IV87&+r#{)`3Ed4L{G=Wo}a%gbA`S?
zG0`DAd;gOHDkcwl7d3Bok}qp2DgNGtkdN+rL8bUKP#sP)3YiRydMxWfV_14}kSzr~
zn^vhd8)^P90i@m(L_g7C%|{%lzhf+J<ZY!tZJiNEj!4KCho3@1ISfrU2C48@j&w`@
zy=y@9NT3JZ<x#$L4zBW2LQe;Wh6<qDg~t>b_!$yzR$7tugRnB2SO)=J8O>RenutSB
zM<+rzZ5$B(=qMRUHhw$YU;*Dhcrae-O#iX5T5_b{PFh4UcrKb}&Fyym_2gIkdmr-L
zIG<Sg;A5TWEPKJO`mMa~dxz$?4Oc@o1UeWj4j^a=zjbwB5ma}04U`}t#Vpfjs*h~9
z(MO~k>##~p4_hxgOW#ldu4nGFcFF9e?@l|UhYl>tsi%a8XNBkDJ4zo|&vj^reRtH(
z4MwG8Dr;4G!vKpwGwZirBz)IB=7g6tAw~ws>#l&T$+KvJ5sCB5A8t=DPTo_1sV?Nl
zudl%cPcO9<nRjV=9_Zg8SpQweEqU>?xRXon*xrO_cKhDasm*aEe&lX*rymM=P*hul
zdU#T(#8+_giZX7!z<wWDK!-{nCaD@ninf0KZ2ds=JlSY7ERWGcpmD!!rm3gL8d&ww
zE+4IP14&6ip$k)V=nYbIuOOI-ga|2ck}yC1I`1qUIjX>LnTo42B)vCQGeaH8nwCe`
zziSk_@N3r!Re0UzJFKneLh;}yS8)Yb1M+?4_Un7Z&Z99Dz>bJMe`;w3n*$I;E5YZU
zR>CQ{9pg~xXCi5xgpwOsNa_eM7K5ad5DaSj{63Xwm~uN)MS**zRT&c><(YRrf6h$J
znk>>anrSkk&O$20EVb;BC1$31DrzQA&oECnw-)V7K64@xTF=k2jR^k$JT|}(A9*1J
z+%Cev%o6?y=j8X?I@}LbYgW>2zb}n1j5M$-?(r8quioJJwf>`*=9R69wIY26<m7!7
z6c=gE?+{U?mETGfo56RQhAGepxBQqBJtbB#=b<<{@YeqGi`G$`5?A&osPL*I<Os7g
z9IbHs075dl*U|wRQTOreS};mCD7Bsn8BV86yD~gBrjhy{5@+tPXn3nZfCcGAs2ZX{
zP{MV(%l{f2|6F$K;^&92I$L58TGf6AB~$J~4X4wxY*(4KBkjb`pA3wYzMy_@%_(_E
zN_4L=+iYTUojG_0!oJ8oL-$0hq`$twb`%3rJGzCmwA~&Spg52=a&YWs3&seePFz=_
zm-mEK!JeaASBhhLvG;<%@vSh5jfGrChe3feAY2ik*S#*mZZPJMYatf$#%pOB>}VR&
z{YhK#95g+ci{k7Ju2=rks%+>zg1N`)kFwR{>FxXF*CpS{BQ0I<8BLUTp(d>^C9TC4
z3dzjuMUN-;94r|UJ{^Q2X6N%B7$7qOTo`1TV$ufY<1Ae22u_foU@A&<MD!s?7jD!<
zvaEo>E!IdHUjnz8`t|D(4_^x^96W**H4sfBkFNUdTP`K9FXrrZ-BiBEMEfi^T)Vb7
zCbC$;+;BC>b5=+08WZ>ZV=jjkxV>gJD9mp)O*l|gc;-PIy(`IQiuZv+u?bHZg$jK6
z6++zV*zoYgjJ%2Q_4UQG3#fo&#sX|3GG9@D>udwoBgL;@L(xJ(vSrfNVeL3lX(ib+
z6rF2M><;^#DSJE^LiYE>bK8!|im2Kw?FKNep|QiqJ~|wuco1MP+=G@5RDC#UcUAV?
zyT_vY<1jx<<C)n@?Dg}xw5sp+0zkca>lS%Am|l<$$G#T2^9rr}ZfT4L)f^|0DxDDz
z(bZNYz*uDCjOnm1Fv?g(@Y2Y+ftSP0&EXX;QW^Tou)>NYnKdmoX#75cEyb$;qN*rc
z<Q`yqL*DO2{K}L~%9wnX82AN!96gP;VO=>bNyZo+8CTh{<KbiS)ycJ+UkpgJ-it3s
zPH6a5s>*q?-FMWfBKFK3cfsgz6_%4StQojAI_GicafpjCam65Ps;RAQieaZ90VD$%
zhSUu0K_UZ~6cP5@a~QcvmXaKGNf6I9)NYr3^MGo`nTktCjhahAZYHZ-CGbPsLE1(U
zq)~ybJJFP|^K1UaXux=oM1MAZCM+CTr@o67ARy#=dwN#bNPiDXn49ACSm^Q8&d~P(
z8wS)HAj2>uNz?n()&?3pxvx4uy;srHJ*`B%Gb|-`a8i(y=j7%>NC7Y{9=Cvphna-$
z1b){3gbC#1!3qLIg)E~C977nvLbV2v9!nea84Mp#24dz3xCOfx{(Yp8Tm<Dqg%@Eb
zm}Ez(^P@N+!DQ-OGwU|PN~@|cA4cWXy|c3_M`^MQcv;1>sq)|w*>W8J;f<ua;wsl&
zbZ~U8B>ha)`LGU8D+Z7?SruR8CB?=bb1W1ytRd9}e{(K&eahHqX~qc9Q;aPb{j2Sz
z+7~+3S&WG<m6fBTrL6J%`Fjj+x4Cqy(<qlhLGjy6Q-wvkRzS1l&UL%>SnZSdDbeWw
z4#YPm6=7X#drxof-x7zYIY>JM&MIRVaXif~X!rUqZwv_^`c|ZU;yeK34X+VXQ8p3u
z*r@YU3ZX59{{_;Su~_#er6Q(f6gIMOLc~A^n+_UpP})&P;G#k-o0S#ru8AxkR{A>n
z6c94usmiVB0eZeHmiwDGCaJCKBAnBE1%bwR^_7#Q4#rEC_^PV%^1tngjETQZJA9kA
z?DaLkw6LHt$}oGWNCj8pbLaR*C&N+ARvS!d-n#&QAK1f;3=OGeijd9JN!Fln!@hi4
zxE&6lF1C;L)5Ui0-uoe#4?Q#re&`kA8*c?N^ualq7I`#*->Dh;96Z689i*2N+OtPU
z>?Y9inc)-TrX|}j&QZI$DWG|J=*JJj5du>cbd@|jj&E2$R$%a2`ts!^H06M(U%l!q
zY9%z-#|n|82E9T~E!Xg7gjw<(0Ik~tCT56Dgnui}WfB~BP`tvFdFYVies=&E2}^w3
zB!o<Rd#}Pj3bYl%6g?@4nFK~6bYf^`f1;gw`!*mo)f#>yn2s5K5@88~$_AbK=+l!J
zNMHq^>=3v<nr9>iq3|y^Y&f7a30xe16onMG7*|?!H^D~VQ{G_k8cVfZ=*~7I8jp<u
zH|3{57wQJEslhSUAt&URt62jwVzS^D<LgVRci%hkIzW}r@Q;=bs)$F?gH=@*rYtw@
z46NF_m}~yEK2~banexl-?iIx@c(zfrcfEps0!qc8pe!F{E(O}B&wuj;+!q1q8UF9E
z;<U!*kKGP3i}}Ql<8t!ygTupRh@Zq02|rDIkA|-kSaz`T-ayLgu=igZ9a05s5EjNM
z=p{zJLJ_(#&Ijp(vDxZrNd3Yicj?m0n|m>~CwMN6KSS@_k@Sv&g(?PVV2R}L-m%3w
zQxp3V|Hl2}i4Us}=`??3(Ac%Obns+&ce*t`M}$l}?(K0!U&1FaEp2b%K1|${m3!st
z&GqXf1yCBJii3+N$IRCgL)GCWIA%dAAq@B^$;qVVAk1|xHfd4aAt~9?_qnS8=K}h`
zk3oldSvovfX!(|pYiYgA$k-+3{pqFYf%LUG+%(`)fLCDp@6HBIp4~|y^i?j%@~Zmw
zX1?=cL)jSCR2as&OTQr&J6BU>m;W0HB?)Gk+Un|ZPJO<E^6*!Q6j#NygpDu({g<TY
z<+<A1Gm=0%odgYV<t4T(bo>~XwH$A+uRkutig>`S(M1O?f2Ta6140lC(y^H#XpRe~
zN`3w2*tO036&0^T4uk`lEqD$l9)g^#ba50!NkrRWd~M&#d?A`v35lt-!hpc4Q`!|x
zK1{(a5!K1zfed6LY!~Z;z@>%(A3zw8`CKB3$wVSAh<?en)&}x~@L&^h+>xZcdu}1J
znvd+wuP$UYqkgKeAzLvq``}l6mJAWgEc4?XGpWjdj-|K5=57?hT2dWnTX8$rVx_Fr
zNtz&n;sq0N;177BB0~S8Lys^K_A|WtGB-EU&?z}Lm-Oke_kVE#B1JtUmp=CxpoK#6
zM@$S9rbdOAwjT>a;Z00w!Gm>t^$*KOfI$WZ>THLw?m$9BC^Xis3^GRs((z~GQ$NE@
z)L2R;p@-mpa;==i%@+`!fG1;RD$u;y<#q1n)qdr^k_A}Pn%A?vZV<uBtG2*#sr%^q
zZVf=3+S*C|2VWe!&(|>64eNirV!JL#?x7|jK`^Jfp{eBI!*XYj;TL8TXUcwWS@Kgz
z+FSQq#CN0WarUs<uX(nu!01rdySo<`D_3gb?ql=;{D*|TXycYGgD?Ss@5rfBQu6YI
zgpkC4E7jjAP0vo;H@eQUft2XVNtR9l*7E||N(5NwvR*;rkk3l352f0VBB+;NNpyC!
zG=2pi7LO|S|1n7a+S@_TO>MV>MiyDXN!WM&qY0I(nAjWu(D&~VV=?5hWz_W*a7Mqb
zTSU|RCR(Fn{nGsL`bDL%#2<6<?XKvm4+%~`w?R->)Jjfrx_lWF=mGDQfq~@s_^;oa
z$oO#UrC~FM$_}dy#x<DhB*#Cc0<qRE9|;jZd&`uOjA;*y^TYf1IJLsi>BHqJ>fSv#
zy$-KiAz(dizrUH^cVWFje|R8d<3N(7X>!zz_K6daZ~Pn?L2xwaEg&x@Uoim%Z{`!}
zh)%Ly3935)XCs{Sz{BxEbGn^}2ZBKl3yU3TDH>R<I1&%IO+CEXQowrTI=uapPdVd*
zsIu;_{`(9tGM4)+J6b)truqcG>0y4mxWgBG4}2oh?7{+P+wns-DYoFL_W3h6`~<53
zg%Sc;K|KXG5zV%FIZk6G8JP=wg%$@(kjV@cHN(dlM`!10B3_t;XJtF0;|l!4$D=@-
zAQX$-!?NrQ*Q&#Jp3fy5k?<h2*yRJ6L})p)7$Z06>MveU`of89zlR_s#z^(>1XmJ;
zmZ4!W+Djrf9;n;jUO(Bdj%fC^_4!%g*pmUR{-x!|A8p?Kmhr!|QwrgxO-XAj*7#Wy
z6V3OdupS?}=SJe)x0}Q(C>R*aL?WU3f=E|PSQx9xBKyn3hl)pZ;0lwovE;|xZ$!So
z0Tq(inaa@h7@p|S_`(OkCp-ZkRYFp7ChMiR+d}uvq)<B7KQ1la)O?nA%`y?Gi=;jr
zYTh%A`@~pJ9Hx~gf)p{xB*6$V`Chs_O0NA0OvJBse6vqfG~|V9YUBXQH8PddR+xhM
z+Vj1}7gv9n{`p7jYxPK0k(pszV1RhYNO)J(OJqRtIFnhzKf|0$=8_x@k#WO6ZT)m%
zGx?2j)w#_jW<J;qVPZx?BEVUU_Fq_CQY1hP)puG-OK&?PlDz4ampfFos-+qGB6dk~
zcrEgE6_F___6<^HXHPM&k-`+Id$kDzIs`J(To!0~oSBjWQ*;|>^ieSZ$u<~few2c$
zgZSqb7T-afK>;<?Ss58POqimbVUvbD4y||JIrmQRiJ1bQwA+C;g#%e3BpkYP5JQ1Y
zofbZQ^eD}43)Sgi470`lF~o)?n2L<#f41Xjsu>*lM`IV-a<8H1ZrcK@el09uHyEJR
z1Z9%VYyt0ywK4Dm#vZ!rR|+Bvsfns3VEmjHSF-L!8mZ1}VgmSjw)?lQWUf)I4X^pC
z%`Aj&*)sJlF#I1%9xoC@*2u@pkR)BOvWn}<2WO&5Y54_&F$0-Kt?A5s12%xU)n$rs
zvJ)K;w`aqE1Pe7-W`+}7=%}T)g{G&A1l6GH8ND#&N@6B~j369!$BbILCOpQcVCLG@
zs~9AU3k#F`f?>$Huth+su=#F4{U9ucBtz<^?N(r0#TJZwR}c)IeYC-zKzxV}t)6W$
zs>HqG-YRBoB`YaerMq+!Ssz-L3gLo<TV0(*az4C&y8<TgL;HGzdu&5P!y_s7T@syt
z!9@|@*n`~9rSj0il8r{MK+&rEK82&fztcJIvj6<vp$JV+w-{zHrxbxC%E%ni?K&33
zG3|`GvnBfyh2&SNDe2!IUtONPv;nt*#5?#SR<!qodA0#oDukDUh5Saw#>e=6y9a8o
zAFg0=Wf@4E5^P^<>g9Pb?otcwnL&!K4r_aFuj9*ej7El-M(ruhk~a6^nx723MeP<-
zk^Q*n_3N1>FWnG39Ix%qPotiqPa%}`;4^>+k+WdZeD9y6q&1Akez?BEtaLch^Mjl3
ze<C><3IZ-13c=Nng)mX+?(OB_rk-Lw5@E_ew9hO<Y7a_jfSqVBi^=%iy=!)?P=lVY
zlQVH0)gtuJOh<IEJ<Im`XOW1Jr0P6kpTZQ#qoVXT$e!l4zDx;-LO>eV1kJ_B#I4>n
z@1b+_R1f6H9Lq|kHaj174YCrHTIfba3d%KhK^$AK6g#A<D(l!|Kb<>`IS7tA41?jz
zL;?&-ce64iBm`Ay<WU&FAWeh+s2Nm6AR7c79-^iwhjj}S{qSj5ebj815B9dSblY^H
zf%cP+sG9Q%3o#_jH)<So1~(e81LJcuW4O!1mr*FX{up;S=1I6T$VG!K#0K#_X*0kD
z4?^1jTkG{WZTLHc)$TkKQ`^tM<5O^kF@p<~KX~2|eK{8wZ0qm;B26Di*G-%CD5+aM
z>``Kt`I(|$`yd(wc)kOE!WtcIhIgo1X*8a0<tcQg0=;=I&}bl$0O0bEZ&J8<x=*|d
z$P^4qn2CNGhY=)9=%rw03yX&tNr!EJS0|<u7b)D9Gm@(>OpSP=Ps#6m9%22VoYT=a
z{Rg@ByG;=-`lyTrE}k{TIR?X#S><CkonYX=0}30-&^s)e9QtT?%bTh?piy7>q3>U*
z$9e>TJagdX&ECg`4rmXi0stf<XJAgH>@w6KabCU&geU;h2BIJ8>Y%{+I<B?645Zm<
zsDVSKzgrRqnw;F1uCA4LJHrr|*5pR{-;<TPLGWXPLqZgs2WY=Ovp{2}cx6lv{n(Nq
z-4XEakXr)&TEm-4S(CkaE%seS*`OFdItlmPZM|;vaIf*j7J~V|RrD1kI{phrqHd(c
zj*o~DgaaE+G;F&<UEgPI?9=0ckut6v9OS@GD+=&NXJ)v4BTDU=<FNPyy$l>esKqc2
z#$2Vzwx&0bOaW-;0r{(pKy$e7+sj-8>m$&|;2*kJ6{UwH>Z#j8BDq%w-qE&wQq}*Y
z`k`i1cZnw)u~<p!g2EL1TagIW1Vtqsolbf?HUvNhC=`BGvlE_>f4DLcnb;63Md}+E
z$ioqc=`}WJ01BX&0=9v!9VijpZ<>#mx{m7NG$LW>vP)19Y`b-hxGaJkpjg>DOb&ho
z0tW6-exbuV>FA&p!ScfUfsK!)NHTX}RF;(`5=!%pevXenh>P3C#s=Fv7+<4^Ap{Z6
zpJO9n{B6bg+GUt2sE3$@x3;<~Aif;}0f5rbMB;+JcwrCrJaz9JJ{s~wl#=$a1OeD=
zx4Q!CPSC2Z?C&PjrWh`_Mq3bTL4}J6T*fs}+>A2jF}<Sp?En5960z6+?#}7!H{<y#
za=OuV{rIxCP7^|y)|niKE-6P)O~Gb4z(5pVM_|q5_J9ilky3yVh(`l8MXJQmi2(c+
zJa;LOxS4RAM-JU2CQ~LQ9#@b?T~y?TiR6sLWFSE+;j<Kc`SPRzeD(Zv9(^ujU=Exa
zuJFDn2f7FjO9%-8C*&DrY#4O<nA7rkUvP!U=+h?ndt#%*7pvRc-PhL#+qZ4Zr-WF6
zX`oSZz8(|5ar5RG7?DHS6&4aA#7YkfMX)BG6G{e9Ea7>mqWl}!5?)9g8pqwlS%vca
z7^8%ooDf?unQzlD&wkh2T3(Gf%MkH&eOcD@P%uJfpX;Y1+6;z>qXqy|<wZn(JPA|U
zwsmU}mL!^}*d}}26x6I_&}u0hv~DL?(a?C0NFEtzi`v?x&y&s)3GfQslQ1*Ff}fp-
zJd!X+M4EAmZa>K^MfYZ4AmIQvyX9_ZC`?CC?BE8YHcgf(+<-f$iN^?|5SfmRiKwz+
zyN*!tQ|awUlsyN;#Ks1j<7we2JYxI|ru-}sg<L0uSnst@vL1nC##}99<8-{1M(t%e
zE-QQGGPGo9IX@&Srt7C`Cn=qn$N^S_%n;x#0^#xT2Wr=TVGwYAVFK1OJs1jM!yejj
zSH6;Y2T^ed@{vt+p?X#kmCe5PLw|8(=j$`%3`|Tw_XVzuBU}8sp#QkN{;rNUZ!o=_
zcSY<M<_d9N>n*Yo4+G_b$q6wYoUu^h$bN~6i__qiN!Jr*k-#w!uDynfOC1Cf{(ly#
zi;!HC`Qo`c#Y#m263}oV9pi0;1?r}?@%md4Guc{VCVM0A?@Shij<nWMVG)sQKRYjU
zpI+kDHU5e|g_q__ZLJ`TcGL7e#Av6tjtbpD-wn-xx!NQNrtU0Tw!p+VBwmjKF>|P<
z!z#9zq4%UKM_m^5)XKBESp$|bcd;esr?nr#M}%z~#0^~B+?P-+n$LRe@}VaVUVldf
z)Uwl3CM&#F;WekO!}_Qnyw~ZBim6~rc%iPB*9zPQ(br$uZ;DN%PK^&>Hqc_2pbV`F
z8yDCkc2?nt4x(ec0Z4^@e)EO8_54Qk_NVPd^-fnT%}n|8sJKcm{#G^N!rTR}Q=eaU
zD0<MCP>02^u)Nq_XHhZM&LF>qW)OSm=&0?43&z&ZD*ENj^YlL{+aEvpidgUW7Q2Y3
zbwg9_TV&<_G|VIwA=bDT(#7YhsQtxIvk%ba{+d2RqGe#%8JQfWY@9r=^miWn8cpyt
zPzQo2Fum!E{X9_FmEN^wpMCO&tV~H1sLJC+6uyew(wBrYIE*G-JmIdI=(D<n=kDP{
zUU)cSU%>LkYED`i5Jb$HjA7T^((LL6Q^66=s|<yt#QQ)nD;f4!T3P*GT0-(3`Y*(}
zp{hV_bjj@THB#Tq5<l}<@#tQ-E1BJ;Bj_g}>B!4hL~2(?YIk~aKmADN1!KwU)_&uY
zziPjgt=2z}Ff`24oRLA5AL@BPhgC*enm<AvE)W<AL->^h!=(PE34!$eoXn?$qbw!5
z8)Ad1zJc}*%ZzDSqbDE3PcIA!YjFsEzbEO5vj4ktu|V0bzlFcKdHFyO2Z&Hlu_a@#
zgK_|BQFIO<3F4$33Na}$!;b%7<Gus7q37<^J@D}tv9H>4sJ<Pt!t;$;Fet`KbMZt-
z0^E@B%JBLKHP+V5RbHq^hf;1?V5iAdyCv_X<b$>OiE9ezLsYRT;Bk;D_Z7mE_Zlj`
zPj3R|)_2$dQ95R3A`y>=r}mQuZGnS-DCeU$UPBQ>%0I=f;N-=c5x53$<|p>Ddv%uR
zFVL6NT;}~F^hJ`TJU;kkoj52Vw|4WeeVm$`lac>YUmx6PCB=Rv(h=tRxl`GkBvi3#
zPqM8lsTaLg!B2vw3T9*|7SiZ2RYm8snS}*Z>QODv`~RN#wM@Q`b2PkAEzHdMa;xgQ
z%jP7M|FnkRhlb{H%L(4cmeG%Q=2w@)qb{UP?;t6xDSXMh>XGJh`7%+;Akvtbgn3~N
z*_%2qBhK(`qpsN^^)#6OU|&t%!&<S@DH-o?@c%Zy_2Ur1$oBd&HYdvzn25n`V(6C}
zN<~~3%sr#@=li6qZ;9O6=*&f6?qDfmhJaVzf@t{p+d((d&mc5oWK2@AubLemNo?1&
zm;%h4Ro`cQy;a-4ftfkdNM9gg!2Mo0rb5{HP;9m|m_}hW`XA{>M~5<={6=4gbyC7*
zNjA0z#x+*C#OxKOu_TN%kO;LS0-+lrVPQT%F_iy|YL_%b%VRDJD2jxcH+E^L&>A(%
z_FWp05kVyu5Z(`%OH2{wwF1>=h+f;I>)t4A31IKOon-JK!k9O+Z1s0r+dnt2D9OUj
z{@P@Z1Nc9qqZ~XusY36;_^Is&{~AC<<VC8V!6r+8@rww1kSH;M@?jvMClW~CEXF)*
zTeA*x*xWE1gqi);ZQH=SLEL!278i$2Ic0g$Ha5P!f0(oqDPKwYMQ$D#C|_Gik~noe
zXRwh#(yA|jBIoSfGUPwyrU1P+IB_C6KZt@lk#dTQiAW8G4`O|TSX61L(knO4gQYqt
zy1?clwI5-^yTtaj^HHQRGm0|T=L+k>X%yc$_6XZ98)-Zj#36Z&Qdc^$>KpixZE_-s
z3ik_RafI>>$YEI=|ALWebYddLM~=O=c-rtfATJ*)LM?R`Ly~Jt(--`g2MH4BpFglk
zCO~}zw_zExxiaDOBaw<s-2PlyT=1a@qV({r=-&SQd->DEZ}m_p|I^sZ$;PH{YKn+@
z6mTK+S;+RhH-71NSvNHFDAF<e#VOmr77VcRLe@>uKmeS?#1(wu3bRnP8Av^~G&P$G
zk6QnLG!xho0B*z;$$mMq;8)<~@6Q?w@=MLf*JhcM6B7mJ^fhx!ENJ;^Z#WDzvpj7b
zg;;jnJJLXuElI(B8aCG^&4f27VKQR+SrB^ZJ$oEMU4T1&GHoMDACNd>u8VFmvMPo3
zGko-Ul)b%bTAC+ImnYy%c-H67Ty^Z1E(BcMI|X7|&YMm0Y=VODoOE`<5fRd}p<$OV
zB$a&aZK$R;cobhjlwFzkc3dCv(I!d~yx#=E-UE-@aLENjLv?-@|9)6VM?9N@ObEBB
z!SrvL$seAYlF}i&hJ%45(QyJ}tjIl@^NvrD%ttzzrpJFT6?y7-bmJ(eT$Wa(aI9cO
zV!Ib}kc@p%Q1EGJ%_8~^kOwHY?mRXzHN}ySn&N%eah-n*L{Y-w@x`3n`Q*v)@*AG)
z#jjqKPQMo!VDN0ZP+V~_c%w5VDpt%fhCZD!hPNnY!jo|F1CEap85a+m0W2od(~sk0
zbzZ>$XEw)4DJdx~&gj&sozeA#3>W<X+==pyu47;cE46jM@>+X}I*aV96J1L%cS>s?
zgV0a{_flFKY(e3u_bB)85v6(v=W}CQq&g5;Ijw&XKzic%@y)c95R;<xgHSRfV*^1F
zRS$R6f*$6=g%qvSWgwHX4xOwJ=<BCr){V4iR40bX$VBaIYhwjfx2kug!|cEIgVQMB
z_9t9|{vO<EC8Zwh$URa<LMZ-oo0Ogi-N72caDFkdBr>uU#`$n-B&x#JQ5=a7o!LaE
z8TnUR7`6^jcaSp0RNrlafj;Qh81|>>ZQHWt!q<a1OmW3eo0tG2f<`=c=>m3P>=by)
z8h)X|<Hi&Y%=h&6EnBwCUX_h6aqLx!)wJ431czhFn3t91<mOg{S_&{&5Bft?h+^y=
z*`ojZ^pS)ABR~osIj8dJvERUtPzc}?g2@96D@1<Z`tY*pfjk8XVL90+g%mtypKXYI
z@N>%{T6tQ=zK^;v_`DYtbp@(w^rkQY@$>UL;%SQ`SxKqXCPEj@5&>N(z$cPTasq!3
zu{lC3z^#KsQ$fk!bYNx|4mzz=q){biCLX$Ld+fo`y_f-AYe`S?Q^L(Gph$~F){**P
z4C6i5e#Svb>wn<L1JtY$`!3J*0Tz+%2s#4w!^5M^E+6S>dU|beBf%$o=fw!eHbsQo
z1It7&zVpIQh|&R1sPpHnxOHv&_ZX2IdC}b5p9lyAn|I37r<m{?&kVbw20%NC3l0x!
zFE6n0r4Tj&hIQK#d(dI@73d^uqANAlPmP_#sc{bVIsDBQCVMk8GJq`qd~s<88V$mS
zsjx8e3y%H2V=34yZ4FWQp9WJwgwNcp!<8JPEv08KCwzjNDMosDm=QoF9?i+Zl7j#~
zU8-wp1Eq5C@%`|jKtU24uNNge*HOYWn1c9(4hlm$@n}o}2teugZ#cqan3Z()@%i4I
z$^SOS(Q22!1_A_=T2#xikE<wKA;-fFgFpo86{}oy$i*g>2OpC4&!9ZRRiI91<O+01
z?Lg6PXn3qg8Fd>A{mmT_?-uRiUl#<-P2O1gZ5VSvNzzKv)1v2;Fwhhw#C1i!My<L|
zC-g?ki%D+ox%T(K`yG8_%f4M5rd1Sx@pP1!DV8FpJp%rrWsB8&8bXILWVzcgJVP9=
zKUv8T^MO}LAu)$Stx@y!zbiW*y8fLp@m_G%!ZUNJ@nFMtK5_96eLgOopb~GOA2%$j
z#L^uyBf54z%WERPd~EU_U}qWz6Dv|bonSS&G$~pS27Z9f2}T@NVW)^em=M+=o|?GG
z$n(>$@n?th*lX}p#Ae*G#=l7m@eQDWS-lx0fOWQhkavu1wjL**`x>5cp*zB&qVRsO
zzQpfh^KS?!zEM^wu6QQ@=a?ZzeTn<8JWovxT#1~e-^RrSJX67IQDVl~P%e|G;$>uD
z?$I{jF7IcUFodQLU;2b2B5`0noR*ncsPyA)!9w|)6;aW$+VFme8CG@>8Iz|^Lvx9%
z{KZoj2kb`405S!mPC%N_=Algzh=!|dea2`J0#7vb@1i}#6DkYDu&yo*8}rMTTIub?
z=s`(IiSW<vZAZSkkr6@{5hq>U|4UtXIZ@FQD)s9)`fwSbmjAOdtV~1le^Z2q=BDa5
zIF-c4a<H%f=(=cUhx;#d=Qo&tX5LE|>fK#knc^F}rwKak4iV64QJ-^%)mmhu$aQeY
zFcF^wc|k-(s`QdT)90ad>k?=%)lzhSEqOup0RKGv%8M6qE?`Gh!g`KA>mBQT|Kh}(
zw-(u|k1*sCXsSPG;1lO+O@^RrBof_FkpOk`(<dYYltOnVl4SpNo;zl_sdD<)&lF6*
zfcb&$ruwuT5k~whbYQam%>NuyNCm|$56lFo=R`8!RC+1A)KB+Wc-;#+Fo;d??rHv-
zp{Xemt9VX6#bVuxw5lrB7>GU=nmYfsP<-O+Nan*}a7TCLHrxVLCUOe5ynzeW*67JW
z)4@gnc-;8I@zno>Qb1ihR)E7BuLN9TbhY68Vt=J(W(2PlWG0*-?^X`id)!`WwUkC@
zjiU@niia~TLW6>qpaB8Z0Z<pDdP&tE4)Gt|PEya5;{<JjoD^0I9XsT<oF$6?K|F7h
zE0T0Cl|8HY16h&4fl`ya<4R4i$k&|?E`eU`<P`pr&VJ$zP7&0--H`G}4iECaP-v{E
zIN0KJ`!BfT?Fi>FL~(!+ZJgR_IOAMULNm0Q3-7J!qdn!Q{To6Bq^LkE+r<*3pZ;@f
z>~ZffsodVW9jtPDo^J3@TTTOV8HslJ@-_f?apW~hU?d-sl-K$y{PAfB9j>yn&#&QI
znw5UEc%*4fqHo_<kd04BN@}vm&NFfVV|K8Dzm3KhjvGV6&P0S2ssIrZAiP&6L_!}w
z!yeo*rGR8(xM6+m+sR&{8ZT;C==(vhFhe6$EA@%cd&u&4?)2v{Bm>`%q0UcMo+fhm
z;!Jxk0K#?TV({^SKP@MMM3dM@Cl2)I=kFBJp26*5Jwm1eYt)$!(EW90pQlMXFTFFq
zmttl*H_K_I8j*Y_9$yoO!+HGjTi{qr1CiLd1owSZe@fywM)M8ae|JhNlH<v5X-}S1
z^?Aky(cM4$;qnWk{4pBkMY3XcTc?S4rRoO*psJI{#MD!qz>DaT%AXAK;^xc9oEBv}
zp3(v;5a#1?oX3=WfJ4N2{r)E|&I1}oJUbr_KMAn@FD}3bEq^kA2?cR^s19mEEipxV
z$1e5qncy$Ztrv!jgBq>Mpqm+*m;9W~De?EgfTQ<C@#w)B4BHDzem~BhmgZ4#Batv@
zuw1wSC{n_cAyz7Sp}fn7jqR>7JND@dH7zwbH$#cUPjU#ho5H+jcUm2x>(?aH{cz&Z
z$>wwN2hG~5s?J+jRPy=?u3=1^y62<rF-?i`LFXl}Rha!^N!sPl=r&#|2xZi_>i3Uw
z|Ag;t#eg-WIWRZ~0x{rVq>b*v+BPwEoBk5sH2k`#XzL_kJ`6e$rnq?HQ6tB=z^UIE
zYhG$ew>46Lqh~uXN&j!4fhW@CiX;ARG>VZ7^j2?78fISk08?FFa7Zlwo;4&bv#k*u
z5-B`@{Cq+C*4$M^iG`$_gZ)iZxx&d&s9?-831Ka?ICYw2H&rXVe$>2uiv(>nQ)o=5
z8(i$|H6NWMft2?OcDg`Mj4}iag#pfIW-=3|AEVK^4`R%xoSAMuU_GJ(+!ROD>^yki
zD8_*J-&8&5VF7w%oc(PH$b4j_mC%tTzmCDMIlbLP{4;E`P=#nQfy{(9mD-n}Sz4aE
z8!4y_Cn|&lB>fq;7U=x~iS`m^VPt)PN(I(d?p2=@W&4p5ZT8r)OZ4RHz@@RhH#Vx;
zq*bAC*$m7PI|Xn1M}*bFE^2RNl{&2*-}Z#{vEzQl7&RZcc(M4hh=)m=w`te$2OU<`
z`3F<C?dqiRziD2Y0>F<^5PjmR9msPi_4pKjx>c0(1GZ(CxY-NR*JePu0p?*`%-D15
zVt#`+o4xEZJ{NF&frV|>^#4g%r@g7_wz$|`TWgJH8U8D+KjxUZ6?nHN!uRV~4r79{
zWAxhq_eY<9)M;=)17MY?`j;#-@|=q#vHs#^3r5yEFb*p(p&y(a4OdrC;6UVF)sWA(
z6N{3^ePC0LtMZN=eghhqjd0RTp`J%dy1aZxmqQvVe3o)xK={W-8TVhjLr62fj|BUx
zZvSI<P3*wjXjUbPJsup!z^YMf;lJ_c0>vbLW)VQ`6<6bD)Q4jjkDa;0y>)BM<tY@}
zIA`8n`$)C|FaYR;#$yKX`?<Ip`hOvY13Ns-RMhWh?*01Vb62}D?a&7ixN%?yeDvtt
zrlG8dBRn^X3pWN=1I&_&6}ePiu;4q#_It6$3~|N%sim6eG{v`Xij_Y9ROSB*_nq^f
ztA(qBT(v;`)y=s!Zd55-_^UTNQUG%z6rW)KLPrE<Daa?_@{;>rW^}y1)88;uFlrt(
zc5bc<kOu>U7)UZq#4}&K;Nsv&d;T0<vxcUotFyC)y85%1FH6z0B&Z#{Xm8J7uL9_*
z(Uks%ii&qLOVHaQ?L{m#nQa>+?BYaSU(-b1*Jq~1Y3YKYl#xCFTtd#aPLZ8O3M>aR
zOek@G(8dqlRk;VgU(Q)gFQr~QALw{M0N=Oov-9o)9y;PBUZ%oPZ0zh5$d^(<`vk^|
zlw=w9rWXSJa%Y+<-EH%Yj^4fdf?&_ABT2;}x3FF>au0l%^fdn!rd)<{1cp3VuyCt*
zuv_KvFVM+6I25**Q>Xw-(MWmh)BkQO%ZwEB_r)MO=}@&NC={PRAKCGbYlSKa>E-{{
zi(w14JP&cJzW%49JlCZqB%HvdCAdTv*gx%)jA>tlaW!=RB)9KDg6}r4V*WFjQP>lg
zn0U#~Zbzg^uGNVvSHL%##dQLl)##N$&@yQ0>2tgUh4${vhJhzM0?-UR=myk6Rt|So
zKE50Fh9g&*1;jq;GGN=ZzeQY3JYM(|h_BcS&f||yV*O-D`GVUbb8Ah$*RhEma=Kt3
zfWeA4X&-e3IZnDz+V`Qddn?arSo@HC|AqbRjC&bwy{op^<9^e{nik*V<e*hdPs77y
zbZksg9*m8qIH`o*h4Sw_K8Uo^2kLB?v>P)d5dWb@1!sVYWEpkf*w3Jl5FD5o;aQ$>
zblfAsR=znnDaO16m_xnd0obsO(A48LTNVINb)D4$wgM^G!a!Z*{)Cq~InLM%8cK$?
zafB~~=%z{2p2*<R_TH$ar8VnE<spfV5HTWVh1!x-ZFX0%QlxxgDL8P{%Gp6-GhH8%
zoPil@M32^RQ>J0=tv1T+;;{<*OqSMr;`tz)y_4OF)77fKCKQY;M1bYxwLNU-Ac5xr
zI@m>N*a$x-=osU`c6ZScNsb*5)l(m{%K)aKsVNN#I>?0swyB5%zy3X$ep__C496hM
z(VFZ8?z+s0CY`|^rK)Xy84kRkK7U3#2<Iw}v!^2qpM^C3efZ8M>+arty>@s8fHj7R
zRt>DT@7|>*<uGDSrNauY8;~|;(!iOJ$3i*5dOLXgN285ZgXlD|b%Qm44ajxX%h;Gx
z{$PKv2Qh7c;RhsfQWFg37k2u0+!dUZzVjICmhpWthl1TzXJ>dHp-Xi@DD<f7tgh4a
zVnVeA;UAE+3A}KL(WA{>T@8$Vlijb6P7jYh6=dk1s7OvRCicJZM&N(-8_;_;H7Q^J
zou{8Jv-%Iji@IiJn9u;lLLiDug&lZzxMXC$^@EZwM>F!Aw6~d>nljMSqcJ<G(}o=p
z`&lk{3HnFiTM9RZAg=_N22O&^iv}9hc3?LR9(qO@1<A<)lxIFEXPOnI?=xomkX}?2
zLpfF7C%=MD1XJ{55zk?{x%as4UwG^ylVsWCjcVmpFTN(J4*U7@VTx>MvBG~`nqJrk
zdGgn<h;U5CNboHIQS-k@iLPihoQ1vZ5{2{B+W<lvKlmHF2Pz7n*wjlh$@8nof^QJ}
zhOkPQr%1hQ-HTIM|L$3=Qb@GH(85R&1Mt65g1?&n;WjZg1{B%>VL<OWJ~HHA@#of*
z%41JJi4W>6AjDa>d&-l@miCH|F~FjA@zTNmcbuJi%7z2`s1Xd}mPiie{|AniOhW7!
zSkU<I<ljb!B6*=v6k+2Ib`=0u#JZsT_~a$@APVaSgD_wbko3aFgAl7iURv9q-7yZ(
z2_3vgY@Ry5QLfd7KcWAY1o$&6!RqZ40uKLYNdUF{1A!*jp3FoC(IluAkR6maH$0fN
z=;e(E>%<B4L2LDAS3!H)v5~gTZVtE)?2-^r1J9+9QR<mxZiAKnCiZ)~`w-ODbsqf}
zq>On6qTs#ppNQj_-6rb4Pd)T9ZUVZ2etpZKaOFLOKUx@1p@e2?tF835p)fc7P{tQL
zh-gGfV4$a$tS1cMM?GKlKlE~%oqe0YbHOb{bpetw<rE-j%j~9*r5Q@PzB!K)<Ty7p
zP^zu2fPlSW$IqLW7R*~%go$0iW`cOKTM>;2zEbRoTQ=a71I7^+7ykt7_x0DTI9g0>
zs~nNS8wkSB9wnr0kiI|^D%I*-N^CLpg9LU3!52O|T6t;;$eB@#z`3EkYb!!yBO-JR
z4MBKX#MI*BSLj}W>C)eXjWwC?zf}XSHMR}Q?YW^hXWoNlK}*Z~!!tv9PhdXK0emFb
z{^c8)>gXJFw9f57GaFfFDMdXsv&wBSuzr1kN=PaxeHg!IBY%olKtN5D4b!V|!KsEG
zHjMZ!vSXO4H?WoOiF!+U(V%tIH!COYTNlViNV&?!{)P}1FdDG>fYjlrxXFS^@oboX
zqZMFh*b5)$zlSpiVV5nB{i?n-R}YI~r$IF@s5quh>izUd&5oyeX(qt*Ik%j18WkRE
z1~L@z3ho{rFp~fwqGF&2?s?Ei;XwzG9}%&s#BB2CPcc?{)DvRS?)!Fw(Sa`ou>?K?
zb~hkjE{UbKKtl1b0(Qg54jnIM_Le%ZFGxBmqzS?FQVVzpK`_#M_AI=95Emamg=GmR
z*@04jb$&oN@rj8At8a%F#yT3h+?SS@nKx_zg%1cw!`K_H&42*L(o*B%<Bx5HF#^aF
z|MuEAth}0U-_uJG?6u;4(3+mGq#*=0e_mmS-bjDba_+zp9WJK8QeztC=!(d<o0$U7
zHy1jm@e3GKVWWTj+L(8v!t!|Td}4hs79h}^$QD;b%nL*w<G+XM3b#`md<D#=c3<zc
zL{u3RC-~mjDv8}`SrKT~!?AcOY#tz90qQ`vfNL^9i&n0|FX3vMN0#exF}Q!^aA5No
zUHZoCWAiAILk!yt7!zG)N=apK=DWB~ux<}I507t~m9{_dZogUl3(h*ulF#L#`#}YO
zSMKkZgp4tF*Hdt#0o6nnSI5odua0pf2|)hfKEVbng~Gx*$Gz@m{14yP*XKIlVtCuw
z=uoP)nFM@JQ%S6A9iLks@rip~AW^?i?)68sVSBGQS{{rB0O3SwT!5UVq}%zaFPxNB
zczHwDH3j>DF#<zb5xWx-B5km*RF7rzi-VPV5Y>GCapGm;ryq0t@#W67(@Ux^i@}~u
z5N?9cuYwt>W{996yGIZO0$~5RDL_Qmwb%C2C77lp3%`z9U+}x{*f}_y(A2bdaG+9K
zzE&xInwjfO8#&a;H+=~b)T1ru@U-D_;=-f&!1dL)qDKl(IH{*-%oyB$aH+P6`3l}Y
zY~d$^J)Q)23S<by8d~_HEdxj~1f>eL(+3A+KuU&$=uS|Oyw_q@g*Z4Vr7;g4B=2*U
zdr(wV1aJe*9ZM_Byf{VOvl@_O1Z#f;8se9AG4wgfQE#6c*PSb`Sihe}wCm$iqirvI
zVtmHG?2Y`2V*@WV?~L3r#&V08rZL7ve#Ar@FHy~^-)|`+J2tcF3bUQUkOM>tkOLsE
z15`S~%R%}pu(LgbC$2zlhK`;d9~Au~O0k9`8%SuF>e%tTHT;Cs1S!sXX?n@_wQ({R
zNA5f8qpAlXuFPjmshDab4^NY56wTG=9$61G)*m~Tb%NeTNu-ke-`F(Ck6)|vOJ1@n
zRe|IZ77wobyU!v;DlZS(<G#^R`1=b-sN=LC5da6|`AcHxf0E^?cDE>{uek5nyZ0A1
zYqVo;0uC<i&pOraKd84}ibN3;`V0lAH(>}kH}_Ca)tuMTs<Oi0eCXJ2RUv8VStz_9
z&O-x%!Y(=dK*joOga@|EN8&XByrO=Dyg(}zqdB}RCNN%Cg5spYj_^!}uBL?@ooKC_
z%-awgb8z-W-AZ!v@YuE|F6;RJ@hSNK{F&<KIi&M(LDuXQ<=VXC)Oap3Bk1(mpQ_m$
zfl<MK>!Q@g%8{!JD;VT!Oci-<8W<Wv57q-O3A}}yn?ZUzRs=+}m~dh6lhb0Rs!C4R
zhx6<NQ;ZH+`gDD#bagfF|H#E>xNz$4l`trhqZgCMHF~MqU@{lDQNhbI#$<p|lEc1}
zG9kgL%>9oCEODqVQr;d%R~FJHEA?kyrM?qmTw+*Da{G3Z9XXOAG;scl;3NxHk{K8o
zaF!<_{o2cu_dI4G;M{&HOgStPV3dJz6pimqH&0QuT6XL`${B2U29HdVrhXQhKV9Td
zwK=x)4>@Kqnrf5iMVfVQV5QkyaGO+omXYB8B~}Xm9)mJxK=b!e)U~6dWbyAmuj8du
zg!9IGIIn!VtKja;H+*BG5kI~)pcAZ<XU}LM8)I63;O_4iwFum3L{?M#Z8_z+Sn=ia
zWY&HgjYP4CExQl=0v{7-+WWgRkgo>}F(rX{pzm9%lvQeOt=*)mMa5@H10oaZ=J`Y_
z@F1nntoyEzySX)3Na#O0YGa7p8fIGBt=!z;Gt_im!Ed=wLkNRJkXp6)+fU;{d*9+B
zRpJI!o4rnQ4^XdoHj(2S#Ay>GY$LKS_FrCq?yR7q0(}TaAkmKjjwdd`PnkZ02p86k
zz8M4)8pZVaN=Pd5B>DNv9lfQhw#ss9C6MpQJQBP9QyW!t{fVEeUR-MW_Lj~(TXS-5
z$~R2O&-68z68Ld~@d1D=pjl%YKIQK614HETe-=V7U2>Y@ZELT%K`|fCLA(uk7v;jU
z*zByTsEa(dP}4k0qhqC~WejN3v)>-L{$5->xU!G7)b;YnIu<PO|9$^BD$PeA8UFna
z_ety8sP+H%XxCp7^!ne~(5-8O{{Dy&Ik6`H|NjxW`9zW`t(c65qTcC)E>a;ODp>l~
zhtzME4FBunDia^~ot=l`n>8iSuq8Ws_vIL5ou`4YD?HAw4CXKR#E%BoElJ+(kBpoe
zh@5)DbA68o4XK3tr3qOCtyWS&Ys322YQD36S{M#@SbuBPZk%YhfkAX%OUjW4|N7R3
z>)%?$<P{~4w_MfMJAUVD&Aq&I1pa;&eXjM-LcjKlM}NbQ(Hmud9B!ogt^cUL5b;+w
z0vfxzN*#WhH$47(@fauwl6hQ^CiVV=eE-RA<9BI<$rnsDi2J=UuCA?3b%Waab$(|@
zoE6QQs)wxi%8A-n3<21~U+6=pLHuHJGJbx818&RnHy_ZiKMO|1e?P0j+MCl=UESXT
zl}-Bo^~<-=6TdvCQrnw+bX1CYd2r9(o1lc@_la%%`whg!!8YBy_kNAS3?tNg7AOAR
z>I^yJGlXBMtc0}sE*F{Lld=aZYoB`(Xbv0xT%%ZP$YT6jE@8jpp&vQDaPZ?84tm|<
zEBT_)EqNh%ZAn(2?_Vsi87kC0{`2WRnPZW8dhfSXY}u!KSkFX7mWnmz+!r<Tn8AhG
z;>H9k>aFJHF&<^B+nQX4+?AT#hbopD<eTKhXD5AHKV`iA7PCW=gx8{Rmh=_;@6aKk
zWFGH_gY~v+-mfnE+u1!c-nVV9Urm^xQg2%Y8UBs(vs^50Bwzip*1**z?Nc<wKO^}Z
z*s97*{DDL=r<McRpl$EUg=73IDY~{}LaTmlJVs23`>3~%_upuu9;r-Xvbm*^Qb|KG
zTj?Xx%(ya2DluA)s)C;?=<`J`jE^Q^_`qBi=1Q5O+tJ*tau_f!PP?{mDxWKrxB;1)
zbf*pVO=iH3Ogwt?CME8N>G&ha0jq!LR+Z56-M(G*t*@W#75<2rH!-mY-HIgms{$6<
zAvx8Lqx;<AhhMfoRDb|KW%`zhL2)i+PfJS+0GNNh377yFrmJ<FLsyGGpg2Y?edrKH
zQWGR6u$_Zo6b+l~dp^zd(o%1bO%F?ZuPlO#7Qg=ilb*40J&W5;Z-u@+{?`?zeCS6k
z=oF&st7|EUwg|--`tVaU<fg|>O}8KKJMg{#1|>I1r3@bcl@tW^Hf(rHypXUAlYLKj
z_ajG-uF14uq(CB#{0=XtAdN&sM&_ba{dDdkf<#f^fBSas;zdXpcCpl<-$1}8e3#K5
zErQ!N`}6DPwqNA<+48eIxOF54?Bt@J!%lLbh9V_209}mLO%R1Z{epV}{D=|Nb)hDN
zL(y|y`gGpUxBb6*05Q&2wq+9^K*dEOn3+_bgkTw%kl!x>?R6<FHL>M=Ptd!X8UbNp
zyO!i*TuevBqf?WUWBB!+pjN}NfR~#n40cElgAb82lcofL!Qf&XW<_bm#nq+R2yXBE
zXlvl@iQY0;f*;sjD)3D2IaD>RHN?A;w$mRPAtS9GP1A!ycJrpN+Tfrd+iY(=y?-Jj
zH*Z4w1HPt>12Tx-U`P$H7Q;>xanNCLOYt>vZh9;b4ja&qf^LF{xgFnUmT+&;;k}wu
zMpK%iD}Vj>EG!-{drDHe{_4dGf(;@mscV{ARI~^uQIth^1P~Mjc{^c{hB*u&Nwz^|
z1fYb&{L%FoAsjo=pnCS?NvHmr4WNgx1i?WR?FleoIJ{sMpUuLFuc9JIT!HORAKhT$
zPCj4z?AdKm(RZJtYwo6eF&L{1f?#(6B$dO5@!4?I#6AID#_Q8-q+YoK)f3%fjmZaI
z=v&o7wnG1(aatKJEf7F%Pk_y=MYbH2x^2&NcB+Lmd;SAl6nOq~J>iULWV`VT3gRi5
zZ$di>-;T4duR2r)(1|6oBK-xmd?w<^4c|!;lv%I>8)No9IbDCBcyd$@@FB$D?&Us`
zn1Ra8n?AxOWr?b)Bg$)!_zr2Z>NC)<(SB)a%2Zr-M+J#_amQE%2mmv4b3prvOg}gf
zBf=A8nL~%;<h`z5oyUy*_U-z?TLA&EeRV1BLIVjBV6h1n(PgL~@G#hIhx6h9bP@RO
zK&-aVBk)Y=#R~8+^tOiw`uaAquzVjKHN;pmD(bY6(eIfiWllGG_C02n9j|f40e7Di
z8gP;WBL#TvmY~CE%<;^@wb));gxIuH(s}Wnm=JL-W^~xcb#3*SmB5iB9}^WVoz6H}
zgFSF=k85pJ;d>~Hv5$O_b{nG^rD5DE;X2Yx5Pn|3pt0Bmb8bujFX+^qoh3JQbaraR
zN~YlLUU=h%<E5(VrzfoFK5Yr}2>q~Y;OJu32bP^oY6l_*1hpRU?s#-k&SgmRhD!N+
z^eYSF9WDzKrWD%WAOW-&L<^6HaBQ0I&@KC(H-@5Y5VP*QL4yxJqN(0TR{rJg_KT99
z)?+_xJEZ{$V$}LJi}7(Hjh57f_O9Q9(Pzi519Ue*o}8ztp<!K4FMg5D34g-(_FQEE
zIJmo6?;x8zltb#)?U!fEAbr(3jHl+L^<)TCNzZhiV3;lxp^gVFFHr{`?B{DjKF0B{
zp(z12cbJuvGyMZreP^e~%Hl5&<<H2v*sdaP1B_7^3PQA@f6Bnv7<?Mcl!Fepe23l|
zo6zl76OqqI=J#3eu;pA+6Hf72%{@hKxBUwHH9$b|*D(FSXTEKN2>MsX=-(g^ygdSl
z<wZaVMjNkO-fhwWuN;(NtLrk*MqcWEWxFLu>8er^0t=tzALHGaWjkYRdG1^`oQDc~
zJdRK9wo!Vvel{ggD^8Kj^??2UxPMTFndj-_9d;Jfa;J0_td6JXb{`uD{R30Be0nfQ
z0IsGQR^I=|#mT98ob-8MVq$gbTNSJ~iKRoo!};=Mq%(fXu>W7hoq1G_d)V(EZ0${<
zj2UVZGDV1@QZ|n%Wh!Z+3{6TInlvbE6dT)|icCeCG?P+<Obwbe?4nIcb5fbo`P_NW
z`>wOrIcuG>*1Ohu{@B~vyWG!x-@o7WyRPr|dwnlk+Yi;%hKvO(d)rW7A7L8YkVQ{)
zlTr(S8X&v-)vXeyS63*U@D*Pn?jQw8bm{&sxqF27cY{&9@u514CalPPab9k?UU-;3
zvZ75zE2Ey`;_I9J$%B?PlRlcuF9Qb%ZX8y}P&5!R@eB%oZIGCwI%Me3*sTG@eO*k9
zW@GT5dN=Re*-(0y%<p;LxpfN;Gs$|Xuz^Lgy$Su(Z=O79XfSPB<E=TROFMsa=S#9p
z#yGxgkYzlfg#59i@s2We%V``isXBYs{l`JIVB3-R+`ch6k_xK0S1S!_Kt&MSgzn5t
zgjO2$W4wjG{^xqU@k8^niR34}!PX}H7{QC}56BS}7FOB%CF>Pd7|I1ElaD_sDIr=T
z@TyCAwL#_GC9KP-Ch58h>sznZv)6>j!}=aTvdcBxleR`JfBhE-A}tUyCm3l0sg1N{
zwztxB&&eLbm8UD{Zu(WU*s^$(ZPyU#+>JZpWzIdf-2aXN?l^56X*(`I?`GMI?3vsA
z#8?x3y2q<#{$ZPaGg=tmfDxflt0oqYYLifGjtE#fF3-a$<f{*^^rJ<~#D5qhh7Sex
z8h8Sbk3hsY@Py>A{my5~3)lUte;-YeY~i1M9#+a4<lo<W<*LS<fJB;*E##VDWM~uL
zHo4D)E4*?->Dr_ez(}!BUYmZcZnd+q5q!P`Ftq5Ap13=(p_)!^gYJCx_?^c{)ym&B
z1kXC4rFd01m63Dyuh-M-Cxg<U{QP%=$!mAK@@N-=*qbKN!kWsJ2==_YIy=Z92<(hn
zWVv9{IbgtXfu?UPH%)zDxbD{O)@o1{q*%vB&CN6V#;gzkKcr3Y7?9;*1~A5w2NtlW
zh!Qmy#{3h{BgMgI4~L!cGB2-AMC*Pw>~!~6<Lb7jdgMzg=Qfs4ePtGZ!})E4#u0kx
zuPr;}QeS*|O&IJ{V@2vWTq+jY-RnSgNlyqWCWW2YZH|a$vN)jjdIJL|d|k+e=$C3!
z3|aG~%rPy-vKWTfs4{(TF9e17m<Vp`RaJ#omke<3WN$#*4?Hn{wCpUKzsIy3I(Sf!
zhTXfzWbzQ6Z*56Evx>mFZ7km3Sv8S4et7Sg<j9cvA#*yLuGX3_A-p=LU&b{zn`4X$
zS%!`q-McVgfI+#iF&u%_m6bCH*&x>|`bTsF9fzM+jVSIcwt^j<2Ut}kel6&aLP8L|
z;l%Xd$%(OyP*$IHFwNJwuBN_ZsNxMt$*Ua|brC)$S&T*DI)G+S{v1I}D+AVKeBHSV
z7wE%r=K7pxB!V({Rx$6M#yR8GhEe_Bf4Cl8k?d@@etkK+jE*e8r`|pPShVO7XKC^1
zc8cgWr4HIyIUvT2xs1Y_RfIq-GxJbJC3AZ$w`{pJ<}31F_v;b)@U$wNtmJiz#tqz7
z5<4Er9WD7k$MDcs`RU`wO&?ra(=svHV)p8?%d7AmA0Xt0>^Q(xz9jG=CLQc&UB&RQ
zuset<*@h@Y0|SN&kJRMXKAPWOq2SJeuM!g`Sa-nasG~eT)vP9p!E=fg@P+7XyxX+F
zyVJSX?+<66MNvz}$yD<c1|-Y5o`!$9p)r90s^Ay%M}yzujeq;Lk-`bwxXA%_+%fCr
z@kdvY=DbOh{1kOBefR7fqM%Uo%KT||)UJ%J7Z6wA<(+Y800x2cFRub7;<MF|A&(qf
zeWaweT+K=#-5?E7ffl}bvz7xBhm>`W|8AcH*Ryh7+i9y#I#_3HemqLU1yRbi>(}G%
zc^w@8j=NFQ{LE$wH@FGk9+{jhNVRJ}GHlhkr=z81;R|^@oLK}+$S~Nj0Xly})*H@q
znu+P?TqFi=XFhOX$)lp4zwop^hcD<kKrCkODS5c=+7&os^TDFNdC)NQ6D&nW@N^67
zB~Zn>e`fljpsa@^ghuVzZ(GR82cc`cvo_2CyB1P8Iu3u=^1RKf<me9Vo3MyhKOBIq
zly!$M**-nwdj9p`JlEGJs+)JZy1Lg*{p#HnEt#zy{hWq0aO?>a-optZpXIG0Xhqlz
z`#e$j_*SLU?NeVIlUy#ejUZwK;x@w$HE=N#yWQQ~AZA_aiu5|R^Hv)2udifv@C}@u
z6Ldn@;-6YuM{0SxsVsk}W|{8OgHr~qH<@R$dGl3lc~D(4l##SbDAS7Lo%-NU{2oE%
z*|Uns&)0;p(xaoBn2Z1@?>CX;Lq#PHe6wFnch*)$)~2A)b~0N(RQDxC2U%qQ-?cbg
zzZ0K0HP~Z(<qPtgm%uF%6zER`#$}%Qz`zx{gX)}VtfoB{HSX-$&b0+$ihT;Dsa9Nz
zX5?>o2~|$Jwo*GJe%55Aagvf$xDx%YM9n(2;0XRpf=pt0TveQ3oihuTLXD5C+Z^Wp
zU}`dA%`VJRrWSoDiVSSDK9hCmeND}cACK<m<(<?Ey%c_am(WmyrJ?P@1-*ssviT3;
z{S!gN<D`G1e<1Ltibc9ewX&Us)}_<uJH$uHvtsi0qO6KSa7K4Fk&%%bp&a?>*mg%+
z7n!c|)Hq2{-wvi!l+*Ixmw9!<#hT)U>A~f67~%3O{=^wW->9tv;oW|yd(jchI4JRZ
z_tBJ=nKZbsLf&18gn)DBUP1aU*7Iz`;b^W$yFF7kmmkE@6kkeGIG6{k1d4CE?N0hC
zuDlrWP1kY5h6;KM=8N%#K8QY)s<$wFD`Z!=wHXbt1zvAK(ve1y@e{H5&NSjlc^Ya@
zoqE)la-80h3eT(_y(Dh!&84U^DL=_8gRYVG6SL73Pl0X=GYbzKXlp1t_q;QI)6rnh
zkA7>yg0#HcvCc*mMEB@=)X}7$yVpplhHe0Dt$KFmnu=s6)H6elNwC>?@RBYRAv`eX
zeMSZb9GSW(>QQs+zNo8g7($ZR3CWbq*u^#un_7*fSJSLvyRR1y$j4>A@OIcCYz$Hc
z1kkm;pT=`U2--w<H59576cuC7#L%ORmL46~lA|L+&Mc5EXTI0$>mN3iRO2;k^}_$<
zzf1QrwMBKj8|i<wiLrY>Xjs>csv&D=!Es|WTGoQ03xU)*_4(<`uE|Y5i?!Nm4+}3J
z#gLrbX~UW|4}LI|0S1W=-YzY9%av_a*9GQC>aG|q{fmS|1>ykiSSMuW#EDO`vq?5N
zY_Ele4iwU*@X&|dw=}J%@X~qQB#Z-BlNfcUer^!hd#^r(N-Xg#;4B$8?qOM3Al}lP
zA68aOn4v)jZoc2BIk)<)=AIk~35e*lkTD!Rnnj7?#}2;L((?szl)4kMp1;Q|6gK+0
zb*{V-avz`|mNcWGZ+{u{Z+(Q{jjZQy*QdChu)IU}3^A7^ETa+w2D-J^jvI(1n=f7)
zWq@V!PFw>J<6mBQ2(zWO+|FZmy44IYMiGU62RE(!{J5y7M)o{OOy2{WsID1iY}gT=
z)R1!Nh+ss@!~20YmW(0R@?Bxjg$vhp&+qHOkW{w7Z{1Jh&p)vtuqMG<4HCe_&U4S(
z=uid}=flLutv$B#rh<ZW#psWSHXjPZhfQH*EuZB}m*O)sVdX`1$_@)1r@kLhAMCE*
zpPqH>aIk3$_jahszn;TBq|l`|3x~eI-8X}Mi)Eb^cDlGYY;%(e>{>Z-Fg<OyE9&NL
zdGa?h@gtA^WuQj6SdGb(=(Pw=T(|Z?|82U96;6;kzNO3GK<yl7BZe?mR#vDeAY<Gu
zKPFxn<|Kn-5Q#mI1j#2gFI+mm4ekM4L1UoW_IXUcB2VAiJiEWIFC;s;ntMm17u_j%
zjts&~H^sYK)XC|%+19p%#1C6J`lBMFe{e=xFU(!?m+a*gbB1HYzxwK8IZd}!3Mb(E
zYWi!b8NMSsFC3wSg~BuU<KqV|YweTbe$Bn??1MEE2cuJEu(^W50=w1INK3K9jxl;6
zKey<L!Q{-phF?9Os(*bJjOlPShakgU{YK34?R3#!z54#Z&#J-0oaAs07$<xl7VH(i
zzR%0idk=LD$K{L!5z~U!u2uT_3&lI{XexJq=Fa_cVAzJ-pqrgkjAD?hPobwFD=D<y
zcz~Rv>J#?@rx?192oJNaDDrw>c4ST17Ufda;ddEqOpjP#>CG>r?hfTID?gL2ad`&w
zW29`O_$tEHMNXr}HDA)B#GFxvMW-K&g#>Ud)t|v+RQ!xo)FOW36=rzn%9VY-5Q`C!
zk<PA+`2ATV7Vk(+aK9d)ra3YxE^f2(xPi+Ej%CCD(h0d48A+-dQSMNYQevQ&I<}ol
zw;fwz6LUvOTe<tvHFK7}xjYx7iO<4=Qeii4@DS$j*@N-)oSpBU2oCU*Zw1AjZ+VaX
z9OjB^>wYtNZkBY`y?S@Au_<2P5)tXe#(wPRST_t|{ePB+YH5Rm11fHDIn#tv&ym@Z
z-4|2uKhW(GrjRF=ip9Moq-<`gFpYWz`<_gIVcVo+lZ6>=BV%`|wgwS5Hl+uYs(xA@
zNo#t2Z&S6qy9u1o^n*OksqfYH{^3an@yNrD8#h*B4K}zh2_<zslzB$zzWl;O<A>_W
zGiE0=q;y~U(p%2=WX^3|>V0oeMP)hJm>L_CnBFpA{`$W0)NTK>T6B$Rsc`=_xemR8
z-L~e2chm>(-eJM-%-2LRQGGpDKYH%<J)ZM0pP;Z8Q^m_8v%^<8>&16eK#_eP-#%I*
z|9IZdM?Af+Qg5j)5Bb0>?blMO*X3+A)fvp-cOz^}QQrgm^uir1vT@DrCt2bT>*~Ao
z@+wZG<Szke0qPeKeo7=&ci;W5O7Ux9p~}1APyW{njBB1xx9z6krDV#0wEBhEB0D%t
zlOYp~F*xO0{MRRVS6c9kyq=Q+P^rp<O-R+(S5A)#HJFS;Rz-WhO*r0^i9x0W02MFc
zgs?IN3@Nlri7}sr4MJlay?(e}++KSk{?3e}*|*<!Cmf>1V9c~4#92ONBt2<gTYCl#
z@ocVJZ5e=X1uHM&Q}MT+{T{+@V&4E0@MBIhzonz4rPz~0c-wK4iUfRc)x=qKtb~BJ
z>2-xWQtvUhhcTjLP+<l?KIN|z8e-=eT3J;S`ikpdgxi(WJ6~E)JF&E2^B`I9JrbF-
zHHo)RUof#78TVGU<L6sXpV%zL`2@}_v`E}_U8ywGu`Dr*{1)q{?CfKazmpoG%``I6
zu*tnd*4;Mpj+q05G!g@r6AfozLd-oolrf3<4|pr&M^aAR$h6osM?f+;vGEFV;3a)-
zuI#yYzS2;T&j#u$zPH-cEhVwKiZqXqB-2V&+vEG=fq3Z-8iX@~!nx=ClOG%=Ej@a0
zUwjt?pdv1=a8MdeBoZKsuVtmBw|~eRe_%li&99m*4+jx*sq4t?sG11G&=vHIhK5%2
z&<Q}aduJVbS^Bts&VMaCR)rYhL?`K*;`s4YZXn+zw3rx_tM8ZSH-e5u!`&7ZZ~SL$
zCaK{KaQN{l4CDbkKK}dbG;i;$uN>8{Q+CH0Oio365LmE&s;cVQtP7z}naq~@8Fvyt
z8zLZ3c*C4^o$}t_ckG?N9iJd3oST|@R6QFK*_SMt3gq~_c7LnOX=&bK@k4rbNjr#9
z;nRMy1ssw*H-#%O1E0Tq$qJeq_h`CFnZMk_n*Ug34TpY_|8XSq?Y!0(DqIVREJb+a
z&70(}Cx{7<(?4BtF$L0J|N8$_JZzS$H#s9GC-*)lIGENbOd~lEE+*5qwTF%&WGBvp
z+)#*OMk~ZIIuOYv(yQu9FL8pQ?5kh)nH&?DBsh<$sVU1`TRV+4lX^&=Lxx*3F#+n?
z6khk`>sMNXPIoraRK%%CT3`JAdwYOMp}$<_f8NnG>xVPv3vloWoe+{HN?VD0iy({=
ziU8q)bDEyiq<_D4{1CT>v2EDqy-^{YyZ7#0^lfett>>P1-WjE>vu|3WPgF?Ac&Mu7
z;|3yU<5K?W(lsUmvivSzuBB!E08M|{;G87`yK{6=uELA|BGd%Nr~BjIV_fskKZ&eF
zz)#&DpW&pdKmTYU@xTAky7c{DDv>7iKa@J%St+;bhPJ%Oq`Ab$3X)9k{<{7)5?ZpC
z)*bp0`~VhBF5d1xt}%V6)MWu|<}<w)c;(MOz7Po{Qujy22jZ&xlP*Qr{`trM)*m$r
zP#5_mGKkM8cFQvQ77q`ZqAMpUa<V1&($&$KJaHnRF!dfVg3v7xw!f#(v`9REShRGk
zlH}9kgyIfcBo2x2u($m0_moCx(wL3w1%Bg{gu)nyu~rU$07rIc|9<_jJdZtaU{Tqz
z1HZ_>IHRj*ig=aUHLO<;pJ{aGcw&%3pc}2A6>R=9(*?*tSw&xqzq-1p=8DjkMY4t!
z5Qx&=`Kw59sosDLj%$6L$jdKy*<=2zlR-gd=H}6^NTE$)HZ3Cm6pjJDvhcC8x`%_c
zr&e0O%m>o$)7K$<BY6)!)>D+6Oda_FLW($p>p(f#KS!A`A{zrUx<b*&)YgbDO*y%A
zoB+wwXA-Jc#%&?0M0R$O-vy4FZC8Yuy1)8-QO#tbAsQ+GnPKQ;P@!p{adYOlF<zb?
z7h#?;j0LHGqC<IU;#h^fzs5O{*;AQPSe{px7S-edN3)9r#C>PgGfIQ*b*2KpP?>Rv
zPQD;*?}~yRJ?F1Geg6Cx@J|kg6@jW*&z>FjC45z9U#M$roHo!8xX`Xv)Fb^9&>?e1
zzv3Q6k)^u!J!}&Brz=7q0M3$LbD%ecT~E5k`9w)nw%u%8v&9}>FHRG7^fG#Vb<)r5
zDOz93@H?TASwSKG;xZ3f7U|&q;h1~rkI=qSqU2a`c7=~EEiaXr2m(nc<{fPbxW$8;
z1OPH=5)PzreVh-Daz$&WGjJm7<wvB16qFD*kWBU&J6fa|kjkUc8F#Yk`8oVrx-Sb9
zGC~Hq`_?mnsh3yT9QGBZOS!qZNw0QsNkq!%t&G89{UpQGQwBZKUlT*|PCK8mA?*`R
zyoH6y&gi`TCTeP$g+qLN2N|)qvx9q6Id%H9Q_H`PfTpD6HDxo*dtBNbaa9Per0C@O
z^u2+b=HxUnbLHW>?IN3JzA|I`bD{?ar;(#h(htQnjCoP0vyi0X2n)j&!0Vqr0Q6UU
z?XG+8DI52DnMn50^yk1k34!Xqdv_|P1);}!i(ZJn%@v3@I_nK{G#)9w^e3>x4(X1N
z6FIF0b7Ig)CohbNP(Uh^oWJn@D%{>44SoEqr^tr5DZZvrma>0;FPV8lkNmfXD$N$E
zgZq*xI&ez}D+`!fI^1H}H&BIK>Pu1|z{PLC!Ibd!AkVa@H;<konpMTJXU{4q&`XJh
zGiV70l|=5){{2KRBuraMk!y$lL%#6i^VvL;rEl)rQ#w!$r!+s1<Y4H&mtJ)f>*#3a
z7a6FB%#jSAML2TEkn!i93(id}!b?-s;exp+Xa&g_@~t$htN=*A9eiERcI=)@sCf0{
zMvMG^Z7*Zg_yxncbsm}O&CTo1(#4215l!pDMZ2$f@aWNgGDqb!gpTRB@8Cd+3K6|~
zi`*NEWA^)!vq}t*kC--U_SOQLx(D}t1Z{7VoGVHXx;1<k1yjN^V?XH3k?nrgZ|-ZJ
zK7D%PnP&{_S$0a%b{leM&+7BmzBT|EeB&95oJ6@oL(;3FqT1|r@A0z~W0o7BuiHx`
zy&8_(*m$_iBC*(J+lZN4meM-QUTV4COC&W$`ZwPjSFS7wH&Zwved5TG<y?)(w;yEy
z0tCa2US7z+A-t@D_fT}{|0y9_zm6P``bS>N^Tyh4-jKg5lZBVd6hu6)QKP~VmBNbq
zfl2iDRu`p9^YU*KB6zz(DQTQ-m5;yAwhXNk>HBxfYP#L3o~~NWXf34pjjy6#3z2@{
z5Ruaku-myV^+s#f<Y8a%sOm>kSPZ>E3Y0Qx>utDSV<^VYpK&a{sNs5Sp2OA0MbVG8
zYm27mf<+7*G)UR}o?y{<@lU!wh510_W2iZ{6;HUxf#s7HL7~`D_lknA>4@?@i{;N|
ztEwuF8#hXOPf+GtJpxsC?3!iyY_LKpA(W~F=L?yqRUZY97Fw$@;s;?+b@mZi$bao<
zf`o)H#R(@zqKya}u^}}%CZ@%2!o^!jNsbv>FMDxPds|8lRdRcO!IdrLiW?u&Q#eIB
zBj2`E=L9z)<>(IJ?(i;bgS#X((M;WZ-Sva2F>dKNYHE@YaEAVU`_2@ut6Je+7fG(h
z95=KP9W^iUTv@4j%9-|a<PO`)#t7&2=P9l2?G?8+W^h>WNX*u*Wn~ny*@%+0+4jGR
zE~%ns-I#HZ_I&<&YDC6^K-~%*<S+}}JIwF;_4lvLy;C`M0IG-vpvHK%+JjWsPx`F-
zhBEl=!QXyn0iwl+tpC4c3_>0<-Gi=thCuY|(+3SaJu)P&Ajtfec|ccT3->EhhW3Xj
z*<KW3yt&pW`LJVJC$p1z>;K%oJLJfth%my;&1ti)wgb<<-9CqDcz+DjYz%a~_cUO~
z)05*(0$<!X;G9Y$FsUa5$}H#6B2lT3eEG+)ewje^=<#EUc|jtwDaJ_g6h8LIVIc8$
z<FU#5qp)rGo(@?DWk)uz(Uofv3Mz+jODY3Kus`H1k-(Add(!JY%f&~APMw*llsX(C
z?g@j*v;|xy6%plH+1B!~FCB%Y|4i>6lWi0j5b1mLsN9_Gm*_a*l;$)!K5DMQ86999
z1RlGne^#tGjD>2-&MF-i9$PqDbDDoN*$b8kk*Gim_WJy>Wu~#2S&&YM$t{cf7_{ih
z9Xmm;3|Qrb=fSW2NBK|RxGyZDhv@U~LOyE6!}hamV%4A}@M=o!`0RPls<)pwmY_*b
z_R}q~1gi^LUh2%sGkDo^7TMM=3Vd7}J^}I{y|)W+_VEIoi${39_crD6;o<Ghg*f&d
zYmL+&-lb&|R%4oR>=xKZ%$hhZMBj%U%;jkd^+8FTbzHp{#T_;q-7=(-lHKGe_C9tn
zsL-l<GLwA%z;4J+0gbldlKP;GO3q_G#igY#4}WwWgij@3qzyvNzSx2*9;V?8<_ZsQ
z$WkHQOz-MmRod`Vp3H-ifo2dYoU8&*Cf`zPtPl&Pd{d`3T>GrP9*c$rmuqqt(&7xI
z-GAPntFfG8Vo8#>Hbx`R(V+1Suh(K)c9?ONO>f>z-|_FmYP%72^_MPy%aIURWe3@w
zjQX(R?r{2CmU(|q{)I5t{S8zEAR-i{LCU5_hlkch)88!&2dI8QbJ?qMcPeqxLFnCB
z5N=N`?THg@koS-mVLV$9$(~Q(*SIHl4-e{U7-H>?f7lck@*`9nD>VBbiephQs8?)1
zVr!&hI?ND(s^(R6Fh@6)4ncWeZ%L}(e)fA&@*m8oL=EHPiy05&Q+Vf;ys$gaKcsAJ
z60dK*H8D2E2a+B!g;FpVDs(_&IJAC!qSxx(42h3Fols2lbxy(wD%##|<HpO$^r%y@
zQx{@PE7a<M<5W}JiGso=D6xnYG<vc{hRMp>AUH<}zqQ~7e4kx|S!ysO#j6b-g4-AA
z97;5LhM49%b>hU_k=r$t$Hzfg-Z9hK@voS;7J6lCP6<PB%8hFRC!EnqgN-NjAxdCA
z70D4yIeGOG$2J3lWRYlKZ$S%Sxv)`k;V24M?A0kMV5C~w+Jwo+fsRiNd)q#IJEH2!
z$UTAaZYnPSJaPwr<}FdJ7nQ~#*PB0?5^4%8*e36E?ay!ax7ROGJjLk}8P+8StB0x6
z?7HB29z`W3#*S5jYGwU4-fQv@;{Fup1L*87uexEH39%n%qC!{7QnA1jLD9B79yDYK
z2YX0cl%d=!*e2g&^7Eb@dt_f4|I}A+^10{6UpS+}rWq)kwvm<8GzvEP+ju6chN;h)
zBmELi>e&XH-DK9St*stywIO$HGY<9((6Dwyh33i|bFh>@&KdQkeU`0Cu}oUm507%a
z#dGhX%QL*q;F#*N;HOWXG^0){pKvI{7{6sh%U(@KsyRe!r)4Y_fOB5kE@$WQSNQ3K
zxl6~&ziK{GVIY~*uDr5UZSxZI2$ti(mjoFp`!yn;V+ScH(_eUU^wp1i&(TyGCuD<h
zo#a!V80G?1zXtLG^=HW$3DLFi3|<RwgiHsW7y>ulrAt3E=}7xqy$WZ#;TLe?01fj*
z`{6RUaBY?Q#Q9~}MqWeH=sgCDP&Cr0MG6u3kw=;&JNKKsBMHpGCSt1Fqr*d}fsEp9
zPF%k}p)7~Gv22&wSrt!)3K8stLL^~pAx19*?!St!J=S{zKIChIHxC+4O=35bQY<Vj
zALYpwDY8YE>FB8QFH%_JyDeY7)2tBL35p`K8V}|lolV~6DQ+#q!>-Sn+Qolhy<3)l
zO>bkyEZ-XBR>`JEce+ICyaA;um|JmrlD-hZiJ{&<T7%gm5zQED>92EW@OpWgdg9hs
zP9?~>#iV#JV#OKCyoZQovj=_XUZ{z+SJVn0L?-JQWL5liVS7x4SI2sV5!!oF#x(R1
z<@V`5#fBrtLVf1Zp3i^|g<3?Bkh@m9am}gA6H?BC1GO=kGzA2lp#itVwSyRJceYON
z4I_96#Kw0FT^spN#z`As)eO4=r)3?xz0TY#0hw-U809%tm_3F$CV|JF@03`+{5fbX
zg2w8%tL@gF8lkp#!WBdgTMdgmQ#Mg>1q4z|^fP35d0JhgHRq>E3$HiMUZ3GE>TXxf
zbSQkSbL15U{`L#?m23Fw?5MSUz7wj7LojkKP|C^486YJ!P5~^)`CC%^z1%v=Dmtiq
zd-m#u^FcusLA7^=jbdJ8-#nwtZ4HYYZyqpp4!jYNI$Wa?R_j>Kd%GVOdWfcv=|1fi
z&wDv6aI@}m{hY~zR&UvIE!l+hm=QE|dl`Prxb-xhMH)SAqxor8nR8qO>z+O}JJyoP
zforGI!+$p^8j+F}L09#*@|ev_TBlxG@y6p*&6rb9;}fl{t>gNaQRBsbxvgaVX6GGf
zQ5!AUgd)|j9fLmm_=qCyzpxr2cCs^z)@tCfR~FBmM|L7L++t~IE4OZ*q$@*+rpzb`
zLC|CWTL(T*QNk0+ZvI6}TRXUF=kl{kFZ&5ZTK!Zyiqedh*)>U~!t1_p5k`kb>UGnI
zzT4JL$cyav^J%;)dk~qbXQ21`uM4|k$QO$DK*&2M^%Oa23Cb%AtKwJ23IahL==$Jo
z3+bh6*AK6(DAk0IA9qHcr1+FMp{rOHYk8z7bjTD#PhySIW=HwJNjjsYQ@+-QC!4Yr
zjS{Us*PD~(Y+4vLn|4EOvd&MLV<j5qHX^%2{nAe=G(m5x8JSbPwdCx!^f#nPkzo%C
zbEN84$hH@6J|3o`HoM4DM|@LTF+}a(kS%qSD0pY<XBt_*y21%Tb*^B-Ms6&^H=fcl
zZ;tZSe}<8TjlTknI6zL_z>cm);}o|Y8vADz?KrygWtnGS4clW^OtDz>Df8DMqV#t^
zcShSVoz&+KI4kV)W-zg&x+82&7`D2|mA)}sSPwZgyVzt1*0Z}+7rH?Lfus^=-3jO*
zFNTZuAn+7+RCGnvZEP)yhc*85k6ABQSs9tjNkJDJkpk7_I<)R6=~cNcf77WE=a{x-
z*k`1i-iQxLjf(4Q-~SyL?f7u{)2I^_yI&Uk<?J%`+0&<E6bN>ytyg4ZzB$vCKe9f!
zAx)V!V<SnF8nuzl`0B)~EZr>ks-{CZp<4ANEBpN>45Ny2qUGg!i7ZUch6=c&38Np+
zrQUY>aMHm#Z2$fY=8i<I^$!Y)aAh!vF`$_a8C7;RL+0(OFVXT%pC#!7hLx!d4+8`7
z^UZvFRCdO{%;gD$8$kqm<V4&r^P7Syb>X;Ca!9Qd1(crodUp?7mw8oz$LnbPwHUYc
zpqlM<+qDaxyH7IR!oBS|v{h@b<6@6I=k@g|JljC?iWIn?RGTiZ`^T2*b(s`EPr69h
z{v#Z^NtNCA)3AwGqUqKw;G#}4H9{4jA}p6j$TOrok5mkKBP=QocY&L?vYI^jFlJ8R
zB1-1>G$&2^z$m4ck=iU{a4;&u?10%vLUu#OklVs0fuy|#ZekUlcPy*_;DcJ;rBP>I
z)P~kc`&CE1LROc))W1m8c36Suv@LFy8Or19vMo1X)?3kZW9WK<Qm2orMw28aUK?f=
z`lHyQ@t{E}Z(`=Qds@Spt?Sag)LdeWlv68q0Ha<Qz)kp^A3n=QdgSU9RA(fTL+il#
zwa1QK7~af5#DIlkWSOtix&U&(zMumHiHRWj2LUor?i+Fd6rL#g8eJfD09jia8Y@DJ
z^y+2Ho=PgG(8U;d{2FAwxZuSskNQIrs?x8d^%``_?>>?Eb7bdj_k_r0rXgRn3bVI7
zRJ7b}aOC%k+4m17NsUyEcr^FV0{zvizBNjM6`mcaR=8vRh7D1(%NCXGv+ol2&QZ5O
zr)<G9p3j!%H-rE4U^g-`p#?NhT#%dF%$U6JM_*B)ugjSi`dINaQ&07lEgQ}`l&o>=
zWy9?$O2Vf=cQg&@?_Zo3N*BD8q|w!{$@x(}_^gOl{pF6oL;tLIaZ#P!BAd7YQy}-}
z+_0-cEMP^3&(S?*sU~IDvhq{6h@HrDU4F6AU2LB>?M!-HLMEZ?Q=+yYgaN0!C?DO~
zaR8vk^qJPwsd|nFC_@svcg)0L<-SSF-kT};i668Jcz5z|!ZA>7O;%L@yo1J4BJ!((
zj&%y9WHHo{p{5%Q48&qe&gA_4VFr`ioo<8zg?7{y)}hEQDlY^CEs-o*StzBNB-m@Z
z|6!%q^(Dufx9GRjbwz!sCk^q*uRr5i4q#inr?^+%V(q96k=xcRY!6+gcq(Rm{@tEq
zRVR6bGX1&u$(Y2am;TuAdm}#HplSWsfqn=bs6>FRvUUF2X@2X_CIr<b16G;{g*;m(
zqB~D*QT-5fB;>~y)z)Th(R^#vXd;U}{3bP!vCCJ+YR_vc&`<utG_2iB5(l7MIKs{$
zEG9diP!;*dXY>(N1{N0_d35VRMOTQ~CdvAypo?+ClJkx28u$95Xt}SRCD<B922}#C
zPce)beT5?|_4Tc`XE{0Mw=DKnuSRicPF^Q8=6rp3YNdJ&%Pw(Q8&&IOyB&1M>bEBK
z;dHkznVGt4xc{FrfleI%VkQlaKck~DuBgb3RHQghB0A;cGgKF|)u=VsZP<+|CB?~$
zi}M19BsSd3{aAVE_t-@rsYXLKI0p$HaEO^phhK|1W_S9C!|Ad=c@~edsnFKB8fuF~
zr_z-^2}LctlqG}iX(zgLxJUF~6op8lh<<bs!ErXFb!|XaU$v8C`bp(>2)^rI+6gy%
z7_!VGf=y#i0v;Y1bKgb=Na~5K+?t-)Gdk43AiAG8xJU(IYw0843y?PYPkn%H(a_9H
z*!D<uZ@GPKZwRl@_Nalr--$v491~=X{hpV%lfI)<ih7fkm5(1j9DrP~E?<MRlO<WK
zyBZ<sZj}LzTVoOu$^jP1qESH<v#fz`aC(c8kK>!hzWBTM=6~MS7M3a&-W`NR&X6I_
zHMy7h5l{%Ahr67RAR)M)5E?Q2-<DBYEqC8nPnaGL`9VVs&9H6191VR?xm_^yWu4<I
z^6>FvApUJA_(6`~G|&No`m4&Ml5i7St=ddI%xi|o%FY3C0%;axu223RgG4V6$PUI4
zK*f_LZu#T1JGf#6r26ZBEZZtoTT_krhE0NkXjU@)K)TV7&hyGHJ?CHo^svzPuB0%4
zLZGBaA=Zt<-|L_ZEcB8*_Z-`lu<BI}V0f`#37{+V2d#p|MLoB*(adgD9SLekMZ80~
zda8K~7L7usI=7C~A~tE%+@0dl(lw2>7=qCUZql^1aEEagIqC=(xzfCAwf);!CTn(0
z07W1j&M(WZU7dxft@dj~=AM#ROOyKrMMbgC&`%YVUb}dxo@`+24sGTbRV@%ZkvL_2
zY&X}jqOKHw!sX(F!^~$TVN)DSVi&EC3~R^RMc7>X4U0p}*fWMzhULTbFrgetplp<f
zLILAbajrEeUu?7X@5P!?mBUiQhB`EuU#xa~eC(bPs?JaEHF}HE<GZsLwL4}P{r#UT
zPC{f?T2j*1SpJPmGSmGA7M#wfb*d<-yuY_BiR_>c%Nxd+H>fEeO-qY_L1LiCu+0UR
z-6(re>7w|inkj4phWH3~4eHhBz=B_YJ%qMTH(=HW(jfO<-w*|$etr731wI>Z;;=g(
zfCq%)1nn70&(u^em~CKR@DM3YOXr^Er0J}Fi~Bt}T;z{W7~B{6AWYQDf|32S9m)o}
zt+fk?KmW@brJ>5kv4S$FqT*uVj&6sZVSeKJ;PvD7u1ol~W+sZ&Yl@+R)Ry&(UK55D
zEu@k<8cUiaXcdu{*JADQLTZktt?gK(vIj<w?SR<;SEu0)TIQ78YDb0g($PoV%z7!i
zt4#gpW}4>IiofJsV~X8}LJdK>=&AlLn*aXa8vz%;f_h$({EC$}2<}QTv~TU;j%fD%
zQl+84q7oRE)K!P!EWEkr=jM$FX+0K?xHRQ6gX##7Zx<aI$~MRr`s=<AQ;dP^tw~Nv
zO=Z^6j|U|hGE}Fis?YiI5H{8tiGElUmMC93&3^k2SV$xq6sX}EMZ(6TwVZiAY;%v%
z3hH~lyWx6+U+b%XoqWhC`g9Y<{{Dv!Wl|K9rxT@KpI09$QV?o$!jNE(2&3ntjX0HL
zFtmf)=|6nt7O*r>tC`M9&8{7AGHkvU%1z*8Rc9s8?0o_U{6c6L;$Q;@+eUD#HA=U+
zZPv)ih#w73r?$05Z<_FeaRoaiol@zBOQ%cD<1&0;um}9#liB|dy!U_KkM8Xnu-~xR
Vd{4)uwZb=EwrGVoYJtVR{{hKz_H_UN

diff --git a/01-exploratorio_files/figure-html/unnamed-chunk-31-1.png b/01-exploratorio_files/figure-html/unnamed-chunk-31-1.png
index 233485f7a35560fce8d6039ebfa2ab6101d7d36f..7fe56708cd45757d495131db4065f64e6b80210c 100644
GIT binary patch
delta 163144
zcmaf)bx>Aa`|mgMAc%y3ppw!eB}hn1cS*NMgMf6iQIHf6P>@aq>Fx#z>FzGM>4uv+
z3!nFQ&U<Fg%-M4@!ygRy-g{l^itqQc_Rld#t}qMnBw<G>DVN~Dz$Yn>AKxnfmi2CH
zTcjZGK}`)w$SbTy=#DrF>b<6y77u5zNsJnV!Ww2~B1EN=ciM$kJ(3QR%t#iK61ny3
z(R?UaXFLKiI3J2;j`ccEl8}<4L91<-+X>UsoAV!eQ;Dvgm9!0V`-@PbLPbBc379?|
z8B~$uVqL?hi}J?1Q-2aJsCetRm=Lm>!!c3Z<O$>T)03s|w<hpJFev@JS#uAov*U;-
z+v+lR1TXrE1#pvYlU2PtM;4J?BF$Y?-(<CztgNykJ+5oQuWde(5#c#(T&diRx{ATE
zj@V`xXsv>awQn2kMYo6@33e#$jP7k|*14b}WcJ}$9<<@y>9$*g#r7yK^|b0I@?G8B
zB(#Xbi#cxl41Y@44L3**@O0gyV{gbZKI@Gtm{#bJSrB6JyC;HC#;luCmZM#NbWid~
zK}xzWq4u588_cpx6_G??H}R1#;_>4-F41E<>02>xDWjW3ND+u{InGIsZ3b|M$ErLm
zGe4)mFD2<edZRkg@<38@U6#A_4du~z$q6VBu{_c4@Fu^dv%Z}LPmZ{n@~GkTvzQi=
z@VEN6AQM8h0-@Hc)#1Od2hnt^UKSI>?PPpM%}6ptwyT`1zCY<KrFZGjDg1Q#ur18{
zGQx4MT*FB6T2qeB;d1sl?xhISm=Yd61;^@9iG1iu`dIDv-m45HrM4K+o9=^@6iBM?
zW_^m{gW?rWT*hH?mM<Kok?$y|a{c+^UZ23MOcx_8S`C)wo``%i_;!1b?kXMDfE4Z>
zhl9IQ^pA$!p2f=H?wERW#a%9(p;`;M|LWm`phNec9!WDZ<5+IPg!1Gjso%3<42WKj
z{GQJ5A~9H&-+lrW`m&s1rch{|7FAz249&|guZ{0?k_JqB`39k*9(i2W=l*?WVt<GE
znhZsQrFfjI9HTZ**}}4C7RjDU^brP{39(Ugj&k<erWS7ExOlo8=b3e8Hu?v<Xb>6p
zSC6%mz@vL-b(cd{j~sdRU?rbxMSbBy)IFsqQ_<I)wnGt%gZn>ooE0gIOfu#|#OTDm
z!dy<U^HT><UU#uede{Hp@{4C{I(F+o=dE*|PP4{Y@_%$Z5E*^H>o?=!-Lrf5$+00{
zEO!EN!JE)kiobb1PZ<kNnc1@98FI`A9?uT4GtI-i&c7}<<ym2<UsbIZwkybR4R51r
zNf#8__aI9ak(WwUS5@omlKT1dH;+OgnKkD(k*hu>I{7XO!mlGN?d4i=4__o#UiGtP
z<fcW&#7N3Z&lZLsSAQJZD2P#4S9a@ZIT@MybKlb0&LXz>!>8!%K`KeUArvo|Emqxe
z_VQEVv&lM}?d|7q#rokSUpVSE|CI|<T~=WKLTE_aR2=7wZ<)pQhyz_>)lpV?wB4FA
z)NbzSMRsI`&OHpT{JSgpnC5F})|2qj#A)T*JEWw`71^^6K3g{RjvKQ}SDTQSyLdd<
z=I6EVB!q^)uGFz?TWy;BG<i0<&Fgf%E_>YOci&P3#^ij)(fl^0zP)Oumj8ZONw$?a
z!2{dSv*?!>Ulj^M6&1o272rQ7zqh(s9G1LuI|9E|##)`+kBKhQ_NMuMEhPgJlM0&s
z{v)areJR7P>ogn6{Ec_g{oS?INm9erz4+MpKI0Dwr5PWKu7YEizAk6z47U8B#aYTl
zH?TjtOo9`VB2-?c4}4wuKJsPxM`#(b;gzIqe<0Z1smEe(*yAUtA2^ASoFvt6Z_kMN
zq(Il<+_?XsolY@-HzmA(A4EBOEE7Y}98+nk;ZU?!nCx4s#_^QCp(*;Q#le-86%b+~
zX;Ir5zlv~cyDrV*arGg{1TD5*1LtXn45#xVd^Uyal3a}fJ;f}3P8Zg%r2IJIF^@pk
z4E-}UPE?4C$$7ePJc3Yh`n`tQ7ulYUQB`H#W%DsbjhxET?jF8I`p5Sn<L@+7$+~N)
zO3>qi!tpjcc^lyfNGRDiF;rPdPZ;H@b@&r_0MnIK)W#K5l6G8fTRrw^=h@QY;^H_s
zju^W4;KABL6DwRp$(yME>J=3wrEuyyZ_1C=D8!-a+S-_q$jwzbD0!l)r1Y4EhUQ7&
zMxnFf+1VNNGF6Fp%Zzzga930_rnJ$nInP7U^3kWZTc%UtTdE^`mfW(Y3i@T2Qc);T
zlgzOST|?aFU8y0hMuOzCbP~O&K5$oECX|xTbbT>u$2I_Gl6NO_*_?KlNn6C)QoR|!
z>ndU?ah!7HwVg1@@bu!qx43MgQH@V=+JiVD%HWWip)S0f82YC^sDt^*EuWkKuBkYZ
zB86mj_pGYR{cHsBT*|X!;`kin$}Q5HT!rCe|H7<5c5pz|X|}MkDE+n<>~5{aSwS`=
zoY+xnS56>_ZM~V*;)h2d!R3C4yElG)R6Ua!D@Ugc(f=^n<nt&gdd*u`9^LdgCL)f{
z3Cp7i3u+d*Mbvg1`YhA)6Q+ul<1Msc$Z@CdPT-Gp{hW{_&L4Q0^bf>;|M#;Jl+B1>
z`$0xi#c~Rx@7Yhr5XO}V_&@jM?TTKXqs?m-ocm=tjQT-|!(Gi#{`<k;Ry?>^2=Lkl
z?BFNtv5pIktNMtiW_sId>S`sE;(lmjERMQ7w%X{cqtb7AUF-A1ctClZOzkmRwUO%T
zu0m|{O<ERCyu3s<)z<PWE9P^Lsu9R4KSUf+Z5NfOM909C<j)q?rfFl7@FAtH*VAr)
zxXpf>AI~FQj;iGSb9(>obfvYiiMy`2xZyP}J6x`Rnm^tR4i4UAZfa`s_V#9AV7R!r
zKp+q}I5>44C#x@Vs;iN`iF~$m^~}u7Bg4a+a}5o3bzTn=cepJM3yZi`3+5!@vm{MD
zQI*ITxQLx3>+w_?TC5Uxn`r7q-P_T$M^XA8CFwGhq-kEUVw-d|p}t#j6VW3Z$}+GW
zQ@O^tdne(-?C8t)saUzyIK52$>SRJ9d~>8OT~ut8!I$XRsLvu^M=X=vPjZW-<kfy5
zLyuy3OG?P>5p8w9vk~&>ER}-ZY;C`XyQSt*DquQ;sG7L4XR!wB|5*u<#5$k8-d-E&
zhrUfOJWt=by1sw{gM-!8)WpTby)I7eju0bUTwJ+@;Hd>Xj%n!WQGLFB8;@Z%sP{bE
zUmwb3)+q;nUp}B>uHxxg%eC4DK4gg)2**~Bz<Kkt+m8+h>;<;uh7qQyM{BSmiAPTh
zx4`qUv*j{s9Ssgw8BhCakSTTqz2&98tB>_-+Y>HsyOeOK-q#JYo~pdo#pFC68uL>$
zAb@;gV%l3@y5yEkd{GwN*pZ#vH`X5DxhXsM5Z%=hDDdVun~8}@nCQjDHOt`Y_BJ^=
zISVy)iQ9pxyu3WLdvsI>+pqlo74iTZ3o9eTJD=5Su{CIU*;qiJ?r397R$AJ`)O35Q
zV&`D3Z!BL=r^eaZLxA2-Nz$IDx-6ih!9lLPAdcW6afC=u{bi$V_p>3xAE+-D`{ukW
zce>>z^vIyyeWgPuj$yU?rK*IX$aQ2utwYsyZc|RFrvaNRf$!J}+>7ofaT-vIixd94
zmoa0=N|gEy_=Foc32oJmPQHDe6Sa_iP_ClU?s>+7ly6<TXvW7g6mAjJPS2AQ^Lv+Q
ze6y`7w;e-o>1%0ey-XQM&Ew?cR7~PeReIuavi&vcyoxR=F>z+Bz(7<~w57SZEt~=u
zva_=*Dk`F)qJniQY-qPD7@s`ZoUU?MAN&r2`nhTM=jgVkH4G=X`Fci38ng(*LRNGv
zv%ihm6OR+T*6sYgyej(xHQ!8efsP`N(M$(>yRWakofo-pJ!~r-6cb}sIx;_??u+vH
zV;mXz(b+^-#TDhw!Cl$tw><mED>ojCPIdaB=H4j<c=Kdgt){m%)3{LEMsW9fZMbJ<
zV|6mT`0bD^_1}#3Jkg!d9VB8+%_l4@vtma$5QsD;R#t8<E>pD~Kk*PEGG0gX`Nrgv
znC9kYsIj9X(5tjIH#avwpCzi(^ZeMFgdf#U>A|xOCEL`N7f{FVo;kZi%yY^<dSR#$
zZg2P4X-<lWrrSh&4h@fTRev?Qt6XCLQ>*INcavNqPFkM!z4D4MTcL6<TIh?Ks()>`
zjowY2m8fscl+&vUsdK0Q@<lCo!fU&C$%m_s_ixw3J@oVLX+i=iT~w!X^w`Pv3?KB6
z-}RYfO-4orxG$^Jn$}h^=;1RT*uj%&K!j8r<1)59|65*a8U!UNen@I?Z?~j(c2kOu
z`6-l{*lFy2fqxw`mRRYU6EV}gG8H}|WNa~*?5d=$KT!>`Xf5-z*zQnt))+NaPNI2F
zgu1i7;S{aUO=JxBI3jK(JrrVPVcA|;DYG2Owix=sY*2sE5kW02Ee&rPsy-!zGQNLT
zdG*R>uD;Hw9mmXU4Mc?<lTS=+>>VigY?nz)R^s*=HkDt0vM30KWg)8$V|?hMZbw(S
zv#(|p@e~vK4DwEtK`Xgp`hW{*4RWsw7s|@9y7Y4OBEyDUFDdD`!VdVvxOX13qEcVZ
z=E;K8Iwb<u9{yu1@5QaHrsr_&CiP#p%(E+oc6GcpRbe}~zq@<*XTVqH`>UVBg@uJ~
z2dj9`r%+N#OYcMV@b&h>g0hN=D2If&_!$z(qEr5MywFfgR$Lqn8aA-j7ZdvjddgT;
ziLIups;Z$e=GYxea?)EpjSh{K)Sx3aU=BPUqY)uIHak$FVjho0ul0-gr?0B2s}&?z
zHCWCL&&~2^e3PSthDSy~zr#WDshVn$dAoFnHT2k_>A1OB#9bcs#E<i<v8f{Q`r;9P
zL80r+wn}?B3#w1kZC;j>q`~#FC?Z=eWK3do_8SYgU*X1xL`l{`c88DuoH6FWmiGH0
z)8yx|*!|S%Ce-=?+1c4hR&nKlzUG?T+)pofn??`8?s0VA`c_3k7h2fAlSfHQ869PC
z{~^&4QWOiq`ZFb~ul)&QmFrPHuPxDDEtHbs=g@N2=2v?@{#H(N*{q$7h<Wo~neXD_
zVweI-SswU?>}-klUnwl1oyM@bs_MI=SRyALtlQ}m*OeZJf10{NblRMC^iq5t;NfZ2
zxtB#m;Njvbii>}&sX1S24=*rin9WqA0L2Y@9DEESWqp0UCyrBAPp_`9u)n>14b%X$
z-h_eCQT11^Zb7_>IM9a=2^-{h@7yWkwMeb5<{dSgh&Jc_=JZ8?31+N#%vj+gM|Ctd
zWOMdWJ5kgF!5^Da-hVb{Ko>Qk8JV*x=YGb*`!VkFV#GtI1SQ=+?i<NE-?&!I`9HMZ
zO>FpGg9xRmB1{!ZIXQ=Q_WBqg&5E^zpN7ed=v?n~v#!4LkjIm(v)VuO7+GSdi(WW8
z{Ju66v$<ETot^2DPXk9u<-X331UWX^%FR<+OT2{b*Y}mgYqQnx54MsO0k(Gk`Prk$
zS9!2oUS5uhj-#rqY#8eCfQU%H@U^hAva*y^U}ompXr9i(^%UpN?CkBCni!E($R{mr
zaqzn=<I|`5b?z$&M77<L1ayy(keP)g|2psao&c@JIouf4xoo=ASYqWa$}p;Y{H!3e
z3A8sS_wcTEm)*NqhY_TcC>t*F0cIa2G4_aDu1;das5qlDAfdVXqr0Yga|Rmi(;nMl
zMZ&fWcZ$du&9FJyDi8jMzj9aiRu1~&H=c4;{hS_WY(y2sM=Q!-2nQo4{|7^djGd#-
z_0vt}T)5q8Z6ci5a?$?UO+-dQU7s>kLF3C;b>+{oFPc5a=dVZ;-^F8TBy#5wUi(dq
z9oT;I`)0m&9ed^gT7OTz>Z1I89pW<L`d>37-q8|%3VMw>f3GoH3Lbl7DJcvqx>_Hf
zmu_xn-`I=-5s4<lS;__abuv;?kFTCGFlfQLy1M%My#pizRFq9sPgDuC@tF><^Rio8
zTiuU0C&XxQ`}$no3w(N!2p-$ZrX&zv0KIs_cwYf1S+Ch+joGzc6<eR=ijaM;txZtK
zbU0G{DKI0jH68hkBqS+zhr)qFfmq<};2%TmHj0UIL`S*WCbM+G4(F&_Uy?uxHC($<
zfxAB85Ljd=JGjL4$cE2WL`&-t+bd}~xlczrITq2rSh%|TZ+Y7see?saokjlncSc}g
z>D^0aK#-er5VNf)f$L90jYXPGdP>SmDXFY0bCqF%<+U|b=%dJbX>ys;`JdsCaZU+!
zbs`A!?%g2PG-G39DCO$S5iI7RuTPFg$HJnJ@C#3(TjrPIXAH_ikM)hLld+MvR4=7I
zQ&fDAylH7Ht9o@BHPqFSr>FdG`x#dpxQplXgoK2P$LwqbNk%<xP)Ao6DujlHMkRNX
z{?gKUxCO2b*9jmF*BlBkiswW~Wl-xx2V!!&vv#=kkn&iNM#8>-i^Qu$7mT`;l$6X&
zGim9dgoNqF_ox%NXz1vuK4HCm-)AJl$+vcPcGlLo?U$w4&*lnfZr!>Cz0^5hqk2b2
zM;A^m05r6$pFbDY*64|4Wo3<wjLuF^{c*@9aTK#@xi$q?7S~1jSr`~&5LR~~zA<Lq
zy2kf}1Vni7!Mw#gPkU`odv|e@uOD*1WEVGol1^_M$4dQ7a`|di&2{)YAFiPHmA<xq
zYTDx|1I4u+6VMD2P9Yyr`dw*ym8$L}UYZ?kmk34v+}wJT*6fk6D*q^GqwjC`V3&G8
zv`d3;u{mCppPx^!m^58z===UX3Iu9_2MEe@_&g*yn3tCqKO!zJuDZH<uHK8!ez~LA
zq|5#Ak8tMD&Rhe<!-pMxeSMf4QxyeaVZA3ibBf7=Tg%I0sY;fXmN?}6Qkg^6*47_C
z#v8o-`8><ynYuNfeIK_^9h)GJ0Qfv!p1Eh_d#-`A$HgIV)L#`Z$uhnAVr8LZ-XjJV
z*2Y3&9~j|@WC<gA5$`L)>pdS_(k(jJ2;1S}QU2TJ1?rvLERvs6m(%Pma_EObqPCja
z(e6SE8J{!gv@tO;@d+kDbDo%x5XiITr<S8`Jz2`Y!NI}ABqt-2U0C?J>GkfxL4j^n
z<hN&v@9ga-EA5sx#tNvZsnuXr)yT+5bB?$VADV034tN*OI5&5XM*_+Vc-<F{BkooJ
zRkJ-n;2&uJS|h&JaVjKhR*i7QbTWKENtq@>s?#FYCj0{b)u5=5C$HUv)6)I~sh7I|
zwv1fr@t~3xhf$<3#a3EmaSd*g+j+n+`@^mZQsn)43A!cAXpGcy5jr?3R~|c0ey>L4
zrrAq182owH74svz?QCtm*Etf2q@$tv5*=-6Wp#oWu|giMzj^a!GeSm21{9dR-Q7SS
z+%hmItEvtU4%)waH?KLNps#;5{8I%ic0@$PojZ5H!<m|zg3&C|l1H6AR_v`K8bpI1
zWF&Rr#BB>HX~4=@yD#YJo;F(=oR0Y!s>INJH?8%i&};BfluL~xK5`JK(NJApdtS!I
ztzUjNFEHnmJAIMCksTdpKPAZMr!<w5V<geqChm2mudhr?OUczLWuQZLZSimKN|9LY
zL;sNE=He!yV_>LyJ5e0%m#15`@3K6CVtzQO4c`_OR(G#&aXm(mL4F_qVMfQ6lIf9>
z@5_d%g2Z?$ooK-@{X&6Mdb(Rtx~ugTd20UlWu`63apGmfm)==(Yp(f@Q9>|QlYJ$;
ziIIEA&`QSn0==5skf=Y_`b>?>)Ut0}b9>hTLt|c1I-2;eP%?I|Kj~w#zUj)+`u?zM
zW7TgVk*u6Mm>~&Er7`+lS!3j!rrWoZHv-h|8<X(k{i_>LP806leZcaMPsjt-Hz;$D
zK);z#+dTYf^K`jX!!SaSRiB&a=g%Li4R~vH)WZT_qBZiJ1+^O7>6e!u@NDroTsO};
z_hSD2Uzg5$uQ3D`cc~e+`~zSYSL3%L>}AzIe$ZylL)w1TMFHLg`*&sIFkBBPS9L_P
z^P??mzW*I^{3dS^?UIFeF2BJ#=XuaK(_X-iK!71IK+-@%Rz+M{R&S!RVfJ9Ykc3<i
zJ4xKe)4|nGN#g<eS=Ge1!j5q|%u~Pn*)@u<yDR*PgP(#Kgi-<t54BAk-dh*T@u2$q
z9gVl@x4?vb57++SZhTmGFEa$;EvyPLHn>q-_6>AF{W`MN=W%B(-Jed+&NL-`MX9xu
zSwymc_W9QF@^19>hqT*M8;Eb`eKWl9xv<_$!yJvE$Jz0zTKMyt(Mc>eq_x)HFG@A=
z=Qf=RPJ=J{XwwDTX*uxJb+x18BCw&~!9A>OD+`ympwm-V3S9wy%Wsc7z|>Rr$=|Gz
z%CEYDoT|B}EsztxWbH7tgu4z%?2F2VcI(sF=Pgu5uww8{6o~wB0Z)>riNj7){;uJC
z*G)k`oqORVv08!up&?J!gkXvNKFUjRVFg%1CUWUsROhGQP%E!$j8vAK0-QB+f}{15
z(cHOHX3C@D9BxT!D?3>r_Ab4@yICgiH_8y2d5=UWz%>2%aewtjPwewdCwga|$VCB=
z8LHOT;}dnB8Lb?vcyPUYyR5FS?1sqZW3#FY2;6OMpc~!sTl%4HhvyAJk0DrEOjxzr
zKYOujx1;n^ZbGey);etEN*ucFxyi2F_#Ck-$S(p%Go<G!=%0Y`nB=aK+7}Z36x+*V
z4@KV~G)`h5cZ4D@X0j-3_&)rtWelZ1d>X)%&=b$C<5DM->4WYAClHByB9B)*-W<kI
zsH;S4Vj&8~WJQG)e!!I(96SOWYOcn@%6Fh$r6+60wvpQmP2McHxM<MRr#a+jb%P_X
zHYlGmu~Kt3+xI)U*b>({4}QSji0R(U%Tw2Vb$lebzfyx6HN+0~FXPi3`+ZC={n~PN
zvM0fJl_(F*lvE;hDqbTOLR?7re~YG)&Q>SH1iZTLU#o02<%7dRWx=k#S+cCCM)A!_
zC0sm1+ZdyEmYF&PGEgEtp!V|N{xP-Wr=Q9~W|`us=j-;RCqpB2<V?B^=#X>1PNn;M
z-@KKSGn}P@8V{fc>l9lEF@6)HVvgl;%UbtoP=AZ=<F?_9%}F-zwDOKbu;SYO07tvw
z`j$rKrmg_&v2NgO{RLM9CeHjgBu(?ec_PthJj@k$K&98p9|tivuLaDOo?~5Y{WFhO
zVt-?+A(s>F0u+n4Z{K=*Je`UBmesM1ElYICIq++U9uzzVPU^35v<$PNqH5BjYIa|~
zuJ20xsGrjQwy9au5B`L{;N>fk+{}y@1KzIiIbtE&mu1fFVRAyWjuOp~O7ok~3yY7U
z4xP%N*PY#+<g={VmX=>u;tlX|DTfntb)I?sHfO|5ZVWf@1}b=$3}*vLf}`S|Y0cM&
zNOP{+e^X2uF1|+m=O`%}sM2+BF_Knxl8JWs=tAm;|L9Z@@*v!pd*?Cq(AJ783vRFh
zy;4*sO;ji7E1}ne1r_?NWygwyWr_Ry_PmKKl<$x9W9;^}sn-a`;=|l_yu`(YiJ;%V
ze}mc&z0}oB2tb##brN;tt;@&|6A}3!V)42e4S_J{(6+E(fH3ad!Mnyw-_HEo7fLB9
z6{ZsiTT=-MQ9w`bV|tGw;g*GfXNM$*uQU+f^IHc_X-=7y8*v|vyNqTmfKE;fg_WVR
z%1=-vOH_`+%wd1BZgG$uqMO=3xF!akNY?DX`lOl9R0pMBKZT5l+~Jo4VsgyyO$a@4
z;p6#Llda@`>w+Z~(QcM(+8F!O97}CM6u&7~zpH!H%&`2n!FWzye)p^tRYaRsynr1K
zRS3cS-+#*+<xD1GprX>!*Eg4>A(p7C=G4=Jex>+fj_$k3NWVFd>`@O(AMkt{8Bkgc
z946<=Ny|pIErb%aT{<V$pLJZzYl#!GvWCZ{y}cIH-dLT=Vd%V+;jaL)xt93Ikw>4%
ztQRcP*}=fVtPveEs;Z{>!Yda{ZU$JZTv6(+)UE8UKC_JBB@xLDl5h}fK`xDYgNoYQ
zc?yjG<e;shK(a&yTnvoe-CftiKgxD?Y-2oW`RmfVQ{p;dD2=J|AtVW?*UXeo$Ns*L
zCeBFdBc(Dp9q4U!7aM1#?k_($&V}pVjq%a31TEdyC;vSYHPsV}Eay#{@&uW*TN_TP
zEUmpFNEUEwZ3+$1ChNAQ%L}xnt^^I4f5^C#$A5mgyz>sPW*hrKk?23$f_{2+?u3v(
z)}z75*SCH5_E%(aadBB$3xGB?Y8+5q(7ag8(3Kbz=Un<Vs4Ai!7Q>DBR!O6goHAG(
z9K~JxO>Xta_b^g3vf4kO8l<OEHM~QPd)ILW%XSHP`g+}JSugvf)|)BBU*C8z*O$l~
zzV<Y^wmA|lJ{)X@ySoSS6jmm~ijqn`LTsEvzq(i%E9OU0;ga|6OV-`mc+<feKb#6^
z{=Pm?v=;JxU;{A1maxgu9e{KR<WV41b@lX^2>BTxsH3B}w;#krM|Ec0^%$L)_=X=!
zAo-2;8<*)Q64frqVr_@NoI*fVMeY4%?(ip40^iw+a9cvctFx`*vY(VfbFt@rB0$_&
zjQR;??Yt0hqbY@1(;wbIm9=o))@PL5E_reb*l@kDJiB_;qMuJ)zE|xXqkimYUF6?1
z1B%g_3(K1u@r<6H9_YBTvRw{qy+zqB<m5>UEdd7Ajtq>98{GOhPI~KFZrl{7sEz#R
z1G#ZgDNpE~j1TbfgTx9<DqYX)5~S#&a4zaTtA!GnDv&DhOZ_mdU;&H#rhy(tg>XNz
zcb5uD?(WAD{P<L?s{F-})1L5Vsl93cZi9OwdcC&a+E4eEBwbuwfD14@ta|LMudlD8
zqvPRmeuseWaC72djMc9?w6X9BZygat#21zWJEnLS2u@}jcxTjxT(#MC6ozxP3phhc
zkx_Ky!v_7SANQtLsfS<{QV1AEHQ~2R70H{(YrPj#!-lFbZ#F=~|9%abM=tY1@Kvo*
zdl*m>upd5~2hD#jA~7Z=#&v#u{p@gk=<Hz4=qT(3O{N)&kDQ{NiQD>ZRL9}3DDxh#
zh$TIIml`{lZ<iOYJ_S#``Bn}{PbjYs78ujlU~9px{l`?te4z@qySn)ws(La;{wqL`
zI2Yj*Lj3&v(=#&--j}~;d;9u?-n3$X=iXatCy=Ese(L7;B15SJ<#9!drs1<SW^?C5
zZ3(dN&mmU}X}f%u;bvEx17(fq20PcKAHFdv?Dm3QvKNQUco5(W<dE<Yxe=%N{*$|G
z%7&v_x@ZpZ(m4jw!`O0~m1lZ~8qhoK7hX$)kL1N{1U%QP*9f4Fuw9#(G8<RH3Hwug
z$n&sjx|)1azA-*9*2)F7aY{0VWSSZOZ#msq%=h8u;P@04W@?Lf|2`?F<(Im;y2Ql9
ztgJt_J=jh<ZXaox;ZHPe0VSV_>&9jaiP|ctDDEzK7fdVO_7^{E2*Sd-{uX5*mO17x
z;f)h|;kk&sjN;SKj3+McwRC${5nGD<DmK#@8?PTFp>3JN=;BEAH*mA8?KB!ovRw=f
z7uby2#;2yj$@qYLUAbOzZ!sn9F~KEgDol}=Ifm=#$O1!oJpn*3A6AM|7L|t{(^9Wl
z5674l2-dn1L>Q<RzdaAS7p_wL3V-@Gcr|7ow8{Y~Lsr*<@2(cUU((n#EdINf^<)^=
z8gWUu-nM;uL_n|5iXbcd5S@Js$}m|SHgFD8(r7mx9$`Zuc+CX=LKWmiN4&!F%A-wi
zFLHH2Ub2D(&8Ijx`|}A~9#9gadXHt&!uGikApwf7CC%RdojxatT+sM|8pC&7UtbTt
z<IC5tdZp&d@7}#5i$)J5xpcR>_0rI8Vkl+@LBX7pHAqqSa^fvAUxk~`OfV|u%81?W
zEI2U}fv)rL<osIe>wmrnmPtF#OnWSwdsMhF9NN4%7^Z;!ZwaT|#ykZoKCoVNG&S#`
za)yM20O82e+Bz>cx7Pa-+0oHaE{*TR-O_`O(?4|E2m740{_I|tcUE-MsmJxNv`BX7
zU1`a7!r9@7GP)0UM6yu>y`AIAHh~!b!|@6Y#+y(tse>c4E|2cv&90O}P_WGJ1V#Ow
zKaO#)G}^lvnYRBh{eAN4XWOfdzzhP$7MR~mRG*Htw6yf}gg%*k{`@)E9^Swdv$o!P
z^gK=*>k%dNI6|r>adm=uTK$#!-q@s<bNLU{^W#^MQ3|Q1HSXwFQa^4rHm@5YY!_yw
zZ}^*m<FlL-An;bA>5y&aVxOB!O9z{+5emXbf=@1}A<(MI4sZRR<^RTSnH8WopW`x+
zR4+W#K7qwHXnR(rHp0nrk}PlsQ&z4gD|G;jZa~~-y0WSb^OrHoynp|zE|7y+m%SvM
z>r<eH^0>t>1L_W&k`8cpqXsd@&*-K~2m8#ts(}NIoiIJ;5LvmID$^@~OsLDb4zEdi
z7c8k{-C-CG&VK6lh(S%9r~OWOdfsi7j)b`z&jGL{UPoL&qt(;Z{*n7Q{d;z4$3U;+
z%eiR&q2Jk-0bj3cuMb1z=n`kPINXMnj?;Bb{%zHjE8$s*u#}99Qm~-FacggHuK`z8
zmDl+(Hnh031a@oCHxh0eFz9P)YJT1u!p%1GNh)uTO|&{m=;>x~{d2OKI!7`}i66va
z^da;ht}2m>!b+uO#DPA!>g=<ASLNmIudM*q@tVQ0flBW`wxa>wf0G>HUX~7TsmU4K
zZ^39~UK)ztiw+iT$!dTl$zVyvG;b;@BC*{vNLDUXR?f-m;?35DMSL{puJtC6?-4U*
zpsocrjP{#Tyjc`%Z9Qx7hMg;-0K$UIoc5lt*lTnPGa&|Py{txI21Z8Ku~B}E`z(Tu
z75a=MYsb!pn-%w?(-SH>o7G8J0N6~|rjcK+s2MIT-kl~eq~ls^jUGPFOC2!f+w!_z
zCjl}*y%(X!QLi=K-}FM3^%adjq`dq%4Cp_b<9q8v^xWLspe}v+a)0qNs0rYzt*x!A
zy99Nir@Q;zN8{Gs=`7p+M5DEBejKdxZ%MH&#viJSYxARj!*2qLyh=xS-2~%fcqg^u
z^&OASlD;HGL;)k`E1~f)U_!zH6Jpr9m)@&(lk%v@orEl9>^F!M_*}0P-dMI-#GI25
z@o{;`fEjd|b}BH^wfmV0$fApH(%9zkbtxL2Fyy?j>+$V0w$T@iTFT}0=v`fYaI}*y
zV4oFK4A;Ju)BT|M(p=p}SV?oZbcN4gTA*p~fb^2{YME0JXxblfoDskG+iq8jottRb
zrkMh1I+4xJ&p-H9&_A8)THAu}gML0Tj3~;?_b5&h*Eig|Xut)%(4E}_p7Du?M?-}7
z=N098D(!YG)NQWh3(Q169o`@yo3X%{IpBgxfy{6l!m-guJe!Tae4-Ki_(lhCVtQp1
zqEVsk>RK5e9}iSf1qB?Bn#jmOpe=zNf91g%P#yf?!!5cfX}Ty{T3SBd2g7u~ySmmL
ztNr}M4pI{(L@<wz&J}si#0A9skn5JQIGa-++hi60mbJ0^I*wbqinpLdec0-9EbesK
zQBZL9^~q4&)8O2WyZ~0c6VF@F1m0|J8(Dk4uH6A;lyuCoZD|Qvvt?gLZ+wxdi|(7J
zthg)R+qLA?#G%TCGvlrWS+Pa~0)GlHUIy$R{BQ(sq89s{`pqIV-&~IkDir=!>pgeS
z<u-cP*^<xDjf(mSxJAKxA;5|PcvA8lFAE)V$A%8@RV`-u@-khd{4aL~R+8hXt3A+j
zXS|4^K$Nr1pK(*H1=;zJgr+a{--n7B{a4`y{16EFU-7s?*f{7u0vivCaOZ8s>1Ie0
z6<8S<B0hgkQF_-Z%q^BaAR{Y_0%US<Wx?q~!q#d3ujxo;44M1UyJqfM888%wAe0|H
zdW3?4a+iqd^`)Qk&tXu4E{?l^D#~cQLe#BWGd1mf7Tpp9WP)k0$zR?R$CObq_3jU5
z27Z>|v$FO@Fn3AQ;D<`gRbG1zHLb^1fpm$|cmZ210RaV4`Vu8rZH*Qr1_mUXX^-D_
zX=%B`SD_(p<Cy#w(tBO#T>Ja`P~(8C4e4Ilr)n<}A_vtSM|W#xUrHbgxF8-L$341{
z@0dPeLyx~AcQhJXZ+-q3dLjLp|4m@R`_tF#G+kfClKikL`*yoD_qlpYg-b*tJSF$H
z#_+1^%vhx<Jpp8eMuC1$Z}0K(v9qINE7=mBNTQgj!qvDNj9wcwZ?2eX%9oF9=YCcy
zT8wn^MgIe$9C~Rkmb(^}LPv!vO8w_h8`9h#^FvC8B9BIg)X{xhC>q|8#yvNoQLLLP
z!7P5I(vF4d(^~pza9`#a7DYuK4{m(##p~<pA~;$`wr7r!)Vc#PL^4E9@eTy!>}~$T
zr5B}!MN1Pi3RCL6`E!bTdWl1`&%-YaP#a&wQ!=i}{iJC5pX8L&gC4TKw>L2{QB_^N
zy1H6jWjC83oJD{nPA*<dRskRx;*Xe;&V=Kxh^d0NFB|}vPz8_P<p<6qzpdBe+*)Pq
zd+GBD;xV|MFbEJp#_lpt7&|x~rlkM81=$7tB;9fk&CPC%n@*DHd1Gt4Is1YSa4(4(
zr&mMVW9_WbZv_@f?ziE+HsceMQi32ZvdU-n<wHL=PC%L-FhBfJGvjuVUE>1notn6+
zI_odM1!n)2q9Si6yqY#(3xHYUWOG7AVs$N3l#-GX9X)`nnpC7=;OZb>c!9#}R<oo|
zY#_#hSn)O;C-wQS4P5zufZFMfyv>p<P&M-k8A{VCmN?dHu;p1?Sm}83lZ4(m-{u(;
zkHB5X{%-K{r#9{ZlDcSRb!EVF?Kr|(w4ha|RwFdg64*EyH7;ZP%}I(2?RBn$_^dlW
z<)r+ottV<lg^f*!7GpHUwJj7Q(sOmlxiB>L4>Z5B!p7E5W@XJ~GOQFviS+Z~#xrKO
zsLea0z8+ios#c?*-%x1V!fN`TN7OcGl$4P0|M>CI!-tyM+Qk*{6@1|~Q|%-yEbQdu
z1OfY}0!&x{5V*Abfl>CrkP_u40Pt0wV>@!jhf3<nx-PGo8a1a0x(wVH5uqi{CwH(y
zJ~^G-Nf7Kc-G%CqY;I`3m^VWoTFfz#CnaCU->dkobxJbPzkRV4aNau&x=<8|6<6Fq
zQWWSDvT6fxR&qYv!S}{nknb-K^Z&tN9=9{l;IHpSuRNy@?f%W|`rz^?)u5wWnmXX&
z`t=*S8C*`jz8=13RL365(p6c7J`+Cm&*`vZqMAn}z$};Ao5hLd>F$n9;Boli>kCGp
zRk&vp%~MtNOI4Mdni^rwN>>c)P^RKFn6H|A(d`U%hH1^8bP?c24i*>s?0E0%JS*ZU
zDExk8z2<1VVb2(#2Ku%+&^uHmWmBYN(>%E3tcNMK)+)m6MxJvrDyqnT_x6xM_C*{Z
zp6>TJ;}hf6oDX9|{R2u9)%N8h^H&$Jr)O?D>zo9qD-lRi<~dEEtwMc?$;ro$I4Fs+
zy%e2Ydl1buVZVHId0;<y{gcOqk2!(If(9ZTFLgZq1aeU|a{%k@>Lk;-;X0>SM|zEE
zuaktjo-0my&Ec`<73AA?aY9yufjDwlH*GccqC1W!P^o|grd;QcGGGoz(aBR~yKr)r
zeV2_1qY%<pRlN^|kqaP}mk&4AsfKZewZI!QZ8!Ze`5{Z{;GF*G-uh&Sjb--J{8(;6
zXfbN2Z^m0m=%uV09s~*r@sxCFepqK#4)wNd$vJWI;A_v}wH@dsWt7_exa30xrb5u-
z{SR=%34j}Tl>NmbfNh3zb>DQke~61rSaO}xX|JG|s>g?ZKYc1X>tI9iY`XcIYrmw9
z^=xSo1bl<b;COBitcqC2=VLnpQ*MhAm%>Dx9LP4`7=BNZ)9}A)Dp^R)t@`rv(Sd=Y
ztSpQh(+oM^<m<FCS`;Sey8D=!8T&@0!G{S592S~Cwue)+|N8ZanD|y>mBSh^jl8Y)
zG0N^>6RM*mk2umGf{{ly-m2oVZ+NRl9JYe;<`s31<nm9VrWG~rLYN`4$^Qh53035#
zy|O)(6ORf48O)Bf`l!FbTP^u({LU&F24pIM+&rluiAvOgpx^uKt^50%K<jCTS@%do
z0r1ONtW)Xff$EFKB#*rAK5NLCa~Q95^e@eL0g=`$Mt{}mN61x2U8F@`u5EMR?e<B>
z0KLe$>+h_@+@yzcz4$96Fc1S8FSCjPEJB8Kl%~}%icd($UFdV?GNVTRJ?>eQm%_qL
zQ&aIGsm~sM0x08Ec`F`l`JDJSW~*-o2Y{JVue?e<tg^ZAz<Ondme^EyPGV3{?4&Yx
zK`H(ccX8H8F$J5OU0=@WLxtWu6Jt9<MHJZpXx<T!&3(P|!uv&g5D;Sc!t@Ob7VZYi
zMB4~xi>9MAMz54;N5)qN{~bqxA5>H^WPS)58}EyKR2V`auIkS73er&^fb_Sv(b_w%
ztX_U_eVs8^6U2CY4{x%{nH*A8Rp(2toxOPnnN7Bg+1c?JPZ$_a3MxjZZU$gpA{U4G
znVA`2X^5M2MKhBI0R0uf2M`z<{vZ&=6t7;rBIdLlj-po-^g{lF;p85<n6@Y&7Ay;Q
z5vbT<cyWtSMTEW8F>G+;Q+61wu*|OulRzD+tsw=>H7J)gC#CP-_X8h!$N!hn+NItr
z?d{6nZWr$k-XnQs_SqSRb8EscV4Cl!jZ~5~g4ou?E*tOX#{7l>0tbj|p<lFup5&2V
z<8h&-+@3&Iv^dV`>5i3I72dj07&IzQcYhp>sxq@MpGSXoADN-&5s=lCN=Yl`rEloq
ze)9HD%2iskNC{oI=kY#sqCOum=i2vLu+NVHbKeJY;PJk-k|qs=#Xh_-rIWrMqCob1
zvlq3&Wx#;qydjCl&hKN8FeAG~XUWD>!D}fCF1lln;>$m&*2#CltAqWtgKKVG0jvj_
zpPz5+>Z&TdAwJp8z4n)HN85}J4SDP@OIto&i*CknQiyws8IcTs(NJF@jFyrOc5>wX
z!~J}F?EJWsjM0U(kNLFEOM@KKWb1>)#sjBhY}UlF9(a%LgV*tkE}#0k?Zml~$uncn
z%=oSqkn_X8c*Uq-K<*G1S<Q%qskmEZVO2bI-OG0olhYNGBR!|gY4Xo>$+I?8=Mw$2
zszA16zhmKP&wJkJiXU_{U=C(n=Q?>nuNT^Uqv>uZt(|DbW9vSL5$hx;Kqb||O`3Dx
zO{n9(oQ-3Vta#PYgZsotr)T?B;rSzda!fGF?A~#{1FsOu%k!GMVcmq$B!LT>=MAHg
zm6J0jbpQT+AQcEEICRkdOI82@J*QtOBK+b46^a~WFY?)mW|GI66|7!9+8o)yO$I}u
znBPCxN2QrzWIwXUFNojjNh#hbO0-(#-QpRVR|c`vv~AWqvyX`tD%aIv$R3M$3S$Cs
zRQ4(^Wi03Y+`)bFeSFCpfC!IGd$ptqt`QQ2VBTHdFqf++iRBn9_Vy&evC+2VYyAUk
zOixVCf<jI0SJTs#b0X+`Z%Qw~xZX$KG%~_LdzM^MTF#VgV#c?;m?$|%FqAdn#=)zn
zrLV51<z1cFBPC0KA2b{pay$U5#|wFF5^^W|-A9;=wF5ZY2DI${j&&B-*uMfRME{PQ
zmOQVFWbbj`*cX6pzPdbD`60LP!q~=!oS672J-y`pGt%B9ff~z^r=USUCR^>N_@0vD
z(`g*pX)N8zQdqa!NunRXT7}wJl-EVt?sB%?zm@)%){&KhISdF>^0R=@^W(>DsP)$`
zt(yAu{H|-K)&y+yxUfC`jkkumbAo3n<r4ojH)QGbpbIkre*}0oA3l5lbg?DDe{lTW
zl2MAV8u3k4H8qsS&!ZgPyt~0XPIgclC#x4;(X+DR-oTDJQPd(^+jDvIupK4xnuKRc
zy4Re4513bV6Q|aaEG3J26b-M-Vxl$hLrGCnYrYF1I@9t^zWq!sk9$Rcg5)I@y?tz5
z;d#@#590%C9Ka*cJ+hvvnf&l@+(ZR*h=e5mw&KO^0xI-`k+FhK(r=P%O2NJ|Ob@jd
zRU{q<kV}fbbKt-TjFGB7zml3TRVSfZxxnWHl2Y%swkTbNxy`M=A&KpRR(k`iR907;
znwgbU7*yGRbnTxfJH@3CJUu^0Z4AVt){dnU*^(92djCx=gs+pcBeUw3AmRY`jfuUV
zQs(!9^u1+dlklYz&m^-vQlQ2zE5=>rWbH=V9;SpV<J&6%6bdpUAiP6~IT1kDZ4E>M
zsrfJc^YiD=xH#f=1~oM`ht+PwXAh0ptFF4;{CMdpg}t>Hf;MZLmk-}c=;_H4RbGki
z9_#^M2(;;U7*zAx+UErCU>c!0km3^(eL8ADgJ#~U<lnwjRwQx~pR|+-XW0ssh2=s5
zFZGpFsTyUA*vYceC&<c`Z(PP|+U?)?VC4YJg3_3&m<;esKr`P+e}*l$7ykf0XWX}>
zxESahSXfv96}b%op>SvFf!R<hi>Qh?3S{o+ZmJ+Gxi%PdIH;ul`Rj*OiGlOm{wO|W
z6F$fQDVoUAj7so<At3`2+6lLdo4eAh<5U9%LA~dyA457*>F|9%Ua}iCfHQdGJox;0
z%fa6M7DN|yc6sjd^s7v1A3)khbG32{K`njy<OvXZL34xZ^YiEWK)NKX(yQmYb-41%
z%Q{F%C~`0^bk2UBiSeq}*DUvVa33>m??<Ms&MJH^sT-CPlFfKMt-!>9o!N99B1acm
z!iEYZ`*4T;YV}r>>FP%DDH=81iqSIrzja9^mlh|KzLpjTfVoppBm=D(z%ZSiomxO^
zuB@D{_o``f_4d9ZWK;)%5)yP-6>lMQ>@ng90W5lOI^)f7@30npcVLO*^t!l&6Wz!5
z&QBnvVBbEy?SIP$=!U)y<`x|8Mt4hq#FtdN&112r-okb;(vzvJpu={(gSw!cOj3)f
z(FA=?)Tq4tLm2Xtk%Q@ng{B+{yW7O7?}PfWb9?53!f#SLMb{h1#9Hs`pclWr(*{$d
z?Io5_XT_k#0}Xh`zfUvAlItvDZsSKX(@n_V6PdPL(XU@>3kAZTE&&RlvpkMBl{GXf
zii=;VUJSJ_xt%v4JtlJLZc!O6vIA4pu!Vu@1}3;+379E{@qy0Kn7E8!8yt&>S&#jy
zGf(ia+WD#!iPC!N#y3T&>qLS_!0oV=1iVyOy(>K9kV1N0(YIwy?va;ySFt)eYGcS4
z#poulnx^h=OKG_-a5AwH@cl!@u{moK4qq*eY=m~FT<7_tIxm>({;JiKt*KwpJWD?P
zy(Y5`Sp1Tc3x4<?GYirR^62T*4d@+u>erh7yBCnh2y$?wqO_e#a|S-Q*VQ|mOI&`3
zaR~!H4=l9CjF;yv;wSSB^xyGZR@*1Ofw`6FI95ZGAG~v*7T&t>#Bt`5xO7BAJ0(zg
z4&;isukrBE#>749%e|H!9Xu?ERUiIVW`ldfGaMTQJX+rXzW^82<McN8Uzck(005l`
zFGBu`p@orF7oAo~ukhd8CEEltP$>2bjbn+wtMKPRnHows25($8+Q(myCfvd=^{q8|
zONyS&J)neM8`)fNe0F#&5F`e0iGgyhE4G*=L|QgxG5>uTX5aTN^BW~4Y&D4OgB3~p
zSK2??t^=Ol3!K^k3K!U9M)kC%g)NQTcvUc(IP~tQ-&E%SZa9zs?{iBxcHp!0e4D;U
z51@M+l#BL`IqN>jvKG9bDx64#Jj;iYE}A1-9_%PQwCU|Smu<jR)H~r||KoM=hpUL3
zs$mfK-&hK5>M5EeiWMs+vAF}yhAR4oUq*kv;cT2+m?v3FV3zM`#lgxNW6pcniz_S2
zkukR+OLg+l*+ITe7{ka{5Lhq${r<bHx9ykSR_WBU9DCs3yw;Yhns;?+35UTGQCsbr
z;FMw+%lO7b7YHP%(D{l({Yc?qf5q6G?A0xw(r!|Xod>$Kx(3YYHgo>JH0#ojxh=OP
z&V*G0h2TAgVbi4wikh}ymm?hGo05YF*Gn_z+Bz1!R;N!l;S}}m|6-&epo0$(4C#zL
zMva^Fc=bfHv7%~^ni*x@%F|v%#@!cX_KNa`PRfgsK-Jk9G(T*MmSbfs;P*b;KCM!?
zd`tQ@v<)Y=@O3ZoT4CU~+XaKo^W`<uzu~QRK~LmpWS5?U2h+&d=&%hrG<|~2Q5K(R
z`G$VV)-UYWmpz9>n>qNma)JZNT&u%!N-7gB1m@1+Pt!lBuQ`m_gES43(tUt11&1o4
z_}Fnx%lpKvGoV0aSZ#>`-lPHG5FdZdl*QEViM+dX-U!emft{vyT^;6Jjt{>{%Pllu
zOR;6G(?>Ln-w|o(pnq1oev@SnS89%>|MDCSWJJ5xz2hn2wV&=^G$<8>e}ja){b)62
zyiX6E8ap1HE6!nr1OiwjCp-Tv=^K(}w+B2kiE}dY>pWNWbja9_S>D7^Ur`P-9M;Y{
zr&oRU(sRF$RS+{+25zQ$Gs;6j@!SFX+|oEN%$7*O*Y;MC#LOCnbmv<YEdAOb2qHw&
z0>>C^JP7dbLq6aD)u)6lM+CACm(4CzK6UNsHz`F?wTKrVQ7I@`_kL+%a>nQQZ5TCd
zg)2$!-NCZreT<Ox1oHMd%?)b{P85M&Ne7(O(bRP9OM(Re!QkTKFh4~abQ7a?XWK39
z-joU<X75NtWu+*$yQi=#G6r}w3a#Yyg$EKid&Ox%2Dk_VR{TT69ZU=wnoI+iliu!5
zhrL5KieY%MBQ+1hrlw2J$?Hp^#{BX4T${bpPz-g3*(&*qrCZQ#FZa-LE)})NkTc@z
zKWA_b;2Ar<*&ZZ13srj`YAot}D2vji+NROyOSy6%j9hg4_^bYXYR>AF?ys3^`G25K
zOZJy;5ICp#j;Gvu>B@m&coQ7O67-UXAOJEGpw55{rXqoDLJ<eHuX1C~lMXIt3qrS8
zAVOv+!AH#MU!tylk>5iSUU=4Wa=62~Z?r{Z{G+wg;~V(zxOZW$zMy%^bh^Kd$HvCC
zN1kRK&p1VBl%$6WDK8xBwBHAMZ%yjZdb&htaA+v)mA0nlc+r_7F4zqLd(=#<b6%$a
z@_J?I?~i~5HSj9tI7hxXmW>E{T@Ezk82Ht8^|1cq0~9%&3er}~3JNnL*pj;-ySi;m
z$WM*NfpFS6rsNU(>%&~95zbOmR>`dVpC~>JbfLV*ecJlu<j^<XUgsW>_KBQpfCH}d
zod3~S0In^1DQ`WkCSND*60h^o=T4Bre*p2$4gG>?{<*`x5Uh9`;v?CnYa{^Eg8m#I
z#XaH$ihaQs{7?A`KEnzH*r9>oBj7!9S$d#tRq`GzpR)=#Ccw`uExny$QN0Cc4K`6y
z4F+WOs!}RBp3~3KKYKg#;b-VafjdBGH~it>XHd8H0!*;(!6%<T-^)~b^7pNMr~Ngx
zrVWzCqyZR(mFsTW6<}Y2R+TJK*~Lfn&y;VFA#9*!Aq}~1_EmjJWM%XAe&MpTX0(J4
z`2>g*adDg4YDC6pal*)9yYlSo2B&W8cT;)_!s?c(OQ(0Jnd0k6J+5;t6CTqtc(43%
z7Stz)UN~;AvbJ&G4e00Y=^Z77-Ynq<?XC5R&Z7E!$W^@-oXpz*7S+x1FakaeNg)T=
z?flPvP-2dxW*qd=+psf|S+e*5QwL9uDKa_f>(IFP5W79}^0L8do%;uXahw#fQ_Cgj
zqS|msu;dkduP+36n4dp@+OVHnH+1s~xHG30XaAUBJqH`jj*)U<&0maOA>kprabN;z
zFZTuZE{><i<0xs6`U|QB-8WAxXW+C3%`NGlj|{wNP)-QuO)_XuQh*uib=^5_;k+5Z
z#gcr<*ktWq_0?gk)3`Vz!&sID6OuKRH!Nto1E|}fP5(K!-OZNF0!~g2Z|{2bc5Yc=
z2^pW{f@*(%(2ASyLIMoZMc~lhTx0)NhaabP?mz^1+bu<dU!ez%nW4Zh#WX*2I%~*F
zE67Xt+SP8&NM3E6Bmn~*%Z)<7I+wGe3AkUjf~_<^j8)afFfex=bOEG$8j*1OZ920i
z@dZ=z=gABFISi-!d?Y1|+x2t!DxfLgR3kZ6_|5%<6msPFQZ_>J&<OZ>N9`;qJ}vPv
z-EkZrmdthEOacT8)`tT-e>2}B!XH*uD=gtk<h1a6xzwO4I@_*$<cbeag;IcI3b(z~
z7_w;Mz%Ji;XTedO{ZZ6N<I`2#yMdAOnRZ+8?<Sb+0+QHdrIx)J#Kbf2rl}y;B{%;i
z9)C$Mw?it98}~C#Sy_vjbxK7|c+{>ISk!d)_hB0ib6Be<D=WL+&N%9O(fb327|E-G
zuBX`m4H>?_f;<_}u6UhYjqlwmhLJj#QeGbI`j|QNIW|mEi#n$eh=1vz`^H}Tc22}Y
zp@QrunFTq&)NBbPuxQnnCkxNNfaaS%zp_63YH*xNUl+~AzpLos@rXs?%(xOR1denS
zUFzp7!SUjBQNZPd9@95u)fNZmI9&g8pWW@#R|d!x-47U*;$A`M+hfV}a=Nu69X1R5
zTkQynVvco1{wItDZ^a8EyltT;1-aEn^KO@}xXC_XrJ;veeP=DelxQDl&dTxFtzG#3
z1P_7r>CmH7TDJN6?h(?{w*}>O0_Rx8Tpv8=#_u`mKZSd#k~IPHAuv$`6z#wP??QXs
zsqqtVR-i`Fplst;d|++!A~CT5C2h-wR!9H3)vQ=+`EqcKu2&oL`2MN5xwE$+c4(4Z
zPyM@-J*9xM<VE}q^fC5_jERc!`sD|ss11c#tW{{g2-QaJFg$v1sC<{3BDTCa76!mK
ztnP$uJvc~KL4gFFJ_>LpNs>kCP7YW4*K>5f0a`3|?qk#I<<-S}Mdtmf?M8ZFc+${t
z@ZGXr?ucmE2-fSGM?nq>(C%Xf*IpVFX#IJ~fvm+qf%sT#8&G`!+q*SXGqxib8=3Uz
zazzplxKLwzm(iPXmxR13`0@qGF>PQaCMa2tjz=EZ81frtm~`#@FUH<Fs;V{o7G4Mf
zN+}{CrIgZ2HzJLMO1DT$cWxG-l+s9-bazTC(%oIs-5vYBdVcqQ-?)F>J@!AxbH+IC
zwchuM`OG<=raUfNJTbZTRh)!xN3tjNr`17LcW?~PtABGqQu2L&91V|!fAUN{siF`F
zIghelI1#0bpl7C+An2Y3lm72U%MCa89*{<7&s_>1rcK{8it^_f7kXy}mhjmPF+TBe
z{ga!OqQm+C71ma<J{mg{CEB^=8D7_&-ysZh9up8fAjHqk&Tl|&EvGip?Hye4lboGB
zbQ-0=z;t*8V8Dcmb>)z!LSA}iD{>1w7lQPHFE}4SfId06rsbZqam-}U7pn<9rk~aC
zeD+#YaJ#cJcl8P7tyk2F?6#Rj)5j~bB*C1IDbDKUVVSOB6-ABb7P5`AkAU*5aMr6q
zWC2u8468$7v(X((nXT(*^*8hGowva(?27`qKIrGe5|Wn5@0~EeaOfCCZ}?IBXJ_-6
z2@O&rFt~|(pO?a<Kou4~R3?~&Mw@BmD~{mZ{!`(xz!6-At79$K(+-pyY8(i2VeOtj
z;cT$e-23FF=dR59%feYyb*%g4V;T?u2Clb&^b)kZ`HlF^>_GeM_UHGVzw@Sof&#?N
z)j1H4*4Nbmfz~0In3Ng!v8b1Rc941VCKQaywf=p8ToCvxWM-}awfy$hR{IjV<*T;e
zzY_%AxSl_M&d#p*Pmk#YZw%)ENx{nE;)gSf=snWAXF%c^%Wh2f^l25%SxBP(-(Wux
zH@COvvW5KeY#xjX#B92@&s=`J8A8#U9YcLJdtQ%?hkYHIsQEH&tO4ZZ6~xiNSEt^c
zDHAij!J5AXU~P!*EaJ-cX7tk`1{b&6kIelmIKf@nK~D2GTow^dKuwYG<J_e48P@9v
zk7GYQWQQ;D`s**J&zM;1w@!#5Url1}>mPM5{$C;3hd;sOH+aNum1i4(y78s<@xY40
zMpzp^mVo4PX4~c^P7pc-CaEh(BrEI$G;CaNkYJ7ocqeqMv!(7Q1`lsf{2s<GGG@Jg
zH4E66H8sLV(Zn10{<m&kyUV<DjcSF7$XxKR2q7SfcQwoX&0$YjE5h;Ux~ZiWcuCN#
zx_<PVL^eJJG>NkSvr}AM$1J1OAFdePWt2T0D(^H_-Zu+X1zS9PNJygr`^G*3zBbFp
zOf>vO&v^y&e%`-9ql<7^^Cq<s@3ZTbLC($Tm88vh_xR~H1Qo(xx~^BnuI!?l1;u>S
zw-2*|C|;O}zH=Ro_aDEw?fy|BIKXQs@xhduv_?FR+H$uWaw0F`DwVQhSV$w6**y~E
z<%(V~VNog6>KFjJsa3c;c5!8;b@uF^)VBdhc){>#vaQWezkcKs1|Co(C;mL+oy`TR
zQC3z~8k(i8i85fikefTRae7`vxCh{l(a}*(POUL<`1>7dzuCRLfb&Pm;Ym4}`7Z;L
zRlh}8n$J6xojJPzNg0VgI~w$EkR1r@lhA!H>aqJ+VJ4lU(^>0XQd>4bO`>VSXtyIK
z9v&tzx-IvKM)g$sSXmdSbnQ6GfhMP%p&MYOVXEG8vN)zLnu+-I3w&u?#H)&8*54b?
z9C*}l-;EWR{I~~FxR^Ts;IC|l>;D^(w!?NB+bvkPwP!RVWCX~`b9ivjVAoe1Olq@+
z%&dg>YOk9@9J2lV`@;$t=>m)2?=I{`8CE94fWhVN9}r)qCw!T*4XW{w`r6>HY%ZwK
z&DJ38V30jM?GxHmPcP4e#Dr~i;o8lulR)-fv66@4N@%xfX~boeYHH&R4^P`Rx0$<G
zn3x{Vy9=MTv9a&M-F4RULYKV)cJtE(J2TYYM<`g-g1n{yrJZrnKf^W&D_t;=7YG8V
z+QH`fYTMXLJR=gawjRQ-sLV?kWsdW>RDZ8-C?}?MR}l(nUcj}|k--c6dsCW|B320_
z9q%;(6Pjy5Mm)%|Nd0@aHD6kBXj2C{tae@b<y8h-CHyq3-Qcb=C@B|?0s;s4A%Dy}
zlj_8VMBkPkmPqLbv@Hf(YPWK9+-~F*Mm)F+{kTp0%v16n+IppSt>gA&#aD{tGIxK{
zhYuar`k#o4w{V(`nwXjb_JQ6qkcj<Ebs{HE1QQ`sQ_}&H@}r}pcy6oAtgOHCkgM}U
zz?X!-pW1;i2bfflCjb+(va)=9!~wq{B%}}gr-Orobh0s_;o-lgSQJK^TDB9(c&wqX
zvsv<qHGm4J*X&!OHRuU>R{RJnU+VAN&}gjELu#^JKCJYZooFB>X}tTT3wo&}ruOP;
zjz)*!%SVPUhaT~3+l!5QbFsdyEB$Nmnfgwzw#Bqtvs`(V^A&@0Uq@adu4MEP^UTL3
z|C8l?-FGEKva!vzlMSdia?C?2$a7A?=$yKmO1;H}E<@!CK*U-u&J*eD={QN+)r3OH
z_|mWCZF{%XDw@C_b!yL*=U@FAyh|03p87beF!gy?kBxUd?NE_)^dr=!!3o$BRHKE$
z?j(fqN?Iz?X#)fX8C%RjV^>0Q>LO7&CEoe=pZhWntCD<MXqE{Gxj~u(6|42!MN64B
z`CBOD$XJc#B3@y(4m<b*_Iw$ZOoQQy$QXB%-)|ZePv0-Uc+7@Q^vNg7X&8%Q8r}u;
zPQS@Q9Ocm<7g0xp$NnJvVHpSlr|ph@%-BczuCBleE}SoC-D>S^ZI=_fl#cbvBf}+m
zY@Gd;!qOL#=q4FB-jAiM_}$NnKYjWX5)xAF91lyf=r&+OuCA`Y(c;4gq{XoI58I5A
z5*tfPOVIZSdOp2}Ij(^~Ag(<<A>|RuTEl22^-@4{NRtf7%z!gy8sWL2h_vAu)LT_3
z?y|Djkh8P%+1c6AZ+CQd3IUJOh?(uSA{{i}9$QMv_%ocE3KRg55Lq%kW3ekQbGBc8
zrZ_^JM=6;&#?9f7W>bV~tZ&rbVQsIvbH)_Q?qf;>ME@i8t0>TOElE)^HUIb%avq26
zx0h>)J4+*ea4=UUwf!*fmW4HDot>}Yk57@NrU{e~2NzfPq-8$m;MlKR;{C(J3fA9&
z{R3y0O%k}PLm92m_mo^`gUv6!Qz%eV|C0}HW6ZyFqE(bsM~63Z^8o+m*2~KdqE-Rp
zF$E0nuaE10J0u+UJhf`bTwNlF8Z66c^q|Cna=(;P!}hf&evD5Wi6QJ`4mFR_?wv&l
zM<qGx5AS6;Usf>zRPIZ^U+hZpJul@KK@)q*fQg|p=WX$25*Km366x_)lqj{rTQBjR
z+(Q$&z>bs=blM?PJer-H1a1wHk&&te8ZE^2s7;|yM9Ij=!2K#MD_i%8X3=I~VnXu-
zdcuA<&8gc96{Aq=$2&#ELoizUNzi0>jQ21|qN}s>mM6Hn$Nlc;yjyjr=~nAR^%O%U
z9K0mbSo)tVk3Ja|;Dty_-9-IVazLJkEdM5?L!U93&TBqDgXax+YP|oHGGp|G(IXA<
z3UhNbE03+s81F-l`P%J<ZjP7*zuj29`Ap!K8H1LBhGBzbB;S?{X(ZQ7-VgA5AHiv|
z*akFyx0)DzQ5STb7?A?8ujrrUKoBi2VQmE8_wkmKK-;@Ad~06XE)tdtJ44(Fgyy5f
z>|^!^Y^Te$T;vT$f3(FgP&;I#?cKsLy&q$qnp;@-;ZgP~J$w8(G&xxp%+f&oI)S5=
zl%7-KAN_L(5pu6iOQTi<nwectRsq)d%F@!XUD?ffwd?E2;bAcve3X|OHJ3;w<hw12
zSSe|L)sbQC{Sy>OHEUp|>ui}9Q3_k5g^49SciNuLc_8^{1N{GiImtz12<F|8+YlqW
z=*57&_D2wH>M^nJ1j)}Ez&v~+C@8yGH*<q_TCEeg3|&v*F6WySR7dpck@ou!+9s_h
zk`kAOw^pL$2l>QW;KgEYG?CDT&uxoZt-&?h56;+lY`1qCQC_|^H5@!;T62X!^;r@k
zc0NTHyGzjJw@OA;j?bUHysS5Hxa`J?i$oAjk@mH9vA<H)78MtP?P9`xQY_z}Wlafr
zZL^|y7Fy|wg2Ut*?klm1Apn9lu&qc)NB}x)40QAckL$P`5eW%&C?<|=_wZMD9Ou_d
zmlJNS+k3!`C^a=TC@2VsexstamLDR;_vZIm7R~DsS0#bEGhM=XkbD|5H&eXZ$ptEY
zto!AD-c*P#{9^PGiOC#qRr1YE&g&pIHy%{-V|l-{T00Nq0kfa*Qy7_Wl~^*}<8pNV
zNweCjPhMG05W`}8Cr|7i+WC)ZG7HN~G!N;nUS=j`#i%xlM#fI*cw+0L1^UF$@1Avm
z2BzW;dUp;Vgv_QDriSiurlr??F|@EDv2*#o`VDn;7lu3weLI7d56L8ApC_4~dMS^(
zqKVu~7k&dgCRC3BCmL`!KKuEB#8g#YUTg-+;lqOizMYSI3tXPbAlrc<2{4=Dwwe`q
z`*t8HZuIer(9hYaHcBX+=6Au(%!65++EL~tHob1~qNkvE7N=b<5DU)@PpHWqOrU>2
z2M1ReP3DhhwAOPZfd0qoK`T#CB}gKlo88JE@lp-$w*J_}LlsSX-O1>KhIUl$+jA20
z9NtK$FzpKlZkbsYF}CE?nBGT}Dw*hm_<tff9g1cp3-+(JNBm5n7=h7TQeH()B?1}7
z+(zT&vi$4So32%><`%Kf$y!$tUA#lE9*Vn><^4z*vqmXq<}A;gr2t@w2-NC=H<t?P
zIVrLg1io_0-~vfCQayXtUV8OA-~qqWPWqMFtyW>+MF52N+S=Nhnp41`V5ZI`Vm2I%
z_Bsrpd-u+Kf?ZY|RabT4>tmpf0e(A9PN$&N*x1}WZ7)an`Ysu*78L?lyeN{@t8bhf
zf0>1N2y}V1C)7tAtdvu|3+_DT;G%%R=YR=EW>p9|wqICelt3L~WMMmxhb%hcBoUWS
z%Z->75AR2F6{6KOj!gazcyl59^ZOBG5-!5<B^%F2->;%twvX(uA>KfrFdc)dv7&X}
zfje{gvQz1-AecJAx^-RM%(mqu6YPiR5cB}iV~nDg5jkQd>=9=&XJn%*?V{W-xq|JO
zu{;6+@yL*nYu13tojEo)Ys<e+#mWtNb6sgM1-<|6e;+&p2S-OE?A~Nys|a4eeNlV+
zwxqh6*bNu?b)L%;4-0Fm1Ppe``Gwri-kX@1=<BD!Am9Mv*^d4VsJ3l+sl1Q4?%0&v
zIUyupVnK)O2?V?!U&n6pnC-4wMJSfB%(;oA4J@V8);M)8cAoF(2zfc+1o6x3)Siy9
zw36AX?pzwK5TY~@Ge%Ivb2l-K*<=+87h0L{x$hU2BxQQ%6R60W&|30)R9s8uZRFlf
zDS@TWTb2zYIRUqAIfEKs8UuMEe-?)V4OdFGfphivIx0Gv=&y_t4EsT?|6=4r-xSIE
z+8O~7QC50-@zyqA>H$59t*tE(hkTfL|M(A=r;~$2Mq}oiXAA#6<mHkm0&I`Bu9J;r
z|D4Kjk<MiCOz};eEIoJ{Wx@e#oMkOF^~c0~wla(yslW*e|MS6??Hv~o+*VZ;>y;s?
z*7T>ZjxnIelc%|pjZdW<Ud~RFgJSj?(X%_mXVD#;dEK39^b1KSugTk2Kb80xbMTXS
zkE9eX`8#b*czb~nvbHJS(CJkI_I`3*MmvIbfEWn>+SN3!4PqZ0f=!;>oAZT#2`zIB
z4r%dNe8B5<_s`x7uwc9yN6@iXwHxlA@a_atmyeWU$+*vnyUX>xATJDRSvvakB*DC=
zE(twUYKbHdsnzIQla%?<0<n17T}K+Cvq}*41@jJ9#l21jGXU{bR&fwMSU7tA7Q~jN
ztft=2L;XO!P(Ne>mseNwWWF*p+YAj21-=&I5)zZ+<<jbM8Sj9J0C@P^`mKRFiFTBW
zT%MAbvtc6A;vy>@EeJ-(MCm?y1E#hG>qEf=TuZItsR#aY7Y(OJ$G<Bj_al`@zbTK>
zz<qVLePe@a$fnE;I`qkU!3Ss}=v1Jn8YJ!l>9;r`2QCbrQ5OzT=Q1mWD0gQYEtN-y
zuMt^2lj4XTJyl$*rN!k1#~KSco63wqlXzTRhJx&y&DIC5=XLb*bm$st5Gzf4$V0x#
zKrJ3b>e!HH;PkH%x4uErMeAxq(1lcT98B92Kv9;$rdFm$Yrh(j5*a`;9xN*_F9(+(
zG!l+f*3n^MWPHsE%0y7{%gf4st~{6HvYG{EcFU8KlOuVmmzS5oY^FaCmr^K@xd7;W
z+$m>8iogT0GBH^Tb9pQUr{$&K+n)}{hmMEF{PqsNE8-!9DA1!=+&<z|c5^e?HTL2U
z*8QpS<u2Bp;0mshx2g{*slf3-GE>9c0$e{Lr2^$Hb}9{~blq;aXIfIKRzE#%f4Q_`
zVz(=P50a9`CNcb_?E9!M=dW)t%iQtTK!u#_!iX?}F%?C}N_He-63tRRaI?v3aT{p<
z6tvw4BF+-~WdN|G!=eAje9UIm@*IM2{Ya@Ubm<1tG^2PY1M6<q9E^1C!(4*=cFh^=
zPeb~ZF+9N|6F$MCEw5zg>N;vwx(_lv#nPWYw%e1Ox4fQPPFDjz2~uL>|5(z={7x+V
z{F-87={Y&{zzyp8^ZwRW30Ya$ckgV0w@y5lg^a8$2RplFsX;(&th&DbBA{p_3%a30
zz(wgRaIY$>YWNLXKp~F>!AN0p^cOLYkaTt4Hg8*3^Wuuhc56#So~832q+VCp*T0Bi
z*`(d%q|cD}=c)9yd@~b1yX}PvKMxvwm!4U6s$bp_Wce4iz91DbqUbyYk)6a)3xBb>
zv{Oo<>~CP7iylJ)!NJ~Mv2$u+5%1PjLq(!I9Xq^4QOm$-=#nV{_(S78_aNU}zg(F9
zH<`dJZ-%cVD5p5Jh7bac5!xh$+^&Jyih)^cX)-D)5#@MwOtW-nmz3jk+P_J=%Ej3k
z^cG8ixuDzVk;rb0TTd3!W;IhYGCl3S(jBYB-~jvufy!1*z82)|yO@~f@ML*Tdiqj-
zifCYcY)s5_y<7BhqZz8I)$Kh4w%BMQ1fQq9oV>@|KUIOPnJX1vjWl9#iSEs>E-?^E
zY7$9K%_eEiwRf^kx-NmS^x>gP;v{FVvZb-ji?D0bXke6eU2;g{Bclz^*9CY*foAS9
z3ItVD`Il;2`=BSE-vR!|G6xnnrJ&GqqFjWUq!bocWpPD7gN*==fgqCBJMVwc(9m#k
z`2mPlX=#f!4x89GI5|oiUmh}22)YhWP6ETSx!iGZPo^d&68RiJX%E<3vVc;qpwQRT
zW6|>E;U^|nKML%5P`DIo*Am{p9~K_|PFov<7*h`TQKSqwRJ*C-UpY7jVl|~C-wl>l
zQa0QpN9x{^MnSo`J*0J#uS-RBg|$q^hd9>5N_hyW^dN%_kK|WZ3@-N-4=j?o9gc&D
z>fD3PwFgw~uFfCDYdX6)qw!w4V~B)$8CF+nd{hyF1aL65tBU1pdx%8tgPhYpL%k2`
zOgL1i(iJ#b=BZY(J`Iy1*IoOUGzJ>Livv`v<<3aP?YzLCAfOzr<?i#!FurvL`bG)?
z=i-bEpPU4VG;$C|02&Hv5(AAs&&T$Z`VWi?rY9z444BL<XmuKQNy(8deVf>eI&C8P
z)%gc4Q`S&uM3h>swQJUPRX!YZ$@I89CFlI+s}q0(iHJ7WYKNy^#nc<=nh!b&P@JB6
z1t~raUN1*xpApP8?Sc5#J^d>)itx7jx}MB|T_%s5ZSltYs#j{}gB2Ja0lMrPr18oG
zhH}{DTr7zNDY@efhNj}jb5)2}`uJaE7@?OU5+BPm22x`q95`dju;KUWJN5FrgKyqW
z+a3*OrCTX6$f-%av5Oh>&Uqk^C6DGw_-eC-zwgRhh>#!}3+v8D97PSjsh*yI<H5k4
znMdA_5x_>o@t4WhragtMBEi%**_Oeh;eSLAzI1t03+a>2J8&vZCt`HuGJ+d|ww_CY
zIUhpCTWU7OvQ+`%01T3Max!JTimYsYO3DJr<~TSwtG$Vf9y|u2rY7b(HRtyfkL07n
z<0>xCmJczwPxtdZzMO_7(<?DR`I-U|rgQwS!I+wHMUL0u7$qf=`274e2LFO)fLLFU
z--n0>kO<go#~dHkN6>#ezU&h`#?xc!?q&vG>|p=07cjfK{XW!HHv|ST9s^U+>tIl~
z=Lhdrl|BDpe$H|=RCB$WYqq&gt9~vHzO#L|stOX2jp9r~jZ$5my0xc8zL`c_T`PpY
zqKESfBv~?yiSo}8lSh%Y>~=62N&hR8g4~?1(pQx(=LtPstIr(=w>3rYrT1x#H(E}1
z*6YH9r3yL-=YtK=)?7UqTW$sOMRNPiCm{0P`z$(*z6bbs;C7%x+S)TfngCitLdd$B
z(ys@YA{rPQv+C5Hfld)dUY~;N2$*oyWcNu7DC#hQ`{nlcpUp&GwPVatkx^S^^w&&)
zxx4k4*tnVRp`x39fn-w61qAWInDf#!2dO=h!X7f{p88YBnC+cm(<;^7OT8zBxvA3v
zA^VSVzwnSP5quMHCrOA@B{)Ct`Vk)zdrKd3CLxGbR7))=EH#MAD`_y2(_XwL+SR!&
z8A|`H;_~L=YJ={Yc*AKSnb4dE6RHys7<wAQ$vwQhyl|cnezG>=MipicFa04G=UqRU
zI9Aa5A(`@SCUHAc!8SDf9$tVvkJILzFsYiGz2(i*FxqA1Av#{!L=#pzh664pfNbfq
z`PlA^M%-3>+A00o+Qkfne%crJQ=p;;LMSN;SE8Ul@5=q?`z2t*EJvP9pLc8~fF3~(
zvcLD$yK0<y<i#fIqaYYKuTk8)_dPRnw85*q{0N9YKo6;?s0eBy#9}KBi=-HOY=5>h
z1bs)DSL6Er&UE2Lzz|)?W|#Uk!R;19=rI#bb|zEb0`2${X(&`zskIa(_LX|c;nHP$
z-(R4#bS_KoPr7E&(;EQIr{1$(#a`^ZYGKz-TrKMBWxaukqo;*Bd+3>7a1SD^^%`UR
zpR2I-8ZE`u8z)`(efy@mOh90Ut9T2#8|k{|O$7Pl-9P557;;_0%-=V6%HrMWO(Mlp
zQ32MVwpL%t$SHN>h2Ib`MD7RF#HJMb7tbfdGiaXS2e|+<F47fvkXOnvy1Ke3Xm@=-
zK<p+ii5Y%+J%>RxQ_BhP8)Ax(_=P{uQOP3$z(aa$VMF>?E_AC294$48U1C5b1ZT}^
zy7~n#Z*14XMF1d%GDKdK8NX=iH{pQ3zsmu=QsA;ri`r5LpKF`J@(Q~)=QCxckz1bC
zF;dZ%DpGI0`}Aw*s0V3}^!Iy4jsb&6OnA>y&~aJLgb6a&V$<JnCUq19V%DKNDH^sg
zvG9nfgatNVSsEqf`^1_W#2ZGdM5{~D&mDcPt!;Rwdm|71KSweAg4dtZO=!MXsB3gr
zpk!dcgjl_)2FtLZ_vhxSiF$Vfudx+FH8|l;uS2JUuU&<Ldx@MbO!`QMyKGnj;Eqm(
zaFGVr1DBrU#f>I4PRDNpv73Mm?<NskQm9d7g_F|kaThp+sh-*ydmrUaP6kJbJ^Az!
znA}~ai?`1M@ASC%zaRKSEK~=0!to)}*bLh52?>>}nf8Njq_3j`1!6CgSGn3)TZ3M@
zy7EOgA3yN+PTW<Hh~a=aIZ;4m_JEMkh>7UoLnFHu*!K&bUd_ihzZPsD&$h){`?5L#
zXXyJI({W|sxu?b}ER6JOtS?zZgnv?QQ{M+jB{4>jt7EKo*F~$5$*J9M$DWuqvklYS
zyLT#hIbUikug{%hzK?d7?m~gy>ZSM$J-vhV0?E)5&vsKcRt8UAw6dD=el?Oc+L{B1
z^OFxBEclTxqJl{5A}1Rc5B3ji?aVN>&CD2MWziv!)CFXYxryICBe><|>Cy1%;mP4`
zNJvQNsMn+Tqot*U@SF&?8*5P8V(hG01@{@QyeKg^MTdun+?<>a7cc2RgH3$Wif5mA
zGx>O%6TE2Hgo6yX?P!LyFztS`wJRAgn&oT!b*iQDN+9cddS;>5VLWXZJwU!8VCoU1
z-O`6#w636z-a$tMHtYDc(T#9cws#fxP&WbUt>qhF6Re^hp{o@Yt4aJQ;*5?19hyXv
zfxQ;WB?w+Y;+56aw%ofi4A4vcnDB4uXyw#2Fim<MFB9E@nTYglMb!hgcUyEhnW^nc
z;+IZxk|EQNZb1fYBv|ve(a@Y-+_o^^r6Ym7UkwVGJ;#v|qq>hZ@8aSzp!?(JU99<m
zKBwYTVQf+1Ke}X9R=>(CDu4^(TqE2FCIz^0;*yf6FEP5d!gNRNRomk6uGPi{AnWI_
ztoqMiOVRC!Y#P=>dftj3KB1~ge6j&)7JE7BF5_6Yc}+8OzlYd33-r!(^qyKc>)UnR
zlyS3C?!00{1}5Ap%Ch=^4>0$K_~Qxpo5bF#e2FmPy-f>!&wbWb=#}ejF8gb2`WHTK
z)?Mi6$T?xHwcrW3+Tbe%H(5Z!7NGtdA0Ph<OvbXe&CDbk#`pL4W8hIhP3h8PwEeF<
z=%eVGjTK0cdo7xT4m|Te7ejV6E{8%^f~Bt<4+rvMJ{N(m`OgO`!q$sMg=CJ_&T?bP
zk{8@Z2vS^Vms*H5jyE>qwUF2cdg$luU`)<kzvjg8l8-l54`e#?KHxXPf_T{R$SE{U
zJ5l+X`laiuJ)f!ZSV#9d0pFTO8@XRJn!j}i9Rb#)vmm1RfB`ppX0YmAS#vWmx-2N5
zk$2q$W~-nG`S+WSFhIi0Q!Q$+UU&t(H_A<iMuAtY$F(cqE4VTTNq}RhS?ByBK<u50
z%504TV|4SAPw-%@F!#G>!%vI6V>*3&(2y5)X+tJm$!v_+K@w9=E3G4!tZRD4E)VyQ
zFWbj<sf}y-6KBgRF+!**j0N#S+TP}rmvK|}+1i;m3MWWgAew1+SE0wxns)nlhCZW`
zM)x8uZb4)-Cw_-@#{>)xZ2^owPy3u)+g-$_qfu8aQD6Z7ZS-;`;i}NxLb7yWc!sW>
zSx!Hq#!+thQ`+f`KWcioFc)|Mo`e?eu#VN-|M{&9mXzkebMx%{Tv|#BNW2uY<bfMb
zUT*FOWuPGjo{*r;7#SJiA?xkxdi~0~B}x@-VRf<I-Mlt=EOtV%b<;E_hoM6gCMGr4
z5@KDc7*h}y+}R3}L{1+Xp7)=n==LYW$4e#top(zMgDasoD(ZGWyRwE(+F0iD3alW{
z^G3OQ_yq5B9Vjfk>Yw+kG$*ZQ853!_H5pai`qWl5hYEq7E20bI>9{7FZ#DFiPLZdE
z<YZWx00Mr$**^V~4N+p5>8tP(s;3bv+Fm^Z<nMp4aWK#Iy<urij_Jpb8<Q1gC0ptM
z`cZQMbOGSY3IW-wgoFg3gxd-eviVQMkOGV8P06{<eJ`iEY5vAT_$z@y?5V?PN$197
z88?Mf|NKg`>cn}AAkUu_L%ZVagYeVC{+Qpm8V%TA_D=3VFtq<9H9op;r+TUY?v4l}
zy9<FT_8I-|(puQy0R!~rZD?fVhnwF)5^1#VJ!@NZe0rAI7*J3Tvs2RT@*>0$TXty;
zF4T?f@PjP7ezpPRQP5R@LRVR^tU^si<phiwK@CU5s-xBDQ4g$feSBKFW7&bicyV)+
zl$;#4Ff<&YMtvfs3q(QdJRKxHG6)iaflt&MCfQ^&mA4>`fZykvcnLKZ->>yIE?QTv
zjjamDUA$jq931Rj)x*tT_><3f*>~^lLj5KY%1cG|g}yRwe#^iD?LJl(#=Lwzrl<JQ
zZt}VcXK}cY(#KOu6$Lw+{&$M#o{fdV!E*3LuUh08^CF;4Lf`Wr*S_EQY>D?%{{I{_
zHq7*_ELOGRrlzKsVq(Ah`>g?m(8tGTJ07^qTZ|Wr03&^1zd1X51Jso??}`j;J`6`l
zqU7Y|6=_uDV(^QFCNok;x3DtU#V2?tk=ySOtb&FjX=qG}25gzZD}<+y$<5f`DcHq;
z8o>ggFQegTB1=U$a=3qxFxTkSpwytxuGKn=AS8}B+iscKJEvHbta|M#xoU2_bNoA5
zj7mZGhg)-atZQ8J$4$}(;}vcAj4kWCn3!5|PAoXIhZ~5cn74EvsI-OH5~KZ$`;de;
z4f}7vMJ*S0>~XY%%byRnE!ZSCrVBdXAC{rH)l|rtB)d7~Ciee5f_`9S3*MQZ7!?@A
zb#-)PWM*cjr%yM+NZ^?Tboxg|;Dmx{Bd}A49#d0405=bO$jGj?CfKE=EB(0b@Di85
zKF*I+RsqA6t#T4V>yv!iGOgnZ-)&eBK}{=ITk>Ps@Qv8z>Xol+obgFM=kUmUou;q3
zBhua3*-M*?UsChXhh3{K>CAYEAb%1OrxV<~5Y=N-R>h%Syzm%ensM!(9X#95_@vz^
z9P{4F#pJLQ9F%Y=0{}gNo@HHGRLn<VM@!J~YR%GqDo3dLf6l1^Jpb2)xzylJfY>DP
zS_jxL8XDRoU{`82d(|h5>`oRI=HygXJ8NT4nuC#oKwko0HSo7)U|;}#4B%&`q@;i`
zdNtzGakuB*?LEvv33-tK<YhbelL7Eh-{mBdIio{BBU%3fhm^h9>GbAqfV?y54L#Ss
zrn;enHjGHi_AgFND9;G^xl}UhNGPGbJ@iu3Vn4rQjG_*7X?3p1mXCcoTN)HoCw>tH
z=f{4m?6z2pxn_nmet;;zoVe|a&?z<#o?0BE#6(Li^)|1vvR*f@vf(lsqin$bD#yT*
zmp)(uY!6PNTFpF?Tji<}H$H(L?B7L2hqHm7@7T!5#Q6Bt`D(IS05FFF`yfzn-2#RW
zVDwb%w+9TqDTUktJnDS16$?C8AQLmQM%QB*Ny)aUsY`I|kGgjRamUtUyD<l;fwelY
zgr*4D9igS2iAsn&UJ?C<^vfRZu4B<YTRM1h`IfGpBhL%Wr^CdkQkv;t^gLf*Ur-4{
z#C-P8SXnRjobFMDNdYxJK0dyHpc5b6Abc<2_r--PbdQiQQ&{K~dTeekgPxw=lP4+3
zg7?!^l~h!2LBM0qwP+;oWM{^0Z~iCk>nIRy5=-0x+T`Wsg}gfKJ6P`eD|hb!QmQ0j
zxS^_wTb~_R8}!5d{eZqzs99AQZR9`u`8e#NHFH7!FhC3+GRdQd-5oPNwc1i^qr5*w
z5*BLn)Vnp$xL<v2D`AA*+`CQyA{VzrY*!_QZN-CAC%uvz%GA<>FEJl8X#IjKC}LwZ
z{g=7)JEo<IKJ_eJYhk4viz;1>9VMtxNlVL5&UT6^D#Po>3U^Vc0<K=%Tt*+&fqx_M
zZn{-WRh9f&>)$IPk*a$q#ay4%6GUBh>_UrXlTf9qs*2$`KCZ?P`K1lnel|Ne7@m-T
z4Ly1!Tv_CCb43Et$-bAb5wQ6K!c{*oQ3aYy+`IFa?VG1>)sAo>k{>>EY_>S%BOV)p
z9<X+S*az8PRaJT=th2t+sw2-wa23NTt*q8p%#BU+0i`AP<3=XaOfQ>aVyXAmi2jsy
zHOrOwnlME@_AI=k;$R_IkLK4bM`3K|eo6Tb|BtZEhsNqA|Ifm#LQHwN`b_r4#5M-k
z0xqUUHvIn*9IB|;lZ>QsRzlZW(|s8j(=p{bU%d>>@S?wcTUuZLVcbWdtULm!Du59M
z{)mT<FPM}Y*DNY53@k}vV33VwWMo8r>B={?3QvooxKp}N-0M8cuzLRt{6saZ6ixxI
zj@_9yP)7h@B2$ALN7>1^mcG}K?~%nx9vB&8O!L}pm7dnGTE%N(sy%^0mX;l}sU2ha
zc5ip-s`6(2PognKpWF52LmUVL;&`;JY%TUtWwwj74Py;~V;_P_<(N=E<)NwAp7m{=
zR9@)XEU)gUKbh^tzxx!*l7dn&w>5mTuM;f_EDyoOiD5VHtq0n@nOaF!U^e(Ol2LW7
zZ)p=a_kO<~eGliOO_dJ6egn-BG31#Ex6+|BscnYs)0;Zr$n~K%I~JZ_8~1HSztPsD
zJbKX!epr7D+65IRHXenQQTpnR0r$j!#9R3|ZM;;fSAJZqbBB#S(sOK#W*anqxGx5W
z2naViC0(wiH5_jSz6QEEOkRnDahrhs!pZuK*@d%;6f7;?lN?P^dYM&N4MI(ZU-aqC
zKwH_X|B|?GB;D1OIexCdm3()0=E=O;l|XUNmu-GK3ioJ!_meGk0|T1B+TgtGY=Fk7
zDl6YcM@PrN00vfi7MN(aApd|D!bgwT3@{-MZgyd&Es<xQyI8EVi1ar(&~xK2Wr4$R
ze_Ii-c!;D;Pmd4oKc7$}EeD*TTdzLM`nE~5BAzHd3seSjkCQZ6y1<eIo>m!&FGG@X
z;7@?!x&HZ3gX85oX{Hrs1gN-nH5;A%pUKz^XHj5?-P7QoWIW+w%Slg_5J6{72b3uQ
zP&Z=rQdfklt44C2>_xb+J#-&}CmI4Y%)93@W_Q4vBK!4s^z=CZ_^qw3e(y6oI}0$U
zWG;(Izy&2IBZFQ7H*tXNg4nfgyEVbjhvCuGJlgja6Dzo?iU6O$`D#ND62QYIl|^ya
zSHuF&pWk1Od5F4dX{g*x?Tv#gmK@%|F4#@RCpPTyE#T1Pb#w_lno9MFOT<hYu~<*E
z`zq}Lm=uFD9LIsvs5-dp`CuBizx>&x4IT~t`C;^M$tbm(OQsZQi7p$TCFs0`G7PFi
zDz@U!|7!s3H8mdoXs8H@>Q^)g)#^P76ABuw*fRidB4mAY<>Cr1Kd=w_yG8~mMw0~H
zel<74e@zimz>lh_iCaGah(><p>ral-A-DGk?k)ohko#Z-#=j5REfL`N2x304ejOS*
zR28zzE>0I<;e7~cwN7H*xyz##`y~LAn}hSR9&4`AMp{+zySRt^J`Q$II{+K$%|DvH
z{T4})$s}z2<q#-D1e`~)#6W0)S)2udFBm;T0RcXKzlvfKSBKsLHc^zQ-E5=p%n_I@
zZnM{NDm|<sYaot43X;jV^B)BJ+#+?gTp9xSV!iX!YD;Yhz@ofwT>-5CR;GN+<<#$_
z8cx8+xeSjw24I=XOOs230}!16*=ss+yHz7(X_0gZN$KGae*xMNZ2gZlpNrN(McZvn
z;^un#XiLC5H;(Uw{OV1I2x+=8FqgBR%q*=u<od5h7pBDdjx_gLRRgB;K=cF2CJDsm
zbONj&FzIY;Y&ag__%(NcT?8+IA**6pcsN+CrKGfE)JeV>>)bjv<^)~{=!=b+m{?d~
zTpXNwK!gFRV}%5s7lxhRC`injTquZr1@9*DnPUGPY6Ma0uP*`%2c3rzXFZ+vJbu>z
zV05_b6*@*Xsl|k8f-f-$4z1IBpa62;7J7Cb6}$Q4=DqrI2%aCPpB1F%glQ2lDWThb
zKV3(PX!VvbGb7xS#D0`u4fimAUK-8^9x9mtxmD*uTzp9c2|`U7E2Sc{@X3vt#9No+
zsE2kx8ab)L2om>>E!6VoYD11l)AeQO`u>pet^KCWVwtv2kCZakltpb4+(*8oiYOwR
zxCM#+7VBb&wcBN{pk$z%%u^JfTr}wO=9wJx{Cis4oBCm@GZq-^16f{J0Lz-bIg-i)
zhbqYb>sS3SA1^PFmoHzwdD9R0t|alGzC`g12$0^DE5*2V1`@XAcrl6_GXp~|;A3ZI
zwsv*JM?^#<CF$tuCMP9<l{ouX=XRhb*zqH^E<Qd4Fa9LW<LdDY*|z(YDAdILJn#^@
z*{CN{^M~_tX}IU~y=!u*aQ(XIpLray@vv7tXWQxeK^nqcg%gPKeju|MjK=AjE1m|E
zWTZX=)Wk!>{x*Oq0u<gjf12DLNhVmmt%C&0x{NnDiV2bm$pR`eAEcy|eoe%O1c?2e
zOB8uigML5r2ajyx+H*0bGx7J?V?z)CaN(U5{XWwN%jw7|lEI*qqM1{Ek8>79HWxBn
zWApFJY@@A)`H_`;hbg$9W(yTnr^I((_Pt%b<Z@4HI~#~<rMu8zQfNu5YFo6CxX`HQ
zkp4_gy{%_MADdV{pu8FJ<G;9I?IW=%tiDzYJjdFbn&v@J1N*VmOFrDzkA#6<ABu^M
z)zhu1txe{&TLFxlo}Qjxzl14DKxO6o`7_`i^dxqXaU4IkL_HfG9tJLukS7?~fl*cz
zt1b(}#avl!R8$ne9sW**a&td|C0gL0>uVCeQ+Bq~pT#AiLZE3mJh+l-;pufu_b-00
z$o>;#vSv+H`#M-C0btl;|FB`_)BNff%zF0@Chn{s$S!~mpV9w7cxr2@@WLCP7x`86
z(iEIXE*r(!2`%{uq7&l6qqa$jPo{pqm_AZSdB<H8^BPXIG|GNT#Z<oA?EpQaFPbcI
z{dqxKR?HyxkXJ`1iShJGipEw(gG(Qp3rgvmag~KbW?fvI^Cg0<NsI*te#!(O*v@}4
z3)Vhf*L=D#7coXDq4}sU_RmrJ_6H0)5IvUq!xj;ilg?(WbEirt$l=(HJhQx$KOPMK
zrGE<1G!gjDd@WZH80z)KahidW6U+_4;CFRl0hlxb>K~u$(K9@}KG10t8|k;gfhDD-
zUxI^m>YPhJQA$avYiD-|%-%q(C1Td#us30)xSl#Lzp&7*w)O%p#fOKN=jRsF)nPAO
zF(6O^14EyB_w(@N<i_0GjqwtFEnMz-fy>jg)0_yhH6M^Cd()UvzUOA*^6L~T(%{F+
zW<Kkb{tHUI_<b9?ulExMhX?|Y+vLqFWm;a>qu{WxN3Ppvj^^BF9q^9bpfC>P*CzSY
zBhR!-PE}4_uX#{CkanlnaU82hPsWT;-Z(!_VDzrr+c#RX+nRY+=26^EMfo{I4<00~
zy?5|H`hNRfD}*G|Ve5))UL)TWnVr7wEraAI@i6Du?D#DXIbmuH$h;eS)g{n)Ou}|8
zR5U*NbpH1}+;y{OtjRx5R;j3HoY&yP3tOT8La)~M<9K9^5f~-|Gg1JNXjEILWM^lC
z9>U+h-4}<{nrFHr;wcbZtTBU;rJ;cVn9<TOGh68EQ`6EueDnwe4B+h2L2)h5ws>;K
z-5R!j4l(J}jgO8-X_EpL2ylVYsj>gNH{$P;1Wo`bo`5XIEBeS9vUaL$&UK&j4Jl#U
zE~f;Bg`c@~wm#;4+?Y@<h*|wfT(3{CoNg>j;WT$@0X+W^`a}2Lh6P_5^9dz}9k(yY
zUNg4K=&_pj1jXd8Da_=QBk{!@FX7RXLd`ASNC%-h-8cG*t66efTovzGgLl>F8JO(1
zQz#=RS3=@~eLEwYHHW_~yBU6wGOlDg0Tb}cOD|!cLAr?CFLg%~+pmu}zrmxw|Lf{%
zj9yxLAr}Bu44`6yhzonLFa(J6l9G}F0y@CK!^WN@qc6zFh>VGWtHGW|(CoV3T&cf(
zYXj6Mpj3TGM6^4y4uEU0;^~%WMTJ9K5AeEY(XRQ-dE-EX=<DufMfc2kKpvAzX3+zx
z6;MeReWI4%QX3iBTH|<BkY7sIn9Tvg=!r^1_W$M*(C{rNTcg!RWj0r+M`;po&YdE#
zpdYYLhJ<Sz`Khj!${q8ArVgnWMHQJ=xvDCmhPl5M`4#Be&I_Id5?9d_Q?nkX;jrTM
z-2}ZB$Wm~pxOYO!nHMSbE<BpV%z|&=XYO^Ipj7hD1?x@1YWd{%i=<=`;bCLLgAA<9
zdS<z+AB0O0_>%wKQjf2|OvS`J#Bmb|gVHBSz-7>s9gyeE#tM&rx)a-6W`;H|p{bR5
z#dLU*_93-};gi-hfToq0D~|OO;FE_GWbZ0T$a{@FK(2i*iv8j9dGutxQ|*ni!8M1n
z+`6QTyvPedPvbH8w6xj|zi`OL!|VgqE8W>y6zD~mrP0z#gP%wH;BU{6Xg?Tu8^(0B
z_x|$Ve+?8CU_^R!P#4Og7#^u5Xgt`Lbo+lmK}zs880qQ3fff@NcX4%f>}O&CYv+`d
zJW_togDwS()#WvW@URzYJR9v-CsGb|2*6r9Ap5{M<^cRdAOi-}8B|ey0HHXUy4KYR
zQ<UGuy6ZRzBMToOmk3J7-_I%5*qiZ?$xDRimEW5_?Z*rF8qj@6ynB3vQ;QDxB}qeX
z)cpU;Cr1q)-+o&b0|us(qhoJ-d$q%6R94p0TAw2zTIRN#hM=cUpR)LozD#}4Kp3rj
zt0}!i<xqgG+zOPeDqB1Q)2}cw1L(@}fq?_ezsCGW@c=;jC@)nN9meA3$n)iHpCQeV
zPuP-Xe9l`k*xoUDk0f~GZ~6G<e#p9B<n4->6P#(NxbYRs43(mJBI1?ouys{f(Fn^+
zBg5lvL*q6(M+_7KH`5Ic&#C}_?{DB3#JK2tVSD`nU+~Km<N?YwJPHA@uF1v;=)VB$
z1GbdiMj}$NAhcTxfV=|ia;z*YmKPVn#!KetW}L@I$A_<te(2>4PuZ2SpSqov<izf=
zTt&n`m}2sT_Rfyop!wBRH`nE2^ID5lg1b{-lf)c*5BBk7bX!(B3WP^dluI*aik!JT
zHZ%D+f!NS6k^S)2^fuHk1LJ~;FbdgW(LmxV3XcE<6xHLH{M?8@@hK>1c?25LQ~8+r
zbyY<lLjHiZ+QR|yl-NB~#IN<uTmdXMABm?cgi)iT+}Y1M0Wm>`tE~r2Mq7=@k09vn
zK7E7N0;uJ{`IKX3zFRuj{B3Df)+TQ0eT)5k@;}<1_ZYmc4R9g&#^KQAy_%>ZJ0)dN
zT&Z`C*NJaN*|4%2*+q*BQk#~JYRm0-?dM_DPQ9}vRoMD6fE$A?HGdw-z5FlYgz5~z
zrQ^pDrbon<sLl@`Mgl6?&B>G%;J|}RHB)9B0)h|dE}*65Wd!o3zOs_g%mo;0k;CNV
z;GjHfC+O_@jtRCXk^ys(_QKfsh(u6jLVJHK)Y?$orClCE*$kZ}frX2WnPs(m$+fH=
z$veLYNWkF|a=woSajp2O+2#jdTEVVl7}jGHXBGD})^&mYW`!{FkF|5=Txx0Oqpg<E
zO5=7tQR5|>xU=Uhybd=PpDN&Eq#P$m_ky)eNYUZlAv*H8lF}gx@b|g6UIi2A=Y&Qp
zf4}IGzb9fygle3@rR8lSO^IA9aW_6Vr8F*%kXi~%@CM&iuM|9%En~1*x&EdW0I+-G
zCy|;e-9c9g@3eI2sc1~H^Gwr=&dUgI|96WXH>Tc{!Jix8?o^E9mqz-aUe(o|vzV&P
z$jBHO9hH#$`(fwm9~lvJ+=}t@!#LH)AxIgxGg5I3&2N|t{nS@p9Uy$=eT(_?Xq8^_
zqj(K5@z3KG!*7(agXG&88JP3(jk>wzX~J<6Y9m9#aUoer1wC7=*Nzm8wgyv_N2W?6
z!`E=hq8MWsCRj=4Y&!_cX`|LM9huKtI|HyP62Qe5OC>A5HGsNs-$uG&mkLa!=dJJP
zvaiYc(RYZq!M8nYd?i|0ic{F_?%I$EL>FckYS%5BSRg)qhmhvnl6KSxa6j|wQZ*H>
z0E61<vaW=aJDu`PU+>7Jhnjss?lYLXPruF=6yF$NYKNzbZoK?Z+~dY}XJzq3eJ)t%
zsF#spL`hi#9Be7wyvl|JFWr7{Q1D|*UHjJiimV0?Yn+OZmVhs@dv^HvUvvkv2f=IO
zJJ<6=?2-@UV%b{26liLSpN)+TxRtZBKY#%B5q$Cs3W%80n!x_EM6V-($JlN!4C@dW
zJ)hBTgPf*&Te2KKz$@j&kQUi^g?(vR_3;^JfLK!Hhq2a-vZM0hkDHIc&`}4Tk5+s`
zMa48eEc+}k-{T<}*em>v{znZ+4ByJPr1cAc4|zm`u@D|+3)gBb-()aZN|!Ve=s?tb
ze^lwc0PK92PkvTD<(&;6Scr(+V5}n3i>4hbIUQ_@HOh~PA(l0L;EU0U+Xs+t#L<Bu
zl<?z3N=eP%)CBDHR>3EM0Ac{DjtLy{s~3fJx#&<K<=l>eu{=O6Oao_*l)=$Z-KFNo
znJ+PS*<iSf;?(QU&V->RFf`(FbFp>xCj_?X*hbx|OC5Ymhwjw3ozZ8J5HvAa4}(4q
zOb;TGwBLU(Xq7{@H?bLR53sJgjBi$r-3-4oL|~)M+`JC4Y&`R<7pm0z*&VPJLG<7A
z|6D-vH4iWW30btNtrvbJZCKBLq{W~6EQaEl|Lrqah7(L$%Ri+jnJA2hy$L_}<ksb|
zJ1A%?Vz<4kvco6N3MjQ*D>Z<>$Og}#Kx(DS+1Jk^o<`&8RZ!b=YLQ#W>mrDw6COn;
zbdINds#y0yCFR)?re)^mTOft5tmY=NZk`<7s2{jI@Xk;u<7EFZ=_L3{63nj_tgGW{
zd@ntAJ(9WZKwFa6Qup4xgLNi6Oib_W>qi5JG%%K#^?^TdA4Qj&1F`P{i5Yu%iI%jE
zQN`*o7$O788SV-;Sf3UvoARt(hBaY<YS!^ZBVd`BU@d}&S@)~-gy_$Hb=uU;`wIiT
zo8dbOSHjRsd67GWO-{A2!TbRp@~Y7#g5Lz>!CTb$m*zsUunfb;FI{>2&UHBr7kzQ4
z|I*bbMWp&G;k7Fbyw=Os+C7CJF#ez4F&|r3UjyYl*gk)!;CE>D8K5qth>A^A2vj06
z(*8~cJV<zb9*$J7g^NmRK4MMWHp<Sr1A%8L>~h@T^HNDAG(-1SS96)(y7^Q7ukcR2
z|J4<oky@i7D>nW{LBBG7$m&6x^(>N-U*UrV+SxhDrSZDrpc4c=iV1ojc`!e>v<3hy
z<m`!3*#h52#c`0VC}1GyRb08dlVb;2%sf?m=9G-V=FqjkZUM^UvoY?85tV=XHS$7H
zxK%#uU{}c$X<57O2+q?o<)}2290SK`URdG!H=Oc3|C4Z<_ayaKfRM4rDA~WXLh=nf
zDV&Cx)oawd^KMCbX%^~NjJTwLg|{LusLx>F^yAP33&cRCJX{MM50gXNi|DV$H7rT%
zU+TGrhpsN?CFJRbM36D*O;h-PK8$Z}g(#?3{>05@XX~G=Hs?}|4E+u_oqp&4TA3Iq
zeHJ&h^Zwl7tFb(ZaNJD4$Z4OR`kK0}U8$hcrnDWF-qb+XscxN}O>yN5ZjR9*u-$TV
z-jNTyNx(1%v>G-$EKz-Y-TDxNv3GVc?=FAIN3`0^3bWk<M(h>%((6B3)Ng0RF_^0k
zV7Lcnaql!w{39a3KyYw`)=*{7?AaD<hv5md!FTW6c@br8(h*Lp{V@caM-2kr4*x$c
zLpTMQlBn{)=20P_+kxy)hv<^Mh`Iy0yVpgvUPQD_$j0+XBwP=oOOWv1)=`-IP2=9x
z%}7&@&-)`CF?*A=O-bTkuKX}};}JM|S8%gXks<P)(7gUFDDVa44c9@||A3`v>n_(O
zcM!s{5+Umv&|heD%O@|7k+5m#HP7c;XCg&IpCbh71fc%O$>tEUZRVinWUaej{M|?i
zFP|_pl&8jtQAz$r1o&Utu3R)9!iB(WX4D2Rx-OZr!J70nSar~)pPiQi7v^Sj%7PII
z(n2k8Y>1nw=+D*X%AMm7Y%v}I%kY`@_`Xlq&5IMa_jK8@?#eKtyiZLuwL$~rU>_+u
ztUG(B`%k=u&9weSo{Pwzgxsg3CU*n}o56<6uU!mSGUF@yU91eKkb|<gyek;oZ`BPn
zf|1BmUNyiQz2!L#{~F9b`v!(}R?Qz6^EN?sN^)})UKdDHYexO2fsDrq3$o`_aw^=n
zFJ`+xzY12Nh_%@X1tkv(56)GOht!=5@GxgRt1XrljcxJ%#!b11^3p9PH7fTN#`QPl
zk?TXAJ^ibyKL?Kh=au54e(LJFkc1oYp;bR-h(ix^0rCp|F*vHU+dh)zK`F$^^#By?
ze)KY=>)8c~+%7Sx*rT?6=a;8CNFMsMh7xuCUzAX>Lx&aIo$zv(nU42bSxQSq#1woV
zq$rDMyb>|Wj=z?2W3RUi>B>d`g4ZDX({a8jVkhe(x_^NM6*cHC2ce`E_4K9&q^qtT
zP!ETGySk6b$>sQ@;g!d2kRm9BX^oc&LH*=|c01;yw1JGPybxqki8Zh8Arl>|+TQWq
zhjtly%s|Cq-yXtz@v-v&eC?(acg`DT?ALZdU0OW;J2s-NWOHGE^Au|8>pNXK_<54|
zr^Ddx)1kGOR;M9H+?f}T4hBVEGn&o8pWBL(*mDq{JolU-YZAdy#KW1(7*<XBxx$>w
z-9B@9J#~4#2~p%;P-AUTf?&xt?Prz2t)47Jwu2H@X#-m$4^tzL9NqS;DL<sKMe5Xg
zH$A~d)KC@TlRhz4iPWXOdv0t)&1uQ;wgURH`!wzaUVvN)+u#p7dvaD*g{(su$k@PO
z0GWCmFS&PZ@JKMr45bbXITZtdn{ZcDJ8ReUn?9iN2DGC@V-&^}dA;0Y1gVa?HU?zm
z+wCQ0V|rWh4&W>h-fQjnPSNE=81*frjKl?bl`N}#5=A%u6{wtXU7s-*-2yIJ6DI?l
zK@zXe;(m5;czi%g!JnGp1-i+gqXWb)<;f2OlpLm-psj`uMM#9ugts%fIa3}QuFnGY
zfAc3~&%noYFLYKsp#`_#zS?J00qzG;Mc#p*{T82P=E=LAuS6^5GIlW^5OcrMtf)*`
zU!Qu0EbcR4Gi=_EzwZ=sus45Nlz6xrH$k-1AO!DL#Q49t8X5xnJET*~WzXtQP>xZt
z3Yw)B#Dy~fD3p-}Cc77=skMLZ*1O#1MJg~;`@*vAS{OkZsjWtsnBcL$5btoP{o`!*
zuS^N^AYy<RSO1kOp~kC_qheDQTq4U$gB8LXhwGAcYx39n2v=np*It(*<Nsk<gtXMW
z6KxUy4Bw8AyQ47onJJ#%CuE^xD%K1S)B!vs!*5j6z<l9K@gRDb3GC^s!SL8ECR*|B
zehri4<T3cR(yS0g1C>p?G?GaFK|ng{YCV2KCMpy;dA+){bz$|S+>+1z{$}zkJ@WP^
zsKe<4)b;Lja34!IGlQz)M?1_v@ga8r=*R?TrDaONAA9^?Wl%KUU_-yfeTud&^}h{-
zh>~Q5RK>pdX*ALBTDbxTo^j!I2_(^-@>@8kja;EFx1`XbW?=G<<WUP8|J`;ez>`^5
z!$RFCrl5T%q!|U*uXz!I{F+}dt2bV<5kS85opsg4`u^F^>)khazLLcE5qwh-f0_YT
zB|m?<=s4Y&b*VNs;%E1@GYx;}4bX&uhl;8xEc1bT9HfT1GGn~xMREb>&@PJ5<h6ya
z_hIl{Vii~%#6Rjv13-}^B=Z$wF=g_?*A(80+g>rzOAl#Yq!OYP#KHti-=Jc%pr;57
z9*U5%7wqmtH|v@bJHyNTDlYP{vhp{uyc1MQ_{4-?+ZMBjd*O2;*~C3M8W6y1nWhEW
zW+v=E2L~?o1H>D|$z5t;)TU=c3qU2vxAt?QcX~5G7*xg$wx_<d?8h8_Bct5zeciP;
zdxoIyIk_*YZrGD|8m2E1>mPvW)c?@$T2*CC2eEe`swCj^3WW@Ya60oo8t2bC*{;fP
z(gI@H6IA~RYdLl4{eulYaDekBg+8~9_9=X5^tfrc91pz0htvd}@C9AlV%BnsPVJ}c
zZahRxmlvDbT_Uzg%=USvy*TZ}rf!mYZAh*Kc+eq0#9#ONUyQv~SeM%u{riH1G%B46
zC=E(VBM1mccY}0GH!n&F2uOppNJw|5DBay5-Q9H-`*;55+?<QUv!8u|dkfz;)|zX~
zF+YPtnuZNjZ5`6mOdlQezZN2ZT<LlHYLhpxxyl=Jl`9M$vMS$OAZ1y}9NUL<9Z56r
z#XtMsk_my*LuMTDTr?2dj)ww$0ks@BOPuBTI$JFAPu@ZQUv3lYKX3hHcQjia=dOat
z8Iu)<fJi6^+9Xq+ekB%XQ~v6^RbhZ}s#WFM2485&Ng8#5wFPvaUXGD$n!6`YP1>0-
z{r`2mu-w`96v7b1Pk{<jLouPnS7?SN`4{`^p97>Mf_lFpAcO<70(-*};Xg`>exhb@
zD9&ERLujVqt3<e7D}ptkr&KhrtE@f45?;-%ejN|_ewo#;R~C=xYA8@~`m8o`CVW;d
z`Bu8Cy_?kTjIM6V;#BfExLQlMd7C7>36!QEo-~1~>H!q6m4iBa6=IxNao?H3LGXEA
z_JVS%{;M%*pvIkacOR<Hc$Mr|vh>zc8CYq4BA?Hh1xlKLOClQwz}P6TUz1HC&1=;X
zGy`XG8B<YF5$)q?G7DxVrkR=R`B*~$FaQ%&Ha$4RX|a9Iq^5?>H;diH6F>izN|A)V
zVB{Q`8qwd#@1#CXm`4RdtHilBFT4d<tctP*c1r9B&|5#`vjRQ->CuQNhyqN(Yc8xB
zd8`z|wwkKm2Ij~SZxt23DXA&MudKdqnTw2iQV?c+)!I3Bec2iZG?f2U)jLdI!OvzN
z8HId|9cI9KA^a2<gaq;J<PP6TGvwyrkt}=!mSx}@@M*u_X|)-H>W!v@T;{S}-Q+2X
z5NwCmBV4XU8jbOn@zn;RN<bwDqeA#@8;hC@+hA#8L4}-*H5Q!{o+9|9hO0@dEGkKv
zc)xO3eTZdD(9K=z^@B82oUy2om8qbIfetI^=5TSC+nKqJo5k>sa;XW`^Zoi9e?mO=
zI04=QZ8?aoI0B<UZ~gtJs?fq^e(U<ZEfZ`lYAoTQu7d8vT0)OGHQ>ts{Ca-RNkYdq
zCWqdgfz_rcN8h{bA#^{(#%)Uh5fKqtO%%7x&FKPvI~d#l@q-GW2|*lSOG^s?ji6$a
z#S415_xF%&XsLx+)z}+Js}&8&tMHdPA>VWZuMSCQfawvD=Q?gl!xn)-ku+m2qosw@
zsLI8;D?%E~(HCCw9N40j&}JW!hc@9@Z99G-KVC301;^OF(#?e|V^r=6xA$@Qr}4A6
zz{SKA*nG)aki+Go_r{k?!gApwzt7d9l5CTeCbS?shw_bc@o%^NzvZ<wKfE)Rxk=pb
zhkGs~I4Gu}`IbwN5YJc$aF%XksR@WarAZ^%(SZN*c(tjM?XhTeGKa8<_eO(Sda5E@
z8HO|(u-CF|3{}srH%Wtvk@@TKZ=vT(IU{~W#2L(A<4Jz5;eioP{i3qNV03AQX#Dv8
zTClT+?aY^O3`(SC-B<VT#?zi3Hv~-6I0OC)=mb&YJ_KsmJtJc2hyd8fOifK;Q`0pt
zlDsof(mFgm3|xlI%|6|aBHqa9ikJF6m`44CGZ=(tL5pgLBGHxsdE+)OeSBgiv)6!r
zf)mn?c>eEj6CBD1G@$v7z%dY02n|SC5#BhT8Su?kWbet&xQ5-Zf><VYZD}zX1yxZo
zT|J#=uP;O7<T1CpUsJ^C+`-gX9G`zgg1>OZ{+5We?9pOsB*Ie``U)?Ik|e6*c!eLw
zG>Z#3*$i0oC9wA*=ufb~jLEVvE3_`2i~YZ{TL?4;L8IZ(LnJ=FI?$Xg)ndL6FatdO
zw6rt;)}E{VJ7{l)MEobXo40>ZMLGD;73(p}DK+?^GH_>4KkF}fm2WMoL5v;mh&C{{
z78p36PeulR|IJ3Q@$aR7W{i4SO0u=G;?$s}Ae1KX-maS}e9+_)V1U(CL`6ki$)=wl
z&z#`^Y^xkVk9=wPauqq%NT4MoWbY@dU@oR%jMWVZvhald%0oZ^eff9TH<M-evZc2Z
zkcnWgruetvn{)+)v~vv}s1OKF-;NbG1PCwSVy64jy@+ds+gyqAaUO*jmkFyPGawkj
z2X^I$>Mw|Jpr~ze=XxP+83h8U6r)R1Gib*WRJpMkx;h$#rlwc?bR^LEh3g)L-ky>5
zA0nLx!r6=x)!V7>H*O_E-xpi&L%3(~^Uw_}qy&+m%WHg~NhC~!vmy8tPtaryvw#SP
z>v2a2xQt(pjc@+auq!O7@o(Ogm{)pdB;|mSo?2gVZ_`1c)@-`4y%*zInVFFxWPe#U
zlY4fEu$vnXaCmn-P0P%ka7#2GPd@R-e9G^7jimJi+Q+%l*3}`rV}?n2wE!6c6y(sr
zh=}VxZ@t&fu!{mHcp&}U?76Zs=7M}=$e8EXsmGK#$JBWKZ`~asdgi$9+l~7VAU{74
z9WiY(g+{*tEWwbzt@*aTF(L%!5rF_sJZ-dsKRU{d=#yC@ND096p4ZX&`itrMiw8_T
zCsgoRCLbQm#LqMn_e%V4CjKaJ=~kuIh!1;LX5Cm#o99(hHhng(H2uM8Qu*0mLfGuR
zyX>d-!{l1SwcP36k@8dl@0J&rZzzDqH9SrcmQbT=c%1)eNg$hP`hbn@^Dj!O2*2K_
zgL9m=ThO4=3fHCBEf*g4K7Bb&e9At4z1QymBuALZmf=9F4M-jdwM?pI$>!`%E1w*6
zIi0XF@l_^pHQdCQitexSH%gC%tG}?&mDb=e7@i8mBu0b)Z-<;aBq#_0lI(1$FCl*O
z+DNEN-6tH3Mw>mj*8l9kMNfP&p3(?<O{J0lvQS9K^$ru4=w$HSp!2InA?KfDI`**p
zZo;%Tlt#nI({9YMwZ+qRJV2@yr6;ecqN~%cE=v<3H*?sE=*mFYzsR?Kw-7_C4)x32
zeYNTl*Jl-cUZEidUQ;9psK7v{y+3d$ei>NjLfaS0{%Od%PP$OgD-K5S5uu`ai7Lvx
z>U_3)$4-_kk*J>50SkYh9lU!9$ZN*ScwAMu-?bi2Ufh}Pu`K;C`wjUBxu8m9fAR70
z^YM8jiB1(9|8C;)gA4FX1nW%9qC4vwP<=r}p<)<%Iy<Y~Kq=mT^cs5~8*L4sjkf!C
z`zV~sj(U!abgD007yL3&`{$P@)~c6}9}V-*iF=DaLI9o`y$?|{a24gzj0Z^iFx!V2
zPm}5ax=dF`NlRV(bXTcoo3rjLsebBs{AX5WpIcw8bX6+I-avr-MP%M}Rv+ve-<4qK
zHQn`$o%<EteKDuF?blsfCLA3b&HQOfS?5`GHzB6*N;0W0)PN|!psD<_+#@g<Uh{%%
zt*kI1Kv4?{3WBjJB&DPdPfm=v+KORff02_3x6;_(%txnm9(j!oP7_?R4)edvR&&i(
zHB}Y<9?&@+qDj{)l)b{>sx*6O!{~nd0&#Zf>}%54u5?uV*<eJ(V^kl9Dm}<YR7muL
zA^rS2njh_$g!f*Ar1jmhU7x)IQv(;>`q_`?UDfy=7yu{-HG9>vYdv%?OY7NG8Lhz4
z6QlWQoKa)PH>?W@rz_HWoX>b%uy8$RzjnArzKg>y<Gui(SE~0YK0vl@FQR75wYXjQ
zq1^}H+N9-t!3|$C{i|Kq_g*5f-klH`oKoFEPM>|L;^ovPgnR6v$;Uy08k-H#P}$`a
z@IY`q=9T&#_tx$(F2}(x+X#xa8^K|V^+hbm4@G!ve%<lOW32aVve>NjNS{sPn|QXZ
zko|=!6qs;5+kVr<m`8@Hp)*mn+cKG)xBC8R6f(lX@hGZjcD~eg$|Bp(yOo=iGg(``
z*ME_{P;pY4I(u^FbI2uF;V?;1Ax!PQT7wsBbQ)2)zLFUsLwF3<izt+1**tQ(4b#<k
zWTCQgdo*Ruy%*<|F%%B8o6dW&ddh63y(uqrv0GnaUb(zlw_eF;4gKe)W6B5LUV{)L
zLPA0?U3&<I4opl;tgyh6&h4~00xA{=)F`~+ajnJ91^M;%->>=x!Uxx-5kJB5FxKW%
zM#Iz-jN64}WndDmZOF-~NcgntFJL)9vJeZRhrT5!3DZ1s8X?gm*|(*76y(c{Rft@l
z#ZENfckX8NO>7VAsLh7YDMTbw*A35pWw1!y9qm;_f%J>e_5PY=gD0=@2Ec$iIy7VY
zx}SVY-*U>~u;Vsv50y4+g$bUui=B)Ye)^tdu3ufv>mo%OdO3gP7n8o5NiO}}LxE*2
z6&P`ia{#2^k#1jqum;xc&6*zjKH!)}=q({Mx0UyYvm}suhDwK>leKHUq6oe4_wPV)
z*X%Krz%D;-xGJyB;-X`+|FUCyx4kRqg$04<fR`i}63!T|BC7@SABvIx)|`tBWS7Dz
zc|Lw#N~342B*2T=<=tI--4V<@jT|bg_sa6RYjSBYuC2iloNTyvg-{`$3;(<|8BI1X
z%&F0;@InxG(!a5VhtA0>m>H7vUP(7<q(@un!EE=HKmyO^4pS&W1hTCldKWXD?7L3+
zO(rY|6UA`tl^z7kt0~lb8?l$Q&20yTvGY5l_=|BldSl+--}DSwxC)`ZuNmA@`^nNv
z-wV*z{9np}F@#G>Y6)fzY;E^8Hr#e+YJkPY!`*!f#5*M<{QdI>^2x}^0O?PZwO<W(
zlw`dly6dsbj#-4Mb~FXn=6)4qY!@u^Nj~s_{o=obn(sZ)Qq!KP6Mw6oQ$w3QOdP^Z
z;YoIl4Sn^n&^y%#e8GPAnepraxoJ^iwPW{ujW!?W_F9^bs?-nL{;U^biJL}e$3No|
zGYYbTKj##G`+}=ckazv_(G4Z}*|HrlT~>N|)XKSRGO#kzJsrr5Ql_OxEB8A3jb>U;
z2&<EB(RX%7A|nI*mIO{^wTo-PRk^|qrc)bE$+yRZ_*59qJBNYNKS+3+)@DkaO&YRz
zrVm`bU#zm7{~YQ0H)LP_iiwJ(mz^DT|E3oc@{!7DxY$Rw{*{v}-aT@4y4%9rT)8<d
zt*J-`Q8Z#2Hptt%kHt;UKp)AcD6MPAVO+9Q&>ED^4mHRVF&$Cz7~Mt#ZP*x(6$+0J
zYMjRYne1)FHgDLM?}=;W`dzknqRFUWNLsERz`2R3B8q3du(kYa?vls{Xn!RLiD*)w
z08$=E_`MXf>AA-l20z@>Wl#OXY04Y`R$&C;JXX)WV#Wox)K92B0AUD&4skM{OPFl-
zeZaSOt{Vul1#fI?AkL9DGp1$she$<Q=GpJbY7e0gRKIB}3y93jk)Rh16QV13kZ+56
z&CYsqb`e5acwyE&g52s`s?)d&O3hZgx{4Cr!>A=Pmy;1iYcoP9$0=sMTHjNN<>qAf
zQ-CvlhonROpBXz*Sl-E<>%n^ute3_`$*Q`#*iiQ!X`|a94Z<hAJHFe6TC)Cd*FN$s
z9v3T7S}PW1tdti0x1K?yBmGE_Zo?H4R5}#tu4$1IYEG&1$RK_5`_#g|95z=LgiD<4
z6Pu<*bYdis2n8+c8@lhjSs4CyhR1;Qyy}5Zoty~d^N!{<%y<1xa-W3(WdDeX&1jQv
zo>_o3moMysATG8kB-qa3X6J;Pdqm>X!TaFD<N;b$@X=kIE=ofB%q+$rQ7`M&Q1(}W
z>piY%aR!tKX(35pMd8yNUgN)g512wznr;XFFn$1=-yMtg)=%U8MV?w8pXQC>`-{E3
zxtW^=`X4j0PjRE8qhA0{I61l3<*_-Kt(YtFH`p1>b4iux)eIEh(^)p!P1+aFaPdBv
zaXIa}5rziF73-YMe1BG(Wd!O^H2GwbT!$O)7tM(y_&Cka^(Z?Sl|KDWgM)onf+_gV
zw4=fQz&2_Pdif#kITRhE79j3vU{im^futi1ryqqs;_T{68<de$_G=XqWYdt>5=Dj-
zx1QUr&jdC2@s*|KvoF5ONWj}1URIv_{LV-SM@nj5M<!v(s1Ag*I246|=jz5m6W+X8
zlL&0B(j30LyabLWBp)!r0;pFY!xA7B_|4jjEpc=Ur@X(4WFq)f)|SqD_ZOnu!=C(i
zW7Ki|+AtEtOTsXY;~^>e<>+7;?7TyCc)4F*CV0%+V_N#gcP}U?)+Sw3;Z54yuzOR0
zJ~k%iO?*;HOl(YE?g(;b8X;9w=yAoQ?8JPJu%dQ({tt$ybYy?$VCnOeZ;=QPi>Ot6
z?*OE(@=IEjTuC7sCcW74Fky#FUp_|jbX{Q4+e1K*px@wu(nkD`^l}3#4Z&?AQj);Z
zv&A@Xij)x%`8_ZoucbvAK34*h`L;|rY)&cewVEB<7|$Ht{<6L%Z`k8=Sla)~2c=EQ
zY_vs$=Jeo8jB-}Ad+miL)u`I;E6Yp=J$S*QRkNe6ZSQ^)J8Ml#{~fLo^gb!ftsV1f
zN4sfJ;;^Q5E~Btzx9R*R6R4=ZiYefnqRini>##|3f8#8$l74U-z1WoVRzQ=&+q?VP
zAj=8))!%(S9orMOQv7}u_Jp7iV<f+Qi|@TdQp4`ZD5}i+Y&{j7t=AsOKN2j}LI$f+
z{h;P893d9Rs{_WcKsfExOQN?RRR-6RFmkw*z;&0Mh<E4GxB1c|p?iTk<4Fz&NtxpJ
z3Uk{rlbc?w0a`O}3gcq--AtLnxMctlY@VxSh4``EL%dA}j_Ys}_-A(2HrmDEB>9(~
z*34BFc+BZR;fod)mNnbPA<OMA!N!8E+2E<&UO?RraF-Ua#U=>eM#Ur=sBtiT97ID$
z_YiXA^#W<!l9JUXOsD<nLuR{tvfBb)<X$}XX_<SsCy|doc-H^1oA;=b(?te?jenv9
zw^OY3N1Gk354k7Re-p8F&-b4f0$P7=Ef*Jk@A%z?Pf)jWO7-u9Z{MeQ*J@JYcKMl$
z@==soFFoM4P>DA;?abacix};qt}e%=O*b$j>#W<Q0Cjd{n4+@csUU;1Z$Hm8nsn4~
zSq&xd;3t&YUtQ&>BYrA&ne`+@wS<aa4&5o#@vU3mh?d%82zrJ+fWWuO^agf~jMiAs
z1ZUvwAAt3@LV%}NeG+fTu8a!q_=^{d9{ba<9F&%o-^|5bSWx&m_B*P-#EWdx^sg>G
zi1t?UcwRPdUZHoN!QgbbT_B?#S7?^QfG|E|Z>~`T#Pi&VB)(e@Fw%+u6<q#eWvp7*
zb4CRlw`HJagZoTTco^50wf_IyR<*UYT2(e!=;$CODs3318393B;2WbuL$GF(w}2jt
zpTUlZu~Tn!3?Iw9S3qMDkInf3bGIA)q|Us*mK+y(Jp?fXA99D=D1NhiY3!mPnXpxl
zmB51xy&M^{n$As-Xo%}gUZ$ge)~5gRG_|2-sao$Q!<p>piY)Ob*tfh!^8-y?&M)&x
zc?KTKH<%SYpbI}cRVicp4cjoKzCwX${)!o<CSM({FMqyEy4?=5L9Ra(bc~S7=#-M~
zBN6(Sb5<%~xr$xNhT=2QGh>IrJvzquT1PTnKHFnkM2^6@x}QYoH?}qHqaac1UTDEs
z9SrBq_Rn4C%=SLM{#xM$Qbe|#Q%fC_T~_8Cjx*?37}!jQ=)yu1;O>QaI1I~;5c%Za
z+9NOp4n1_cyRcs7F862lB_D<`L|-WwKo4DsZx1Z_NrKsr`l&6t(4lY9-!*hv4Hyzd
zGSlCA%Zzeqx+&<0p+S$EOw)|Kt^O&#e!HuGzkr;&>>EH$z*JOWIj*Xv_P_^NvjOsM
zOBY}wIlk(Z!hYGojYMO*%LI6Qy&9B6mye)#Z~uWyQy(W+l=>o=aGVmQVA2cSi-nG^
z`MEs>1g@@zqUsx($FeqBcFx&X<%6IgI6Knhf4+a%Nbx-v`glAv`*Vb;GDDHwLFQ^*
zP_W5zbNv}`r#r`r(^eG_36u-fwwy;}LQfx~(SZEJFGJdo9@0=nyrsFFRCd-OH@>pd
z1GouQS67p(Rpa3#)U(88BiMXi^K5bn;VCukJcx&8_LIu!)o8+LSV)YMb8)y}-aC88
zhXH}O60FOFA76tqow;SZMg^mzRknHDM!Sq}C%16u@fe_o=;-gADXv#f71~o*#wOB?
z_hI1NyW_7)BT`52C=pROHeQJ_Efhe4*fa*J39k@c-+p;dVFVP7aNU6P>myJrKVS~k
zn2tZ*MSwsF9%m!;3l7-A(FJ&FL>~n?LA-P`G!*w)1dyh%Zr;IHsWa*3-!J+R-5snu
z`H#MGWmM>)-7Yynl%%8~CgV<yjxgeb2TMy!H#av49+EVQ_V$MWri8EnI6(7_UP}k|
zL}P7#pJ(HwmRwqDqu*Ck$<sTpoHzp6bn2_mu6qxeC5SLl&^zR0yQiF&?L|X}%Z~cy
zy%C||c1%&F_4hjX%g#j!(ob>el&3D-7u(-b8Pr`dp_R;c4c@F-g$rK5bLzq#4m#C0
zNPi!?7R7%uVNHmAPdY0mivrDgo@?F-^Cp*cZ0QRRXhG^#`iJ>;t&=3`3CGtOb$m%q
z9`^31an(kkmi*A$&)1?UZmiO~h*8y>w7M<eRPiI11VTc77nO|KbR<;bvEz9m%Z5eX
zp^ZXGo{9*)rE2N&6GevkNQVX2=}9e2eikvmeZlQ0izk-)J;CE*R|z8G8HlXV=7*}W
zcffv640-lx!<6`~wOhkge#!0EkTAf^z%0Ir*NE3DOXApXVcklK#y=$Q^pv#LN|uO%
z-J9>1dmBxG3(xjmk^U`c4sg|5^xsKJ3V2_;Hv1r)kaobGodBCz29Q=@)!&<I;0H&`
zsZuaG`?^mv1m(wJ@=sgJqsl$g-)DKQ+x%}bhI=s`=VxC3%%#BrgYrV?H$=T92O->)
zgKt;1(Wp1~I5#qVgGB=24{$Yb-(-$AkWPym`=_4#aI1ZQ{rSE=3JCs@?Eh?<1=P;Q
zf4{<`4z$=gRUv))`4jHPyJh3h%inz;(i0tS*fGAn!{+irSlHOYp-X)FIcq-X?dKzX
zejnRmI=sylJ~`&3H5&oXlKgrwGxPR$#~am_ahyF}zEq8xP>iura0fVye$s4KMMOld
zXJ&LBe7i)^BnGo;2vG1hN`aBsf=^ITxi=g^3W3+8y6;H4VrTsX^%FNGa>1nAZ}YL*
z>mQ)iR@WTb&Hj1pIXz3QWiLPU?mFd_Rqj>c6L0WBkT-tx_~RIE`iv6uq(Yv(vYp_5
z#cBI78a3`T!Q@<jdj<+GWFIh80<5;2BLVMUfR=&>zzU`c!K?R=w#h6e{7|>?gpPzc
zkC9P(u1QxkQ;bb{a04?+m$&h1JS6Dhl`9LGz<fxHKGUq{`3gQ|iMv>~!Y=oP>=4)w
z$ojJ~(jSwjK8}lStJEnY4IHg`D=U_MLG_#&19CqZSRnc&^t<o#l7g<_T1r!pa4M=l
z+(l3KFCIhRFZpi;Dr<Ka%qS}4y>rLzNrJV&9O};CX!Ftw8mk4bOT`uIX-e()C_TSQ
z+=qT|RNE+yXtykXao+Iv81D}VU-N&kw86!X`XVS1vq%a&fJUa~$1h)ELbVKzT$}?1
z^KC6{BOZTydQtn*QE}t;_>FfX@d#%toNwu_yk(mFn{1=Kgm{zTk`LXBN!KE&shzJ8
z^dqAhKBcG{IU+)7*+lmjQ&kT&`P$}*s~2BNWF#Wg*48$YXl`lwK-klhKRnJ-LcCWu
zd{b?Q*Dc1Q+sQ*VBNom!>}4~`<=adI<<{@Buj121^Bw_{b9-|kO#`A3U0#leTK%6x
zf(uqYr=7)rGH~F)`s~lRK411}Ei-Ub^>XAKTqlQ^LXWe<iP52rNjK87)S~%w4{>a0
z89(-A6RQ<vP9urGl<4GYLiHb-Bk~w=>3-1ME9}i@y}6b_ig@?2bPqJOR7MkVv==7w
zo6ISrpYbo4vRv8*{aRlm&MvY{Tz71{S+l(i*QvIErxjM$XMu{bgyF$*nQuD=emHoV
z;zfJCy*#)_@|!NgadF<fK`8%{jA1hPI_?=4){0j=kN4Yw6Vc&z?7hM)CI%`dDyE<5
zKRAq!zY%()PMxol+c_6E!p>{BTI|I1Xy8~jSNmO%lwe1rDy_I04+4{^mjAr|lNe;+
zE!xz(i^0Vgi*CuEXrns8aTM5ZXA;4`-b67DIKK>JowK8uk&~9Gso?vacg6Z!zVRRc
z<VQ;+1v@!eaqxQW><$ycS-B)!&DR}S&*{unF?g!FeNViOYLUUS<9(eH;c&5Aw)mjO
z8dfbAmoK(alBdJ-(PNyuTV-YRhegElEVkliSHpy#ZXd?8cXkRg7VYaO_kdF^EzhLZ
zD26uy9$s-3R}KFDrrmzsPM*o5hkmIQb;yA@FW1I(^ktM)6`a-%^3GBXgQdFoI^Uh{
z%(l9H3H6%@r!+_LJ=i`8Om6pVdhTgTF;o@4yXh8H9a!P;jE?RbnSH^a2}~e<Ik~sT
zb&&MjcxC-b+_Re+j$wI)sagwFCE0kQnT3qG0Qct%RN63M0#iqcK->|d<E|eznAfwq
z>piI$>D$q59izBWQHb6fF^PljYYuhPwk^#1*jNGx2?fR8mVkl+8`A#0&G!=Bf3tDv
zE$X{Zu6JT|XDE~zPe^iM(isI6m2p{}8ebmwR8aQ)vS!qB1P|heA5G6=pFAR%+N;48
zAlCwmgeFbTNP#6G^m1S;<>r*&3}^S(_C*NKxxpM_uwj2kpY50%7jq(4XD0<zS=SGA
zI*>$oMnC4i`=0KhP*vJ_Yn&^z51*ny+gAw-^naSRcZa_<!eM2qd;H#vPtT1f+7rV+
z^E;F6ZN0u~36Y0fPgpEEqGOa|ZH>d`uudviwzXXLj!(hld)2qRGobJ`#*#3s47BiL
zd-LgE|3g?JG+iyelVT(K3RXlYt*Y=vw&<(-r!0-xINkQlu-M&|yt%Hn7O@%77WZcB
z#^n<i7T^WIw+0CNbWyi@Z9aU8i=C;~xPsXbp+NNb_7|DYz5ADkwhi=OmmSR0JbQ7-
zv(U5V!)_QbdE2|SiXZ#B|KzP@vE<u-V=?R(x-5l3J(!NBVdN+fC_A>!l8tBm(PgrY
zS2&N?LS#(`)*VTb=R~{=KcZ{p>Q!JDy#Dg53dW=4S#*Irl+S>oU20ubby?ynQRGo&
zZWw?l>?zf@VTrSuYT#4zOw$n;_xRM%QAAUf0_@0t502o`+3aUW<9mi4ZAeh2F}}mt
z?xv8<U^TLCs$t}SxV2ME)U=uaf0bpi(|pt1u=rbX+BEOl+u0@_IJB|S-AYl@N^uJA
z_0g}?0m41_H4z~#6^#F0OP##6!mVP-ZE(Q=dqwyje50+S18Dj&2^Z5)c#A#=bOj^Y
z;6*YvHg<7wp<o@~&0*+tiH$+==|yEFh>5OncDemlSsScK$YW;p_;#`TC6trOP@f52
zp1#tdWaCUN_szs!$1^xN)i4SIkq+y}VJc?K?v@(t@MDq(^R9Tu=c7eKy=JlbIX}NH
zZ3Y~6ZKjB3(QAF3{9}H@uV=JUifTNtempHFGQq`$;pHz;y4Gs?k|^#bRqQW~9n4j3
zaq4<pS9vz#w3=)F;#uiJns*ztPe*qxv15aF@QFaYw~=yNE(TUnE|~VpT|||{O-%|D
z9G6$%tu_AJorfxcU236(QPjnd%aEIE_*iMSit+*UjBOIkw47g@N{cDXwY1?urz!QN
zd&d=i(O-Z6D-G9Ma!IW&9T-Lhsb|XM)2(dxI%@vt%PHCCHA3~i(s3*rT3Wzf!N|aX
zP0owAVq>ZVYiQJ1O|k))4H#S|B_*}BwMF;;J-juBRFAy?bX0&Xprxi3%~S-$o(wQ6
zm3v8@GB7lx2w>}-oxgH&z6d`}5`uNbSNAgiT2Ryc`3tC<=nx2UO2oGBD;_<46ogc+
ze102J954km>NjQIsZl=463rb%gP<*8TH;QaRDuh8!*;4R`OT+ev)g&EtDt{Fl&>~@
zuM&thED@T63))5IU;_bA@E6QD<5iIDz5dpYJ>ZR_y+_0BZ&GiIj8yiwe2>qo!CT~y
zTNHUF%^w8)-d!b)8sz-Cu&Y_objll8Z?0clQQNWg12>eDUpVQy;CX{W7c9uhepU+K
zIo#Koh=eVdi77};*2mAXzLqb~onZLdlG!^K+r;P-?AjnBnV+QUa~pJy-xs?MPUnd=
z6nCXyRPl5(T{3;hIkox&yR$)mi%n~G4fs_;#j{<urdGC7MiFTgEzcg^ABpQUy{3}m
zE<kSbCZMXUuAW|=SUwI?R)7`mofO_Tm&rnKN`;2Idu>e2=zYStk`fp$7ZnlVvYODh
zwbfJpBN+8Z&yx<_*vyQGl(f!lkcuq<6WZ7iGEo4W33~?zgyxYE4h{~-1z}x@hda?B
zzMQ$*PB#VviC>C(c`IwPPOJACBG|__u~Kr)`Rtni7&BIphg`uta{lkNDwGC$RC?V#
zt!HifJ%9`opU^hRmJk!GMwbw-%Hfc_Tap(f@MbXbmE^Az-_!`yPX<t0M*5G;xtMM@
z3or}OkO71Z>2c&>lR}{XfbUH7?XMfZuKmS6PMOZ^<b=nfv2GYOG~<#~{81;(L_l}m
z;QP|HQ+A;soC9mpWT$mF%xfgtNdx)GVoJ80q|EH3Fv@Dyu(sNzx9z!7s_umKj8bAZ
z^Bt$x)p0zhHxjf=7?VHe^w^pP_u(9)2`JEp4ugZF>j|wU;ML0v+Gl%mx-~){hYmgZ
zMG^GJsuxeB8@b8dDfKjTj_B5P6g9cN-@JH?W;O^PR{c>PcS{NPofo6^Y&zq2-0UL1
z>YK?C;);sI1)UEj=+BExp7(b1?#qWagz4Zo+^gN-anuq54b%m$Dvv*#@nCZ|gNFrc
zp@q9`tm$EN2uRe&9y_~XJT6lvUYD^GW~6Z~V^1WoEC{JZK>?Ezl3nS{s`<??etii8
z_5keg53^NOr5{Y+y!iyyCax^6%9?)cM(=l`{-Ozz|E-6aC$w|0H^+RIOd@D?3>Ros
z0csB+9$s{8EXC`OKA<{QFZSUXLH40n$^!F&0OkWS-d9#u{<gPoZES!bb5^a&uW%7H
z>UDxoWL%t(>;3{r5C=HL$1uR`JbB{a-~gs%z;P6ie#yznr<IA8=c`%`D|h@dE1dAc
zlW7rI_bfl;?tl``hOyFk_4<40+mOdMyAA4Wy`zSGNacfLGYbcKE)`DEUy2T5nJPl&
z#s2~TsuHH<^PPRUS9IS-3p*EgUo@2C^LB$!L4k*uT5O@Bf9p_T*4DH|_t|&XXFdLj
z*rsbY?sR_-!u<R@uEtnDCR8nYi1(!^jgPTjhfg-<|FD=%VTbf>a>k>-on*w_R8gLY
z5%(da{U4p|wC6;^8<l^Op+Ugo#^Bf4qcO?f%jNmtpKh0wn*QWs*I<n4&e3T6)%RqR
zW|X&CqYf~wAp9gE`n%Jf+xknbBZX5{NdIS%(@pVsWN%=ZJ%)`CngNy`V53KgV@*{%
zlCTo)WM+g3YW(v;g2tHR>j!#?A!qlVTOx8j(6hdEnBh1}&JWDjJEtUxE&I-CQ~fQi
z870*L`EMpDBn@UW^CoxB$5uWghK6r}B8B7{o|hs?ZglSdC9Fp|Ci>RTf13#tV4FZa
zjc%XiMe^%Oz=4{%n^=Z*S6(iu^>FZuFp?3`GA(guU0WR)@=)yfx-hJWcSP)-&GCy5
zH0!AE>o2|$8;lkJV<G3Ue=9G)4{%N(^ArMJH3C*$vd53@ffEUcZ_GNic+hY7_ix=s
zPkL(V`xGeIWL^Xy5RHK-1VC?O)u~<H+3CfUUgHz&X=VpVv;I_}`S$iTusng2@_qc&
zYHy-MD9K59_#A>T7!IF2es!-*Ijx^XNQ#1LRC~=@IO8n6(qvk-mw&UtYL`Yr673a=
zZ^_+uFM0coy!7gK_=ua*JJH_rj6UFkXS()ZarmH67^m$flp_2ztp#`@_U$@1%uG_c
zp~mWR&&$mQv*W&-FsOM(9!I8%;;i9}lf}Pg_G;j7(q}4#2tBv)nri(0uB!nt24D(g
z=CQo~k}=nPnxDYe(M1FSMXH^7<_}5P{`LuK75;RHR2Pi&?&ksZHux{LmTL4%qXZI|
z-hs|2<w*nney%yIisfShD)t8(<J*EUYUk94KT;)&;u^0A<m?8=6{B#UJR-#3$jo@<
zT)k5!?E#dQZo~Z2<jldTi>MfUUJP{fj&3QJ$Ow^X{1v{2MsKsZp9-M;N|!_dfj+8B
zK?=Je-19Ol+up(M;p--co|n_$t{au=t0D7fIrtVZnhMs{OXz5%(<cvZ&0z^lN?xb9
z9d>zDHnhUGp0%&lmztzu4W%x!j+^U6^F+T6BMj=BSFNbm7#S}2%f^zJrCeQEp^xpH
zgCAf%fZjL`r65`8Tmm-P>wTbl8?<`Y!$A#BTLcj1e}C`@V7CBW3B}iLO*Ts06GmZ0
zxRG9!qj1_f$n)DHMrECG21g;$iiq=8;DL>eR>2MOThfYUkGkQ-+2wZF-tX+LYI~7M
z-r(rcBd^tKBb6(4NWu`@xiOh86u8n+mqP@7i-@FSVJ_nKxdUj@%`P1nfBCyN2K6?R
zteyu^^1c9%K%(_7Nw0-P4#UZSvY6ZKCZF4hg1MWq((c0@&8m-wW#g1=_J?uS0*^o1
zWJ5uC6c<2X=K(0}eahUhHnIO#KiubA_;u=V&fVELx7HJ?LtA9BynHq5A!vOp52X!m
zXkqIiLiA|xfkWL4vo{LcS4so!7Q2OL{Ar9}X_w^sf3G`A@zrEfr&k7}O<puOvbEZb
z&9l%yt2DX&$$3P}=^pUs={#s8I>z>Td*t1oiYX-#%Ged=<fO^d<^N*&`TXx-ts~Sy
zPshN-FsiEdrs!Gw5y-kRsq)ngsXqJoEyhkX@Imt`yfVzqn%ZYKUNkj}{rc{D-oz%y
zC3lV*p9>f_;B5AL5g?$V_>}j--rgQa@R60ZnvMbOx@hmv1u7~ksJW%(Q^6$2m-_7M
zE98AWVK1=UkLf8U_V<2<d8XQ~;FV{Mwg+fy6kDE<A%2oqldzk&`|PoFyj_1(l?pY(
z$FEu^q)r@FzD+lnuzV5*oZU5B0ddkn8qYuDuB$`4*FPeckFB71b^{vd4^1Z9l3F)o
zUdECV)ZAC<Mqdp&%!5EId@b`^2ov*xPXv}+Q?}`vju###^h9Ecb0rKCe*f0Bzt#=f
z=VV#&o4={$?dsZTC&CQykGL%#42ARkQG76}Qq#_n_wd<Ayi@CS*sM8fBfzTglcxIS
zT*Xrg1!78GwP-2DM1Z~?SE%+<phD&vZfd(t3$6wKB1;Ew^gjQl`WHFO`=LbgR{mSm
z=(wuwY`vY@UXx##Rvic$Ff-22{j*b(7!a&DZ20+J6U3CC#}fO`O8~!pqii{R-#o+_
zZ*FgEL-YY04G_~XUZ7=aYb)fqJ^+Ms5XM=?CH?Llu#V#6SGFpAssz+wA$WtNHIREa
z5QkE?%tS^^Y-x)WK_6W5dcBflHtrEpp%2Y5SD$RD8s}W`{$!e60G04wgYHRr`Qug>
z2wb3TJa_=pFQ$6;<V}AkO{%&5B9h*CM`k)(MF`c01S3P4EcS*0C-ZhQ;j*US8x?z|
zh|pff2l>Ih)kZb(n)Uv<ht+pfR-Gey-3y>O)bu<Has1qEmAr-r$D)h;*^SKAdq=W$
zm%YDW3zd&<%=z-I{}NdS%^TaD6d3|9=E?X4{3OVanbCwM)1<TW_RF{!Yg~2Cmqd&!
zB@T#4=(kP#0VAm8_XOKZIloBDPS@`phj8z098<#N(fBKIJw$6<?nGe&HkMQwPvl)6
zvfRxv_;oCUWQcIR>R**%qUe%9e0h)<yHE?=KgeUR9Alq`MDobt6RM1jDz1Ch6TqJ!
zbj9CTyVk3dfY}Ny0O%XQZULjQ+0s&bIGECR-JU49yS<7?$}2C&gN%%gC$S~8y1F!r
zJzp2?#ZrLCFf#?}gF>RHHL5DwXOv*q;i9O1+0msgF7?@vox_b~UQ$y`A;s_kIabX2
zd35zpj^oT2J2^Iu;=-`Q7yn=klIErxya&)KO0AU6a;?)&<o5*9(*shH=0Gv;2LO@s
z%;CZ8>M0Ba!%0$v!2^1FVxAt(PMim=2*N6ojcffzCRno(Nnee|S(1EDCwISeT6T^!
z4|F7b{y`f48ESS&*UD%)dMBkx03C0px7a=HJ<SJI4K@T?X+gmc;ZfW7FEA|f7{M<Q
z^*y4y_D02Z8J6by5xx4Jc+xGek_<iM+}<%!f;8Xv1cxCnhw3pJ5ao9H<RBkhjRB{J
zeM0lRJy<8U<Y&68UUbh^s(3#OaHx23F|iE{TNH$W)dNVfdW?Z)K@mLR9A|=(U*0n>
zR&Vt`MEm9(x-kk5=(@9yH)p)L0<an|6J2FJJq8$w+uPd@fU#S^-4T#hfR}e?Fas3=
z@E^#>59!KGR#7q1Et25(7UT`Mc91a5V?eax1DZNGjZ;OW1|9Gc0o|?WuTrd)pt7>!
zs@M9z#uJR!T%t>z^}ipBv14D%CD18rx*1pa!SP?d6o1o5rFgd2^fIm`19?B^XKZGK
zw}lCdgFUHn@s{aOUQp-0MgsA=Q*2BhD}&oUmDM1gY59S{lf(WDJ_b6LVM^XSdY3;l
zR6wJi^&;pput^*D3T6x>KD^A~&~$eeggB1tePpD>xd24K&RZMtzLjVz^f#?e7(E1*
z1_-AvE_c&fa@-LuIm<@UDqG4^ml=1mdgn9q)i@g#Y|sPHDiOMSy`;<gMh&MS==@NS
z0c5djt4a-x2#^2Hq4l2fzrW1nLG353&sCz^rVm81{CO{y>jF+wk`0FP%hFq{fby+Z
z?+RqMhY+Z^CBn$)>FEJY^$0#Z6v_L|=d>A}n5ZT6ki55HJT+mBS{HTSETS2Nf}HNn
z_lMs)dKv~!>w4f4UB0^9^o=vT*Z!i*g9@zp2=UO7g^?f>b)ta!yF+^3^xWE5gs<!8
zdyu|K8U_^GCQvkkbNx6`TSG?_LJX#|aYx`$1zn&Cs0^bjWd?>H1*7bPMk?@04opC|
zIJ;la{;mOeS;}2p)>~8ottr(nSgM`#;F(xPo}s*f#n`}nncAKMYlzc+PsL<Z(v+Ck
z791GjTLE6bvE9E-Vee?v&vvY5t6VHykibIQT1+tP+||KZM)={w2T)!B{+w`zOBUz}
z0CIEsj1vr{<>bVJ&#tWe$JrI|(x$G~$Zar-9=PJ&2)Q4dc{{G^l~|t4AVkF~&?%ph
z)9Px^X*Ip6D%275lI9rLqWEZIdecBQ`JL2C&^1CNv*C^mbh~mmzDN62hXtTR>O`Ah
zC`|ln|9KdMgdP(xb*HfgjB)iJw>smhm-mQa9?(uTyVUs=lJrT}g+(HrZ}V4y^G<WE
zi9p0<Y<w*^(%m;IfiE+!lDx^?Of*9x=pjaW?g{bag-lIwK=21un>X&xx}sP^rS<@{
zzM{6;Nmw-RAY`f_IhkE_@WShRgDk=l6=;r6Zaplg5>-;MD|XWJ?+d+tE#pWw0*i`r
zQqwlM;K&@4e){ia4XB_)@f|c#<l`<PoduL`7M8Jr0sZO1tSl4=j1gd>_zDAOIvpge
zs!E_$w!Liy>G!JjyVM8NW%PJ}Q{44>Pm@Sl;?O^O7U<i)z|y)gV9Y7UKu?(&i_%j)
zc#}aBOM!*=ty&o=Nv9ty?s-EjH2;A4Za4R7;OEv>>L!0N@%z;BNX_9~i*Zw%oInds
zp@(2^M`dkA&_9P9c6p`;$bn;+1JvO@BALvSo?v(myq@#5A6`I6O(C_nFLUT`-umx;
z*ksO?kUgKJWE=JB`5d|~eQW{t89U6NY<Y7!iH5ArB2#lXfhSamq@%-pKL{Gr28Ydl
zFVkL~)uK74!{wMO+M=1W3AdpU^TP4lBPXXB#B#rm&p@#T0_eX%zWdS$EN;m`L1<9%
zN)oD1c{v9J#3_*s3IPED$o2M`M~l7uZ}S4qnpEQFafEu{2ZX=|?Ms<_5y+JC@3U0P
z6`a)YDeH;G$=E28&yUDbi8r=elv=tPvkn~`#v!i~Pg6~2Z7fY7$W|R?zegulTmS)U
zK}ftcwXL7}KL;%_Gy^}1)snwnbRhf4{IuO&;<+(?zbJL0w*>G!Uy{U80n4EdHyoTN
zIxi>yHCx`*WFak|?aV%aek>{driPnJc0T+c*%0jpvzU7$^TB_u<z4;)ICl#N5tO3{
z%{kdz4u(pfj{s{a*%(^-BA(s9KjCyaC5k8X`)&G=ED$>^cPJ_)gzm2hI09OJ7?|#o
zU<Z6=P&20ROJ>}hRvrPA9yCy8t9P6U8~_kMi2vf$gLOcr0mv-}^eljja&tSV2Hoi<
z=N(N>&I&No0n&Y{d;KKK9~hM`$c9fs@>&vT=y!iAT|Npuq4r{i)Jsm(JFZys>vA}z
zx~wHaqyc5Oeo1&tCx^sS`TbBJ>j7AY_AQ+Ra;<M)@jg4$qT1*32~6#t6>5J$esp5`
z5+(xXvacQ+ABBuQ3PsKFZbHQdcYj~ZDLDQ<1AKVIW<+k<hTYkRNWXVS9gvaVfvdZ9
zmSSEWb@YNyO7hEt=I^uF!1JQ+LVi{kcP{V*y1dhtSl}JWs?$`1_IW&xyp)+Nm*nOj
zyQEa3LtW^}cMcRiGjlpJHG_g?<_jO-rR&ADI{>t84>%RXy7+kFy)eIx^M5~C$Qiw!
z9B~m55iv0;Mn=niXiwxNYd|KNOd=PM>vRG4ySp2Ma|?h&YiitNW%~fPjUEp`nN(Cl
zDFkXOE1j>;4~B-A<Jl6%Irk$YBLRz8RaG^FfSK3hG(S06r^;p)uBQ2+SM1|3cLK``
zMd@Y)o;DHVf@aGn;eFj#hJU_tX0G5riF4({F696VQ~m@GCv`hG@1U=x0Bu3{%qCxA
zGX4}OS;{1#QW<Cw$XCbb751$|>XeZD&sV{8tC?%6F9eJ;pM#^Wtd181wOaL(_CA$)
zimTdBE@XZY2A2kRWz9VDm(nV|c<C=SEDu;Kz|)(PkzU%|>vlJpc$N}ai&<h<yxYAJ
z3^<C6CZV&%dse{f)rxm^K)vIBHUL}~9wdiAV*d!O&x$(aZsCro`@Gd`+Hd;B=yKm%
zX4xm<_7t9<6ZrC0<j?py<C*Trm;$7~^SM)m2`<~sAhXg95dXO0<5#U7*A{DPDC4->
zwj@q6<H>J6lGZd#d|$BtXP<cSGsq@SHdSyI#2SzS&Q!e{5J{i&@g<6+e+dpw;dPSI
zxlx<kzSp@mHFW{I>FHBpJv|{dw$iLDacOCc7;{S=AZsHmjQmpK;o$)uO=6gsxbvCt
zcYKQGvZ-pL%_Htd&yC3<6G%5Kb%eGnjE)xbXFdas|6T^#rTTwVpy_8Iaw))F^U=Kp
zTV5}OvrS7Qy)5_#ztoV4`HYhKmmcXV(3Kgu&1$^u-*7~Mz7kEmyXGI#lDHWe%iQu}
zSzadObeBNwn}xrI$vwk;=rP`BNY2HY%yl_e#tqKPc`hmCFGu3=2v_X?YAgj{prKps
z^Qg&dyPy9fAJeJ*J{xAM!$P;uC|2+5q_+pokd*bp>GG`eUd|W0?5ix2c7vdW5>{IE
zL7;Ua!naSy1NVeEJv}c@O&L46{`j{yYp;z?O#_R%Aq*M}z++oe6%+<C^>PUa@fYVX
zEeVM*#n-dX*KCF0^Y77W%9@%A&#CEISQ(AS@~(u-bWIvanF!~*oy1f?y)4;EAV)H`
zt^%e3u-ftweV8%Y<%q_@Q@MTqTZ*PTMdm#P+>g(%FUqQ_&Omw?@MQvdN#^}~egOfH
z|H7$}=(IO?1Cobyb#)OC5CC!(@akAtSO8c3TV$lg;O+r07)IHhIo+AQ7>a@AX^AEK
z#efb;u~J?2KRqT)yY`Ce*iciI#TQ|dA-Hx{#Z->yS5ao$!8Zhg05ix|mGz<~v*fk{
z)oq`BC`E!ARrB&vNTSLo7#`*H2i2!bxch~`!&PqykWC8a6p+CLKGp}uGTOHUGZiMg
z>leB_7h*s$n5uBcK4Yw{d40Syyfm@mkxtpPN*LoD8#AP+1WY1;HS|fDRn};=dJ&kN
zfRD)Y<kmse2Ei90IN7*B#)H+-K~;6VXAHa^wVr0f%VN4GaQi+O&p<*<FARH}nq(qm
zusB$-D~iIgL5$j_pdhI+{5@s~!4nntfZ46PQ!8}|_ArW4Q~s`|e98H0rzct1P$2ZA
zOvl60#d00<zqP*t0VZgkQ>CM)2bu2q^u0jSJ0ZMm4Q^ig?eyeH6rdWsdDDVL%Jmhj
zP+%Al*8aDKAPmCCG}geX|GlV)wOt1!q;5@}jL)nBkr@bmS0MEWEMDL*`jwjtoP3*`
zrQyt$78W3OoQ#4(S5-9&gjMM4zXa9VW0tYmLv#!bVoa2XcUG#zHC%;ZA`SoUj#qsN
z^_C(5U$K^<Yg&W2|Ej+tx3<lCBMUC?A%mXcKE%il)o5}hIeU9UE-WU7>@NaV-Nij~
z=`IGxootRNrrirM$)MF6x0dgj0G|Z#VGvNbk&(`E8&u{;eYi=(bTbCm0hZ!@CD+(i
zVeZrCESlPykFEq5kCn(Y`>E)Jr%D`4qsXtQNcbQ0Q=|JUYHBi2Q~Tb*w9faid2nz6
z|1<dNV)n+eYyuwvbYH~Y?k{bJp%t{hh<N7-7mYqnIz^6C9<6X@g`mFv3gxIRdD};B
zUQY-J2~i=yKh*_RMxo1AY~(fWO@L7aAVslXh(Lu82@VGNtaDAUkj>2NF)|-Ovjt~d
zZ~%awK8K&<1Hm19;p11B^pc>V1;wrljg>?oBWHc3;4<b29Q@7R<jX!=<zIjENB<>u
zazjdNgoFN(y~3l#ruoiOSxrT49gSzteWz0qf#BwPJU2Is=2^8nfDa;opY{?9nC-^5
z05*6&Ru@Aa&INK((^aHMX2FERoA+-!=6;;PlzOt-%I5|1&c|rs0{{7Azs^z```A9(
z;fic}#*eiCB2~b3f%Ml0(mgkE_&9v#>s(kFUiVAATW|NM8yXtAxjF-><lx>Cvzv5u
zbq$%71MLZfWA1|V5&@6X?b)mLM1D%jR6x}LA@ul+svVu3;Cfrn)`E)!7n$}YY1BHJ
zW6UC++D1mi);n8mowgE!4~h-l%IQ7#lxwP<m#)YYv2-c8NqB8B0NmO&9L>brQ+3fj
z{QaV82tG`7A3wBOks3B!TGRY}JJxIeYCFBragB#0d3VvFCT>g<xkjkxQ0gh}^3MYY
zt^anDDdLo_4uxOJcrZQ~_3X*MEf&4U*5NO@$0Kxtvl<9086F+gY4qel?+14u$Zi?p
zp`cC!f`oAI2SdXq&kHIRmPSw~-R~pvii%Uk1`mSXx?da`gSoQ%&(i<+X#{`Vd!gE1
za$~G(g47Q}c6J?Fr@5chGfYA$4RYr7H`#y)1hcnsP_Tto7^8Yos*H~zfXv`eWD<Mt
zTl2Bnu}YA9hJtgSaHrUr2mP<CGbkr6xRKhYlzZR`(l$Wb=AI9|qy&2_flvTtWk5P*
zX6j-712Bw_AP~4d8M^~^LbfwRVKXy2C_X;ESiIY1V@OO*Ev`i$q-q~Ez;{F_P)6Xi
zKy|U4h5`i?EZx=Cs};DqT$1&K3gdD|Pv~cp`HfC1YW&q@zF|R)PX!-Hlflsg?GHDN
zUp0m(+*95g%ZqfoH<wWZrCk<b)^l&$dZ%1xto{L(im~sSYte+mqQ{u6ocR`=a<2gf
zXCQO^;kwgJs{@8KcWO1rE;E&8jazBZj>Y+2;no^qiS}y}w)ePS%T8G>P7wEDZM4RM
zB`2(FfXFV$&-e85IypWDqt54v7caQE!RMf(<9KZUK2vdJc{w^JX0j5<?jSY>jOX6x
zk?0#4{Tb`&=vdzxFEV2dsPLPsolVu@;V@i@1;Swt2mJl)Q`xJa`7<}>m6DWsAu&zG
z>lv2tj&(nug(|dpb~L2Vs#S%+)QfpsTjf0d5~TZiW(;mXFKx-LYF)y3C8giHRPwD~
z<z$5&o;rx!#Oz<O3h^%!Hea*R3vkm5oOyrz<Ijx2d}+q*e%tJ#H+!bTVnV;45Sb7c
znc%{?0Qa(0JvwURy8VK1U*+67?bLkLCT-t>1O)`VKrScP61T<*>Rb;%w7U$T2UnM$
zou1lUb}C%ggS0--2n2Lt(EJ6TeFQwOIy#&j9Pzy*U8!fMJ6e$b*w+0dFoeh7k8a)B
z(_`h{{xZLYxAUEjP^I<BDRms!-%)%LaULv14fkDH39}!mKOx?<v<u2?pZZmm3Xgm!
z=Nz>C_l&Ep7CfE^!slzQ02<MdhC`p=Hmia+Lw0qxT?lFe5HO(Ly!DPGwDCj#-*Li*
zAnmO;NR3y#mP>Zf(}*v9&RGGqo1hnNWYx8^X1hf<BCxhc-_<!H=MxV9cZOpWM}1E*
zuDZ|bL;30h($NvfWR8wb2@xV{4*c&6<Ke@1Qc@;H_g{Vi0kVl)y@P`h-6bHJ1k})o
z5E!H{&nzu{3~ibwf;I;#5OH&}T=GG_rYK9!IR>a^KunM2*^g1Hqxig{g58QUWpe*Z
zo)>ll@VS3`l}9`&r_;bt8wi1gudk*{US%X~Zf@o%Gf?&?#p*!NkKaaa#0M{Kug&N>
zZ}_=VcfNt8=ZQclwPiN@aChRS<->0e&wPadj0?cf^IT$3eR#XVwLl}FhaFf%$W3<v
zFgw1ZWmM^E^Un2_@As5`biu)AYg3R%3VX85Qkvz`(hVSB6r|hQSz9xo*%zzx{k_J|
zQv7v}<AfT})I_eUG-nynxy-d6S8o_lRD}$R*unKG|1^a9>gw}<e?vemg@lB}z{m(f
zz&H2yV35opAOMa{Ja7BQ#>PM;yaQ?{U~nD!DcRtDA~Jn{N*O5%4a!Z2e_Xc7C@Uy@
zLfwZ0TJ+Ppkni$A6O<u1InIh-UR_&LS&_@UByR2BuZ+OSO6R8YRm!{DTa51~)n)X-
zQr!r`itm4F2=P;id>*ZczW%|ss<7VA_DNWss}o^ed8~{l{;Zzz`VsWF(bPHiF*1aH
z(ypIXSs6QdIJhMTplA^Ma&%C$N{eue-HfiQ&%Scj?e>kh&0vhgGZyC;{4;imZD)I|
zV`Jdq9v!U+`q!f_ol{U?J?5dAKbt<h5qet}4JyPN1W!`*bX)4#1yUfKfgRU#169Ho
z1&fV~OI1b1xUB&UY4^sz$^aJ&Sg*#5^z^{lVEUQuTs;r$unMAjKNT16e<fxoBO{Bi
zqoJVz<!5f*Bs)bqH)Az6|LHi<$w;GJk4Mn>EvsE89=cQO+%TwS##tPxF%B=Uz8LI0
za#)zn&v*yu^~lUArlU$qHFfA$QY<DFJiGn68IOk$gq77h{FT04zawp3szQ|R&J(2l
z>9l~rA8`NtAA})biU*SNR6VkGZ%zXQg2F;wxX#zN71R`wQBfM)u`w~AnA--Yd1wF}
zeSRrrcXi)Y)`EDg%AY@f-cmD`gV^${4e}?1gzfD6y($_So%$~yB7KgJSGPRNJaHl<
zZ1S>b|DN1#D~O5Y12hGtp)$Gc{clk>t?-Zki>I#u%A$++UP@4;k(Lk;B&9nf1ZkwC
z#2}=V?gc4nDUmMe?k+*PQ}U7S?*0z%z4v?O8HO2Vmff=_{_%@(%NJnMg}jucVSUp9
zU?<*BRzc`<eyP_KS#IcJ1r!uCMkH&q)pevbupq|HuIFJAVGy^*qm5*l`h^a+sN&-1
zQ#}t;iShA2R1UB)qCq?vpcwq+CTvhYL%yvlD*x*+QHZEHE6{}o9uvC=gZvP2gaJey
z*=5M7lnPIAXlQCibQwndA>DKu{OK!8$9wTMsLAxttqOUZ*Y2mOBdC#vQTeT^Djpdw
zvN#ZN_xG21|2`G8X@2`+x*n~m-ZFx+m>uD?gL+#dC|M}(dVW_~bdKQQWkg5v694y>
za1TMl!s2EZUst0#Fg6{kVp@Kx%0Zi1D7;$M0;@FbvrJkNM9XE2te;&@s{(6BZchLN
zMMl25LZs#ouB*+Q=El`tMLc+fb|Rck;V$1To@^H#V{{p=w(~^m*8lstOuWL=^aiQn
zFr@mF6zVc}7bP0VPHkgmFsvk3V)dK#GtwIA#|Ntk5~;vA9P|GDiU3Csk|r>rQ<3Un
zPfv0GE2MTP1#Dk1EcBe))0%HLV_J!*u5t)dbYVLFIa!DUuwe#_Hag(rM^8UCI{FDb
zu(>(#+yq7|*DZAoV7CJXYrrPyeXC!IuO<MDLSP4EngW6?U@>UCdBgu{Fva<Bh3DuF
zY<UbCE5O{fw6fB8mJ)vS2o0jhy5Pvqs0pFWOdB-{?Kp`j4a}cG&RWlN<7KlYHJhHg
z{!PtQbmq~SwYHQG27|3Mf~bMV|IN|H_naP`L}&H_^cFloE}x%<X)WSxWIwm@9R)<w
zTrTS1mpG@M<&#|#D?h=z!)Y9J#QX=!f4GPS-`jKCor3`f)KWp=;buDkmjT5P@B+7I
zYHTM8^+CX8HZCA7B~@t9^j1d)a24NKSoGFzf@}nUkJ7yxfn>R}v2iAlGzU9iO~T;l
z=n+7x0mmJ%Lm2h9gIK`*{J>Y?Dac})v}aqNdyu{}y=ps6nT`ng(q#&mafjnuQE+5<
zDF>g$w#RiH$;0+=a<2}nhF?P-1x1b5oD0p-Cay7|yXghPf$p2%EfkKvp&}=?@Vgub
zY*O2*&cFQXD&L)_gmA%zejB!UyrfttxcM3PdKE~40-I_-pb!Jp)8p_9Z@%3fax>3K
z9@Z~Z`|iC@oCXMr#Kgp*HGMr;Zw8n&O-?%Z{^WT2^eM=*AY3Y{-K8Zv&{;7~{2@+#
z*P2hHQj9<mS5yosD%u5kGCdvLQ-z<NyrS+Ggiu04!Wx!%pUcTsPI|hUo}Sy@zZb?O
z31Gtj@JN8@tA9>zF1?W}hZKYJu%ZmK3cSM&)Q6`;F0!_3uFl}SW8(aDM>^Ufr`prW
zpSzO<!+F^GdG%X^9)5Se04`VUBQK{f^;}e51G3wqpH2YAzU?z>zKnk8dqTLDeh-Zd
zU$mUOLJtyP!U=o+mq6^0Sgxd}&)eS<<>hr4%hdpmhs<?Wv$d0DCS{o?IfaFy&IiA$
zsvLnwWiK4d1&(IW(T?*t;P}`M76_uowqs;OO+rFKN{UWQEEy0sWMx@Eu32E3HS`In
z=0L^3(lV4mRb3reT%lgL0Ybsf&JLh?g2Dk<30+-XAfW?W_(-PW<)RkUdM2T#uyiFl
zkasfx&h%B9%7WQMSZnKa%l7mwA6#BlTI(?K>NanB<))AA8WYqqPOi3NrUbj!b=ehM
z9d@<YpC_%W$}UCA6d{Sr>DleKy}#P_Q^h&XL$U^zO@+cPBD0gBmb-nc^LNUK@3Pa|
z5u10M+cz>Jei45w!makOin#Xtu^*?s&N1XX2bu%`jA0_*%lza?sUh4I0kWmYt|b8C
z1FwvGbBPY$bgJAG$Yldh17JI;nVFHn^P%k<sGD$lZ-9=``f)b~kCdwFezO+}0O|lc
z`Bw@G00`z_V*?!Qs0wNPE6`Ap-+vwX_blBX&-qPT(Eggzpp4+eEno7&2tL=rjShT+
zH>>f8x1Ssnf&pYJ_H76Dg=nrA`rEhK4cAAB^g{i^Yj*en)g~G7<*3Q>l>C#iX1R7q
zp0rw2GH#azgI%j<>d8SkPzsnQXQ`T34QzJ>Q~T%tpWk%v27r#B=fHV@Iqx1~@Xeby
zy$J$`YeVTEwYv0{zq9tTHo&HL#zTYAd_kVfb8S&tT8c@`cI9pL4#fRuaX%-W;(zt-
z@PCU#v<to&9UXU0t_Dy`>@!h8ccJ2@fI<g8`d1v`5rMhZuSl(*d>qPtEACltat=FW
zik#1(;|HvX_L=st5CGfkNKLBG6?}Z-W7N3!lRq>eoXGraCJN#<7^T|j<KX|`0OpTD
zM5TIG>+6elY77{1fDR-b0<hat`K7rV?bGVqbN4P$tU|BtN24ke>W^6k(o1Vsz??|`
zE?dcaQHB;f`hUU%vl<R8vA?o%W+dE64Cl;S6NaYyXHr#6wNU2tu$)?F=xLso+7J26
z^S)%YGX<$A{BV<hN(IARrgal)40wM(TiZYQee=y1qesy$DD4wts*t$6{&L@}8tjl{
zLbLmn6^QHaQYNUWsV<oim#ln`wh&G>VtAHtSeo(kc)ZrJlXmH&#lLBGsz&<dN72!m
z>H$Fyn7d!aubB$IA$@Z+WVJ4;Kx#DIRCI%Oo5Gj;0WLcXB7L|v{HnXVxFdn%M1R>x
z_hREN`x-}j?%DE*JZNBAis3A90?7^R$prZLV4lv*%v5t@Vq$_m#UnkB`<tIAMPaqm
zGDbTIxr?eZLJ`a3@ooI}phlMs>!zo+rT};rY(I9R>3AcJirmuvW}^K~RSmjFO+{6z
zel8qn^C}Fx%!&r{L9ylgjLgwV^SPfZ4$xgPR7|z^&@E`5Cm^w*0VrP!IbvfC3Alh}
zbS#$%^lmWa#eZb2xalzOH%;}2moabZ7R^jpdhLA=sntY$m0rUI|Jx4P#vyC23J4qb
zcy<;tfpoPomIr(k%*@R}*#!oWC_Oc`*4nDf%oRYrqm>B-UGPVQgnH$l@b2_Ncr#M6
zmyV~Eq36xRZ{#E>pdDdst8JzC!m3e`Z^b^+c{SJIdcHO>9U!H>wTby}Hi3Je=ka+f
zEximnF8%%epxy&&{s%`#TSMvocOv+|fMroz8vya7{D6VG{rI@menk+;3E#OGK>Cd!
zy#s5vKhdxd0Vn;4J=`}>P*qMG-yu6$eTl;zPDn6kbMoS?D(R*JxuoT;GuuIFM8Vj9
zJOdzZlA8YGvpFV!&7*imMved%6A@uyfHwxaA@KSFA8qorwKagKXzl6>I5mI&CLaE}
zu=+<zn&KiJP{Z__Zo%`3Vq0B6NlGS!L@{<dbR#y2OQ65^no%7<Is7u?Zohc<xlWOt
zIuS&ZBqnHUs7**~4yRRRR_02WpJLv<w&(f!@5*DZzF7VucWd=W5{81wc+A?q;eqV+
z0TvcWRlo<;6CBmx_@4#%!-b~X8v@*a1HRAmw>vxgFL58CK&@LW(muYRt1bU6;D?sx
zx!C==@()BEUL_GjpP52dT%F1qD%KZc-q(d0geoc=>xkN({JKYF4VWasiFG@(S)ZPS
ze>Tjk+bl-Pca;=#t2yLn<e(RscXct+2zgChW7YkDAd$jvN;?An*2MZ<wd!+Wer{tm
zqNjak(0rh48tKI{==i)q;(0`m;=cW;j3GCVsYJiFzI3BJ8UB3MvAaGAU@^IJ`rKnt
zCdxLI-0Ocz@~ml_&c)oLlcFa+nE)*rm`Ypao#(4Cq^BbPgUbYqY8y-G!^DR(-a=4Q
z0+CLlpgqjgr1b5*<xHX{H6fYLy)nbm6z_ubK7TL3;IO7m?+bCU5ja`slL0^wGw#yt
z%F3OlKjaAaz^AwvqHlZN$m{~inv&Hyki%<FGyM7mzG67e)<(mZ7Xg-Ek0}#CO(`U<
zk!eF0`JyybB1Bm!NLi_l0N3vLFOj(#ig6E&qskFUWC<{NCRPFRN<s2Uu7o6lM!kEp
zn{bcG&`du#k=}F=L?)awE$C``aA>QDYc=F0YO|Z(9~$2w{I5~w)(TNbAYMC>HLor&
z&(_$YLlk%Q%rDDkTL>_{04E9f6sNJ$q903o{}l@Ju73DWOuFMTq5cU2Gc01hz-X^Q
zuZH3t+PA?tn$y!DIp>b-slQ32^qH);_H=1kXiX<q|IyRBheImca@68-F5?=vCIJX8
zz)!8eJsQ)nWhhIuolZT>A)Snyt?f@*ngIlsFQ(^@ZnI94{8X|>&&?gAMH}eEpCE~Q
zyx$*Pjt6Z@Ad<5vgT>1O4+4%65G(q!Gj$DGJ(zryCtTZHX2puv{fx}rbs?1P4P1R9
z0^{L?QBg%A|4tXSiN`}gJyo6oCwYe(Vqg2;a5gjx$ql?vpm{I*UnL6ARnEs49-3b+
zM2@5_Mj$nt%EH`hXpj&y0f%aMckYI(vSejJk^!g*Q%u4eW(Ft?n3oLzHApTf7X31Q
zyK(yuP){!Y<GvHc0VuZ#4JQyZsu7)Z@mT8qUO4IEWAsqnp9w$CFZ9(x_G|XCGB&jl
zfoFZcgo+>u<?>ixe9nEbC1~br+SxB|^|l-5&dd6*A0YlR$A*TAXNR9B-lOCv7;8+7
z!lF(marVI>f+LFB^DFzTv+7Ybj+r$K9fH9TrQP>;E+e^YCM&fxCV;?ET}pNDM?t0N
zuXo|>V`P-i819(!Il;i7V0dH%1u9>A+<Noa5&7#N1CV?7t-!n=jz0YpSsd7rF|l{x
zMW2Ydo{~<CPNG3qK6K)LU9GhPzj?ymssAJ>va1W~G<7Q~G|df69#=@3&yH1E{i%I)
zswssV6~fp#C-gpS>(Zk-uAPn=SEcRn_MArabg9NaBjJ*zrHGlBN38+Vw{61Qqv`xP
zGUTBrsO;h^9MeIJf%??XC0Y4CZV|vIYxr^$!`4tIE8~wH?H699_I-S58;LNt>oDVi
zw1WrU`IB{xr_f;bD9xOE8*N`OR^z$b!LK1nmcE<#FSK&TD^4&L&zR%Wk=%Y&t%HsL
z^Wm|&&XypY1p}j|0XJL;44TpR-;@rs^1>alt^m50K{2iz>~PXSM5dqqt;?;cy+}H>
zdT%z`kx}P#`FCD)DWk_x!W|ZwHEf*dn{4T7$;?Ue_4z!RnP|xH@#-F1ET6RC%W$Qj
zCW29Qoc!*MOtzX6JL-<Ve_~0=zq3opM?X9&NAhIHO{jO0|FCD^qHYkSps)Wpti#oW
zvtWb5o)pW%YVX>l4;2@6e>dc_mQ2^KvJbdTceg!~8f;o?Br#wR+W@<sdYsNW5<FG-
z3p9oWsKZnye%z^=Xkl1a77_5-i5oofvyr5|ytro2onTY_)pLedX6E2?-hsHSAtsV;
z#18oO?3oGa`X+BO&I{a(-oX0b5l35J2Ro}bPg~;y%PTlztkl29ZHPIXHS7ELb90x%
zXU@A#Esig-_78SOdf*KI#Iz?3=L$Lz!NQ4Eqz2U0kHQJbV&bAyRaEr{h%s1$?O1~C
zTqk4uNal0W@)>VgDHsZnJa<hl*fEZxo2pLVtu0vk8b#Y8*J}+#>!l3gy)|4asfieH
zq-(41b;F6J1`nOxYM-RPCiIGITV@XV29zfp92`LN>-u;CHUNoOLqmfl$SwQ&0Q=LZ
zy-2@KvtQ`RQA@zez^<^9%WBIA=f_<0O-ku@?9+92_pZ^-l-T}VU+HIzD(zUO_ZuJU
zXd$Y<T(&e-<wual{)l$mnhDAJ!VX}{lN46+oD4FcQPckWsI3r-*<&s;(jZ}vlXc#R
z3}PHMy@p+cLMS&xz8vAMpWgylEBb8%MfCTCM1xT`PRlhtrTP2oVvlPhv|iK9{fyI%
zh|{&%o$UEIk>ti@v}SU{+4ws9v4-YE16O7m8V3dqc+_N0XUsD^+LzyMd06f~`LRwc
z|Asn3slS2gri<@<?vUM!{PtB&O{(z&-%?9>wyNxDC<~L3Xb|kYKAs*+AcMWO`2?@>
z_RP1hE)>H@0!kVA)E_!)!J=SU(0g&R^W}1A<l|~#&klu7ZBJF0zq!+13;KIiM~P$+
zH+;3%{R;<Y^FsmPJB^btRd;@wigHmY5UXj=mW7G)f%nQbwzqFoe0pN6Mam%0s}J~3
z8b8fkC63Z4&D{h#9}*c%3Z&y!eA2_*x}Q}jD4+U+DQk~~s+79>{D5eWJI6Z{Z(}N{
zy@8R*2)VhF<}DXP@ONz72fhXcGsA<OjgRoVRz62Z>+p%V9^x4E#qr}qm4)ZVYokSI
z67w4dry{~l23HivduweWLfA;2k}zLS1;Ru0?}*~7EjAjw63fd$I`Cd-{hdN67SMpq
z*&n=;0c(-qdV-56ZPnxDTMt*8Wom&cgQ(!diu2NQwT`p%a=KU|2ezzgtr~J{Jt7QW
zvxgZKhwZ^sSL21K$X<I3tv|(-<_dF6tVQj{hPta`3bfVqX#|efjFqxfVQ04&TpSIL
z3+A*0@7~{rioPvMO~3fufv(E{lTU4>y!Fb!Kr-)-nGgC?#a`}3b69!SpXyJ~CP|vS
z3N{S5KOtJ@T;|F-(3SOH^>quK?JobZeH!qgU%w6q3Jg37La&q^2wcUC%L~^@OR0Vc
zz$qkCaT|s~(KRK}HTy@f_-$m26+5E*&Rl1I<m}l;zW@eXo6~!&SQ;5sRBvP%z<#&f
zvb^7fuzyL7%QR%_?{QUM)!-`yMdv>MQ<Ur#+MdSxLb&l?a!}YwL7V=OV4xq<$5HVr
zW}2bCHihF;59{;)mLQ6K63?D~s9r}W%~8%gChEkFFOp{x9Qs2G@mA3{<k;0c_<UI?
zY`JE(yU1j7(+s7#vgwwU?PI`H*96g1zu_^V7r6Vcat>w)p>~hP-<EA#(s@;K1_#HV
z=?z9l$K>hKi}J=I40Xn8guhb#dBl~6^loF%<8G6tcI*(HEr6(5XbKv`X>Z-9W4=?r
z(a<+EI_q|Zy<WORAn4Mv4%Mz3>h&Mb*~<O*8ZfA?wyN1&za?DJ?%`v61?}{^{>i6L
z#Z^^R#l_#}{*k4IX?-%(yDN&KJ|q{nyl|Z!$fA<e&I#|UxjfU>cTbOFO1G>UQBj+%
zcDtD{uNvdywmMsE_CjbyZu1g7O%4Cz>`R{k_dbvmd8{fN(=%Ous4^3z4goHkW!*D~
z-?567kqt9U&M7HkD6g*IEJLluv6I-{#Xd$hGgu>swenU^t*boO)){s78w5!UTHf#{
zuh630t6hrGgv9M^PN*Ga_T`Np@u0Jdg*qjvs)PECUoBZu|EzKFUNH}!h4WL^-8R^C
zL(oh7>~urGd*(5v6vM92>2ubiBe+bapK_Z%FhHrl5${GguNZvmz4|zY$GG`MWBJ?M
zKQ&QuHZe#pzE(y!$3<zs(H$5cd40HHB*fQT7kd@oJ+2}@hqmEY8<$n(QMGmY@qUiL
zZN+O(IgPUqSKlAmws%j8ffLh55XPkN+I_yQ*(@V14Qws<htg$2LPFAB0tN#pl7R~S
z<izPBzt;WK6|ftlU+|r>_Ni!+38w5GU(QotpPr<5zDoBmKM##W@k^@ye5ifZ1GYI6
zMGD}reoFrLc5QM~j}l(^s?kH_Ha<I^TbjvBFLL=Whk4RjI+rQ<!VhkLZTtsZt1-!H
zBe$Ti&RChkdGkGLOb3;~Vcoh{tA66+Bm0UV`PeFnCsaN%D4zbh%By@8RgAH*2rC_a
zuHvvDwL|b44tbdivbz`=0R`f1|2;vnK6K^dO}??BWV_4szgl^>^>4H<5q6#lPmGoD
z4rdzHGDiH|1>UR3!<?qN468fRmi(WkKb!*CSG5bmH!qB!jl4@vl$)(0IkGSOIC&&n
zegACxc22<NxG6+Bs5_5M_*=~d|9qA8Q!kc2!@AE`%>!36x(76Tt-s&w{tn%3T{I90
zReZ*1^o)7<alXRK%j$UYiq;?K|IKZdVoxySD$Nt4qrEUF$aF_D564vNS-pAlH|2%*
z%MdbqP+5KSXy!eMdmBH+PQiB!z_}ke`Mf#lw)b})rHgiu9f2L4<%wQ0yQO(uxjW?B
zw8nk6If*5ceLmVcF3%U<uUP~B(LJPIqLhVPF&2=gGvan@ao%M<8KL9a0geO@K6q-U
z)x;TgyV=L&19X9B9u-=S(~CkALHAsxy9}2I^Uux){)@JzS)K~SyS+D`n@M%=b0WMn
zo3~~FI)@r0tBX_9Eni=1$jbIKT``9CEbj8#Psl2p=Z9@6;x271NgKBq5KG!|YnX%`
zls+G}w=Jie4nsIeXzm){n~L;5uHYWefFb%SpGOq<@%nASF0sd_e`F8}q)ZzoBQ4CU
zJK%Nu<sPgdv9UR<ENfJREe@wemT7H^C}+rdM8pp`ApWR4#(H#nbD$Iontdp<wK#U>
z{;!Ds8{~Y!0n`(vh?jNYRi4BTd;zX$Y3PeO6M(gLf!ZD~FRz`T2rVrw%)|sdH`Uxk
zZvmGM6u>}9Q_6KIc+md~J4<e1BOcU#an8tCzbITeFM1svO$Z_iSt$z*#YAC~_|xL5
zn}xCx^qYo+q{Lfgl|8;><$In5=8dGh#fTQT$Z%;Xri2O0t^)nYR{Ocg;A9Co<R(D7
zH&byLLQz=rg}i5i2bNz6eJuL<c0VX5wg!!US&waCrfqQQ_flDj$BnlzV^o!>8)s~j
zvR-#D>pM-&CN6N2&+$t4G9CEFdMO=-bKdv)E%wz43>4ExM0c#a>zl}Jf_rt!6bp?`
z^0;}mRs*rFv<(UC+-7x-2Viuquzy@?_K!Nb?*GY|Rz_-uf6Jc!9xviikCxu%l5q*W
zeQ#~Wu{jyUqKcHPs^YN!o&|anyLR1RRv8^YIVlP!3{<8}%M|*(vcNz@Sm5ZpbGFSk
z+9%8X7umW90+YwYCjk!9;fGwVOP%5Djc3Tv#nHMgkTulR?FuLYR_3674*H{{K$-}x
zFChSq7lzZ`un@h#1@sd^LCa^^DyslW(XqH+UO|64c1c~CRPB6Z#t$V-msSH?Z++|#
zgNjpOC_h1D<FEvUt}v<Q(qhqz87kiI`7+|yB+AqLWf;hQdt}~z6-V$oN)&o!sW7%o
zH-0JSBnn&IM{8|5_ojZg^ApJyh0YQ9-#Gp2j;$LAw&S6?E{m|xuDu>j>?DRhuT1z3
zyPheYA(FearJ?5x&928YL#zW1e~waZJ1!QG()97eI`Rq&L3=mzb-WvkujpGnHc-R5
z2P{F=Kfg58`tzX#{6-SYyHoEiDDX7$KBUf65qVWOCfpR=!MypGN)QwBwB3X=|B}F{
z$t6D`9h1ydRB$0cE;nd!(WnLX!sO)fpi&+G^C#3CPAjui|5yT}!bj!oVu<co-<IAm
z&@~zp%6*t&dIMall}WY=GYX$PFh5mtnT&m<z6*(GsY>3wJzq<X21hrfJ}SF2t?aP5
zKBKsTeRbF_n`Cv?Z^1JL14>|&Hl6#UyVx~5&*kQ-22_HRlar4!fQ+M2&kyj%a&mIc
z&(197>d!?eO4+QD0Fnq50@Zyk-XB-?YHgjJpTH|ldk2E2Pg5I39We)TOjCO&_4q{4
z8`4dVInnW?U$R+~aisNShrEs-tj>0{fm?Hw#%NnZcLTgfZG7I(<(S-v+oXf&d{zY1
z&SReT4{UB4&Q4iaDjY#6#q}47HFvkSMj(S*WHJPQw(+30;B{>eCZz}sj`RaHd5}Oj
zZcjzW$LFP`;S&*6f(C_{SW{qN<<-HYjg6u=AqFZ75x}0$B1(8TNL92Zq?Nc{V%XbU
zOu<EtNdxTfE9h=s*p-S?tH1trSt|gETXA4Qy$|CPJ{y^(JdTWzq!gWq0{8yJ{mlh=
zESgc)u4s^Ae;3<_(i%n4g+BK~Rb7%U52BQ!0;TlV=U;Dsv@~()eiZ3DE5G`Ekei8o
zm!0~|tqI5f?J@$rME1A!-<?hS69j<%b%3a-D1&kmHy@v}vNF!YhYzu^^eZhU0oX=}
zjlSl=nV;Xi18IPNIt(}TjMdK5N_#RYVLFKxvn!KnxMbtxrY5PM+}Ka+emVS^el@H6
zy03rR^gLttp|Qr8d&4muR4t$rfp#Ty+e35OwaTq8(K8b8ME|6z7p!4RF=rW=nh+TZ
zk=uT6`N1^b+%j!<<K6!Ar^`o)Cc`2s;wKM~(H=;vg|y@n!P!%37B#{ihwtyUR|)+^
zk{SRp+nTM@GB9YWuh(064X`1yvfnBzPk@KBoxMF$bASI1$iq)gP5{ImI9XH{e;2`g
z?O_9ibT&2sGz7Ggv;h85yT-;yZJO*pD)Pbi8rAfdUA96#h70>=@i)TGPtF?8=wgA0
z!K6;y7NY!@YgOpQ{ENic=zye>Bu$R84W*@^<&kT}qj-vud!E^RCjItRyfg-H$TmUz
ziP^&x?;`&~o1S!Bp=x_LptFovyD7OE4XRx6>&e17-LcG~(}Y+388U;=<H<Kbx9%(T
zft-=D_&ZxC{GJs=tR-F<8JGpNqrA`mduC3K*xmuLThK84-0Ft~O3D${F8d485Q&Hj
zw{}A$s9S@12pkc-y}d!Lb?QYxSKU32?Wqa~_wnNfAQ%xVe{$}21Dd6v`<t7QQDHX5
zBmi6ce=mQei%r>Nr|c3S51<$28BHrhjjj-p{j22jQ*?;Zpm_Jndxe(Ww=II*@5J4m
zzs5J54KgI=n559Y&A*Qd<|9Nc7_XS=ur9S^^NHdvukGONB_FP;8=g)-a`RHTDw4B{
z^Qsht=rWgkwzf5_O4`DDh06NaJ|M%M&Lg&4cz5$tHqrRL+qB66P<%-NNDUx?2X=sA
zt&bl+A~hor25Ih~h792n6HipeSNhbWU4b?+pT%DQdj?u2hUVsVwyXVndjL)Q6|^Y8
zFQC3WJvsUKKY4@%QFk3l$@Cjb3x_OzeaOB3cUmOdESB|#87ytTo!P*EFcHl@{K@g#
zdxrn|ApLC5=P2_xh=%?DWFdLhq$=Ge@fxP*HK0MGd^Y_vK@lE38Q=MB4xFuB4~_9Z
z#VAS1F`Gfn>IU-4^7;$-FK}P2NBFvkF_1&woRVnD{KQDEsw*ZVXir2e<7SEo|JzG=
zY#z9qoQ=Rw(Qzp2K@Ssfz2f2|Kxlvst*k^e{Rua39~mB=azOm}e#&A108V$kKr2wC
z=;_gIbgStt23lZXSC*^7A?Mu!Jw=-pPf;TiD<58}J!;M*ZPkLT^mC7c7m5B%lBPX;
zk;K+m_!z!HRrgqv=f%u_F+^XCtMxFa?!!Q>l*XXwr@&Xe6)XtYbWzn)#A&>_neg>*
zu;`na%g;vl{1yCl)?^3CGoW#tP>htJFrZwX;fdx;7rX#cG^6O{=!%P}+p#?mXX_4b
zuacrc;%s3I?jp{eeGCI2>CP1!xGgir7HC!ybmlsEVP`{L$}>DhB2=s;Un-3O)A$T#
zSzLKirg6E`3x6BvWH%p>v|Quf2u^K?ubuKE@CCLgF|4}A*4DQjR514=HGO?7NKi1M
z4lR-uV8;s@OaU&snXlp*mBg;hYcq#KBng`Es2^N^TbHkwyal6M+$mxPNE)w~9<zlt
zE&mOXjG@ADa4dg5o~oKdpT7Zr|Ec9{?O>iPGS!19%UTOj&;1=3OG{*pm0M_@P_{pK
z*0N)TO+@&Bo`dyvp_K$X3<p9C>}i>OsH$<;l4_fnogYoi5^j9Ali3>d6uYwQd48d}
zY!9o*7p^BoJ{f8Gf16i+XpPUgU;OU*qZE>AXq3kHzWDMSeP~j;_Z9Og9Cp0D{lVm|
z{xr#t??hxVa_O%v6^=&^<aDC9e>Woqa?g<9-_4iZ6mTi^Sp|)h%EJ=})a*P)9`_o1
zG~wg{hJg@(cm#qxrcA*JvW10g0_?9PB~jW=Ty`-)#RR@N&l0xoa9<BkFx!~BUQ|`n
zy6q<!)nLUq;u)K`fzS6U0^?kaHMYSm;Ume}O$`rRt%N*g91{IyeOhsdTuIsVXA!f1
zHb{_xw?t20+Id9FE4>~eLNv>k_}y)mBJ<KUq4towEXCr_PFb4617A+p@)#}lY2T(q
z$kY0ke|fCur1QcVDz>^I(NvsmU5kplXpiKhRhFPTClU;SFAmf0b94zTfHR^@pJy6T
z<yb{z90U%mROQJo*OKUpklod{dX1k4GM;LUUevy+%B>0^=L6=*fZbkV(6q853wCqR
zvXx7YPzr6359_U>1qHC8rkh&*1MzH2y9%B+8oNK!zaFlNeMzr(#oAdUxng7!rV*j9
ztIn-_0nQ(@T#njUQYS&jc<yo_BOcew*H5;*ytwgs!B}ePF!+Np50X!6s2_MHG_S@!
zYjea-;`ecOlFh{xTsy8OdBq7D`dN87EAtjy=22QW<^<4Bk*UptbSLZ>h7HXCrRgS{
z@dguicF|zPZkFsE^U-mY?OnseKh<8Ln(Z8Lmq0<crhaVq{AZo$=XcwnsA}RW$=UcX
zJYwQ15dQ#apm3oIzSh5gtAT5ATAD;!LuqM5{Zu!L?u$8O%2b@Zrl5~H4llTn<N}Z8
zX21O1Q|nN9ov^dZF^fpAurB8wzuphxU=47ypVvAUa&|shT|v25BlOV55rhDJt1yze
zD9(4Q)6Lzb3B;>f_rtXPc;b*4D{H1{Ky+HMmApH7T+$EK_X@rzC9&aRq@|^;+10jX
z0MvpSBO&xHAmEFmBOfOxvL|p^2q<1y_y9fU;8?GwEAf<}suEbd2DBr_M#mVRoOJXY
z&M1V*fAtkmhVGuDm>(yF%b#F@Wvr{K8_*881rZVW_q^YmnLU6&%Muy{pa*~yKs3QH
zfJ7{`03;8(=UhYljJmSN0VS>fp#So_kCg{`Ig0xB<s1gbV^OEEICo<R%AmD49xi(E
zq2c6!!(2;fGr|XytqzI+9v#)Q<YNbNGnhner+FV?&e$+Y5rFIlX@T{`z>01$xa|h%
z5fq1f?d+@qnz`&0@%u-If=!NzwPulZGEcy@!9P$jKfkzeadHAZ+}?P;unu7RDJj|d
z>s2H_MSr>JNO`$V)qrZ=KSGz$M+%ezpL@UGgr!Hm&0Mu{7A8OuNp2KjRm2jrdy7E>
z-(An$MsS4iA{3vHaA}#&qDS!mzy!<oCtH(1ss&)ofs#+zqy_L-0#d&AYwloh^?iV;
z%pTHCD|$7&CiUN4fv~9FvkFjaWM%&rzi$QTU;tGI4Oajby2H#U_JxW-n3zug^9Z;S
z+gZn(@PJT3009FpPkeJ*+tR-l?_B{ZJ(RHjy&Q%+>St8SyYakx=DKxZ^M69p_}jU&
z56%EwD>yjV(IfHd`g(dZ^uIv|KLiu|xqKTV0fcHQ)idndx1R#G#-P|P|5d0flH~vX
z@ni3`KyO6i*Lw4Ra9LRyFw+_Sz46$cvEmqO1;ykiPQw0qx0uci+dvm07AK*^)LJDT
zi1J-w9(r*(P&SONQ2y`jQ4yz&oDC)ioG|QSTq%oPx6OZs-u?Eu4@kzv<>l%s%^~~$
z_aLbbPpy$|w<ftDjE4{Fo%TPtE_vVm?QZ<nK8`z&H#-X0g!~{t4Udhz4}lAO%KERT
z^WQDgdF#>c9UmX-u@OT@M@Li7sQ&Lss5p>)P-(6Pwe1vbLYjfDT5e7b5)kKf6I%Z7
z^8a1^rCppDD16fH1qKF!uCIvO*&gVAoAmyCL39YOr5`@$oR`|WZX3;yo4SXh--qPx
z!uFGMcy1~`=Pc{~Gc779<0T@aB_clV(hw{=7KMkIp(n>wfBmP9i2RfZ)RgsIkDkyL
zHF{ZE<s*7pR_X>ibmK{1$^77CKD~9_{&EzQ_WCJAvcBn5C)AhtZ$jaJ7htLqSk(h?
zB3+0B?e0?t5KIH{CopqJ0MfM49e~?+=h#eBPrcc3EzVcB#Imfv={%78X;LUwYq#3f
zUt!P{e0wv260_qM#P>h<Clai%XSeNP)?r2#!gSrf?a#g==<sTR6+A%FOv&(c-z(t*
z52>rurW-X?eP*++Zub6ec39N@cs}qvvMH9QvX6@8e-`hW`&ybzwf#MCe1%boy#<hE
z0NDUqXbGTIQ(wQey&V!9%t#p6dj=4(J<c+tq2pF!E*m-+KvN+nn0#e<s<pB{ZR5S+
z@3J;6#PtDR2$Mj9$JvZ!sSAJs-CFd%Gv@*npa?Y;n;Uq1vA2Gq^DL_2h1L&u(H9H)
zF}u=+%fw&^-FaP7A`>%k34!de-J&X?0rlC=YEzQ_-@KH)i@IeGDPiZ=A|Y~#|HPUb
zvL1)z;+}FYUKKoN3#&SA)+Uflyx)p$ceaZ5KdWcdA<+c2F%J}ja4;}%8;CGafz1N=
zz<q5Wkr2+nOfIPb?*YWS(k@fwVg1ocp^`n0a<9iA>sG+syCc%AXaAoo^v=lGY4P6x
znHf{sNezL2SV9$VUCPp{E%|k1QRb%Vrl7^p=tFf9x&KNTg5DQ(EVEY=8FlmY%h8i)
z^s_9%NaEClHy0D8(4e}O$k8Te*c{;B*eZ}N3)z?N)@w}1M)SZFwa<-<(G%HrPg(!x
zZDgn0C0#V!E$JmzpU;5JV}3c0iJXcHOcc-M)EiKM9g}vIlZ(_=&#!WdeJ9E=c3g~*
zV|(MUQ-Zs`Yjmn`N|>6%i&t+|V~;`@3A<N}Y}GZQ{xA$HP%cZt*ek0<KO)fF_GTH@
z1$?2s?JaTDV<nsI_jPV_zj&s@mWX4LYw!e_KWEQSiAaDhDlTrMHc`^+sGqGL#(2kh
z54hC{kGV+-$6Zh;Li8<<#O%J_?NwTmxSncjLf`7q+$=4(>%3L62ot03K`1kcGp<vw
z46bda*678y39RAN5z_&-9n)MsO`}M}{T(HT**fjwH!S-vL1KG=w!cyWEF{UG>>S2{
zD2-(0$J=GwgEie7wLgDxv;4!8U_BC!saSpaLGDF{6K&p)zFea$BF&`x%DYB6m)oCW
zc8sT}m?Xg_kj=ayEV-wTr6heLHO5oi^F{;uw?UWgf(}Cg9G^`=Iie&N7?pC_22pm<
z%{~k<V;Y{)Yu7w{@Vbs}i7*P;v!bRXur$K1{z;;hi4j>e=mn}zxqZN(k*dXyZn!q9
z7ng-ROP>@pC8kz*)e1_M<a$MF>phOt?q9}CS}i5IKKyhn<#;<sL?vi&cRP917|nQR
zp<Zb46$4XNv5(t(BQt8QW^t{cm+1%R$N?t+z)H}9la|!f-OFrjTg_S%(-q%x^5d7`
z`<W#>O<UUQ?6%PviMt`#br2HWJ?;4d9EF&A_ezqNH=9-5Hk;}Nd3QU#qYm*%g}wH&
zY=<67$0_4P;za$o#8raIb&<d{QA&nIua~+ad+$C9yz!PG&jF4;^Ya|8Cl+K_$%J-n
za(05}X*q|;goKPb8@avRWc%%~65?!omusopnJi>#q`!M=rnCKbwX$+u-TapqF}pom
zxv{PDe;N*H9!JZeLA+XK)*>K=(G@MTnqKAy?M%YT-lKzJzNk=DwM+oHuGLb;OvExt
z_mNnb2r3E9oVphme42E3pQB^AiEAV^cU4@|<l2ja<sS3#e84Kgvu0sp0$3&yprbcl
zpbOt#pMF_vujpw|sVjcY0ZPHBFP6ADQf50>N^|FQ#wqb)t`2Y1fIFUL<wj4A=ygIg
z;d)wLs_MY90^Bv#o1GmTZSdQIiMpve_+jMXON>-DFm#5z$`CYyKGl;yC+)`{|HnAm
zz~hI;bf?HUzy0~oxjt0P`)RpJdFf9-#wwe%5@Vbrr{X7$w@K>Uy<f?Bva`@(Y_zLF
z{kk8lx-T0w)zrXATUJ_{HuMRYb^-u7h`PW}7;w^ok0<zffWWT6q?fKuP11bjY;@K0
zFcztcv<5${k8K7U1@xD$dke=d1a<KDK|<djmEZd?xvinMo7v3Jmb5!C;`$J6iSl@X
ze1+EMdwJ<b`{gS%01qt`?wOcR+&}a`sgC(>Ffgwl)^$ES@JjLgIQ><!eGIC^2Y8yo
znjJm8cqxq*N3B%OypeKU_)m}K(OX_++$)yNp`hPx#;pGA+(VjQN(U1+jLa>wX%62$
zLyOL89l^K#$%7?~qUZ-ER%|1-UL*DA9z}@1!L8aisn$uj%+LMci<#X^(^nd_(3&~p
z>ALarq4apW+g9GUrb`D^y+>Q-=5X<;@o+=il?!&5lE(8fjcu!3F2`@_v200X-~eB9
zZ@Fk_E&pUy)*1MSnNH{KC~CzKgFHh+`vIdSC@5%gVS!D*)()sL0$?3LZv!~{`o;!w
zGjJdSw9hGKYdSII2}%f;=<tb>CRWoe!W^HKzE=)<&Nm)eICuXmEGyd|#YJVqbya)s
zq$>~SS5Z>k@y~>T-4#RTyP~L6Qf4%UFDleay2MzrqyN;BCnE>N0?D(Lpv@>+9$hb>
z`|Y*0p^j>FTV!`|#KyYN?v3-|d62NCJxWa9vPtx~UQmBh#p+y(D|^PJ0725tGJ*yt
ziTJtsU%kc@PYp+yiN$iIur9|}gOkAZAlK<+?sAou)27>OQcpC#XTW3}O{^!b;nZU=
zT|L%%eh)1ON3@Y5T4P`y2~wu+eU{hoC|T&^M8jVNQYx~$6@!N%@mKrCZB1Fa&Rid<
z$^Ft{=$()$U=h2C8UV~PILARrIfkDI*eJ2+*B(v573Lr+Vj(i``N6vVozsi~(f3%n
zP|4tC6S0K7_e6$$u7;eF8Be)ie}I{dp{j^@{=kX+6qWv$Au?1W+-TX3b6Q!!jpQlP
z!5%3y6F`?~Xenez3H1ga83#WeUeTrY5k=d#4*ZPc|D&*aznCw=si-sWYxlO+VueO@
zBiRmIeqMIt?HU)cJ$9Txk4_b%$pd*7MRpyOH1H{HpH4zTJVt^qnoGWGq0bf0x2gr6
z&bhG>;@56;bGdE7apSO8vKk+4U!Y)rSXT0@tfM83{aYm|^$FYT2T?;in#>(_1+i<Q
zg~>dLnbd^%Fv9%{1C8>9{_Tu=tj-(L!YugY`tKY%G1i>ugzL95QfLqDx8iCVs{Y=@
zU9`M5GQ9EZ2++6iANT^e<kQ=We=R7`Pd@7;0H0U0-ak45;=4C@dW=9$dPPdLu8e~(
ziL?9k==eK`YCPHDgy(La=3!PE74~6CA}~p;c)F@3tHN8O<)8(BmlIJuHyGqD%K8_Z
z!vbE|yv7<l$3>+Fp?cMkYAU(iDuvdDF(tWHelO3DzG$jh9=}xKX0q@5J+M*$ynxmC
z?AO`VUS3q6mTRe`fW3g<brH|M2gOsvK=F}*6G$P31DVNzNkh1CZu->ks=uV>5$u6X
zoAhjt;iK-!*N#03%J%KBuDolgZ+))m5|avJeC9E2kRh~Z%aOcgRJU*znLiu{Q)PvG
zuAt6pBS8IrTpwFFwZndo%rxiw{z$#$#MR}IcJ0zzItiMR+Y?Mk#f0&OccgXf;&`7T
zeYOC3rr?H{P*uA_z=_=%3~u7Zxc6CF;Kf@_7r1x-wT8>Jat7?*5dK7Q-ocGB??<A(
z=122ilfMV<-_2wPr(ivQ(^-rwP%wQ)*bSTu8qZhAOEXeZfZ4W)NIaY#6BENoVow<#
z)3;Cl?;bQ(aSW#Gxx?Yjah;NIutLg+DV%>XT+yK$Ip*(lk|Z*R19945#$1(KpUsS^
zHC{Vk7ZSiVzD5QLl>4jg69{~Rg6Zq!T7Ao&4r|Zp%#1QqZl<t~+tZEK<))kKSrDsZ
zG<M7%jB03{SnF|G{fu#=;*iMR%rpB3d1eMKX_ED+{8Ej6WD2S1vpWlBiZkjF4eF-3
zuoqfjc>F+v2t?`vavdX5e{^^6TM<40GfiH(4DM`~VaMc>&qaE-|1BwP<1rNV$n=(c
zH#3F0W@sj{tv;cY42b{zRaI|@bosg2OErZ81u}11w=K!mt9nVCdstY5aE)wYBF;^<
zGuQm)&=X`;6$EJ`SL<dvTFtxDRqVpT^#D%~VB11~t^t+rIYFHHBMJ1LZ`EG^qJq!2
zFHyu`zTO@UWtm8&EYc0rFsdE*>*ke!8<by*u(s&w-(6<63xRmD4W{9BKqQTnxtI|K
z9VBZLtV5gY4+j{E06r>Rn-l&vUM-HV>-PQNF0j<7s^C^t)%?K7tP@Y#_khCT^sn_t
zKbOw^k|#c*q^OX%IiEq+GZn}aX0O1`_KyuMC|Sz3H#Fl=rbhM5nR6J{KQ14v;2B2t
zY)hEvFcVf}MuPN#GwdvVSRCK*>;2-$83wUuF=!BzT>1WWe#Rx$ww1JKl2cMoMD^&b
zg04&OtnB5n0_v4A2`F%0zOW<H-xX`pR8BoTI0y_2LzEg}#`K^;YHIKZNGLlyI{`O_
z@3645Qx=>IeAc#UvXmRSOy}=!dF}0gStG(5-7}_P1A*Jq=tFukf0tV9W4q|d0Wb!Z
zR}Md%dh+W0<7;y=a^9;w(~c*#ys=VJvzEe0T`|pJS}xnjC8O&Yj&C8x*!&9&qDQ{#
z-O$=ndLZRtLv~|ubV!B9uObj&roMFfqM>Pubv_}XDFVIO$dkvZjBLTnp&!1E-^^j|
znequ!QY+>hl<j>_q$#U(RyC*<6WS5E1m1TlTRLk;=QAcIK-Eg3#<=68%PT7b`K~np
zfRB%p1$lQFaP&iI21SXbuDe#B3xgzU^e<tec({i>H=ECNlG8~w!;)H~ncS?4aR~Fx
zG1Q!E<XTkoSE0(yt2n7uO(QK5ttK^d%gmrQaSQP)APWV0mtPR^6UFyYgxZ4rSaf_B
z`rUPvYZnF`8HQ?~9W5{dl54(Bg`BQ#eCvGhW1$n!J25PqT;Kq`J7BK!u|0VHGd~Pi
zv0V}L2P;7;=-HO<gp{?npWmSS?~4Ur8OIA>e@{%PIV^uXB8gEEnAf+dAslW>-zO6&
zEbrP7qDQ5R5+EPQoonYN{aCkF^{J=<21JpR5=)XhS}eR@XyDSxB~>llb^n%sJQT!a
zTxWv>tG9=#Vg%both9Rz0B;yZ0Sj;f-MpufHHKXgRPfah_;dq*CqU^|d;K~$_brt`
zrR6lQtE;PkwZDo?HF<2?@_yHZwy#l+z#nYQb}C+aAC-9LzvRzU7$v^)Qy!R=$7l3v
zuPvC7)nxr1hQtGuwizlY^q&$u9*g?g4e}9xqN=rb@?82AOJ4&^<L<$+TJm}1^g6xA
zqusgWUE4=SPaJ#rM%2uipu5CX|Mj%YKWX4m#BR{I_3s}90cA$zWKj`YAPdN4@`RSQ
zYh$*hg@&KsX?-~30D+kr`k9txzM#pEf%{Pa?#5Mcx%$0hZZt@fRfv)MIB_3+mwtlV
z2I-xJfw@s_@V<j7yfpt=IvPt+WRe))M@1F`kx-|kIaRY<A)#$e&6hxIl9_i8a_-fs
z%ujV~6qjxMgZy9pDsciG0aH4rT2EDfSp)*c&PQuQXqZF~DTV(6cFN2QAi=eOA`ZXJ
zQfCk`JGlSp<(C*0LYnroBflO(B>y&2QoA962TnY&-*9C5W-^67760D`5j81{ieX~O
zx=1yIvTufrNfZyk`SYBAHtIxBHW%S4*7n@*o5zY3NGY{R1Q(?$7iJ><#Ctu*UYl&&
z4d;5oN!qO*5wBVB^fYJBmZA+mDr4$HpZ@I?Ng2I5g}K<ON`Wz((uyFf0@xIAU7Eg4
z^!$Ux&XU<==AIor0oV`_MS>=lWf3->i<p(jD?$uk==m2}0}<AW9-0!gj7I|_hpv+q
z7lWt|eABY%KVp9J8FV!^O4$b?M3a8t@&TF{1<dLCebJ7}HDKr%Bw&AAqso*yeNbO$
zq+IdWgxyQekm+z0B0GIP@v6V*FFr+9JDhU5euC(B_WJ(U=2>B_7bcW``6fIheU>#$
zShn!v;Z5z~?-uoG@xx<HXNshMVa)r;sY^w~Hw~;xewiyjCMaFca_eH1*yZ(yGQb4$
zK7EVq8UCZ2U)#U;ZvIf|Y3ajrq#)Z18<g2_d4mjPGcsfCOiOA9cnHuDK~}I2LwUU?
zF}{?v{VY2B>^U-07=wytn%Z73qYs39@utN2>+A%9GxGdAdKU=!g||X19`cMvt@DJ;
z94&0D9lf^sDj1SmZ1tMTx<hlz?L}&HDHOwgi=AV+#3n{2Iy%|gB{3l@pHZB)V`hgf
z(BvK%Q?MrMz+-fAi#$7?><s_VeS5iVK!fC-zZJ}L7;Ar;O#F*j>-hP|E4FM15<EUG
z=g&np+Q(Rpet`%6sWpxW<+9bfMXG$N%G(t!y|AwN12m|3uqmq->*TJWIO_WzDNNK<
zdO7-jP7N>@8;VSoq8gljt0-Ml=;gloY9Y4<iIaJrGE(ya>Th@)?jv8CyM8g+BQ`K6
z-@~dc;CQs<>7_@BaPBsP{+54U;D&Lav(4Y$`nPjdqW-^Zo1P8dUcO?<%NB5Yw)E<Q
z4qn*<=%GEgi0foO$-1QpO-<F0O_NQszX{iQt1ic-(>7aE$HhLi_jnUuRsalCC6|G)
z^UsWuq@(S$w|D+a@|`R!Jxq+|?`%Wtc=|%@o?>tF<u?)36!A#tx6=(+%|7IDKsY76
z)f;P`7gJzz`bkH+_%5x&>}y>^fTp?7?)$<r*Q9{~;<TPz^4+H^2*qAv{WKM@z&GjB
zT?0CBwvC?BePfOJ&|^*(CaQo}5w>nVJF6bg@-MpW<{8N#JbP9L#fVwc@1y=^R-M#J
zE0vDv>_C!qjjiA{xr=~B)}=-!?vUG=pxfn~AOtBSizc?b`&LyxNNnIXl<VXpI2uxn
zYU46MLZt;n&qnT@u7BD7qW3IMke5D7Icy~}E2-hqt`!k8<>;UytI)mCV8U#^A<Zfv
zy$rta>W-+c;OuQ)0Ov`av8^-L`1LVondcRA7<(rRV-E{s-3|Zzp5HlTe`T61K<An5
zPUfY4S+E{p0;9%)8R9-xWKxw*PCS^+pW<Ijs6~FAj(8Z-&FCKSI^k!J;aI6TxpS@@
zx4WWWP)1TwWR5G}HLv5&oTAvL97I<~PB@%H|K|}FD(*_H8Sl%uk@{_m5T1<1YrWnA
zEN*NuS*70hHcrU$C?Nrqyda_t^{+>U9-xgzAug1iOJfTBy@$@f1^)~#9o~3Jjp2^<
zl1kSbWWp~?m9k{+it!<ZHC}oqgSE&WH&^FNL7tvasV8woIWEy&wi?aqz~-lH7|an`
z-Iq=R2Cfw75Z`ZtxM%+i3<<ZyU;`~R%kfXv@*hR`WSC>$C!f29?rqV}CPnBUU5CQH
zsPl8PaW}^J7GOw!8-zjt4vfp+*-+(8uknJ0gO#11&hHM8=iKuK0P>3BrL=^U=la==
zv}4x>?vwl*?aQw%oL}%>gQxKZ&XHRJ6SE|73NspGqg=OiNVFT`NjDTI#|OHll99jb
z{;oZGE?8GAydk}UAg1)-8Y_B~Eb1eAkMO}!71tJ`^l-jS<i`28&_}Q@VEUix@z<MR
zUY7HWUs+~mNzppz);HdhbdjsCDV#dsXwV%#ybf86tNm5f;&iat1``>71IxCnh@il0
zmqNRoHbEjxR3AVZifjU*ccY$jX-tkO@#%r3uTkP{FJCRxo-fA5Y1~Ylqg#b)UTXD?
zseI(N#bmXA#A@H=Rk~5@x6NSv;J$>iy!$B*S69FVa_U|f0UsnT8B$_BHshhhYq-~z
z>dt_@UAH!op1)BG&t~lgA33BpT{C-lJnLzwJNB|a9956%BBJ8!xZM`DT~@VlHd1`)
zk+Y_7sa8b`0T$x!?@H^g>yuYpGR|mhh-lqjx45YNFOT*DCYLc!5h#!l!8Sq9%kJRw
zb%mvZzd*Zt%d$zhIIQ<4DLm83RB+EzQvNED#m=a`Jcb6|?B}sLJM<TV=U1=#{|WLA
z>;j$gVw(?5;yBZe?deL)EH!J*4KdJO2ZeQO75cz^<fM8gFFv6$+Pm=J1w)(!VyxV%
zi}?@>#d|_%XT0Jyx*G;Y#n5rF(}A>SSYEbjfD-vk)*S+DG_s?oiw^g}L?6|J+c8_@
zcj_OzFu#V1&WrTdr&rcL8Y}Njda!bf(l!x7M2cigxpS}BPKe}1$+01D_z_b|y|T*|
zPqjF+sn2#{P<%X>>=v&3o6cbG4B1Qk)2}=Mcb_b$V{C}2TW$%}=L%obTr2@6&)>HY
zA2`y)(raK*B=!DWJHkKsQ&x^2i)3#>S+h(6ONOs}^7FSrf}%8ega@9%;XPM_0d|io
z+jX=vc)-^Y|A3B7M7Eo~FdyBsMMP0d%IvP!b@8*@3iYj)mW_p+hDOD8#EZ3tTD%_+
zWb7wci2Y+>o3`FHT&4x_Q2EHAB<B1ru@I&Vb0qJK-9%GW>!KRcyX94+o!Q&?G??m?
z2^~@=l}kRCS1LmD43NCoxj1#uu$0u33zZUQ7s4DOsrwdivzaP9l&&loa<TSrDzh{<
z`<#)ga46?NzvczUIz2!<By?<iG@--$uCLRVZ%Svd8a)xEK2B~be-jy<(1DMS3LTY>
zXEFV|cLPBa>0CIoimyL1&Rv=g!~XplLC?oENe5^7g)`3fya~O&lgx}L$ZM2jr-?qP
zSUOABg5FtXbChf4g%wKOZ7K{0xF<M?(i1*vYeNhhM#e9!I0&Hjhj!7`ON%R=Dw-CC
zw!1H%T3(qU>yN)sTzP3pn%i)C^v6q{272D!VSloQv=Gy2n6Gd|+Rdp1r)mBgvFOA4
z7*Upi-c_JzWYLnbR~WejAooT2tc-b`OpLA(KOGoen_2IK`NW63P;j{q2@e>_5$~y5
zht<ktL<esp6ra50?%y*02qgyR;$OF4+=o(f6>bM>sGSRvhaJ0GI&!2n{C60rYLi(!
zx};=|t74PRuI{=76JD_EU;CdY&;9{m6p&S+6m{db_*)D@+rTLlWMlxK1l>)bZ7u(2
zwm{WH!pZ<0g7uyDCVbI3Kbmcjw-O9@k{;s+AM3cUscXKU>=rS>_*z`WJy-8k+BQ~?
z{}lR3%6qQYJSxucR6s~M*K=aT>d*P%9e~eTrj5?Tg!l(sR#Uo~08}7q8|GpaR@|&@
zZt5d_txPpg$};!0bxQ02+EJHnNMrRys%L2JLEWNC{3y4Ld>+}FdTvf(?izFwnw~2q
z1`0kx_noU24<5X+&)2CG^q%XJwS9*Z$>E;AC*^pzQXkNuN}Q!eQdca$_<guK;JEb{
z4<G-v4y<ElX11}kRBp2@2Rh^d#Q<|<zc~&pW4K}!`1tq$j8P*`b7=js&<-dky8)5Y
zZ+Nt?)zpwuTu-ay5`p|1P!$8$qXJz+P{;>B48^1Ua8nD5tHEb(Zk-{F_rI<^Ocwb{
zZB?%gA?owOKKwtX-a4qO@B1FUgkX`<tsqDv-AD)s2nYhwN;lHYMG%ngE~UG>k#6ZO
z0qO4k9p0b$&hs#@@gJE1&OP_+v)5XCErG=R^cmb&tc_^Z{2+hSvvc{KMuF*P=F1Z!
zH%bZ3w-+H9if_6s%~k?X?}1NXW0NT;ghK&&v@S~{KRgd-gqjV?w6oe){R_#OZuNk^
z_@0Y0`k0#*RpK3180DvB=@Jkc8+mk=wtBs}e{p?wcTzE~@*!1`hri`0@zBRd{2&e$
z@i|tC5(>C~a$aUAp*cJ|Oda;$A|`6fqQPw2>d}sqEgRLZuUg4y5?uj}vlHSk)}+%y
zKYYga<%2hySxOX`q0HPQ=!AdNc9IPf1bSZyU2uaCJlvWNu{c@M$vLCoqJ}VK5Qt+`
zARLV1Me=gZ`Jv-%hwMIo&^0z7^1Z1!_r%CKhC*N#?eROJ>|UdpronZ};qZkDASu|_
zwgxvH2Mf?-t;j`WWatB*D`n;Jd%L{rgHGT+>-YQc=%_Q0Fd{bgAq1Y-|F}e8PFHQa
z5gsgyYLQ1}i7>cyDx047e_DdCVq<kc^av*0&c~XYn}G)>Xp{E$;eDX=>}+eZ4YH)>
zOkl*Z50{W4C-kHsA|evB#S0CM)>iv~1PK<N>bhy9BYhXqx%ozKhR#eyol#mV?#FZy
z@q76oT?mux?xaXUgGuF@P8iel6m<>fHGs@6FLaOD35NH8&WP@d5`JW9{N;71&G!3~
z3NeHgHmta~AGiAO+E?zCv1%13KDjvrh*+_~d)e*kq#TR<;7QPXgqkK^J;A-Ztu3S6
zsMywA!y|>GUWVgQv7PP*Q7Z%XvYA9aoBrV8ZBEv&slGhbLAAv%-n>US)-5u-SBq<O
z<{G<-p1x;d`aA~ygUAU{z)qj?y-!L&TvpOI&(V)-C@}h+nwgd>XIXk~NoAN}c00@1
z2hxd&(yW@5#=n2HE54JAN3G8uhi^J_F31ObF-$+0g4Cbw+*Q6>d-9{%+yP-R0ZW$B
zIl*t5W~X}a-{bN3wmaq-F6%9MowP5}NCcB|Kc}}BvyET()g_cv28#Y33CQzGv({cw
z8Ra}YW{-pgcAio@sU4mF=D89<SWIC?jor-Nt&4D_7fcB(GCcCh$3XF@@!F<w`V@ff
zMXRH%?HEs*($7%z1<9kZHas9>*T~_CbctLezmt9W&;9MALF;z)N&c3GHE*n(3?Y=$
z?Jwh--5i>M>j&hUxsLo~<BpoR!kFhLJBQyzmiS<J0uVyv_5U{cJG&}4{eboww4jHM
z^~NKa<_nE;ZEfEN*Y>8$)_lp`CqYgw@Sg!$BVAp<znzMj`ttldk>6=}b{62F0xT>n
zAbo7ARKGKX-0}=VObkp~0*ey8lSYI7c(7-Oj4bV{L1qv*PHql+nSr_MkB|@{`0HXb
z@NrtL*ERfE5afGPP887o;-QGJ@GsF6$qM;TI<GEtv=%7_xoDcD9SIWb`Tjal`X)pw
zZSK8vimn6H!EthMDwZRFB@XiLk7jN`G;nty<F)@P6jxPN!4m8im7uMblbYEgAD-w4
zXfhxU2hk<JKaH|E-I92<z@Y=*@BT@sYicIdU$X29-iud-g_y@j!V~S}vgi>#fS%M=
zh4FtQ0rPiq5-{pW8wnn?qH>@~dfO|+QTY&x++b%mOsEy#J>3G=BdsWQ$J1-eh`X$*
zv6TRKbc56oO$sV^L1HI9CW&jN2l<zX%5E}vBLxXdyAQlzV;#0a?G@H0A#CN5eeH{s
z&ir28@wb^#!Dr=H3*Q_6-h^s%;NnI}mv~F;HiWL<SxI#lJ`lZ}Y)tMOTyAoge|13m
z);={mzjW7ZHqX<4q6P7bV4h%#(j9k(-r99bs;p!ddDG>QI6u$OJ#tIT^mdhBidL25
z;Djq3oYrt*p-*XPAmP=DgA1?$^KJDTU|6uBugOiOtX2CoB>!pXD__YX{hd~sFZVmR
zxS*#jZBDu`CVT$vCeO8Y5p{|xu+LE1aLXKZ+IOtq+%y3Wlp2TQFRf~3@0~6_vk=hH
zkuOdWI%9ldi1zn?>@~C$;r9o%MUNH1Ln_0a3XbaPtbKS3#J*tc3%H^JskWZEblOlf
zv$oc!Pu04e!3{2FKpyb_7k77Q4}*LPfHhiJT9V`7lsg?6Jb4m;-XlM6$Oeivk_S+6
z-)}MOF+M-34DfyKz=s?9x--oIK(x?nWW=SNt~J0;=8P9*!7KhoVPW${68|ju-$kz3
ze>uo;k)1>E_*RH6orz{imk%16|5U&8evysN8(*I0)q*rrlIe3xVOoGp$)J^I<T-d!
zBxgyJnV(=`6==#iyAbl%K=bY5DY8m`loc!4hEQ_McW!4|3TNQ5_`&ZDFY+9okydz9
zYPiP%|Hl|rB)p(a^It^>kX2kpCwU{z1C`NI6;fDZq**Cdk;ZY_{Jd|gY6=NI$oBQ0
zj|d)%UB{??tb$bz%QnHYE?)rr)B3&9x9Mbo2i_0W!UJZy*g{EuIF;LeCqh-vv~fRv
zTF)yYSUCHGeMhSO^abD2lkW+#5)Z%M*<qiSeuf7txPOKIQBq0f$?p6YH?MtpwrP;B
z^r2VF;n2@9g}&p=sr01G$z1>he8t`BZ5hf{x^vpdhfJgl_Z_A%wW*BVGhH5yoO#qI
z>U`~#e_Equ6)_+CHCeor;y*DmPu^=;s&WX4kRRz1ARby-3cJ5No6rJ&z&K-a@)HEd
zzSi{-Gv_lNrn=V>KEL}vMSJ-<W<W(nMO77-FeV_VtzCELh3B!(fr%^>%Wb7|^yPoV
zwH}KI`FKQIc~87EM6kOdWA5x8R^;N~gWV9;>VlJ`vC#EFzg%CQsVTC;wX;dYlZ9<`
zs?pV%Eul5!$d1pE7N*K8HoLn(ZP;d<B`FY#31>ZYNU24Fs4u4#eO=0*qd~>&L*NFg
zEU)%KOO=w>l?<xpzpY&6O3qqCYIb4Dn113!#P3UkX)b{Zp^?n@@>4^{Ze<*9u*-7!
z@1FG8Llktm)qWDx(tj$ciggKhXCZqTKWrhE5k;ZPf+iw(xk(g(+GT2R%8ouol{n(V
zrFRP36v09652aod4{Zaq<K5eZZqHi_<q`L$mVXDL;fFr8l7jZ#^hG=7y8E4Ljj21U
z0wjT4Nx{{mF^XDq^3C}#$?@2WLY3BZ3Yy&JDL&972UCL`QIDGdT7y2Of=Z5YM99+v
z-sPteN`)J=(XmL=x%xY;PW*}ubK3v90V3{=lZ`)u_lw3)Z(dOa3rL|tc0PZ{YXl1|
zId?GDO`rX#n!%6|1w%xrd;H{^I?qd<Ydkq}#jx%*JyPA{etqyx>LA8lw!ibgx!(tZ
zJ=zRT{K;`~acu(@(9iUAWSuL&P`k#arY5mT7+fbsL`4CO&dVg~B(b=)xe2`iT7?T1
zVB!V6Vq&roeERF`P5Iq*HbO(fBi&a+ZnGqS{$3Uebf1#ZQ*?A~nNSa&807%JwNM|K
zrshPf%J1-7#hGRlh)!wAcPb6~SzGw!Gz2cX+tT?e*%u{Il7Nmp_^B*KRp|+|01pz;
zHV`#=qtrOiD8`rFrkI8-E86&{Y3ZTNIbs*vUilF(frb*2*HX2iVa+LAqEttaPDa6i
zoK`u;(2BC`I}M}41}QHaXcw=hh><+rNVXY7hBKm(XxYExfVkB+0!(C+-@L+njK;1+
z?L6(LvIjNGr-spt;riizGxo-Xp*KoD9eD`?P>D@)21sVzz9&TG1h4dbg?`E}UNsA1
z`?dz{9ywdUEp`4f;fUU)I<DG&vK=Q?ZgY>yqFu4VBV$xG9zfb5AZueD!Q+>tI!8&$
zvGZ*vzw3wC&U_?~DX+S^Mz@vZ2=86br2$C(;|B);x-xU8M%4DGz4@Q;uk6xW0X<KE
zf!LXgWerVtK2#~x=pGy-XEzG9<%6ng8wFur0Rd77&=oI#1DOs85@9T0OsR+tb3mS-
zoaCmVp`oVEybyM@F8%iFOBP?oBR3On{@WWvX6TKabb}nm);})^$C^;o^o57g1k20g
z(=JLA9O?gXkURhY09!mUp)=dT*EK{1)|Ah062}L~t4`MK$h5zn?v9i>r1<NMDn-I-
z?8c)I;nhw>Xi?qrahO(?#alhR$bv~HsZrTq@&>3+kF878&wH_;^V73uT1@s5T<%l<
zh}|$(jQ5cdn5|V;Sh#tVw4{}~$InlAU*pre>GKBlFN5w8ern5#<gvn!$D^uQPm-aC
z%T$DBwkfV;7cR$TJ4SeanyMu02cEf_3OyKl)li-s?L6E%;A5G)g|~Y#+$y3{`|N`s
z)8?|V$!locn8h`*FWpOkG%zI{r((sP`G_6@kA#@U4u>2<KHfPU+*un#OI+hw#ZK|X
z`*%TBF9;hYK}MH3Z`nrl{zH!ykcWn~jRaAE)Oi7fzd-q)N<7VgWgZH&(a_L^XFV8(
z{sLHcfLt;_sDY--?BH+vupEUXu<QgRE7%Cy+V@2+k*qx(%&LGaONSzd+f2b~IS2i>
zya~Ne0_5_51YLe>U-OGW2-)Hm{Lk^Pv!j@f^Xt6F4}ZUr80l@_?S16+{&;n+Sg$o^
z*=>&0tI*dFH~Mwn=5hHbf1<fz5ZRwr5IqbQz5e=@PwP@QoOUL+&3<lO>Oq6wd{)AN
z?H0Dd{$RoaC9KFlIin%@F_jwIo&lv(e#}E*D6bL#3rtn{)N}<@RJ7Xgi_bp-+7Eo0
ziEFMGnBQs#^c?RP?~eEX8ItjpHMMY$@J<dSJS<=&)t4=cA(N-^t!6yo&_l0OCpM`l
z_5j==Dx^q?>ysLI$Gy(}z*MyV`!?&w1`=c6kpRnx(gUMHFvs{rp!@xc|5jJ+3ci+l
z`}Xv5q#sTSkd?CRgk4@o;tcLy|H{d;jo08S<78(CgN9BsmF9@w7PC^^iTE8^rew*c
zMP%9Vvwr<TAa$jS0JBHoFQtP?`IHqD{_gEH0j2COQC}%!F)=Y$U@6kx-rm#Gb1!o_
zvVYj!y)eSV;i&$^3#g?04B;pvna1LZ9LKl3@9S*@7*F8n1?e^b>2*GM9}EQq2;JBy
z#yqv-(4Th>b#bs}YACPJ4%EDApTwh{n!2ufD(L$NTIn&tREOj?EiG{_a$lvWs(<`o
zn3^We6t0r~A&KFbiWkUoYH-4$Lp76Cv|5)Q#n)8SBwvE#Tb)@Pw{%nCF&+3o9<oa7
z`WZrS*VODnztqi+(>axBf&C<+nu=ye@WF7G|MX8kzDoWGL_h`;m~dVY0JABM=fsaQ
z9-rUuTmsX)=>($;TqTZY%gcJheM$ghcv#Uk@H6k^eSS|LEuHB}{i%VKV?767LgH4f
z%^1G<?ghB$gJl^U98^(MoB=9)&j(LnvHMv)kV)OWy7ITH<V@BL9mMqXf??>1eAS5X
za8|QP7I4CV`!_H<fB7;4G_AnuR7^r5A`Bo7U#*O<kip=m$>W-*g47LEJSi_Ru~+u_
zpyLu3(<$_tr({s-uoR+znf6L4^fN0Xxy|UJ=@;#lC3PrG;ZXMD(@c2!pYvw)B@qD9
zzA!&wdMhQ!wrgr=d+F=Z6`5MR@;fW%ZC|{2@*9E#RiXq{=at(4@+@aZUa9f#jM`4~
zfq3>UPByi_WJd7}4+ky~^i(t&SB`>ZIbOPj!Gn7t0>&mOFePKuyX5!DGT*=o{lb&#
zWy;TXmz$Xdz(#s`BAi)>Z64n}sK*Natb0dOD3(>fXRqnb!q}LW3^<-@r~!lR56=%e
zdU~V|O^7X~MSv<ADt10G0cj2ZsR34X_v2%$(%n0aAXVcYAOMy_-7xca%y2&ql!dNU
zLmZXx2lY9H@!d1TU8=+;L`J!hN?c+YxOFo`;V|F4A2|goR1%i~YlVQBU(tSVO3h0?
z0($@cdgpjL7aCb!rIxI$I5{~s=#}1ByJ3`1<Ew{qO!@h`cjtQ%Y(mKE#;(PRPW`rH
z5yYQX$4axTqd13{`aeok<kyyfPYz@q+~Bm9>^VO;uzyw3;GiY9^5i>q%+rdGm4jga
z*Sz_UX)ZV&V){1N-+yv?%EZ86Yi-T@b<qPTY-|I=!muD5+;@edR1W4`uQ!?>xgG9`
z$i=f6$~rq=0u`P_Ai+?9I@jSJenrI*VK0;d)iP=}HXG21?pOnmI((yPWYH#84}tkL
zoGga^8?d_61~Flb*u=RD{@pMN#QXlspo%CeZyar7gM}1DWd5ZT-Ig@nmMKwnDxRWd
zg-7FDrQ2?uvjPld8y<<VAqqaUakB%_o_S$B+cJe5MvnlweplwOVD(;e2dO!Q;o<Wy
zdkt9(SR{+c?FDvVj>ZHwuecwl`2srX*3+?Z^nx_Cf*F6}w{PFTVxbLZ)Ni^xw-c_e
zmURN3BZsoI9o?KuaG$Nri$089OSVZyg*Ip2Vw)>oK9+fx=wCf6S?O5cxr+v}ERzEF
zRG!%uvuN?n?G<t9%$YeC&HGY3^rVm^f9i5`Cc%IVM7F^H51=_fry3I%m&4(V>H%mV
zsM`4?SDbGM8z#!l8y|x<%yYj@{*~RfR(XZP8W-5|r<Ga`Sva&&-rP$lL2HN_QxpT?
z+EKgC#(|!<tyH%=1E_t&JF5lD5gL^!5Fr;EgR`U=ZglSolAsQ>CN1-S_OkmGcL(ho
zM7}RvYf=6@4->lN;{1Foix;z}RQ0H>+Z&C@Z(aR76^{JvBbd^8?P?J8DU;hJ>u|#u
zfk<M|@S`;!6~vOMF7geltFvXhOlcW-U4;|McN>k93S%M*9|yi#IwhE{bFh5(uDjmp
z==S<Zj&u&#uT{HUodP^EK<tIm<jnZE{c;ByC1n~o$({Gbvr`Bi(LT~CQ91)h204L~
zQFTkWfrB7M6f>PhiUi;G2l<H5&zsa-6z%b@2N0_t&TdAAh5q*{(y;BMb-Z<*jMtsA
z`H-c_9>p$Vb{B3>%76jhI1oO&(am%YkwC1yQ!mlt6MlZg9pc!<dRpRyl`JNFrHNmv
zK8n{Ty*gP%v!~NTwm?kOi>Bk5_=8~=wGbcvr(BgwPM5oUZN-BRZTAJ?%k2;Eoufw4
zwnA_Bz>|E)`q0Pi?+upW;@S2T5|qGc{@eMDZ33^|CWxboPe=ekr8F_Xhn)(WQN2=z
zE~-$iqNh)$%%hMDZ3s06fQf8|11toADejlXMuk5p<XA~Mmm^9+U`~>va=;fU7>S6Z
zZ_uuPHe2<XbH0q=W6YUFzfQs6f0@(%F|n*q-AMcWG!FilBk@BrhYtL4YSf*PMR)Ey
z-jlD%><tjZ%2VTDhMM%o_eOe>xYD@hhzBS2wUU;6y1eEGjz#I$&ODi}R%=lgSY+Rr
zqW{UHDL1~#`FU$^MG1YcojK=y3LEMRQYTgF8@F}(DwwDtGw;@yQv7b6X{j%K9~-Kv
zINpW!wUTOAN}d*DcTJ&755knjov=0$*p`X2jjy2f5}_6l8oN4DoI(uSu2?AS57>Wb
zUGXAJ@<E09!A6wWZf%f_2pWoV{)78X6FRi`L1%3VtQS!30tDNf!xN5)QBhS@5nKjV
zy0Y=?8^EI$9K>J@zOSbzT#^bP5({n<dvqUfQhgz^)?!jV``K!%Gys!jp#v{&=(ACF
z0|A96g)>zY{rvT5(H7y`T)a39E1IIevRdt)v|kqLOZ-I$BQv;KWh>CgoIeT4{Wz{r
zd9eS5EWYeG;+h1GmOT`VQ&XOMF=@}OG$#9^rKh_QYu-NLY#!O80G{*Tubep;4)k+J
zH`PVFlVXtn+{0$PYB=08Hu!JL=-}ii;w0S0okB-c!|J80)3q$UcKMk}W3Jnhqa2)7
z2y8rE_7}bt=}wUPhM>PhYVN4{H7JV#c^u4hGPhsuG-$zed#C-Z8|93>FLXu8r2hEd
zpB`NikH~qX*RR<8r>E7%NL-5Q>+4I)%0M=#J-BWHr9LXE?@Q@u*TZg_%*@OX)&W6c
zBdF*j9X`XSuizURZMABwS0QF!3dthDo1Jv$<3c7u+TQ6|z22iEUq?w%%I5ZP*)(k0
z@eU{WgrGU(tVHN*0rW%ZY#Oi1(w$BxHD&o=C43W4*u{6!)0lmr*ZyrebBUV!s@}pA
zLL~(r149d$V_4r_Ql_|fDdBCDnP38a_?OrQo{?^U9?Qhe6(~z9uLvzD?n|LXCAACo
zp*5X)hIL>6xZVa;3fPKQMYs^;hb%)Gp9>#}3HhSgh{`Q6EkA6_?C|;5hkB-k|GiEI
z<AmF(?%G1t=84?*<<I6OI7{Yd+sDw@6Yx%gC3<GLBGYoFO8_ReZ)*1z7Z;0xkMS*r
z@f%5`dJbz~NC6fI@<>RS0poyb>gpgm9RmVM9&9WuV9*~}4*(Iq47JSy+~@}4%3cPA
zVw81-E`R0b4hFvX0L@w=>ornV?$z2K$KZ7@%E9RwTbQq^8onDy3=T%G(Xg-(OzdqB
zyPbBYUUrnqDIi?#^9`^&!vVoYIXDpL(tapVcD<sE>Ai4mtf}Gg_B<616k&~%p}>S7
zua-B)J`(ikzOA2G0^vZPg0Hl!*`PXsFvnl~cqtXX@tS&;&_0kl&K2(5_005JRZIMm
zL%P8f*IPR}IH++sE4&X}yjO1n;~ILE!bA>JJZO>o_HwJ}4xnz}Zv$Fx;`1L^$V1Pc
zGs>N5@;vZ)InU^75PTX=(6t`%QU2~KLj#&@bk8lAf<+@%xY23f20@Ek-!1Z<-=BD?
zLX?Jx?xTe3%9DqO){x^tJHFpKDFo^m2uynEeI73M!VQdEkdcusz;|vSZ%1xT7*mYd
zB!0$uM{(}Yv?;mX=)OHmk=Z{b&Lbv$WmA&cc*T*)c?5{fQ^V(r9{3red7>{;>Q}qI
z!w%^09oPQ!SPpFN%GmP7eqv$Qm8LbJD4c92#EOW1$ol5|$Lram@3&~|9z{h(s1P8@
zlmeQXgdkwKovXHPdzoO__->_^cySdIc?hjFvf=c@E0;a5K%&!s9T7-L&}TW3`T6hH
zn|KICifZz`Q*JNMMksTj#V@8r#Qx!f@bibPmvdbOQS`OLieu#udlIT!`N)t7cYGQ=
zX4f@PSpK{`29y54q;QiFxST^jrHbSpf`A(jkPU&pZ~f47{h!guRpl!h;&QFa?o#uy
zHWj!|dH7G`mDKJoea)sX31ghLkdQa)tkWSDvanBcRKCdx^RrTu`fYNrhPH>WC&eZT
z2p3;lL%k2T&c#DTMdeLOj)?&g2p@yQ=0S2^CW02aFy>sQ_tn<!n-^i3634@+g!+LS
z<qeh{ZlxxFxh?tlY;=jd&010=9A_GiMyp#B)_BxQd|W)hN_z#*Jnw#GiQ@675uJe&
z36gZ=PL{wFOZn=WeobtL0{{23L7h4V)BW|w!_xOYDP4M{MMajtbVNktH>mADEP#VD
z^jTMzt-1I4!&LRyl$jNl1$EVz0(P-NJbkTLH`IbwT~xz~U(a}!cH3gO(H(l@_o6vq
zw--rYBu{KgV>vtGjP+RRy~Hq)3HepZOe3X8pnnq1YQ^g?J}VvZ4Bul@9t5W7RsnWo
zLT~Yb+as*k4;zTqpI(EJAB3Of<?O7i)&N-+)3vo_hg^Z*AwaqBX~?GQLum+1hLYYt
z{f#jP%Ysw8vI*x`eS<q?Xwp88{9TfY?f;5yHvT5mJ;RuLRUfNr$6B!+pzDAKU+l$|
zA1=XU(-gG{Dx0k41-<1JHd@!!S%Q{sU>4tSCRkM9#Lvy0euyS7J2gvCRPI-^C%%Pk
zTqWn{-Cl(_4F^Me;z7I1qudygNm}a0G|Dv#^x^%FG3Qs?Crbj44yh=|2Z;$4(vV0D
zksb7R#Ul0}dh*+YhfPjQTn9!ffz>q*K8z1v?Notrd0q$*-%C{s?!CRepw&9s-_Op@
z4wa;W_NPsYHokO$ou+4@<GP(N0Sm=B9$ORnE(o(vtHl?_5m#6Y;n3yoU+d+W<C;y{
zMKOn)lsibREF{X{_u}Mboul<Oqo`-6pw9hy^CAjYW!&Khl?+OQOUg=Lp)EY}&Y@yL
z2Fu>R*OVB+LzwI7p%rm?i@mX#v8CBkX!lA9LUOH(LQZ6jA|-m~!aC-=!iDZ60UiyB
zwchTyw*0Q^jpY*_4P?TW_?n~(jsl3qJn+e!2k-pEe5~wNWLMF@(fSgW#L+=^3RT?g
z@7wVvoK2Z;DZ;zWWe9_#^B=%r$%>ua5p@|j8<AZhG%>TZ<|anZ9=EZ&*yySdM?^*3
z{N^{mvSuuiIrv9DVV$SeO*LGZmG-xzc7wq0<;b6*@CwJ&knu2zuF&669wDLm-D&3N
z?mvO+H3|1u3M#6V32j(uUe279fZ)6xU%MVend_9qe&Pqk0OZhAe}$Kt|ErVhDCJx#
zoed=>%I0hme`q=NSv9u*U<)Hdi!%`pbrO4YO<G;aB#Z0pL=NpEMx*<0iXvUSV-?~Q
zH3&I4-=vAE+KE2#mEbSTHZ}QN#WSsDeB8~HOA);&j(qX7rJ*<tKiqv5p8nq5C3I$5
zl!V_Y%TEF$L4S+jg(1>rZJ8blvc43uoRBXsx^yX?=f!{OlSXc4r+dlUym8o&3odP0
zxi3f4iuih2UfwRi3e5V-Ib7rA55a{k=PjD9fVo3GiD(A@r-gWxxyyl+i3h2u#=mkW
zujx4N#I@$a-wEz9?yNe<XEx8@9En>f(fc8*tH{WQ%vS`%=~}tjYqs;W=DUK}$b>p+
zY<E{(KYgKO@icSL!GTJmu;GQ_tzKzH2m(L9A)#(=ZGjnJfDcp|4W)k5egc{l$}q{o
z(|D|hhg4t#8Ymar*@o)&fKEj_?is6|D3R2&)@#LGT*Nox?y6_dt#XIxSaJmGW97kw
z+#wpc?5Al=%lAK6RkB}Pi;Eg6-YS}2Rg9XIh#81YBoBGetk*{<B0?>VyM^eSP_y2j
zy(CaaeAsh8z90)M$zf{IBS71G7fORy;yxevE@XdWT8EReZS}B$4wCFD+;xHO<P@;5
z*!VuN;LGk?SL%dGreAU{UDV-nFnYS+Zr{KYhuVp_GPV}@Qu7NbRG8gQjUz$d&|)!e
zizM|vU{O!X_=&bHT?PB>2xsORU9z=IjXu!39L(2AzgUwHEXS7El9V!?^1BGCC=dw(
z+mMBT=h2FI$E#yW41{|BU%!4`#(*?9@a4=ebl**2wmlx6zbr><k@RVAyZQj1H6Sw9
z-B{}5L2qK43`L)0dACZ}CM*beG;w(%Ib!>jf9*J52+_iMcjL2Q@_%B2bU)*%qfFzJ
zsk-?1jAj=31b3P+N#%c;5JgNcDy@f#rp~0CCx2Y-)OD#i6R<F{Q&Q?rTwnYAUylH;
z?+dCVL!A%SXcVPjgywGlioF6JhKjb7-Za;g8TUqT^H&|uf1IEgK*TNb8gi|%mx^pk
z%e9&Pbf&b>A+6`%RzIDL=qTd{yY7$g{1u0Qn_TtF2-i#2a7>A1+#z@7yZG$a*vP`-
z&&tbDgfUm%wWz3fOg+;Ye6mgM(8lQq0cq!iWKdpnF(1s+4zW0CEUs961A!*0Z}Ov6
zTCDi<vazASb(H6gyP{k`ABw~F%@H%jF)K?j#zG$(1Me8VsaNo;vbqT2tZbBj#S$jQ
zD^iL_ndX;!eU;C`rIdycg-yDM4Aoz)t~$T$jEq$<*6sO3wA$)iZRQ*HJSuj^a?v)M
zNSbIG0rn7~et&Eyzh`D>QE%Yfy#3Wd;z{QC6I!e5kycu+!%b*$%;^g~jp_XB?LA3M
zJ%Xn$C)^*!Wx2dXtbj#QTU*=Y<Rpjbc=q(#V9MZU7H~;QhHsr7OC@=Ohn+7&*nscP
zrb}=-z4kg|8ijcsGirj}xlSM!vzIA%8}l6mQ#fmUR(t|%)^Qq(*{ag%Mi)fKvPui%
zn75HVtaM$oM}2Q4o=NKgUt;^$v}vV<s&o;2GwLB|`x;IB0}?yfbfigt-LkMP^l?Vn
zEBC*X%G!@757=)-_s3%%F0x&Y@1UY5y{963!Yjr!GLG1c%3LtHf6Pz1ut-48C1-dA
z``2c+@zo^h_?3hiqzZ<K#Ycu=K)Oi$N_^M)*4`6+C(!;y!r(v>N$%!0R+;R#FIP?T
z?7AXdQIRe}yF!WzN_Ru_dE<-;8JiJ`8Yi740UI?h`GvE<i>z@x88p}z)3XxRJCEqG
zh1zs#TdC62rjn2h8!Xw5johr&&(gxt+ZTWP*Y}(6c6kr?D$0XpUk8cX*%d!Juz`<i
z-dY4>l)9eSVvDHB9!!}!(3ueTnkI<p<Rn$p<azb`*C_r??NsP8D*XFSB0b)r_%lk4
zYF<_6_i9YvLW2phOQWLs)L`-FcUM*+Q5w~x*g<&G&wzw8@;t7bHHg-h)Mgo$B5CB0
z7Jq6cUff&8hKPE92s(=$^UtOcBpqWm1Kbltrqlt>3<65DF^rO_ZQ!~A0-Qjm(cPb4
zqJ@P5yGQq}Ey@gm%9h|UVZYNeGBYpcS1K5bmijA59$0e}{LRvfFF;Go9ZaLYl83r~
z_IR%-1;Z(u%H+e_Ik>h!h_jd{H#$7`5`17Ls)j~`mVYPGV|<2qkxlz8_Q^jL#}t(3
z<dx^XZYckY_hxIHEw}erSWKOKG;@{|Evc`fSf=#skdABkh)RJFVu&3Ixjyn;wsb9a
zgU|d9<~o_&E@nP#yWdW4qMYtYFelfCU-tjQzvM9R`&fAm2QeQ><@qtMTEBIAuj}-F
zxoKh;-L1|^p~~z55e`A+#g;%8w!#=Oy-5O{NdoqWHMP&>&CQJq5tl|Njg*y}i)I2Q
zWFU~(L<%Td9wI{dcGmWj*=dIu@!`w3L$uCslsH`0yt(S<8yW8>Y)m<L+rC$mRmU?&
z(>4mkk)>$CuQ|GSW=F;cyFD|H&@M*e0wYsV{#vQMQ%rbcravq_k2)rKNKbsY={Rw;
zd36#*(e?9<>NZOM8zje^kGcNy{r&xOv$LZiOj`BTfUeHWq~_td$jJifqFoee#7p1;
z+}hS2kGpG}M;SsIB$7<NG&C~w;f)({{Ux5m3olJrFO)e*yn>zU(Y`h9tGsu$33nZ3
z82(w=tdsAbx_r&6O=8x#I8_mQBj{r7I64WjJu}cUJ`q~8e9SAxho0PyFyBDl3Qi(s
zw|Xy{^>kt)h7?j#!H$`(KR-(gFN)YE;?-s_lHMoc{F5_#e|KH5c6&8mcE`8BCx<#v
z92jH`FFPPWH5eI%c<nN?jeG1+8_oO%S+llO4+LTRu{)TEk!zPE{>@E79aM{t%U9<e
z=^wJ!2;~||4h=t9(3F6hr+?e3Y8(?-PUNDT)@(U|;Xtn2wHU$V#QmM*XB9yOWcR1R
zz~e)^<wv|Bn)P<K;<X~*8V=g`j5{m0Wa0C>H9vZ><6Se$2Iu*Y4m9f6ejn$W1>IlQ
zq!w=2+aC;w9#{%BRcniih)2+?YC27yY4<O{6~2$dbf%KRGuI(TX@lB4gCn`j=zkw8
zM#~?jTgynu8PbneG0WUsia8@meEQf)o4X7_K_ck(RmpfOtbDVYx+i?kGqS(Qynx%k
z*HY^#1*9L68nOwgKNDyMHS{Y1kN%2y!Z(zC0H>p3HCTeETA{lOjHt^$kA$xGThBn>
z8Nl2)S!u2gWO-n0>!63e=;^*cc-mc*E|$h^(n2)HAKzIl&;I$KDk=*+qJgGaW^D3#
zxE{U|;+x~@!i-$S7acf}W3w>*E0rPwIlbaEVgRPPcea%A3)95OT&_k0<<?sT=up=e
z>-{_0vt-W6y*uf;cv(h`Mc@5w#9Dpk0@|N}wGv)03FqFmmpn7fchhDF)0C{0M&8$B
zedHnbA`AlmoVa0C*F~8BNO6(18Vfp}1?Styc}rW8Mq#x*A2HMc2OU&aRjw~Tav<X+
z7VIz`T(gWX4O9x^4u;Rx9zkDi$^}Yq1jbCR@nbq0t`7zh9F8s>twqdrO~pb7zZG%i
zi1UF!0d?(RS<QFGOZ#Ak>_#BTj{gSiydPnSS2v2!s71#bPje<~?#|vR3Gi{`VLsTG
zV<7R(Q{`~mhzo_4=3=1RJ!0>s#K9NXdh)b-<PuK|O{4QF-W4a<DG1->j4A}(VhNt!
znW=`VjfQH-!*$gq?<t~nd-|;H4R~BEMdoSpXx<;&{Be8?u9eY2+UzZP-ZVWqCN<$o
z4Mv_0)iDM`js=t@Uxt74?+o{-)rJ5d@PqD8gloV+LGWiVbfxWj;608G42vV=1iJ?Y
zK&+art1JKG#ruXiETE#hFsLMjO!lQUYaYRfaZ&#hb^aF9Y7|cPk28z(Tk$oX=r+%&
zq1kR)YV0xzUxx?4fh*|ZWZ&t5&)fiAJRvR$MEc^Lu^kfh9o2crNb}BaoU4lTnhOgW
z7sQ40!cOLf#t*mpnHijRY`hV#w)!gWz{CRd7z_Y;Nd1xtdOEy*-IMFYfo4`G*g3|_
z;2#&OsPQs_zA$}Oot0i{EIC|>Y8yvXXkt(|_E5+&|3k<;!lI6XxRt&D=v5CND;=Xl
zPivPZ5>i66ZVF8_-Lyly(+3xvWpA{QAeg`G^K=e9EhInJJdV}eyZZQP>vjQSvN}l`
zSp$+s36_9~*=Hqzy2_HXzfZc1qBO^bf8!%GkG>vwQgP$f8g<_38{W?PRzW!FV=&>c
zE#p}igW+Ca6R?f6Hz5A!6MR;|C9nQ8MR_>;M3w!q(<Z@Mj&35t&u^6VDnr(6O-yjj
z&*Yis7wd}!Y3g_;`Ju4sQ1#nq_0Zy62G*c?)6b887gD5z&IkBUQg;P{zPBy`EZapX
ztuprEApzk#mqm##mScmT2(BO=@*a}<=uicc$AG_O{CXgO>7G(zsE;67_SdM7Cu)t4
zQe(0`H3l5Sh*x6VG@ebcU>hgwNYYvMnEr?H*B3<z-n8_K?mxMix;lSYP(*hDEs!51
zm{lDfyP?2L5fd9t8AA^(-bD}Yy8i+V7$mpmc{qJMTbRRQ`a}_u|2!FH!!^#3_-kG@
zM6vBjX2{+31f|qVcg}-gX9(U<MuEV(JGp5AAn?mA8A37>Q4rTKd;~Rj44Ww*e+m)}
zu2fc}^a3^R(%ny9C}j6}?PjCLZips35y#GChybbi+M`Rsr+Ocys#9Qh)pE=4RY~K5
z5!a+etD`?M{3BC+Koh)YZwC+!1P}3yZP%9_;Tbi|q>3hu>IpSTP%}$<+^ihMrct+b
zxgO(>V}6&c>t0ybI0@Uy*?45u3d3AOff0GXluH|L)n!}R3O4~hRz6@GAzSfU|L^J<
zIQ(><vgM6Kx`VAP<dI{{oS~}9fCsU2aG>2IQ50RxVUm~|x@Xh%#d0cg&X(`mJS7Kd
zsGLzN%=dEzcDm!dAxvDdH?>XOvmW9vt$)+ZX`{h~oB7G~jk(Ea&^TVEto=0G-9<GY
z;+uC$Dk3>fBQFSTMWM;so962np_>bNR|EGrnZiHI4yaJ2vcWTv8w3hk$`r{23<#|A
zOP(iC2kI5}8P;W2drPK^uAKleRIjmJQMhwv=dd~9A)ryNcekti!2*ptxNOFe@IwDC
z5pKHJG^}Hy(&zvEF7yME;$J@=4(Rhhy041~ldwu0)-Z0lb_4m&S$Y&RB82*6dG|G2
zu&8jszxp%!CJoi=iw(J3*<E+KvAOTEej1&)+VP}1EvZP&$@DdCW8~&k4_Tk?V;xxn
zEc0o7LxUU8G>MD*?7lk>415X|SL_@V!8G?x8G#-_@(4xlQa28PMQ^G%4{KkbI4r(9
zZr>>R!fp)DHMg?G@Lvo9R2QV8JO`O<tvv$e0|d0HD-9N!n%GrVQFAsE6aD#($gkv<
zOFY?D@q^V;;KM5x9lqMSh@E8S`RhgB8;&5xnwLu81Gu{nlpT(P$*j9M@T4CR{TTVw
z-`GbvXOD3oZ(AQW;M!0lfo7TZ!|w*|tu=(^mX1;v%7dVp86)le1w|LROFL2Fyu<ho
zuRD<35&dzL<C*a(7BVy}o#2?H0L7ARr!Gg!M>bK;dC-N?hCd+|Q{ZV09aoeyG*xys
zR;0`lrDUtu__o6zqx7j4Mu0M=Bh!k~-m+D(mE{8_X>MZU*FwPg@#DYWc7K!0@a~*E
z(7m(aR$@@P-ZT5RHYhtTfc+rz`a(qZ?7vc-F?95h<qzR`Y%BpZ37i5!2)3LX5i#-E
z&J56ofH=_{jC)MkAx0W`>o(_>;nzm9=D@<)bk%}h)4m*hLT+r%YQ}z-5+57QcCq1*
z#eXzEHk!XH;b*=Ng=&7xX)ca*c_8L;+`8J68AvS3Pk}vZh^q`WPeO7?(_X|2Vv@7?
ztZygSU2Gjc|NT~E58rn5)F_%2^Z_4Km2?AKx5Fl{kz|*<%cR*`*EE56J`T^xZ1NOG
zU!xjZ`e!g8)ejSCZ$5!j!3Ranlh98_B~)g|kdX<SzelgG2V9s3T$frwhxNF}6qf*R
zeTNrzUP%~ayxf}UA7qLjx;hFQ+UjXA`P<SbuL)3Dl#eeW5@o*4(fv{KVC-4If>vS=
z0Wqf4sG>RF)}YcNn5<EHS!dH}feWxfgrD$xipFGh>ApPX<n{N*+D*&DTr|6FSt{u`
z9EtDfU0fXg&VT2J9-b5**ln-+OeB5{!E_*os^c|ns`6FOB(*M<$8%;@=Oh56ZaKL&
zC$Wco?P0j64d0yHT%CdWk<RvZkmnllV{I%WGqdq>GY_PFf{}aR-i=gG0-g&zEaA&K
z6Jk(h%1#&Q=!#I}PkCcji4?^8v)NdWHzD?PJO+M+LjZl`DmU5@HS@xMm)?$R_YSzs
zE^dBaHnS|4@0{r$nx9V1sSwz(m~Uf&3x6U(KQcDxn9w<S+LfNW?;bR{O`l;68;w@4
z?^myXg|dy!-t{syCc5sKnBc*_^o}QGniCkV?~zZDzWmnfBOtft;bB|V`ogDe^rhzo
z;<?|K+h_ylY}UnpMhT4BW+tm90@IVJ7lp+9IjV29uP?!TZZ(g^Mbj#HC$kr<s^jLH
zT+h3VUMPO%I@*Go!}W3g{?dOPTpNq2lIrgwqHG6njO5z)hYlJ<t2QueMtOYyLS`jP
zvY0188(+9?bKN<SLW10(I7_-sN2{@b`Wx&F<Oz)}YL3jZj7e(L%yj3}rU|f2PMMCG
zd$BpSeLK&eBK3TWpnKR)>lsoahxqW;OH8r5uLQK9J_01j2-HVO$u9WrY5`b6;^E@9
zcXXhjp!j%u^YFoZ#6TM`I=Zs9w$R`bvt9$}^UzStIYZHVl0CKht#!WoabfrZgJ4{*
zCT^y0;u>Yof?*MSK7#OKW;XK2>p_Ot!ioeFVpnltCu;+QY}`;qu9rG}Lr>mypFPf6
zi^gQ!B;1BPUZJZ@jQT8dzf%vh&nTrwWiC;v|2BsL!5!k_2Gp9;tD2-{8l+~*14-5^
z*ZqtOJ3H$*G=_K#K>?fYxwmRKY)jVGnTV}Sb3IQF?20Tcyl#t9_WZ!j*8YS3Tk|vp
zKL-q<p4^JHzhqBon6n@cgdk_VtLR<ngV#@$I^3-@a5#UYTSKu}e?`Vmw{n>1nukKg
z!+J`};c)6-pE4`>`IhEhhJjidGDlZvjP><9wlto=Y_G+to;J9TLZ5i1_C&-pJ_b)E
zINSs`v2xjp|JYhe@Q&(t<t$RnY$oZ8lU2@l!Xg8Vj|;I1KC{cw>W4-NOWcX_hrv|1
z0(=6+Sz&$XmBh;yhI)FVZlErV%463Fv;Bb~GDz@z5eYD=A%d6Ljg5^z@4w!pdDYZh
z+*2YT()k|c1oT3{Ib!ne9p@nKfy#F^eud0ptj&qmwWfJDl1?iq7MJ5R#+<l<=?(={
zTZzF1v!#9LvvqYP<USh<&|hTI)SE19sM$lDG_MkY4DxiHg?=V6*TatVHoK8lL~?uU
zI1&Z(eYm#y>YQ666Uhi3`j)O^%@zrGgvS$A2^_q6C=}~B-vrxaA9HY~I{3dl@N*!V
zB7uG?yegyd-CA7%5NXyk3s6^8EDihponK+ODVp+qFn2d%ud+gDo*JliISyWSw6|If
z4MHX~&P1(CzbVcPfs=xsW^*-Z6%0*|jHF$Y`r&TRH8eQR&IEAW;hb~F!<V*u4y&ox
zq;^j~ZN=4vfE(cYdI&|=I?Eq{{qGf~6CfM?X1^7q%y97Z_&8kwFl+1U+~(8eV5cj8
z{8)c;I^$e}2@aZJe5dz1bEM}w`O#cb!){C5w`1g;jhFB;0agQZ9PY`CVbKXrgV!|Y
zx#yu8NGE>>P~$p>05W_@8Rlnmpt5od5c1Ko_bohwDqX9%R<(XzK7ZiTaCW1mvvkwL
zqxhra^?V@|8m<(ntyW-F0G9j*YFqQ@mrp__xp{xk(Ed{UBE3<wKj@~<xd$96KAgEP
zD)+^~A~>3Nn`gT33o!7UCDTg_W-wm)e4U?}>&Amj64d>OU(>PH{2fWT8(_*s_Hf6|
zkhp&@5j&$OR~FXQMT>k19T7{~lO|US<zMNs5_+}#ndk;>OK4*&!g=_-*Q;bngnuD!
zC)_eQ*ZZAF0G7ejZUh}=<wWv^ojPp^8l3D%vc8(jQOd#fCHTOe(z+d~xlO*GA_Z`)
zmmO>YT(>dI+pec`duwtbPoM1DmR95%oAFrRQisjD&y}j?zF3oEj<x6EQDjnKm_45Y
zl~3e`Kc7)1LA@p>g3a&PF07^T+dM|8Io8x7Z+p1S#0XH7s>=8zML8{K$F-Nh&p>Wf
zOQe*}#_jb+m>XPZ!JwN~#3^tGZe;-Q!81QMTb-pv&|(@p*)H=lez02c>y}Jx#q7=4
zjA0;{=fymnh%>dW(HYCMoOiVURaEy(Zc<f4wQ5gK#11Lplh>c$OPZ2BvDjo??Bfj8
z02B432uTvY7zBCm3aEEY<}fm#j_I*Iu!3f3xAgypefgR4_EW1QlS=)`7<sP77IAX&
zS&KCN<zbtjN!?V1g$Ip{A3@sVMVjyD_IsHB_MBFH&p6#0>Pj+h`oOPLIIm0!ptig<
zxf{LZk=*EwWiM^@vc;IDL0X(iaBkkYoz`CFbevc3++L%d^la+FuCqT-iSLe{fL3kT
zA5Cs>tJh)T#LXM^Q+x?G|MS_w#EG-DQ<Ij!11S=CuTCgoruyz`OIzt`(@9bEx)#xq
z^77{bS^VqgD>dz};iMX6t&#JNwWgfucT~f*`eoF<sEzwAA<%vg4PRO-62{&B{O1tV
z`_shpv>Ag4#{i~$=ou)I9PLxXt1?6^FUbUKJCEx=y>QW=k=5Ca2fN^KbMvJ4@8$Wd
zybuOu?YB=ayeuB9jK23HI6C^}2>l$&)~#c-pphXW{;I!+-$cd(3ouqy*5H1f?e88K
zY~5&IwYqP0k``R7rs|Eu>O!;_ax9GWZ1k+XBa)(ki`LE|&k!4px3S*u1VO>`o0w?w
z>oaF*zQNy-#zTY1haVs7-yR>J%4$FG$iWhBOLY(c#h_)Tf2q^WaM(tgiZw1KpmjP+
zN2M>W{%BDZq6m{@SpN0t$JvGYV1F!T>Dfw@|8yp}NmqDpq!)YGNgHCD5(2!~?$Fv_
zjLrrw9Z@nT1`9I8Oj0{sjb3-p{JvPaso`J+Nj!gBE{@h185l;@FV}(6u)#q@oD$vF
zYneMvhI*c-Ykg#B=y^<UfB#iU8$KcbM6dUZWvQKQ>~8xs`&i{3&HpFI5dTd?|0V1E
z;;d?+pZ)>#IPSW?ZVSBr{P3)obA9)!wP1>+g$0nZqgW@K^P%tXMJ;~T4YS{ta0@~s
zy#HyK4<V11&p+P>r+LXUq<!eJ^+JV1c}UQ`T3A>`>fVZHRaSC>k%3ns?S_nZ8+kk$
zw_igLp~dl>1y=#~y)2>e%d_VUjEq5!fT7_~yVpX6>BiI0m%uWybJ+qJH4nUMIe%OK
z@J*L6*Jvr*bJLbrMuJe-HwQwv`Zl7G$pCGxu<KT}Mg?u(IY*@Jh+0ySbQaAIpm<QE
zkOdf10%EECWCyhfKOL9LUTOte)NA)A-l}vqfg<L0|H&c~qMsWjUXx*6;x`wmRrou;
z$r=H_hMqi3%d$GQog4Z4^7k}np}Z$;RIoAw4SmGx!tvhJ8v-Bab2<9M0yn@yds+>)
zBV#d=VowF0-Jy7#smQ3YZv5IQGDjR5Jcx_afA1P*dot>Qc63}*bBwi4u=!`wy09^n
z5y3<0%Q>5_9b@VFjr~@Qh#=HfG|@=mM6mPD^r{N>K!utZtgMFO=FCT$?DWkQc!ON8
zc|v{>Y2JlO7ITd6!c}StxxDP?HSxq|abFC(*S8k#R$(Qnm7a`m9Y+n$cW<{MX;nz`
zkv&w>@(r(yl~0EQP&f_y|LxIj#)CHDN8{BIS84_BH&GNhk1S~bf<cM0l5Xbxh$o^4
z6Y$<BM+IW@Yz|f{8#glHSTIe@^|2*NrYk1?Sk<Gu`f!KWX`5Lw1_VN<e(ys+An$x5
zXTN!}7(&7`Z2H7=T~p1fwTi}KqUzBYaW8SrYS1ik4F{|}2ph{z4!>$jPA+tKEOXIe
zsnSDS(-z}8^eU%41E0paT9H3?AmrlT<`CTSL);zq(Zo`>&Qti*A4MBkO&Wl=WX9j!
zAA2}C4i{ycs+C>Tp>U!?GO8V@jZPA2oP4aviQD-G;@-f8(R=|``}XB1>ur2OsY!xb
zVb&@(@gf+}!SDN{KJ0=l{E+r7+RWhW7|}D1N{9LezJs-s4@M5itRg=z&RW>6uV4Z!
zv8poT3~e%LB|0mpxaza`1MaB=wzsTrHR*+#G7G{eSL`JldE663s6mUu#xB}ZsP}Bh
z#Vq#9zxy^g`lgqGqn0-EyfAW{d5uS~5aDn3vZeG{RR+e{t7FQh^2;ekp|915P;m)6
z-|<|XJDF^&KsL?jU1eka-<_f(X*jQHkFUXP!T3a!34Ody2#(+_4my75X<N0W5gkm}
zk0RT7(HC!+nTGqWtBb|j;J*uXHjHrH9YMp#%ZtaXWo9y6!Ey*DU7=$89km+_G-1d(
zSFMrPUOlHRXrtUAIxvNNqre?Uk!~<E>V_ocIZkgQpViFF5dWWcwGdc^i<7<jctHvJ
zeEs!x`m{2_!D9bgtV)9p=lZ{tTgKU~FPc<c3qE3Q>9Wou3Hu<ldbsWzI}m&)vI9gH
zv+I6ZvEox`;B?#<b{rgQ5pRlla~W)GX1iM@Cr6W>T%D*f_h)!`fK8w1%89%EZHOj^
zW$rJ8>3{$VXXP2XY$N#J#%_dKk2nIqXP`6KSod;L%!jOOY@NZR(&hYj*GC?Rj(T6a
za3&>CEj~+SAOw{zN2v;1#FLNH>K!UZxE8!8+}p09>x{g5X5FMtp4N#FD*3r(IJ^f}
z(LQQ|(=a&FuRqy`w=F-2*81gs+vL2hY<fV+BhvRR0#=fc0W5xbS3{GzYIc5Xn7Wxz
zm9V_Dw@_BWI&!5H>i#_9o%C{QbjMq?vp1M4em_gX<nvHu*DwQ-g)jA1`kF&(Y0q!_
zL;o1DL)_rBZt@>|-j&7^qc!l)4zW4>UE^GuL)7}4HF*p$$D}z*5#Hl#;p+9qZ+S;#
zr=kuwR)Ik=5Z+(lM+rLq>qrk_N|iKhn70Zfb_RYbjQnzWfgrsiwc6zlM(Bf)<N{AJ
z)~4FZ@r5qQu^N2OYf>glaPI}W{Kl7uhHh<lOA^Txjw;AP*LOZwQGc3l`<o@i#58Jc
zH}Z-ar>iVez~FtZz^hl;A%<POy;~EwVKh@i=^_x!@>3x>X{eW&foV4F8zr)b%WD77
zw}DOH-6$^Y?%dojM}WK1$@vY;>WsS8Sms^SLg{O&__bIc;5A$@%S-PxGCyCNI6C`C
z`Xfy`^?F`g@Wa?-5B`lE<DTq0%7hsNXgo9UA-V5r#Erl<yJ)aq0{UmE)ez=(@R}|3
z<;}r1JawslR~^59C@BI<z3FJt7dh8;C$GaA(eL;o+%L5Bv&3Z6LP-l`5H<40JH5Hp
z1wN(CyJcTGV_ll&(rlT;vD69M7RYe7&QMJ@l5ZiVKfV^h5~`?8`^FrjTg@1-X)3CX
zW0aYUd3%)i-ob<y?NvJEAjz=W>i@I^u<%D`0(r8hjo+BzRWY=K^Tm<f&4^}Fh1C&X
zIx7#q!VY!a9n@>aQ}7xjejNHR7&CUHr}Mx<U#3F$gW+YY3`oeTpA@q{3(?sjl`(Eg
zFUKow=SH`sn;MGboEiKsxx}Qjm~~iPMaQScyVCaS!8e1UU73@kMCO${`?|4ixNmVH
zgNcjoyF^*d2|pne`-hgv0A-~96Pue%?Oozq>(v@WcV(|^)mQjZuZgMqyCW;V22K(1
zX$qBkDqjS-Fs_;_gvl7UraGw;7CtWf%HmY4r53y-eX8=Nis3lmOzE@Kb%g3E)<6gN
z+c69)Cam?D#nhOdy6(IDIzfA54Ab76t_if!OuYMY>(b~L&`m%~uMNtxPhi7?vI)V;
z%+@A64B^_J4EA1`;MCz-a0L!=q)ah0U=U_@Hs1sc(@dN-htg=$YkwT;L-Lqp;W8c3
z3eRPVTbU6sX^-p~M5od2%a*syH#o|hx(0RjD|1~E@!xD50n3s1Oxf_&_E**-#Z;3V
zbKKF&$Fd7dPWro(Yr0+)f?KvG-4mUP6$6DpyWY#kj5A^NgLyT!G`=6uZH2RWvj;u-
zR`U(t(^?N-m?i4nOy{T(`s_&=&s{{a1r2ApAdl^o4q_9j^Wo%16BXwB&`ZC<%70k#
z>DjzeyhT3cvh4e;r>(HG8kMk#1M1@`3}q%9_JFieky*sbSV!YdUNno}h%$b6d&<2h
zhrY@@<41ZY{Ho$fAP#)JYzlqUBFq*|uJVRjooa77=J|(L^xVVR3TmcKhx#oB<XUD_
zsa8-Z9ZMrJy&l(uk1Bsbf$~C$*=~zW@TT-P<X%g>HLieXQpoVk-)&VN3=0#SDoMe*
zVX8>MSg2^Zd(*LWhU5Qnbyh)jbWOM(0>Oeqa1S96+}(m}2n2U`4Q_)=(BQ5CLXhC@
z?gV!U?(TZ#`>W2mIJ>ynY>L`7v!++~ySkq@<J+SgmxRUGdxu7K$J)&f&$Qc_*pGaI
z;olHlb`J`BWqbT}Xc?AQ4+UCafVvu_%$am=Z`G+ey;>Wwv|>y+cUJeglzA=Yv;_{j
zfZySlpHiT#i;}({awbQ)p#h){li)5;08ZdQ4<I8Z)>c%shb#*#D_Mg+!2q7uhcLbn
z;LJf$B80!=SjB0X;=G&<tY#x__#y=dA%;rCvAwiQ<bC0;O8vj>ZE-L>4xev(BQ?Bm
z@m1^XcnNYE+sK`sZRV2wH_e{5DrhB@-~JRz-;ixPd0Y$#Q^f5^xWPBfgNpq>A5&*8
z_O@7G*B)pWl%|}eJEXq<12sf?{@F}si&RxWvadA^a+~{2_H^ZC3M;v5Xj5rvnVnUY
zFH9BLrtHGdr0GX1#FLEkvge4bp>`{P&-Cffp}aYrC<B_QBjIf&cEi%NLE~U2HXN4q
zK@dD;X8Nur@p`*HLF^FE1W5PZUm@k(+=UPAw(4IZfN?zK%=Mv8`q3VEAupdoT)RkX
zH1wSQhA=SyN)WSUvC_QCOP*NHoXVHl@6j@)+>dF<3X5!P{M_;}W8lj=#?ff{aSv<0
zawVZypJTlvN;K12zN)jm*msM}lA=s{BUncdo1Hx`(t_^j9ecBf6i}UCyT3)8Mksy&
z24JG=I%DjurPq8dnW{bCfk`W)`Cd=Y>e;aD-FV!G-qz*?s(3?q9Ho-zuq@Q&7#Yj8
zAts!&Mc%h~;ddQq5`=M&K#V4x1(qQi{h+@v8yV5NoUbX0AEk=+p_Y&9v_tRWB`NRv
zFUl*_0;wY$9J%y}?BF-%oT7;hR}hvNRVu`8b9z9Y7T#$w+gu2}a8)-W3M=g$;(0ga
zkG1^yju)Qv1!aV9)ZPYx2a0G&B2Kqvvv^$sfnSPOO*SD{3@)36pZBqdYtTGm>6i!w
zB9OM25ZaT2Zq-}O`$GxJ5>%-lRyEpLJ<8!d?~Gr}+Jv-b!H@kkR{M5`GL$f3VoD5r
zg)wCOsvplX&Io{ZLwv6~M$?iclJQpk<{SUBm6OBP!K0W*9~L^!;f(9kh?~~HbbZ&f
zli9v<d`+x=F8jb7(psTb_Q~WGjx+9o)xF8KIL4(?bL`OHsdu_)jkfkEz2SfHWc^2M
zzak#ERc`PsfwZLbl9Rfo^3kT>f|Me}{BH^ti$3+yW|q8$F@0&6Q142}7~C2;HRMSp
zoU1fPqpp_l(|((Xjhxm4w6p$1%vblR5_?kgslo@+M^H*qyr05di+^{;meQ$iN>^L2
z2<mks0G`+j(c$Tl(~jk1=*-?1ayog!nTs^KBn5Z$;Cr;<Y11ihn_VZuyD>f+pdFJo
z6SYnobK&aUbzQzP?N6|B;jK*bZ=C|ncI<~8BX2p~$V%$%yF$;3M1(ZZp|mKJVZa8<
zjO+U+63EN)%k02gBjD60LMQMdKm}=%UxP?tM(VAbjEdw9&?+5zr>V4>u5WGElM6{z
z7{}e<oKKv0i*`RIE#Yfc$;lNua^j)w`^CRP5jAFnf|pv^XYlRugXkc?J<D^jUq#VF
zgo+az5yeo(vL8=sE<h?0^Wi^H{yX(b&Oy)(k$t^TNINU*t?e>9HA#ZKc+9w6wyc?2
zspfxXe=R7lMW!718V#8}ha{2Kwvw^B;z3T+UfH~-DOiZ|g+D>*MGW}PasFXSVx<e^
zcsVc){*~+633V1F?F7OIzmvqzQvH1<CReo7Y}`Bpwd6xh*PcVvzoEZb57T*gy}#g8
zDo=QiClVw>Z4*&R4b#u!r6v=8Jjy_F{zlvjNG8i}Xj)8__GVZJhRar5-wAI7TQv+R
zTk6GgW$t>hjUr$10Q96r5KCLzBBa_%yiuyGWK#-8KmD6f!RpI^<;(kgl|G@XgcZ0T
ztiJgA1QzGawxYAHw^O_9j-jxyKx&X?Ed4^-U)5kiCp?KOO77H&M0lZiUW4zFvlO+p
z${K7%zjrhAEydjJ8cD_-Z*}>=aj9l<99G)Le*&*(<n|}@FZOil`LG$dw50$me*%fA
z5J8e&H`n)u=2+&f<eCODkNIJ$gyT)UM9}F@^VyvpB&h+uC2?d1ozrfxki5iSp_?Xj
z-Q+y^DxE=^iagh~5yn|Ju&={)JJO+M|7aePTi~BE{br$9kJ%Cd3yQG+sw`EKigrv3
zlc4AW8*>Zun$e^`hmdM0Kzssb6bmS;Yb47MT6vg#b90+9k^UPu63wq{W+h61krmy(
zW?!(cY|uEbiM0D~8{&TN96`^0GF3dXV_J9ERwsdONv4PCR2UA_?~WVIa^y2L7JO`L
zp+4{fm)%vqA%uqoTz_+i{9d;^F_0U-;|cXtwRehOr6eGQm^g&8mb<eDANW3pwO-2Z
z5;HrdJn$2WU>>}Kvz$hs5v$p9`I}i*Ch8BEzQqWLZp2JKiE4u9;jcy-Pb%tI3lwe)
z-^U|=pT{ByhIA!Ul#c+E3OOQyQLCbV8ls-iR#e$PV2^-$64SZCf~M|Q6*)~>=C<O>
zAZ+OSO10XfW}gpufE3llhd*VyZS=<@Y|i_30Hj-dnBNpG9Pm?kRG>|pmCldJD3v$A
zA2kUe@0a5I3q26<VFXnN%0Av{?PK$fzGhtxQ1~HS<g=V>=6fF*#xo5|qQ~W2`U<eG
zU}0hX3!Yn<2<O)?rkbc#HqAOXS0A!ez6R8Y7+4q@n3^J+2>kIxy59UqbhYGrjdchG
z2+UEGD=N!<!aGNZ)SWxWT=S>em4)rsrj%8KPQ#M_5dk~0bkx5F`P?#~RpS-l=OUrn
z`;`y3RD)*B>KJ(~Ea5u%t5UX&kmJR3Ww*Ehc;)Sy_FPJ56}P|Z3u`ZqT&oD>xF#H-
zO2wXXd{Zr!jr)aGlQtu5$63vFHP@Gi&$X9<N-${cxPuuZD7niieN_!uL5^xNR-eN3
z{fXH6NBf5c6<r%e_-u$U?jIzojV$gki&p>%s-9HW|4)2ESzZA^NDvJzs>mriGnX9+
zBA3#;Wb%Ev7!wLZfc&(ofBg92Axl=O*Vd}{^y5QfzVxh@h4<==8&o;;#LI{ha(rx;
zo>m{(&XA%VEz_T^KFvQDR7v+SzOsR?Z}lz<p=-IZsaoYh6N5(`+iZ<wdo0J_SZ2{X
zN^^Yzr5SO}r+Ba%SE}55FP)3pYAkAzCLR`0i(t5|((aqbk<q47V>LGx{YcxMm^n!^
z+SG=cne2yNo)0S=)AFTCQP&<M^3sI@zds4Sw_LE0urvRB52X)`meN=pYV$8o&~9q3
zvie0@EofxqRYO&ZUjWikSBguMl4cs*c_k;U!v}vEYVh^k#D_+fLibAc#Cq(Kh^z7!
zZ7af8(?n}~AR)PvN&izi3Ifr;>;<h)SMd0^%us3${g5g{rLxLAbQ+!Y8SeCHL@4@|
z+t27vCm)Rdn6F<iH0aRy+>BWP80)r>RE++9`^AefK>hpY{21>HdT70B+sFkhTgRST
zeJ%iT4qV%L@v2|}jJ<Ue<PnSo-`{kMlmZjb*)K*L;&P8$RapYkd1|+s>}$4W4Na?v
zi*D})2y%stf<Ooe*T*?1<kgp`eEdQK6Ef%mxmqs}!2EZ4J{4vAzf0&(5cdDhNee^}
zZC+gjPP#Wgxh{sjefY{c&^7V5BXHZ;6kIbC{ru;Ms@5I-&2*#3-jgSC4qAkctPbnG
z*j`OW%2lq*k54(>-_p7p<k!K<#nnv1MFArrX|gSC)aqpk_dqTHI$VKzKjM%WawH{a
z@T{=;zHOM=ni;xW^}7GN=KVIlN_X;`XXjw5W=^oG#G-G$2^xMft99_Xh0~LS1s4Li
zc}nX=PGDVXISnp&i98wqF7#t!EnoWVWuiQVXHOS@nfa6PTbbrehqjb2)A4wvBTAgk
zD#1gG$X4&y)-P291r#X+Ln8h#xMVCd!q_nJlWqa1ZosRP@`dZ(-e>L;L0^NuzUSm3
z(B8AO*5Ht65QTE3&rz()J_q$b0kw(-1AjSt7<LEUcCr&QqGYJ$?Kv{nEKvaYSMEMl
z+@%F<Ewyn>y{+=r&(0~~LAA=L@YLck0HuP_e2|d*|3on}YhTkE@&7Fi47huucF2Pd
z6P=#E5sSpx=@P$;ydPkQJb&2IE+AHZcISD>n;z%&=q*gbL@xm08ZgjDS4#l^_TyGk
zWEF!vyO}pjzS2pab1mlpjn!|9f3n8N9(6XS$Z{I<!_ebl#o)PsFaQO6O(-RzM`sa=
zo!0G(gOVhRBLJ|@d39p&nZrj&H1(rMT3UnO`+z}AT4?!nUskr2Vmkaym98-vpZnDC
zJ2^d%c0^YT)O3E5p~r?im_OZnMdi8I*&M0=p<}<yX_m**FE%0avap(!6@fctAVCJ0
zbANo*K!Q^HYwGG0L25M_IKDV7F-H@ZJax$x5rILb(pGo6>2A+vn91SZ!N<RKX;1v=
z?DtOKfBWKpTVu+7UCQ5c!Rbof4IjclI~anRMDX)9mjV8(_J*b{-hyzu?dcUpXv#oR
zc7~hDGq)mP6$!Y;9X|G>gPLyS(^e#jHce3hNTK*@DSeZ-Gcgp_>1MQ4a`oR@nGbKr
z3>8=60cboCm6Jk}vGdoUwt`~Q0F~wv6v+6j!`5o(k~)?y>#oUie8n-By-_9^8x{nh
zopPUO4N2r$0#N)*PyMky+L`cs8$a=nSAcbL_3=0#n&QH5x{Zw1$F#Ao_*5Z(mmlkv
zMF6lwkFv|o3V7MK`aF&Ali8LxWl>C39F$Jb5`BN*gTy(y)#ubnMO5IcXm6j^>Jcdf
zL4_CoU;Kb@lGeig{XNV-Z9q>`)-X7k_pR<@HC<fDjh_ki>a{A46tT{3Yl1fVZ_IW>
z6ZRvC4bs+f@`s=j?ddYZe?e8T_-i5Fw_f+g`P5hKD|9?!(mV;B3$?3^OR#|Fn{DMA
z*?M$83O7vK?P<ZAu6drUN9BdTXuz~}2f{<yY^^2re%==dOEVAoU!ZiIIBUs~?in6h
z$s_n?3^4FSq9-7C#ovx<F?^zh2iUwx7rce7Cm8MGaO<Q1VSxeuitnvIRXz>#yRME3
z;D-*_)GDqpdYDVr5qG&3)?mIHKj2dcPekkaQ&hUo(k$%n?V2I~b5aX?rKY31YXSps
zB<y)hJhF(4l&{0{4t%n%{!l%e*Ik`%w)&tuM~gd^MEklFIQ^VcePRa~2f96F<<^s=
zbhj;BuY@ZfH)<5_gGKs3m*OdUztLB@*CU*_e|xxCcQJlW{K^F>q|gPHz^z6gB^akh
z;BDgtm&VQrSYKaHrd$SuxsSt)j?PejMNSzxl4JZ;+>|v1dQVUJCS!0-W9Pj}|4fB9
z#In_^$M`4f?Yf2CD4sMOVS{8G^1%WeifF3TM?36LtZ%>j9(V~SugY6fKT)^U>Mv~z
z%)(Lb^=2`RlAVwonu=2XIJeRqU7?K|U;*a1g|xl#?)8Zd@&4`HMzob}71gZ)vyQ)&
z!}r@fS*R!aK#?cK@_=zCrkN{829DQKzzf&-{&Lk7B!JZpMM6@Q9#JD|+yZ@!9wX+|
znK4k#{C9X{J*Ux@0!^a(*i{Y*9tVph@-w;V+7><<0KGZe(hs-iY*p-5qdALa<Saws
zaOL2_k*TmOqi^+{rHYxfFb>n6XH@#B>Ke$)3J=mIG{OWqHHaxLGp4OsS`X!_*#DGH
z=$t`&4ZzJ!^fft^iX|VJrK(OADXyoD&Gg+T->MU(5`5v#P}h`~G;KiyUL8g7Ga%IQ
zjzjWjUF)N0PCG!}J!`Y5?31Olae0918zEdb*f1uxZ~$dEKfjvKO9uQG52VPhpi~3=
z7eWT_*5bP2ZsuzyoXHNI@j!}6n@~#cQH(v@IyfoA!HfTw=Fc!elW8yhRqrpL&}7`_
zLu`2Qd~-DV%Ga7Wsd7t8;h)i61<PtMYfs*@@4Sh9R`3gxauf(Q&PQX<ULJ1Bh?O%H
z7<fu+6Emf`ZuPf*rWhQ1G$@jx?uUc+ce?o!mt~ZRw6%xw{1whcy%SGY8>P_p*KG}f
zAgu0pz7d0RhIP&#ZB-HNRQW?MqyYyXYNn{V2{+`Sm~|t0CJU(3I+{bLMJ63J&a&%g
zjZ;4idH;%()%Ld*qMhO713QnLP`$o5??1J?8F0`Iq@U{8tqP<DZud36fxlok+P)Ec
z2u|ibFwZ<U5skbrjg6fNq5kxtfn_!QJUdvg9@|Jf^zMwf>{FQD$89~2dR)K4{C^=8
zE&+{;k0({LLXUeu#l4YI@(a}HXUMvnHI{SpzpE~7(gJtdGm>FLQb|GdUlW{vgdmi+
zaM`W!w;$vHr0#HRUEoq4Y{pn&QsG;!cuxD<3?;d+G-i}#Yx#Fx#JHa?S<EaojvrPk
z!@j^MyimQJE_Xlr6!tn?aCFgN3kRfBe)?SmvvZOtyOo?V`gr|R<0#Wk1s3K>2kqrY
zTHAg=)3+H6y>uYA72Q4i<jwBJ)Oj4P!5ak~5Q3}yppikK2bf2jasKJ&R+JWlW){|=
z{^bQ$Qe4AQ1TUr*O&<&-^w{FUhxaBoXqjiBqPZGkz<AvZ+d#2;c5oA5c9EJH!{(H$
zB;6SfWm=oPFyUu0H5EguHKRvif1g;wgu*UMkJpb=Kcl(pfvGRKS~GPehIV7C^Py1L
zK=^$_GC}t}nCztI^?b-ZMB}=h;-!nKHnWfuL+7AHJ-rf$dtigLXUYGIi{D4U>-k{0
z&w&`hSL2HjIA1&d*B3q-3g;4Nu*e$0w8cE@yk*_<&3#XJ16$fEI~2qu|GV6zRLw#S
zKM>So34wcO9%FXMaOQsl{wheYn`bJsAJy-{9_eN8CF_wW@6)Nj5j#k(tuZf)N2&mK
z>r8+GNy#wvryRwP39o!pQV0?j9t04}vNudUbfMM;F-mvtdHF~bLjof2mWSIvA!FDi
ziNoLTeTF<fn(T5WkT?FnQ{}xe^sOuKyNs5YE+<N>`uN9q0j>^_KB~usr#o=u)gIO7
zswrerpqXTYm(O0Tq%?H#2NH0QN30gOh<c;m$X_QE%@8B-*79O7zIA4HN}f6n3GUT@
zXit>hzsvJt!4RX)fEiyxLbSpG1sY!BVMY!zga8FgMk2w=FVy}x$l@m%(BSovQ+$(}
zh2?DJGKo7D4bEIN!swwO=ps$&U9P?`7ka#BnWd#I?seB0o3g0u862_S8HsD`YyP=v
za-N~iLswjH<)bC86b$*oB)_x0OXmjnX}9CI-t45v!#6=#YWnpQs4YW3ewYt6+<&Tt
z15|@F_}NrvhX+sUw8q|^!;t&5%wnM0>JKjdw0qG)0ERvd&qh@dgUA2KL!Y5YbTX^s
z4s9KDmSloIl=^@5_rysRb?ZG@&-VO8iM&>*5PREV=jO2j&9C8~=THP#eXqOQJ52l-
zRi8=cFj4zGQ(fr9(qhvvr?^+dHsDl(wl0dU$SO%{Gx_ag`^pqEW4mK)A;H~06W?!O
zYY6Sx3*+T?C$<(HaDdOLuqgGN56i;9@}T)pkrFV6wIkT2CTm=|=O!(p9D+Ku_6N`3
z<4J4g=8XQCLzZ66hj*22cz!cl>uOeZQJghGhJq)V@P3X^2;1O9zrch|%7a5lu&$;=
ze2PLPna$itM=mjNp<XkW;Hz<HdjCI#GkxB)9Qqu@^Ev!nkpA*h&S?XL8FF@h+AGWo
z-iD@FP)N<Ju(6Pw>_QzXM8w-N#rhbjougN#OVin2i=X7hwh2srm$o{HS^c<E7WJ1E
z2`o-W0ODGPJ$PXYy1{x)@6vjmmI~+sY{*O$-*qUI{K56KYVmZVXtR+tjNC_8Dt|XQ
z6*`DV7dr$tEW_Ta=9H8d!2+hYn!8aeWH$UtL5>##v6YEs9x2Eue{7PbYGrNXtHFvn
zzNxT$pXZh$)DOo3{qFOW@j4&Pn^cGA`6uOf{$hyP;mAOG(?QY&@wgHsziXp>0kUFa
z9k+)G-~;?#xxo><wjPdv)hWFw)^fi5@@GGUp9j3oY|d>Gb=R5v8v%_UQ$g$Pf{lkk
z+C=l!E#ZGR7shu$cUvs1?^waC4r9DBBkbRyFr6fuC>9c-`vjeMx<?sRo$l?}@;1i?
z$C&mYE=lX6ac?#Qa)kps8%nuZtn^RmOjEpzMOI5VrJxCDViSXFu)ce%Mqw1)i~qq6
z<l;QbS1`To$>Ch#M9KN;kRKEGwO57a%#Klw#DgN#QD}SBJ@4AkhX`OsrJeMme>{w6
zpmc}s@SmKGra;R<-zqc;DY7lD@IML2hE4f=;uyxp6QV%0!5jprvt6qHqM<Rb90K{O
z1I+AVDS?>A@owH7^{8#ylpp<^cl5ylO_uj-m<n_WbffQgmIgoCSZ|I0`bvhk=sw=D
z@wg1nJ~KW#kt{RS_yPUqgHL95RehzdYgS38V~VzcFj4MeU%z7Np^Xo9VSc6j(&@`#
zhO6*bwVC!Y&q+GA%N$n>5ljs6l8Urnw1jyDDV7UK)|K<|@=KQ^rg0+$$;MUSc!@<X
zPF}$r(RJ`EN|HUB<P|gByJt6UlrDgcoW=tLJG^(6Jx!s3Z`WfBO)a+@`?IcTO&~Pp
z%;*)I!bvqz8!jyjxm@pq`8*BFA377e@wv4=PF?efg;9=uPr_lM&ow_kiv3xNU-KA@
zhr7GJAmQ}qRW8iJSsmE{vp<2Tz?n#=PuCPe!RPSNL|keg?=Su_i|&KB=0a%2E_Nw0
zHjY!pr4@dVbCH^sYvQluo6TwQGtA1-YX5wV96e&&$D-^=-+wgsvP$+PDg28HDpoWZ
z8Pv3hm<BnN{31~!kp>szvs2in2qg=)uYD2v>`i+j4woh{koqJPHQ#^@&p$tYkfGX&
zK1>P_*8YHUy(0Hz3IgRpw)e-8r_cmEd;{?-zj_&bJhDd9yUjWuZcEqt8)x)4F<z&q
z)-ba>i$g)-BcGoR9)fGje5d`-8&6!b(%B3ywW3!(Vr7#EkI6;g%17YpSTF#<S6=Tn
zRvhu-Y7cRYeF+CtR!xxN8qZ;hTGA)q^zPPnYYAP&0@CRW|D0XGNyo91&}=ixp*z+j
z!U(r_uu<mAYnp_QilV#6=8|RLZn?-v1d4T<5tVfRvn(x!S^Zg~Qj^y4=H|ios)O0h
z%Oif$!%RH}x4us9H88?UWYi8iH>c!U0Jj&1_=S*B2Y!1%9QbaOgYdV<{C0f=POb96
zcM_WdTfWy<9*w%6Gp+qriwE8h9P+$d5+<Sg_-5H4#{mUvORfI*5QYWi`ns#hU7Kh2
zi2l3TtV1_T%YKve$Ay3b5wrIu{s;H6e)WX-B`|Hy2dcdu(P3dJg5HfvMUPK6t1xS=
z562sh4R9jqg5Kag<hf;7aXZA+)_hdE$QUq)2)Dkk-*Fo_^{f5Qb*7MGj)+n9D`Deo
z<LBGJtx1XSTR}CyGCSqfoBJf*y!@RUoewf{s*pKfv)O3|J*Mb)v0ardhA_mM@<z@M
z=HztpRn<A4U%%7-yF5CF<v{yppsp5Kg~l9L@q{PeZ?Nie0q(>5uQ+(SKL0B0>A7a?
z)Qa}I#Ti7#N9CM*pP~$y1&YXEeb(*+Lk&H5u_^aTFt)7=v?dIp@$4V5Uy#w8>|}d`
znp-XmR|(N#y{|UGHzaZxDlr&Dco7IlEOamo)QvKtHE9+Zk-BP|`UcEw_xBuX*Ryd#
ziP)5AwV==Q+Z7Q~UipF^+3T-_qZ}zWs)V*ukT>$JqVTxHUod_@YKAb)2@OFyL~3rJ
z<FO~t<jZ+ktef4T1<x0=^~&-{*B%EA9ier3-oE{;{p7YbUCaZ^w?sBm-|Z~TeV5F~
z!>c@OWO<mI$5Z}oGZjmJ*JYDr5ma)@9lc!Ea0Xj?TR(o#AKx7n5UC2dFBNmG9ayro
z$G0{vPsaY&rx8yr`{wGA)T7GjOV!q&#WY|oWshQV<GsKK_luYu8F~0{XYIpTy#EU(
z(4!9*i58n>2;lgylLY~Q%$f@oGGb-&0sBRpIva@W2|*<w^EwzHwIIuFJ*)<_n#~mS
z_JAw^17`94gtY~_1N&!6l2HI$K7*=ci|@qtb`zjaOIP02sQu6K{ph?NuI)&I?A%ZT
zivCq5cEy#2oIO!$K_v>@Yn4D^mfVzB@<r7--BNZgE1S{W1374DiI$0?f3by#P}s(?
zdfx`s(n=&CGo|BQj!))M3N09Ru9N>UFOuGp-B`kz`i2$$an4>kd#+Ai;^gW~-EY<A
zV%+e<G&KV7Gw{2BLM-X=zm6wXrLuxn8Ye||#5sFAUNaA4F6@k-2B5&*^6Spc>{&d%
z+<f)Ugli!^6-4gSJ8e<`ac(vg=4$t6b|2U5qV#>-yTNGNu+8#j4;`?^^1-SpEupt;
zq=YxJMZsuQr>sq&>t(cLRG@O${$ChxT*l9##nL^-CRn_MMxQfTO$PT0oz<XN&No%#
z2?P!)0YlAG^N}Pb_n)Jxo3^UsYqazhrwz-UUsG_JAE9bKRPatd;<QO#_`}yarG6sc
zs=8D?S1&Cpg!xmeIBf}H0no}T`w@=LLJn`$uBM^_(+-Qy->!ZFFuNWvjihmr@i+#u
ztg)bY-YmO9LGk&x>14%yCj;`X4$#Jg?pP3Fz4b&CAI*)u2h%tK!^2qo6+_ilyM}SQ
z-j+VT%h1w5md}F~8Ef5G7OV9=d2rSxbC`Djcrr=IZ}y=|kT2sjqG3_;$^FBX<$+z#
z;KfL<MC7!D;z|iV{n^s<M>eI8>y>gY=TbF^ENe_I>ig5GKs`mzB1#m@?uI<xJ#|Lp
z-Mf;c<q!ZfB+^KXNn8RxEWjUqlv|3Bo4YxANPgcSy^1;c?a29#(MftaHqN7xG=Exv
zM})H<%<WkjNP8enu|+?0`f?xfTiWcUNUF+9`$Z3<CBf?~n!49(Boa3g6<fy5En%>>
zSlg~rIc>DaBKrigi1aN99kpr7#g%o1UD<zJ9Pt%eov>Ks8*8C=`%^`yLQ*Tg!|wBH
zrgOZKnbzzpA{kJPdtNHQ08cbvVo(+PleWW)l*noV<awr<c1{jGGHGNn@E>6O)Blj~
zfZ~`Z&Np;rMkwzRyX~?n$1`qhW{XUkaS?hD(m{9%k@|s{<m6pq0uD<Y05Zwpmb~?P
zhGu8)Hse0~-2My60-x{5)x^j#W=U00ZK&qQx*RNQE>39z@BtV9`@64CQJskkoS16#
zRYYJ7Y0pHQ;jtF0S+cVz$_IRS_p{2??+JHFD7P*FExF4mfHKwt(%4;5w$1VnYHGIh
zNFyYXPG$lorP0qSq0!UkAl>!Swg1rhA+c#yd6<Kb=g^d-Rdw`OXeV&3_KV<7kn#@U
zPc&DUpbqa1WrP9wc13jy!ajQNyOHF<Pq1M7A-?M8gJo$`#7r#Tq~E_fT?Rk-uCczY
zen(4gzE<hsxfhTii4vjJ*N^s<h_L#a><x$V2Gu{Gr$bR2-a?i+DU^V=VMJ%m-^v^B
z-zb~>IkqgDoa4f?u#<`jd~gt@N@u1m)HtzE-9o|BWB|zjRiJR5dlgS#{k>Ct1BeZR
z^&;y-8j|Cl89HP4|I$srGQa$(!ZFmHP1<q_kT#-#S-VTg*?8B~30Kf3OM2HZD)Cp6
zkx&QKEh}*}{Ot+->>^@g_V}Ud+O}98ERbt*m<jU|dhor&hXon{A24<#35$$;?F+GE
zXFD!0oq#!LPJ%p16uL8165*vF!8h=QHI&YO!5nlk5|MAh1nfB_fAAEyaIfZCxMHVq
z3O=oZ-pPGe4T7%>jafLpJZn`{{$5_nt)hK&i+B)_>Xb*`qa&;>A3t5~iUcr7$kh{w
z65ld7*9?E2-Gk2_fHJNX)|!P)8XJ~^kKG7#>;N;x%;)YSp(0P3O9!m|KClNc_r)#U
zSxk-0zL%;ZqiP}o0B&x2ncW=<dq?i#WjR>kqVIz!d#3wVC6_ZFbETx6lP!Q#JpDbt
z+a+}uvFP>{&*ySYBsb2GG}1s&Wu1FMN)u6k1B0@2-Q?Cr`cM&*6W<-42&u^`4145j
zcvx+mQ#pq0WuINdMK-!97~r?!@PkHG$@0yJnUR5^wIRMBUFuOYdn=9OVcAgxEgR3S
z$zlmmPMOfc?RS*DHoEFj`f2rK7|5;~M?b2DX^+!Lm;QImmVT10D`bh_dt-V^Nu^Gf
z>gHtu8xUy}z5Tl7)-x{Prx4G?^S1w~VBlh8l#q-wV6Xgpcv|i?=u6J>a5m5Z-qrlb
z-d&ZwLYId)=p~2-7q4|Rnx^-ZW;FXvTVMfOg*b0-36<gs0vjA7iZqh1k%RES{Ewu`
zr{m&L1*z{oG+ckDA=b5oC~lOAGx|9?R+SymJG*hdS7JI@QGaBkzaS=iN9NT?sKK^s
zocx;n0Wi~IO<%S^=A4H~LJ(NFrKP2@ae)U57q@rk0>Wb+x$q?|HVAE=oS*{?48NY3
zUiqC0`)2KkDXRHz(&!&r2IlT1`Fu1kafRpnoZX*l^rbF&fryw|N6gn;$z$n}RwIrT
zVL}6gWnBs#ZwSu1G>H3aeuDtrrF{Y}K;1WVf1Q0ys{#W8Awf?YiQSoBc3YgHPC`n|
zi+lsLh^Ky7GDI1adJqA_myA9>&|2v<`3GaD5HkK{!Sb9&F|Qis3pP(Fi4~^-5JS`(
zd+~5)IriPa-8xS7uY|Ybj6u2U-kjl2t%%J`UPle&hix(Ru|$_xpRLdqiA4enamByJ
z*=Z=7Kd8a;p0K5A;TvHfYxCghQb=KrPkW^b9(Gxu{ATt?9aTkAe*HmJ>G#{#czd5v
zG(MfOf}!cDD8lm~?Xu2Ec)ui+<>4-2&0qSeu}-c*t-pFI{|wrTe)`kaP)IZ^=hHW0
z;Uf@T0C!;o>&Ed>OWLY%$Xh$`boLFef3ECeiBT5_XHN?wsMy^~@#XxA;H|u<MVg6M
zJR%^oSR!{UwQUlab)#bTLAv8n9dr%Qpf-jDq{sP&oG+9c<>%O7I%e}VxD&D8FK87B
zY1lcIcBJUp002rr*pL1vpBA>c-&C};v=BgNCyNLrc$C0_pND7Ow(ex5Wsw$QmqS>q
z4zMFY8x7h;<9>Dr-iJLDdGQ%IEkTj`Gz^+|ouGELqY5_Y^exZ+g56oU6DlDO-c=;@
zd$jqqv(Pi?q5*dagf#;WIX#HF-D9rF!~h26)>U<)h3~m8%5v+!%F}dpA%I({)MRir
z%)R7noHR2hM3PLcH&7N>vkq9vLu~{1<POMpP9i^g7mAQlkYuxH{FQjx2wc`L)WDa6
z%);0cFQnkjRN2D=V8y$7mDUyWLJVnEldw26Whi=&$-3OXA>RarErl7-1LU-~_3h)|
zM7urR12TNB%nPqF^YoXKM(^6_N9E6~F%j}$IMp!H1{-@=i7lzxT7+ZRrRu>sf9~rS
z69-blA%pcd3&WRbl1FIBz}mCZXm8Nt`vGeJ!%=_Z(=+s@)d;saeO?WUYQ7o?*zMK+
zdak)$HT-ErBfJPe|Lc-9-R~k21NDI7pS{B>ZV3uw<c+sAPMng>e=T48j_3FbWAc6N
zgz;UM-KN8~7&9`;6ZQHP0<JQ(?jHtz8H*gKFcc(0@ckS`leMyNDw<PJJ+l9gmvnc+
zy@C>8o~NB{4tOMd2mFMq-17SR_1wd(eDx(?Hp%DtZU+K4^6>CL^MkwCpZQ^cxDTHp
zgtpheAiUmaGN`j~X=dhnM#~OH*wOLaM3Dfx+uzDHN)j%pea<Jr=M^}0Zl78IrGH<M
z{=E+wTo&D0W>zQJ*=OfDJ5KP!m}OSRGHJ_FQ>{${Kbq&kJc_4J=?<3iAtjw5@8_u4
zDW?%X=a%(pBS?b|^%cz{Z~)Fxm6nH$|D%3Tfw<Yjh8HwBARk}VQd}+|cF(QBBjS+#
zNp$?OsmV%DFL@0tnL`J(-YiB>yOmM0xJ}iLCXvw-hr)awKRtSl0EBYSWqso=g2@Q(
zzNDFnpqb(OK*CG(BuemusMpYyi)CTh>ATSLqgWYRlHfahJuToa@@KCM7s2kYV<P00
zA-gj22p7VxwIX^nqfJ<0UGCv%qZaR_75(fwRZL|{bpt~X1t7-p(vty%e<6B0P;za?
z%6gfyAM7Bnhp>LK$hn#R@y$pBAAT;M@HM`^8+ej<!}f6M!dCq7Yd}BdRQ+ZP?wPMm
z-tK)Hf6;v#Z6=A`2s{E{PQTgdtl~W^p;$dHg2cle9ho0|`X-H`Cu#UgFKiZJCxWk2
zQ1<j1O+!2gKc;?e)ETIY+jWDb=j;;YHr$0C=lIOacL`<bxGi~=U8?g2umEQi^e?^{
zbwj038O05(x`%eXVt0?c%f5OB?K^EmpvKZOU<M6n!0dmD!mmc~nJLkLQ2PkJ5UChC
zTL~fX+WaA@POs|V;2<X_2O;I<<WM&JAaD)XhwwjrLDhXX(PBf{$dm7Whi`{kGUVC?
z$Kkcb#O1={m~`F(KT0Zj8ELPC)FYfVFy~c<ntXfrOxp9!@DcDO-`nvin$s(5oim`d
zWVdFp26q=@8G?>pp`ihmaU1ptz1BB5i{c6qWYLkZfU>BM>GwJ!Y0;n2afk#UXbT<p
zvRjsXw*L6Eso&M=cs0{`om8jd?O5&?X;c9NZE(BYySrYeZ?I6hd3l$lI0?;xdXBKI
z>J<X4&D#lMPM|t%@9FCJ1Sz=V$%xxF;nY<3`<NX^!$Uz!yjtp>P7M_>9B#r>>KZHz
zbFB!E0836gr?R*gxT+(|49$OfPI6N$!!Fp#8?*7eeoq4H5RVQ;^x<JKBxO3E&q2>#
z(Qc0LCqs0+lSs+hXEOOCGK#7Q5P$@>eOD0%U>;H_wpK+jdH-u4jR-Kl!4uM(GQ8W4
zxKCrVq)uwpgU@DztxNI_j9~uvp-%kCInmH1c1o`w9`Mf%*Q9^pDFC&fzAdNpmg%*?
z(2w+1q|;ym#DCaWL=mh|fc6dwS-<^$a`+=Z1YcAl&i_2@CqE12{~9HlggDi!UKb-f
zS>TFc1#<x(umrw5h2t_1J~~1IMrvT60r<SWd}6os^i&q-B02j<$J}>S?>LnURJwmV
z1);&K@f4)&j4{AjEbd_W?{PLvd-{TRw7Ort$HLCExT$w%DNR}y?ypC~M;%bs5+l|e
zeoSz*slv1$fKw6S@esAx@Py4dps%;D@0KO#(*OO^xWQA>BEFKhw!4$Z&zVoe#p8rF
zo6yZH4a(TC#=X8A-d(5O5LbC=GHP8vl;dKSv~5Uz1%NO_rhRU1zK~(b!dD6WL#Kw+
z)MWv8L6Km|E5rp?TguC35y+*fHp8&-VHtl%B#VlN^eAL@|Cf@=4nfcn4nQNOE)m*<
z@TO~Rh25p<M<VJk@uj~b{Ez&Jc$c@^@cy|%Jh_jjiT#<ca<CNgClc|yA7&}hB?VX+
zKOuhblLiC&n8|51hr3dg_APQ=L7PZp$ucrN^WOSK=L5?cRP90q7mWT_L?O9_k%AhJ
z{abDz;~^;02s=+)E~5drWy7m@%PnKixJBBGVs(D@elM1$x14Nl!kuOPNd>+pg~qBX
z3i%(|Z<bf>bBa+hY^-IJ-7$^?C~jyW7fav9K!Ka3Z=70cEYmxgo{W5ao?XGHP<@4j
zgv7Ptwz1w5j?kuRV{^nd#vhlOdJiF%n@AwT&V_0VWr!u>>FH^1ZVp9t64xb6eEeN%
z(;CgTyX`cH4|i|{ne#!Ikq8JwU9K)D$xeY0I)s)r<o+-7b+KKB&!d-fmC}$w1sIzv
zTBVfBV||I*;(>I99nW#_kocuPq&q3JJ~btlW>t3r2Jm_1qFJwIda*H?=V<s?n{H+I
zSDnK&M7w_=G=_c@Q2`O_RXYPgN0Y3Ii$B_rdt4D-9+6)BQ?}SAZ;x{lO7C=gjaxdd
z9<rjD<L19##!Wfr5QPLitpklMR6%p??Cko1xfh()ciT>>l{Til3fmWZrPPD~j$!)F
zUfFO5g|>e}(Jp6Plk})qdDFP0c5u=;g9IU9%uoL-DZ&>KNwqit9(L<DSTcNZ9h8Q?
z`XShM``?G9=q9HUuMG|7fSJ~Yeo;<@$)l1H;lC|TpMLI*%H!@VV!qpfM+E0TvVB=H
zH27T^iM+QIjf7PNm28{=UGs8`OVN2T*BfRoyuDYdH}>cOr578CXH6^DFuO|#M-s09
zIAS@+y|nBYM;xWmL#n+p$0x~<#Syf9?W!PJIPUWNK5@i5;-&rUsA&QR;sq2mAUSY-
z2dc0%P~1N7>()_5LS+7Q5OPv9Nnf?CYr9$TL<<Q~Ov#gse!j?hp^}arF&#3lgy7GO
zjg1hUzV&>N4*JaHyR(y1GX&d)F!gW%WX~G=-wTUcM~_nEbs1)8@v8<gZXvw8MM;!~
zFh2yD#e0mkZ|W&5y3<b*5n_UkM)7@TB{X}ZFGJ^UG%V`i++t2Tz7Vxb%ZK7ZiL%Uv
z5%8*SBW1<hy#CS)6m|SF`_E+DZ0OyFN1Bl0@SaUrvzvx)GlJ}WQOfw|k%(iPPMD3p
z7UTAt@(KaR#so@>)NrGXPxkf31=l}>$mrV#fra7YxQBhOYo*Mv{MiNGwBjpxX}=(V
z@lCubA@B-Bh!^}j`*^81;+3z?tD322WIn;enW0=ryX2~m{mLE4u8ZxnjEpl}s@oYx
z+xL@qh9_z;uRat>O0?df4D|eXDb9c^NE9H8bc(WvERxpoWny7oikvO?PhuR3+&?bh
z*TEN;vRp_W4Rx_d6lBiu+z47FT8tD^4LnU~Ft7zfSUdSv8RIethbOdgIR%y6e)YPQ
z%!nxgfFWrnDJxaa8_ai6Kfd<bLk}kO$O-gu5RO1*m{PHu5E~lmxUcqkx{oB{8cpLG
zGp%ZYZep1O3}oN(3}L7sL#7t!QwXknUo|Ee9udJ0K0h`(9|WH{Lj^dRYuVk&FK|(h
z(D^}@%ywb1Zj<N$G@=TTtQZ<9>02R!ehsdk_9dupT5bl9=dg%)3qVwv@PFSVVsDTf
zA}bC)ApqU4kqv{Xi@l67nf8`O&6zLl+7GNRg+4F5I1ir5nwV)Sx9zNP1HLViJjXua
z<S25wW(Rt~e%a>!`-d-mtxM=&n<&>sf`3;hX$B4o;1bJd?!;smi~U{(Z5_59TF-CN
zn%;ZF;5?1zxD@z59a0q?<dh&%0AHQw1{X?!HBIBpuv{iJO}cp1LgUr$*XPOM#*!%t
zMthE(Tn8jgA1{G+69Weroy;D;y`_VF2#ZW(UHi3oiK%1xZH7&Ql0SGCpv}u9fdZ7m
z_l-J|0CG`pt3s-SDGWE6e?=09bUEx#6-mcYH$?ttLb?$xG}_;<-t2Z(@6ci^^&lo9
z(qY&3w7s)aIN5MR_mQ5SJ#7T4c;t8lfVGP|2$9Z6Lt}1d_viO-!}L=V6BEdq3DWu@
z4zuy&5#*U`HB$m=w0K-SKRrOEX9PGnP8wgu$;(7r=Te2<u~AU|E!Gq4u7M-!ThUHE
zE3_M}J`iwgH!Y-9mHSTMl8+IpFt{j}aAbO#@3mYbU+#;%7BGn1mo~sBQSrcN;wmz|
z_bKA)07p5LIU&gQ<e$E4C$Um!$UqD#ropskXwpAB`Kym$^;}kTQK^&U$vrv>>4D>@
zb+z7GanasDBvUG>Yc&E7(oTSM7}d)~D{oxl$;RwrQ{BOlSx9}@cAC#6@gT);FI-~C
z@Uh98#;6NU?wU^;CMk%;#{#|}iexf4a;S5{^Y+mbi`;lMo%$hP-KtXg^RrSLjeJ3j
zy(Pyrq(gvmicf>|V3921Q@hh~q&uxo-+>&Q_=~&NMhTUL62@n0{UW)|Gv0T~t+r0T
z$e$;!JA9Kyqj1jVk|VQi;7|c5Aq4_#^uy{o@z+SMAMY-pN5BrE%?u{cd6G`=Ice3|
zdS6Yeu|X51kwiNS7(V*DSNG_Gx5CtMrRo*A;LbS2%mhLs4si#9H{8U;-~e_`PS$t-
znVI<>9$L?u-^IklK!%><?B0-zr^v@JVeK()#Pbs(muyXw*$)Pe9eC&p>bw*(l)HnP
z5pQtY>fJ8G!zDLb8&@zwMT{!raQWQ7nB3Xx2|$teb{HqVZ+svEkq|&c_F+6J<OZ+(
zWP3KNMEOv26fxle28ltVX!}@yYFMblcAVM!Am!_w%eOym0_><b%PNgbUn5|Iy_v=l
z+ElQaHW&4T_DN%pP?H?CO*f{`Wj$U3g~R$R5r6@aN3QJ<gzs}y;$$V9UOvzS_HiPU
zf)l&>5q^6!&K@lG672YUepMyKLQt8PHuPlSJ)`=t%eB;JHAD}($K&1%4VaC0;y`vq
z43R?7%F68hlLm)T%W4d{2LFNpn+d>sBLT&)>k3B{-|SrylSXv(nzq_vlcOIl&{-Gy
z+LN>Dui`2%ec6jS^_hHYe{vQ+4xAng3SNRxgy)BoHYhwjjF_?kw|DN~DC##y2!RM~
zLyt#xpyyVC8hkl!7et;Z88?TD2%|dB`kSD@KtNbr{I&e#T>b~~rsdJi#*ko_+RWdO
zFP-nO+)aD~<TwGD?yyN>B6j=oXpNsfjK?<FraI=o714&F@*2&nid7wxv!2!*aiLV~
zj6?cTO49w-$S%d>2=w#VY2zwD7!3M#qx6C8oq?o)Y4s!mQU4P2%$)vMYx-pARD%mL
z@F{R^w_PDc4D|X;VnLz*nZd!R`GIZI4_?VE4~G1Upd9I~x)3pd;&N~NhDYR|t)(WS
zn_={bej#3W#j*4TAY4h@_E%&Y${XKu{PrZ`C)C#?nK%B~2VT*<J3)m_Yv@|HEmmeH
zwk4D@>da}69@9*~KK+;D5-r&mh$Ps`vA;t(vIqYytwRJhP@0G3d8x;uQXS0fQ?8^8
z+&WU-%>>3`$}kQ~odDqiyt4}wZQNSZsCBmti_iJ3Uk-Mec|K94(-64oeqWtq9;-bD
zpl7YA6878>?9m#vI$(Fn|6#A_6q;Nb<7oO`=3B17p+2hhUNipqxaK=RE<7o_#Xb-_
z3-$Kp=>#lREj4^5h1}0{T$iVnx1q|OfsSryZ0tYtjc5`+c;HX#{qIjXZjca<ki9;d
z!52*?Bna(1(9_f#e%u>RtLy{0)P>_QwL;4Pgb3?zkdgn(Da#E4B=QKL{|i~N;#X{M
zn*zHUNtpx&OSgz{hJYB{5Cdn3jG1;5&r5-YhGSS=TWhEl3UODbVuyRTN^gYm3j(#I
zwgp2O<t_dFILaur3DFWhyHB)w;dts2FOF=btV)&9?UxWVQ`XR(xyZm-X}upe27hv?
z#UeTK&$kUxmd+@p4I%=FSo&3l!9|DSsx+D8eZ}lk;3y^~S|aXTKDu-lILd?KLI{|O
zISU>$%IP2delL&aq4Q${(tv;@6#v!VVbt$X0VDsVgbyu<j29NBw0Ij;9Ww1X66d|#
zoI^cjFAe_Y`z`!_T&ZpYLbm7NzAOu$^n#>jeQ5Y<28q3EZQUL1P~ROB*$xlxc2QBs
zXL;0XM^d%)$ff3P&y2c~?<ZR1jJ`LQ)iF8h3t?5c&zEjrt%=5i?BJW{t$iS?lK-rj
zo^IM+oVU}P{0!6`|9hPWKoM_RO-}%V$Ep@zrI=-EG&&Wvwb`+N)(}A$@`ZVWj^1|C
z`rr}w@664v^=`ZAq}6i`f(js@arc=GC!g)+Me}-}L!DQEhiV#Lc+arxg)UvB#D=v;
z^eGl~eQGXEf{b&%vxqh)qCWr#CV9WeW9#X(>6leFKd&-?EoQc#rHgCWL_DZ;f#I8F
z%bF;$EhfCtM=&%qND5;XOS3ej?2D*KNZ6c48IyIzZ#g)G5E>pXCW&-Bx<-1HR7afT
zm&3nk?=TO>B4Zq!39r_6noY~q=~;+}*6~LZHOm70|Hy<gyp6g9SVKSm6Sg`kG2|$#
zc3Ft$3gp3Zr=I0o?UcGU%jeI_@HzVBZ5Z=TLB;@IdkWl;T%3c8glX@jDb}5whK()=
z0+M?SzZ5M6(A2n~2qOaU0U|x`5vV~cs^`y<?VH)FyryF)RM2dZP_(M%nmJ!#PXrs2
z$mXsG(#1P&A)QT5{M{Wi^=|=TG%OZ{<%S#&CEs5IUkpEL<vJs4F%Sb~3BEd83(OP@
z`jf1T^5EYL%Vas~gx|#+v7gyKNV)BB9C`>mDYd%0)C<I*mzqL0+IU6t`0&F=s>RdD
zj4G@*zof^|lL{>21~w<z7}hOKC5uP_%6xhj6MJ2LD(UM7cFF`99%399r;+tEn3sVM
zS=H6mK>MOy+j82~aqqJ0s!fHu>K|Pj))dH-4ZJz58uMr)mX4KMtrvRQ`)n()zr^iw
zs1Dix0sUL&?d|O|GpYd{iqK14JvlC}tDpI42kw>YNsJ;5958Em_4Mfr&fpaZ^KNNV
zqP<UQRb|`@x68k;7R%Qgc-!*C2`M=9C$tLIF6&8;odG+)i556G`7m*JlL(1C9wHpT
z3Im;{DYBRJHVZo#!0YVs`z8%~+NIs2-OgxSK7R1i?fq|z7X{ajhQZ#JLc0BhDd$ql
z@DWV<B$`)0xwi7<NwNDn1>mGfnTnRZ%F9d2M$9W---cDjzDrt;At1tS-<6xJPCIWM
zJ^ArMVsqc3js(1Dref`#kgDtsUZl~1zY|Oa5nk?FS#%Wyaok?saxM;dw$j7<h7^d6
zDxmHfVK`hd^VjVvqi&By6_-RQY;?b}Z=LpNoo+0gI32Dg7kbuzs<TE>@^{1NhP}+2
zgxy9>Au+w#Q4*Rujesj~)tA&<CPSRZZ(&{#q#OQs{hU8UB5PDJ;;ilO-(`~yb}`CR
zy|(snPiLU>J-toj3FUqFf13$z9?7*4_(`s01U2OZSr*Qn-|884DbUUr#_O=+3+sk%
z%$TPggt6hca7w&5{jjpvZ%X9`bR-m#W15);f{k?OB*XZa#+|pk0Ci%UH0+hD;QA+;
zTlNbQhuXR~8WOLH$2VTUao47cRbrcm{KOs3Hc9zxyuTcvjba{!C8S%)ecx>KF4%q7
zI(h~j?>cTG0ub;276;|!uPFM{2H8u!+u8a`i>Ej18sv?kuA$KY8A1O1`BShPjg<d(
zT#`(hDi*Q^Fg7z|3wXFW(Q9%!T77xGnbq}V_|MRn3M)*t7_3pJw=HCQh!?!_GpkVF
zP1~TOns3k;HOvWCnnjZfd=qRyk9#miXS#B=56>skTIaB;#<V)JRqiSCDf&(J?=N?U
z3k)`vE#H0?G~^XDY~19%zhO*1UCZS+wRah%>Gc)*s-8huY_HsPOpb=90xPWQXsjMX
z3`oU#eCrw|`2aH3f1j@TeV{qrCu_+T_wMCs)5aQ4UMlwfGD&pxNbrc*LS+vW(yV;=
zdMA4`Y!ct+uvFCi=i&y&XJ88Y!rmaChbMXn!cK4gO)=kVomJlGfY4%?)B@V?gu&?f
z+7?f@QV1gMw*;lC^D6?syu5Ys-#E)QE-_~@L7r8kOQ4+NCkPe1Siw2-8g)g#I}$A$
z2D{QCC`KXihdV`J{2PF@EF=ZfSkw1Pn62q+K(P%z(C;2Y`EExH3x<DTgLZf<aFP5#
zY0YP>$?x-d(etTefwS7Xoo2$5azg-v-q#+JgDpK2<BPym3WSgzJZwsvd#ky$)=ovt
z`&BCc0aTvEfr~OIJKux39ASwPWw6eG#jdarA{{lXNT)~^n7;Pe7N%%UwOJl-r=6@=
zV9<ZZ0<2Z+zOU%+#@J8N+`g(zXPVEW7FaclrOL=+G7L0y`O6J2!*-?e<1KWMDi?*x
ztmc0+HnB@sUkj}wvyUzEnf&vGzfe%Ra2Yf&!43qnbRlr1u*T<puMk>1P@HP98-Qw7
zLV(m`m*T%ZEjH6(#o}oqPV3w2MY~m>2OHq@=4RHiMj~(g$jMZUG0F2!n%uVTJk~hV
z$As?%7Q}tt9>thyJl32<fI5AmHsg*x_st6T*+r@3{vx{&3B8Z?9%@<aC^_Nm8&KGY
zS?KIJnm6w?V7m2lyP)qO?Dcz&5uqbK8)l8R$0zRSKJYwB1Bm&MBndY}w8#+>Zkef9
z^H)C7yKqQOl@Fj#;mJ$^pCA(60WYO`tFftzlQ|G<a8>$S4-H4#`k02s?$!YOQ5<ab
zT(#BLQV1hVxxO0DWh!TA95Q1MmS?X!SsWd_EOr?F8<w^IcyjRQ-M-<>8v(HdwO7<A
z$u0sS4d&k7VG*FfQ!7bR{$S4XY_qq?;3y0Fl%s6+zttAfl{2~bQ)y3e@s!9l$sLlW
zEx*=6iEH!tN@9S7TiCMg?7<b#zH^Z>CmNgvyJos-7B@ukewc3H4`_SnjpZvJo-EX)
z{kL_H554p}AFo^+&0^)PggRr$mVjZVhjLe8$`*jG{r$tFI@HD$6|*My!9DuIbE+FP
zLFPRMqK8Bpr38i~)2<-j%v<S7{djunYBrdLygXP&0Vf@TEeth;R)SLw4h{e+rPdBu
zAv2gYqhqUY-`K>z{2!XWI;g7m`}!bANC?u6v>@FLN(j>39n#%#2x$T725F_cJEgn3
zk?xN7e7?VV8HVBBxqraf&$DCgwf6#{-cj9>L-=A=XJ;oQ91GLl_Iy<aZJ@lqb^-8&
ze410$Wo1W#PzehQ3zPOQe@vICHqmsNvqJ?GqHuH?Vz<5qkJ;nOB=)q-z-ZmpD!04z
zkL*l|(@~$5GB!;+6wi}CCKnEqNgae2y-OV6Mi=<fSebPcUJ{HWGaQ1748r-kH~iDH
z8ZNU@)U`U+7f(CnRz(3`k+_a@j}QIrO{W3K=-Af^+Jwk4oCRP@(zwoMc4ua4en)1O
z*%Qo7M;xWz-~z%w$WJGCPn-(9U$Q@(_*6MU$AS2}r;L8)PMZzJ`CMvd%E06J4cW$h
zmW(Zv<xxTWnco{!)-{Y-5ZghJdY2;c-`_6d0s?23Fn@HUMLoAWa0;R&9S@7(yy+xf
zW#piqQldMmAu_RLeorcBy^jp{DzH;Pe#WB4f}Madad2*t2u*?veT`dIA&#xE8%as`
zE2N`pK&hah07;iY%3-Ojd0MGOot>Y6kcfz@E3x<Jh+N!OGxTpkhSrC8XBULCf!)RD
zn>t-fxmUumidFWrlF-so+iZwIDGxWG>*;nDmiY4Z=N9xlbhVu+fT7H|>z)}eV9xQp
zGk{CZ`OP9xck84H#F^Uo1`~x0p!rjNp<@?rG(3h=J0JkQd)e4%dg+80R<u|$o#c>S
z$6&n%K=4p?7L-9i>B?xaTsgboz-8oYlHE9XoMpiC)!tf>9@$)LC!@U|DWgN>RBZ2P
zYohscGSgv!X@A2N&n~PDyyQ^r1%W-kXX|My-G3_FFa-#cp{xBQh(9WtEUc0-HRDz=
zM7Qm!_dWdE!MTx~{xRZux0W#UJD?j!#9=Jv*3Jbh2y%1J=`%wm|9vA0rsqj)A|D^$
zA_ry0I@I8Rpq=aA3OTyq6GT)tHa6DLS&pWVfSAC$)F_N+ub`GEBn!pkeP2Cn>L`A(
ze|5jHVaS-sXT89#r-${*Z%^|2cam0t*>m&DP-sj_3Bo9224^W(|D*q#-YV}=nOF~!
z;V;lwL?jwRku{Hw>fn#u-XJmhSNodEO>%Md1gF8HDk&d10lU&3gQ{zA!Lk9hW@Y(7
zO<wmcfH}Fz7L)BUP^Ou5a~Let7<tPpWEi5qsji!5Eb(V%<JZU(p`+0QoryznZ5794
zoLlD2uQIAQ@!*LsYk0ZewUf!T_{PE4_<Ue!;Sd>mNKs-=O|R?QYM#c`KXmc6vo0KE
ze=pl|5>?%GJ53DCvuKfCjpIU>cz%9<dU_h_+x2;#wCp}<V2M3JJptfnbo6?v%SwgS
z{D0j(v#0wjZAMo}gOWQE6YdPE<sUC6<+vl6G%CMiKvLUZWn~$em|mc*shK|Dy*W;}
zN!9!c<ko|`SS)8gM*j^ai9v+Re*5gTyX0N)l>`1lsC7Y*flXuPtJdv;DmDN&(7i5_
zq15n>B4E%$i|$}Co#MiAQfSd@%}d6yNQ#tT%sj|iND~DJ6Q2J*p0QlWE<<0_+v{@G
zlOIVTrqU!ES5LyBVI>s-1F(Yc1k^@IGZ&)jYESNhgF8KjTVrZH;{WKHRL=F2A?dxx
z{fI}%$-{%#rrf`iM~WP!Oxs}fIBS;gIIkmiPB^{OBq%WO*mC#hPwBCJz;wlJO8|R&
z+}4`eZts0H5y1V$Vr20FSo?REgJT+W!@{kfPji1nJN-)7r>4~w_IVE!gT3gBJ7`Kh
zK6Y5HD~Km&JP20tRtYBN_3wU~9Xk_;_Y)>7MP%^k^Av!nE(B|fubj03CYm?N*dgz`
zxnO-d{9i+wyma*R572{7WYpQ;)_}BjKWtWKvYADWzVYbT>W?idEcChCPW*2HbEemM
zs5axt-X57`OpjrSCRjtq!7=^e4l>ZoHWSF>v^SQ_W|p0u4OQfaE}rV@bA>8JL7!OQ
zHtt2KF`l=N@0g08G>n+r9)?G_NbH%6I^<UiU-_by*iRiXU9k{~#E6H>d|ZQRf0X=}
zX)P!`dAXv5t2&a32yh})yhua(JQ*qLs*W{{l6Oy~m<3<K1B+C?;%nW-qp3QV-bRTa
zw-wg6F<<9C;*3*B_6^<z6U1b{oE-!%?`mly;7{I*sz``eKW5UiCigCEZoZ0Cragjl
z&XP>TSgtWwTpF9cY5J|pXRWN3*t+4taPd6`lXR?R$=1YEO7lU-#_fv|BCJ>%TI{BA
z`nb3}f9>uu=nYS(v1ADE+fCN`DUAGG86{H3Cm|Ruv-kHu`y;OjIla*l;W}x27$+`?
zsV;T=2=Grs-uH09K1cE$t5=!IKt5rAx(M^#yWWTc4(^%$R_}+T8GAoJKM00AX|Fy^
zgzgEnhTIi`IUw|2Dvy<%oSc)><=;eJbo8>e#Tzhi<G<P*B9)G$@hGXNY@eJ6nNQ^M
z9oKi&5YfN#!{g_YBfJ^(^N*#%<szl6cfYScRA3P>QP#XkKyO(uexHbi{f?AGc{z8<
zv2xN^$f>r|ss^@w!+iOb*UXE@t={UfA?xD&$C=;iV$$FArPbH!w|+krd`({Rgl8<M
z;shmZ8Px1hE-)`t8g2Z}Rz`)U9-S&PDps&rJ0kE6)}KG2lv78zODH%740x5wUK~Y-
z-$;kT;!-?XrRPIjE|E=(_ROMLW^fDIbHfd+K}$D`T5g(_PJLmdWq#T)LId0T$}RQ|
zX}hPlvCiT#+3Sr_nQa8^ELIw-d%MA&;BXISn5dz9dmWVv5-voGQ+jn@Z0#%jgZ2x7
z6z^{W29x)~BYldnr?ej$+a>S{2!_0om5Zz_*jTVQj?fX|S4$Zf6aPE<OuEhLC8~1C
ztlJfRB%V9TX8r*I0&loDIa|+q2_daRS8gHz63!$=5tR0x$dz2IwZ3qp<>3MEpxL7X
zS9lP7f&7%$PYcKy`nYd*^n|}WMhB4T-FY1_y<DE=g(IFZmD)%}y?t3fiE<M<YL@(G
z2`;(re=3>LPmuh+7?qCTTM-&v+OsPgH}!aNKn(mu$A2u?c0BK;KGjZ{AZ09eQiAPb
zPRS=+;7?4iS|GCUuG6#31b1#So^Ul`sYp`Z@r)-O3Lw;Fn8W$nHJnyGICMWUa@h1o
zdVO0C6|ZUdrRq3}1ptB4=<7^Y%BiE%=GtcZC~kjzbLzO?9MqHsCxJPzVjYUD9czGZ
zIu)@lBbSx5+<!KJ`On*M!M4)-wBq`~2v7d1ddKP)oT(5VU<exEC3?Tk7(@!ArcL2W
zQ$TNZ9IRif<W%y>>$N7@wG)GRlSY5r;19RAR2v?*QPg+#Zac$>9z6s@G&<(qHY;a&
zY1o~Q8WE%oz@)5CX{!ujL=2lrdU!)-C{GX#6NUH5tF_jq@n8Ob>|Dq)Y{}9m%@-rk
zvu=ip#i=Q9Dk&*xZcd%3ujqeYkWO)%`6P9|IdhPKZSz@g7&d*7FbMa`KK>mc&g~e(
z<jiw@S)*iw&Ek7#zwFptsR4n&HP)|hDf!Uc$hywx@dZs@*^@n~qyj|59?pzweR0AE
zmiSw=*vDo9b^+f(1Mu$L+mE$kv6@n{;dv0;EDzX@;jDSq77OUZRaBkr{LNB+*RosG
zh3hR4NcJHnJ0Q2YM+2jDhaEtyFCM~sfh-}mSo81`@+`kO^`LHH&M*hHXU+l&@Rdfi
zgil^)(uQ4~_?;jEFvOW%I1K=p8?fnv3hZueOQ-0hf1`fd6V_ZSU5SSWyt7+k00rnp
zu-OS<$f!ioKmJS5wx;;K%JeO2YA@Bv1wkbE3(toR{~n{qoNfL(Nt~y;AGv#KI}(vm
z??}ryyj=5hWUZTQ4No_mKtR2+b!p?5ZU!O7PT$k5{6U~9n+$AbMDSq!g+%{O!9a42
z52G?^;Y=MqkryY<`%MO^#GBg9Umz!jP=->vHR}gdGyoC3f~`}GE5F~9ZbzG!;?reM
z#dUxRZ&kYCW>RNVDpq$J1sn5F8JmgzQPzwQqtn6=HJ>{qAqS&B{e6uJziT-Mv9BC=
zw<iM4^5t2%sCOyZr*q>pHw^F1lI?3$kNx=13Ai+85(sG4t{!lPUwBrepTUP80mm5b
zI=Q*Jg2tdXE7zA0t6K~p(f|)$KuCg1UG0jR&aj*9=WrI(^yH?ZGT9!8ha?<KtK63x
zoxKsE2KCj!?9JhJBormxNgCbvPoVlti&yaBPIy$&E@!h71}gAQ1l<irzK`2TMBpqQ
z_N`7547oD>^<G~gOy300)<q;0-eX@`d&=sfBe~(ErVko0CB+?k*raN{1=d9>GiEqs
zCVNzlE1Gz6j6eZpx8nN7@=61%C`A`NE9X`3Q|ITbfZ<Lq0hWfk9L=hxJ|8db_8lcu
zILUZ+3pS(2^_T|%^?}UTQgzo~Df#~{P#c%Jg#lWdDFt>^#tQJp3g`LXfO4{Quc_zl
z(O$z+Ig3EYN$R8&pGamgo9tIRuF%n`)TDu3rKY}7J>_}b<?&zaF4IiEl@|A>YIRH&
zOp^UvBi^3Z&qOL7_1!U$kgn{r*WO%nr5>g8ajW_l9+&%g{jSk9(o4ZO?X-1{SQx4M
zEACr5tvz9(8*Kt=B_Fkp_495%Af#N)lvX0k@Os9Mw6-Wa&fcyq_PLkW9l<><)i%=Q
zL6rDkd+-q@B1U}bd7!n(*4$|kBOzwj+D=7BcrV;EErT^YA7Ni>**ZHuzYrE5t-5FL
z^@+wOM%Hx<3e1-D=jO41tC3$76PdWY&ne*pjz0i96f96tHny~^lj|dq(ia6P*=}^!
z_Pt2+$$S-6)l$8-=bR`(IwqzYT1*U#3P>^-avMvdZ&h+c$Vw&ZiLO^kw5#`Pt~e3K
z*~-1dRnL>UL$U{Y6NMJZvCnDp`}p2QQ`YQMF=z(h12*Z=+b$c;OP68ywG@FWaFQ_X
zsPi8b6~MY6`tdN|95I!MBZ9nMbjlMt9vfG5sD`WhU5EjEDLjzeAE#n_=2nzeoI<t)
zpAYYvvOhJU<l5Gfu6h<u$-wCC-o!AG4n!dB4~}WJJR7@*x4wmehHc7g_C8(_?@Z#V
z+?gENyxPp(Cc-%;`%_gzpav@_posHvo8H(T2Q0}e*X_bP4Rj^|*6h`dvW;Q2Z03>@
z$ewnp-vQ@ER|2TFIk;Dlpu|>>v$whj{at4nL_t6T8Y*Ib*hPQuD^%2UN%#j1h&=eC
z5v(1PaC+rU+<S`zXZp6#>X1-x=LfkIWu6azpc3nfbNly0R7gnhz$(XlavqwfvgIoS
z1sxR1STWe%J>^q<5!0Rt<?Uh;vxvzCxr<UV>0BeK`8CYEoF$;;!ghy2qHS^LFv!7`
zoP#TG(gSk2(ySnww;5&QOQ8+-^h%*CG;n}h@IV81hRW&SntW^*k@0KK^@iic*5nm(
zA4xjV{ok#zlM!03jDP6%XD25zK`UQDA(Xe=K~!+anEruO#Ky553&F<0fr8VYw#0t@
zRPGL;Z1ZEk(%7&5BFTaO&yK}mF%md8>=wdmGy@u9FsRY(6=vQVBI4iLW(kwI9sU>k
z4=aL#f*v0qgM#4muRZnke|^e=Hr_%tN9&I;a`ceszundKVoGea+nERMh+SZL+luET
zG1HTTZ6*KdOwe7!KO;<vE`MWG>wVIjj8Ypb1m7myEr^iVm|m~1cpsd4Jx3(^JpTqz
z=bvs#*|{9rir|yuMs%3AJWyNZ6kuT$HLF&hs`GAnQui*B`n#J_;*!@}ejwVP%YtMG
z!z7O3Huz&G^E#t-ZmHrja1_LmA?BlguV_|H#H$t!*7^?{_5>oRPSv5hJr6!+T>mQV
zsxJ9e@mqKC7wddT>#JLLB2|yv;$Nyq(gCgnkU|g!gBgn34*G?J>5sIo{cAGCGS(N}
zxs73mF7YH%t`A@2v#Qg~nNyD9#<u%If9*km2l#@D4IkF1O}>ijga=wt26bFib*9@N
zPQJ?xpLlwyZOM|gez;59hs=$r!Ie3e?;ok~-am0yJj9*Es{;#_zVcN%$rW@GG6O*S
zVrah%^~Uj;$RI0UEQaFe*F;2%>SH4EP7wq9Ojvwox96gKK!|&B;*NMOJuercDb3(c
z?qA)-@!vJ*dYOu{U}$0^RYToAan9$!vY{xa*6YJBXPyrg-q^e~$-igkAl^D2JYIXA
zAULOjz2=^k^Md^E6GAA8TQ_2!^7IR~Zeez35A9!LIJ>WnUUGRN%U`Mag9De-1(w_;
zLk=PaPqpyy@f#er2RcR}bf8o+qsjBR!`~r&eBxpZau-)1Q<;as(P#<XvTZOj$fv!>
z|J`72B6)svx$4)h2ILtOD#9kh836P(nv|I6J!)oW?TPaSOwMC|h2>nY>n>8A-DXb_
z8yk;KikDjgAj8Yn$j0OGK2Z&({p-zg?hO7-I^$&O$5ksKG`lpOz%J94iu(kmyqZ$@
zM_`APmlGr|d+^f3#ylA5t>~7Qtk=fFCL<z#&!_pvSZb*ODqrN~ix@3uG>9o>KC7g{
z0C=RH;?-WB297309fmY4D0G^aYcarRc}e-xA|}HM6O{6O{i#aJfEFXDYgzyP&!C9T
zoH6VChdo`eb$IQzMHCi5mUw6~$wCO_dErty19ur}Ai9NawmUad_Z6+=c+tfte$5jA
zqLrQeJ-l|2iHv22_o^qIWtjHT^mj56LCE>xXb0w%>KeiNLZL_YApJd$dO0hf9h4Gq
z1{kR&yNDDnUC{=0P{e)?T=o{c5bwWM*#mxw*+!w_ok!$${HnVlt;nvqBbxRX)}{jY
zte*z=#;^rHp<kS;?FO|p7BzHj*>qj-+-4R8cIzzPXmxWy94>J(^v~gK%9n@f7rOU{
zxKR=``90R2H=#_m!~;yUwpO!az<3r66v=Ov;u0siZjOyRa-Dj8+sx0^aRxvCT1a`o
zl_gBPyt+2xkRhf^XK9z?XJI%ly#pI^46y;b8~-H+gwgrK$nsw!Yjz~TE!Rx0$Ab0-
z7k+F#Dl#6iVvHEFU&5Pg$K&Gm#vt<rq=6yaS_OAjwp2hjTqsFe#!q70c4zbNK851b
z7jb9yFquasl(hF=0XSXEG1*jJveG~CfTR@%O0h)tf6Ng57T<lhVK+4b)FZXMHg7O6
zvS_{x-?U~&y_@z>gU$FO6lK6XPv!IM)DrFP%oTNZ!lNP;LdNN<ms6Qg-H5yO6%C^s
z{&L|1naq19dTlJ)<IB4!B-|^PzZ?wLpNgP8Y(!ong~9Y@E|rHxN>Tyw&=7z<+gfbE
zYXO-{eYvmN>269T8r=;kpdmauq8_?OG<rnoDTe3|p_?47?ArKmZ}r$YEsofQe}$xD
zQZ-ut7;@hfBMU+^Zl^#G@w*tdeCt8svWTitx4f1ya4br`@nU{=D8!I?^wjwFW5#{^
z#vkuN=|Sm?gnvBIyMdmzP7FVGjjz8?&0{KhMz%el4wngeU99c49fF5+7BB!BGH5H7
z&^Ffs#Psn$*6Z8<`aWw=o?bLruE7oDiruqBMU1fDW`9+<#CYyxc31QE3)r^MOC;oc
z6gy-w$nK<`oEMKB7iX1qND11kfjpcN;x;3vVeAMR;P9SLbmLBvej!sWr*h`ca*JtJ
z%U`y?aK{<$yBBVwpagGJ|1JYGx%7Ro-m5m)`U4trc+7>A<`9F&jfda556OVux*BK4
zGVaFqM_)mu?=Tw*G(6oQRI)*tmM1h}g0h^3t>uUSbb7IqpXphjJ`j8f+bJLG3&s=t
z2@?eYk`(c10ocGNlgEr5c2xO=;tRK#+j`%lz+tlkb}uv#nUSiUO<O8p_S4P%1&VZE
z8M;u0+#o99BxY`)47GoD)HDI)Gl8{FMH-gxe?fDiy}(QoQhvOqEjrdwf&wYK{A^ev
z((W)XAMjnugF&2_d_AsOxM*d0CO+yksLLoo!$B~=;<^!mQr`%(OZ?hWRC`o9Lf`tR
zxM@Eb5eZD7{j8--@s0?VNaG}A{q^!k2=6e=eRs?R2aUz4m>llVyZoH$YZ@3#xdMc*
z+>DKf&d+hYC1cTmUQw6VMo|$lu)^lCrPcGFUwQo|eG6zhb7m$Ga-MHcN}Ja`?vH3b
zNm!x+zD7T11;%wOsXa4}F|s|?VP<2EytoYv@IYYBa}I{$X+65nAEEdBer~6pFa4m8
z@L$<G`=@MlqdMyAM%bQ?_4SIhXiS#>TsZSLAwmcf6Q^rH^3xh+e@A&WRQK9dCNcu`
z-2p>nipwUHd&>}1`3_djNyd*S=*?shA24k>ubfsHJ|-W!#*jBy@k@@!6OH6{@O^m$
z0`*p2R!()I2mg^$W6=IPRHE85x3p&!nx73~HEgN!yk;uG;Q>5M>hn!kWSh6~4s>w|
zxnvy=b-gH6G2WOveJTXR*6=ro?ldLvd=#8Nq);(8v5O0OFP<G|Y#f{x+1D%9GMrY1
zzFpS)<IOK644JN<8Z|VL>W=i0jDlN~cN9?s&~O{hWq8Q{jtdf!Uw9RHZH9UA0dq^m
zcC#Z9Y&`=F9ewXh59tU@`>GG)K-d-NDn{+7OuqKxO)Ru)0!COWDbq5s9$m=icjhk@
zGZ^0=-+0QJy4<znoR0U}AuvFYfh!G)3?xMyKe4<ov){9XjAFTEF64l0UiCkaA@(ER
za`;I~z8?#DaKmiMr-2Vv{^BLe;_}TTD?a|&gzPSRRGUu(`i6LX0wK=7{b*Gb#Q7Mu
z^d+Bfr@nBWw@{SAF#Bq~n}=pv?of?u5-Q)3Z5rjko;D+@GI=xn2cfb3^@TbVAs9nS
z7mC!>aL-<aS*=~UEa3b$xsB<Mmy%5i%+FMSotErWX*Ld_NYJfwQF;a@X*=F&DMYQs
zA3rA+!vX$=UX1l$`}JceiGQfYVt=THs-^du<ZEj<2+qGX%h-7V-$0{SLrNRNP(LqV
zOshYKBdD4CCdZV|d9{}=AYMD|_+=~?9oiKXi0B$;!GLA~KD@RbeRJnI;4Cc#N`<sf
z&d<IgN8@CQWMev(I`9)lLG^oeI74}P7$Hms|GF_0`WK?V_sC++(k)h-G7`!m#`XpN
z757+KGpdRuNlo91*fP7>b|cD6Lg!|AKen0GXGSA%Y<}*U-@VM%Qx$!=)H$__<3Za4
z=|g!5>sFe9RWsH+XvnKCvl=$6BnV8n+MEf5^z&+2^cI{P2S)%iqc-JwQ&XLTA!UDu
z|GCRx@Erp=Nd^X^0}1(u`>scvfOt3{|J{lmD}}=SW<#WS$<u>>WUju5#kWKc8o>4c
za|yC^v#PzkrI3J*cO%UT=PSFXZ}#_3J_du(k%O#*`9JGW!20LToG`99{T?ztqU%V<
zqXzGH4(-pn_Nm3RhaoKKxM)m)y_%2+V5V9T@F4-1eS<GPW!&Y>e506s9A~DIjd@3o
zh~ubv8JTXf;1Pie;1oRE92N^sLDfSmQb*a575m&@T%VHE_`^g3x^7@#0c2<ztagck
z_w{u2ZwUD9Hk391_d)dGLR4GT7f}G|zd!C3i=nX4BIK3e*2w;v%@A{IjBjx2&KJ8M
zRtbPw8C+a!J5qar_4|JjaqrGFc<GS<E26;&Y)`DOXgJO8N>dLAnsMTNesoWMbS)R?
z@XW4phaF4wRrKud#n0_P-M@`SM{r^%wA075)4Xp(>u)!eu$2U?CEet_<jq{V7Clpd
z|ARM39H3q&+$(5&Eu9RO#83DKz<J2<&%A|A9_wl5Z6Bm*7&stC@)~^>`y)~6JEo@+
zDA&@6Cda6oJ(oA?60=O~eRh&{E?k*1EuUonerIp==w9~&Jc~lDXXSZB37>eZqUc#1
zDH`QW;r#)JicV%2g^bJ4kBFFnQjXCQ3#G*^nEo&IzZ6mt|M*fJMeF(g&vY9W@Hq<P
zM270}f0Wr}Ho1>aPIQxz?5*|L35d~xwFT!6molI(#B&O11Y`7tfQF#;RBU)9<HH0Z
z4SYa{DOd&LD@|=wHNJDg5z@6j_Rq<oA+6_=>JvxE|8<SYAQ7og>+z}xER?^M>9XW$
z(Y33b8Ot1bd7;bJV~E;Qg<}ZkN79Lo*gY+)5m0CgL7%cJ)s4?hRlly|BP3eVgkDZu
z?dG_FEx`w3fWMJ!vlQQ3#h#E^yvXQ!5EOdV?pU|I#Lv<(SY_jFX&K&IZ6!6J6kaN9
z#0l5=RhL`;&y0&py$_gArad92cfb^WxM<1SHVYw<HucZEEvrwd^HO79mFmAL!4s-O
zfuXtJ_q6Jml~T$VPPAy&k*#k&zi%6xUdY7R_dLX){ve-JOAj*s1@kw5=(FpfI%Wf9
z9usT^Joo%3lvKfx!N`G6u1Oz=y52o43+)M_y+>EW$qRj)#gO?kam;FZS1J!0Rqkcj
zD{=~UHZ72q68<mBms-$<m70%V(^Ic`w5ARZ{=g?bYUoM9#JL|9Cu5}VO9`d&_J}W<
ziW~4}y%lGE*bgBcJS0cE-UcI%$8WufK9QA@Y0_WKE)yS7PgtjAug|i^u1Gz7c-xal
zArAk)%c=HZ*kZF5Q>87lP+|Z6rj&-3zwVGE6R(kM(Xs1dU7~yQv2{lkKtEIH{<yL5
zz3UYM5~@dGzwi}?7}%U__l5$p5cBCUk2kb!pua<NFb+?|<u)EH)(;bQLxNh-ko^J!
z$Sy{?Yy=dX8!O|_8wYllhUVT+?^pU7tTBnaMF_Ek5{A;caXMu0=HzCE1~u4cQ8pL=
zzoS@*;jcgB-9Jah|GKbzknCI3F%gI7Hx%_naf6`S8u{7Mmm}pXTbd#c@uCAasqdhT
z)Lb3H*P|9U@Sv9_Xk;J+9}OD;PHa*$2Q~RPq&4dsw|d^9R#wq*xn}5#*-7)}BJGLL
zbIasPS4?no-DW)0STsX;SBVRUSX4YAGFTRvfb%U2S;`!|M8JAujgOQkuS!Nq`5sB~
zuM3gcg6u|%$=8Hp0o$i%ZixgF4HE8(bp8{yXZd3ia9QAH&+Eo3<0PBoovuQ&v9%ws
zNI&UGX#_y@OSx}(T$t#DjLSIqhTXh!T$~FQ$m3;vcM6TT(Ty2WJx-xh`w_$nIl#gh
z`uoE1L+u)~M@r}VMsG_EZ!sOS@*W-P4W^6~uf@m1OLZy$E3fZyM@XWWQSa~US4XE&
zZ!>8BTwvFnjLWDe0%P2)3qi3BFh(cJ#%jnlG%OXa@Vz?irW^c_;(pd4=2)<Yrc&3|
zQde0qc>T99JK~}WDMt5nDwz2T&b?9MWhQY^%iPN5>1~d2cFfQmAI=sEzmG3lL~o0k
zmgbLZoSfk!njGv6w;HFbq02ybB<`^NOsjhEajL3mk^ahioZKT-&-UWvPsOvH&Effn
zbqY~x>b$cwVpP$h^cg<$boL(@t;GRwzM}XfJW5ZmF`cUerudT3XpX4fOMMp$uON(V
zpt3hmlyU!Jss`)cn#x-Ltsrwe%sw4UD|uva`um4auIyflBP5`D{$CkE=*6;AR3eT8
z$efPCG;MDqwSI=cG^#c)b7V!f&rqFeoZaJ=@~({|6&IJ?gOl{4MaKqks}^OZ5p+6N
zotF{toI(abosUu|1k*q1Rgk~4kbAh(v3dVob0U51nG%BOQ90Rv`+-{1-|Ww6Jbv?<
zQLFn%UELopn5=Mpo7MeSzD*am2dwTG(DOeU{Yeoe`~zE7NWhpdn3QTqSQ_o;f62<G
zxKX0>?4Nw5f&IhdRufyIt}r&E*+kQif91C9?@c4vsx|~Y43B0i8!J;r7{kLyd)S-{
zw<nc$cc`_^uY4{I`y5H*UzZVAyYOi?5o^g!z36MKw{YzQKF!Qj-nV(F3Z{{PK$z*u
zocr}t04}Un<HypYZB)H<8xPaX!}6odhrL_1g46)6TD&6$@rR19_y7K}lzE-NGa_t*
z8SKmexHO4TyaJ1kb+M2o8y~_(FXnQ^$H`kcMO1@+aK>svJjJlyyeGH!0icZ_RjyJ%
zY4%xgtuN6_xkp2oT=!*zr;`Ie3kpc!c?HZ~k1@r$r;Eg%xF9Sa`<~10=PMuI<iGKc
zV*5X<Srt({duEMJNytk_Bp0$M>2bVkOJVqdT$g=zM)gvVcu!~I!o7!v;9Z>lctvE;
z>T)^YvKH^sSRu=`Ta0hR`tYd~3(!2g%#8As5$S)uc<7XxzOwoWT!|T8<lbKN@c;2`
z3nQdLj1liX{WqRCA)F!k|87+MLVa`hO4TqQJ;5lKGlK2xUom>q++}_~rGkW@r|eJU
z^|toU-r*NzQe2)q9u=Esu4s@@dY>v*!48cjH@EJx#e%~4m!Fq}O(SCTRD%Tt-@OhL
zo7^X7Rp|blUN?aANCX-NQn}K%FAwwmOtMq3zRKNtybL@{9|jIjNt$wn7QaC-&n6MS
zlhoPz`qyHQ**|%fX8AN0Ccv1D%Q5%-+=8{xvc8)uc;#wKhgbDpSld9Boo~ds9)58U
zcs!^#Fs(e4Vp5L4#%nKjJ1&LasLyCN@x}E{4S;SH0Mbs|hl)rW@)07cvPpzj^n7Sm
zM+!0++cs`Br@E4DToCnE{R(GMu;?`uzO7r3x$da1@Qz6<;)3UwE30P;VpZ_`qUeq#
zcNJ4=vBZ>Ht^ka0m-QA{_yj-+IP$v&e7dEto&AjGPPis^EThC1t<CB~2Tc7@)v@Wa
zyt+YU;Frz(4?MC$Rc(PECu6Z-%hJzVDx~=vSH0KbH&Et{uSt$KG`;59>78$rS4sb=
z%(88Azt)V5i{jAy_r#DJu3tNaj?zx7bPJryXRS2_`pa29X6ll6^B<<i8v^=#_y)6I
zBuM=5v=rm`G2ULFN6$<Zh|1-=QFULTc&L)2gGNKOm{+5qHfMI(qp{4HPI$g{*0VG%
zW6h0~fNtG`bZp<O*?6aYF*k*aw|Q^2e=bYeo=aiJ9u0zX+SiGJ{DMBwTV5G9aMd0M
zFeZP_JkgVn!x^^G(}}5Vq<q3nMbk`M;d0H?<L3yST}7#L*K-yLyJos!`WYIOgwBo+
zS}n?PmZ#sXBH&P&d!m-yd>nXsh-5n<ZHAE3r<X%@h^P9^IMt_8n!xFCamRT1=Dd_M
zHYa<<7w(aALq9;kA%KxjVfLs$3`ZyZ*~?SPIZX#uG(6(iY2rQSE+#P+&@Fi;%Cqtf
z^`)+%30l;vj`Un0_6e!%B)ObF;-MSlT9KvEnU;BNd|?i2a>WXqR=r!loD;O?7m9wE
zo3$Z*ZQ?=q3*~>NegEzD<h9V^^Q<ErOUw$gURvj};%6_X@zo>K>pJwoiF*bMC*FS_
z6w+wSkVThPHs!+Cn?2s3)kO>F=1*b^q8NOE*Zs39XBOU{JU&%b5C~i5^c6&1zhuLM
zj~YRTs~wf`#s1eUM0`^_|Bq=u<aMZaI#wTZI{<wHu!ugU5_23A(Gq06I3x&KF^nrd
z=3{w^fO>T38&Yy=gguQ=s5s|Fivm|066B($W?VJ>>uA0iLpQE8IUXaYS<!Yl9XkTL
z#t4f|7UG)5Fl`Ugj<H||JaEG_g8bVxp*i>W2?N9Jpc&1TzzEYIp-o0Qesf~>zswSP
z62%tSb}dgj^As3gGtLpT->yJAvzxKAf>=vdehy8&fXFzOj0Al*p{beF1F0V6+Q{D@
z)@(eNrn^6P#iWTA7(ScHT2$TBMEi~^I4XY{+AtZooEp}6dfg0W!1*dJ|H@dWM8$$t
zAh*&W;8JIU_5B&XZFwe3W9@YF;l-356(B;%@44j57yn)|yye=kM~lJP2*tAg3~K)?
zA@J8vQu&jQb;f7c>=-yqN+#KfS7nx=F#O!{|Ggg`Q(wHjx@PFKP8<`YUk(MUB<v4K
z^P@<u6E<8`U_mJ0yKUZ#j$MLbAe2Pyw(~PM+rSm`PPl_sCXRx@9d1L)&MfOtJF#q9
zDrmWhS|WxtV1pYy7M5}iBZMWNOzh#7f$BY>CEv$I$a%HLnD<ZKA3*&@X{2QbYhgBN
z6UUp<uzZ~Y0&fwdEqCn^|K{&kdH8Iso7DAcr*}J`4S<g9_WnN2#98uQsEyaw@nM9^
zHJJVH*9qf&<u0L~A47AGTg*1st_(wWM+klXb*J37GkRsDhu>zNksdK(U%db{;eNO9
zu9V5X`glVV_H7oFk*+_v%&zd$_|l!i)8Tz5I;=~yiB{A(_rfv+*u$x41MvWPltW8u
zTd5<^bxwcxT7jL!TyBKWGadO8I~^nh#4rDMmJctG`7$~&Z-;X5v-M%hpoPm^Jp1QR
zyS#%*k&nkT%CdxCuAep2N%HN46A0u~*M>RoW6Uq9V8z+b5xe7shq^MgvfVeoU%6?u
z;pX{<FJ~|{QMoFAH9_ol`!BGt^yc=`Dk6Z!XhL?a@z|$5_!S0uYaBuS$MY)V^)<z<
zU5AJ(8AJ?_BWX2N4N`Q7pJ~XPS8sF>Ul==r&syFBd?`z$yWd_^Z`q$s;q}Z@8$AHw
zH~I#T@9Vw4_aDRfKIZ-|M-7Ajc=NrTS>MI&QeXrpr=_{}K-gr_ZX`TO{##zN%K?Zk
z?pERy6Pl(-yjvumJueQOuDQXTZ=Bvb)p%c3P-Yp@V9Gu4bHb19Sa&0^EZg7mjGJ++
z<*W@mwmva*OA~Xc1r&VXX|^-NQ`n5Y&YkQBgbT^N$mIVpKK`JWCCI<4i=rZM%S(Qv
z8DOOSaVSPOF%|~ML!S)m%H&pv<N_D>RQvYSqFaamQlJ|JYWoUR2rnB_$sTx%<M#Bt
z*XU18CL@v!<AXEY_`d{~?OPC!#$rQF@w2`PA-=KlmLt!_#k-kA(>UXv2%{Tuk<@-9
zjA^$tO@~?}UU<aYx%Jntt5ZGN@JO)*``qHRySLZ20A%Z`@#p!9g(?|vxn%iT{{+6<
zIG0<&BVRz#INvuMcaK`bZYPd5@{{kpm$U{#jUYYQnq<f~pSNS{zi!y0afjnpBp)bN
z*6bA@6Z1O3^eaMbr+|2y@_QKH@Xxg=Y+w75Gv`JhyhS4<rZB_&Y`9Q?6cP4QeeL(N
zXCkG_Nz^%Jy)5CA$!qlBk;6Eke|3%HLx>Uv|F-|T+XLA2Ec)NQdlOx}MMY0DA4B^J
z1LeP*-w=g)VQDdH)>~oaTeZ7CBu4yuc63BrGmacbdLnTf9j(}EMnDLkw`}(!NaZoK
z?+BYSAn@H99yR6}!xQo5s*u~+Rl3>yUw1}?(J(bC3*&xGcMyt3cFqrGeFs5~7Tw(j
z8j<IdRi;HWUdGfwF_SWKL+p@nx5xO2IeFwk|CI(mq}`qbZ)y8{_jh$!SFTy_r~n@T
z(9r(8;a3ei$W^aRQ5qMk7+>hX`o_b=fRoN5kFB$%0c&!`N{Kp6UAQ1>{pY<8yN*28
z)}&x|Z`TPDNXYyapy2qhsrE6F_Aq{C;Z&2*%Jv*i@6R-sw%n5~W<N_d<ny80QJ)Ng
zv!$_`ea@?loDt}>w3?25{3rh8x%a0)wMK3a9jvt%obye%UTrGTnL_1Tp9;+}Jt`li
z`TMP%&%o(L>Q|f8MB+K0{EqS)tFKiZGltboPoiM;Cn8cREjBi910Qn1cBRQ}DnTAa
z$L!jwuB~#<M|(O+hib3Q$Fyv_{*`YjJa(;D;h0z0&N@!=<@wUxdb(+irR)?GHtP0k
zm<gg0L;B9H*C;8Nva6ehsS(GwX9|iaut7*hU1n$Ra;K-|S~4w|o3FR=fiK|-k9Iv%
z8uOsNm(8lRyy_%RV+L6-1`FxS)x`JJL#wOrMMazx3La|qxvbwY6Xws-1Y|^G<2md1
z@Q_AQR&zwv8|h+tOE?uyRPLbp&!E-hzRbdHIHBDU{4T<d2JX0Sd#Md>K;Ld0SK7&)
z#=S7S@-8V+?p{}yT=Dhm%(qo%As*ywzM5djm0PmS&9|<Zw$~j!(*aOaa49;G4u*2(
zvcvW7+U7s-Y|F>~(LKr`msw?ZV?0U!2SFhsL+7#k`>)o?^V`K$Cyo0hT*Ag;RGpcl
z?W-H654RoXe-6cAqAm_1s=&+p4794vlI2~nxsi5k<sX&q;~^zfii)lB)~U6CO6(Vh
z!F{c~(^y`mkbmm=Cgf6K<l!;2>QkOXsQgk~+3>JyOIqIa&}pBFQcE3hlsZtWwPXC`
z&^7earmgYm5`3$J@`&4|uc<lIsDaOgMU50>=AW?!Wlh*uK?yd&iDZGzV+hoe<5rRP
zqDQo^`7;N{|Mo0%x~hbKunGV@S~w`fr%S`AA^O8v+mpV1oHX7<MmCL7A)Bc0lKS{6
zXKt&C=|@|8I3N4G5YIv7tKmQeBORSHy767HvVs|t4HJ(yfRyuRH+@noDj473E;h%&
z0T_o!0y1Tdj*lDuajT2m;{JVKP6=Zg9yI@g8E5WacD%T1n4|e7dMUso$uz8!GsTkY
zQ!qep22)~D665W|@Y0yjA_fL$N%{(Yqw`w4czDG;%hn+tw>gGwp{akFGEEoIzGsN%
zTi%A&b;(N+aj!}=4mnUnIPzC}BB8=Rgj5qxB+5=qOfIthyYZ7N94gkJqT?YHKDZAW
zbffBG7Sd*ku-1;l0b2}YuL~qBSA^Ww^Ru^#-8h<_YK{tIT%tloz3(G=U=oMkzSS|F
zZjxot(MpN_9c_RIH5CrI=f@1IGi{iyR@AP>CO*~LM|C|w*UY6(uYI$aPLfG_{LNh&
zPLB8BxMt4q6Xa<(E|YpDV*)&FsvG9tS|r;^8gx$lLAv_~BB^;bmy+N0r<9mn*^pLT
zxjhjhJAAr6{{783eeohz+?FgogQVrb>7u`}v=C>qUG(yVxE{SSZ}Ax3mpk6+mxmc%
z{g?%bvmE+nxNOJEa3x%Dq6M>O3tUO`=>&VHB;B%x9g1nM{WWK7AK%~Cp-Y-<^lQQD
z3s8Lu{KDP@VL8}f<!zB_+Fy=C3KoZ@!`B}zT=w)tvuba&ZT#`YprA(iH9{@*w<)3b
zxU@{@#`o&g*-ucZX?sFtbN$`IImM6SDC-Otqm;euCU4v|&s+)(SU+Xk_HdZ-1rXl$
z^H=t5p0d$Ty5B_A{#XhThmcz`bc`fp7H7n6j(#w|CElfuw69!fh3_Ts0e6L$%hOz8
zKyC)B?dC$|UxgHKJL-qbAny6=DM}5yr;w6;nEyrrUyKc~%OAh(3FQTD#|yJ>f4Bay
z)QX&WCG0PLeLSkV(#8#*O>Ge(#~!z^*k0t&Ym}WRIqUfK_A$C75t9IJ&#kZ9mY5yg
z?ZJJhq<O&EeK`OVknXgBIlXy;iHp9OQMYgA>&Q1uOV>jTt9mJ<WeM69A_Sxfa&j_}
z39#+?-YQltbpY<-(942131Fxa10{7if;|CW-^>XInF-Mu#jYUSZhbq-+Q0sJefV~E
ze$#kz7J=-`{kC-hV|6l&FNyQfhi*3t5Yqt~G?7v=#$V_!S<#sDa*j~-@D%6(et%a_
z@IZV`NYeq+zL(Fi;#0>W(zPQ6wgkchyx-))ULp7v+7$K^Slq9yP7vmhaQ_P-h^nTs
zWOKKouB{un`)qujcqJMf0Pp)}JO}F(nsV7P1nlY=*#B%7a6CuC+^th+|I3O5S6W|}
zR&yuB{^^8SBQ<-BUw*=&riFjD?>#w%p7Cdg-?gNBLCkZd7!eS>CQL@|`w{Yd+>`U0
zkB90Ez!EQ$=h>9c^^tn~R{x@@_e;$(rI;;`>O<8P+uRX3%#B)I;`3JKg9V?5Ah+9k
z<Q)%_mtrrod>jd;T?PWwhKInlNYxMsZHo!LPnHpVWh1Goo!XB<<e@V;4hs+wrKXo4
z|LYat+Jd6mc-R>K96`fgmPw!y^BYWTw8$s7?P@(6L1Whk(!0J^6)N-B{RAjog<3L;
zgM{--&O(|j4BT(Lb1BVDioQ<Nrhof(R4a5_w+z)pTU6$UB;Y(1_*C)LgQ)LTJ91X;
zSZJ1R%^pgnhLYd4ZL1TDj6+Iz-futl%Y~~fEVl+RF7`=SEXbfm`yQM{Q#b8GiR93b
zdZ=B@4qCH69&xu?VzElPO+jJ|;%OS*+4+Y7w;tM~u#mr7=G?4;JEv@X1c*S=pG~yn
zN^6ggX0Dhb6Q{vo&I;;$vnbE;C0mIL#5|ba&JNTb`pSyK<oHPrN8C|V+e-4yi7}1X
z#A(sz48quYd^#vAgJM=}`W8OyYd|TR+po}>!h(My<BQvzHBa+!{4*!t(-v2TnB)>W
z)p5bE?4r6I?UZ+V3A_VBk|+WEs6K_^78R3id>cBc{ox>vkq3=;eMIqI_M9?72THRd
zQt-G|!|sq<&1(oTf~{|pKfM%*3fIpN+DRuK9ai4dWU;Xue@N+fK|lqdTitQB&w~f}
zCK^LUN2r7LP}fxA6ISOj0pyOUndFXGbPlg!+mBvXHoc*Eo!Fvbu&A*4%D2SzfSkQ)
zAg;z8)Z?Cl1wJue-!FWtp`939Udtt!x1szN6wvXtC_5I*D5sE)w|Ufi5*Bz-$0ep0
zlS0LSchZ8u!J&$95C3eKwGkMS1wj61_VCmn3%-wLgb^K!W~MzQ>G2guqEA{)5NI{8
z`NIVx^Wn00z9ZI%1mR^PSqq-e_KTq-5E2811HReGB__^wL;Y`7u<fea(2_F~Pv;sp
z)Y|wz%U?o3A>^3-bQIs8gld^F{%bpAedE*W6V&sMPkZ_J^j3&R^Pzuyp`9_swAtq!
z(0FtVo|HYGbc74(s#F`u?j#sF4~sw83UT9n3pFUW_i>RlJ={DYg7ejTda#3j2UlCc
z&AP)yj^0R%!PGuf(VzCGpAyXu+%I(4s$FF><3cy6xr=e1T5)3H6F=XN!U`*(m}$`2
zx?+y6|9VkfN4Dw%RP!(1RHq3gy2GrMZMVK+*iCNe%*GoT)@Uw)xJ6wnrVh4_LT2a6
zRL{2`McNY$a{tT90rljO&Gk8Ohc0ut&r4yJNy{gO2&@)uzr%TVmG#G;H@xrI$in0I
zi@^A(tdAT?=5QUZ{!jYtz!pW|uk!ln5ccy^_2{8NWOb~Bzc$G1+w)=R;rxNYA&(9c
zj?Rk}1nT*s+8AST4Q3&qZ;yJaQR665No(+f{YWK;T|xaF6aW1BRSY=PRB#kE91$0C
zv&`n=ZwlzkL^>5u@x5Xp8e0w${=sHV%?F6~i`oZZ5z0j<%($o%G!?&J1S@NA=X%g$
zm#eC>F#r+`0f9MpA2JkJ(_jf(7fwR!?V5Cik>L9-2>gFMo>d_cx=Uo(hMWQp4kb=O
z-xlR>_tnm3FKfF8u=#MFr!gvODy{f{RdX2yRx|K7CEGP0*H&*~FPVk!==S6G^XpaR
zT5tEtGsC9&_3I0FqC~E0RU7~v>*>5!NzN+iZX+NZN}4CFs&BXjEdG2G7auyhsI1yh
z0U(24kAzW^ibP1RmQr2MsOt$$aM<w;w}Y}S7j}m;<_e%%o7yaY-r15j10}RXRQ-$b
zt-hBlT^nssRZ;Yw-54in4!ZuG=17<miDi~f@(ACY)V;a5XT!#Qdc))V+0xOKXj3|;
zJDxv$30`=-et3)<+XnahLOZDPTEGPAKg6dJ2ilfV6v|*d#Jl0^5)Epk3_K*f;*k2@
z(yV^z&HyUR%Q!dSG4%rw=HgA0*+v^mNWRvk`$tpQGVrof^mr+bd3yJvCJomX%esg{
zmXJ0l9L87uLjq-=TERsO!d!2-A-C)H`3)x~4im+uk%J)T)_<|Wu=4Z75aIUBuFZsW
zqu%ucRoDf5W|T1+r)PqhWd66V2QwY!`=}ZZ5D;eAu=<E+Laqg8=%kPGCSFt<d&rcn
zi<vvJXnH?OOV5#rd-Z;*DHvFTE+B+Ev$@S~+LCfRWEpWI%-{4yyeAn)C|byvbApuR
zFmqO(|Am#e{uzc}Ofah1P5Hqf-V}v6+=x|zZu=mU8UJi-y^EZg;F|P6Ug1vE%&Y=7
z{U$Sax3>r(^+A*^h$Taew{Vz^NJYHlZmw%BAKoObrmI)}yLdju+i+Xx;mt>cmN|+5
zNae*0rhA3A*$6$YAzY#9-I5za<GuQ}E6JtN+HE2Ta3XV7obggU>)I!lF0rq|6D#hL
z;P6L?GgMYjZWfPR0QB%4L;G+1SR3F%yA>z&B6@Ei0rrP*dtk;kC>0+**|lngISIYE
zKpt@&or57$=~zgBHe-_klA<EYJr8Zz3x2WmL4JL0`!;*;fweRDLic_no;3{ZJsE0D
zz>b@UeYSB#WL^BfSiOnpY(k3S|7Q3}sqBPfkPF{RCxJPzd-Y|fQwL=7x$BQkSwKS;
zW79a>hjLWX=09uOC(C#nl%t9<*f;B#md}svog8eF91a+s6G~S>Z9YEd*SNLBZMmc!
zfijejl-5qS=>g$&NIhIC))-TK0C!7w`<6?{-!v;{arqNqtssd4H4+iEl<QEpa8dG>
zBKO1RAM|wZ6SDD2JwOSvQfd~(NW~D?_8m5tCfjptE*(Cxl-w<74mNO3>4acTUoR(B
z<s`7eze42%vD+@mOBD-lgWxkRRm+&^h|CX<-EnM+dqeL6lgZh5i0G8mC-?j>7888v
z*jR%v*hek15kdvbD+YqqWK-jbanmeIim96kcI{6C??RC;KunllU5DJu`uytCF1d}%
z8<&DlE9+4i>Vt0{`a9347Zcce!*v)rb-=Vw<xXK9UDe;7;WDmg{xN*6{c}Rph)SL$
zelbRG(q-;Aym>yUMelh%JKvZN2iVzv6q5DQQkbfC!-|Fyr=Y;3;8g*Ydr*P_jkM*i
z`iZ70esehojPB@ITTdR)J7Qr0UU+tsQa{F=KkT6B#<;(v5kmv~*H7<y)YIb(+IMp1
z1K+}tvwndnfSV{<dD5$!#(0PauySlD%mCmf4^T;tgSRL4G1-ooIjZrwFDGW(kQ?Na
z<t(L>M6Nb4+Jw-@A|Y)eDrA*qEMg1wgUWh982q^Q+N)|a#~6YPARE`E{|28oSrZd!
z%Z{{pC{U?kQQC`EtB;CVU)vs5->onlJm$}w7dYeZhwgtQo_`$KI7eHC$M}(mxA0~Q
zUp&>4Fem)}U?qXkQ}6k8jOmC=J=BP?p$}`cHNE7%M)ebxCdbJQx#nsdb-9Or+I|5s
zvAYq2UUW$Ny;OF2XZEN0zW5tv=cQSp3Q;O7WujfDeY-j5nFDRYw-NH>#}@lZtxsg`
zo7_KJm6N#K|2y%T=mFxW-(+eti*oN`F@TamZniCp>p4Q)yNEdZnaVsLNeF<M?Jw`P
zp5m0Fg71qL+-nET929_GIF4_O;ON0BDp;i@>sw(tmeWh}s~>Lbp)My9yTlBf=g;JK
zDQO~!QZ1Ahok-a$0}w3ADN#>m!m6Q!ZiKA4BSO}joc5g-C~x#yMfqKc1~4*JY}93r
z-1R(^S@B^vdtbB`E<}<*S*nH6KL*FjH%c|Y=|F9me(<ABDk0!x0dWIHKxX-6i#PMP
zU2P{CFH5^^?k|7aZQ_i3;qTcdkcpTC1*{PjPGB@Wn!&kwb^}abl*bnMY?WVJH-F_r
zA9mydp3W6w4m=c6bCX~AERa=CVDNmSMw4!|-^k=a{Q&fUv##Vfs+uo3OX4Wy_$cC4
z@ormRH}fa231U~*b6O}TK?|yznI`Z)?<~O4TR~Qg7Y4ZfkUulU<h>dA3-J+IT(3<>
zTxc=$AMd)Ihl5_UHe)-?S`n{WRn1MuH#@~e#zuflYN(+FJZ=m_JIUc|Y%8{}S^UD(
z?<Og3OVd{9DA3q-YwnyxCXpa;b&QL4&8`{ZXBw~nh;{MUk>NoP2Eg$D9R0li#EXFB
z24|R-oS6DEF&D;nSBQ~MW0pybQ;cmCR5W$G1N4WBFWawqTTCuyRe;lU!VA_7WjuJf
z72fc{jw72Fy}ArY?r&-U5VmzmIm}q=f{2L|xY)Rw!|f?IOtHje@V-z5$Z;g_K7C1d
zrQPSRTn15IZ8He>BqEmNwF$$BuiZyL^PAwqNC<u`XNZ12yt{8zZ;GE|-k-*Ug|A*<
zdd?)q2bp$Hnx<d@BnXk4<*QIUw!TvY_R3?m;MK}I=DaT_J%^VWAv6RtaFt2n(W}&A
z8SBsYC^mppjc28;cs+c>L(DBNE7SL_Y?hk!VIBSHDwthw&;a}8?)GVN8xUqY-H{}0
zVv*-P&N2-AxoCE0#x^?d&5jVz^KXN6Sn*lzkArzoQm_#_><f>j(A*ZjTi89xM&`C^
zAbOiw;Un&avSUnV*~g#k7oAxd7drptV2Nx(XN|g+;($XI)H)(Y3CLXL-^^<B-afr0
zeVT0rsU~ACVrwesB%9_dN{U<i^<jM@E*x1mJlG;*u$~}j>PwjH!~hf&e3z4OiXKNI
z+*7YrkNFlWy*&0Go8f>YJ}Msw<?zpjDzMqNA3uDFQkq+S4w5~P2xh8DB_fBJ-K2o{
z6@C{bP`jS;)5`sPG8C>6!=8^<TQ%9(KmjhvoCKsM>*IiUYpeT4bMx`yznQmb)~;@k
zpX)!9T2>GLKc>Dqs){dadjKV*8)@n8P6d$$mF_O-?mI|JN=bu~(g@NmA>Bw`K)Sp0
z8-MTnuJx^T|F~=2VTL*ToSn~pwsLm(NAK#Ym2uM~By<$yl=!b43S^V1vx;vu_6355
z2G)&jbZy>55ulT$J7>n>)289kHd=3`ONC%41jZZ1lNaHv@~e8=rsykooYnU!BC9&E
zRoy{2LKA1#@-xu5x)A=xp~!_pj%a&LdMie?m|R6N+T`d;Y-%BIQFiE#U~I>(uX-s(
zuWZeVI3x1C%(06@RAO#|^Ep;k|1uIZ$<v7M2H_<`JB5VE>vpZ@e!%DN%OB`F*?&5l
zrFj_zHO*~R2;9W}^qQo>(i3se%`o=(JX7%8xcZOc^_3ckKIJ{u;v611(XBY%4T{Z}
zosLsw)Qu<;x!e*b+~xJB=B`WDyrm#FqeAR%kCbIAJcp!88`i{x``ifkpDn--Xr>#2
z^!mvj82h~R0A85xneBkJ_lrwgkoB~1Hv&md05s5;H-ScS?Ocp6=ea=v?c<G-YkJ=?
z<P(E3+uZM8I1VRXVY*h=l9$;F%)rRjNMirRDk}1t3y`Y?1Qafct3$594mQ!q@EE*J
z=n;OeK_>&knh#D_-8(V|Ro*^cKZIP>v6=hTZ2i0qwacx;gj?z_FV9dNq@5m-8HtL0
z3&})y$&7n<P%i8xq^vSNcq^MODLlIYUYyJHGRn$R8pd<;?)y|Y#UR}6A;UYV1mpik
z_2Nl940Ina$h-I&y%5<qIjH++{~t1>GdR)=dqiELW%~6f<Ym@cCBOE;e{L{Yt}6(X
z_QyY&6phsUb$!gXp3tOBFtk_fF3u;70LX}ZYz-~t#)U<g_@9><>K;-f7=!jPYxk}(
zM67?$CQ1c7@Rbrm0!(g4P`tbfPR26fbVGI`_EzFM4Uipve(y3(VdqjeD+koG8<9e#
zMIF+I%#bQDaUMfuXBYYMWDo{3<QwvH45bN#(sEwc-}6g{(1A;$KH}H8N+qbkniQ15
z$FVf;de}Ukx?~?7nJfb7q_VWz2UOlxu*lzDj%bQ|&zA=1mQ3gg9F`ykxL5N)hR%;;
zvu|Dj;p_ay!)<cmY;;?_#fA}8qlu=6eyPH*RTn7xq(Q%ZC{cUz<D7S6y~i;ln->50
zT3;oj_%iyWGihSc-)`z;WwO+(Mv4Xpt$zDQKjiy>-o}47WI-(!I$Aj?-c}68R)3mX
z#KPaqOV@o>sX9M8$ha{!*0u!h-Mr&9C5~9>hQ#i-uRUY=o|bq5*4#~5H(Lua)d@9E
zmPy`E?_-uM7WOC_Hq!GO@K))-Ykmh0J%hX+22SrWzghyn{0`5rV-(-ki((dKp{@y&
zjBi!^cLapk0LXE68O&(2F>X=k{pYS6*VRD|9|`DuX*`|pAbES0T~0Qu#PF)4MOf<@
zU@6$mfv}Llis?*me19P$W~XDBB2?PFZJG*%JsjM#=_GTqu<=0<ib_g{W1u~QZj8NQ
zx+a!{hQ$0M<Ypp6x$&g?^Q>Nb2&d4=S}5M^q8khfq>4<%wJq?^EUrYQe6&i1)Y(De
zRpmaFp5!XDoKgao-6b&3>4jc{%&RWuP?+>C!QG4bC|WeqS=ObdA46;3?G9?|{XP^`
za><ZU@o(MUuHu)iV{TM7kpyUYOcOA$;OBN}mA2ti)MdkV>p$M#!N*DR!_Ce=ySxx&
zd;=za%^3Ckr9rDXJRLS*JY5lO7x>2kvfTJzZX?##_YIq37v1J0XbpRdXFpc<xF(^-
zQg(HV%XipU;jq^(%IpM^AYlJSO)KKNLN|K~IDMN(0UE~qqEql`*GV$elR6}64N;Tt
zH=pXW3Ho$l-_4)oho!-h8J_~WXq$hUM?qre<OD<@Gm4CykSZX|?9;P@r@3vmFXQ7#
zZ0M{WJXK~{MSPSGtVCMGv*d&oAaLzJofB^R*#?d01$L~ya4DbrV1J#VFdl!^>u9@P
zOTJD&7aAqPAb@Vj&!=kt$NH?<bd7vZ0e%mqAR6H9ypmp6Tw!Z;{fbE*FWVs&BtWhp
zaCSBQ{P+*O_nZd-_ULxK6r;B<aRcYQkX7ik$~H-`))hOuxuk7~ddIiQK*EiU@AQLt
z?C#W{^(Bdz?TRJ6&#S2RWjB;x%d^;}8&o!r*W`O&^tZ{@1psY985WYS<DwLFK$jef
zhQ=7)@Y4VeE9?8|-mxn}3PTJq$Xh<AlbsT~13Kt!hd2IVv^O91w;(hoe|UJ9BU?;j
zp*XLun4^?sB^d=Ka^Vnf@cx;btF`mT-LN@_k@83)V0mSR_S7uS2YX;e$=lp@RGgPD
zV^>refZm5fu;A0F*%P3BBN%<YXM>_VL>LF7`=)a`4XxHrPYoLZ02_1>fXYJKz2oYp
z9bZ&{iSpZ{AMm0=i#w#|2ljKYOF|fp<tIff{`RK`1TitS>r?MyW?cz@+~Goh=Q)ii
z=ZXk|&Oh*IO&>+~51$gp!+fq{enaM~{r;bo-pdTkmYbv7zm+<zU6rNA25rL{XVCtX
zP2|L@Com*+YJ`q}9hB+0UdP|zYtQ)c5!!F0&Li}vrVa_Ak&^<D4qPb|ZIn;*&Rw`U
zQP^)s=6>&xnc5q*a({Tj<=gq5fN=)n<=7x}QbQ5C0}^8ySsH+LhJ;*_Ew~J?Q-jZ4
zN^XLyvO_<~L0{jm#--nVZy1^=*)-XPaFE)uEm>f-)#;$+a!}-?qm^04Uj~TUy_y;w
z+6VQ}<ixA%_%+7ld~=ydBn5GWUrp6?7d6qsq}%+2{jB6-VfFy-p@{z`l)jiSK0ypn
z=~&XvdwzJJWW8dZq2(ao+VHlF;9P|d8f9iC)20`9)iN8Wuluh)#B=PzK_<TFKr7jX
zQ^4j2HfNv%$eV!ZxdjQbHFBU@90RNOJ5ww`{kfkVauYdww*RUvJ^7#ft-vuO`y-?P
zmEmZLg@OBd99GC*3(Gu<Gr3V2Q%a9E7Ujz6{6creA(+SH&00+dK=?4DIvQFr#Q(&@
zKa`_1m7|n2A*%z=3y+nRFK32RQp?bS9o$hdTRq8?{^AUk`yO~y89YC-+awu7sfKTH
zaif&ly{$b4J*|GH7z8I9=g{fnVwasm=a1W1y7laBq^u*OWRc4TrW`|Z`36D&TsFDJ
zC#a;fx*lx<#fXGl!bdq~t*P6e*{+})?Sj9a6%!`FCoUvw0COa1j`inPiERSkA7}~d
zK4kgSg&BKHH~rZNZvAk{?_+033RLmus;2l%tIx&}#ZD<YmK0k4=Itfh5kYB5`nTU=
z)!XIYqLphv1aS)7pie&z=NbL=bnesh=v-v%S4i!@Fudi*&+CJ^MYw?I;N-7swC4~>
zK+iMEnET4t&`f~{;fZbZSx_&YdT{&*Vfm1;bhGJZKcD347OkWcYIxR5<N#O)c84BF
z6Q}@}8h@Hp1KRKWxb6jG>H$;|=xW_)E*yZL{x(Jtp_Ubcgpd^qP!`;I(;17!jok(|
zA_9H{eqPSuiGty_l>hWUPPja_{Fnq!yq<wcj*AV<E*N>5?l+iD&X`64P;abaCmF`>
z68~#=_{J1P$fiK(Ke6__9v#W+yY@quLjuIl18ehksAAIVoK_GrP;qf=X&QQi*q%l%
zA5pcWdp9YBKz6SZhmjqWFMGbv9*zqDuXz6@g|4}^icOhNf_x*_MyU3jZ?>&k1(s46
z`o~V3F-o|>>r4LndjcfjQYGP(wxE<OUZyCAlT&-QOv#(ky>;NX=CHJW_SYnLsOp@u
z06F&aBPA@*JTd^s681Ku7<TbTYL8T5PLRBr?tWwa*CF;`gr?$XAU{;AiIT)7V$BO$
zv+o-}0^5WXXgA(skbk&n)-myYFdh*t+_M+JLjtwaUcj7utnam$sYW7pG&WM-H7~O`
z>!LBfeJ>#y`pACnYHrDZN|p9)d6`6>*LMdFiaPjTr&wr}4XARPQa(U_@?9eL_YQAW
z%H)<-F!SfTv$qY0xYHcgQvc|{5=hca3};|}y)w~&dZCMwng%$Qk+*bg$HDrFw(d_i
z(Ah!BreNb{`~U!s4o@6{f`9<FGV5*v0<M4ESonwy03^)28Xnu37AOUzrD-GCwp==T
z8>JGDFxpq&r?5B&34q0qTt7~w{boe<IgVB~746K@7IHvNJ?pw8m9cxRd80~YZjXEU
zgq4dL#>>QTH4@gD&y3OfhyGV|pd198KB~~$@(^v<kKf(UCm{C74CI#n!||HB2eEo~
zkP9Pb^+SzM^KD=~jZmgaAhkM6USc^VM!S)x;heV$HXsyw?_+R)saBqtTi^Nj=-LnX
zs?O4|aIAj^ckR3B?-;IZqVrQp4A%RKZ)MwEK|Zd=*V*wS*+E5qItsP}*6w5zBP@T^
z8LLx?Bt3t{@bv$3hp%KJI9yc@y?wNP8~bu&@ZmXRL?xHw+A+Ole*JCLW4?{7<%$2a
z+j*wU>e0&Wfh_wQ#vdH2*KdoEfjd|m+yMpnq0L4VD?vmI=)-PeE=du-(#fBsr?0&6
z{LJq-jnwW>|E`V*R7tFR(;Ke*X3(-4y8Kg5`%Gj?0K|w3EISEVm%C>MLG?apBJQj4
z!OTN-k><qkSnv#vD8cxiAa+(}p}HsnL2f$y0VW$WplW|L?G0S@<#Z)oYnglSGZfZ8
zyEiuh3?GU^H4aHa#TzX9lhh6qo+f9zTo-9ZI-+RLu!FnLCYRo{(Bj}4lWcj!!&OfF
z3azH;Z!axvzE}?YZc_UPx2W%Vg$R|6^~LLj1SAb=s^`{@CkWXugl+m!qhkf<v{Al2
zO|{M!#&{HY+jg&n0vH3a90R)Xp78bWP0V_!-}do035K-A*Sa}WH44*PT<e8uIjG=b
zkQUVh<MWs#CO_=I*rp`*&Y7}si#H5}t!jIQzWGcP!lMZB^g%$Rz<;d;^~DRH^XGKz
z5d(tU|MVh0$Voc}ECfc$4V9}dO3Yln(_y{6MiI0uWsaa)dWb@wz^6wam&VK@61;A~
zCKG@16mp1~4J`dk`_6@g!hg^+zNu#s<nslI27;>jIALZ&b8B=qmTC>xvxJQ$3xfak
zIcfca{7{#(Tpz!WPPB5tV@$J`h$nLD^c@pzR?lBV1E5YUp%4J*@*4)J*Z`o9de*j0
z+{`7SWOU>*MxYHWO})Qsk=d>rQD~H6OpCf({r*wo(3J9kVM`kG!a^-^Vg;mMg>!cC
zqZ9vziD~IA?$9t5bdc-LkwU&ZPWX?fNpaAuNG(DP<97wmfaWI#i*{6{Iy$b7pAw%w
zFK#E+FfCUYT*uD2{+$O+W+l*S%+^z>wHKM}g|#}ub~39ID!OzSm<#9iWl8q|>XpB-
z@R@WyTiF^5eNp}w#w3+Kz2NadH-Ds3`zNg<p#5)dDdmUeCf9qZ@u=k2-8B&Ei65)#
zpTAi$l*2YvXa!N0R1WiiIG0yu2#?o;xVIzKx-#^w??(jVE#)F`pJ=7G=<bP3U*Al-
z?`>kzf4Uq}q?p+l8%f(~Y}wlc7qY&5C#9jeq>{y)FY%cfku)Fy>-ek^5zsj#{cj6M
z0WWjdUc`cPxY%XF;8N_FWno+4Yf(m-ta5I_l;(AI73^1$UiW|0{ODR%1JA8<TRJxw
z<rv-!4}2mjV^wFw=GX5~_h`hPnx8Li^(a|V9DN5J+6+*w*2SZk&B;})si~!Owqw(n
zO8hg0en@@i#^VDRhPk(Sc{rpVapRy>7~5#(%8tY(s&R2P9*mnAW3iu}%#?NwkLZf0
zv(c$u4N@!kLb()_Keqeq#^oC}`A4>B<99rpx<mF0mgm_APU#tR*^&9hPkHgWq!zU~
ze#l?9FmIvwa%4<3)oIK_i{Kzh!M<wo;fW_fk^2dk#Sb5FVXbhp%t~|tu7U^T&}9^f
z92YZ3&}mqIV*3&ya5)@VE0Qq1UZ_Dcte)xR-|AA?11!^~!jt}(4Gm+Fy?Uz@6_rlG
zuAfZ2&D!}}dDKXAe%P(Pi9EPdRz}eq?b+g-p9O3W^(05j-5ZNvXx{z-S{G>^QZzx$
zo)%_B!!(c$^SPnG$m4|KN<mNLqJ)XW<gyb!AMx*%wH8`(VuxT^5!3%{^2`4;fIZL8
zl_yDG;VZ0;Kd{v`fo1wFHG3?mFE#1)fDfa_lkQUQRW^*?gVvq%YJ#CV6Ze}zf@>&?
z5B7`d4K9Xl^i*UQWBx4SZ>-DU*4)m-;so>KPB!wG9XMlNFl`(tLz%@t@CG!NSHWrv
zcO$Npfe-t|KXabUw#9h(so6T;GIO9z^gY$#UTIP=b&rsk5}bLZ%_5qyG^sPkwkQxe
zqWG4SSyFS@=nEM?qw+^HX*SfoqCFqyY5G6)KU;<V{#N7(Bj2HojuDNdWqyym?3DnT
z^{y&ClMtGMTR=>w{0fx}s!xG@<|^cb<h)d$ggWYSP9&X#*~{CCxyH=b#)^uu75TsG
zHsjq_-G)3Bej;I0BFgY``cHU&9VTl20pD?~yHTwbL3dl()TDQI#sCT!75dXCU=c;!
zUeDb8{l6c$p-5=CH*V(l1$q8xJBb2cZ_B9?|I#8ZtDx8+xBg6S_oTHvWIpo9*Xj+n
zFk%8~da}>G_2sdp)sc|g(<^^oz5&-BHk;-}6(Arub|Azg;;02|-0b8P|M*zd2LJYE
zlj_U%Gxz(%{XWg<fv{)~$L!~Aoo_s6?2HVm;A-UV^OLY$R)V2sJ(Gv#s){VIc1-@D
ztOlG|_Hui5l}51dag*ed+c)KTM-<uCy|Scd*5;pfleOZgac~enb{i(1iDk6ZF<<&1
zQ1a*%X)6=6BI_D<o_yYKvik7`hE4BL&oa00)Hz3zZ5p$A%AeqLGA71H&G8OVo_Eh1
zn_mQ60)u6_|Jzby|F#tPG73r1yl6>uY@2Z#U%6IB=Nm&?gl1EA9)j^-xxqux66Fd?
zYqIYe5~8+#yWg`ceJ;1`TUAF$E~n1!&O+ltVs7Pch_k1sp8L8J%X60zIhJ(aDwnM{
z)ND`Sna8!Mj13xH)*Mu9qb<%@9@pwLGwSVy9FK2KTLe!TvxR*C8y`J4WkO8=V{em%
z#|6HhJ}3d5(HGGENUAE~_EO@YkqYXQitN6;zNY*Sa}Og&`3+Ei#%})Bq}o2;GO0GN
zryb2Lsv*E`NQ3XR(<I)+Ijaxb>6~CdUS9mhcSoEgUYZdTcl<^9T@ULOFlxitTK=O}
z?>@=r5*jSv2@={;%LQjoGzC7r3*k^*cGW0k)Na=@{BzO$DLyWXBc;*dFsxAd8Lp@O
zM+L^s%^ob<ClVrN$QWcUPZ-d#sLrAF6)2H3Z<>&Oy#GG7K`*F67W8aiHZ4Qk0T1|M
z9;$;YG_$-_iTg?6VMkvk^v&|on6K@=Y<dQCJeOW|f(4@Q5p6vg^$eg%RA><3r|E_C
z7ksF*{q4j)Tx*k>X=91dGCJ5-`pr`AMzk5c$#xGKqP&VeVWFe@ty?<lgDa$23`ML-
zv<oJ-tseX)x6dX3E>RsG4aP5MM?Te#s?qTELP?SNPT(|i<@*?B;FPrZqkCbe?6L4n
zJexPD^7Czbw}@TWaHK#&<br#j=6C!d#iTb9y|o|6N*2w`oAYN~GBd({)cIuQc5YPc
zo%y*8yI8&3QoS$?X&89b#p-2ynz&8+aY7a}u$3H#jQB<c3g)4m(uO|lXN+4H<Up7r
z{^vlDpb%LUFN%(#0Vum{NKfy^kXZp^5ICI(+a1=kb$hJKQDojIXN~&9eC_SrGSOAV
zFy9vA1_US!>c!^{k4BtnE_L|WR?nW^X${ftoNlaUW_lPkzjr#F+={Y}>r~utu$Itw
zq_^4N^wnEz;*7-~D||S0g_ruoCY>g4eppF@`1<Dkd^uC+*;`L3hOt>OE(b9{qh|-!
zncEFz8g}mkLWIX+D+W74d8SW=Q?gBUef;({!9bU{)1$e{F)vrYAseH1`Q6?7@*=z(
zL^u1`oCCGMNiKAUww4d8t#_S~F63CWELICR8eLsNdj^Xt#N=o#{wl!EFy)1j-q(ly
z@><AxeW~6c<0ZsJ#speRYXJ}n-NEZG1ZZuCK3bbau&R7{$J6Xxrymg_Z+a$}uYA(m
z-XCH~NVz9N;}xZ1Iu$kCFXp?SUC|}IWL=LwbL;SPo8zN@b2eD@r0DAsC(jIOX$H+B
zTx)+{z(t6bOXtcn#QUoc_EE=nR{ZO}@91sbXpQqSlr}CfO`E3yKbgP|gTHJu6aqAo
znrDOKLdO=4xhqeOA1V0Azk8;@^A-=Rw9KuM1v2@dzO;ZxA9?-uO+WcB(C2QpB~t+V
z{9!xm+UC7Se#OwvlAbL}fY;N*)O0sH90|aVN!^}B(BII8etR^_uH;zco-ee6(=i;=
zdi&{i6@Nqj(&1R0Hh7nvS+|(-!t53rG(V?Lg#8py<%nUI-%})OU^2Kk`a(q2M{e(e
z#k}mP{}bV5M#klMW7o3S*AiDkN%&kr9scHdmT06@+tfO>i&=1+K><3Y$>OWdk3O=o
z>2}q4%pWA8KK$<<BqD>tb|IV!0`pELuFC(#5`gM5kqBRNrVcJH2d-bg$4Fn`5L79S
z3q>=+7ziUYfLP814~JXK-V6p}Xe0)s02u8XRVu(J^OK5&FXwRK#*oxvQAJauR{jzp
z`GdkY0y?`tZ%<b8Di*U^8eb;h(=y9{dGzt#S+U~>&g#Q8BLz>U+e-H=eNX0kGDI!A
z2EmcH!y|7U*kZiTFUU0qvu&{t`(edJo3mvar6QTh`Tjlzi=T%tfX`k!4-^jMd}`DH
zx&2S1m(D9V)jNL!5SV22^TP)=4fododm^gcxN;)bL-uXgN37RYLjEmnH65=ih4=rg
z6$M2v0$-C(U{%yuNYpkAE{me^me$C8c;I5OzD8w+<w&n(vsLD`&dlEy^P{W2MGd@?
zu*vy5z{&L^DUjydJUn3d+d&d;eS+8?HGpTGSBMTJzOXlYg1=v)(teF|qGF22vVt@s
zW#FO_F|$;#rSUYqUNADkqPUIWsfT}W+x!FkDwn;CD&yc7_iWtmno6rHva@>%Gy^14
z$}neZuBysfCl;F;&y<6L3~cNDi$&;6#q?O4eH%#u46MMUufWm6Hl={Ud@i~vqHdx%
zJcyTn8%dMC(O~VPr6j;M5+90cQ<iw!+q|`y{a~8M_~r@ekOZ*2tnDn-mV>R!>?!?i
zClo`+uHNa)ST60G^&8s1<!Lz}YSGGzwZCfarofg<aKjWg24xmL1<$VmSD%50?fn*3
zAzBLVf4PU0|6-9Ils`u*ZwHSHVE9|r6j;gGfw(s#5QP0G@U7x%+&}Gv1T1CdqIZo$
zJzQarXQ5qxdzX#5Fer5rsShf?YHnOMrtg%+^~=T)NRNGuJJAG*t))bmSN@jX`WK21
z;ug=@kJuHNBR#XsE6&!D49=7UVmdO^p9ADJByJ*$#wVoYPk=#XvR{uNMgyKrj@kNM
zsde1v$VYRN3?b{qSOkx2X->j2m5a4q^iFLBSJT%;@1*MOD5)?wV(eQ*SK+=qzbNTu
z=0<+`{7T?+?HC1pXLg_<9_j@mt0sU9tUu*BF@}SOTSP)$uBbu4-as<G(i(1W_)7lw
zr(Bb$N<=%2{?d(K)IBkOoC%N%2^A2y)CbjPt#zU5pgVEt+Ek#^A+?A%TOR=d;b9di
zcs<yle1Sg0GSDs)bm`*D2`Ei{a><bfEr9w4>N9BmRx!Nu;jE;p8)<&?HM&~vO!gwL
z)1lMbhz0=KU<YZV{Bfd@lahKQu1irbGEN(WYPVw&l|A!H2e40%>by~u53}<zd4G+K
zzh#_=iy?v1K1xcT(Z!+MGZm3#@uJvV2Nx7F{hIrs6T?3P0G-B_PtHyHVv_P~q^ila
ze$IgiFWLVNsi+0-MSMB7q8p%@{DJNoK_dS_eTx}!7cou)F*&Hk6I?X*wGW!SY#cQK
zPjagb0}%sq%%`gBh)&+%<qAt0Jjht6o-P}n$&x|5yQO$^EVQXvGi#eemY!kJM{!Dt
zUBzms$PAK<OqTb0$;K~<ZuFLlj&aU-2k=0M_Ha?eK};KJe6!(*>1Mvt(fvy?!Qbfg
z&`8^lH?nEW3D4khYD(YAJXU<tviis56bZ73(QzSy$P6V<qb(g3M^KBsc>yBg*>&o<
zvR2-G4iEJp=E+aq<WXN!PJIMzxvbWm4NK)%*Sq_@4LowHCd7Glo&lwWEdi06-C*cD
zolaqcfElN0p$uH>o#SZhCk$-3bVdOtn*QlYbj%Uh_cC&!+#<bIM9E|FetyH_ldIrH
zLH$ScAQA5rQNLvo@^oZQ!w&q=l9zp}#_}vM_vFF}$U6@EI{W>0&q<d6I9ZjV(kv}9
zdeQzf{>ZhBj=pVd&$b4G)EOt$8!Q@E9oB`CcQ<1PE&Lbb1U9cqnWSl#o!@N7cLLfA
zc^$86=kKgU7JjE?2GuCN4Q@!4hK0Y2`d>-^*CsEW7%a$d=ur!IZk+mi_O8?*%>;@p
z2!UsA?~}ahA5H-rVuy&*x@&;?9X+b8ovLQ!8@lDr;)S@hV;ztrfRcF3;<8S!xp1Cb
zmb4UUo{uC`)!!v1nzJ`rOU%>un{jZ+I&_fU;g!$IP}=`=%%EgQf{!tXzUFHSQ-&>?
z(2Cf-UR&ElI!+kGpV!^toUWs_DHvL@t-iwgGyho0<NU!-GJC#PX9Yzq3u(J`&%Olb
zqIFMX9|ly`_kkLZ%_M&#>4xg>f2L5`w9m7(QV^F;Wgd8>b1sun5otze?>lMi7y!CJ
z{pdP!z=9ArK(Iq+;9*ORk#*0-Z%9t#8wxf8b56KSW<rSgR7J*H57>p60CXa?)<esY
zs{E`%Z!VieOB%%(=k=$MN7{XC<`4>&D=YiOe(h6$?CY9HZX)3?&46&SycXRqC@oo8
ztyq}w1W1pay)yC-cK8exowh1}0C8h|-fgYS7w5U~Opl!5wG!TIARyALf&14i^SI1*
zl?j%6`c&I)PTAbDecer1YPvK%sn7of8{Ge8u%gJISYx<WR4z^UkF|ja{WAb87R@Ar
z3S;QJvYOE%Gtv^k?w|6cP}N=n&BfQouhE~G*wlD=s;3p74GWZvP;nc)SF77S#vT?D
zdjgnu$vwK=s^fxPseicw8j4QrHN?ZnD3G4|MRrd+oFlb2^9ghJvIz@aU+3_}X+8mr
z3$Jc>_i*F0!`9&mJshIA70X}+>su@95G<;i%lLh{+|0aG;>%sZrA%B@OcEYNKlu|U
z+Kbt*`heW@+1x_ufk~<!0`-a5L$&LMGGmk-K1WQxZLr8w$yXiOc&s90dkfAl_yU@R
z-nrCvYS^Cn$fue3h<>4BMGgoQI?hu3gPbreJxYSSNBTSJK?m)Pj|dS6t0&C<LHbet
zl}{q@oBG7v+@^EhEG+hO<dRw`%taOj$e0MjJ?hk3MYQ~WZ29;GH69_L{btx$F`NV$
z55WQbgp(vj5)oaVo2$4G&X;+0jww%Qe{iBXUF6{zn+|#7eubORr?Xl-mOY$V1@%x1
zZ~0VVcO#*oy)27}9vtoSMpJ;P{H$!AsySVuMnezf{*_De|C_w%hV6bW?Aka2*UECO
zb_kJz0MR3#j>{6oko>6=b4hD7K~6v+Nh)iJg~Gxm;XisqF_C85y6>8$QoiD?^~m>=
z+8b%6BC^l>F0|bpMGZ*<FVi${Rx)LH{xW5QQ>WwR^Tj@CgN7|bR4YZW(lG?!_!+E9
z7l$h<3@76*<D$V|f(l?t>aThF?w{!L2UOO&BfvTuZsW<!IHvnu1Us;-?6-2Q?p;wb
zuN-D7{exSHZA!AypA`V$h_TQoviad+XkW*u$m+BGCDDY^oO<*8pl%@QpQc^_1=aKt
znbuKs22^W8winM{Z_iu2iMQl%^LD~uN{f3YBoEO8R@SUDr8Srz8EaSAWl<(}`4SXH
zz6{2tM*U@3U8K%}WmR7yLkA^T2|_<A@F{6Ul|<<GrY6w|FXGvRf3b{y#eC(ad<eIt
zpd+^w+9D83h~x`Z9PjyMSvtxLdI31Jt0sWbTdZarKRUC+3K_G!1)l^-jc#uf?<j4t
z;Lnu&`sA$2UlJJ<fv*!P>h9kCy87#79O+><0}thb4$AK%QT)>~56_y_91sKO2p<=p
zzmqv$S=PH(x~=OU;EKS7QkdbN(EjTnUHJS=h&N&VH>thVJcS1*Gco8AE1-*D{2@@C
z8zZ^*#TfGwe#5%H=P;>-eR7NMn@mE}2Y$QOC;(ka2=?F~W!3oOAWd?WnzImYvi(IP
zKd(P~XGkvtb?=fz3;Sep>+UvtBBo3oV$;PIte3PWRc)#tPG5NU2BFEr4p1XLStexf
z+CNnGjdTW@933^!;3R{4>uQO&w)Rk9en@yEZTSiGv*qQ_%g2Lc_grl6SJ$Aq4K;;-
zb38U+fF;QrJ{HrE@t!S^9Ij>i6re=8UEQIZ0{ErEb`2wR+e3NPcMM9$hoT8hk}Fu>
z^zaIWLqr$NfmGypJP{Kz#zM61bCR^^CmWQHVOG8>H4QCvASQq(VVK4?E2K;zs9tbf
zGANy&7WwAT_~U_M^?FDnceo#si}MTYOFXx)iychRX@LR`53;|s&pQI%W6=%$)V&AV
z%e#(+Q*;OUoBBx2uBPe=D2;kEza^XkU+4M1%l>k5ct9n|qyxgeQubW3hJ}PVRfe0G
z6<Z5?9(HiS-1M&_JK}<t7^KymUoe)PD>pT}N?v~}k?Xwcq>B1KL&xChMFu&cdzB4a
z%xS`N#<p7*BzrKDhQ~YO%Jo?&_g0DD_9u(GS3+E=wVWFXhz)9~l*r}N=r}y&#7Rl<
z5cP{~0eg&3p-M5m&-MXL&6?iSN>Ak?4tPn6dH+6h+x~%}8k+{WJyzt}Te~V(rU<E+
zPsS+56!q?9=OCUW9c3VF--+?o+ud+!Kd}fGF#xVefu4Ljv|o65=eeH-8u;zjzxdTL
z#DTZ`INLOGRa;z*uskxg*m;t88D2*r>dW|24$-`l6g5#wkBMOJedA(X$oMPt5b(?U
zMr*}y{>=7Y2lvGL7LUVao9Pt|t1ens@0}>RznJTNu^K#mY**E%fV&?2C|lUfa?L_Q
zbAee44FMQLl!FR!2rolAx*0MKvvbSUS)T^5<VL~~B&@aCwu;FHN5z*fc#ZBnS1ZMx
z>c<{3%NkdQutsW<s8y)^my4lsgP{ASte}#}#)`!qv7_qnTa-r%XVm&??en#qdhU`>
z2*6R4DWkIhf8)IBAy!>A{9}oShleLje`)ERd^`!XNA*oW6p2z)L+!XSTp*zHcg9#q
zAMfGG1olIdcjr8&W99*xzaXCE!nenrTlNY{tkZP0FAJB6UE5^Epb`fj2tZFB-O5Dx
zen@F=+8uX6B894~Owvvp%rFs%K`$!evsQeqo$wT>CpUd3H-5D>7@iCS^leK{;rU4~
ztR7`k$|E<ug9zM!&tGwZIr6N49g$23NI8ftV8o-l0N^kZKioX<MB8Cp&>Lpt0|5*P
zVQ{Yu_vTqF09r+3(D|y81c*#;{Xla$J#V<UB=P#>v@*UkKMh+MfkZy$%w%_5&u_%R
z(JcEd+rYcCk<Wsh(pkiu$N@`K{DvQ>xT@1~AYvB&IdKADAx&`d8I}j?&(0021YMLX
z<_*W+qPtBCI1a`caHWF}bzyLJh~?4;7#Yr~*mbGS&P)&hsq>B8rZ6qHD7|h2{bwkx
zw~ikUthHvE(V*Uf5+sVuAXUU{f+7}ZBP;lNSd6Fnt?v=^{c_t5sy1D$7rZYyHIr?O
zvD9_|&w}b%tJe1%a1V`mP%a|O4E8TteHWhq$2^NWw)5#rMwFcbk2SdlJs(|7Y7Ks5
zVd)5QG0K1$Wg7_XMkCrBWXN4dP)_b|HHW})SJ_xObMH7R!g5sy-1=F}&&#Pxe}iCn
z8+$+|?Whg*Y01MzQPzzEc01Q|%us=GN@>(@MQDYikkk?ae(NOWi@crI8#^#g=owK+
zV$Cr)p)l9@xUpm<LT<RTk01KnIex}=wM8QSp%XNr;QuJx_><-D(@dzGvg>|hmNzY}
zCfR*=|Jv7}5FIGx7`Jyt3{arCIxje(o0{3PQO{s{_sN?`(7So|A*8C+KP6TWlNXer
zj#%3?35lT$BQ%L|`y<mbvE3rUpJv*=!eA>3CqyQ`?Hr0B6tAKC@hFQL@SB*??=pcX
zEE)8H6*|~_5tDpH&bdlv{~%uTMWC7(5J>C_Rw3GQbV#$4tx3&GrLuQy0yXU#b+?Y*
zv$tfGX;=nLF>bidV3x!P&JL*Mebhj*V16G6kx4P7;W~ZFfW@jP#M1%op<Rd^@8anw
z{N00|=b(5@r??X>Rf->Fqqj)k?R&}H+=Gf6j_lD?jnDC;sHrQtd^37w$YX;9-ZX!w
zBS@AO<Iy^e^ncTod^0wDk<@T<=fxcsFvl_S;6s)kd;eEBt-b2VoF23&H+9S`gYVo9
zYr>UhPZqr5;EG10q9fb-VpwrGYjtPy_%XnYI?c`cyReDhcjmDs@&#<1h6J3e?%>Rw
z?~dJGEX+TS#ZCGgwX9Do>w|kd-)T$PO;SU|g3vdGL4M&PwK6OZ!#D48gnYN!O4YZA
zNm=E2NA`IZv<g(T(Dx0CyZ7cb?YZaf_Gm&3p*m{!yDw~?XjC8Md@)`G7cgKf<4Ps$
z6^NaGY&1+WY21?lw7>8QoNgk#WC539t};7{zPfIfBNJ4UWgL^{za!?$7HQ8ndMG^&
z1B0V8yvH2W$Q!Po_wOf(({<k+7GkDvFj!i@_Q0O9b8&1~u&?x)B~yz)z>4ME>~y4J
zQ=N#?iC_*4=uSPM&>GC<pPK<%=l{0N?IQ%x2aRM~J*Aga)jGyAUl`Q3i32r7paXw2
z6ke*mCOnQ{>{w>uxc~K#qwl?+t4@v*i5RgUr{Uc|jZFoGfQRT^7@1udS-EN#HK0lS
zol-A_%dN$A!W^omd8((oTOL+m|Ga#>3XM;xGT$EY*aeom{coVJoYMtA4*NqEn!TLu
z3&{YN)$V)GPm#%&>_I4C0A86om(#r~qy$X1D7BL+$a7G{==^{9r^ufa>SiwKirBFh
zq(6DyHvj6PvG-Y!=#KmO=|rB99?tE>b9{N;CR_U^+loi)Jj7wO$kk*^Ol`yCt6xGT
z=_2qAH?M7btx~6a$70-AX+K*VKaDU^Ch{T8)7Cb*i$C;reKHiF%}=WBx~fn>JF~*e
z4fa}?xic-3W>!e;zQcm|-`&faP#)|pZG%V3mB@X1Iy2P!_Mi4=@@r;xU#7#<j?n_n
zWr}&qt~PI@fTQt`F+62!A@e`q@ww!QhvG=7;UFKOev{YCqS1=*)`Y?wF^(TjQ5z%G
zUq;^YVl;bH4uNZDkDGR~%U5Yq$_F(7Ebq2N*2JOF)Lwq?qCZQZdUL&n{p0GB_Qv=b
zbY=KH3=!}$)zJ;4qxFdTjh1ZA>Y>jqc;OW#U=%OaJ`9UZIJWA-CbWD)FFEp#8)F*@
zfSRyYkKd%-iAlwoe}Xc!-uL0$dwO^XjBkFiPAO;&f^W}^%qh(+tIkx7dhd?nx9!ra
zCl1U=Mp>Rd=l<2CU`C^F_P7?`)nHRqmee8=IkLi_VY)bI-gmP?;vxSF5;fAfCYAhe
zY<*OyY}<o;^v#}iOhII~)XBc!K1E$HxKjLc!OBv62kTHcnQ7<KgamRAU*V=?mmBMC
z%+|0wP%eo+vP5J7I4TLEc)T{XoZQsW%1g^+fos|~m^)nx$sv<O2Wx#{YbaEwpsLDl
z4p4+*^6^hZzaL62O5l`X$(rHy`SL4GiW;fhQnli*o~?MbHHgDDHAaLX{_X=S;O8^g
zn<!)HA1u6q=giW+)-jgO1$z@LiY2qt`_%U!Ujp?iJ@XZ-71Kt(BCU7C<++Tr>wEh7
z;xeC-k(B~<QwN)PbN{&b%<Q&SIrqwr4cgTD&*kB73I3BW=6~|-S;rZ0nVe^7LNHcI
zqjBqXOVnH|o-loVn8mI6Ji{<vB@(xc^^<|d*IB=EBmfmCa-yM$c4Sk{OfCl(-jC(;
z`VMil>Hf-2rD<wmx)=&NAEF>Ce>=F24q=!>E$u9aFSRn<?1}sn?iYD5(?Qs~0>7=>
z?XPQ_9!>ChM}r-?glt)+c=B!5Q+4Q*tKZB<=tPx8{CL#-SMk8BnR@RS5?_POU(dY9
zxt!n%_Gi4|_w9Q#9UgFB&@$vAl~6jGDdu})S?oHtGmEA*cIA98=55DFQmXE8ZN<jx
z_b07YDYyNByqvJ3{L+qoM&_%aOr4Tt$yTQ)v*$z>y+jtl*YK;2_$x;awCXRfelfQ4
z$)4s5r}dVg(hxc!+xmTZbCtQ?IcQ>n=@HscG~V!+(hQ;94BeDU59C_rv<|2IBj**Q
z!Tp~dWqyj}6+F7;`@y?kZ9pxQqjM9lsqG!5kKgYEi_R=bR?WM+{f<j-B4RY)2hS^N
z#yevTzmu!Mj^Wa{lOPG68F|3iFx1qr=iFkGPDb+4c|Bi~7dWzBQY1SW81J#WehHEz
z5?S_p8gPSu%$q=AEuD$Kp{PXdvGymYsOKYblxZ)P)^ppc8afWX{WF8npU1o=<h;eO
zxX-AY^?UD|G23W^ais8(A3w>UD(f+(m#Ny7dz-pqHZ`_NA}_l|5(1%FZzN#Wq1Wqw
zSPMy&fc8}vn1|M}kjx>QM}x0)-R9~512Q#q0Od$Fz+b-wuDyeM=oH0BC21%bE9qy9
zd-%1;rBGj-35U@_+QWubk=B9R8(22ZynYZyLbi)F<oBHM11jl!$vZbXMlnuT{;<k^
zXaWL$zxel<=Dn@L>Nh#+tQz6<P=UtAg0u2;Gb}xjGVxAK^QCSEtR;G3$sDp(w8}8b
zZy<LcUTn`4guzLkg9l;$iuB63?PvJw`0Pb6&P_%;AVJEI@acK>dG-XQ@`Sj!cmn(x
zUrfQ|NHRJXtB&6{`2E52T=;fIGK(f^LnhXvMUd41p(v2YbS?A2C2)QaxZCHw@0c?L
zmT|h&Sl?TH|6R2t7jOPZ<~Drt`XP=e0=(a08RNQNs?ombnW4avRaITuRYF^cXO|e=
z!cvB{vdHQaaC+?r69R?{g`5$@cg98WdEb`g%NY}_+SuU^o?HBBaR2Nd^ZVlB{=Txd
zLPtUQ{cYKvDf;TxF=y+kIZc*Mg_9Mtd}adJ)eQm9mRZ+V>Npz_C658zSTC1yFnG7-
z;)&yvm;#*Sfpr(m>Lj4iSvXUr5dZRL_ApZf=>^rR{P%sEvT1(^;7(itv7t%~osgQ&
zP+{!kdh$vxU#Q8&=ftOlKt`U<toCs><!X_AWz*v)f2tUjhOT5W`RA{!?>Hci4)uJR
ze&yKQPN3Pja{Hw{Cx8Pi>8I9!jb2gHF0Mill*37gXEuOMdBD;s_9s;Kl!8(S-wv{$
zsH~hLq#(RxBDi>*Oi(mG<<%Fpweu-i+WAHxr2=_vpYQI9pJMUP+!G{xC+Iilk?aQC
z!m@&?l>m(F5$q4xz?Bd|02=U`CQixYUeGEFBSK9hTV0pqr*9)UEEN3sTA0&-U0dsW
zFNu4O9Un<V&VobTYns1Kp{c6Qwi0U$y^^I>27F%4bVyL7Y-L!ym1QcELw?9wz;v;R
z4uQO1Qm2d_iu0Lds1&}WZtt~kbb%(_wL^d3GQHyRTZoD)?L8{S>#;85LUt#ijFgmR
zu@-9|s6iLjW>l8t0#;i+N5+~wKd9K}*KTRyXw&Z<8Vx75VAdl3fVjQOFt~)p-|AUR
z2%rWyR`ls<Wr=lQhNr<DE`3^@?8zGX9I!}z?TCD8bSbVgspNSriDhu0&c=q^1ri{d
zH+#gYdKP#Iy?G5|nRO|I>ihJXhDcS#ftYV`=A$PxEA=UOSTp+iS2|blD4#4&C6A<~
z#lxDUu+ObEJ{d&$w<@@~@N!I4%H&w+^`xlFzjM#@EG2~UMKHrJ6sx&K@x@=QFgKZt
zo@<+gIdbS%)E9269RT!u51euH%#qyM*a)<d-#|j2V5}2Ih0c`J)|+4iF)7sO3OHep
z<Ut8&WwgXAcpjIZnlMlKKQ{>a_1_J~T8m?h4ofW(LKY>|ZrA#sCoE9^&jp?TbHRX>
zJG^4=yNTefwn6+EeEx(>RS^0If^i~9A*=L)k7JlG74{=PE$2DpZW3R%%%7qmu6q7=
z3j*<fE~)iirX#rR|B$O7DFcmceEwqut`pT{{_ibOzrFjPhiKO<-rUJ1Nk9zgv#H#Y
zt<@=E(0crP3Mv18_h6eMv3opyps$NX3|JvpN54!&^}14y{I?XrJ;0(SP4H-n&7RBM
zA;{U}M*t8ou)^GpQo~0w_ul+l4iHjsh>*g^Ate(@*hM)ssA{?gJ||H7+L)!j4*0)E
z@izB408n6z;(>jg`KZq&>JNM-WV{_JaA`noGsXqlA%8#-MdVj&_c{(5k71R<+eB(z
ztvoy@$=C+GKP}J@JnWIF1-;^|z1k<PH(R%lQ%?X@_87f{3NT}D<KMFWi?~O%M^p6~
z!mxq&{P9MPENbD`32+)Wq_5Uc;)KlZq1t{JUQ=6`+)d0F!PxZf-x{Pfas|i+`S_f!
zv$S9Wl)5Pwz42!+6dwM*TrDKV0GyU@6&b%f-<SH|2hYojHw)b>MvnLFWY7(f7wt>H
z&CB;IK|Os~gym0`I8AyMx%YNVs(7j`-eYxlH@twCozBMnsoD&!3TMN^B_qOes)n$w
z6@#+nssz=uctUKB{;kT!?0R8Yc0wf&*HQJKe2SWhT+|YtH`^YOhvp>{`-P%=56Rdb
zch^~b4Nm?3o&R3cI-(t*#B@A;4j+2zY(on6tqJ7LAMF417Jl;Xds3{ej8*zj+tQM8
zrPmFW-HwNVyJ)NdzrT~p<I&0))FXhGi+~zX!7-Ayr}>q+j8C52gY+U9J{gG;#h*zT
zneMZxe+aqZ46XJ|Y@Y&7D^Pf5i4)*N|8^9B<VAoMZ)!L{e>AV0pU@=Ys%5eH2vRsE
zp?T}I);NjTZeQ(ZA*Y!@Wg6zR?0l8C;R_N9jvmt_`xKh7-Rq#g5mK`B!n-sf*w9N<
zf7th~g~Kwrt2q>Vf+2Mh{RB9<8wJ8wywzY-=xI>7?TCqn&}TQRf0&;hEdl=dSYyh;
z*WM9nhzk$llB~|Wf`ihjg<uHWE8g=vm8<Sle1L({9vOU+Hh!CkQK7GUZ;$$}_agfo
z(50+Nj(J)&JEnE~C_LI?O?&9lDdc(ShETmTP)B6O9r_h7{*Ra-D%mt;Ov_LDxee#P
z&7FXLxHraymqf!UpPt#;MrmW1f}fkwTGVhya35)3LokHu1!g3-z19D&@q}>MCA@|g
z-^?O#Q=*)Kc|)?$O|q@T>3te08ltgJX_tpjpL?75GXUMy6ub@@cP?K4D4D%g>2r{p
zelyws{GQd{QhBlQWX-vl;M>ER6c#XwG1HRx#^v6~inDL8^hcJp^H05dScLrN(~}h$
zvIupXon^@*POy0b0uzsvDW#>d!FGRcTjxS>A#ZXT`HpN{*5)=b2KkQIug&(AC&2Q_
z#C%B^@w@@F2yn&jfh%T(hZ|E3>Py3rj{CiG58&z`BA9U2gwpxJxT&>Z0-tmB{=PM;
zlKWJz&{v|Y3TU?_d+%x*^6e?iH}CXqP$uBD+ZqK5=@`jsxYx%Dh@5F=FRr*ID3vBA
zGi}1tO0++7pg$Xuom7NHG~8xK|Dzf2WtI7-R-PdrS(EU*xMWqFFj_FQa=m?YslNst
z1~My)pChwzECtJa^DP!>+BXEWMR}R|e6Y1YQ>ew~l#W94TG9<9K41vmWG(Mn#PLbL
z2Y-H+eAT3-1BXGpMb8>0uH<o~-WiivS6=}yyEePdA{X_1{3VyHwmd$6q%E^FDw(aT
z7_R-nwh>uvCZ)QbpF(wD?YMPZ^{6-=HM}yR<0qR=E~!_>K80-65DPJ^ByVo|Mtg7)
z3jO*bHoMczZgjl{(eg2=;IY2ePY)BB1y?XImwhwL?F&(WFIEF6x5GMV^kzs`3G6YW
zz@jR5p<X8NHixi!OjL%q0{r9ty<@R0WpQOf&u`PMJf7}&WZGK#_q6Ps;;Wnzf`3{W
z((GeL4$R?8YE$m5-OY8yuI}&rbkbN(M$*VzFI9GjYjj1*D1q>}qYX<Nfke=j*EC3N
zrOfA?M25GmxzPHR=HM~%gy`>iU7w8Dir!{lkCVd-e!C`f^GM-lbL*b-0Q9mivBP!K
z?m`)cJnvrgD<}Hfywj}`$+1tBkZ=S5iDcKu=H&_GhJ3DJYiMRq$Z#XIZ#(9hMdtZi
zd(ZT~{9X5^d-V*ZlsvpLW_eJXN4oO`^1>jdbIZ4uX^T_Dm!(QvtOkU&mr~cvK0}VN
z(sD#T0yd8$hb$SH<vy!#Pdm~$l4Dn#(aIDuvYBpIejULHuiZmr#s<GA_LLE0HW~{0
z((%7(Zq;x4#`Y=lvK|$|Pu;}B%Lb6sLPrGgAfGrlbKkDLnae;#jTMYT7sg~bTog&4
zkZl8tk`Xq(&8B@-yWh<y>O4@@Jeejgsxk|~;%sSK^eCN$x&TUKMV4AQ#kXQP-af?m
zc{h*##_h)v5Rw|uTbPWzg+xuijK}IqczsOsBG$U@#`BN1Z^l}=u5FQKGo4z`3uBW;
zH`UDjW{B61XxTa;6U@w^@^H#ZB4bqed@8#CkWe-pw(-OW9^#Qapo-Ma&U-^y?NC-*
z_0cUbqm4!No5m3ukUP}&Slj~NDdB6BthZZKr@YHE%WEN^YgE964EI<kKq<2}U$^0R
zPPOSzci-fX=_tVyITFo8by(E1sO2%G2Y+bC%^j*ck9WQW*JC(2<w0m(%Z#dh+rn{E
zi&n07*!9S2*`vr9(bZC|p_X37?pyoI0V5_3s)KLRv7OT&ToBqL?zE%rQn-(5coU=q
zEpx3mk_^soE8NZ{du#u?sT~fgud|XV{=(|T@PNcxDA;svc{$ob{DMkevbrIsb)w^e
zQ=Z$NK`R~XbFVD8)k11w-PwKSbo<>(vb?pl`a<&iuKVNm%hew@Sw@Qnd{67v6HvXW
z{B6;UkrDWY7ht-M=ZD2V4%$AL#g+~?N!)MVD_S-s)1Y2;D1TDqvDoXdm_2b1DbC*K
z|C#1brKRmK9LLCt;%DJLb0311lU@2p;mY4=vRaJ}yrazwBi(D^cy9%zPJ%f{Q2IaA
z6i=-{0rTj|rkhH2I%2?M3%#>LG2hWK;p|BTIJrZg3KbD*<05%2pwyvZe8X2ycDdw>
zie2kmB%q4#*XZ@*J^N6izqZLA9SA^ZG&)6L>jJ@HpuF&y=LQA8CLs6z$Y}mE%YHH|
z47C4tvaRZloZj|p8Iq)ax12AtoLdqNWIb}jE(@4wbd$Cs(I*xkN@b>Z7yWdp`eJqq
zNqw{L>q()Kz{;PTD>wRNq|wn;3BReG!mT<9+fQ1kT0}i4{>81G#|M1^N$O6QY-O74
zEGn9a?Gvo7tu;KinfUgQbQuUwB<{ftM1UF@>yEO1JYTSBf-k<w(#93v+ylFzx@y(h
z02@*YbMGz}@lV4j8O!U53@xqmq?5=NWH;w$nFQZ4H|Fwj>##EbbL<D=8kvoTm(VGP
zEZ4Vtq6;5s--@>-%XqJ&$7&0UZNnEM7xc1gNZ7LoYOW4u`x-v+G=(h~fV0kd<zacL
z$+gq=FjGV0Gv>zKox_?I%vWQ+?a<AjZ<4y{#OEPV2{s9@0A*%SWbwXm6)sIs&n(-@
zQ^y%_=yC-eG6ZA$Z!9=1)hp)Bj%=SEI|Rh#c|||iL21JIM}xqC3eE`ZuKZH|YXy(4
z8ROHZAq5{#V3Pc=uR(?c;(xUjkNyD8KB4D_S?_DcHm2on5114Zx?EmNALOh26uLn<
zf&?_S7V;u&Nk_5DoJunVVc9(LbTM^WZ#Oq(!d?Z?&R`}y76M+5W61FSc)v*TRAIY4
zbMdv`uu@r<aka<HJ&tK!%cRld|4J4?EMn5%RqetE-2TAZ@ZhT8>H12)|MTg%P3ZM}
zoF(u%(o3?b+R3tb+?<CS!__s*r!(UWatvzg?;eHzG}3U!2s#C>x2e_=+7Lo+!ds5U
zhXHw<mq3Tg6a(;Uc=k^^JfFGWXi<$oJ5fSV?S4F_@m{OZhg_m%Hj|WBJ!n>V2cl_s
zmcR=#<LPlg<DYfdA5Nt$s$iY9o?mRKz)gti%m2p|9_dzS%i`#wS!@NfytZ1r(Cb5)
zVe4apLMFq;LZMNEQ=X}|Ii0$S#(}l7nlybK=kYOpt(S_iyT;1X#)pXQ9ckst=<j*V
za{V0exH=-Qb6utR@RXm0gx{_=r0Ka3ep~{hT}6oHA*Mwu=ajP5(^Z*NQj>%}J@{WN
zA67vX0#d)80p$BliC-ry@=JlI;_iRu{~v2_9hFt|wTo^*0YO2K5K&NCQb}n=B&3w?
zZlpUmg0z5ugmiaE$4g2{cc*lB=UM3Q`^GtEoOAEEW86LbW6QhWcds?qTys8iu4m1K
znr@95P1n76f4@H!Eq}bbgi|#7PG>LYI9R3jaMv3{i5LFTK8#+9QQNS@-?vs_pU!1*
z;W52<Gyuqqk~2D3Bt9ptmqea%t!qiB&alb<HY|9-Vm!sP%iJ+^0Hlpidn;_y5hd?j
z(M@f9JwPK&MV*t9e-nBU(!5$XG19Ab2;ifl0=p(7*E^P(3Q7dbO8g9YHioX)Q@TBm
zQBqSl_+Eq-vO`Ddk2ODxK9L+$C{&v!+Aj=qB$BcXRkDfz#;hprm)q3mKz!m88C?&o
zIwljozjM$Yo4BQ{5?xYIg$a#!nX$VEpH(4CvPOBU-)n$k4Oarm-0S9t$jwL(f7PTF
zvkZbr^OZ#!C|>NSDp~YCHY1$B<Q2<?_SHaOIs@T>tq`K|ZOiXl#Vp!eagotD(ujm)
zg*vN;sBHo;T-jx{^FE_{!I*_}%8gc%QC!;3JSIyml$iLy=6RKnkei#04!|UI@6oUP
z#BRm1mOhoEL|10KDR%vuVX39#mw}78%$;cQY4&2K6$gK2@zZB+mSpsG6f#E*l*K-&
z*Dlr08gZs?ymKi&#y%jvavYo!U;UiJo)MhST-5=Go?F=gf-m%1G_(Q^x@@3t3%xuJ
ztxTGjn5WPw2;78+Hz4;{L<PFvD6H?gkG-=S7p)nxFPySXZhqOG6C}9s-!tmb=YG7!
z=+G}};c=1U^v*d5dHj1^UWlqG3<wvA<I|VQbc#QlI(h8nb=gw{vWc(hdV+%W;TL5u
zlyDxhVu^e>xR-&ANRCfG_qRxH&uW>+-R6C1{?k4$|IM;Am<5^6Ojfr=_g+k;-QCL1
zETGd|*{Ln~O6h?9(n_8GaMPY?(Eec2gX($J(84D+PHcJmS{GQ!@$9gW8=SDhGkG#I
zFj7omqs+_AM0_M?d(kuQfu56M2rq>{VLQ*}IwRE`K#MhF4NuF`7?UNSnmH7CMAUrR
z7nlrUtSbC5&AWE|P}J05Rrw=t8lnEL9a6^$82JRj8yZLG9VYk}2IdkM`@3c-U2tiu
zb7?Ck#Dn!Q>Zx+nGnEjA$U6n)Njyp*6&uTUrkfLXZx&!$t^jPZDE=N;Ihyi%o00{^
zB73N^?86<159~``y5$~Y<-sz#PG76WHoPVoJUzGn5l8O>2y49jlcj-gGCTl*FZq7y
z;t%RaPI;jE5EdK5Z)rNzvAelb8%XYWrHW11VF_P*7E?_q!kuy&a34=mJ|}N~NJ_i$
z-HmK+7cVV!Ti`B%y&3%L#Q?WYGjzb^B9tRMvd%T4&8a-YqS*FxgxGfb9U~=`ILWyd
zf^KiU%B|Zs`hlz9*>DI|N^=}8d8L7wv03uu*3owM3zDo&GNYyDyrFI0(pNvDuI-)I
z>{OW{ckw1f<%^`(IV-kjk{nUAHW3=81%5Axu7nM2KiR~k;?BT)EPk6po?|tL{!Eri
z#?l|xuE~pkD1Ut9V6NIvF>0Hm@?p9KK@`oeug&K|u6W<6){4`e?(-S}_|0P4z3-$@
z5I|v+{%a-wMb!*q@Vv9=%h+>Q4SJ9+W|DBqY;emSI&DgRg*+<~_5`}6s{K(&P~#_&
z@tf@TiNz<9WnAzmbX7gQNc0MhKNL+{OfYaBHTVfuEz@v}@h|YgpYp5Qq<6I~;XLe;
z^TF3k1JO@!rWO<|B=l^FzmRNyQSDH^{MuYNx~`n+qHJfAfCHk)MvV)K&%~2oYq-yR
z@TkDo>W))2Pdj#zcD^1xOBA`ke=tWBSf$;2wO^di1Ydn(sVq}xmS4-OQZJ88CFhj-
zXN3sKpER&<d<5pae^vn}F)*=H^DBY9sQ+48VA^M!-TfjqCR0RFL3ZQ9PU+FB;yG{}
z&G`JJ?HD@r{ADO%-G@gGZkE>`2X!Q$PPSLsnTzCdR2Ttoy-4|@h1^>2l7MfLlY4O&
zhGU10s~F3o-lA-v6B8Yt_LY36S=QJ*>h^tbWp|lb7W+_z2dnU@T6=mr%lL)dlP#IN
zZWdF?oEWD?%~s;$^xN!(8nMrGSgJ<soG@?&6WBN*O|xiLn)QkM5xVW{8XI}cRh#PX
z85c-QicnswSJPC^Y#d{xoNu!5Lf_H}TwtT1bRMOCnDW=V8k4X2&hDaFH(`|ajf_9Z
zS^nC}?(X1P-)ircYrd2ATuc88bE5>cuFqrFXMKt8gsU}b-&WFV(Y~i1+*>U<xC?LV
z*`38-FJM4;)O5&57jVzo3{SrR%Ux*am7rub{W}K9Kpa9!I$+7m)F)bj#fSfyd_kTI
zyeZq;^xb{cwk3MLm}C$^5Bl^Pr`4@DJT+V#hQ;Nji^_J75mOubI`gpGaHsmO>wJb4
z{&RQ8VQwilAa5yWWT_ncbE+XbdA+yNbLNj0LhgGurL`mYcAT%j1b#icFjUqqms&3g
z>wKEhe0QLHTu+-o&V~~f0Y}H)sr+fK{~^>PA$4{pI%ibH;JmSrF*2c6*J37ifa7f~
z&pWa?YaTaXJ3JddOCm@iGyhLlJSIjFpZ1&8qd$XrkfyZ9V#)|r=(V`BJgzoI15w~f
zvuT^L<<56X<58x8O!4<Ox_cj#6Tk~JkneIhNLUv2cQiGJ8)LA0Fc`ii6WCAG#CaCL
z?8;~Qe!)6)!es9lf)vmZpTJW1lt!r@#a&LE*`iu=C*AJSwj*HRVcyBR5K&=h0b^HG
zSke~hj|5Zp;F6azsmDUFn$e@S?71%hl6vAR%Ef-mzr&KiFK}DgQS*fB=^tgPU-hmw
z;nT=aF*7{q)ut8CVHz8G)R8yTINUsMHSKKNr@5`H)==3bH2yoYCpbMVR`D4$$}`wT
zAiPpy9hu;Ea4C<7RMtp!`ZUXKHNW7)eV~SAW#_+;kyHA{Je=~_m5|eQB!L#AqG?jL
z%>bNX%^>NI3Vpl7Su7*G(;A<RtDAY7L4T%=AOnu-+5N_vAV%Ny_<ntiDk#&^agy|z
z73IUsW?*&IZ;5|mqigSzJm8FcMDC{PMhlhvP~A=#@*xnB4}e~etRI~<e#H%AZPZti
zB4TE_xTT^IaSIr%i*>>EUP^m~Bd?SI>of15V|jZdD52x@uqE^zXLO?FVF=c5@i=&z
zIBebNY&dp|XzizYSN8{gSCei-<)}vws1~(8WZOjsGjqB?A2RPH@PE>eruEMzg2EbA
z(#p50KRNsvSWB_>+}eI3N-k{e#>&to7Pi#v2hC!)Rsl9mXLX|C5_W>^Sj`PHC<p!e
zv@jwK2f1F)GiIuzrq1=y^?~BvJd8%~qZIoL#T^R)OU2=X9dqqM(H@qf5qBZRGQ`o)
z1f=N}T>@wGfOF|?cJCb@SD~if-I3{x$#4Rj{h-&uhrTDNvi=~!SZHyJ<lXA$+gWHP
zRUq6%P)~T2aiO>i<Y}KTVnqgz+l@JF$Ihx%9Ld@$tGYony^_6o1~3Ktcas_Zlyc|a
zymD$C`*TbB-oc!}LXL(kQ-o*I=WYLTe#{4n$YP+FtfD-0SDxP)C_azjs^Op-p<J+3
zuPt{Fe%Bfi;Jsu<%}x4>MY1YM)mhkRj;8z_rm144o%Ku+`sPz>0#;_E;J4pGrYy^&
zV^x?+6xQgu!svc|h2RiYE7oP<0bHb=(O;w6=0n`o2U155ffzuRZkJizr>icbyG7qd
zLn9Sqs?#Ml_y5LE3EFoegzzZ+i`h_?7`sHnF&w`{>VicZoaKaVbEAlhsr?$@+B2+?
zUTeC!)ldExhXwvWkysvI{TGR4H6aIPrAoVT`MzEtXu$;vbd(*tU8Au|zS^5LNQXH;
z!<}coZd>DA#K14A9_zw8n7jV-dR~zZqM-R=BKAO7dHaNc={tJGwYBqfFn@kJ(Nb%I
zs-ulki&}BXn>M)ZSkAttw$5BEqky5~?%m|{%y-Kw^@fL0v3wGBhFwy9F!y99YL!hB
zIrIF8GdY|5$h-SL?eyoyfA}$y>e-%#<GS?{9P|Y|y&|;msr*cZE<iP>LKc;BH8Wre
za}PfJjsAlCP5fhW$-of2G#V+@%`vAW-Zi11wF|@O_J9ZQ<y_gqNRZvpt2Xk$Bl5rz
z(jPcb&OUx{cY<%*e(u=w&_I?UQdRaa)?&nP6~e9_G4|+BG2H{tmJu%=Z9D((Q-r*d
zEp+0KcT=C`c_gIau>Yur67j4J(r)$CCxp>f*1WDPZpkh_akD--lYWA%=^BzPPIq<6
zM-ElnB&LYNbdG<@X4|g&etf^l-aI1!d26RdqimIcIUpFXn+1(vhMtpMYK0AwJ;b&`
z?sG{I1q&##Z-G<MT&uW%w~F3c@cH_v;`n9?)`+b^%lk9|_1sTiec}o9TY0MJ_q$TM
z-2C0<KMb539AFmdgvx1?M6MJ$O0gZ5#hg<W!PC6lj!M&qD$A!zFF_<I>VG<Y9<HGp
z6-`7}>ij$1Y?+$y?xudZV!iD-P1t~aNv(mTx6|k-TgkD135Lg0<F91^Z}%Ol+X&gH
zg~Wa!n+;N0At|H?kAgHcd#f+i`SIH@_Qmp+>K=cB+`lTWrqC5F>$cqejov%R-EJ2I
z9|Me=DNA=XKV?KJ7X>I=ad!MDiE=wE+P8@P0jDEn+=(0myQ`1B`@5^Mo)O!5)sxf&
zeO#d0x8cRR&DAzWqsy`ZEN93ki(?Hj<-&$W5nu1kD(H4JnjN3~-?$w6|IOuqUyUhF
z@BGI*fw~z&tA6-OAi=U3|Kc2<slAWKR_sUvn=J&d_|i)Fx;#~2+FpJ!_jfSyM>Upw
zy|2z%CJYcy8U6m%>GM8zmk-B1QlhdL*A|I=b9={JuH1B79vJ8ju^`1Z^UD79(0;Kt
z`*&*7N<#p;)F;&mUrEByBp3j@s<pA;9HdzT6=VIh9JiW6-BJBphw9f@U-OTXS(4Y!
z+BRiuz_!!iln_kkjp=ARl~LqCwIC6F;f=hto0RK*3|jkJgJzi&^?k1CW#KTrkKUgq
z6jFK)pW<8mBV#n05AQTY7NI(ek^k{3(@I@LdMZYK(GBl?(Ne+HJU(JTA=19O=#8^B
z&pMm#*NVsPmI&{bxEE%G43VhZC;q-*E~wD8jO)o}F8DX%(g8FLR$AIS*Sq*X>c(8z
zHW2nb{oOsXuuFqQQggr2{)_r1k(2!*Rv?lmTwTtJAXfmYl8oB58Csov*RGXj_K52g
z)`j1>O?qrZ0o%K*+<8w8D<&q@Q0UoQnh^yEi_0UavVX~4=7n?a7isfxnj5a${)lrc
z{jL$8UgccIPOXj#+4o3JD@?6YuD;&d)jHg+VMIhCQOQ$*RIxu09sAM<+w=DR!SfvM
z?(ze5qjG#Iv}LSO7ptfeuUs<(A%qLdCBI=t{sME|uiWM17kFTPEeAH*wxACTJt&bU
z1&Aej`tK5->p-W7|CD^FgGyUsKw;R>ocVO<JqU0kv>sylJOEeO!t^#HQCYskz6?i~
z)IF|>%hfmb)BV|kXti}VqXXW!yOqT4VPNzFS7b1Yv<Y;*>Z&r#x?!iaC8S2a5LNTU
zV{d$?A1;j48>QHNYJl1@8Z79bhUG&Znl9<j#L)K2AuUWlD$-unaTk*%>|C$-d?EJg
zWGG%lGBlYW!5lTES;y*xwFrp_>B{M6(y%59<b=ohsO9_uUE=4Nnbg~0LdXn5l+Wj8
zad!o1`u*rDX>AI)Je+E?;uBc-l)BpAXrF6B&il1R)x-Kjk|WyTF_{Fp=^DurQj<sQ
zC<EcRsQ=Y8%;;umDWo4exU|b*&&ki1fr!Vc|HG|3!b|;@0Vn8W!nV36xVOpT%n%DH
z4r@=U6@m1+o>Knig<#bir~0-ax+|lLwOsz#BDZB}E4=lLpEKinVce1=a{qE-pKhGK
z5p`$3n2lJU(!8>F+SuyK%EWf|6_Q2I<e(DKngM?%<euQB%E?W>G!oM;H2&-Fu+3X|
zUNTW5P_CU-YNZ+CR|dmxA|)H(vIJJF-_*}v%uV<AtX0`7JTwVpNBGG4pDwPhofD+$
z!L1#%<Mbi13#&tp`^7q-KPD`dostR4v{B(nwI-77UvF_jlPnyz=4pLm{dd}FSdhOS
zpW;Mob`&oLcM|hmlH4qGlL@}4NXC`$)f%Z!A&vu8kj^iSZ3WNUl8A%;QWc;?=`05x
zmlN)Lr<ZOyL)r2<m6IRuMV-^6!?hCwi%+dtf3bF|ap;}6@WphYLb!mY0N5JVy%jHh
zh=c#o6JTrLO}RTcIYI6ni*q%@%a5nO0NoTY>_o2+v?Oc6zUghnthgNt5c16jVELct
zW_kYKf0Bry8%Dt2zklhy2UcSL-2yfOjOu<vNBH-d4GLHr{?B9gtB?S;1+?Y>ty#sC
z#ouz)CIvppKW=$x0(PQAk`UR$A-<u1-<$FUQv_Y&BPjI~pw=aaDm?4kuvU{<4BHg0
zkFovm673iLMEvg^53p}Y(&=QO`fk;Eh44W)tPP1vUI&-6m8n#R^}qPnE-K6$T3wdm
zG5)|1qr!?MG6co)t!=kXP*;fld!+>9P48s-lB%YYFk?+iIjaYoSW*8rJt^y^f4#DU
zpZ4ymxu+7vIYu%6=S$Q#tNiJYQR~I}8wsq>eV9D&!#C~fE%^+DVA%c)(x$l33TS?R
z^U4480WGamIJVkr^2Ll6U#j$%|M|7se}63*(J@!~Slbtw*nslxA^X2>4q*N50NBWQ
zcGmyBLEs>bGut+;@WJ>$2FDjFV)G(m_xl2W@?Zu(i4W*kcDXfhnPa8zxtG$N<A_#k
z^v6J=H{D5K`umpr%2j_(BmZ;~`{Fi<K5-9$4^cGuI~HsIhWm`?hOEJwhI31|)vtKI
zDQxQge*{T{rTbswDZl~!A2i}$|Nj@2_}?^w{s29x&0f^}I=4Z3;LnETiJIGc@<r<B
zUqs(Ggnl~QU_I3>H@o}Fb&wa%f!xjWL~3eY*%ih76QG&MP4lT_irX&NuFAa46s|qv
z*flb`>_xE=*%FR-LSY|5LxVI)pBMl1E~)QE16hr_%f{Vmi`%4LLQiKc+og4t3biZ0
z8Q7h3=hp6<vX>|1uiNmJm+n?hFaxR(HKOJ}L-LPlJs&#VEwp{BUQKToyHn!(oytqh
zSJi{V{#$U|%l^CjIlK%^X1^JS+o^I&Q(y`l3l2E#7L<zI$>Z@?oXKCxL(wH*5`MX?
zV!+eL?p2!PGcJlj0jCTM)Q!!w3vKhOhV7P8w!P`&hJJsUL|dYAgY{66W>Z$mr5;hG
zW$Ac8AH*Dt??}}*nydXC2t3NipM?#%tQ~%HU~_C`v7U)U^Zr`4lE<fp`RruM1UeI%
zyiNH=`sS0ai2nPOgB_ImY=N$#$A|YR&AVktv$uICj1k&4V4}3q{t+yFdDj~$AVyYf
zC+5N5A(lX{zW2^>9E58#ZpSW&v8n6Cgk*bF_E?iu*UNz7Is9Z%qlaBDfY_m@I2NQz
ziBHNdJojPT`9)-CrRZD$#eqrGykG-t%rq)@aK($uyuUy#vP~Nx6?_M_rOLRjBjjQ#
z8$*C#s8{54@%f%@nq|`B_Q^h_zrI^7yR*9YoNn>q6ibVI*m7TUn|=3J%<oy-R<1UM
zIqkDQoii1Q;Exej^BcGLn|V%xX`z#IPp{|3&tOn)U4M!=Qu7=ALHK4+qe2%~v9Q6B
z(}irV__xgaxr~GK?b250+M`!*{>_I6K%rPoYfAFCXq>QDN}K@0>Fm%)U+TS$qN&9=
z4+WG<=NG8U%FX{KI^pTPb9tb7VVe?ab{exH0c``?dmEM7Rj}Dy<9?V&dxBXZTw7Ha
zm=`x>whQD6irru{iBy@}35+V+$sb?*`4)7C^J+IOzmAlnSA2#WHRrf7Kn-opDWUec
z>1crVN7ntd&t;yrHJ8|Hog{H%)@4KT(e%!iZ?S?sU;eYI$UvWyR<m0)5xv?Z$*;qw
zohrLjBiIKI4lSj%EPgV<?b1S}oUvVRuV%w?>_GL4fFjw&Eau8@_N5c0%l>VD(%2I=
zAyg7|3tn&jDecR)BT$*n%2#Di8v{2@6>HG0;%F!u6q64Q^L|Biy;L2QsJw<yjp?z?
zq^Q1^wyPYTNR!1BGM%NrJ`+@m;;#Hl;|FQj18R1~YT-xp3{=N(078YV`im5!eQ}?2
zvcS*u7ZGt9nm?V6hoTSk&EFyBYpmlmVk2KgtN+Kcp#sY$A!QIFj`Mvr?mJXFvTC65
z)#}YT+-*0?i}$8JSl)oK>ti9B!a_F3_ZHr|n!@IV(^-LK2FI5h6n4>C(_C5*^yAxk
z*hKOB(DONXNA7mCoYzbKscBuyu<clO#*UBs6(~a0Gu$H|qi8PMo+w=oY=EO7c=T>T
zdD162+}5K+SCEB*(gh|Wx~F^$G=4LDYD3hm?q$LC=}2UY>-^=oUBGeCbkKhs+f6R7
z_h8WU7)zst-Q~xwrrqJ~9V@M<f~Um+lkkC^-YB+l_^)zOT3rqMaTib+4bpTqQhn-V
z(A;rbYUcY<9QL|s^qBn6<zd?MoBC^$!|O=y64tmVecW{XZ_auzQu!Yb+m5<88;jfD
zRZRgEk47TXZc(=J4{I8|tOTb81eAPUc>;m8zyXdlzP;}7`s%Z1-$4C``*iaK!2aLk
zs*CD*E*1i|ptp#g5&d&w&(S$p1q_}1brD5v<NaTqO+S_|E1O-=0ACjZs%6^ue8qMx
zfBG>RAM7p8HW@f9co+vgC9v2+Z+nIY<zTR4;9x`9f!&Gad--wMb=Icrfbp?MsQ=t^
z+CLI=DOI?H)xDMS>QDskLj>c^s1YpT=9)1dJpDV!XFQEK4qwtUY$Q%{%~9H$-cuz~
zCXId|g6Pd)LtJja>l)nqp6J<1oELd;bR|5_pbqTa$sG)&FJbvOY)2I;5+q0PJ$WW5
zHB*qxEvl8kg)xMHJe&E|*Wp>5?Np0pgPw?N0`D%|1QK$XZvc0d{tQA*99iA|d3vDV
zb35)*F-0%c;Pv16K0y3e-vX?4u3cuc&3B%ek@zS^6Y9^}BU3y-i>Mv3uVQ9v@SG4J
zF;Ej#?QPHxqo#2;cxzbPQMVLK=#v?cN)#C!ieB;QwCPQ<D<tv_GaC{>Ik{q4C>Sx@
zMScy((qD|t>0UgG#!Kzm82dOTqi3-~KmE*2{hv$Gg81=&#MMf5uM;C4d#l^~3^5Q`
z=%oJ2br_TI+O;lbl`)pe;#TLKJNk&t)aBwJx}+EvminWSEiQvt0<61@t)>_X<q(+2
z&3WiH#}!LAA#d$uL@9FAf#Z5e5YL_A0S}ua-vP1w^UaTt!#@g_s8P5Om5Zj<89%=X
zNp3wLm-@B0jiK@Vymjdk<7%5k6&h^wrD^E1*Fhth5h~=Z7S=7fas07i@nrG^`4OLS
zrbrC&F6pWdA!3@ZuN>z|@LK>2!orjBN`Jl(b;@pQXWaV-E<j2Lz8Dl>>EsV$hi&AS
zcV#|%s)vF#W{Ihw%b_PqmPMp9a(HbUD}_UP`!-|Mk3DF_<-gug1f4WyRKUTD&W<vX
zDOnwsxwYq>169Hi`uQLM|7`c(auVeaK-JtGdqEskkPv<`w5Sx;O9g>Cbq{hVd5Qyg
z`gB)scDglwg%v<4$M9zkMoyv9iZhGtfsq$(I_ebRR5@Qjq122n4#m${F@$82Ukl_g
z)_L+tgin0Xr-I&?tZo-g?;g)@Lf=Tp&-Mg1b6G1BXBs{B^Nt$$Z<2w}zVHQ`{e)`M
z(iRajmu6a8e=wP<@2kBGZr$HPY{zriw(A$`85l&u;o(!x+cm0NvP6Tf`xf)(8RLLR
zzc?v-KOu%DI2_Z|4EAZZt|;p;wB)emj=xy0taEcTd*AsbUwv9m>!PkFT90gc*dQWM
zg8?k;xgm0bo&<`^`Y#!n;C?474>smNnA~MgX*RVybt}&Z>Ug$AGZ3V~l70_)HY=PZ
zV!0KDozykb%9W?wynGd6lgD3hSF*~nT`afWi+r2Q<ab^-0CuBK^BJ8?w;v>RTpvd9
z*)(y#qrByg6%&kv&OO})4o63UHkfVmq#Z=fN0fNfLkYZfx5<sXyFQX?K?%9I!?sm6
z`q+)HS7;rMWE;md43KY<^Fwi*#-ckicVSu&o}<62zGA&7gfZ*NH<MFPyLODtVs7WF
z4G8~C)jI;#)+{$0@Ynuc9g$4eJ3jp1JILD~@#cdtadd@OjoejFMk&_i_0!$(MJ3Ys
zzRSZ8o)f|~6`Sn|B=Ia>`u)LPv^9sNQ?{&)B44V_b7KqBS+lMYfBBa$t~Nh-HomQ`
zE$8D_k0^LDQwgO!!`b6I8^%3q)-Av7FY~?mdBcx>>Zz*!icL9Qstg!5HP2Qqv9$Z8
zS=ZD0h~eDW>1}Y&&BoZSw<P8z^leyfB!Bik23K1*FFo6iB|2{E$Af4@zHjAjS)Q*v
zk>gELngR9elqV@57>`1bwMI{<dO<U8Y~dimxR9~)ac>PT*7fRd=FuOI-5`+<=mIJB
zsEPS34*f{~>|-n<q%7pT?)toLczC!E&v~<SBswN0z~6t`?R*oyT-Sh4J2a$aWOA~z
zr=8aG3e7&TmZ|m@Gr74*iGDc1f>g0AXFOQhbMx3nVaD=k>MGayHmSb4QE7?3ukNjm
zi?u&)prr+8(>U`c_6R;MMDgzWT(Cozl|}lNyE5$uvD-=HHNKP&tm<GBE^fO^;QUEl
ztV+pstq>i;mw$e4MIxe+UO!;&a&t>(=6{0$)p3s!e4hydJ9VE7jsc8xYj|`fHbYV_
z0D?95D}xei<^X5_&KG)Jt!klSDQL2%fNwY&H!;JfwZb1VE!+>k%1v4EMeQjap~5-f
zLgTSGkm=Om8yK+#$KMpO{Bv17UTvf#UwgeU3Cql9@?~lB<V|dvjZIAG=;-vA$AaAs
z$nC1ta)$V>FZQK)cAJvk`jEKnzzr1@vrQupa<e+_KzH7M%+~K+-7UwX+Q@%_4A7Et
zM1FilOJFe|%QMEHLBrwHgjCpEs?{&3%P8dJl8R%E*7Rfog5^*idKXq)7_TNiZ*N5b
z<lJ|dr;|qPKgb01p6`+7cRT25(22eT_)_@VHBFYTL+l%Eb2W)iK~^iQHnV8u@ElY=
z{Vbn@TU~)jyKcSXP8~tXzyGGDQH&Q+YN-@N%l|R@DdEH1=_v*A4T>XfgO=Mxlh&L~
z3WP>*RZ81d($(@nPR<_NZBc0~B(^8rH408+&(AcJGV#BNDG|P>{eRfw3IE;Xd9Tj#
z?bWAc1DbM$qi=ZobqVaBqGO_xaK?CLi`!}zt;!qL@b@`E{jK!_P=BNIeR%L6^M5S-
zsy9;)k*d4&oi7hZ<Y<R3k7ur1h(W4uqsi|9^unM!_NsXC0WK~E21a&vwqzVvso7MO
zr7`?P1gu9Dz?mEHu%oDd;Qj#gl;dS*EF9K|IMrQ0)wj;*LV*gc{^Z_0j4PI5)<)~Z
zVCQx{zPNe_U7U0$$6(-kNLnQ|U(|Gg;?0|_O9oBvH73-`H?_IXBO2d&>#JhFn(05^
zRdUViTRI5!d;d{AbS|lx2q%B&`XrVFI#KPu#O(KGbQFU>di2eqx_6i^d2tqA&Qwk6
zb2_BS7+975h&AKK2cw<URNghn*cNziJokJ!5*tA7IYNEL-+hnk#`n7oy1J%<TFtv}
zr~rxLU*d=&{t;UQih7Gh&owH_ZS1_y0#WXv<=a$MQv`R2%KPn??D-HFP~~_Y%M$n5
zPQexti<$Gf9{ePIY<37y|H{N-joy0?#eyR-<5eSyp2`rh3U#$heXF%G=*~HxsCrWS
z_HgO#;Ydc>eW)D2nXgv2-fa()Ib1^{dT(jhE>s)D4I8<&@$+h6)YsDo7g>aMK72a8
z%hp(Lf;ItC9RMfa3MzO$xS)<X>l5z*zvNIWETgb<){}}E+uxgxxC*SZ)>n1RzM!6n
zfo2I2(H%02Os2p6s!jh|X+!0zm&hmlSG&Rb#iyss3CE}2?VCI{g)J>YP!yYqeDAJO
zvA&#~T;aB4{=!4if&l}|Q}zVod-sBsm~poKVj5>!6F#jt>Ny7vx$<`R9m7=zHTj!_
z_cWB^59@zv?R^RC2dKksdGzo}$GCVtd)WGHc(=k96CJ5p`s?7SiShYc?)bNNAC0|u
z-UMc&k6EWsUnd>6F`%>@{N;Y8WYJ4pwx<{UoY*$4({(UaFkPWowA>UgNH{w}5Uz2b
zk;#Q7t4MUqhwW?HE_c9xnV(gk@hIM3SG>Fy&pA{~Nh@m{T0b~90u|aaOa#TOWyOl+
zkV()W?bM5mh>R~-!I5ddXGXtYH+Xdy=at<!LMo-_?SsBsv;jT-_x4$J8#^mFSN$3A
zrwDD2%zT2ip31i{Nrjll#@)B8J5I{`Etk>kJI$+;%lxVariddTS60A9WWO#xRYe24
zvZVjKm_MCSjOSuft4vE4GdPM?b1VJA&+^XBz{tfe_93(tOH^K0HX_XYvE<&r$eW+z
zHB@KN6EBk?Ib}CtYQ5B{CES-nr~l$1^pu^QedU#NYE6x+xXP2j2?Hv*HF{`gDXFDc
zqQ<171je+|CXeNCE=pP+!=?r4y=P_&lJxK8tvH*)?ZEB@EdF=bmlis)stzWW+IV?q
zM{AQA3d^<Dxr=c+UEtuN4i+&a_lK``oD-{LX4c=2h-^oPg~rg7J6<3diGv62z<lX1
zqub^$^?hiSSRr7brJ>vOu9A29h5++DCKVT+0R)&QM;uX82)aP914laMbcR}4z&^WR
z{b!AH^*c)ZNnUD7C5Ah?PQ-QHx;3_`ZHAT^O(J01K{R*`;u1mo6>L5fiFUL!Idqm{
zz}g@qnX;X-Ku2COpjkG!8)js#CVKLY5wP_;FiSPmAFj6C0_lse4ry;Dw4d{BaFJGr
zZnJ+v9k0V%zJP#R?utL2?lgVwp|pcjcs7WZ^k|m!zV!S)0Un$kh4fzx`r;79!Lwm!
zSjC)@ni^iD(;(7&nK5t&N*Y_zV<v(sRP%qUEZQM_m{_fV@)%&wVGV=KW}5sy*a;f>
zc7dv0Kb_Vfod%=C>gp++V+|;m>kT)n^OZe$9ZaIDn5bZP_Ec=a=~#qv&ct<aiUcO}
z0uLh^#1mipSudf7(1}64fMe&^=2lJ!#aQqQb*ZS!LNJTwbepAym(|3ox|u}VmvB29
zqPIv8<qevWemE?5QJ}kxUxqU|4g=qrG)=CY9S!U|LA}|6aNUF~8NRynLI|UVTBf6u
zwH_c|*0up`WAXcPd!7L>&PWS-A`e}^fW5xto1IYI45?xD^?1C&B{u3qhApzZxeVfb
z+sJ5*<jr-)uttTM=@=Y3qAptr0{`S+S1uW8ET%zl!aB11vV2-g8!E-94kXE@0)fS!
zD9M^PA7Q-nU+afh2x9aqoI-x>;MT@JnnztmG&|Jdp$j%=#sY74&K<`C7SzRP@a8y9
zS=rN29Q>0<@BN8)Tp~)c=>0K-L!eXF*EBb)REq0&>+D@#h()Dp=gFtOi>A=jU$^?E
zqo>a#si6<7GCDs;9&T*Mq4GrnD8>+zq#Tuo(GEM=rA|-k14cQ>)+re?o@aW5lTovG
z2f_P_&OSrMm&sSd$oHU%Ns9CQk@ZcDsKrWxWP~bX#*Wj|_F8n>o+A{bo!=gU-UnQ5
z>r)l4WV-pj8m<;)7sUB!MfE<CD61`ab~^#f)n1X}T^L?o$ewvuFjD_K1zNRA6paxF
z0uCgwU-|Jo+R@<+*~=aB8wrCErV&>cdqOhU=sNZ8KG4}oH9CEo(JkY(pG|&0`cy|2
zqK!9+sPB+fb#0)yU$UhidZCbDf>OM;<CoA2I?-1Q&M4-W85|d<lFoby<}0`TR=mCW
zXN-c=ed<(}b{!tpOX<jPxzAKP?hnB^3LK{6-(zC9&$O*7C$$T|Oh^CJ*_P2Fbr>!p
zTGL}0oU+p>?NidTkS*ixWaPg&oiKSyz0d6Wm}uT!L4SctRSK(hi;!W-S{~nf3~4>R
zI_kLFcAQsNErX6`=`kG(E``y{<V4TH5pPN#*I}Gd>wpJCR%T$eWm)3Xvj#S?#y+EQ
z67rznO&xpC`-5e!S*wcfr^I;aBI-vC!e|H_@|UZ->t%P`Dktx+wUp+MFJP9@2#?K<
zSv6xpU=NT-$%}xUe=80C<Ae8RiHV8%FLoLyCnp1mIf7~*LQPFg;A91#MjV@1-emdU
zr{^W&RKyuLCvI}D^`+s?aU}>kQ$KEI@9^`UeqR%)|FAgVVl^mT!<@cb?0caXA`TB!
zpK5+2cM@J=^hneek=OC2X#KW`q{ap{uHPv6rC+ma@dgF1pZM@`)a$1&G)zX__crE^
z4iuGE9Cax`h6S8!tM$<1X=?wh*H&SBt#U4(<cgz^KN8qR_ym?O$R9bp-kfq&l~K6j
zzp}>_LBIc$o-78L41K)n&oseEneuu?iQKrGGNJ#dKYz}CVNB_!d%JMXaGG3enqEmM
z7h^?0@=!KjUNX&v>~{$z*&|-eXEXegUAm#roi9zZ>!+<e%**kU`?bpstFFuO&ReC`
z?P8U3^pcyr&i=4zDdxRRas1D(6KOw6Bq=bZQcn#vzZ13>T`idDu*LF!)2hcPoh6&}
z;pcOP9*??UTt+@8euQk%w*F#m%=O`PdpFtDGSm4hE}P8oWeQx;w{KB!#>U6z>@TrI
zlRgZNjOe&<)3VnJHZAX;USJ)L+>r&4{M4nkwA8jkt^pY?jP>eREOW?fp8qpc{A6q(
zrk;?105~mueSOJjR#6=pW##cNcOLi}9<B|&d-v{4784!a!pe#)xu}Ra(0FliQE*{S
zyMAg=pK7h#2a<|`q;_LGFEQHJ6`s_&@)lIT>8n`%wAHejlq>Kt?YnPv?B=IWtlR6%
z@!At#J+gh^aB2BBi4H|SBT7q;6izPcsGfsz_e1g^Xds)jc*Pa7vZ0%!#(1A5VdI;*
zj!vWg`O2gxVqW99tEK5%8K1<hxhdRtDbkwFPX;&nQz8_UooPyEmpuK^(ro)tyA*a;
zKaIT2O=-OgL=1jX?#{L-azU}r)Z;H)Rbt(Jr{S>t<H$-@aIv%V@$Qyz5EGxdDetUu
z>xq_kV9=%M*;<9TFRzn5UfVu-?;8-m4Jy)GRmM{8xbP)hzH!I$&KrAnAKhYc<XM({
z&ByBke47`grKJyNu)b36NDY;eMl}yO<51yY@bd7mVo>6WIy!QKP06Qa;Gy^--%PPc
zm@z9WYotH}-fbovqRU82!wE@9_z}5RD<gZEm>c0M<W<2@3bwW=R$SywhfFZ>)8rv!
zWaU0m+ZPS33(J>fSBm#d5-&|8+se$y#qXb*obJYIcb(#lx0~o`Y}LFGtf_t?cjqEr
z?D+arMLtDZR8U@3(%wEKvbeW(tSqA(*4dj1eVbLK@v`3h=*g7CO96wmDV02K#ur3b
zK0?P9BvXxJlb<1Q)wBFrm6(0|_hQ2c3k#c{oAdDW)X~;f6dlP|JIoRiL_xz67ZItj
zTJ)Etm75YM$j+{E+@}J&&@nMh%HQ^NBygrp^yoe4+vk&vdmMeCCZX93Kf#?nOFZDi
z!H*c>9OJ?biBvqPc)$0vs4d)wL|(-Aw@U8FgUTUX8GhBx%}*0;4NI2c3zM63zAyX`
zI@bJj=~LJ1an?v%FoTK(Z-L9e4`;!Jbz{Pp)hd|@#ZLNH24#R9xkOaXGV!aT_4Z9a
zposhwiM>xk1?n{(9v*OeN5@E>N~?N0S@?U<qXnx|F{YZ-OU*Jr$5&V3_x15)*)@tw
zmI~53o_v9)+>UWE{Z`%>#mVOP=FjwIb~0p9k{uCP+H#RDz#ok3i4VUnnqOA?X1?h=
zDG1=~8sV`_@pXMju+q*KD-MeY4)bi4n;$<JpOTxetuprF;e|`9SOUBMyJ=fQ?`QoC
z8$50Y1qChlB*4pY_5=*d2i@G;sTmp0PEOrXFB>P5J|^UAM<ki@#xV4Sb1~!F?$M3u
z2@bY3w5*F_5zQ@%2--3?|1t3z*AT7#DEpCGMovaSHSKh-S}1MLrX^*VRLJBqudZCg
zP(siT<{)9WHb0%43hfw66I}R1u`zL2V2c)zlO{Mg*48p6X)Q-W*|(WY>q%y0Yw#$}
zG8pr(5b39i!|R<L9CFnVM~jgf1Q=e&TT|9nR?)94=hlXD61?qySB~y$Y_Go)e20G+
zGI8Dcd&H8oHdXnoxo&`VeNpgW8nZT=Ity++SetBaj_|dkSWdq_Cfvr8Gri`G@tu5P
z+H$<B>m^w>EWj7+E2+l=c*Mk&*HzzQ6vh`2loQ+e1q=ztt0oPPlZDc(cBhG*nfQCn
z)@9Ec^?aB|mw)3-RY=KWXcmtJ&{CwaT_4WLR3`79%ls(E%<;|qxLuC=aQw$+YX$sY
zO0l)-imIM*lav6Klwl|~E^fIy&SazFtB!osbzE}x!A1K;QI#XN%(*Q-nlsm{ab*tL
z@Ws-->i+Q#r8N3H*Sz56is@_7r(g%hg0FkCv$^a@2jPX)=Fl?N@qyif)vIgx&EhWc
zyCQEPvNKHwj-kh&uzUvKjDVA1XUef)6TT}>+qFbS63Jh{c8L;UhiylIDhZpB3x(km
zYkL00<s|2oV`F!6Ro@Az`@<skBfdCOWfu{Y3anQaXIll&9p56W^2T||t9H-R8nxGi
zdhFH3izP7EQ|Ag(4-ulFRgj1?7=B>Ax{Ng0#@au$F3t)4ZA}7?Y}M^0CRVHeULNYe
zOJggV*~arZ^EbWCa1p~%!ZH#3-vj=H@hpLieYm=!+{v8Bb$c|BEKS;xt$f2ce@WMh
z*kt;oHa?^gVe}n)bBoJfd^UZ+o)|AsIG>QD@+x6<#tu2>wvMOJEkw>p!nJX4-ex#Z
z+pibxFy$%75JwkDA6SG-XWVJ?x4ae+p-#pYvcV*sEWOv7t#UG(ANRy`IH5B)Gc0nb
zJt*BsQhQm&(x7(XdN`4B&C9c@@**!`q+!ql3`LrhS>FR3AH#Mi!L4?V>B`m9w{{vX
zt}EDY-{E8M)-p6Gdyqm(_*aD$St)j;U_zD8L%%F}sjzoWFOTNwD)oZtD8mmyhvE?_
zfe25DL?1%|4W^-S+ecJBPS^+b(=a7J=(T0Itmpz2?!i935B64eO+F$<<UT3q$uj?*
zv~>sYe(QP|WVm~dAIVYFU8mo2e<mI6TJ-!gg{4MElNZ8?kr($RT1t<&!xz>(euk9q
z$3c4aL(OUXql8?IYG0<!ePvtA!u4f3x4httu)Lu3>pBGwR9ovQc(LKKziLnU$7s+@
zj{?v0K?DNf>zjUehtsiGWW0Cu0>p=R*E)}M4(0)80$d$wIz_BJp6^Xk7fY=)6jbQc
zoxAQbF)+x{hO07;j1e)xU0YGc#>NX)m5TK2_D_2mvvYFNq@Uw(a&jsu{b`-S5@C!~
zkeByw)+Pxy*ix)N(PdVWlf%D(7~|t*=K-?y^zX&QOifL_e%ZkT3{meyMepG3I`F8P
z@XlM{pb82y9w8Gtlk*Q?{cR5<bnq!p0Q=v)X41EM80Fu+B1EzC?+zk>?(qD(b-(=@
z`M;h;`d0scdUX3U+Jiq9ud=Vbu>3&i;==vn>vl2pcCt{*g`4NDWvjP|k>;!SY5kKg
z|MTmwF<*F*{JXD5e)<_k`*;5n5}8H%cS{f9I{fEhO!r&=)3g7-J*ug<1<j6y#>5~j
z;6J~<eft&__0YN%89F=LET6%B@W9&I`Zn_CgoK$AqXDOb6|JpA=(Wb}+qV(++uiN$
zP8TPZz&3e$c%)0jEUm2>Z&g5vq9D~b1|F}nUFWjj%B_(j3rUT+%??2%n4*GjWH%-&
zStu#fKn_r)+|2dtz;Qm1!+|G$bsoMo@D>?6ojj~fK2Mo4vZLz>I<z8|^dVQVa4wkt
z>M2=>h^&T2LUgp#=6I>&{!-v0w)V+Bq}LsxVPOjSs#zasa&vQIV`KOB_xJbqy5qQy
z4@WePH^v6QL+)l7l5wFT6kHtP-@k(>U}0h5en?7+j?mu=;7Wr8SC|R_cVen2r1W0T
z)uz_g<;BGx+!)Y{`hvQvvsO~rW^K?2AWv@lt;rhaV>D71aJU=@m}I4;wYOHF#M=OX
zt`FzQ(BS>~^9P7$y4LkPUk#z|dMKAATV*^%n<Z<!1tGnbQB#{b+MHlvW{%GOsoxb1
zS{%>oY&KI1yS4^BiD5V0+uaofwTH?~W1i37vVMb%O%)3Iyzm=GSXkH^4Gc}_`!wNT
zlEqzz!oor%B&5e!K8=5>ug=yRz3+v+Pit#yTf({zDSwNIkXKL`Nf8d-*|9J&F(E4|
zEcEmDUmMJl5fyC%J?MP$6uJ<WmhMk}7cdQ9U&3F$d`VxM1kGORia9zuN`S98o$t2{
z3|O~?Qu2^&m(RG-F*34iRcn%n%r@YG0UXX(TVGf>IGQkz;U<83N5+Z`OpJ|D?><c2
zg{FakcHrw{`11UiLqQWt%<%9)CIfx@$OLBi^z^jg_k;xV_@B`0ND!a>`XvC)|Cybg
z1=V_NhPCSAeZ0N1dm`-Z?Xj`2<Y~j7KYvc;kKc*bQ}6lw)9BBn>S`CzDCQrXlZB9b
zEu&hMpr9uj_Are-D2Zxk$1W@gX+8vBCTs8Pl%);Nc6m(|R+wR6%NTSUTEfM-xf=ND
z&bJUCJvnY$R8-`fsZj^JJzBgxI~0?WGEhm<0jjvk`10jTN=m(r(ZZypq$q1l%{QY{
zoE#kGpuzjgJqG&vdk#D>d>yIbMl^B>5&UT;)Bwyn3rWfDXf~5{$+)$pCHW5@?vs*g
zsH;x{l~_)Irzeq+-L)_<G5ti&8E|vINK}0ex_5`Cq~O;t_?PKKnWLj)S1jkiO}Dg0
z&98y3jnz1p8yC`*#W%-tS_^%Bs5K0fO%M6A=4yZ13|Qm=Lhp&^&7mcIsoms5>NYy~
zSPgP89vU9T4LLI22jeIxtDs=u9aUelbC;)t+)JQgn)rCPl@#+9vb7Hi<gF!oUUzj^
z;O-8)S5#C?A?(d3D=fi3Tt8hP4F@NJda?`fc)8gW8CjtsP<?y5DKH76BO>&-dZBvF
z4cAVAqz@^R!^08rpeQ33JH<|Kklp6EEkI%A(@u0RUi9AdreSod5=@7dmKN_XU(me-
zn3xJOGc$kw{ORrOt>JdsJc$G&zs3Z!w%qN~iH<IZQY-?@!2;#dH6XRCllkEF_4P<5
z4Y&Oz32QWT(CoT~p^*`j@5RpB4%ipPz)+V&g82{h8Wk0__HbBb35>OJBY2%}OCWLU
z{O(|f2AGOuzc4eQcsFjlji<@KfB)|7{UG<p7l-^;V<SF{(cHs-{i(JWN@N=)D`I0~
zd$gDf2#+z(s9rGepkQyNfz}olzGPvs?oSnc6w}||PbD7pS};eU01F-cg|H3OTLhLh
zh6V=>3lv7OWT?q#85!j@Mjk=&?BU_z!KD0D$w(-uhMyk2ve0yKxd2VIAV5(hpr!^-
z($NjTG)zoP)MSK&ge*<B(!Fkxj0b$6IoR3hrTH#G0gTH{oab6tR>m%D2qi|kySr~}
zZ1`Z48h;1hco342L7jh&6t6Fk^sQqN1kJQ~7=aS^?%fmAH8BY&epp{*v)WI|#&&pi
zxXx62lI+B(GW4sbr$^lK^Xk*jK1aBZ9>LjsTEU`{l0g9hm{?drets{-WTC{mLY>xl
z9!KVvFL%Qg)c;sr@BfyQmd3@g-JYIuNdngQ#HBVkSPCe3a{%F37!&PAUVhXqG)+rX
ztjAH`zF91`<7dfcE0a}aXRiUf;=;#IPv2Sn=kuLeBH-G3^MDg*XlNK48>532@^uTo
zhjlC>>^8Yv&q`iDMaRbGuy==gJ+fpnaQlGK>;Y3_bW}tT^veh9$vDEEDYC;A_>UV?
z_Ap(cK>f{|H~-AAPMArmw2`W6tSTeCnidH}1)eG4p8Sjk>JGmX7KZoQs0o?W#U&-}
zY;76n=z!UV!JOKftlZk#S_;B~ly`n%GEIU>TmKnl#`Sm#oB!e|L4er%_jnk$gIi->
z>IjL5_*SCaf;<@%e`@w6zXLH+oztOlB%@kUaj_orkBp4vuS6`JPaZ`f-#Br@(DL$Q
zAbj8$To>EJ+uPrR82|>+XdvCF;1e1ol>x5~WMC69Q#^b24182kF*-3Z(bKct#W8cR
zyZh4K5bAwILPBCPl79)JoP6~fr|oH81A}D{CrHMy2i9PaaPb65XR+Gf{SRX3MFvXy
z?_Tr&0$KhSbOZl?c=rGGQ5$hlQPGbdKLRBO0>($bb#-;6)AAKKAb<&CE8lujJ;%ei
zaa<O_+3Rg7#uE_39A*m^uQH?(*tg8Y#Ke>o%qPku<P6_jZZ9ph{IRIL*z3snITg%z
z=4(6u%L|i@UFie1iSza4akKVrqYw5CT&QOOwiL2q1zC*kiSq5Jmpbg`(=PKZf&Kl`
zKg1(Irr`Vc5&#|kH2U~>d8t)cG#q?hTwd0wvRPdN`OkFoNA$9Qp8>!CyVb()k10Yw
zxCwg2tgQBc3Gp+$x;Tx9O5>PcFR07U%ZuYMkFP{Qf*jJ7esBYM=x~9re19&Z%S?nb
zHZc)Q#Ighy#RvnML8?kgNvSoMv>-3<60C{}2R^dh7=8JG`dti%c~D@WEl@z<SM^|o
zW+j**3Q9_qpT?Mhr*{fz(+vp;d7sA2#pTo$&9;Qq`upPIf{dKJ?)vH+bQ$21^(8Dc
zXqE_A3y3SIsYx)RX5XRuT>ydW(U<`MOf>?Ij;GmjIS~vhRAFsY645q5!-0p7rc-ub
zQUgVBXKGymLLgx^{MHBCZ-H5|B*6XZ)kD|io?PG!zJC1*<ZU=rWxEuVrsi_CnnBEA
zrdna~=VW)D^KnpRBj`SX76vb5bAM=Ph(*6sbL{-;@;s73B|W6(L5Lzf=$_sK)tsE1
zNO~phAQ7a4BD+mxvS>b6C+CgW#y=GxXh0w^y?z~yUbDa=S+amv-U6(?M&0yeVF{tR
z`F04o@HEgN@V*;+{-2Kej*gBdCzFx+fBhQ!LQ8;!?ZyLs8ht^=WjWUbx+9L4&<4yf
z4W7T9Dez4^ZWk45F)#c9KJ^pqKf3(AxK~xhl_ouQq*hA9&wmYOIB=5>U~0)vwdsXH
zSmVBpe*^^utB<Fh^()y-i^6tmj;B+o>ll$B_tj^g(cFVxgB)mcNwc=4xVU)0p^lDD
z%1v{~tgNg`%Huh$sc319Pj>#qaoZF434-Z-10Wkb(Lg%7lnz8NElo{#N%<zh^zQD?
zzgeqG7q4{KO@^J~IIWjJVZzZ-MKapw&$T)}n5ZHxFE0b!hlYj*9uE*q1`ugwWu>!I
zoY#5we7kPh;#Cjt@nf9k)>brBRDVG-kXZx(1xW4cU_gq?lOt0NKtC*>u&5{z!Skcd
z^?{7wz`(B0C};<FdJ-9jc2@exhjqXQ1Jh3M&g`2qHYO%1IT~EQSC6~<QiRQ?t2u#O
zK?9OLfIk9cw=*EcU&sYuQ-bS}(uWV<^78V0d|H7A4SK?zUT8h!1|m4VlUWqtleoFL
ziHTzXwgqSiG)lps-t_T%{r#mbV7(tuOO_NDo6u&LJytU{%{!t!=*j^{XQ)+LEq;6U
zR#jD1knH33bj@<-59{UbCvoA2oJ<Fn3%^4~)SdIdUx0seb8`d0$E=q69!9_K<|Dn{
zLw*Tm9-;|zaBOG%CleFry#>*?!64N8`t}Z_dD<P%+i{iddb${CIu<NWg^7)AzJ|Rt
zTBt+(=+WT1Muo*2dYE86w_39SS*|oQlV%eG82z|x0RaJz;^JbvCqRma#$BU7wNP+C
z46qJ}N&Q4pii(PQdTA}?ca2Bd+x>kdV_!9lYqOh<>ozf(8X9UY0(h2}{o!yt8)&fc
zHvh9{pG~{Ey5#W3Cng^B^{0uAq1^-6#u2#Of`Wo0IQ?*aBxgI;)fFHo5E%94)89u2
zvCTw(T9+on4FnpG`D$eYa7nHO2sFBm`K>p{+04z&H(jk>#p-oL7|9OrR~yEIXxi-_
zG0q*YUjTX|s;Y4N=9&Xko}}VT48grIYg%uw_gG=(vSZDAq<pcy!?{W+29d~+I}ZF;
zwP$0@Yqb<&fm)h&xXsPYjk+PDpn&koakl<5_`TNV=G&;KMSDXz3OTfd`1rPKgMPli
zy8^>i4qOue;fxx!lN}uaa(pf)KhL^ve5_!8e*U9C5*~+dLm=}HEE+%Pj1>PR9uZO7
z^1jk3&_?C2v1)KDrk4P_D+kDcQ&H)ISO{S3<DDJN^B=D)wN=LTI2#o-TtNwG9$sEx
zV9G!XLDWEQmu30Dfh1U15JX+VL5~4`x!UVs0&1!BsPT*fOiBP;j%VNm4s()q4iG`A
zspS=t!Z1vRfrWMT$A?r}S~`e?_pH3`5)NcrTV5WCU^K;uNq-{<0B`@a@%I!Kn$PgX
zNg-`(443T<f&mWv@`i5((xSN*r+bUp?cVW(5`MnEnuov<0R;Y#l(Y^Q@S`07-*nZo
z<3)(^fb7`}Z)EQ&`9QCjI^_=PI+#b4$-DEd>jMc8q^Sx*emIf*!_*uj4UJ)kTb14B
zdqG~OgUsw~X{LhQT-|b3YVoK<TG=jMKHIhLPK&v@^mdzkIxTk(t>W&ufJ}HW39n*h
z%jP5|A>+>pCt!3Q9nX|x&G^Q!nM9~EDwxvJ(5%eV@lOC4m?mG|VTB4XHgIrqva+nC
z!TxX5j3?n+@xIzRIzx0h{OW~fQ(T7<q;4nA=;~Zfi;mp9u}SlBt{2+E#?ry7<#vRK
z-ls`d8uXCd7(L@pp9cRUs^Fj?{kI9<z9p*1hlM@B!xIz)26*Tih=?mKBg48<l??EI
zs+^DWvsHl&)qHp;IG*@4a40{?a~hV#Cbo&DiQy%Zc%itF`oK?c4K@JDepZlfc`cl%
zyb_T2GCe8FAl%PzXSN~R7}pO32(1$nD&(hHquR4IPKVOrKQZ<A5(?t<7}TpB&JI>V
z{PxHzHM~BDM0q%>x+_<?)ZcWfP)B&5H}U-}!e*5$tc|q{EH4!i|Bv3zG#;w9kK=P!
zPwtWyk<bujmo`zhZV|G!Op_K%cCsWHW#ZKBBRrB8OUO-wv1T1)Yf{}xvWx6V*&<se
zOY;1t`^EF_dHKvspW`^!IoI|7{eIv7bxe`%E8@1BcbTH&hf5{H!^5>aNj6T=BC0pm
zty$CZI?AKF^l$m^!_E4(E&fZ3K1I$D9Fc{-$t%Lb!XjfC)IL*VV>_uT5!{q&=0!HV
zqgt`Z=b>--7S~Ic%$15S_8l6tHcil~eL97Zn9HwI&l}%!rJt{;Ga^DD%p@!<OhST?
zaLtsztD<aRTX<n8@yqCW9kY%0%`!>OD;3n$<Hlrm>@cA#kVq^AMH?Ox`M0u%QjF5l
zQtG(w*7aR%VhI8>QrtoLcWCH7LwGD>-90g`C=()aHa0f?KPL5~@>Ihg3w-|cN%=yn
z;@D#>L&dYr#7xfI^k6}H3tmVg#(xgPc{JVEU(?a_HTV1rU(RT{<F1=nV1a>wc?mlx
zl#K!KE-87GCnGWE$Nu%==2~*bn=Z=x9VaA@Yn_hh3D&9ZN&PS|&}vM!C0m3-5#Tw3
zxwfg}ES4sMtj}?ErKE>KLQ3bhG9WMOL@?NmK`-V|+d4NB$$1_7-Xs>!D>rHho9s1r
zl_2>}trH}Xj8}0@-HHre!}y#)FHg_*Wc$~5Bx=V`q*FEt2*8e<uZh~>Q9fCPQ|^;5
zpBspC4V*x@L)cQZDE)@&cRal%@%4CjnLc+-HO#5G)9G|tWqf30%HUca;EIBWWDh^W
zPqXY%_Xw?hdEet^n`n_}SQEWfaby#h;bM-oTgKp>xEzm{kL+xdy)2Xkxg-4AQ(>fQ
zX=$m}c6gg4=Zl`Nui66J?ch#;%hT)Dh-sKR21Z-@riW>ZWLhL8u<c$uGyL|%I^^m~
zZYZnIi6w&~O4|a!L8PA5*AwA>v(xnl3q7Ks513zPlP17w)O;fE-pxZw_=h;Owza7?
zJ3%0@la9ea=QBV197OYtpd|nAD-i5H!hvGzryCDX(57vUtFG7Wdm(IP)r-<RoIp}8
zRCrhDPo+i$1?gvY&P+}=?&TL;ENih_W9ZawLA=={>?zwmezaC*qSnc%G&_uZ3VCWe
zM|G!yf~|Omzh&;G^kJ(|nB}Ibtzn(n$5vJdriO5;o$@O$d3wsp%O3(Fwf232)IruZ
zH{ZN)5gM$?W=!)<C{VM8y34-b<P$||7Fd0l?Y{^<QfgOy4=^rfucJ$wxF+FySx>JD
z=9FocP}ii9PKk|(PzFnc<$;iz)LX}vbU2vt7;L|%^3`x(j^(0U0F`wC-pg_Otw$E~
z0xnuYP3-)iI@Ruj`6Uewsb193BJ4$bx4g$qW9N!xzb!OS96R6~0gHDVB(XZL`8ClS
z&iceUCESu0Ris4|mV@PZB2K4JixanTi)Gxga*y7paPGzC+~tMVVzaCUwp7U^Tw@%S
z^_>~<hRXv70|9^&&*{B3U0bu0o#iVRvGT?ryRJ@s5+{D-!AyV{V20Yhx-G?c{`@(h
zNG6NLnlo6oY*}zxj6LW5IzCs)Y%L<TBXZ=cWZZi9Z!~Wgi)fMLIeG>YUc&w=C3NYE
zae#@3@E$vZ(3+97C`MvYybvOlqLlkcmA9v7UX_c`W8VBRt`Q2m68pd*S431)L0;Zq
zyjl+OqWfz|x%v6?LrLMm2J(^q=<)7oO;GpAb{X3(Sd_>*Dr-y_8j2kBFCU?8N%HGg
z0*inZ)VwhY`>m_{5)_~{$I47sN=ho@9Xswh|8)fS;d1>`AhIzrM@%$sK5m95@Q$*2
zdMzfH`|{{YR}5nh%P1~Y(@ys?D6z^4Cu6<Qrd0VGj^koAb#<-nOS;KNWd&9Tx?C+U
zN)RI8>PAm;5TKh*_aP~}t!Xm<?|9H0Z5QK$_XlE#(?eSa>f&pZo<gKc!BCmmRv}KE
zMQ(rsO-i5|GLlgEvCowehEjAjN;itxH9x;zV##mJB1}v(*3bHTdTeEDRAyS*(TY2#
zyALkS)n{Z7@{&!K8_VRl#*ZUN&}ba+s`viympAEooRZNgqa;+M-><eCjT|;xzKteF
zXO<==K6@=uWVrNmF=Pm}N3A~g-aY<2b60!&Lc3~ys!hc7pPB+*BQ0{A7x!-pDo|Ow
zxwSn)TlJ%}O>E`y#TwUvKBR6Z8Sdi=Al8TeJvlk4sBV#KU8#^$&WIn6jE)BK*efad
zXG>2>%5!(7y}iA&@$+bd?Tw9%qXTi!*2wQ)+QNhH-Mhz%F1j5a?kLWRE}-@<->_J}
z^kZF_GY9y<>C+nUQ)r>Cz~;?q)2q*>f;kTjIn7YTP!7@1ef{~8<oqk*;Lk?Wu@O6`
z=#fuI+&zC6dJUS+^h!q<entuP&bGJSgvW>Iyzv;QIr=PRFI|FHZIsSECHnT^f$6Ri
zuaP*rJ?(E}_9E{c*ir}jVSfHg9WNian;`v2&$r>`lOAM8YL@xkh=>+ckawsc-Lk2x
zY-1~J^hXAX!PJTjaR{7h7TPl8S4i))uU&IL7shKoPhmeJeCO>g2QEFk{u}%xIH|L<
zGk{=rS{f8vRK*qlv>r?I*wXwH&;BCUc7$KO+gp&UKBB3CGKt`d2C<5|IyZC`6>&fs
zLqbAyg4PO#=t2tt(ugR7?cltlQpWD??r3o!6{-9G*dLmTe4MdZ5M+w=w06t>^_w>j
z!SMF=owv6a6cSQ(Zc2tt1{w?w{+(vF^T2_tP+ixQG(2zR=?~c}A02&y>iwlH$L!($
z#<sTaXi2kwO!1c4s<?OUj}p$ezSWs+3YD!R;xCqT;XDYifIHquJH*mLFJPE2PB=R|
zi_C-rsO6pInw#^Cc=$xH(FjVDbFPGOCb97x5?e>%g&;mrDwV3@@b<yP#Ds*-P5d4g
z*&=lnd-O^(065FBaI>C18Voc=ac@rLHX1rX;a09Cw#?#v6pS$U$&*<mBN?;51b%TF
zEHW4GJ5?@ivH|W9&B@-24W*TaM~+Aop8bZ}a(su9*D{jz>AHl2*iK)sI-|dH*REZ1
za!1lB4t93y8$VPAyko0#Jx2~MA7~=o?A>4cJtZ@js$~+C3rRCL0C%><hsfTYK2tXS
z^n}})0*A0VqicX$*v2vB)q<yMZFK-C9sq}v-2)oe(QzisL%2mLQ&?0qvv)s>X>?Nc
zJ$sRJb|xNPsJFJ(#PMLJ!@j|RfsBy?bxx?z&eq?VOy<p#J-O!ts-)kk*&&i&2wgZP
zAE_uL$-u2mizbTwWMdF8!-AtQttzTz^;0v-y}wXT7Bc6p(kaqXQqE>)&g%%1$&Zh<
zCr&fuW@cvkUxl~z1Rzl(=As40>|V9<U(PRSGveQvzV92@8np?x#x%_}>73*xtrPIF
zc3dyZ)6&p@N%_p7SZ)7)$Pw*JtJ~D4Y66cMEvhV2Tibq265{8aW7xPDSn;=^MlCI^
zWjCI+wfP{fsk(O^m5*E<Xu}z8H#Idq^yp$w`PGbhY~C3!pEd>1iI0yT8yicYe=Rw3
zUz)7xOL4R|idI6?Ec1`86H`+iC0?oX%+6vDVzw{n#*G`lS|&7YZ9*H#y0*T)ZTim1
zRo6^ROfWw#UuLn{>_4_%q|w&$@^->HpRj6ScehGS80wU=Vpim{SeN^r2O{4V6%^Qd
zZy}TW`}+Y&-!J(Y@?Wu!3%b`{;w7^$@8Ofv{8v(rpQciaoEjO}c~)0hslA;af!6~z
z9UrxRRbr^zsfxX{IPTeB#f_EqgY>IcL0;%thvP$7kF-VdEHrs4V#HL2cAACC*e>vG
zq9+6f1n7XD-S0+2bnIr5t8csviIgjVQoo3L6q3ed>P3E}QRe36Qc_Y7iZY@*eKhvO
zWWf)u6W1!Ie^;XGiF|OLK&Ox(rfXGUvaB%qN^F0DtqtIMxx3J-!uY;EYggYeIrjr{
zF4>W18-zn<!$Ej1m7L>vNs#`LCDQfOV-!k6Ma3fG?e_cCZ^^u*4UOQs1^M|*<Cvz8
zm#%|(n{s80q!HHD-HjkigM1*)K2Rq%kt7*_=%74M(m3_?r0b4CTy%UrwMAdtPx<19
zXK*hdSLn~ym!fi|Bqu8(Yy%9T!9vm~<DT~N3S2Ii=xO>YhrinEqiC&sPRR$-h?#0;
zWP~3~WYxTRa|R3#?WDK=Fw@KiZa<sLf6Yl;LjLRXe{Dl^o7-By{_Hb!!vFUbOe~Q$
YYH75-tsKFZDUe7<4x8#`AF}@QKdw`EG5`Po

delta 162786
zcmag`byQYg)HMtrKtu&Wqy!|S8|en6q`Q%jlosibgNPu~-7THcjYvsLcXxN!xAAx1
z&-Z-ed&hXsxW@1g@H*%0z4lyl%{3RpH;)c39~EQ1fxRQV?X$jlenWiph)h3KKlOJx
zU*Yt$Zf)&P8Zk<=C!z>VuSB1St?%pv{+uOgpWWH{(?553eb(;WXV>e(`CGtkv&RiO
zJ8sLUB!%l)*tL2Rj!}?=n-TP@<~=V>9kKYFQ<<`3g}S?c9z9M~_yfJl&l!23EHYgw
zmvtr5SO-74dA=t|##a;~p7-!!F#5w_LA%TS^+bAxfX(qFq8HvzgtM*QWsK^msw;Dm
z@So0gz){;SS69dTU8mNGQyeLc;--{7ygDXwGYhh}?q>(PYGW_ng*uT4aB&bgAFpxx
z38n=mTvF$y4n|w@iL3029F^}TP52GFVWf~Thlj$6$x*R`3a_M0(4lOaKkP2F*4BkT
zO?(V{&=5#@1o)1+PNP33>mtLw^bEQ`FT1r@2bC7iy1qUUevo?d6}8M!oxXy^IB({m
zWDdSEu2qn{)aF@)g1Um5E|0@L?#M<W?V6$Y^ng{1lEyb#D@McNok3B&Yki#_6jP_e
z!{ZHj>xTlNLTL1T&$EbbSNOofeC?;-I+QZ;9w9j@GCnF&TSckOQZX#NB>CmQ{@BW*
zb~NHEGZ*8(S=pR&|1A$X!(1(8<1Y+iK^sC|EfuAj#j)j}ZWITiRqN_Ds!UG%qst<1
z`qL>w-ivmFm(K{%li&rc2axOaaAMysjJ33CalkOdh&Iv>d6s<dyp+G*Aa?k4LseA#
z47$J0&S`G8@%@uXX8zmiQ}=0+OrOS6XL{Pa_h0NWg&(}|Cisf%Ub(c@)S%5AKiWiM
zckqap3V+Dq<uf$(J}M8+xU2bdq-}oN>G44fsO&PgN@r~LFPh5`ku1JNZXpE<1e-o5
z!P4$z5xe>(<k5ozEoD%e$d;Kek@0?tL?AsotjtdKv-Evihx85Qi^FAw&Iq{Y6XD}#
zp1tXT(cPbcC_^KxWN?dSEP;cJV7C7sdYaBZy%zjbER^X#^6M%|nc}V|n+~0<Yg~~$
zGjkW3rBS@yP%E~N%{c4{0^3IY;Zu3t*Tq=2iZpy`R~&HHep*<GS>8!sF(=pGZE3Rb
z-;J#`&L}F%W>B$JH4?MDA>r?yg!jj+dsu;LZ148bFCUU0U*<`FduY0Xdv?#mm52ZV
z5;?lS6cfB#9fjK8hz;6wjPy3)&xV>KqPQ%c*htdi?lmUX?R=aM;)Akb+75x5j|wtB
zoJ2xX=Vy~H?zK{7v{Gh7s;hNMe5db_<dZ2QFcpjtzUfn-;wW8T#!(HTc8bAY-xm0^
z=_jvm=7P+{6<tyn-~k6Z_?ryrFc*~7EoA1MPM%^)nrVtrW;`kh+mJf&C$bDcSubOa
z#eG}1if?6J3{TfxukQB4?>SBdu6Eq6_@Z<rLT<-aS}WA30`;1HJI`%Yhxg5X(oz;_
zyd4U*Vs9wS?Cc>Y+-x}bw*^+{sAtxaX~FM3-q|j)ge@2lg`;iow#5Hr<BXLFdp~o3
zLSSb3Qa)JE0TIGM#y_R^C2-X=Upp$l>=ksq@hN9vqGe>F?SXG~VmkD7M*T0Vt7$Av
zRaKR#8OMpw$%D?q#Pj@QWddboJ}n%MoL@D}`yj=Hz6)h39U-QnqcyI$l9ITN47uPZ
zwci=CeN)M3Bg^_+rw&D%+bW^;?`JNO2DA(8$k6b1ea|xQn4Hv|*;32p49YmWuZ+^$
z)Ea2@8%x)0EF7=^U$RsN?H>u6E@^oZ8$s7IlKR)=cYDYC)UWBiLGarlWGg}EohW}(
zas>B!XBb&%WwUki9W4i~kWg}yen^KA3(J*JDp6oU!<Dv-J2`@n>vi%h-ym@%6Fq%;
z|NFEw9eZlVicfPyB=uviORNyhA1T+jLS*!mGk^V}cVK~W_=FRp>6Ge8VAoPrj1u)~
z8Ir+czCByAhsLT}-Yq^wG2!n=Dx#oHOi0Seq^TThGVzRZRg}yr%}S1*<%2?=;$2^D
z;o^NogsPFnqR=fq9-zAIAB;_E5TN~C*lZ<CPVl&#KkQE6hpHT%OSiM<8Kr#&{kaOu
z%F1eLU;;=~RJ5vUuX&dEn|L}79-jaA?@BaVBoe8_o>Cc65fKn1baxdV^|LLnt{#GG
zf1WpAKv`Fp7*e2#V7k;rbwG2+8ewF#k8wVoFD2luPxa(+T&z_&4&v<|&{S62=AU{0
zYYAOCv+LubNB#C1&H@Vto)bCwD#EXQc-{_DPz-T#IgKd(+o+6zE0g@vd}P8qB4Dmq
zVgxZMBq52zNdNZI)Cu+Bl}EbeJ62}^p%}?1MyohCQWD!msc6tm(>tE?uh56ZHMP9?
z``v?CL4+(*)?&u0<)@oZMhN25jkxQ%@3!>G#=?hQ-O3+1nPpCkVLm|{$GdQ?)o>lU
z42r{xL23<I0=~4=^b@Q%^3Mn7`h~oGZT{9=uRcS&uL!kg{UE9Nkd>-WP>|^(7afY7
z%+(WDu74jwxY5n@afLz>qI6D0S!3;;yI4(*<zhID(e_DMIKrKu9J3)nRWuQyR9`g6
zmw!;+B(b&jPP~<rt7+}{U~dZpG<)!3==dFmyHmO4i3+XlPo4H6qva(w!fG1qOTTms
zOI;=!&-BF`1tZq=gZY9_s@Tljg8A9`1{J7}_6l^Cia*{Q$#oa_k9}NH0tx0@!%0uN
zcEPHZOvj&w`2~Sq_v{Y+>6)ggskife)*SXZ;!$aTM|Y%;ewHeQgi^r7FXv9DC_4nx
zAO(@ES$KiXX@h%YuBQGXfvUN`^&iG7jE)YYI;*RXqF@jYxvd>6cYXZ$F`p3u0YOku
zP*PH|uC9)bj_%~-<oe>+kS{MccjM;za(A|onU+>jS(%XgLrP+z4p-tXm&??j{ix<I
zg5sjL0Y$l)om~=fa6DR6r1^-KOg745ngS<da=r7)Qe|rLV%UHLg_52sa%t+956dpX
zR*cR1uY|Hn-R_;aC61T6Zj>aV-8bxMD7AUZ^kDDQ{OaxfFQJL*F=3%Z(Ge#nBHpK)
zAwFco5lb7KBRU)1ZRTU+lNj`Lc@@tawwCrgC#J?)Ua1!)6TtH6wQ(G5u_JGv-zAm@
z4zG3UJ^MO4&E=jY$_Y7cX}o_=13h^7P*qLM$JZA=KQKPt7><gH>MjOvjf8?uK}Ch=
zSzP?V{_Z*&Ez+yBnxvwj7?P7u)!E<sppDK|_S|}G+%JXc&5ugbps1o~UDzSZ&dpD2
z+NgNPYaL#+sd3FRS=th?J3KpS+!UbXWEtM?_s7s{Xs65qa2Rigmv&k!nftv2;J)R}
zs)3T$$mp)EUSm4~1du;8X~Y>dJjd|O&~$Xec}LmY!asD*rr#t+zUm04#=*uOL*GBR
zsg;QL_m_VEKKsXyrI7;d;h`bu*W~2o1$?}zUih|ua1aacw!YzLeJ~3P3#+`moXc{I
zl!+-RG}LzOZ>r5~!*rdC-S$K|(Q(2P;lz&pmXjH;+g_IOYV*R|+TtH&`EJ3V^4|~y
zvlF-A!s|2PBCET9Ei4c`<Ned}a&psX@(L+TU%u+%+WWGYj(0qhaenzd7JnUCpblO-
zW$NbIGko>lS<M<DQ0<V3VM9k_D2I?~^`x-;CH~Zc5qFLAWln){-T4#D=O>IsDH;l%
z@{Nrb+;-%zJ>F8|-)-ABw&`?Fbp?gB;k>5Qg5lxez9fNk1sV{q5m7hwRN;YvJ*T^~
zv-NIxPoElCSfD~IEG&wOie6q`@MNV8^~I#Iv9Z(cOuglJiKnM$ESss6#N3>nx2xdi
z+nwZ=QUBAeB#!VXOVgLnHZk5hFS;f<e5!JKr`S?q6_@*cQoWajgf}GG*VpB!=W>rz
z(l0*VBH{e6?TQEjDJ>&C`_-gB*X}1Ip%ol#=4nU%wSG#ohd*(t2fO_%lVQMb@=g+i
zWu;6jrd+maSGN+cH&@03IV?uC$?pR4_@$03jw@ZHmVyE)8QFluF*;;3Q%6fr&&JMf
zq_*qb>W@bv;A+1&H+wDp;>8O{Utj+hua>Tql$5lzf0KS4eA_^jgdm(v_$;lc{*MPD
zWVEuh&1;)@95Put3jKnco3Hl{KC_sJ=OeR0LqqIUk%nv;#w$OJSFj@^ShXXTGl(n;
zc9JGsKCVhZG>k<LuAcRm-0caUUyy8E#YYm0V>ucoM!nM8g-;lbj`e(ZK&yZ;n1On>
zwPkt#OOsmE=<$3r3iOQEo<_SiCMG5-3fC^R_Uz0VddB_@{^ZWkFZ^8W!F6oKt-u>q
zMO<hqsovxK1gSJ}Z29$T@sgbc$NOLQDS0a*PZP_GYCrd+;qb|i2pE~Lhd3)}sr~mb
zZu&2u#RtEoXDqU$(cc}f)vQL|j)wlvqc|H})qVT+t+w_uS2ZtFKDoi5_4{IbFf|1Q
zT<nvPk%8dp?JcULG*It$>IkA7ghH*;ZV97jbZ967)P24uBP#96>xQC8PgHBK{<dA}
zZTm)Z^jaG^b+;ycp&?}fSEdyfDGmPkw@nT38<he~K?=c5AZ)rp@?w1maT3*NXk68f
zYKei@Q(`SCJ$$6On4>aL7|sPZh~6T4PE16dHgB5eBEI^Y^(Xt;N#B!10X<9%H2ceP
z2K*qV8R9e#D)FoLTc!P-xf^eoktUfD?qVaAI8(asGd0~F2<}GsHp$;V4?McrzYD~|
z;c6eXNd?j?Ny&%M)z$gn@NjueP277l42;v06Ws>)^FN{FuwT@{!AB^&cA_pNg%naz
z`SAC#s)mM!w)Uhm+M!o#YwP*>d7$XX(9jbMjb>K61GbVw8l!_(R)H(jYIt9oNW;wd
zebHYTDS`j69-HDQ?CxK?bs8;oU9a!<`ZYp9O_Ja{4wnv|@OJK+#%@L$t;*Z2j_oc4
z{p_U7#pYFZSgU7=ax&`LIK7ZUT}kbRo@M)HP@NW^=gPefCvte)<!djoAk~*LKh^PX
zqp2k&B|FlJ8fUe8OpJ__L2_lR9{Vc-fog^VHj7LBnKjoctzuun>RyZd{vs%b_=_~L
zf%WNc(U(~u)$UVa_Mt-G{l~cyY$WD<nPuC&-EUT4eXEIX)HHMj6_ufJ+LRMjO>m9R
zRXrOjV0jh*mL_OsFwNRmE{p4&<PIUKt8Kf+i1P@s_W*e0sTXm+d^xwUK*!ANWNW*>
zw^!w`sdjsF4G;*ylV$?^eqKR=fZK^B4o;+(7a}ur(Q03EvJk8*A|i6fpWd?r{<XLN
zv&gNxudlC8f^B$Ygmt!y^UY#YcY6N2;H3J7xwGublwWUl<9iwv{q5Eybf~e0h-+?6
zqnq^EBbIWj!)M*UNl10Mo}Z*gcBK3y5AY#eo0!z7a;}$?=ooWfqGJmaCF_YZGR`l$
zK20ixig$jA8D8IvV%QkfDo$&|r2@p(DYauBSC2<m%Lr;F54o)SW{cY_a_AE$>%(No
zSibN``K0@E=^ZZmOLxTw(9+RvUI?+l*jJ86&S#7}`+M3`i}RId7^4}e&h+$oO<z#x
z>F8M7>ro#(V0c?qTwMI}<;%Z4Jyw?|+k3a$aS;&_=*0XIsS405IXUZbxWssQc^M=a
zhmRj8>Rj0FaxD%H&hYh6>{&us`lq=LqmSVNd-(*j3g2DIW{OvA#E0Xa^FoI<vRS5H
z6A$+pcP?i2p85?bgw;E{pj&BhyEO}ymj`^o+a=sM#=^syDgSu#fV}=vUXSvJ$qEjx
zzyBi){A>=30UAaQF(?KWaEb8MlTgJ`Rg+5CE^*lXJ@MPU&aq*73{yNPShtNb0~B@&
zX6!t)55+V)B-4F^pYw)nll>z{gri=1??#TV2+8OehK^yAo9Afvks65vG@Q=q1GUf^
zTLlOkqQ$+LbbjS!-(5piBy1)!`ue1HRPIH(^_`ua`2}!+-$tU@*4R(51cCnk<PF$Z
zSYZ9==;;k>3Hr(CIvZ$I$m6iOZ*L27y}y3F1Q}X_0&jG87t!LGEw7)b`PCE^X5;+T
z)m1^A+$JFCY|qeGV;QJM8r;-UdzIAG#MAq^FH|=qFnYJG<YF?+k^_Q?Ed#^b0~7eI
zjq3N|cBg#a216VRY2m-kK1?%N^2zxx`)anEFOKj{vS$;HxMlB(g|xD=h-Wzz3=G{@
zdBfbLFD`tGYB-1u`6X%xZ+TxJ28z=&G2vojzPs$hK1Nfai;VgBr@g)Xc6ORwF*r0i
z83ebUQyJ|B{=vZk<SCqLKYpr}ZD_wPwgnTAQBop9$jHboD2icWVUU8;PuDa7z-%61
zVPd}MH9#dd$ynf%@P1;hT@y!#R6a!TaBg+6wP5ZaHH|CUJo%-rqH=b58Qx)JVPR46
znXF}1{`>duqO~<OuEN5IlW~VoNPIjh1h7LiCN0gt+1XiC)H`P##+A9aIELw=p~>M{
zTFxDz&9xOVekLYHtkIu{p3|GS1-kG10v<l}cfaB%<zxZ8th_EwNlogl<ZH*3rJo~F
z%`{8*g(6%C2|SK(7MW)~q9P*cJEa~EjZBrfSLBAhbrT}@{+sM4N=7tWVc&Tm@OYb|
z(y8aI*~8~yU;dH)w>p(5a8ht^Ur*2W#&F*J{Jf<vA|fL6N?AGl?Bi6GEi|{d*v!d1
z-PeZ!v9o&=Em#{F-Gkn}dzZgC92pTIDkcUXC*T=NT78mad@fM9Yy=GWZ0ghFM{DKv
zfE3e7PiKgwe2pm7tygK&`CGQ^w#ck+Wv!J@8PDeA)1@F+V=-J4M-U%#@prPX#qqTQ
zTa=&pgVd<?qQaxS-$@O02bSBvg%MXz@)OJ~#%VSGW5~+jv6Do}0!f5sIy%<7W7!C~
ztcL~%kvwyA>71OL^z`&V?m<C8Aybl-m6ev>Y!4;@AH4>57l4Om>=Ws4g0HTxU0hsZ
z-|_Ooym4`HLhud0)6Uez@g^;{l!L=@DEXWB>~WEik)E#x{TUqs4EfYarmHtvmfhN!
zXEzyXo)8|~gZ|u-eED#kbSYvZ1Sy1tCmEX-<kGw$3w^50nvi_aLYdKUd31clk<~-L
zQmiw0eZG12qpTtKyT7U2$YznE#@mazeN_2b_!j~1r`C#ez3T+O?IKL!?4EV=+_PD}
z{S^_>4`Y=HS7|d|R~OgA?XiTIwLAY7Aa!N42ALm{iRUaRDLDWfyVi00>C>kRi;Lt`
zWvCAy7U(r5#SF@6l$#GnF>25JBb{VSOu7mRud)WgYXiP5DHB7Tuh-~txB~aQe*GF`
z1T_^E83~CLr5q5ZAaCymtf~vK#){^LKZYot`dLX_F`CMQLOJ5luAYX0;*X`?(P&HP
zU=$@q7Rsl9Z0%p|)EW4MXdh}mC`jy`)^J2MKL){Y@OP4#jhgudrL2x->ZfEJ;WZ3#
zs&IyYW5$3<eMW&h3IV4IX+|^6(PDRAogX^b+t=0CFD@(q_Q=l8esQwRXV``Y;%!92
z#pN747C2KrU%vv8psS-JCogYua<a6*lA4^{?qo}onmVg5S*X3W)z8l_HYR3la?&Y!
zStF+`j(}H3P0HYNJ);HSu83FeCL;I2V^e3Dj0l8^!em6_4Plz~SkI{fU+^A=E3p=P
zzR5<(E->~MYxRaNHPv)!ndtbtU#W{A!{z_ZylW6ZR4lZ|cUL?bC|9&WXxaT%D||bS
z>*9b&Y+YF7(fgRFhnc~AEhJSJ-fqaX3_M^BYe5o^r|^BUp+eGlGv>VZyU1H*L>xv9
zf8Abr%Sg@GulZkmr*gJvBEW|$vd}Y5ON~a4>j`R-7KFR%QX7a7IPozDoJ*ync`0|*
z<+yr?*R0_&rm#TMW=NmiD9MlEznf*g+I^F24x2G)`6~6)WdDeowRb6^M}L|a^kKOZ
zhhdJdzsNgcvM?z%8BSk#I+&}WHe_?}r18bjm&gd!{28}Q>tne;z4UM4xvTUp6h3bH
zyDxb$v{KXb<bpW3X-*&QQtVCTb(Z!26j{>h(mJw=aynv7A5u=bJVPc6b&r$8tex$g
zy%n5JTUM=fU)z!+G$jM2loDii2MwwJszd>O7-3cyl+>6Q9}gZ81b26Kpg^xbXW|W@
z-aPxc+QI>?Wfj)h^>~F+-LI?T?@mY9{jHV6B%D3@==!RlBPOIzF%WytR?>_~_a;<m
zL8^;BgR;d3e@!k~e_XrREq}-F?zxXve<3#1|LM3&vb|uW>NSZl6L|-O$&PhRC<P`R
zJwj7e?^B0<flB18jYZOyr6Qf<=mO<0nYU}jl-GT{Vx!Iu=KRX~im3g**4pSDfx?1o
zCz`*YrjnojAkb)2u>A51XV=qzXS19jDBqgdMC8biaf*cBB{p$VpZY4bwDKEDdXEpg
zOHcO}Kco;5S*Tsl_AFp8J0=Rw@_({X;fnf*9iM#aRrhDRO6DnB(}EelH^^d?hnzQ9
z2Ryk}N}~Uk%L)PY?1YSJz%43~0|F9F-`i@o3w1>8SDG%ZclXfTtVc%1z2?Z-5gX$j
zR*hVONGS!vx3LiDA}bdEP^NiijT#8?3&XfXvM2YN{COaV6oMs1U#rNHZ4PW_sa5X3
zd3(S?EngwNvV#peDJ#v<nig*VZGE*CtG6IKdzpHXE@SQNRJEP9qk&VI>GeW!kx&}k
z_D1xUHMwTU{?98A?Ii<YVy%>ZIk7e(LaTVVND$uFkNe4tPd|xj_AM`@L9blImX(t$
zwKh8Izry#7C|cjdKdxS0kjxmC(XFy=9?-mIkMRF4-FPcicQPkwAXNhi>A8p*J6eAp
z%_Fs2Y!c>%aZ-hvhiq3&t*iUrb38_->5#-A`k`ETJ%0!ph;{n_^!*B7n!SW7iC*6=
z-NsWbrC+nvcB{RKeBkj(RDYe~V(;1WyBxq9UkK&ob5ykEE}1{?HWs%&BSagk_1K?7
z(6Zb{Dr5hK;?MaU`k!=cW6Wvvxqh1wuTSqGt#sv8Ga6N>gtVzF{yYA)qDCjgUsn#7
z;?FQXHhSa}B@+r~cPT_g>sEef`$><nA<!w7`h2p%E8bt}^lxR%85J`z4puNS%>S=Q
z#QW?&R?htHMbkB5G}DbmSbDs=rGdHdN7s*9nK2SJs+2COl+{01+R_PPF075s%Ph9Z
z-f$IdO}Vou&I=lliC+Fp%zG32xZ3en($YXm&Lh=`L~-!Ng|vy>$gmQMXGUpem1{$8
zexX{3nX6<S=hD`>YqL-^53ezNy1~<RQ(~6o{^VjXUyuEZs8{&t+p(4DZ%4SiwN5?X
zULb7><dxNxbkIjfsFulo|L@=qgY^VxJ3BB?(xbfIgl&}JSfJCgu4#cgUZiQHSNSPo
zJjAE-Mnqfv`9zIZIpw+GHD5E5XNi~P4405xDbTrOH*tGoczfZg;jUZhqoTAhR@(hc
zr!V1a<O55_l&xa%0snSRm?)hR`6}dvh)Hk4_)1R7jzzKctOy}q35UmJ<l5Fb0c2Sq
z<AXK?Dj=Ya7_r7CCzC({9~78`tz;F|%m)V=<49~+HQIs1v&hnnX0IKMeYfCG%-OiT
ziOWAXz;p+nRF}pHl9B7EADzAX;$R|A9GRRfkG*GLu;n%fWcb#Ahf${1f(&%D$oj<$
zLbpXgSHUsnvlm4F?~CqPllwjqhG`VvrIxyleo&-2R@?nvqV%SMY4<Gu4t~59L_kiN
zep=w8tEkzd&+g=FF1LtzAripC=4ub0r}%tFv>^8rjnjX3(7j)UE$8{>U&u=F+CTW@
z;K0GmZ1ZLY)4?`Ub=ZmcQV`WSIV4O|HiHgOy@3%8OCeol)z6GdhD%!qow<|kobd$c
zV@^B0%CmC=2U6D7OvXGppRb2U8H<&pqOE25v(6#VzFRKvWkwWomzC;+ndoShLqZjl
z!*tjU>wbFNs@=DXlSEy>G$7!WxGdQG+BfSVF4`WYs`4anmbPQ~5Y_A7JMrTAr3*?v
z^}7cFI-yx(Z+6(1*8kqa!=tdUP(E3Zk&bSG`^}Jm0IGAN^Q0uiU1Ka+Z#M_vzi-@`
zLV!MaTlA@4naNB9+#`KkiYTNQq9<5<a)8wLC+1#-a#b}AM(@^h>y*I-XSu`k=f@T&
zKFwm3#tzA63AE|Kmbz05`f_Y$uig284)w3cF=jf+78w<#rml|U`R&^SJiNl>0i^!-
zWeW=iKxcdObzr<YrQ%J7lC+c*<e8kDtRahg%ny`>EBNZl-Q7Jg@qsxWphnQERE53z
z5t5)%u6JD4Q;_GoW*-jf7`d!Qm%}ztR8LRWr=+A%j;AK&rDWT0e6JUk<?Ee#U-h~0
zy-gE%bn2o}%8X@G>so;Bj9t&idU-##^6&^~t16oG2kawV<lsg|RJ)=$pZgA&ujQR2
zT`s}4v73#%eDNfGih|Y1t*Dj6%%H$+?P4>mt8Rj18+7aTJMe>c1Y>#5ok%X@&_*iC
z+09;mK-G<nst@~j79XTq`-0>?FMLw)djs@Wo0^)y9$=6Nc6W5V6YwV4AkSp0B*BS{
z2<`aMg)%a(L4Xw)E{>tcJ?#os6=vC|joxt^3nWa<r08;0`{-H`s@v$6RRP)^H}>Pd
zHIk!OuhQ{810m$6pPAN%6r=wtP4eiAd7g7XCbp9wt2N(evEecf$ceq40OHVUk<d$v
zk3nAAd!1!lk^kkb2M&Bz)!(LlF-S&=^omka^i510_ZP%Wj(b~MVGi@5Y@lP9n3z<|
zZ!4Te#>Kj={$z%_D+$~4c^-ZoSZJ?gq70P)(m;AG#-tx;LhNOVb(<%=1QqncYXdE{
zt_7YR>J3ve2pm1~K$H^Xq8JX_4GlIi9!V4OH%$p90}JZ^CFd7I+htAsZOnQNRp6?r
zp`xZPg)8jr?6iU8GCnSw!2J`1Zb2jF2FsPmZ?iH2NMqq;>P~-C&8JHX8nIicizoYE
zwnIGN#}$H(P)KwXDe;ib-_+Rsgcb7fH}@dWT||+;UXj7?A*g%T&t5Ue^Js-V9JlfR
z>{HgVz}{@br^#{)b?^_$V*>jvc#H+CLr;$^{|7rex0|bTFJ#nW*Eg0=mDvzpl>~qD
z3J|SFh@*Q1?XkW1T2Mb@J7@G!KwHBOyU$FJGpY@!R8aBp*s<+lWs;rybKK)_sq|qt
zrvH8IV+ae+Dw4Apd;}bqUoVDE2J29gX}cFqTKBPw_YUFL!k%Zpvgr>|(aDo6l@ed@
z{0pgRqIQ8jz&t-XI5^nap5I*aS@1D4@2>TyZv#!$%E~VTGd7Z6*zz0K2nd4`k@4CT
z&3^N>jOECDq&dnKQSKT=wnd(vH)5|>PJZ@i2fdY+>btohzu63ShUQjQSPB02O}PJw
z_?h`z7)I<a<^G>#%BJRl+dmyVW2>m2`mgcR*Q|+9R8&NQiP<#qyS<$Y^h5ag_y`FJ
zU%W`=?0I~w0gg;hfYIg-Img(jZlg(CB`Jw=GlGpy@0qXn<$W0a!F5=ox<vYL-R0Hx
z^r}skjR0@yr46d#rxnc}jJwT&8^f}xQ~S+1i5iZUHiRhTV3GeYyYjatLSOz2^1XZc
z1qI->+k#1$NlE{nX+HR=7!et8M$A>*cyXxh%g2Z38MV^f)5GED-48??C?q23T9E&u
zbyB{_tNGxOfAlUt(V8o$+am@6Xsw$UQEV{mHmB><Uu|0z@WOnU{}HSpEp8N#r9Yu;
zro9R5W>3|2vyq1)jSxsF$!m8??7Szm?gIj<t3AnaDB0C1+to=~p-WBEB{@E+F~#$9
zx#h$dQqa=HKVvCDz6<$bUCE(+Y~ze?Ib@f;GfXse>s$8!y%*m?zUS(~#kAq>_O^i8
z08Bwkn*%g_lzaDv3|u_9SyGY-AZ+YNLHH0vK1d<cwx~EVCR==NSr#j7#^c24lJqw-
zpDL`2bLKs7lU4h@{`aMv0r@;B(0c%VZOw4&o7&j2Cs5Yz`ol=GvBmK3)-^1>K7UP3
zOa$G}{*I6HaC65r>0`q<IMY?Oi*12~pzs7OJox%}bgt>kB(365cmRQ*b4Z1_v;9^Z
zT=)R>5gON!F)@%CxUTIoimR}p=avjLCW5?ibY_~>P05X7*D`;HRgCK+{6{Son)DMJ
z7$1NgE<BM+lVdU7dZ5mHTS-DTRj1b%qtDa*s&d`#rl=$ro&e1*(;|_D@WH$sb}`MI
zrkXhEo04G_Rm-)AtKSRs82s*tLYbz^`^O8z3Q_UAz3zq6+!gp|2O?CXf7eg@wkim$
z_YSsw!$%Vg7s8#2G>gZYeDJXdZnkse6)css-YRLi7-t0hs0vFcYZlD|<bl@cq%T^!
ziynAF8b84#p%EatD&>;W=M+*jjy1J8hsfC_nTZ)kUUv$BN|7{N67Qranbq;$reYB5
ze{7Z^0cUSvVWBG;K$*LXmR4L|US4S_sE?BoNXf|;hjNs&<Prz|{*B?WGr9cn-4z|=
zTKL*MUX1U%bl0tx+n=^h=VCcM7c;*fa2W%&{H>lJ@Rz*zahPVDs+ABdrP8c-zYG=6
zi!J_8TOat<19q@w3C*AKinbAIK)*(WMDClfzt1frjTw}wnV|RM?drFp7njikti)Q-
zo%YxHa}4=XPK0yfcy_nx&qw)hO{8VqKgzh5gtzCFqh{J@>h_Twf?MXDCiekf6GcOQ
zrjn3YoBQ{G{l^#1?`~aLLT4Q0t53&F59g&VOZor*T^s+mX`@@@|Clnaxx&4N^d9E)
zAqlU_5jojU9<(K!aYXW(UN;5n!66bql=;+us_iD2{Npe~e%v$Sa1bdob0bLej|n+j
z+S+QWs<`?1ZqN3SAOl0g)17JQmZ^qM+6oGPhlU~ortz|aUQZAdt~wP7e0V|5ck3s=
zhn1|q^5Ur<yItDy498)Or66wGMc%3`54`%Yz{g{^BhO7<_ouFuy7KWzzihWIUyv71
z&kyAN>^DkxwZVJ%B2@ZxF?_I(M{uI5CACBO(RZm#hUwQ4EhFTCvX4=|JwO$^`DS<E
ztol~V<06y(OVe150q&%x1^Qc1pg$keGwgmqj*Or4Cx)B3(Fw4GB9Y%&MPt_+O=cAy
zXWw>znp)TR`9ESVsZ>-9VHp}En<(OOdCF0mmOs=jnXZZh8%t71$n@Cn9|X(GVVpmA
zIEf-W)Lg>FW*oDx%KlM?#~fTAn9&3={$nERl_Y|F-5s@yHwH1E3xSp6v4Ao`fyqr<
zGpOwm=KhQrTS{NsMm+q-dRNWgA@u!7P5o&xTBuX&`04s$7-ZSuVGdeaKklWSyNih&
zCO0%S<pL@X&{3uq9{eP*PC2}|>F@`pAd!Y2CPr+}QqoA%GYYF(Jlx^O%S=gqBgW{Q
zxNYhQbskhdGEG+%bJ&qJC5>Iy@fUmE#1j->XWL<Sn^7@-dm}I~TbfnX=u8+N<J!C5
zzx30@{QJki_T+|jU{aH!3NO4V9dCoppY?OQ1dNlp#m8Elw6{>})olt|T}g0c+`grz
zubeCLOAoTrYZ6O^JQK^$b6p${xo)s<@ejAve=E<XbQpGPEcVTOWc$16wAz9`KjW3Z
z9fXmL<oVsJyQ?<K`|FJknnTmk!*r7ZS&5DZ=>A4gJ^#OM5x9FUXHgujQd?0|b^Dt4
zM=D~2kDOl3^%>A}pbz@Z3)RZ)KaoHD?~;;nlsP;2fA#U<<>RZZt~NF`Ev@M4>XMJ=
z+y<xK%8hSGXX*cF@Khs!^*Sp4)8K(t8;`!tx)3(tCFs%9OTWeVN9dLGv@={R#}&87
zZ`(hDfjR4Nk$b_D_(E&O_l>-6O=6Qi8rxro#*g3dtFOm~wXORgQu^-j_mVw-SYm4e
z#Pc+bZ0tTP8dT8xinWa~O#Rk>8S}=bR$$U`iw^oD>DTDyUX91hY~e2MI1?f%{l1KV
zL#)~<m>Dau<+<}yzopZ;07pF-XT%vq(4QZaX!jZ7Ic(SUu*P>9miLzVc9?-O(8zyV
zPc<8~I10ehPLZY~b;oix1#2`Re1O%MqSYxsL8R;q*|$flmIRs+s?VgMFE?2A%W8NH
z726F@McQSRO#2VYxN;H%s=a;7Ku0H)Iv4So_}l;GD}se!)2uHOVjdoXz+^;1qGw9&
zH6ivt-lCmjUDWdGk&|De#+G-N`^C{Zm(2`72qBLv&{}!|4Gav(B?|^3^9yIw|LN|S
zJ?ax{oO?MU4&pI=QZ3*aYjJ6bc}D7&vTZAdM1kv>uVU2#ZUUd&zxd-_G@_>Wo5N6w
z_3$I;KGKDmMYj88<lekA`(!HE^bf;1J4P8TS(?H22hKRjVY*6c{n~r86Z#=?dKDW)
zP*rvH?_>^*r$AsSEtfdKMD`@$y}rRBsbD%1zi0!Uh*1)Ld6iu(2H)X=8SRk&J(5KY
zRFA$In3=hNl$)Q=AP>vS1F3w9h4{sgn32(HU~;2hfh&sVk1_^dRkyn9SqCXKD}&Rx
z^A@R(Zt}BaC#2cx!shoOa6{FQ$b7C>%CeY{!9w`>L^)whL{q|4QbJV6L?H$Bv{5hG
zeR6Ps7Dq|sh>N1Eg?6ray%~SCnLiWKV}`~)o*l&SZ1&B-u4lBYK2jMR9^=zl)>Y0I
zn8CB&KBmjKWrA4$l*``X6M;QT&@}nC1D2pQ#V;T*dFl9|ocK%AgA42oa5MVhtWvwz
z+Xh{sUHIcJ$Hch>#`!`YJhB^ts$xE%m=05ke*Jn<s8j5^8|#%8c5P~tJX2pYQw)r5
z5d5cuIfAp;YRSBez}3v<YsjxgfsF&!r^n)S9g-|X`IGX2<4n{#FhA5nmB}mUx<B7_
zTD4S`Yi>?%mQB~mn>6+FJlP*O9f2X{(={;)1bj~7u3C$XGMPa{O`J_hb9m2z=51Nv
zBXs<jE-!n^X=P+tgit&7e>Ma*FG6epjSCi*yV@fxtg+VV?EHLtdwX&nykaV<Q!;VG
z76k*v-h2n;NYk%uQwg~~SDd4qP|vWEM_e|yICh!#8d6pk7xz$sX%Fb3KnZ1|(m*Hr
zp~FSR)tDWtPA)#d<>$Y7dP^m$&VyxxL^z>SbGr2RobPBt3z~yh($>s#b7?wr={xYz
z{vJHOnmb!V)~<^U)&1Kn?XmmY9ZBf^zYY2Oo@4iS8i?S;n)#2`yIQE>O_o_8ry5SG
zb;t2ju~FZOw4724elDd7#y3fs?3DRWCs6D{Jp)RSuebNw`g&~awQ0{hXf+mSSMD5d
zj+U3h<xomN0q{NS0P9aBr?NM%lz|p5gdH6PUq-%H(%b)1Q~tMaHV!nxZbv?$Iz!wI
zpT1UK(*evFoR=<2>>YWm5EdPDM$A{-*w$4`%~SJH%cL-_#9;d5$IvYKsw%c|T-~4A
z;%u<%{~Wj{k#^m_=u6A<?}$Gg3f93hrQjsB#bOYd3SvWH10}-EjrGolGe9rWeZ4<m
z)&7N8CoJl$4>9&~lsS&W1r2&WQ(`&PU4Aq%VbxxE#+PU~JydTTd-M{J(B7n*cZx!&
z{$1YpTw~D&VtHI5%D4+Tzk`)<5Mzu*N{)*&2i**P{bJYS4Ls(XK|V~Fm>4wG0A{~Y
z*@d1E5eXQ!1<1y8_OBmt{=rK<$s)UT8?hFaGFc4G=IgX*>olrH?%7V+i5rSo8tG2H
z5B>DZ!P^xmBuu$8Gt=UaKeiH8((Xzb9T~X2jZRp1vs9|zpPvP`o3Aoz5mv4Pwf|f<
z|6kJ$%yH+sDZVugjfxGAiWRZcln0`vV{v(noB3)-B0WM%SsmG);k`-CbeYF~DAybj
ztMVimP~%gjBEDNwD!i=NBmPmHI@v9?q?+@Wdxl0secmB^yVTZIj}z+_X0T>^<sRQD
z<HvJVp27b9gM)*hFQ4rOAXqMmzq+IZehNiIMn*Ggn^;)TQ&8N4sO1ts)8o_Tpjb9i
zHqF(Ju#8MAVgt%k^R}{YGhWjk{XN5nTi*)i<YzzRU54|&k|u%xWlBQ!9ucT4s}pt`
zV&F-a1W;wfXMgHvGo35QM%_0)NCL^A*=t@_5{5;FV?d8UGv?qR{~20aG`yx!(IzQf
zK{&>s^|Lgo3DHEnu@a4jG4K6SP<(Vx#1kw8&+ySiHoF7u1W63L&3XUnESu3pN=4ZW
zE5VP$!?j7bYj?HkP&GIGoQ1jh^XJdMS<Pq4#-E(+FFI_FumlTwTyfvEm;uAY#m9F6
z4n7Qw;Fy?apFVwps+{3{17Ts|c&cE9T7}EoXLI)$=G@W{7^L&dsWSZ+=jR1$tau<5
z<s5|6Es-X`Z_o$UQ6WNf-3J$Ek;e*;6C38|kPVRCP$m^EsWVMhc7U_0+^3J3LAT;s
zlq2#H1P-UQoHgC|5-21rG%->A&xBmYfMa6aKIH`-G&FJpPhJ+}<vfC3CzGWcEQ~WP
zZ=HX4)=!x%3gtV&U7D?T6oewwRkj5VcN=U$T^YWFgL`YY??H~>*2^6g5#^*Ic&8Yc
zqxY`-YIHrOV_-moPPQj2-A=75(u0D6pjTAkMS6{L_H{^}`T6--Sy?3|tch~)-dq#l
zgcGC|u=hmxcQ3S98GzXoKaA`hnX<PZZx!{+%jWFdH<Q6;b-D1q@1(FR?n8(kV<K=c
zf9QIlC`mEU6e!)+H>>hiX{Fe_2oTXP4jjwI6o3|x@E-$P_9F-w&M!tQ&8DW?1!ot-
z+G8;yqod9Uux^)(^-bi6A(`TrF9fmQ55U>;_W@%Ye9z(7<y5H6U7sAMq$s%_T(v~V
zqzhUYh*$NU^tmO6kLvY&K{{8E@PtnySeR$(vii;QSFP-ei86`?)rDm!0RFYcpB9M-
zBK>;u$tuw!H*nTI?)cAT@Fzi0_tU+{aU5o4Iib3!tY*EXn@mY4b<9Qi<`gM#RarL|
zcenM_H+1QTcLzll)aSs<g8+$%wOUP9{3J?K%K7~HGo){Bo|}=;>MEhD>jCT<z+iZJ
zaZFD-w0yj<9QqWmtGh#9sxNbNMClKsCLN-KCdJeBv!KT9sBH?~z)dZVJ~`#`B!`Kb
z^!iQQx<}}Si2}Xs5?Q;=ur!L=u0VKb;_2D)B<OClVuVWk&<uaGt6lkRJHjLPd%6sY
zp359QE<Rwr&8~mkvu~<DS;WUif~J?tU_KFvV6d!=G0`)5vQS*Uu^9Zv3H3V=#4zFB
z(k;=qJ%9kRgaxh)pVK8KsD@7WyJ~~=N_1r_lj9cswtLP6jA6~59dpkv7i*?RS+VSg
z4;@iwJt#Twv5wg=QJukZWJ}NIvz{L;qe7seYiC!wi-t`=Al2=W%$kvu5f6-_U`nRC
ziHX^}23r*wA4=&rA1K|JWqj^n;e&KB%~zVYQ)bHa$awLoFJqUfqm<fqYH;@64#yz@
zFYnZcSDsz)`S9U37z}yva1Zt1aQ#F%Ua;x+Zd2^y<SkXWq~$&~(}5BSVUc!wre<(N
zc+}kd{0=#trJ%~~L24iq-pNlZ($Ok%?Wk#?(n<GhL0cjoGe3OJsD@K{lCfgvh@5!?
z6hcrH)baBlur_j$O;;G4<e_A3=l3{psEU34dhZN&J62W(0~SO3OX#z1S%9d-*$v(h
zbfTIBpm<jOD6^0wX6*XWPD5{FIqbR8PI|W_<e>ywURAoF=MN-npaJMsSmIBn0-M*?
zY$FNO0N?hGAYnj2N7AgcP6Wn6IXO9cHt}f{z;S_LF1*9QU|QcPjqy#<wA%sv2ukSP
zdP%}tTspW;3@kyw_>fKGraEP;s-jj6il?J>3rF2lQlaFAing+S^2nt)UD4}46)i0-
z9i7Jz85vo(&aC2T`D(x-hiOrpAxKI!4&!=6;TYyNDs<&Pe%ylqpE*Bw8*+cr67kQ8
z)7RcU3Jl4Pq0!M`{s#L|24VfE@b+gQxdRt0MEc@|;%P`@qYxw^Ej?N=gW&00sjj8f
z_fpQghBWHyT4}ehqcu}@{odhmoPcrn7Of#KkkELWiko<4arc_<Z#v4b{(~ecpf=Id
z(jvSPzkk`O3%;5}$m1S_>VKi`(b?1U^*$X8zEBlLe)}zH8c@-U$CiFD$fmMwyWAdu
z_`6H{-^T&xe$8IeW3&@`lTGbk6CK)iLC|>+cn>fpbi(|>cjy1Mg9jl0ou3)lW4#Kd
zySn_v(-Ax~GK}k~JuZ$O0x=C+D~w#2NKO^A(#s|wYqqux31pr_&r>k5A`E9EgKh`j
zX%XhrY&g3>!lnuJa%4yc1>p`068%dNb>-zTpoD}3AdhKG-a#pUb|uQqV3vGCLj&~E
z^=J(@(@&vYDJ*g!O#y<O!t$ckm66@gD)m}M3?8I1Ndw_wyntNfeu=z08ymsi!byJv
zUEp!wowqkz`2#16;xU}73SuQEKmR$OW4^C13QO9}Qb#y&*F*Z2mH@Q79i@zomqE*S
z-JUmACW+U<oG>F$;c{VRB`h3QUan8Br#qwHysFx^qj1*Q#W^=;DCTUb>IeI=Py4K(
z8_{nx6Fv%xUwF%Ic#!K@;l%MpdJ@A%5PQUOvaY~07%TZH?lD_qaTOj{>y&eGrWVGy
zI3{@c4Hxds&%pnxS#FL4xnEzXiiu?a$I7EekAU5WB8W+^A%TONjcssxT1Q@f7Z`k?
zV*MsWBBBIY-<rn;1pM_CoRmnFa%bwm!>>te``#S8z7l=_X~dxAx_uKRll<l<@y+jO
z0bVxCqNoiL+UQ8z$xDt~$d^n9be7v8Kc}YTEvO6ubxXZ(tM&`&SE`6FXkTDJX!jR^
z&;~rNqL;vXWZIY1gXXq3_Zch{m@=@mWPqL#5+-xorUic7A<j3+e=_=$?B-fZ!i?2p
zavrt&>@ryI<)n0Hr(YYN^~`oGyh~!s7i67#u*>Vn{ryJIh}Fz1<00gHj?9G(K=1`6
zIU&@h9b^C2L`Tw!k-Rb`yR`JxKapPGaXT_U9^NQ00M*yk0g)Q`7=e`-c%F!eh<2xI
z)ou8JW<LwxZf|bVGcZhTSk}^GBv%HaynGg5Dl1DTP83ibqyr0^t1X7Y1cZ6>&>lGx
zk04L|3}kcb1f}HUYV1DOqKh2)`+F;683(l5Wpzt*W?esb?wOek-s%>we|UdZ>AL${
z@g<sQoyVP-;Lb|`j0w0~VBpSvVm3E(P*pYyBfi9aGqg)+)4eK6TS4^4>O7&!DcNh1
z#;0#I<s)DcR?UfDBh1`rkZ!~EOY;>i%5YlwPhJFY$-lRJV4M4u&5c?n=H<KJ-r|Vt
z92^+Wxl@3R4zxC@PqTvC505_8nw@WnBMQE00|l(3hF~_23eJZY7Z><z=A9BJ9o-$L
z2XLih(@_mV%b5#d-n-{IoD8fcBNIr#1JVsg0lg5g*<=*&-bCJ=wTfj%^!uLbuaif%
z__P&Nep~v<O9snJHUcrS>p(Mab<#G%Tx3V!WZeDi>6fPW;dyuG^uddk%=)MmR^T*!
z`SN|1zY{qys|*hhk^Gbdh(bX3cqeW3nC=^Xx1)rF=WY9>FGNZ)hUu?=yfq2i+7H6M
ze;0+1ZpFw#!A?(o*grVP%ksR94G})-mM>;cRWj?TGqHZ%?<MQ&FD3Q=I9gQ))DmEO
znFu17@zRH!agNMRYLK8`VyR<+BxWgj9b&2OHz$L`mmHB&m#Ygp;5_hI9I7=iV9vsZ
zeN*<prmAaXlb%LunBCq`!gZ(KJrOSOwy>yVmgUH}@=605E(jgM=pJ*L?n_ltNCTbA
zeWy<_1laH-k1(xf8E#`HZS9qj@g<3^t?2@zJO%a~#x&x<yL`Wij0p)#&86vb{kG<z
zt{kzMm5F^$UKopagn{1G!3^K?Bz#k;_4xL*wDJEI!n`%Z?XYEv3IsmPU_xMIk~eA0
znt$z7Dej~~4ZR|LC0{5+UBP{;^NU})Pvys=&!v@9Xh#>*JU>OMD)lW2{}~)X<DEMR
zDVSTBD%4Uu7QxfBd2lY>qlGfd6v4BHW-zELUz-*#%KkArQlaNw|HmjSEP(=yCt$0n
zs8t@A3j_e~7OkyfwvC9h0aEotwFE{zy#F1hy49)DmD5_-<x$?s?}=Rcp^dwBv(N(^
zU64ktK5wZjHQK;>t=YQX9qZm+I{%*Ty_uUjN<2|@FVAUczpZzYfz60+pP7>uF6cN*
zq_$yQOys|g-KI3gsacHfjF)LVZK-VS$d2!ez-K*PdR)1354bO$n3b&rfxY=lf{c3R
zXv)`y1~s%(0gLbm;oLgm1L%4v@-la%^q479%Hn;;(Ll*}<x%3XUTvXY1FxI(uZsOp
zR}04ppW<rn>crtsc~ojwA1=XJMM80Qweh{g3%qJ=oQDW=p_Pq4d)xu;<RGMT$YxFS
zYQwwzb`=W=gJ<GN<b(#;0)RvazI!cW#OkN<E~&s%UaISYIQ|}l#lRB}F^|8&zH5YD
z%Da!DTfJTQ@b314h<>$<A?7RxjQYm&Iap3#@{VcvJwjz^)-nu%<@B~2=nCo{2p23Z
zlT0{6Uz(h6Rz%t<{vj`?FS2_r+t4O6IW77v^Ux+4)vw7Q&c(Ho<OUBCc9RvMRcl{w
zF%zRzH8s`S0C|rU5=%at>$YoCZLBG|N;jM%+mpCYZgMm4Z|wLTwwp={d=0T+crR2>
z<M~tEYU^FvO<2=HL!I2h);)aaMu_|QoBmO;e8tmkb;M0F81(O;SBh#tyN?{q+dn);
z^knp94vsCo*(&f&CNm_66Yf)d8Jd(E_T2UnA*kaQzLfqb8}Sftp4bjnFx@}EFyklX
z=jCLhZZr(LYuvE1vaWQ;rh_>LQqm7|zdrZMWl_Vwc=on~M9IbjSnE&b{Cwr-E-2Z?
zgW6gsF$E4e)4Q|TPjI#f!s7L~oqA?|FAV6`)uD%r6Rul;*=?NY7}u^=Hgdkv{vM07
zkk8?50q<XbxSZbc863=?ayP}$lP-M`I2#la7ixD_M-HtW@ACEj7M0ZHJ#gCsW?fL!
z3c_!0G>ScL2eqlSux6h#Y>olV-ullgtA{M1Q(j~s5#+60>$XmF0eq(VP<6u({lAM|
zyzGhT`e~l(5s~I=Bi~uWNE$?`Gs;i28Gy&J6I3h&b^KGqDq1q`Qg4-N;F^zL;v4zh
zu9N;l$K9SUi@$&WCaeEC!KEI>C*Dx?5E4ctstlq9i5OiZy}vDph!6)y%v}@oWU9}P
z&@_HI-J9$;871cTyTMNmNhr)cN<LK;J*|x^sf+SC7~O4*b7@llOalCc(A>%Tbb7bX
z%gJpj4T8z;qWs(w_~fw>ZpU-v%GqrO-^T=8ZaS(Q<sFllxw&}=b44A7@;jPcJ*~*l
zAeBae=F&^k<D_8B{w5Og86ThEm&8L0XDbSdXVY2V0pJ0D#XfEuFzq{ETr<sqIhBTM
znR^hSbT8E}tnPp#FJj#TH@p?XGZs>Ie)>vJ@jMqi++!Hbn+-jVwsRZRib1n}{UJQQ
zMzM+pQqY;6LOvw?-PJt(Z^^*a<$=ExYsq2VmB<KV!id6Hsiy+;%q;)*VEavl2*uN!
zU<n&VMNHsyX{aulQWTc?VYmbSNI>=?^8NcyhwqpQC=?1aGe^epkMD7NH;_MD-;|L2
zh5&o|GG8h=&a|AJD<%Xc2Y>fNf|wq)8dSZosI3cF%0*_&lr2`efJa%fnzo(zG->z6
z&?~Us#B;dtjcl^z*?LvQ{5i(rbE&F^=76!gAi(CS-2(w@WIg6PM)3p(OFlpd#y`V^
zfR>GNGWvcTx%!{nCQ$F@(1O~23d<KI&eXe>#bQEB|Ba{7SEe90wUC|X#Ew6nW~US&
zgj8;7qdY{uOLc0U3_UTo0Tk)#PwL+__A?tHEM?Y@kp6FpMF8eq>BE1)gk}6yF$TzA
zSqlpuHWKzgu?NjPN$ExRW=@cmh(60ufumMCe>Bv*x?R%>!z`0nVT0)~CA5k)N7>p6
zrTEA-p0178t#ju4au&r5#?3j_Pj|XF8nJ51$mH{IAp<jAR*Bl;Jy7YbbEfYIdF-!m
zo^F@}1Dd@5Vmt*U6Wy=Tyt<M{zJEt8)<~^vS?24!48TF6UzIy{xx|F1G~z-eDCC$U
zqhgUDFBs9i`)6(E!2tN6SEykmU08EFr~cF2T8{Q{6Yl5H{J95z50IAv(F!KNyBzhj
z9$%{)JSukAXja*EJFi;Gj6M_3E(PiKt<(MuX7cmogw#xxK2xQcA`ygF%BrtLGLl$M
zROn=#u9P6|3WxQrj`TZ#$(nLKBU^)&jZEYkAfe>Au>8QU``~A899M5R_6(OEkD6l6
zE=qNF;uujLmko_DClq!b;Dg+v%vx@<76g7)$nD~Bc935DLq)BmZwymJttauVY{t60
z1{&04y*i#`_}=+wP<<EloEE|cyXbR`eq7LWNb!tm1rfayeQ}#rD?IHEo46PcZ!c<h
zqaB|lIk;~Ss%YtU^z>vfPzwNmMa|bim8E?~D&Ys^pX1^P7Cw95>t(lO8$+6X{q5;B
z7<V_2;w=q*CIqI_3-U60`-iE5h+pUbDYw1V%)_Ub({s9bhB^J0O!qmm)7MFs_09Jq
zlh)ZqkB=GA`^?<KB`Cte;5v2?c(~CKQFit_$)3Vg+VO-)o}ftEs{6gSrZMM!?ehBv
z<>X^a+=4NelARZ9>7QG^fXs?}k8te-7#hEK0!3pa(CPS~i=O7bXH(6V5I??VwIw6a
zhufJq^}at%E&TPWvZ&1a*;S)LQ5PR6Hi&=^Q`a8r4ei<Egz&DwzN}PNZsq_`F$F&b
zj(@tSsBWI{F?mMRxWZ~U+@Rj{4vf_UyaZr*>HiCpZRDP>7{_XmdSV?F!_><DhK`!<
zYmUgMm?($y&+QK`%k>NwHtz##0{v-HVu|}H(7#^H_XC^Ga-^!Q<oeBvdH)LH<eO7n
zt;h8{nyT>dP)C>JwA+HVmhMA{T}$@=L)KeIRk?lN!w(@KAc%A$Qj!WHAgM@;h%`t^
zw{-I$igZYKcZW!aAaLpK?(WX>ZtnejfB(E^oFQZAaB$9k_Fj9fx#pb9{NrTHnw5CB
z>P$Bs05&TKUnM4mhYwAt+2LA@q9ligJ_I$#GRCjxdTqZ(C~KorzuD1W`mU}LyIi6b
zs_DzUaH+=xQ4gnf4oO5wYInB!hWs>HO1j6__m&Xv1`Bekst!E1XHb1m8L!h(g9_tG
z+{G1l${zWwANcJt+>Pj3xnG4I8eGg#@&Zqj0@pojvN;M<f6l)R#(pMhF6z8iJ~OEK
zx{lP1f!BDB7qj?i-yT0^ct_BkFgT(p+tq#ll7N&Dzj$g-xVZcX;b?566`MYvJG#2I
zSJh6R2>{cq;+Zi?xd2j%1HsdEcm*Az;dsL5+PNnz>;Crpy>r({6-7ap1H%Jt{Oao4
zL~OUs8Z6ro6aW^0F8`nA^B-GWpwYs8lwC%76B|mFoPY!}pTOJ~I6{i{FrL%=wI=is
ze?!o|AhzMRxR}^VSCgE_o)BZUQJ-ox$N7TuZ*Cf@D^&UnePgCBjs)0JFeTYkUMe;`
zyYXzRABaIlW^I!Ni+^?m1q~hCzx;Jp@Z_%B7Heoxl%E?H6SuKJeE3AjdE?YYsJeD{
zap#uyAw*z6kBBqQ%%c&#lb@7-O0Ex<>%{<8Hf!q*x&wB<%Y#2TNWbq>R*KLR(7T4%
zD$*7|67NAAh==O%1NW7ra@{1A>Q|<0&|TS$L;32L?b`d*=)e2j6=q{eLhfXD%|?K2
z1IZ?zVK<4R;~5a1zm<^CuKJ?8F__t{pw;NHv${$ITKM?*z?Y#2a3r&`ej5+a0;U#c
zb<+FwfByLgi<ljVq>S|Sfxr$F&!ER)(Gf;301HrHH|`Kkrg-AszaPtOru_NyXIn4C
zIG^ENB|NeZlmt`_R({ArY+OD}mVA`LawNI}On;A}wCB;2c~WopgbY@e6->CnmFy)f
zo=|0jXin`>&Jt(Cvdo@xKYw(yAak5q8^S?3JGs0=)_#JpJ_JPC9lXB05M!&wKi+UH
zmBZ4zo<Ou=${#n3<c29M41={PIm>b7=ujXhxB0cZv#6mQ>Cs}3D_LQi%d9v++Am+Z
z1CkriqH5PUBVH|YLD&OhqZ`?iUOE729NTmTIt;)TPOLHa0n~{Fi0x7AG2do(xsUc8
zAkVMdyKsY`cI1rJ=TtpYg8Q@gN~IK{%r%}H`2)d*n+-VSJVIZoJM=$V>`-4yVB2H8
zMWUvRj|h#+y_~xzk#w&P9a0jT@G$lNd2*sR79gET*>DI!ip!aay}KW85o;M-j6tv5
zj}}VJ*g;R~eUit~PFrbGWN6rP#YV(T_!K}LeWL)B1p<UU4CBQ_YJ|qbsqJCF$lw0(
zx=h4YrBAiiA0p=Fbp5{dG#9qd)`jG_`Y)wNzrB`e*Q@tNjoHD<3LzJC)q7)V_(fH5
z23d$ta7t2c>vn^(n~v=rTH$2Z&3fw;h@gPvb2AbdcfkxUWYgFHVveFu)nSD{#nXEd
z`Nt+EVk?0DF9hIn*ZZfZR6Ldm(b2?HF8LC{6cg39jET3dbxPo6X=$mWGw+Q>42ZfB
z5iY}XN25ghK;%P2L^Lu&F(wA5`00sHJDm7mQ^iNbr54c01SUs^M;RN>JBEX3Q^4k<
zrh$XvnK2ybsQa_;X6~Zyeh&+idfkRj&YA^m(jf%)O30L;JKn>Ght(-L8L^cpp3A>j
zNu>^IcDhY4Or;MC{j5GonOyWLR8G5l`En%R`!0lqL5Pnxd2z<-N*=JPoo$pJifEtJ
zY!{2ILxSKa_SZE(+36#AQQs1jyzYH(F1WSbB+$A$n(X%(otD`P`H`2WmnYNA%!4j?
zbuVOWN~2O$Rix>{V^=8Iz;HI!`L$iLyHU~t`JS<D-yR#$D7Y{usy)P~)C)8OZl(*f
z-#)qswxVgVal{v;+{5QLu2tn9CV|>ssZx*Ena{<>zoqSH=lQ3d$5)yD>0<}C_RS7G
z@&2<f-L{#w(&F|OH`&X-#Nu%WRCd(i+AoL+jl^0B)D0x%tRMb1gDXlE9HMDsY}Xr)
zuebU`p!vx^Cr6d;-~7AspeD1SM?J|Ze1#e>c79uZa8|MyUL*JcQTen`PS-^ck?0w|
zrw9^c>|<eHdf3$HK^-2Khzr$@cB`cGC(6H5Jg-{QehC-RDli@I_yGd=J9%X|RFl_&
z!u=QrT1=xoRnFq`&b(-49aKF$SirH;Y<t#m0oMq)<KsZi-kD6E5)V-~ekJk)tzLCF
zQF6JP@72LlI|xQV+*+VkE)Fb0W@o8bgzYy8i0Zetwm39?zf)0(md-TP*PjOBUcim(
ziRS?#kf{<~l)_|Sz2FMFxFCB1-KUR_Ph_N$yZcSKQQvHxb9hpcki+(PL<9j+mVlZ&
zaj|0=jPPJ}N#-=@EHmgrdqDQrMhXQPYGIRI(MMAr8NO%?I(O|TP#M2;CiZ!+f6dJL
z6dk&6{oEuM4tL7UH_prNzOhZ{EBAchD#!h@;(I$?x1H^V@e;gCZkh0HWJ%>Q$RjKq
zdE+wnNME-6coxm*CflO*Odn^@*cFE?UL(8cc1=T^O4M0abE)lHbXWcsC<B@eRBn>;
z%Vz`|D<K-L4`I%Xd}jhGJ&yLjYi-6j1VJ1t2os^pV_`APW$ClCC)?8L**6<ZZG%ee
zqPC1R0=96qVe!B8kr68Lxrzq&)X>e!xv{eHp0Yf9p{yJl;Nq|W_8MNPRJ9m=51SM8
zkQjdgNMj15-h(IM!o^B_)5B<`^}{U#cit9LrqO6cw4JBPr<DQ8^LNeR`hsX|`>PTY
zDF+X~gg^q($V@*W+vEJa<0Ta3UbBN)_!=1kyA9Tp;4B*Xs1QcY?PmN%QNTSlB}JqT
zffzqg*3{@LGR-NaI-)b7CQXTaiQz>vvje!HpsfhHU*Tbw8*@Ja1ro(GHC6Z1r)f_l
zgvGFCwM}MhteTdVmZ~ZVn`isd!MM8nlPBp-O(GA1sA~<PS=Gw)+Jjqsa6u~)coxjT
zE#LBoXWWsCk=*XE+p5}HVb{|Fmerx;gQEjkyWfpB=OY0rzOj~ii;o&bvtKr$cuqA3
zdDE}FKUEGXqA<)MGdWtyu~i?ez~SB<T`*^RS4A4U+L>r1V<=}h?yagU@4n|u17VMj
z5fS6BWD24)772b1o;3;9c3%fAYG1g4D7yXx_qF5uQdY-br0S;=)DTdl`U+bWaB~l~
z=B?kqf1i{1^N)2i?eWH+T^Dva&mm*89H6R6yK#V^XCG{ulI#OY#tb64l`8Df)X9L%
zA8Hb~NE=_<zFhcHjxi|95Jcs1)Ddq*T3sz>@XLI&SX2)Z5f#(*ecgV$2zRnEfdod>
zA}X1?{EAm3ek427ZQ5NuQ9}_8s`;jPp(cp?tY6s(W1jRw?(pH}Wb>kb_w}N06`lDK
zB*UtLl;ZOcJQ%?rZ&DKkU85emSecmt=Mz&?(|qN^w2)aO5nyXEFfcGR#m~s-qwECo
zg|swUh!-es*<9MU=io*}XCw>g<E5si-aPeO+n{D<j*Lr8d;oPtv2CSY9NUgN?@7&Z
zvAYOoea!Vg<Wx3nOI%0qQl}=pqQ7uu5FNQ$6OO{yh$VMSv#!KBW1Sb|X&E4him*C(
zB}0pd7OblI?J+U>&B6i|+W%A=+J3q{KZS9og=%XIv>sN1u))tkijC-qWdpFpdIS7k
z#~yGDc~=)Us;7z_dUNBh;;r+e*iWkJ_n|z0RtHAwl)*oz$2M3pe8){i0$jz$V;$tZ
zM?7XB30!uIoRus5Jd}0Ei<$&JUx6lyUr-A5^~3Lu4h|0A&~Xi!2#AQ-nV6VZSwGb}
zupQAyMh5!8sDS%|1EJwkZ7(nX{Xh?pA<*r7A`uKoW{xujuwc?`*h?NBd<;}5G<@q?
zc%+f=O1Rq~Gc(U@tN2_1#Z#*H$*bstF1^XWo}o<0KVLZXPVy<VJ>Yt-7#hjWR*Z=e
zgc%%&0+D0kPY&2onohg7)MH>`g<LfEe0f3J+zboJp)?pRzbCQCPR)wA7CgF$KN}~V
z2O%hs+@MR()gf%iMqPbMYHR=b0)KS<V+^ljIw7Iy@AB}p2hlMnoy(4wFj7iY0rOLC
z@>f(%_<nTTPvldR!@Da3R$6S?7p5>MOz)km9lKr327Dvl{p`AQ@7}o&iLpsO%NF*>
z3QqVrhX1wQmD<B2dZ_<k#|-QxNtMCTQOn8)k<x5WpMqKtbTl4dV!GU1a*KZJ?tTZ1
zTq3NKi46giR@YZvx$x8znCgUuh8mliULLLlRB+2Vx5D^9)q^iSbs>P&#OYzrju3!&
zLrP{#%zI<EaDC|pciM|Bg+0Wy%}n6}$F*&t$*XK|e0LWDPbwxUgI9a2-McC&0_ENZ
z$-U3yaI`12?lr^O+nLBMvY1bMFy$TzK4oK6e)!Ig(QEA?+e81(7b6iR?R(7rHmhx)
z+?}MKFw+*$aVRLODJXWH|4qL#&SilwqSVyod&+)iB$Ns;3mG7)`5~Q@LU<Qg7%fq+
z9ve;88%0)7y#_bjgW-{p2_<)6SOJd4&!0a@Nl6O}3%BDyXCA1PdNogk>(KUrPM{9(
z<|oRHz<tK@`0;?!&}RE|+pBW{EC^CFc^VI&eRpi`wGUtnp>!U&l2UijvNjPT|C*DU
zmy?<L%Jp#EI{gQiIjp?>Vaby0y^_>zU+I%X3Dpklh~+~niysgCP8+RHFI!UfV+Q@1
z<#lyMXI__2xA(?#HpontbBVkun0bj^9gK(i+66a0wasSmqf*65@f+cygY%=2D-tM%
zw=)Z$PoZKYvWwa5V>y=U=<VKZ*P7K6bl(8M>T}!>+pIsx=LK-xXH4(TI!QLc%Fxth
zDG`!80}i2v{SdEhkss=uGg-{w1ZAk;0dEJjJKh5&ze6a@iiA)0(mW<9DG4wm5))^D
zqTh11PRdM*=Ak6rBM1WvA>J-<RFi)M+$s(Z4nVF5a!6thtrwh}n_(s=D1SoOOIfK6
z%+w;AVHda}wp7SAGBlK?_EZ$xo}GH4cg3yB92^uS6%~bC#UGU+6hB&+{^_8YZrGeZ
zRr(a%C41$7PFcVs36x}i4=j>nQ*ojd-hc<E3wGTSC1T(C_SE2?_6>XrVa9CHcHGbw
z)mSodMz^b|pX5g?wLSgGgV7+E)^f}mtiXi)6F6IO&M1&)_0M1{owsUAJyW>JZG6|i
zs)~Q5y*fwHL#r+GCxbqyL>UFo<|=MR@2P3>^4zUTE#LuLPW^rW{lF%*Ip5bMCMLd|
z2QvC(PV@jjzucT0;B}(Yw*cdV?E)4MEEjkIryd?aHms-^85>ji_!0hm_Pi0K-62nC
zB^|?~$exY<@fY3w42ekmd4e$*FH0sky>>CR{A;`GOzT6gq8Q&sZ{lH<O;Y?>dCFxQ
zL*LTaZsM#|vgmH7JEEZOGk_%ZW(|(eV)#ss{kB5L$>MRf>dXvt(hFz;BA2Exrh>H$
zZjF^>MEL~Rb5k<X>fk(O`Q64kJ5=O>Ew8?4t5`_n;$q$D?(2;mo}QvX^6~@o^EzT;
zVn5Wko%sbph6b`Vke5Z(Qwlis_w~ui%O8y7tFV$Ei-E5h&LEox{I)v4sR1vF$#jTn
z1ON1T2QNNGz11-NN}rhDsJ6&$@$aKZ>7eTf3Idpl{A$J3SxHt?FFbG<5_49Y(iLsv
zRh=lCfbp&uj_OXXkZ9WW@>UxeC9PBRH_$*j%dT3kV_F+DF<>1tFztnKT^NqEoq08L
z--|}`pI#f^Zjp|-njdDk+DJ{P67jt&?;^=)N=+sL1+y))42T0YOYA}Nr69VaEU;j)
z3^Q&kyjKuS^j2uMU+AW%{~K(faWGb-03_Y`L{`Z=t1sTlab=eb?`aNK22Q7bJ1L1t
zaCr8;Jkk4XNQCz13-=S@En+#Y;ei}8W9gpG#)EZxOZnKU97$W2hPu4dlPeOC)FCZS
zQ!)6{V0c|W8jt-lIXT%qz#l1nbA7=j3NP$0w6rug3%OqsP*Pf$T`6>$VtU&E6DnZo
z0w|e|TM4ct#Yh5^XTirZo1KSp?7Nz!M~dQ>c<y{SKJjb2+ZERGj&d~YPwo3Uw)}*t
zTUE4_G_;iB6XNVwl9DT82X19$dV)@BTp7UCA$E5_B-H0z@d8)gB`y!gp1?;j23(Ao
zSeV;<?)iDYZ<`a7N7>0Xuxo6-q^Cf_#X~m};F)1Y)@;mYmUYGr@g((kuDy47$uruD
z>?Du(xS+qoVZ7O7sV*4}xa$<o+Wr_N@o8a;=d=D^k=vft0#e;8Yjm?Ai8bJ?r3j)`
zXi;W<{?+Zx)%Nzbzdt4noL|)Ws6ys|zuf1~wATr9`bVohcUXCTzDB)^y@-e>V5}8x
zg1mEl@DhMt$MckS{8}jIZEbDrvFKjMFVQA9|2WFk-4DPDkY!~pB*1(0VvurKsBKnK
zBE8n<^x(mvfXh9Nr_2n@oTld=+}R#Hl{x_XV=vbEr*($~jQZx&$GojDi)=@PpoN#7
zV;}nK(1w3TI!<Q;bfD14sOy%5Y;0~Sg*NqAQmuw9>cr5GY6=`2PcpM}S3g@->ujoO
zy$@WP?|XiFscs6AOMlsSt(ADkFGh%Dsd>x>*UKxDkRTkN@Q8Dc@bv|@t^{)*P5Af~
zOCw2qQhjCzJV=TUu4AO8vRHg-^zTbklYm)TtCx3o4-?0MraORu_MNVQ^v`i;60qPu
z0dJUu1h>=eiG&0SU^%tAx_W$k3|d;ie+vl(4FEmcz;R?KOZLA4T_%dPw4?-F%`Ugs
z)y2ht_&qVfrarvWH|z2U>?$uin9txIA3qD%{3YE&@b9%GNceF|lkxwN2^#Z<zNk<o
zeX?&3_dL_U(L{w_Bi)WrDU7ivFtDF0`1fA6t%4q>mg~A4UD)V0A3Zw*=j=chqmeI-
zWbvbQ_dtV-e}ucG^^!{qIJt*)ivVTn`tlM6to`e2)}$Qr4Q%*TrK3e{cn4-}dy%w)
zte@_O$B$n-aG`qsafz!8w$NiLD9A;DKy9>;9FpbuyZp=Va-E63+n|6mt54l4bdk#j
zyzlPcUD}pF&H}z|HMo;2t>z5-l4M!dmX@Y!?DI1-R{}}7m<T$7&TV>TrZm9;Fjzn;
zIQz>)!2~wh_=<wQ?#ji()98A}-UW|c7VTorOVNdSLYH>XM$EL=;@`bk*&a4C;+_(+
zSHK$0H$I<%cS80{+176LN#M5EVxbQ^l4odW%9&qcy4s3E6d6r})zTvwg_cGK3PuH{
zt+Qk<o>3*`P(4dO8jQg;RPn!*^#Cyfrh2(~@bS=`xgzt)j87k+80OrVG5CNuVb{1d
zRxB6yVq>v2U}Iwg7-9it7@(1DjTUk}f1dh*iAA}<>3CgULnHo28=U_cxI$Q1S;32c
z{ea6IFf9XYE8s8&%5g##<>T^o%9>}Mz=EmKVMkq2Q4yHCfQ-)jP3p&70!-Kd)N%nt
zY=Z-@jVvE0WMmjw2|EcfEHt8qH8;OG%6>my@8MT9ZWUp>zC7Ync>WSn<=&Hv8sHBC
zb+3<HM`dz-u9fkty_@a+Q9?oPcJ1QaOeN1`?eDdHj^L?iplz9epbIq(HSODD!O}$+
zw7hvK8OX0(>8QNGVzj`@zzNIc#B|k$_UAtrZ-9Rgc&*w92{o_)VSBZWcXelXcXxYx
zw(ZfjMh%0fu-kb_bhHW}D~s|)N((ye&jVf%KnnhpBq6;v7Wzxy`>8s;Fi1pud{n&3
zV~&GiA;0nL>1BQJjjELMNI>OJD20i!-eKXF!<3cU^L8HiSzT%M%58w-lyYx4S<^=C
z@SuNkg`w4_kIdB6yUn*j3T;8=m57l0Pd#4$jCB@Ma~A6<EbPRz`KB0YDCa=6XWXJ&
z`R47X!3y+Szd26wIQ3yAg6<%D9`?k{!qn#JvVyYJ<{hJjQomM%NGOGsgTHywZXnPP
zZ8A^|t%UUqo@I12&CfUIbK6iDbj0)qVJeMF(*8E06oeck5tfFXMM`bNJgzrA;pa43
zP!LA<0#&eK{fl(l${#e6xF1xMN%)(YnGT0K%tHRr_Xh`-^l8U+*kNdnlQ~M=*WWeq
zGQw*1-e6PFP>$R~y1rYk-D$u2(V-zOPR>L4>@0f(hvZpGY3Tuou|YvWfKj#L&YO{J
zVs5Hda{<h(--m}q)o-n@*lNYQIw?24HMSFoeS_p_A7J!s+to<wy|`)XbO&RRP*Wvd
z+ESy<`29I?8YgpshtU!KEer2=)Tb<u-_wSV?eSl|pb2bv#!}1#-gKMgh29<M0l?vB
zHL>B#+JzXk@DG`5FIsA?aU&}O>A6GA<!t+_-qB(1e(5qS|8gvUpxXIem~eJROO&jg
zVMJUoV_Q>NeebF)lMunp=DL+CL(1n3xUPx6ZB&$JL!O|hA6Sf3oRs>_@o*&9J71Vz
zzJKKo>YL-3qpML_&FVEXpi0Y!aYMA-CAGOFl@3-*Es$?41?j-P^!nmA1_`mTSpenS
zoulSmO(>r(4g&evQi+;}$1NZP0?}@x+r?qCHx@XI&tAS<wSJqC?7UB#lb01N`pt*K
z%b`%CfQE5MPvKH2Z3*6{A+hLsMTlQ6-{O;4bqPVdsQYheP2hpm7HT1Sl%}6MaF30x
zW#v@<*-hy~6;UzkqAL=M2ZR(jpx6aw)*ZbZ!NCZZ(^b76O(c*On#3AW<fch3M=W1Q
zw`BJ3h7GWXi{GS4>K{(lXD?~bWKfFUupbUxKhT0`n)-w-7MW!5QII@t&m)c{ir?Hv
zT}%ydR3un|W6HOvYi&Z_rB+ps%aQLfv9J*m9j$B^GNK|MT((3Pi>Ax4<YfmtCUExZ
zdVW8j@~vWVyaB&uLPKs<jx?Kzv1h69F~(pxnr{G_3$-f|$t@iXYh_76U5&#$bpMHF
zkoA+owJJWYjAGG;;P#)xEJnJ^YjPFyA|e!ZbYRe_V6a8u<EsYFb8)e;fI*iCY+-;N
zSs>W=m}`L=AA&qFJ@qDlYn*r+!}~<u8b+fN82~V#c*f-TX2m^<j@I#^LlSvy&DVuS
z$LZV}yp%d{0L@pJOnBukRd0-5NkI^(|Ilao#2Pu_gFq?vqtP{j2)FZuZi<yvm?%;`
zj!b~J(aN)mM45BaVLVPSI|=Q3q_W}~QPev2$9oW`_oDqQIg~+1UnjRsB4my+>J$}!
zWX?kz5~QWXebQbu9l!mVgXw%o`-48F_pM)zx8GOLl^4#o0#5y6C#OnBqZ~<JmDJ)%
z@88EvsJZo(q;v23M?T_e59MbsJVOT1o4^oU9(S)LJG*xbcf0R85bVBJugD<f0@Xj9
zF72QsV*^TKAQ|iE=n%jelNlZz{m2xdEf^pKq+38+@+P;aD2s!G`46q!=m0+hf3fF6
zo=0w34Gxrn=Igj_C?@t@<G_MYiF;fLoN!fVO-;?rlBUroDzotScBG~@`l0N=QiCyD
z85yiXe@Qw#-?w6Xu^af}VHoF|rULFVEco9NyuzlvK3*1;P_yIsOqJTOBh9G?Avh@8
zoX}ae{m#aQ1HV3^h4yKAnV)1ogrfT5GK3{#8<JmlmgTR?&ceIzUrSw7pnGsJ2B+<5
zYpQuP)Z)8e;Ss^#N{Qq0q{|uk233MCTaJfMB%si^STn<vi46@=<C0`#Xn1T4F&RrU
zlUEcx6vmW6ZERs-@Lv7<F)rXYgY5#YcNIC*4RL+W>KPtpP@H&q8JB)+Y&C<D?`;2w
zh;|CjL9K=E85oF7<+DZq_U+_ca^N##OG`9}0taSfU=Y`4fBd*@)gOT!5vE_2r`>$8
z0C|Rl6pQk(H1(lF`j~{Z(g4n60XWAb{0BW?ww-_SORqm86O*Lyy2M=9w))E#70^&0
z-W0c<Wrki(`eWwHp<E<&vOGku1WE0(nF~7v$gb=OFSXu99H^M=E<azZbKfG1YsgxI
z+2e$MHo)<^qIBrk=ha8ldK2o;lud5=i&8uxj>;eXzbYzzO430O@$eQzbAgEiXKY~m
z(m-EjW3xvCM8m}7J3Dcv!p6q7zBU!2yO()!eoiuL_IwLQ&whOD^kClm`}er`_$hZ9
zg%2MfJJ+SF8VA_op&|GWip+#QEze=MGqf$U_Q6D|_ORU2+r523(7NelV=XMSTD~>(
zqqEVoWdVaRI(1d<-@oDgw9}ti`-fUculUJQ*k-(X<nqKS-a`TceI6!&78U2QF+J_;
zWWfxPb;GB^g1Utzp#H%Ry6oBaT7bw#RY})`9x}%L7#aB!jUg>94eiyp&v2s~3u~eN
zy0ny5)C?W>h!RO&4x(sdJxoIs@_b9Hi&O_3L_I~KDPQ{^Q@y3d!<rx9Z}jjJq0vFA
z^8_iHMp0ukveWU=PO&j=OMl2YPy1H<GeN<6U@+3(-*3<rWo-=lq9_nW)Wnm<LX+?W
zOQB;UO^JI=e`vT3SdQjk+OyQWuxrqeiBfj+u%4h%D*F8=+h9$PBZ6>+B+pCOZFQ-o
zns4f=KFHkEWX5rIb>$nS<Z5!P#`?{%>^1-cF2tN(8lgS|CiW{?YL>cBA>$mTGP9J_
zf`?Zt8d~yCqZRH!>MjD7c8jAR0N%Ybc;>l1lL3z#pyn`u!oiIT+P7_GvQAFt0H=g>
zuAPb=89#0$G&}kO)O^n8i);a}DE5yIC}*lqqp5o+<Iita(^X2f8%Bn?r5`-3<m6I+
z1~k+0sha%$g_(YHQ<4xykiYf|JzAdVx{$Vykk_{gQc*%2AO6*9)7fM_5YhGU8-;QG
zW6t0;Rel+K%bNP-qwD#Lsqt6`FXU?mv`KX`W{_;K$p+1e|8)^|Z#iUUkDeL){jt1-
zqX`$Xuy||fvX#?&7ra1X0WMWEQ2YZM86o)A4Il<U3k|H$frE()=pU-HMM|HXooUxO
zJ^v}GsGtCHa+Z*m$2Wty@XIi`BZ9Hmi?p-#t=M}+`GE8H6OLFy5Y%=B?Uy!hh<}xo
zr^W5RdU0uQe9X$T7G68%uweZ+-3J#KW6CoPbMa7vzRKrW>fV&1)Pt}-TX0T@Z1gW*
z=px!Ok8UZjAfL*c6;4ON*f$3P4)n{n)eg9PA(z+opyzgn{*~&%K|bk0)%%TIhe$N6
z45|2);&K|InHhC?B@y<MZs$Dd4N*TZLr|V??P}cJUw2oSW$V^AS!-)+7=}S3{gI#D
z;@|t++}yy8u%L5vR2g{L0Gpz$EHs}MXIEDlney>Nf0{R5$k8F@XJ<m(W``uAnz9d+
z=vlayg1Xk=wuaN&mF)J!>VtLo>{?amg~vMeb>Bvj0Fk_`nr>J)CD4D3B%OJTT3$Vg
z-W%b5f!tIl-0ts>!8KeHSkkOTvSsqQ?}*C=vsh54T;pr+(pV~BH`ARjmyM~Sn_WI6
zh9v1YRd`yGqG~uHFf^fZaVo|PoNIAQBxv1E6o%$zKqC3?Lj)lfOPK+e^3@BF`2Yk$
z^|S!g00q~;DI0LxAsq_~z--?EF!`0Lq_ClxiOa(LD6YpCzSnL;wWDSGTtEaN$(j9G
zNnTtH7Y~)K@t328d~j{BQ&=CBO{gk5cdR)(nbZ12aePa6OS*^b8IkdK6<#Ulb*w!p
zu0G^ML7HE#K+h!Hp%DD2EdK!7?8L+@3rwz(&m9K3sHs{%2h}(dWDy0KEy_8O)kz7C
zZ~3O1NKMUuJ=?#=*q?zc<loIGS-BZk*a~zC0GE}Kk*Re$SOmtRjIzK2*?=3UJ7;Pg
zuP!cH|Nb2xAGhm(ML_Q2)vU+I$oE8n34F4eFK^u~PFkjvArbPg&#Zu?TZQ_WaF(U9
z=>$0Z{*>+i5cosF2yb_CG-({2kqMZN{_kmlHY(J^b`JMWFHe(b)M|bBsHLSOqy3)d
zp<{C0?TQ8(n3RzSs|wnCr59z9K?x1FyzEaS8<-DV;njhzGe`tHYLN0LQh6orHgNwh
zN>GDhK1LXBC@n1o(ag`U12jH?v29E|HxG~1c*$!81%)?nnt|OEki1B4N3p&?t~q3)
zVPh*)%2&46^}#qaGgvJ>7c(#q*BGz()*HJ$=x;UUen|FH@|E?;pU7G;O~Z#2-oy?g
zqtP1cj|fY`r;7I#I|1i^?bITLEppBL^1HmF`?t^djX;OQ*XQ1ik(J^2g8ejVRu0+s
z?clG-+Lx@cx;b1qMXArZSr8ywtoF`a{7eg0;xKPgC3WBl;u$r~?81|wd*>I9i25&o
z+EDBt1SudU`MO3rI)^dZeF9)a-0i&B7xJ$V#6RZrtiWpT(ac?zezzkfwa>OE$^pWT
z@Anp1$^t<i7+C@^)vL3;J3p{Lf4+_ul$7-U>u7NwT3%Us^6yLZ%^Np&cLV)#s@ZOS
zyC@E?5@ltz@ME_88AH{kiHfN11pN$t<Gv1Z;4&@?vlW7e9+P|&PqLXXSwH$$B>u{B
zqio`4@22HwyKI7QLXAB^=+Bjuh@s^|o+hKo3NMA_gt-d3+0g}g9i+1{`UGdxFoeU|
z{`yv@;M>X9p03lxY5*U&xqSm~W2LZ1pMk}4l7Qh`#u^J1-0A;C%t1$Bm&UGE27i~N
z+nuQ$8yZqjRjta+H38@s;%W}D+7llg6~(}Cx<uLsEU7?R1C*`1z)!rWC@D4;jN_uB
zqG|xlO3*`hoWqufLTH3Lo-*BcLOiEAUW&HbAdu!qPeSs{3v+XP*D~6?JX1fU`&@fg
z+{0hy8thGlxqiCG>{)3{u#ttOfc5YfSC*mRQ<H-1YNyq|X1H30D}<W$h|d-61P*K=
z=2u-4y@%lOg8Da-+a8?Mox*pZ4s8|UP<=Uda<FygB=A~R4HyGe{lC8U|G%w$IA~Gp
zGXK9jFjR;cm{~S7%)+Cir^)c`A89%|I(Dr(5)zV_<YdIfk<s?!Yd}o_En*HfwmSt4
z3kw$H;^G3(g2Az|GtgByJWMli*t4l}=+hSuWtZ2~l+FAIvY#4zDXDDwOv8F-YW#hn
zSD*<mN-R|E*0g|oWErP4H#j`}mhRDjF8+;@o}QkD222iNQ}BKef!xohaPj>9US7I@
zpnhA#PYF0=7^rVVMNuGMKfeJg>c=<l?CmSS{DPdE_2MLqLAJcSygM&D>iV%wWuW~7
zA4t7SzXPc1H@ez`$cYIEjvEm-R-kf(N}P{Y!x`mz<9UGg&Zi+ZJzWo6^+3EdS_o^5
z$#T85YIcznRXEaFb4EqRg8+Cabgs3~uWCI`2aTVcHCJF|+gnz}hiE%fRfBB*d7t-p
zRQ@rkMv1HsfY??a;Ok0B)|2o~OH97EqnDFj5WUNyEi5zlu}{x({U-n++_ZlO-DeBb
zw7j&0Z$Z_;KQ@b#2#`@}H$>mSn`433FOlW7#5+f)n?$f~KK*;kZOGIq_!iD2^#rk$
z2cl+(eVMKt8yrOT1ok|b-p%c2PnKMP#f6OxCuoKv#v90}fhJwAos&Tmu!ueZvj||k
zy~kyQes5oJ*?wJu6*yXS*QEi-waq22(N4Uv_$Bh=)2_2hJZqR^R$N(ejB^<0qY~MY
z&d%m%D+|N#EeoK%%E&R(%c+nUSJpi8Wa>*5&UH|~tAGn>+1_?xTn(D~eE@&+PM)rx
zHtJ_)dMy0)uzfABeOrBAQdg7I&_7`MPwI(n<Z^%e&cDIr_<_9GGTNJB)%Be`78<s}
zucx4aC*XPtn}7TW$S3XK4068!{VC4hR&!eF>YK~UFI<j|3)Bn^8K9i}e6CM`yNw2s
zlA0$Y4sh<ima`7;O|-}B1t!luUADq=+~lyEId=@~026ad-JFrJ(J!{8$=jcci*JTJ
z2-Qq-oA|SXKb=;eU0cYoP_Ym^dX(_o<hZB;3(m}(<ZCQ+tZ`}|+e$cV)Rss30*PdP
z)b@UxX|UE<XZ<T=Rz>@a_gUvrx8|yatZwqhEDPS-<6~v|>aUIZpZ|-LlyB~t7Ek2f
zEY-6||NZ-Swh;y_ifv%(04}=_WssCz12nBasbg$x3<WwG)hR%oDjSyl0L|gUd8V5H
zHA8C3S!j9mad3+~J{etXTeW;%rEjb9gTmOu>xFVeEw1kK`)_Ie8vK((`kIIPdnP>6
z!VCS=C~B$b{)&MKMGaSb-cobrd8S(iXwq?Ho4$Sy%jtEqbb{IQ`(l^%uY_-cq9Rr7
z=C60cYe$dw3ql7wCt-F${WCdOQ%bC-K?av&?;_vF=wFx$WAHvpkN5X+Iz{Ig>3<4R
z?G{wuT$l9x*x|SJXuxq?-`XOlq<jDY7;|%TlaG%N_(B2?ZE!&V+eKA!0V-Sw1?`#8
zF=$eudPWho`FLw$fCskFT8pcXc@F}h-Op&s^K&KR*VY%&5pZCt=rMdUd|BwVc$l)U
z3ZS;+a0Sw^jsOg-ppuu={DUra=|KtDl_4*@y}uXq1l8(h3GSQ@@z8O`ihgS^7CkG{
zW_|kgBgIKh`&aq%m+xI$XT=#Kl}DMo(!G!W+j<U<FaExKw3Szu#qPl=m*%OBHpL?!
z(DVJ(@YD@_br#L83GDuXntba%K7)=on5hI<HK07$1hE|?Tu36y%D=&|q1;@*1)J>&
z{#WP?eJ!JX0Reu2)+I!Qcplfx1W*#X*l{-0kXx-J;;H?%O0V&;Cc9v5lkj;<t#4I^
zV%w8xbZKecYu#PWW|u`{4eQc|%j-JLlLxZ!nN}TOBYad=d&0o{`E!b(`=Bw4f~fkp
z7>+sKGZDJ5#m1~4GNFQaBmZl<T{DI?GyqmBNM}tkZ6(Fx{&!7hXR3sgO}ZH07mUU1
zAT+1mP{^gkk^`VD5(ILvbg@0?8Q6i~3IoOjzyhA1lk;EG3E-Oq`v0aR?8SwJt)~1h
zjD76XAUTF`Y5}K*U|vSz-7Bbl=>+sQPEIV$%>@JmJl42;{H0ky&_inaF8@1?wmJBv
zC}?PNzpOlO44AI8mXc6N5o^yrzvpeVN@%b`#9jG2kLjNW9~I&RCk(E07Ove2h<z2u
zMw8Dx$Ox!3J}JZW%*+MZxi9VOmYO&4RdIUU5tFAwaT@!Tayx^gA0}+}KY#i^T$Uf;
zvcuL?3hgNW;<7tN9JB6BtWSyYhz7wg5H{FO5N@|0Db!^m*vwv5?Cn@?=T!|UqhuJL
zLr^pzVoF>95It<C-C!_%@E}1*ceq2-tOP#70HJQeabJ{>*dnw@Npo)8GCr^7bn3s9
z{Mtp)bGH0mk{K|=2P$spnf2n^AIt#Q2TKXoG*&uu@e6Rbpof^A>S7Cmkv~WxVq)IW
zPi;rJ4+4{u=YlEuKPo6}U;*FN=9ZRXU;}o%J^%<<vcUX9rft(U2}T#wxl+##qy{pd
zH(q+;>cUtUEQLA2ve%YecF&@3LjPH|ckL5+;3TIV6RQ#tt3m(*LSbgdl8jb2(&kx#
zK*)*KijuGph3S)rLiCv)($enac#a2646$Fs1WuEaG)oN6o~pmE0@byR`f1cRVo<4!
zu_nB<XbtYJqb<mT3FnkG=3@5Q%B#{0yZrg;=G1+{I?B&~obo4!4-r|u8WyLy{`Br%
zli+<PIl7-k6qCqM7z$lyK1rh{_6|Bu==x|bZFki~IgR%z_rGg4$a9>8(SQQ-2_Ss$
z0FJ~aEl+<+qC!ufK4mQfI4uq~HZ~4U61N$Vd-?o4tOGsc;n{1Da(E5FfYJhN=yZzd
z>FB_w16tR6sHir!wg4w5BO?QM1Lz(BtA5aE>b?5nj`Q0?OIf{h0q|!fRT!^~kaVGW
zVuC;J5PUL=#rk*TgHpnpkg|;EJ)f2pq9|~2g+;0mfH4FU<RLr=3KEp2G+FeneAu)b
zeJhCk!l%an^Wa%Q^dD+zmORn)rR?HqLN^2}Th-1dgapSNy#iCXrue&fUTI}@&h4dW
zQ263wfJE&Y+!Fw*luMlFHAqO$_&y$x@?|gkD;?*>c27IlsZTp^D!DT7MYZb@i|d)S
zvTTLN-XMWp*SR@JH)jYWD}wWL+g5mf&y71TGAkV8{Gzb;uzIpq6GEzMVQ87rb*Lzy
z?kXgdko(h;NA{6!yP{S98bO;ob)W_I3s-&i|85_F)Q_%mM3*?=)N;Rkc?$v=NOk~p
z`C9vAdx8rB08Uebovkgu{pK)eYiX#f|NZMB@Hsax4-*Ru#wR39<nnoLy!i$o2q5*+
z_-&2^?%7ztG6fp*Pj40*jT`~{26VVi4iEE+inMC%aw{tAE7=*>y*X{M8V#0_J+ET$
zkgtV5J`es)&cMzvIPG-?pMlzxFk#GRXxinF4Ei-3`E57pCmGFq7H!U<5$E1HzI*fl
zcIQHe$LUv;j<Ny;H>)wp-6sLBiX*P!%h1=p4wI)Fd>*gy;9&9&Sp+C-KyO#TdPmpv
zAHhqmwtomZR|##6<%ZebVYjxvJxc;mYbN1HQvO=8HYa}mpu7sf!}&&=zGO+d%#5{u
z^nJ?uiXT}~fPw%)^It?VF)=Z)D@izYhv9we;Fq14Wz%k4pm%gkX@t~NN|`T^RiZUG
zSE;{Dz~FAhc<p+4d4pkAJ`tXnd|fyG^^dUzmYChLQ%L+P!>UScChr?zd(og@$1iHE
zg$e%)$<e7FQ!7#TDdXFK%?q&Zu68~$#KZdu^3a2$BYceSGSTdmLT(_{hUdWOH?x;E
z%aY{4r!OW3bTzKd&oMADfB*V*H{qhN&;pDtKtXp{8fc`H%SYU~Tms<^xNTt=5k$^2
zD6t38Ff-R!&eVXWJti2Gr=mj5XY;*%;gVtw4uO+7XMZY#i51J7&14XWbfrrxlVWsX
zl5}PMY13(>`0$44r#oUR*l%diMXNzKv1tzTV47oK-FB0iL?sp`Ce(C*m|1IpqX5T?
zhc!`3dvW!1NPRzPMvdXZi_1K}bn_Bo&>gFrBZIJ?_czQ}Jf}b34evrH>WTTs;CPn9
zy|2@(vZu296I4?1EOtxihbB3QP3SK<_L}`3e}Xv&l{JJCclyLWDK0n7*}B>Y{clsJ
zoW>zu8MXzIf*omMX{lB9CH4FF@4tWluBti*UP51ZrvdjF7=@Xzxx2Z6;*(m?g^rQY
z2y`Y{Sb!R}C_CHn`uu?X<~`@)$$d9TNeqZpt8TovH&`=%p~Z(ov*vg(Q)UmItzWXP
z5^CpQLWW)_60&U!dZK`-d4kF-XD@9sdiou%@CXRb58L83E$l5&YV!zfXExFu1Hs}Q
z`=#$))%Wyr0B9mA;^CehqQ&Qd;d(NW*pylvr{X|TjY1BB-hcbJA2>5io^Rkw*c$I&
z8Jk)nesyH6QuE^kyA(w{gz@W7`;!+u1p(;6f?jEy_td_N;OD8i+BQ2Yozuc=DE3-`
zLCK!JnrkYW#023`3wTy{?e_f3WukLxaJ{@V=!^izJu4#v7;TV{kpZ;X)y0L39ATI7
ztEh8K4dws11VxsgKQ97q0$_UJQ&ShGr2#5;gdpsCDk&vpzjcRIgUO?i&iVdALj1(!
zBsk~y&91K`rF)vg(&R|j{`s(?Z}gObd2rAlrhH`IZA*A6kl+eh0Cjo4Q>s`&OObW8
zD?gOUfz1bJR8P&)z|d}Pj}2(cB4f!me@zM33Vb;%3p6V<VH*K3Ro?meq#Q$gPa*u^
zA@R-SDGr4Fa8Lg|$E@qR#d;-ONje=OJqW-&4M8#jSwXH;BnRAVwLv0^N_RZzbzTYb
z)ezTEs&6d|qdmzlED8#>CC#M`J#5Y5fy?a7hnD*|Hwft7zc|h%K69jWW|`n*;aPcU
zsq^8oRBaDv!h>TuigJ|N4sXT3VGqaG>AC#OfTAUUaJ;#WB@gMHcKTZ;>#k?3`VRSb
zT7zb6wKu?AJ~y~^b+R;9KF?;+doAjHGNdRwLcLP$P5%pOz`wC=lC63_l$vVIn^1=>
zM=s;Ems(l+l81<3o-vFB-*B+Jk~0(WQ9{)I=x}02fFq+tkx*Z`52ixo8-4vR**F#l
zAsAG(v9Ym06LP%2Ki|I%us<lJsHk`8i#tZ7wEMx!u=uFHZ?skS=Q(B$1i$pgzXQ&k
zDW0BwslmT%{DZUdB4g!0I2D|Qd*aB-iOjcuQ+|$zXL;EOz;d6<z0jii3T<CWS^((C
z<1Hi!SG|W#J<mU#J)hl%-vQ)^N<m)L_rDKg*Qgcox2SP}`_7H2qr1D#ek%%?qttpI
z13c``8y+$;G6MY^QLrj3Pm7T~!G*S36JlTqiYFB(Dl?N-^dGH&YE0YwhQcM_GNNv;
zTNZlOKQLZx_;qPNwfPOA0;W!Ef_Iy}!Z@APh(h9Qv1Ce#C&`%Wj{DTq40^n$`kM9d
z1G<<&tJ14mgaSw6##!{;aQ3phk=5cCH(sFzrK8*F-=^$G{h5*6(ovQRAsi&eW)SpM
zWVj>|v%uybo|l!=PRGn(7<F+@K~bu_rr2x8(8;oiB5+Vst=dx2q4O9xbdi`K7GzeD
zD#O#0c?C2)l1e70`fCTlFR>-i)_J>)E&sSCPjBd0oc3+RgOvwb1k`)?KrHkj+p<u7
zr67|iHu+`uYA-a=jQDT1zME_p8nn&ps7@$MJvULm^}*8le{drDagyG$ODkOs0qDR2
zk1&vCe*pN$*6QkS$GzE1nP~s(tGC1uARB?O4*q>?bd-UKsS0pL?pE1&Yc0Cl&LFyV
zW@2^_G2^pJbY1(8Tv4xiIfUHwf0*-e=j(e}7Hj&PQ{F>TxoSiVKe|`$Kiz<u`r5No
z^Zt=RQ0T7`1|{7%%Fe)Q>xVL(NIq~261xC{HE7EoSN?d;HJTYR%m3<lZTXk;R{bU!
znF&2;>Vr4Zo-ljNtJ|ElP6t<-pB-CgiSA?%#kFH1?-q{W(a8jWxCUBpypRKH0V5Rk
zY=u8e0Zv#Hrwao)0+%gTGP<`gZ)1qZg3H?F+8|{8yfNRWzIUeP3|^%CDVY{YzKQ*+
z&f;E+1j&EVA7jwMMUJRei!+G5#SAJgDgvXLdV~?z<#~Cg02v3<Az!|DF+4ns>IrCi
z;Ac(EtAZ79ADo<=I8ziAgO3PMMKIycFr_f_eyw3q?<r8qK}`XbcnK)ygZ`#a{|9;f
z;G$eb*tEG4+kyCYZ#UrH5W_QpF}A10F8j13kU5)d$+BwVZnEI?k_v$I|6dQ^&CKk&
zGH~rRHR%4)?wW{<|GFRiD^+c?zaL<80hJU*X2t|~L~$#d`X;7y%(MxK+HPA?pb<T?
zV`R0>Dr+n;U21#MAWWobmc!I*4^yWXrh2u!w$$BdiGDnB(K>LT-8bvjU1xQ@FP410
z$VvHN#GqVLHpp&^jEwpmy%9T@Hte5#$K*iRJv@rE$o9ro?$f*Fe@-Jcl)L?c<(}r|
zJR&B1{=EM=_u$3_{mp;N`OjYd8EfhuW%>FW&CrSQnHezB%xdbl#l{A_VQ+5_rm=$8
z@dJ>V2UGAZ-sRV-?qW{e#M>Q5bc`A}0co1zs($_$pm#AV?q+Ig8TpXdVC<uYD80a{
zjEA+Fv(f|i25aQ6Rfz3Ej;Ewh)0blchT<H%_5l{3-#23McakgZr1nWb;vX+<`_^^K
zPlmL34&Al%VbIfWO1c4qD?;o`4&CwnmQbMSQkFA@+HsJ30Hw$Gb`Ra%ueYOBc!vwN
zV)ks%C*c}!2Evqu?5`!SO!5KCr1(wx)Pwobn^Q1R=d0uU0cz<+r#8B1=Pg@IX(6}w
zhBzfh0bRGxcp6+CaM`kNfbgfn^}X<YCv8B}gnud+7i4$HASmfs*jg`+X#P8`s2f7{
zM)%zjL);ShpjLAp%#Z@O0MK#Vji*!tbCZDMcz!;=+xb2i%Y}7qXhb838`z}vj%ukX
zH8#h$Bs_5?KQsrRY)k?Uo2U{FY>7LCmVzF#bYWRV<JHz1??g3wYA)OM)L`c^>T|$!
z@DK2NQ90k0Z>r_}Hs}{==a(08AcjXACmf{z$Y)`p%QO_k#cqGOCTCLyjJiF$ZZ*`W
zN}@nLG58T>vKpQaL5y)E{rP1R5BLZ;J}gX68}|^Fe})fwPn9c^>e@DedS}Zdmurqp
zd1~5!?b+(sF$MH+evKM7;16v0q$K50(H;TD>j5%iZ#lywDLPIywbzyl$M=s1xFKHv
zb(C;+J!R_PU^Z$94Z(GEn$pIeJ9SoQ6T0t&3xVWZ6`I?){V4><FD=?!hoxgy#W(5s
ze=bQd;@qRDW;_n4T;26su1S~r89DxUz%?|9RKP5AnB!B`mxX3;%@N(X5pkNj`&Sqc
zX4xP>6>KfHL83`%e=ES`r3YD68@K({2+;=9Ad9A*y_ADn7NUBrLEUyyXMcsH(-GBz
z1IGwgyFvMs07U8NI0bAuP{0DDV!&~o*KPgchP-MIzIEX!Sl$!NPS)n{zQ`(RKY0Ut
zd0}O7uKJ)rDPA!#Uu4XtvNMM7DYWxXY4qCIdK_~SKt;lC&Ok;RRSCGyfULCR?j_~}
zNHJ`;j6XXZlhN7E`DR{4_x%Dy+DBS*`YIa#>bBN(+$Xsl$m|X7ZQo-v(aye-xXkhV
z=e%+4{c=B;>y9Qye>(0)5}^!_Mkx;O9~?a39=uet6}e09GT==?nf0s0xvf7=eNMy|
zxw+MZ(DbM6DM73mgYS{PB@Y{d9-cl=ElM>>$v~}*lK5S0S5{tk+b_&jnNybaUdwPE
z6f2qKpJcJ^A)N_|O2*<h8V9Hq@RIo5BcB7}xAOWC#BRNBFhU(xQ{WN#)cyB^YgGPG
z@{+hi4!rX^4^(GRO4@j@tvMN~iPUp>yeK8iGEQf)OfZIOJQ^^-{~|kqBpz%(Y*9H1
z3-X6x$d&(L8O0A<```SoTT(JC!0f9Fw4CGTd8bDQ|8{?nkdzGVy3Yj$>;Zk{J^B^E
z;{%+$;x};mM8z;^M(3yT-7J*de^<%Z0r~tdn7EWTNfnR#t^KvUk-L_Zv@g97|4II;
z=UsLp+4%Mza|L2YXpa(Ls3#136OoK{i)E!TIg+;$+CSbnMArtE*pTYBmTfkzu7FY_
z+r1h@1ob$e7fr;{BCn3c_PLdPU|#~=!c9@s>C0|Toki42^xlhQ(lS}wX8_mA{o|;w
zuRprWm#Cr`r=r-|KMz7|czd7@(<ji@Cu*|N;EQfquHaUzliQtkS^o@18v*zrV^8gl
z6QW~b_`>?v&am~9%yF3a4F^|c|J6qPILBQ1VZgoxswKN_@AX*4%@g<F5;#<c{A<il
zl`u`!tn^+7Lic$eRP`mQ1j_&FNJmk*`C6HFWG9ePKrviK!8tih?)RgkSk(qoPbIa7
zEjuwikUT&WU~4g@_0J^hVMpeCPM_72#aZ|rQ+~nYM^f_1jpFXNE}NGvwul013$s7f
zJ~BdmxJiw&SzrY02iSA&uZV_>sF-7DUS{sVl&|Vuww0<@tNV|}-jk4HJ{Pjik&~dz
z$O<I91G5R?@3`0d_aJd~91rrFE~3FF{~l($daYmI8=~K(ciFLeYS-93J^c*gr79%H
zvhWCV9!3W_9Vqg>GtG}m2??T6JV8gFryg|Dt%ZN3z$r~m67eA0f<jdkPoF<tkyY&m
z9Jjo<OTwh~C&^kL!=iN^fy%b{s&5t<La_*<G!vk(=-$B1#0`*|_aJM`fHTgT4i?&`
zaBD9F=x8{<YC;IM{F0IaOsH)FcPAJE9$om7bYMCamC)OLm7susH&F55+~TSh3R@~H
zQ3E>jhk!{t;oJ)7weonnKh$ac5`Jnx(5rVy0$p=Q1=qBeXl%@d5YV#@d<_#0TQ<vk
zYLfS~$HBG%u<D1tVg>cE;aU4}F$5cqwftW$wp>Phlm~3yyL(>nuJ7Ld)H3jDDT(hB
z9kdOw_3x}0A&4cfy9k9~$mN@&9d`uZv4WN+qOtuo^`nMp#bIR6v}q9yTT$G~pDy1>
zVD}$Y#bzcJS98KFJ3q*%YA%NnH`!T4bt#@rG=zK#?mZQ@j5r^EwyW<7IU;7as2L4j
zyr?h^0d^C=*)NWYN%c(XTN#Y%Fs~9cRiiXjM+2=CqQaG<sMj~)XWf6itj(~`mIn29
z0WZw5O_iTmg9}H(iuUf=4)T0>Dq>*$m%F((_})GslZ<1L#!`nokB@t`Z%1$;3m5^H
z=V;s&zh(~}e0pxgDY}y${RfcpARXP$O&^p#DtCN%ll8$3As8OQ1(HI59g9!l)WW;e
zKbX<YY`4YqU<E86s!rxSBoR%uuY5H?;|TCs8?GSjapA#yZ0<qjFmhF8r85Rj-IMKu
z?gz+E@PXZpuWq-}i_}u>%7i*i4aA&wd9~OIc;lkb^00r=;0tJdihKTYv7m=hffK8Q
z!;cQwSgI+%j)ljwr_d2pmc8dP?Sb#I%=w(BIAFS{hKB*agy}NWNK7s$)cV{wna-AM
z)`eNwTiX;7y$klp$k#i)`2R9bW&<Y8sQ#RVnb;+Q)hH_qG-+yWZB_gI*274FkvFOF
z29&|;$0zNg=vx}`qyu>$Ic-9;8qgurPRyh1Yo}>O{tuY^dp8^LB?h((ScCPB!uIfv
zuA${)o^Np^#>A1zZ#|@M<D!Bxi?f~Q5jd`Y5>!HxqJBs_nSXvQ<?O6fm^KYRfB98p
z8t%R7s>Fs;KD<IC)YV=|xB!|ePp?NzrM9uiTLD`rIaWJNK@kU9lE8lQP>R(RBj~2a
zu1^e_Ut7Bx>M)Ec{S67vMGsFK?ZHBid^46Rufd%8nZ#S2s?Cf_?bjxeQE`j3X9J?g
zFE+{IlN)YG9^wImTONvfQv0)K!oqV673ByuuC42tv9;ZLDjN2<{bM?4|E6~FX7%VH
zYmKr>SUg|vdj%yVfW1a53g85Gvg8KJ_0(VU8h`WAU~F5NMhsO_gL9w<hjFnBxs_Ya
z(or>>{TofLKP%RxNBk`Ej_Q9xd9-ifTSc<fvH*eop*y^<i+p`ui_oglS^8S@qb(QS
zNdS4af$8w{U_wOP4mpVw%_Ah8H8R{uoG$=11q^3UB+YRhWwD-pT6r!eD)yYe@q8k_
zB*m9kNQlCF@OBSITtC&D``-G{FXKKIPoIM=>8rQeT827+I2vCP)4%Qno$Ia7KmIXE
z;qs-l^4Q}MXvt(1PFLu5V2dWc25gF4wTm(8hHf_2R-w;K#d+?8WP*pdM@WmGDX-c}
zHxL?zvw-OVxaVoKZwj4|Ic}cR_nhOk)ZM&Mu=PLmXMtZR5w7RYx}76K>Zbmeam>H6
zM~?xzIj{xTh4had4ED6)m`%Nvp|<h&)MmBjdzqUxZ9|S)?<q_kpC7gCdkr#oJ+9%Y
zAC<hma@{oM)(=l<Itm^4kGt&l`Y6h#Bt(Jpg_nXF2LgY#gq_rKv>@q(`d<eeC51gg
z-0&s9oRg-Ps|wx*BhRHsyYW{IKlR?YRt3fVy42CKZf*Z-N6>T^<R#}u1H#5ncqp;v
zL9yDi<OCGv0;itW>Q{hr^MQYMzz{qe2#Dt{n^q0L-_bpi5^8u2-oIY{N5C-T{r|Xn
z?`SUnH-7vUrBYO~cS3gAo65{e60#~Q*?YT{B73idB$Vu#tz_@L_jr-LH{Z+q^ZlLk
zJ-^rSS561_{T$czxE_OJB?j{d6A|!l=s>yA#GjAmYxYo~14<Wl7?|H#M7&6>Ebh3-
z94c8kP2p6}A5z)(_i_E<U;H*#)A%Je5~ak6<J+{{#a6vp;RQC7_{C;LKfMa~&*;@T
zc8k|R<W(Onbi@>^bBZ8tZccxK%^$?C&@7m<U0A%+Agk8XgDd$r`(U3=q{zcV<ZyeF
z^T9M1-Xp9R==SIJf_W-^i|>VFHTJtDhaU{pg1x?Cl<dA_&(=)tWXH0vZ>E{$GX*$N
z-@rrm8HfrbJ~7;aROxWAJ;am=b;<Mt0DH?oR3j}XqckVu)BXdF`q_@EnD069B`<ZY
zrS6*$`H=nZFvb0XR1U)(DQ-1^1WzX+C%d`m|DUmO=cbgP$?!`re)h!V)Qsv54bA%;
z{{~ZRixZ^(GR_i5LYIIgp{-!BdQh9sC}>Rxf_zN+epZ`<`pQ9yG0DwHwHqHL?G2^a
z?Pp`Nl=P`GN*aCL4TkmF9HP$FX$xz5#*gr5ve%Y|rFKz2ebKjk2wU9p!QQs3NJ6y;
z%F%hy-T?Gdh$Mm~GcjF*g%4a@g6}seiRx!Vw`Do&eQE;pX<X{X+2F1f^K56$l5J+c
z;JLnIhP}6EIf&45uyA28dz9Sn0n5y*B!yR7*T?h7&N*>ERGhn}qqbhZlIC#AKy>rk
z$5PiA9dq>RAwAtM%4pkDC^ME@f`zSp;n#CMSN+00bHX)Evq-aSx24<jE~3F@I|bw(
zjUwyM`;IP3SV(!Lz1hGu_b79hIPX+qFBMy40v6a1AH!J(w_yF6(9Ll{1L0(1s?RG=
zN_s0n+Yt>!lEQ+c=({bz>}qMrAl!WDg8D*>sx4TMlB^B^B$Pv?=UjoD{XZ(pYqD!(
zS0X5Lj7M)R*=rgqs+q^Xjq?8T1wkU*chW4GZ{Kcs{D50h?+|1v%Sr(c3EFm|;EU~N
zIUoDx&{j9a@zTfoCGW@2LJS3Q`cCI+)K~T^YK}jD$Mc!Sr4&aDh$2GK1_LaQi29%H
z6ymQ2s7`ZM2S4mHL>3q<y`{MUXP%rzPL@y^-sU3~L_%Kie!v1HE^_V0&5qv3ClO(4
zkN?E%s%R<tmbj-W+a*8wLBxAi%1~|agW*eSy>axQEz1y7m$G{L@WGo2AI|tVCNI!W
zO1}TFX8yh)<0v)i*`<3b=p0|DwcLPz4r!n99xqQJ07U1`i2Xl8@te;aCoaCrkxY5=
zWB!uc?z{$~r`fn?bR7YD7Yvs15YPZp@R?azSpn4-2-0Aote~JE0=x}bSy?5eHV_%6
zDaQ5j=1U(4GdPps;}-4O8%FXH96uD`k!|L1A|zh?hZ+p)t3D|8jr2|UTwtJPUK0mR
zlF^AQd9b-BCdY#s*^p;}+Q4_)I2HMldgqXtfwd}PebvC(L{69KYKYonhgG$`w1|w)
zU%A{ULgVCxcr2I>R<GT^kHxB5)1cgFaCx@5xaIn@)GJrHZ1AAJ&g}<~MJgroO;dpJ
zsQW$|_117ahmZe3V(J-xQMUdI!`7&G|7{15eL-Aq?esdNH-WV{+wmqbxQ&Cvn1m!6
z><OHJ3+3blR>$z!6u9+n%YIe=Mq`8ne%%#cB?&Ay@oLy7g%mEUYEBZ8=ZLGJrgLj;
z@Yy-H=OA7%hr!@_@|{!N8XiP_mU;RUDuo@U1$ShgN_e6<S*59d?}|#pLZfVCTl$TE
z=VowDyJ$Nc6L9z8pTxMWstE*d)(<?t;(HWv1xXC=ioxDn8<6zo#FUoZE$}`46jT8k
z39D#@AI4X%ktpJCTtFf&{#U*nFo0~6kV$D`-O1JU5a>5KTI?k#$m~NxLMA3A0BIgC
zt?e~3!da+))HK}Qsiy6{;j~RiNa;Q5N|XSr?eX&;No>KHH`Jq|`Fd&oPO>PfeE=fK
zR~zDmsk^Z7H1f0Jm+QZ6)${}NZHzXO-G4+zgY**KrB`RFhtR}9jwDP4icD0X<XU)y
zOSPD6lNPja;9&pd!!5vh;pJ#vSc9gWtn?qD6`<%X)q0|Mu$yNb`(2|tcKzh+n3d#8
zO4e6Kg^?J@2&Ab$8Y5&>Ks-SpP3`kA!~+=~Xm_Rfrkumq8BQgDAm?roA``7XI0&hw
zc);d~l_Jr3q=A{u(wzQof0(QRn@mL%`sr1F+m5r<70}&Z9uX}abN<EbZg&+~&Z3;m
zMQ_Bm=%4Zo@Ie)D#2Y_(6~|ZMg!J?(LV0r6aj%9Fxiu9=+Lr$9$&X9pT!_R|$$=Kx
z;-5yf6|lENe$u$B(}gtvH%WJ+@ABE16#TtnuIj|QAF0g9xN;WzA;_<af-Jxjbh!Wh
z{t&XBa?Ix?5WoFliw>m1`pYd?w6wP>zov>!y5biU!*=i{H^G*$Wj*=5C_6Wlwkr0l
zpZ8q#<A<49@#dOpRLFPcCazVJv?o8t^ZZK-Prq%6P#>oKF8Ylv@lfB^XiT~LBrL46
zr|U8Tq{$$@Ipi@q-VqV;Not%?xawDB(j_FLH2~-Q9@a!nnt>s!BYcfZ^!Xo+M!B$P
zm1vc)CkOk1oq<DQ`O01Wz@Qp7Dj!V;RdsW5m-EeL?GlbNY=*46)oTL*BM+e3zarh*
znU|O}eb=<*NU=a)03<L>*`kz!`k6Z)V;|NzDd^7aqg(=hNi^r)erydtSyMUySZiI2
zK#zJ!M6dp3A8pwJK6l^Iar@xNpC?iLk$a+>*;Gv4N*Wm^{51|uO~=L+ri&EKEI)JU
z@;3c7b^oI;iSJbJ`ki`g7k~Y^PF@Yfi*pj%+LRb07+WCK9LEChj-E+t#DNsZ#=?uQ
z8r@~m_>~Z*X|=o{%o3SSOV8V*5x2>J3LPgUXW<|uE4VLYnmTVA&%~E56CmlAxu6o5
zu(r5DP?5{K4^<Np@AlR#F!=`9Q8v%CwKd9who`#g2`{ijIFGyY3rXKqfZ(f}<zNk{
zgx83!7xCP)%HgYzdhovfC;|}|V?eQ8?++v5Bd<U<FNZi$k=KYk1EC~lhG$RQb(c!H
z_YfC3vgOPY4ZoIW+|(5C?HhA)rTKEjVzl?SE?x4Wvd0DV<R1-qZa-=qV;x*v8K+6J
zC|)dWAX;#)I{GcxE{GBAFHPJOX~$qsOx+E*S3ITJGJL(eU;7tXSg7j20NOJ$&~t`S
zZ|_WB@9v|lvWYWKtTG@8NGw}d>tZsKme5z-^eJH5tw<9k(7SW|T<}fb!9$EiD&{5+
zFJYBhZ>7hnbLX6wPur;3scDzOZ9&eRUmm-1IuGfQkj99B&wsRn69^71Yn%C**#g@S
z#7V^+r@f3h<*T0Eko)%e9U9i{vFmeO%(ZVzTxNS#YBng3iY<+q){hI*RKwwlEdd`$
z$2B#F1c0OCW3s^OGkVg{jI0Phyk*-|F4{Tvrh%3{D$sUxw>h??DYm4qk3UcJGSXns
zRM0!7k>A){?6`1q#NsBft4H-MUYa=Wrs%Zbn#%e4Bh|BL*>z;%ZqK#xXHERn)JJvu
z9N<0Me|L1Wemdqf-gtfYmah3`#XZC0Nx$^xh*ct{h3kC9!<?3Xfr@`+IB#%o$mir*
z3#+;#w5oTw`42?gPK4Og?$9wD|MSA&`#a#gue^Da782hLTc@MtwgX$MtX1$aRW{;J
zWJ9XLF~)VMMp>JzjF0oW5E;^P#eRR!DJA9aHjf%Y`>(&iL4yZhK^e<wZSP*u9I_^I
zqSHK{aj2}H^VTu?9^XKe@jc7YzGa9tpPikI6?*xXF4#~EbhLKVQnYI-CIuBt$1Cr1
z$r{C-7(t%VY5lFYm!6s~ddE^jBv4`k+Y^ufVIA5s76MutE|-{bm3AMJovfJUckZ~s
zdL^w5-^Wwk(~h2q8UJ8uTpj(j*Ex%QQeUJ<YH}VVS@yMjG;?W!%DUxa0Fp*ZhcD2V
zjI{2zB%GRSbLq&qSNAvltBoTl+_X7+rB(h)i_A0?9Db`yLc<^m5{39UnB@F<e+dPZ
z@jY&CeUmkytq9uAJ$?2J1QyPlLe7FX`#;tC``Yct6GI?I0{WNX^WW(+1Kf)^P6NYL
zzxPN{`{aJ@GrAz8PGDZwhd35a0jHtsvIv2$e93YO)+r;kh(|=q{TBoB<nj2+G2+2w
zCih+scb)Ehu6mgI*y4TN$i-tpG+OtmhSlyrvT2`KyHxc6mMvDmbdlZtHu)3V99}8`
z+dp@!Cgxj?x_A#c?vYPh{UV$8CSUxXA#6v5L^&rGa`lu{TH4ESHccMc#xG}Fu_rbY
z9*o`O)_>}ShHu#7R95CPFX#&)12Om~V+^)lkpDucT6tGJp#Bvgckj@*Sr?`UF440c
zuZo+$VU(_A3M@LFO&mQ=<HV9sR+W-hyM%n_8OkD5Kdbs@Hh#ATygd3rD6L>iDd>8<
z;@9*E+xQ%%%=27>vM*h)7fF<k$G`p<h^{+${^#Af;b7eho?Y4FeC+YLE@K8}{UN`F
z+}z5=NZvIVJbL4zByQ4(?vyhEsGE90$U{byOz___Ti~-Bdlo*@k!nv3XKps_Mgk5N
zUwK*}j*Lz>{r&&Q1fWPeA3xIgJ-7IE>GV<HVe(Y?b7Aqba?!0{%csiKvn{Av88QoC
zR)>T{MHzwrjHTu3!UDuU%>oq`s;8mh;kejXcOcP`A(T{9I@;R5It+34g=a~kl{J2y
z_aB3;pzM@+xiZK2Yel32@Z+u#5f|K}vUAuyF!kZbyS(POHN!*h-n%m$oUDiKxh|n7
zPD5aJcaz}Pb;^Zb4f+S4bkI2)^hd7vI|pX#PCD%2xSQrd;iKKQx)wo2Z~XEeoI84V
zgyPJeaB<fW;E7HTHqKKaW<FQ!5-~k`(tg3w#d<XBVpCrScH&u6b+&XZ%rjUy>7^#7
zBcnq<@ia{(trG8+5c-h6QKT}b<GA%#d17&Vl|Ndt(7~gf;dC%|q20b{ab};4w0R<b
z!w-1AIf3iGKL!N2={;$Tj1c4|3-8V{ifbnCu1{E`;NC!MK8Fmy`B?cr<{~2ELf!6`
z1l|S&CYs2EkP&Iotc8Z%S=*BlD;%40Ilfb(^{Rg=mIO%V0(~-kbs~uj`dboP%Fk2K
zfjuU}{%k(FPxDXY-{EWYzP0OS^ThhEiWPrNmX=l<hib7#)&sGWSrvmL4B8@s43u^S
zUa!FR4Ea+1yS3R#U@k-IYz~3U?|fLTa5(RVk(s~(yLN1FDFcpCYUk_8$w_wA*3Gpw
zu+^^b?=Rl$e-Kjj^vN6V;e!egxt{faHVd9*DWTLz%zVC5<mpvAp<~olY1H-L26AEX
z`+3SugDbahp#xroXxyMhF<)P+hmH4g)?4yNnak%8cq7Bn%I57U;$z}{LmZkWmx`h_
zxD0uo#V6ESsUA<JW)yY<^%!|Enqe)h-Zn9Nyk9lVpVho`L<$eynPWe@kkH}gv#SZ<
zFYmF(W%K<KQtot24dceYfVAfFrFz*zjT86DxAW2}N@b>iMa#^Kt@Y00B~Qq{8AKJ~
z^QFF}3su%13t@iBRzii0pf|7)Cy}<nN!Lg3`Wq9mBqU`!T6>cWvs0Bwb5{#cHO0}z
znRbh;s`8ULZ<m3;P8S4Y#Riz3F&<=;9k`E6B^UFXdL|_>PS=r<(h2E5>e$?KL^_n8
zcfBM<7c+#;msHlU2maBkG{+E~Ikcc8>GoZ`zt$+@qoa30-ww=4pjN~67qBcqsRa2~
zVCwzj#}5(??Q7Sr%^wwBo3Y#yI4E4?;$(A)US_q&Y~eTS;m!&*ZZ@P){07H`v?NNW
zS8*{PAIJH?S?u|P{Aky3jZ>Ea+jAQYxgYS<7UvudpDcC!nmd`>R50@{aa~4L+O&+u
zISF%}huTxPb#-T#j%nl0F|~^RAlXyg2E0po`}H!kYbNO|=OlFfZ{xck=0sT;Z-qPG
zrtCIi{1DYM5LsM3t=aCykfN^s?1k_^R!aIn=i(U-a{76N#^rRqbI~fZuK-O*7y~XX
z7I0iRxVgg<6ZO>8h|Vuxv=LkD+)0kk#S+!o#_|xO4xr=B8gTDgef9VbsL5F_?7R5-
z`&q5eAQGvM8x8!vAa89)ZEzOe!e-0;W|yMJnj5>^CqCD%Z`OL1etOK1<R*qBnz?be
z=1B+e@1$Kmu`y8^8PPJLx1MT^lM+XIwpN@*gvN!cUPL&O%}HBx5DmFXJyR;xh;Zk~
zE#~6y*qWs40vsY%(I)Gjo|&KDJwZZNnOYqzO^4x?U3;HYtfaSk=~q>l;?t)+9UYHV
zRSBh58^nW@(rAQeel0gJu({9}@$H?I=!^u81#6ejx-L^A->-(CRrWT*&*uCxQ;nv`
zPJIZ8%{r-gWm+8$))q`pi!oc#OYfrCd(Dj-VoSt4)=P7J=POq;#=Z}}DeA`6P%ezk
zbLkBt6jadnrr1pk=e8BK6T61IStRF?Zt0_{Ko;Bko8GXg9C;aM*r#LB&b8v2SvKQ;
zGn_436*iqL)gE*~(-Ug6>!qkMmfi*2@gDkwoz|f~pJPo?e>um*zesg+@i?d5?mAwP
z&x>Ur1?@k4Nwi&?46v(AmpP(qsUec6aHLt|LSn?H`$<p8BD*paaqsa{7+H9_zYA>F
zCtC{G=DoxocC@*^KBPr}<yo(ios&prwill$L2A64zon1Hi7O2lINb_W<@L+zXsfqs
zmAvIs!N$G+-%|~=;H2qyU2L#s_iS%&2I1-=xHW+<TysrLO?^Es!b5HjmMof@-;LNd
z+UY_Y<VD@_^)CLY{JbLYr@h{b!SN|m{5tBiHlm1tKDfgDjf~!pXZ3wzhI5+AnuXJQ
zy~QX^<x^1c@?G!F+c>6qPmoyYa5_j~Mqt7i_Bq-~w`#W_E9p{>OJ~W4;s?pzikjOS
z%eX~8c4N#{xI6hLj>lu&(3;%x)z;KGWE3=yYd0J=G0nq!BtRp7JJk5pay4#-hZ5g$
zKtI;six<%#U2xAHrP@xPrK>e0vQ(`^JW=@fPkL+p`w^kS#>WZkn|$6?ZWN~++hgIR
zMXfnPR)LDGiQG2HdKI{5E<91W)`~<@OO4N;K6v7A|6bI3kcvdmc|QNjjIWpDqE06k
z<dK|WzL&bCU(;_BP}(-bZHndT7p5(<ve&Mn0-g<FHhk@p2pASlwL9QKQ;P26*}HE4
z0u>eeM@-wz18`$>adqYA<;^uq>s|n!%@`QMMzL!-EI*J-4@ITmsQY~NFjw*!u$V`W
z8c($ZK9oiOa=|nPO+$sEqBOVjgDS5pGj%B|Bg<Z=G_$mb8mSxjq)E^jeT`62uKBVP
zsft#=)mE1|R23$Tc4$=cPCR}CD~qaHn4W;)2rX6NzfYxcbnRaBQw=x&UC!3Gq)CAC
zcChT_WWrUvAnl7nHtSI_A3GLZnHmjjXnu?<P<t>fec1IiJ;TfAgof%KD<|jXF0A|W
z!+;q|8`Wn^PR@&xriSf}22JIGxH(LCha)6PRF*#GnBUWA{lKUFadbyS>_@EZs~<ld
zPc@fzhej2`6GIa?)4t4fl)%QSdts}ENEBs996#PM*(YnB82|RIwp_`ZcoPGKxi>$x
zcYV1W921v@|AhSxdim=^QclQO!#%KxLoGs$S4*~!xd$9$&hc*svqZeSgj&Rrw5s0+
z6g+rSQ0nK-DpD$9%)Ip6#Uawk#oBbI3_h(bRA(sW{gW#k0oD5Jg2WZ3uU`uY33080
z6;bV<z22X~YMmtaPl$nfF->Cj`>W%*x9yJgZ?a@o?+1V7qY(*qZ>z6wemN(CP*&KL
z!Al6Pj~~uI>GIeu39l0fCjVFX#^Kxeb#q<_eh{!f_IRdyuUX}+4)n68))TyI-1VAX
z^s2CYvaM^MGCB7Z1SWJ{wwJMx;QDodkJws?@aw(mHz(Gs_#+f=h3+P-<03@o2W3ch
zSG0xuoqIaSly>@q&8+o!h!Akby-@JI<U`WqAO6MEz5%Ydpfqi@wcTe}-Ng|`@sQV0
z_8!Zs)Kxz%qhOdl5RQmIH$~ijJN!!ry2XpmvX3DXeuEsG7W-i2n<I*&b%6lEBKXtH
zlJ5D#lakwv*xg*GrnXuJk)T~8Tg>P1tY0I%NS~>I&~7=4M4`HXmA_B9XG;-&e+YRz
zK45pe5*}rI@v<*iCSxJV=v(bO4?2mi*!5z4*)pB&j_GGBd(>3*q5jLwGxd-x#k{Dr
zW7Un3{=xtJXyCL#zNDlO0YN??;bh`BcAYLYbplwafpiG*%+6i~7nQTtWnuD$cxtd-
zj-YTADe_lJ>Rhm;?Y@eppB5~YtYP~wr8Ixv%h(bQ9_V^PxJM>02M~5_b}u0<aUu?X
zF3Qj892=$(Sm?g1nS@OcDIBbc|99Hjb(F;K^68Txi>Q3j9XSgv!Xe>*|3;alQiuRk
z^Qg>w$F`Z(Ys<799);sMP)wsmEt}k_?m|W2?{XfYZZ_ul%oUXKHnOprrm&1Krr4MS
zJXpfet4LOI$<fsZr4L|hTa24VamaIyT!FI`RP1+hUKKSDPi#`NjtH<Ry=3!JR&2*Y
zM5v7o|D?W)4_tU*_MObs=v!K&GHrpG_iv+43{m9%C`2j}N^5XM`{N@0_iA!*FN=|Q
zjon6##<ovX>rT3{kT10Tj#16u9jXwt2aUKypCis$=AO;1{LPXvjBiC)G3iNJ_L&5^
zLGSBuc>C@c+gAM52_SjT+m!<M8*oY=0-CYBw*>`TP&q(RYQi%FjBeqx$KLoLkxAkC
z+;(O<F#?{p4>*?Y@%PLgsPV0c%cpd<T(Ns)84r2MSe|(IldM;t4Dz8tAKhgbRgo8+
z1?%j)pXN<pN+l7F<%?-^_w;tC(NyQD+~1L@m{%IbqEN{WgGdLJwzfYZh9uO%!-XCv
zsSMpd#RF2w-TkX#;AtQaK%;T)pVpGuSZ8M^N;7Zp-HNtsj7WTo%Xbu1KmA!I38|DZ
zeQmZ=;nF{1lDpblhH3~s6cZbZZA+e7N!ZbtWaQ{`7(74MyM$f%l%ZBit0?p)NpJrX
zeOFm|`iy{F-*v@SNdQ0VqEfIp3EKN2EdZ|2S}pCdtw~V5PZwJJFqv;ve6gTo+SHI2
zG1FYBn<TrxdxP-0P>wS*WP1?@#ZP-om`TiUI*tEytZyGi_t@q#lM*iU1<dB2KkMJ_
zF6_+MSMd4k7@r>2)z)HFg9RFhpR`OL(q1EGe_jAW1CW3#>EhpaVVkqDzEJwFAK!8=
zL@kT^HBYH;Wz=bBCUl-+39O9ln_s&jJo7-TYLb5B>pV}?p<-(A`>&s`4CkBtU`b8I
zxHUEzqk~y|o~{41MasT*XorUDT|o|p(#?z~f{XKCxJE>`&jPl+wE6x0ZVFMPnJrd8
zasI_C&bnIt*LGK?b7;ELa*?sF*I&%W$F&QB?RanaQf|%fpKzZHythTFr>&=#*hHWG
zx;j7p=0i~wl!i%>`<tX%wq7&su>%+l!@a(X=;gIxO>$Ac09?<LJa@&bcF&8fbM9^H
zR=vRFHXCr&u8yfzxhj*D{gr%=Z>GuD>GAr|a<V<{wRnVda`yPF*1>jMd$7T0<Zeyl
zn1nh#G!n>jj#cn!zkf}Elj>;Yy3nD62on>8DFg-z@^_yxRzcdNq`Sh&W!?EK!vpPo
zYs)WKyd_|VWx8)H)JK<$`K9j;m0W8ByIKBmUkwS_GB&mJ(OYUL#@*>1Y8MqOLr-tB
z1kPE{pG(#rc}>w>mgeZ70tWJ4_B}b!!LkKz%_&l2BS5l({@-L!pg89y(p06!UCGHt
zqv6iW^})|D&NyU&eo*Sz(ieS@LLA|C-t&YigOdsnGFYTbKIbId^}E~j)d%?~jD$#=
zuTk&NlN_hDh%8~qQnD9g8*ZkLSELy!`SULcevn#)O5l&|Pm=qUm>rAx$MB1IyO0Mw
zww({TpUF#MbmH!ZOGCq1^f4Y=x>F#!lP`1DUE{93+tUtT=hHw{RQ(yuv##BoP*;mc
z6g9Dvtj?vTuF(HXmEpy?40I(%W)7A+)#9=;t;WMm52>50)Z9;l?lO&?J>d{NJ9J~8
zhNi;eT4g@{8qPWJ&`wHC^|s9io8Z@u=ZAFbc85eUDi3*|%jIm5TnhTA$B%OPT(A=o
zjnK)>8q3V}{e`x4W8_*WmYU(lg0jHDBz~qDmyfH;az{gq`__t%#mgV#$Hvp%al<Fy
zUkCnXTr9$kRk*NMGIgwck{8x^(@f30IdAy$q}4BM3X*oKJWh_&9ijcriZo7RfkCap
zm?QAcGqp6gVOpx4p6_?%XZt&8CH}^7GmyffEwfBw2>nr1xjr^cV);Z&{aOEZ=uHYP
z*<t5`tNPc41EtZMWW~D?dibgd=P9~>YpItmnEouZY2pLEpiw<e(O}CT+3cCtV~2$N
z(jxrBzlK|P@mrq~%{p3Y3?vX5Nf#e{FD}#tq`P1Q0Kr`I_Q|6>_rn)HA&mx6r*pvU
ziYBl?jZA`%)k4D;y7i-x|EN)%eS_DXz)S`C7!WWs!bV4jg$O9*uhxRO-C<>Z1?~A)
zR+5yM3#(7KpZ}y7(NgQ}QPe5K_8+qr7&$n9N;OpeO+MKXxkh}zko)-ya;nl!G|-)E
zU#8sK$9s3RwvOehU9Y_E=p=yCk)n?G(04R->@PM4@Z>8U=0$}2+Tte-2oSjGHYV1Q
zoY@KIxeqH>25uka`K1@rbYV^W&euS4Xx_Z@zIK`eO{up4$4J3xOMK2eW_=Y<EAtUY
zqmR>HLp};1PU}L>kW&vt0|Uv`hZju71@l?gW`f3p#>yI6CoUo#-H+=Y38GtE03TgV
zEPb=rqb8M-Q56PIg=UKuLi(*czw$roLvUB12}aX?SN^v>+(R%O#+m?k2m%^0vrI)h
zyDebrBQIKF?!}jn=&{f(s7@qfMo0a}eTc@Dnw#2N*rz&{R1D4~uP5GSVT-xpaF1bW
zY9_Y`-6)MMkxIs(zpCW+&mrfI-fy~|)^FQbFB=$r$)BjoMIJvnF<%UoZz;=p(7j(f
zn|ZwdwR8R{dEldDdzJC|rBwF9jOS_TcW{wJEw<YF&sQ%}i`KhuY%H@4FK1SAj&-Wh
z+;!X_MP%EV%de4kZCt5oy_ze9`%)!)EX8Ps1Za4Rq&(;e8eDqHV>H-D&(<CT1J)vs
z&L}V9Hu*HR-AAnrwS51+69^PL*3|eP3Shz)i6M}PIWZrp-fll>-PhmZQR~XvfX2zv
zM=F2GWo^mO!<r1qd0KczV-3->cU>vfjr*IGk9D*$+e~!xCQ{woYhUAiRZ6M~#|a+u
zZ!cU9l!F73#qQt#4W2#W(T9t78k=93COJ}9J=4?~9v$6R{KVq7ar`YQ^ZuSp`!v=U
z!}PT*@0!VjM{-+{Vb$rPr}yyqX#P$~Fki~09bX<5CUf1)Bh$Fz#>(?=-9IrUE>|V5
z>EH9BpA`SdKV1k)`1#_VGj7@TrnZW9oe;Cvqhxf{{YFh*sLDJmoi8tqSiI7%>lz!(
zA~41eDTozoJ^Jwz30RJwH|9krR`c6H_`rK<2Dj`{b{4Kbf4<l3dL3mA9Y_azVYFOB
zB-1)U4;fr)ntn%jpz7YOiTU@T&(It%^PicP8J1=TpJOqVyF3U~r&;;}t%;|u%uS5~
z=s?W5>PcIEXlz;<n2(;R9j@pf{Lbfp=5&B~Mx}>MkLLu?oDC28wh~n3R`^#Z04MM3
z>jNiO8yg$2)A?;vod5duYmG98OhAaheiTNn#l`%w3M)N}efY3oKYaK=Bj$F8nOQFB
zF=QKkgK(m6Tkc{>VB7{SifDMas-YnRzE7FWtlzhpM0x+-hadeeTIzA*A^=%BP#Pz`
z{E4ZVw$`vCr@TJ?JT6$zCg-)ok=4$n8dn;+`&h_hwFo=_X&Y!WSC+5vg;KK`ELZND
zskQD3{!8}$M7m&tR3D6{>5sX+5PSb1<;1d`Ib-%|{q^`IWV_#Y5p}=)H~0Fg@XJm%
z+o53Xi*I7D0XtsCTID)2;I`&wt$BVkiyMm|_Xkalht6300|m*~!z}#B>(_6_JeD3c
zUl?FJDF4b;g*%Ddefj=8%e!qRQD2YVxrYiSa_iH>>M~;@d~I!Capx3D2#HhfnSB~9
z=%uML0Eyqvn#7jvJs&iFphQ|vD^MYcTEw+xu8Ig-rX(<K(PQQTZKO0sN7Z=B!?;qH
z%xh2@ZKyBqyxeUIs8YRi6IcNJ{KbHVZmg|=KOQ9bIL&t?!7SsY#k0qcUl<zNgU%a5
zaOvohK-392_JhO2>d*3mZd;R(_&p#aCr?dD$%HH*2^5AyKvpOc1;^F1sjz~g;`G#%
zbilmX`?Y8r>zq0k_4A^Beu_r}N&0|T8XwUpY2-*b_E7Q;FVv)EACD(=>*e9a47ZHt
z=(V+|4Z5&|ZZ&TvEFHa)PPL9wFygh}-k1xYY0<`*i*&RlyvSz|H=Zs0d=zlWHvku5
zS3{+)MxIc%bt~L1be3tA3`<$G5#(0K>z~yaG&V|VC(MZ}CB5lq4*dXSUPs#h104N7
zxRD5ZYQfj8%?5@z=AXT0X+H;*ia(XxbicJ_8`WRiUH?@uF1pl&Wj0?ne3dMt!s;JP
z)1PZrEL01c(c_~&o4VpOJ~h2Py~=va6>h_w{`ZrVUyXe(yDsvouGDAJ8p*M0`W3}6
zK1|?BGv3idNxm(_T7^S#TsLh!(1KBpb<UC(rID#pYHe^QrRK+=he6uqA4FfjVB}!Q
z^O!_o{1CZ%ci4Sbu4>}ruP-txczjeN&PO0t?RpkBErfouRb~t%GZNCkMW{vm4V|B3
zA@pp49+{(~!u3k78M}#G!)DypJspkH-kWxsg%3Dv$jqlp-Y7Zle#L_jSJ6tB`MsUq
zzK0i>8@6Ysr<Yk7Ug`@u1s{H$|3aAVXV#=}26@6vQ+(}YI(ZsQ8LYvA0i2>?#+3>l
zd}wOw+(53rosc_BMgks_AmE%F|M|0-zWx~&s;di*_29$}h8q<i1Y4;HGP5#99!X2j
z?X6D(G36>AUh{ukCN@@9Rq%c5>f++#bFi~pdaKWjc2x%Yo26v}2LT|Az%Ajfwx`31
zT$>|(tIRCNg5<vJkFGe8qPZ?vi*>c#Ie|rU7lQT58N(WUX}Z~C-;h*2QUaf@!}gZQ
z0>Ux0jxlej(6zM8w=&gC??+d5BwgtGPfXqqCz1^j>gU=0fpYEa&vR>vza3#7G8Z}C
zu2vSvvoC4xG6#ghrw!La=bnFOWl?44bcYlO{(EYh(ySetR)3G7swW!Wwg_!lLu7BE
zWmq@Ig%1~CI&i5xvBIsa?(p^|?Mnv_YX^_q@e1P=`P9r$%YSTa<g7dI6&acDOezl*
z7`OmC_L=T_&DyWrqB+9lOr%;?nZBXlbX#d027+bEuVT`&&&GCyG8TH=#oTAz)Ne7(
zW%nD1NL0HvIyrX*IppVXiT-qwnZBdg6kW1%7)vl%`ihu1id##-kW!pxl`pXw#u}CV
zEB98rR4A;Bc7?>oie!#<N;xTc7tyEW=FLf)p(kQ3J{Qlr2~sha1O$NwUJ<ugh3U#<
z)~C_~zTuo6t|gOHWy&9!d|z_DxrGFMih9G@qJ!3-3$9nB=6BopH90zZ271nLZuJwt
zvjo2l=Z)JmG<uSf?;-jTRO`RMR}5^^Mk?J8+v0^y`!cXR_xAQ6iyb%)pzAGGP837F
zBqd$HapUWcAI`@I_Li2Gbp>`=({mC&etu`MO>S;(@4URg7m$jS^cS$3d<mJ*lj%-j
z<PK{?Ry;o$lDFZfd_*2wkeXN8XTQJ)n6<uXk#WsC)iKlP7OR_tM5)~iGMuAhj*N(D
zqlPW@;Z8JNegCRBx0tb}Iwhc%hq-ghS0}6mD2YGk>s&ZxC_^Y>G1scqdwW`k?MVH{
z1jnx<F%<t^)nvj+=5Z8OgAS~?Pdzu0N(q^0FJGh_x#T+OK1KfFQDoRPEdl!@=G8(U
zK29<d!z;-Dpu86e`D;tFTE*M0tUpbm$xaO_7A|o14aRK5D(%1+BT{<Uwdqu8z<;OW
z>8PA%Vs(sc)&9i(hUK`i?iJonbOkl&H6szdYN+z^{&l+h?kIcjY2Bb$naBLZ90PlO
zjqu;LuGh_Zv6u7QVE}3P?e>AJpZ@Ty!9n>Ip<KH<Nl`^2Ga*rk*tkqNP~dbjlWV3t
zJ2!RB!SED{I(X-oa4Jvox?nGrofc2Tj?)1uwqqx)?vCsjFmV^pj|9Gms+jgTaw`?k
zNUoAcHBi?kO4-xB$%WXxOE`G+6>!X);z0}#l;U>Mjs#`g*`BKdwm9PARUixoZ}pd*
z!Zo@2^=DCp4okcIxex`0<@vvVhp@%krY1in2=DCA`nbCLh2aa@0Jiq4A|9ckT24p*
zbf6VdiyVqUowmA`{|&i_P&*o-{X)`tdhiXBBYumy(oqx5{j{v{M3K5@=APwY<pJqm
zwH@xujRfaVZVK&lBeH(}5cXAgaY7u+Gu7__s=2T;{>Mq!AMb7O6K<P)yhj>!S-C2T
zY7EZiyeSa{Y(+kaWn|A)`#TbZ|2Q~j#OyE44*%)OHg!zAQdXWA?Zbr9Erg$XSs`m0
zVt7A;>actz$6V`Zkf%2du!EdgPA{Gs2@D=;m)lQ_w8H<xBB;SAkz*sU<)T;Z!MA`2
zoW(|`>LtB@OfRPj2E|l(SW2n2LQ4%Jafj*j1?c69(qerppQ#)zM-A6Xe&FYf%XJ<M
z;A;;dn;jqG7(^?dFl4V3j8xhbZ8i<IfqoPmO9XpAz!0{Gg9>@X%2K7)_d5N~t8BXA
zO~f;3$a7kkRdCnpTcowA4|a8%{iZ`!T)jho`_L%<oztU4Ba82@m9_Ll#75qwXeu|H
z{qQ1ltL$-3pxQ+yz`QX}2e{L^C&oC@F)-*w@$vE5TBU=j!D;9i?4q(3=^UjeAiM1>
zkO9!f>LROI8<ZX(BJ4QgX)A#nzDt#>Y?wOe?$uYsusof^>+>T1EpW2BTW#I|jdbec
z>=EYx^S>7+3B^;nqa4u8j5dxtSe~`rj7T*mL!|Dgx70he2h@K#`AZs?S5P{yf<rm^
zb7fFphpr=cleQ;!--T_&C3PZq-QVfBu3SXkpFfZ9%y!-rD!;Z1IN`E0r^}*>Y2EAr
zTXRZp^x1h3?TSO3xT-S<awFTT_$Nl`kHzhs<=qGz$S8&O1ILrsR);+%iVpD%O{nvh
zQblO1ZnnsL_>W6~O5T6^>qy!PXV_Ftt-r3~_!?2-9)oZdKXKd)KE(KlDn_ofQaEz&
z32u!Ijf#$3WYuSIR%#!GHg{N@>Eank9zvaY=re}eUUhr^oE(gJm{Y!e0}Hf);o)F1
zZX<}%8XO#KX$AWj2s2j+owdr4&~kA(Sgo=%pZpVtn&|0`Lnt0hu`4RCBdH1t2@eKE
z-<Bk;^sLud;>gr7dXy89*CxGomRu6%iam$;rv_S<^h@f=t3MRv6=j+{-nF&|L<y=E
z1Fl)cCXz+E17uA|@G3HMD5p#<3~7W2dEkkVV5`U37+UTmfarK--qjGo<5IL<{=SaF
zEqbJVAyMb?1~WH-KwF~mNNC(<!_LTcqnL$>g&_KK)Y6nfG#^{LNN&vR4wjaSh9@SU
zO>SXhHrt;#iCFX>>_b>e*OPI~;w#SRk0D>;@D^2hkQkvci}6?rHD3E{)h-Qkx5l<{
z=a~nROBCeGhkCf@1!H?R(|#$Eqy9#(cX^<%8`!$`^rS)yx|8~%>`$eya6|txe)t9r
zJlLVA=pbj~hy|(xzZ>r_F8TPeQpiej1^V~@nUr{@{iPS8XtboH&Lib6+fBp6Wx2W6
zhxedB0i-rj%-y9gW3{p5&6_uf#G^-kvsPyE=H?_(1glGBDKf#-Xr}gHotygm-?k~k
z;W5Qf1Pll3P?;FAF>^E(k?}ckbjv(YQSvs&W>>52+4{_>xjzv7tPcAUju>ElaeG*J
z6M7Y=SCF|2LeM4RZmmYEKP5h0AtbJ@XL3IWX&D~2*zrL-XnA}sJnkP0f`T7X0(cvP
zffy<w2SUkU)W9fnG_>HG#;L~=({ldEb+7(^g20FdxDf_^NVV+;t+;T^J;D82<V$vD
z6JLmtb6<#-#$_b^(Z+*!N9+`Y7u`OwT+SXt+gRE3$4+k1ne_cz)_7#={U>bWYMIhs
zCJ%u^V|}LnD0*!bt#QEDE<b7--Az1?c$`aId>w(Jm9B6hjtWj!^c_k{w{hCz6!u2A
z6=m{t^v-7m{pYFrN9U9xlTP6UjS00!SZIWLt3&_lBxY9du4YwXMFD=++xmAqu!Oa=
zv{u2ab8X@`7*f@PH3<0qfORE=BYn-xyo$88wt`ThiM$n(N$x3iNeNL@g9QP4?aXl?
zj6ptCDL(zn7td1)7Q~)J&RQi~Enj_AL`73)(S2$tG_!e-uV$)o!?M@L$ZVs6JA}EX
z)SD6RP<X=mzlZKwirtO+t&ca68}>X(+6AuafZyh9UZal*j~y6mjE#)G468qyGBvc9
zl{|t|cA!pXxj{U5R7FY+%^ZvQuv|i4M||&kQpE~hF6A6R%K18j0U^bbNl)+|Ri2YD
zZO7A?xfz9chKU^3gD~|D)8@Zp<0X!}<LPhaPA&Q6x7w`u+1`X?)p7AFjf>5+|JoFA
zY5t8Dz&WZ-X>cHGr;nak10za>U->Vihg0}bNqVZ%Bkh(RiTXe2Yg(6~|9GPnjdfoL
zdIM*QhOHlR$3mUF*tetm$+9ac|K;x`V877cx&`GHuza6x%`}FC=^Yjp798XQ>1&W3
z1Z3XA&+mf*D~c&&F<;v*tPR|H8_B9pE$R|MalCMRLSalndV_42W80`<^R(_Nzj1Vd
zZ;F{Pt^vwLCUFh%)*`Jd-(M5sP0sa|#dUcDhUGD2^N454dNmL@Xu9$!K7F$GtTDP?
zTm>U95o&&iUxU5E%J1l>a*)dJ1-AgOC?Kjnt)sg4VsjIVRy6HP&X%cZ{tXY&OXpds
z`rCYNnr8+5eiYBGd|YA29b>DRF-?L^(rWbwAC+{uNP=tH9%%uf6-q@NoKMZw;L`AA
zW#zxCi&iHU6A`-jg7f`{4|2D=0LRIlkAJy7RSSQ-#)=SR@`Z!l&d-bt1Hp3elNvre
zq_U7V#&$B>IsGOV{;M+<t$nK~@7=B;&!{@G8y~nPJd*d@^W=C8ZQ|)rBPc6Jt7dhJ
zfTQ6+T_v6R80M;zRsxqJSvXj}rlem)h`n79iKHjv9((VwDPteh4xQ>^m;!RjVczV+
zIS8!<OjQTg%G|_FuyyS5WnMb+JXRH4i;?k3kacfr5q4Q1Mu7NK$2jFJYHX1(&^M<7
z?z$Si2Y<q{mrMNM2(?Bvr9fI|dUeCU8Ih1$H=b+!7;Pz#n4ed=b6js1{+miS&C}|!
z_OtaR!&^nS9xBRNL~LIhtTaCoP;`V^KK-f2fqh63`3}YZ^>DFc7K$I!e<JLdL4keY
z%J=m2Cu(YjgIQ;-J^07j2~qjfG&VJzld!hFC!Ze{b^}FTw5hk-$H%V(8ID25^pv#m
z3DFzu8QA1lo&h6&R6Pg=cpiu0dLB(}^m(k9@CZdp*VuHvwClO>B2+<ps-9*OQ<ytD
z4y9*q<K!i|#HsKnXTjsTCGKj4T3pa+z2WtdtizUQ*aR9p|Hx<CHom2x%kr#4C`?RK
zAEU6z@25TnP}VV94?#M!!@a1b3eyrblFHiIiF@_o#t%-I7IWTrSdw4rZD?XO5@twH
zNO~psuQ$R&OrDj+{IFBvrO8iaoRBkPiUb*s*fO?ZyNQt!ir*Lu97~WY_L%tgce38T
zk#_0%5suhdjjM02vK4d&HeY;)S{c21bq-+@#9+~ii_;)HJUltyWz^M45b(>>{Pa@m
z$`udWi~><Ot;&A=LaH;hiqBS6L;4_x2D4M5d~wdCq>-0h5Ar3il#l11TC+a*`u|^`
z59}vehRs{AAhc1;p2kagJmeQOkW&sK562S`Z`@$$Z7v|_@9XP^L~3VTjQVpnEsjrh
zv}mZKdrpLfJZ9VKIR$|Z<|7)_>R?aGr9pv|2kBb0zdwSS35ymkf4T+2W)7kFmE8%=
z4Gh}}4dIgFRw44GGI!yiB8^>{=;jSg6*X)`8k?7lVwhVnF)_PF^TM?v;By`Y;&R}K
z{h13!@4``P%m0qVLh$d#{23$T?4nh&d#=1c{{>88&Q+J#V!rq2jX9hdGMy#ey>nlf
zpTCF%Q;WW8Q&(3}xlTmHaOaMI`@vRQxzlHRKF{$yL;t+I4KSu~cXxMf;qk3P-c}c*
z)3?sLY?Vu^nma+{Z(e;%!$KAv6VV{&&9}}ZA07dcqzc6Xa<|tSXOkm?+-(DsafY~8
zw@hIExnvaWm-PgEQ;zK@L8v|09F^sog_ch}mC|;@qj>uIr`#I8Coy^zJVjm%cbYDj
zC-(%5=to;(SKIJ20e?$?a%Ccw!L3{b@?e17Z<Ze}Z!F~J?}t>+p<Wn}e#AJ&Uvi3x
z)xtq)JTmC5sPvPwMraT>j!}>B&vhevFnc;Swmj$g&TL$>Etzz<H<D;*$;+FCg|CeY
zr~sgLFD%+MX&H|D9g60O|L@?sh`U9-*{LEeE&c4-GdVdqh~%ny)C3O0ke~>;MpCk}
z4b9DXrYN7;z+GG0O^Cz-bGFb%1LU=t*=G=!gvx#vvQK|6Ns_9tvPA9#k%0{jvT)SF
zsSa<Vv8~Ob+7pZNMkl9Uo7n1s`0MY!C*SVXSola4k9??^Ebnip5l3wmb&j`s$AakD
z124Q`Y1nRMW!pxt1u0^5!7VWB0GUfRLltl**Wtu_l=Ni(lABi{nl+Nc(}#_irQYL9
z{>3d?%tdyGfeyfM*At_QNI)U4CASw31b`G6{0p!@e6vU~WGe$~>}@4It`F(yx2qh0
z=9?UqK4rs?NshJ<zY=T<aMcUZ_ccVdQc``KN0!5(^3dbq=Z~f-<xqUfM2C75T*x=)
zI?s-WZ<&1c_B?$t(LTnDewwx?#o47rWA-Ke5!N+6SGoXj18B%nBmkBXl4H29;E}4b
z>x<9VK7B*bPU5MCtn~A+&E|In-LqQ3!>_Fl*8_a%PZf;+ThaGJ_&SC|8B5bj!R(_W
z>7nIlsr})Oxs()I3x^<AEZgndb|**9M3ISx`4C)SVPZlCR@B7AscC7^5fMEjBN-s!
zi!@mj0zez<EaMT-Ya|2)(Gh=*-tlnXtmYoBbWs8%$eumUU{2bF{C=hGPsan|Xtfdd
z8@9W0SM*81uK1qy%nq-uCVuEO!h1cvvPUua2nxf>7Ta<ngF}|4#usu!Y=O9!*9e8W
z&4E1La#!^fcan#k3aJZzt$!k76cW|6KWbWZpv`{#d~lA6s?1H8R}>e*zsK?-1kH50
z9lL6ba|h2e#O);J-|ZnJygp@B`$4r1l9VkrOSDzgrh)Xg$w5_!W$chWJ$)Q|YwS1b
z@^QD+SvXjaA)MJ`Hxbwqbps5Hx9nbYhxzV2Ycbxl{L|3<maZ|8>v@@r&co^NCO@;N
zH@hB8uf?lsjqeGv2HVC#T?|o_fM&F{hE$!4R|<Abq&{JpU9QjAX!`W&Kw9qJbuwma
zZujl$Mh8Y-&)G?`S4%5l_UTeLpoVhS6*)&jE$7upfxPJV`_Wy8KRb8zzr(r9Hm}3+
zb5W54)a}K^jxc3#b}mUxO@#sy(xU;CvgNk7-`-svcXf7#DFXu|V^ec8oYH2)MT*+m
z<?>}fnmK`z0ep~WAE}?*cfVaLqF$Og?(Vldmc)3_ejNO{9Q6=~DJ#s4;_)FpR_Na&
z|5Qbo%o({hSm8Zn^W--TJsRoK;Jo{d(xatd_1hUyp60!Nu_w__QvxVFca@#bzn8SF
zJUeiv)mM;eSaX_9OY4Mt_-49XCb8-Z{Y^TR{RC#uTlW1dIPf1{FiqTXB?a<8f77SR
zCsX8T)8(CqibiEZYXvb(c`up-fb``5Hdwb@j+x`<m+t#n*%5^cUx+N)01(n7-`rT+
ziyydWo1)%Knh+cujT;P<6_6g)wP?HB^n2fPh+1~@4T)m;X*%4Wfq#CjLSiayNe8p?
z?c3=_Luz>*e9h?_oz%L9)|IAfk(A(fp_Cm%WdA$#u{<o#xhm6fa-zWR-Qf`iYCO$6
z0k<|2NftOm-ezRnSYL-pf;o6=93Spt?xNHPRAs_;ClKxpSr|oSN{M2lRVZ2)T~bm4
zZ{>7t@*$WzK??@kL|aG4-cCcmd2Y@Kup}^292$BKp&8gHwwGKu<NB9ZqdKeLcMJte
zaJ}RA0Br)9TdEc!PA<Z?r>3LLZiV*fDb=+#IW4gDry6}7{zT>cfdTQXnS9m~zm#iZ
z5r=p6#Nj{eMDs~lGafD&P1-LR7Y{DU@A2O23_5ncFZ3-dD@gYCInVd+aX2|ij4TW7
zxY-<rf1na9kr$8$=WtB*b?g~5t*<A6^+&<T)URE=oNYF^G7kG%?DF(=B)z03gnbnl
z`+It1A3ged>RWNcS28^_^E?7pm~?bXoBbhG{SZ6|e;LmC&grGGOIpFLz!KfweXjF+
zLDtTm{zsWB7Gh=So04aj?czT$$o~=|ov=_(#W+6m(b7nY<Ch;4+bp}m`x!>N0V$Uw
zLPJ5&%Er#_4%TB4%&IZA$g`(UK{|<<nL)sa-3eM+_}JKNz%&^0S<Akj%C1UJOvj6w
zEG@~Ief^MzRZn6*LP2#3B;dqfHzN`v(Kp#;-|xD#_q@FE>K4o8Ji!m2(UY6N0+#7K
z9AD9$F%2f<M?9mWjn}`0$}<MvxG`gF@p!ny?iTG*(=i5)`QxOYf8HJK5PNwX+T40@
zF8kcht3<mz%zrmt&OzptHJz#K_{Q;1O6S*L3<xg-xogxy%CYT3L|hF+KvGtL_JG9G
zL)S8hSclNyC*Xf(YPtxOg1np@#C%LlOu(Oyd*K2;rGQn(msQIRK#XC`2#l4!Y&{49
z0F_2L>}Igs)h&0jZvDioV~5vJCHF*oYwm}RtCF^$HS<L<Tww}GtXLUR$;P^DV@~a#
z6C-9lvl8w>N&e-E16ua^t4%v0KbB{8y{p6UEtEokG|K$%`}3m2rKE3uW(=aSkCM%|
z*1P)$Zm}N6Y3hQVVgZhe#ecRf4SK1nhBJ3=UVhe4nV#+^53lU}8borOV}B8S!4qP9
zAgPIpD&Bsv?`3zsvy&6dHyM>a%cgZhBt?!9nA^fL0G;w66i5XHJO~s3mgsO%ppOp*
z9s$7O%U`u5PBcZF$}#_(jd#p%+?>GmY$q#8PR;fUvMi(|@A?EWwG{sfKb5vf>Y1L*
z6s0N9iw;_!b}-($gb<eoy`ZSz4k+aN%r$fadGn6~LO$z~7|yC(+R+bV`gXglzun_6
zPXLDB(%OneOH1o1e~j**hjpBh5fKt{?b-xSEoh+#5i@gh&8TI~=g+xz{u3`EFhjiC
z9<|z^ofs5^AFre=D?8zGdL0MRt{d>y7X3=ggpEM$rW$QtTQ?KBGtIPio^n*kCB~##
z2be2|hfa6<wIxo381#NpS2*D!Tu-K8b-l?y{uDcn<~Xn3SDt!9@oAvQSMeav>X0md
zA<jO&FX;JF#Yso+$dq0}S6IW%!d%v#x~yN<5AtRw4aeU&BvdTtI7L@B3%N(z)03=>
zIB?yi&z^C#AULaFPoze5k&|N9-@A74I@n<AOa@IN98o$tqrjH1;onr$)Fg{nijIt&
zMMnZ2B+1Ilv);LrV+MhtQ2D$RK_Dsy%QH0ex;!^MeKM@JSFYE;vCf<R{4l;1U*(~`
zG<*{XQ0_}l-nbOC?-UL+h;&iIj*2EQQ+U*R(cdF7R0|>ucrI)~iEs4f=h&#uLm^h7
zb;3w#Y|j)`jsA-KUYQ4W3HhVer)Y=KjeWI8(#?@;GRM^M7cc6&#i~7yQhQ8H8SiA@
zf7$={C-VgHjcu7-dJi>)^J3sx0$-NiwORHQRa;d>H22P4Z$s6JB?o<f4OHKpckT=n
zn1n}1NBj8HLwKy2nHfy%>ys+r20c^`YE6s|34z&*5_phmYQ`3z1qHBx!AmYZoh4b*
zbbk}lIL{*iF799~gLo<?7<-#Dv#`8D>#u!F2K|af{P3dv{K&p1E&`ATM7>S?Q{gl-
zpf+K8D0&s&b|JK?=V!!fA*s3I){yhZz}z<L)5nOcteo+s2^#5&R%Wm+kFd9;<a@01
zBejxlUd|02Xui!7S#;7-7~NmKE}*ncdrJ@1{&jo>0sGPTWreHuPo12d!n$l$ysK`7
zGW%qSM1)6~#Uf{%g!|Wwwsd1eveM?wkh2MR9R^O+-(g+0YZO>+5I6=Q6r7ydF)<o1
z|NB7N)<}sh;M{_gaS)-4X`@1{98ib4BIx-LiDt9STeof@9gqri5(H(sZpoJ^f-l_*
z09yevVKH6el9e7--aol{n``#r^S+PF&C09x#XNnhD<0Ok5awK8RaHTu&j-$?UmxGm
z7P0k^I}eiXYX-NahW0m;SvZfU2eR3GW^P8t&8H`fa0Q3zjy+Y5j7D)uD37z2w6>#{
z&PAR8M0^XaI?ol`B7hB(B*%4X<l`kg>Tw?9k7K)-s_D$Dj6>v{<F-HYQpYeZYZp~@
z#Kv?rUSB;VhM(Iu`y#aJ=CpdugQPFwJr$$}Cp#|7&e>A5*IcUgHpT?qE12Lx1xr=w
z;NVS2NlHG;d~Wy(ef0-~4ILd;aLM2jP-~ak$*HTW>%_M%Y2FHj64+^XMTb0)PrvSp
z`|)b$b>N<PMit)6ee+$zecB;CJ6i_pXf>&OL1XqE+9&DC%CKMgvzX@R_lzIuKjKPk
z4%17*i522HMvtdHp+|>@obw&iZ9DK?Y~<d{-}uLC7txJP-s_Eetfpi#FVQIAa1D{+
zOC;Y-Z2j%_!fHzxX=q^o$I5|&7&Ps^@P;I-=iZfWiE588Ahwj5_7}GF#k|r#_dGBj
zK30ElxtYcCeci8tSi7kpSDq{s4NdTtycH@d`c;6ti|C`|E1$9MTPPKIVBcJoJVYn2
z5KOpspf?YwyRK#}m?3)V>h+NxIlN$n4aqP7Vb+y9zI-_eC$qYmT7}d4tvh#K1DMa|
z^S7=JZlBz*Umr+Fyf``9hY3o64vf+Pl2cU7APdgnXZ?;ZvQba^&ty!f!nClz69!S2
zBVA@5ym-0L&(1n>(JlViT{5KVI{n%T+Mxou28e;#x=8-pH<$nUWw4fz_Ra~2yDGNb
zK@d7D4-$v)8Kwc!gc3qc)&7{U&~bn4vv_u#1rec+g$fRv6*?EY>M!u~zkcnqDf9BG
z@h9QT%g>MH2bNBPn7a@kA2H$$#2e!_O0rnJXC!ddf=73;o52V3P@uY-q4koIk{uV_
zWFY^fW5JFRpNWlaWYm4Z1Ga->`|5DsslsEh&&T;7`G1Id3$QAh{||I&6bU7iE(Ii&
zZV;tgq(nekKtlRZ3rYw`m&5^Sq#J1z>6Gs7?v6XWzyH10=RAlHy6(*G%zWb$rb`c#
zCn|s1B+sx~Oc4?Uykd?HR5X9#nhb)vkIS{-1aVH@I0+|U0m^SbZ9d*avETK5J2QF@
zD(R2-mR2MUS;&>J_0qq-@er017l$S~zf1JOz?b-g{%ch<sAC^UK^WXnztAn#pU+C1
z4~%C=dJOAkzp(&ZrjdpQ8GLUK7{yDRh5*}VdmFF+3DD%nMn{clCSCI=`i>FWT6v}y
zRAb%@tKpM*?^wAyKr>{l%6b4?L_&ZnjzF9NPtgXC3m{ZoX4rh!%4#b{T2@8|elK18
zu{Ch?1bC8OQO>SNw>Jsh>wE!1Oa6x#m(JMO*r(gmz?iubvGbJEH{yE)uV>GBr?hbJ
z;n?NjH8F?<f~8*0%uEL=Fc_f-dyCI5EiK>RVT_6HG%WJ+M%HVLuym9xlbUR(Y0w_a
z!lX}Qs0DmI``KV@Qn>66Ix1Z8u=ha33!FSgWgJXbm=N0HMa8jE-s1_rJvAi*ZIi-F
z0mjI7MMXs*G%GTXr2b>|_`w~VS)c{twuJu-6#_`f)r*NB1C&es1HI}M<Khj{x>>q{
zx;sFR1;``=I950Cj_&BNu(losGAV<D+<bh__VyRS&Tb$?yAR*L?;`BEIa<uB#=^o3
zR9<9d?@UaOj;<~(9RYNEIK4ExZgpAtN(iu90l*Qro9nvra-M_Z#YJQ2z0hDNjc&J9
zzm^F`l`r{;hWbcM6r8|6Gu3L@o&L#bYKj9rup{@g(R7ySLq6WfbJ+mxd#}l@Qf64x
zVjYF-J%Ynl;aAXC*9-GumxN=A2guVq7k@j#j=Ue@*Bhzr#g(6NB>4kF#8(f_DmngF
z;Or-zyfeEq)6f7|+`Y+STsExi>>(i`w-z@*?ggKMA2?+E#X|zO%V$hX*_oN3`Z7E^
z`W-O>bn8*Xfxqbmr?7Bsc6Qs}zs5s3>J+fjtvUM5^%aQSK}6s3K>@rO0N;;|{-?;G
zNpMgEeU60OzA5Xvb#i#~G%;NMaYT$81x0#V!Oxt*sjGA6e`|-4m2}E82&IhcclsVD
zi1}jjUcACMK`pgn@{3`XvG1$%pZLj;p6knmBjL*us8$Nw`307>wu}@II<`CfGWeH(
zk)(FLP5VgBpAwsV9ww#>Ot2WT&*^U?H(QCUfq#Ju?0<Y6twFoW(aEVL)%pJY`@r7Y
z%xtASk`bh4KvoDv+|{)@GI9VUW<m~cff$?hB>(2+D2)IF9{0t?g_y@VIPe|D%gjex
z07V4VD>9NWpIEEZ8%D#jFgFLndX!?GiK?t-GxZ+EoMSC5m~35A)&C%jsy<6m!p(dX
zuh3X&FY0`yO(p+JstZiXXVN?0|Lo@=rej4{VjScPU5l?dUmN+6ft6;`9jBu_@o~pq
zB)z%-v^2jJKUiGn=!DMdp)g&o=~u>)X$cZ3rpvLYUV*6a-P!+`c3iHZXRB6%La50x
zF%<?)m{3?m1i-dwm3_ExZ;=d=egGE%c8V}4+Wh{E0sg!Xpr&GHpc9f19}oK0D{E^Y
zPGM$ZIz8IZ6hyNBXGtLVQBer^lCrX*dx5%biAX?V;w0!$759$h>#P6^HA2Ej82EXm
zr$-@?fJOvF{DG{{#Y7?W+19rxY}WUSj-g&%Y<SF@b&gvssB^|r%sbE1lJYuS#lMs*
zRFjjXi^@)G`b8Ekqq~~aUmx%L3f<2{8}hIC{-~FLEYgAg@CgvF#m<ixjj5S>)~``T
zN;1(m=uUsQBy2Vu&05(+SP!d@pm?Rka5@+MmtoW?;>#J1ysZAsgQWBp`!tX4D?v+u
z4HCk4mDZH<hI#;~I+Ux4g@pw?9BgLl!3r&W$+$8%7boNl3<t7OQfMLI4Q%n@!#8TN
zYv989P4)=n6&dAB^n+x;V;MNV1e5W|fyKeab!#5j-VU#|h0|yIQS=ke+VYdzzqE2(
z+TLyx+tM2pB5S?j+Zj0E!s)!xg0Jtd%au&TUSom9i0~d+TewbPYi$Iz?qj`Qm98^S
zn1frIzK9))@!e5nj}^8p;Zj%I#DW-YM3pWN<x?V~4uz8~X`w!SDU0<-EZceyZ^JiN
zQrA%U-&7&`iK7#`C><Rt1TtAOlc9BTG8~*B3>*?r0bK{rtKm}s?MqO3{*(Eb1FYv`
zK(^TuO8{lLKw=ipG;-jVj3a7FQa;6yD?_Q6x?FKX!5Sa%c&D_oooM~}O&^Xky?@;5
z2fVO{gd}+dHkVQAj}oH%B#!_*(tjB4F5#kVI}!!a=G2NH+ccu0y5LhwrcuH1a*JRN
zzwoxl+wJ_UJ+19(TziVMJK3)Y|Cdjtm7)mqdQAQK^Vy!O`O*FQ|G)$QOCT@bsg|W=
z5DZ>w*SztT%!-M@$|=k<T<eggvZy6zWX-|Tnxd4S0^AHwxfud7g5jj<`T8L)hasZs
zY<&9eqe`}<vXyxePG46p)?-sGR-@SLqS@`}ftImHj{h=lmrDA{>I=11w*=7S+e(b_
z;@OC{e-s95TX}rh-u?Ob0V%MNATruXzNtG94nl|+HS+>h7dvfj)sO_;SAsHfd0Xx=
z9)DGPItvVWWftb(fM5VBqto|rxp<meTT+o(ulh<xT<PtfSDM|wI~om(Of^poj$cgx
ziKGJ2uYn%P8vEseO8%Z6fAjB^KUA*R;h-NH()^DNq34C?;M%L@K20z9TOHI%flXsF
z$f2mIcYwKVinhDEFvP^1x-TzNxX&aUbC^cwAf@yh&N`&f1sIF12mP`y9n|$ax|aJO
z|0dE6rkkJJmq9>=DY!T!xL8RV|I&8F*0LsQF}+m6Bd(I*)J41EnOe-wM!AEXw9tRD
zylu`O(Ca%Lcmt4ibt2HadUsNse`Bsk>qEfQ1m*o=TR5_o;`1SsGqv<{7_l?L5#A7<
z3nPo6R^lFdMMsm1uQ$cl{}moTUwVMPrnI{Qz}#Dbk?>iKKL+^a#}7dA0D2k;2?=p0
zynKAg!U#JcAp&B|&`_JDj_Cg`PeB|#!~xi{UUM-m^Qm-@?an6$XXm?XtODi%`3H<(
zl~uScUC6ogB=7py_|U-r>lS?PQA*K#^QP8eEuys4{uW3T%_1Tq0?1{6`-DQn!Lb9*
zFA!jAbpnH63k#>SaY^(uBPwCUi;REJ`}l$EdXL9nJ~tf%$<26N8gi@T*3|epbfn*B
zF#LmB!|ofUIm~HO`F!wDZ=H_*$*?Zv|GaF^SPtRg2i77lK*toawX>^+Z?12nB=@t|
zy1Gap%M}*ZI0+6-OK>VEDJdZvNIoA@j9wxPXkYjnEQ40&i{EpQ?ae2L3p5u1GRe=W
zwxMn>*m$f1e9#p#0T{88x98h8PU;Vqd6Wr>aq+r?qYCScPfVZ7D5}X;)LekRHw>23
zzUGgU@0XuM(#hX7`9f3>sG0D4UY-K4pw{MQ6QI3gYKrPr{ogNN-{w5+QWr#(JuUhg
z?Opkf83q{BtJo<9+)NYOH%VV#U>ZOz957w%NjuS~HT{9Y7kCd(ncG;MpCT@Qb9tJ_
zA@nv#SmXb#m*%%*F5kd0F14P?fvpQl+fL4uLWu~Q$qD=uDru8tBI898Zf@SrlSrx`
zG(uDD2H%iR@ULTso->p5;07y>gt(4<CwHY3bmn8tDM$%#Y|YJMviXr!<@wKn_z5CL
zeuRfx58H(RxyxSpBONsSW$TlHm0>DrPYlEP#~QMe<CiMCrFrn=pTs{8dXQ(qk<K`%
z)H}9>TXPKK|3hK|Xi)<$I1WsH0jcu;#4gNw5(Op&X+IwtpZrNJLcGvYQKZ=w7>%Ft
zB+`C;y`}y-r|O?LUaMXWWwpUb0B;uY{eX1s+=>P;!lpU*+F?%V9U^AO=#{yNP`bDY
zvqd6fc#+7bwYHd(l*qveUq2|zRPj~+Uj%uCn8|znh_<N1%}elcGXT(qof2N2_hs%0
zmxz`QD^TtE#-?9#&|tjcFV)hWB)mO4tRJE6P{u_mDA3R1(ZlW<e){v&lGr;`Q#zvo
z8CUO<tpb+8e_4A$3R=9xjJhCkV{9xOJ5%F~f2+W<w5YOYu}J;7J9&rxskM>ySF5;Z
zc1$FHr`;%;X}#d&_-*@bzp|*mz@|+-YEcpQaFmxm92lJ;T5CP=AuM4H1$(mQ+LJ^8
z_I@jOWnxzSO+$S>wCRP?i@k2FUSI=ZG?PhDXKUXS5y!eRX+JNgXAm&%1z!i5^5YMy
z7v}t|*_#EB_UFcTO~#hdWR(j+u^nn0u~D}l-pb4QNjS@>faLEO6xUSo0z&7zcJFOJ
zX*Un)g!;wfCGmG;nwSwS8Z(5F7i4Ahq;OG+2=sTpc{5@qKtiy$j&>kL`{J61tF_yn
zEM(Z>e|wDW@K2~-8BP{u(a8Iu!UP(CKxAWbo$&La?b_JcPyG^<M)SjGtg9H15DTFJ
z?Qg%9On+sg<BxBL?WXycSI-h&-g79VI*K45G~c=ePc{4^2|x#+_puidW09@J+`ot(
zp+d}Z`^sn`Mt63fA7lxJd@rFBYFJ&`Sd2k&2cwTm(QAZdZB*xW(vb9ic-l(*E-a5z
z^HMoQSg2VHsi_V!Yg6w+>{QbhHdY<{la8CVELpya-G9T(IO2>xN||jVXIR>lK^JVX
zFD@jBVofo}l037Dx|^NVXl#1))#L$uyv<ecvlIt9y(8;YbTVB+bRPxl<!@hpl}|gW
zxf)jaLltmgX~=j=`L=~lrQuYd8w(wFhtwP69b}R8v4_~Z;OBR+(TVmhVb8C<B8Y7i
z6oC_MEOt(o9L=ug{MQ5$*eOchP@=?)(xRYLh+Kx##xO3_Cb@&{!SLBe^s)Xo>T9RB
zH4zc>Dgc_?n3GKx8XFnFL2y}Ns7g3gqax#8cMgY=<I_R2Ss9?G(q@lv_ZEL|n0?Zm
z4(+W|LT}U`U0ocbdnI1dttAl?G3jW)^f}YAr3~L_^Z}DTf+IrD1cHL)C>N*Xh5Nfk
zXA?LWlpYDR^>1!Dknb6D55Ad6PzVx~Ms_8)7RqVv&|$GphQiTDv<EM+X?;{`Xa&3A
z<Q}PMo?U>nOn7GSCrQK&a9q<<yt7zRdQlP6HNHPpIbsXux6=I2!(!Q|@Q#igBG2^r
z%&-`-5?WkI9{e#OcMd~-+uvv~-hRCHNK%mP+DXMoq=-3(tY&e1%A3OIp_%DN@p^)M
zHO{^POsHPmcK(lIz;Ad>)nm`t!qO7{zdMu_Ng#LMF(*Vpe{tz7GC8EWFHc14Lm-h%
zB`j4i)uz)3s-T{|Ng^{ICsV&ZEjaB=E8?}$4+cD`SLp0w^WvKr2_T;=ti1cOs@~gU
zYkGmrGKQ(&#2wb=5%_aCKz%cf8$_q?|LQMMZ1J~4$Alh#ny(K`5!Nr@Pj59mKNwGz
zyUhhlj=znQ=VO*pJfapUyXpCowLv~Ac3ra4Fgrnyi9npJ&-HgJsfJHGFaTn#NghK)
zc$Ph47vponec6;xPi5E6med{Hz5D*iwEmH)^hb^z8YD4y1n$rL`SiyEI`#CsxfItw
zbGv{rIi}yLynQaz=r%+FxATx~%^_Ub6FMyg8I~nmu-edX$GugsuqiR~qW^=?KFKb~
z4J%4ZbxKXUZ5pnY0o#6S%L>$ptgQjAhLy_`^dan~@27;ybHkpk-fBX4i4mk~>S`Cy
zIGwVU_1EMxr7#s8Jo#AVUU+kwW-IUJ_l;Nk;ZERPmLf9XuC~Fh$uzS<u>iNEBxLKd
zZ9=4g!1<LPFG*&rbHUAA?9>=S@(#yqLcMfUFFNFKR@Ok<@970;yojV5Wd_3g2dwmr
z$pgmW!+||Z50k=<4}N!6H80HvKNV;OALrzHe%GTtzDP{vqu9Mr-*FYm=tXo~^QG>u
zx33_<Ih@$8(}+*bt8iKPzbdKXUn7)IQFHVc@2V-s6Win~B}_<m<>VA`dDNJq@+}*X
z0-i1gcMi-igM#>|dmf3^>f1aFDDr6Su^C5b(Ga0RNGW@#O+zCC3rmyX8na3r{A3Cd
zH$|tFzDc%v58-!BU+^dKgQ_Xd4}ULwpXY>0Vq5V;mF|TqO;+42CMnL~Z;#XrwU=ev
zpCKElORs0-ja@S_$LkC57|!1weY94(&-I**qoM4xx_ZrCqY1ORL4R09El0$Uv2S1I
zcb^G>N0)6+V|&-5CcM%sOBTr!tipKwcpG7GhY<3^-;aN8{oTP!h3V3wdHEdkxH}A%
z;q}kW8!V|BOYpq3SU#Nt$buujFN`o7!ClR(ij&_}9`h^N{?d%rqn-p1MYUGJ*>R9}
zlC_+n^PPaNR#W~AFT$0`McXV-X*i&^9+tqb9!>`@7PtBsyY@RUZZ59&me3X_J@<_3
z^`8w5lqaOjO+3X6elpMnp4JJo*%f>(kD8(qoKj}Gon4GJGxi<SxAa^H4pstmh8#B)
zu?qR$lAaXtSVmsZnNL-JA25GIrQi2TL0LK%KN?R0Ji>AHl}Yge`moB&<-O_B%=;-l
zi}ML*2f)&hp;9ZBHyz$#asIQuPQESbD%ep72c18}9d|CROcF8~^p1>GgE{T-FP+Wz
zHt(c<2Y^cyuT;9LKiA#;rSoT>Gw-G^w{b9IK<dPP6`6u#h9=e@O{|64`fmO<>vZyM
z=l#ru4<CE>XVX4t&t@8U;z=XJIFQ=2^ml2mVFvM*hJg8>ITn1LnpNlf3X0Bo)?OIu
z9nzB8{X)2Yb_NIkq@XeQ5FQWmV>WsPS9a3=Q~axNd~*H%jOO+sQaUE`>>@+nwYJe7
zYR}=IYwq}=cqN6zXzox2`vxDod3-y|m<V00nhGp5TidWeYaE8JS(r@%#@k1L?^OSg
z&<lb+*{A!HM35A`_hq6;J4nMSynb<GPOd(?Am37#lWqN-(r<itA@I98^bGNVdTm|T
z*sI2<us{PMhzWZ;(r6y3Q%OtvqKH|jy<?Zy9xeE=?6J<w<%;OoOy|VV99kB*P3)B8
z-G99y8)g1&?P92)M)h@me*W9HZ@@;}&CLx52WL&B;Qsk5U#XU8O${{h#xT#pmWHA$
z_w#2_(PX2ELZNWd@GwvIPtLc_LbyCbxScB=b*7fO9!=+BCp$Mr#ADV*1`o(cV|7&C
z{bA)B9Jt?XUsW3{gx4aXi~@|tMynH0z3eq7E>rt=vwh6k*F>tv1N8+--4opZ<w|R=
z8NX>hwzFZ4BXMNU`mu2z`XQ6mO};STU@Q*D)<)PtUp?VCp2jKF%h%?JBX9Xox}=PU
zG3uyZDFuF6_OWYC!T=eWkSu#3^eMhpgNER`)Z!R&_P_YYagyW_7O%74pxtB^eE6XG
zRYO-b)irDB5AexL>VDoo#LYKID%R#<n{3V5!1{9GwbGPlM3kzETI=w@>w|RzA;!j<
zfJ^euVOT|e8e`qBIzFqyy>c@&?atFN8=a@(8O(rR(uZFv-8m+9`<4sdrSP`lv(jjk
zmXx#_Et&%*08sc$y``K0I+FExS^LPz0Kev)6o#+My;$dbM@L;M8f24&C*WM6z&jak
zYP&;^Rel~qh=vJpQ43)AlCOMe7DBBAD~+FQ%$&5vS-zu!FE67zmzB3HEY($YBNJmB
zXBU_9CMDE~rF1=B-YLBtud~%x=5rh(d8k|3g5PjP^SN^YKY4GetTCma2k*RkT=Gd*
z0&Vz7QyNNRQ+~}s!nlrboFrGPQ}MI=;(CO)n;If2N;7K(pz3Tn%+M%<Grz!$7e{fM
zBhuAZ2AjF+ux@rXLWrKGCM-_3P1udkbuEe(rv>+e5<~vR;^tKL(FKS6<~1Dww&s=y
zyrlZsg<iYeza``FQ;9(BHV#=dGgZS@!-Rif?UYtoQQrRL)pwTA2c9>ksBQhPDsWnl
zlh;?K6Y;C=khHk20O={Jn{`<Aqn+m+dF}A+_a>bEty7cPR!$<~x6z5e7lwO{`4`ms
z>S_R?*9KhEnEJm<c1?f&VBUN5%<%75P?7tYJ&V`0s@6PbYvcC{^72lq@YgxX{34CN
zrA+1Rxu%gVj3NAW%d~R;qz=FLMltK3-{H<x^P(5MSwV*Ba$v7y7OcXAlZkci@~cOy
zgbG^rya@{R-#CL|n>9+TlZ~sX+NU*m$hp0<)}9%Dr)$5{Au3&~&@C-SvNd95)3Fxa
z9mPyZWGDTy3CY=?;o6_}7xBB#j)(rtgreQ<I4!N^Szj=MbHc*no!1N~Vq0fC_d~*y
zhCfh`UC^|RwI6(gFB>>~RzKwQkj<GGq;D*LkgcGux&EFv-F>f&5pT`PJWn1)PU$%d
z|6|_VFw!HsgJc~1hB$7m!QK|M*xu<NObAzfDg5f)yUi{jOib0BA=S|JwrYhUn)wW2
zEjUs4P{NV>&E-W05T}$#Idt9_pUJmc*8XpYd%4a}I0_5NkXcDhqv|mcVWOjlNw)$R
z1fazMb?@=f(Y`f+d^pU`&ISgX`)8NreAcY*e(eHggH<?}1oiK!qU`Z%Oz6?pO4^Uz
zq=Qkv^Fwuw*T(k0Xzv#TBEx>7)pYmtdFRixlN;x!Y%Du7KW1&}0p|yVv0UkJGTM!<
zQZJwln5aKmwK&;W;Dq1)C^dQrX+0C!-5hemr6{QS#9}!zB)|NPY8;*^#(7E#Xb2|n
z*n21c_Kp8BUjHC=gF;2rAw*rz6jzovS+d`8!T|Pvo%}Fu+4^T(eddC%8N5|a&HC7Y
z-JQv%?(;NfKFbo1N%51tFOv^#@C?<5wqa2Z=U=g0NGYoWrR%EQzV6?4JAtHRI|U8=
zh>bh9ie*VN`{jN)CD|S=w!Ot-+wD&cMLd$8GS3bi$a4sP#ei4{>tKzZ63&106Dz@c
zV`L5+yV1`AaOV;tw|=vPVpUC4ZmI>`?$o`?tpZX7FCYKQWU})#1vS;nrC5C!RlHEI
zP6;V!VhLzfJX~D%phgauj2K=&e*7TivW;BnB?@Y$!<GOR^U_&77{u$RAZkVkCskA|
zT02oU$1DKq{2eT>(u9YV^R~a_Q}$R+64AVfOQqbOS111+8gNds9)(FEUe-rkCaW7?
z&?p%5FuiunPJ8hWtsEH_VKSGyz7J}T$fa3~_FSm+pB*NM-9cUcw$8^WZdZAXd%=-h
ze=sPcc`(<&(iW33p)?)hjt6DII!tx`Grw-GjyOvRAX6-t!g#Kdv$5-ER=bJ+>Rr4(
zWc^*(1B|`0^75a9Qq=M($`4~Z^9u?vAt2zDcz;a0Sc?^WhnWUPC+z&d3u!k-g`(xQ
zXc{bex^0?^1K%FE_Mk$uo8z*)Ey;~$d1s5r95Nyz^jpb&z%manm_})AX;}siwhRmm
zknQQNfq;!afkZbgQN1U(tjCPQow#|uCCH2UrFme(xAMIO%xcS@xnKW`*G5t~nLZd2
zg8Iw{!(a_z_|n<e;Rp{}8j8?4iJWp3bPlgVyO~stz@n(<(^2HXgdUp^$>~ep2CM6N
zAV12?%iCO8Vf6S}P+$#!Gyq%+Kv1Hhq5ws84*BPG>NGMwd4kdi08`(BpTJW@&pA2c
zd9C7^9)VCPq2hdFFJp3kU!zu}@;e5k(c%i<dm_1(8|V<9ANSi$$i|Uz*>5K~LxSLy
zIl^%=H}T-aC}xM>W2N1GFV_(_u_TZTvNd14tesXGX6qhqd4&(%?Z}x<Aiq7MMD5Y{
zTuH8^8CS6Wn3jSrOV3;zzd1h#NAJXh8)z&D2nc`<1&FdivIm^!SU7CDIg;US#u08A
zZ}CP$pbkEDC89$ZiBevUT;w+LQNaTIde@b+T0;%{`5oy7B5WVIKnHz3ETAE8h~gDz
zkVaUbfLmr1QQF?LLN>6jq>zp;57W~(91bG0iaAP4@5X(!P78IhpLQQ4kt#{X005SO
z>^JqZTb!pSrj~hGq{;D`u!_c*7Ib_2edLs8ghEw$TE+9A=grB{og^ak{Q1Au)|I~j
zgv-mz0KVky?F|%eKoeZZX;bSW<FO`+zP`S4eH#EH3r5^~#$~xD_r@pwDG$p{e_h++
zAq8HjjYamxzy#u#jSX4B;^n)lhOsIUp2Eet-+3M5g_VgA1oKJDmw3TDu)w0mSz?R)
z2ZW$%dgpb!`>EV$1z{zI1ZWy$4_Xg1#I#foGn^OTk>DFfC{KAZPoZHL7t*NwP~jeW
zl*hd1pVQqD)49M$=?9iN?VvzNcnxSLgZJg|=%}-!<1xFWl$4*3&j(9O{aP2*TYNuA
z<A4b-Y-(deO+-{?KEQy316<^n7&TFV-KVvsWkOPtC*Z^rvuJQv<cEvX(BQ|BKR5Id
zu6!xOKo#1sS4i0UWhX#6Lcn7yB|oDbI0aWbnHSnn4TfvSt&(n-6F+CMkrve#?Vfqn
z@b9@5Xmi>xztglN-1p5W*1KPt^P?=axN;GW)d(|CbeLd3eiimeI&0lafZkhL|8gH!
z*-3JeGj#NHF^qR9s!G8FjUa5g%2&te=)C@n#UHQh#z;7Gp~kKsYF21pZpZwXFtYt}
zK8WKUF760`Fn{<*7my1iS^?>iUOHk(sC|4KNFECTtP4NC%UXZNR@C=e%h?e)4>&Xf
zU&$a6wuma@<lYqyz1maIR($*REzn7hI&i|4y)ZcNHhQ=t$MH$F_i19uVP)dy6a2<h
zCG#9uAJAJc+<K>M1osVPSkDY?cSigpbsnvuifhnDc84DsDblpHabQ6I#y)bE%QDi%
zN-RL);Yvc)p4&gBb)%;#GvE~*pLtk4b>Lpv5DLw69%WOTLc74n^(NB^RvNT~P#1Z!
zQRwaPUadcqpVbkI1dZM^fOho9qwF6UiLECiC&z)Vudg3$0bv(30E_%JJXI#J$Tqvu
z6<6wc<$BlGZg=*Dj?M&t-TMHb-O9>CA|mk3&Ukqzcx`Xs>#KX13!qa5&;*V<hNVJ(
z*HBjnh)IwRdrB@wn!po+cUxc&6l^G44-KdmTh?N!TKpN)c}SeZ&P}x3nQz(Thg;S!
zd8@~PM58X+%Hb5#TIIWY0U#?sDdV%D^X6T9m7ZQp1}PksAC?V1VZV<a9>t)Yee$UU
z#p|mXyvAvHiuOD7Z9JQR;mlO#zbm=>x2sW)KC92!M=akC2p<Epb|8F~UHo(rtU>6#
zHT#(kMU|;A4^O${i4`8AcOP_xF(DI^B|ksE$;rv3&e+a-K(FzUsQWL6i!j=k+tYO{
zW_YXiPUo&422f7u7vh%GT?ON*8K?(8!P%TqyfD|Ny%-vAm0vFMefw+IT>fsaf9?pE
z*HVx*fE;`O2%pO@yt5YNeVKAETC4;YPRF(A%@6<lxQPVfBY=WWYS=i`g`WKziC<(~
z_{$RxM*O!!hJ)_8NJAMYCrasGSmE|A?Nwq<xyB^nLyP+^{Ud}}Q!bYZG(>R0<;G+~
zu3047W;!7s9oXhRN>)f!lcimpd{C;m-Sy8CgSSQ$9Wo1jF80<C8CHmFT%;6I^p}~D
zrp~8S+qZAWfwBOo$lP2W^tKp+GY=A%mbR;WUy=bj1{r)-q6nPR7T%?ot4%U;?G}RK
zOpha8mtwj8a{unG;?D^e#V+ixy{tt3>R#I8qptJZ?N9UG4k2j~<Yq4au1@^mfcd1`
zdaG2*^~}xbVO_YK`l&%nQM1nH91C(&8-9MLA2kVIFnlFs6}oG7Y5bZN&>Bw$uA<4t
ztO{_2?(aD|^CK?qgmQc@-Rn!;L)&7#mus=1W-k{q^R<o5^d}PwJX~63Ld@fwBz$65
ztY*Xn_KslkRnKit2Mt>LKY<=;9&Xj6ahgNiK2Os9S~B{w-;k?a4NlIJnAAKgN~Xi*
z^Lur>ce`aXvG8i}sS#&e)z(p>tTW36>sxE1^sXGV+b?|n{24U=$#CcyBXV=)dxC&I
zFjQPOJ@KU$;}X_~ub%WyeA0J)oV7lF0=vB#8Ls2QbksON^H4l0hXowUYpX-Fwe7b(
z%U*dsKvp>9%uN2fdNPi6v|Vj#-56uUhj9NrR-<ax%n*~C%VcxOZ0mD!(V9AuP8d;`
z(8LXgMph;pFUDRMZG6i{N~Cm=tt|~TInC{~qRdW+DZ{QuMIyD<>`)$(FHl|gf<~~a
zx?t{aa2p|kY`w2i+I7i?cU_$y_fWxxSCF{0Gf^tfReBo@^sQ|EodP~%e)#qr?E1eh
z4AWRFJaqGP1}FS*$y3wP_&$#IWhtxKKL!wVAW)I6EX&wBoguwE5e7y*q7P3oAZP{!
ztKo67ypEGD)!%0a$yN8KMKF);?s+}}g|wJN^*7pT90YfHh92p5-4jOEC_I{!*Hlwn
zCN~S&-T&ms?@3R$ukCujG!XiuBHfhhzjY}-lUZ4AbD$)%u^f8*$sW7<ez>+uOHR(v
z(3=C>ia3Ekxe4V!W0P#nam_&3zmw20_Z6JArStq?d*I<is=o{-i<%?6z;pQ6vE<B7
z0hVT2_#QjP)}f&r*D+`NHtN~!#vrOjx$@{VCjI`yhn0Xa1h}7JLgjqFM@N69gKehv
zU20uw>gU?=Kdd_9)2LJ_p}B=YMiv^LgeY=>he^7hMtAL7RbD49?>z6GNz1S)gSm&V
z);d+Y1Kb_2`Obt8pO#}Y(J+4r-LrKLLzo{uM9FQy4zoVl*jOAHk*)R^NG(nv368e5
zVVwXF?o9iaw=2{&ZQ;!>8<5#&AixGFG$51tHv^!wRhft&08RPi?99i>iR$&|k3>LG
zD(Hwld;Z+K;c>B-baKU*5qzbjhoGr<j`eA$#{zqin6cdLB_*&udZsDnhVPfr-~yTe
zpcV(TfCFrP-VpPPfrZ6g2*8n`J^<>_skEM|aklA*dOH96WP1wDD@lZW>W#AJJ{7%x
zU+dC)ql)?5_F!Gdvgi6|L+SW&__uQi%Gx!dp04k5L&9Nl0XyS(BD(=TpjNV92(E?Y
z^)F2tt?_7Mr1>P*h%*i(3dYYK0s5BZK<~0p5xDJI$pO@gLcMIDZ8M$yL;TL+fncRs
zeziqfjWpfut%KxQ{tQ%rTwH)oQ!i2|wsiqT+{5F_=MEu1MGuHf%gWwX^r{BU61fh-
zqzV+%@ZgX7(9}r1^d(zFTcG&X{|hD%CPWiSo}OY;#1(VW`@d;%66_~BKYl<&1kfsu
z%Uy8^d^Umt0?n<hrDGq3glfQ)RBU*F$wm5<M9lYjZoj;!f;V6NcUJ@w$XYKTSgI%^
zqt)Pf1t{A_MzE2hf&%FIf@u^~>^>HOaBy7yHwFR;tE{DsUZFf-YtF};(D?W7B48m?
z^HaY=`|m)${aYTWw&+y<JuWcjgf)K4S)o@Cta7^Vo_2M0fz<WnTIlTJqH>7xzcUT|
z14n*Jv%4NxvT0joFHorO-+vAR39-Dqyk&IfFvjQq8zKnQvHp=_H3*?2MjFjbOmNA#
zv*U(@&b2Jo*Sjy2`F6yvf{0<>8{H(r2Cf}3kA<Wk8Z#rGR%4g1vUd3GiibpQ(`#@o
zQMOn9d&OdeR?5@&Pm|&6IAG)02$~nmy~#G_=0AUUkuT7~|KDI0rcti}2oNFwpjTiB
zUD1a1e*^2w%3H|u-f>tYV`Idoqo+rMzy)Tc{wq`dcgHfARX{Rhef`%vc4&2Ntz1>~
z{~d^^0Xp^_iPPfJ)x<?IZ_wa>qo#&-tD3{x@qbr`efqM#tv-VVDZw>BM#iX|<PMa)
zbZeXxlSPrN%l{ZxUKsz`n{^9#8aKpqgjpSb<?5I$X8x<LT-#_*`9VbCWT-;#Q}D^L
zc~4Hhs`dSm0@d2aOvA(SJ+^zvk24c%&vy~7*194SlUfb_U%H+30YiW*B81gIWAkEX
z9Zp+a@qMBIJ?kIYb8?R7>M0C}`oEWjQmXZ5(*Md6ikm<<>OMJnLRi={*TYpH((MGC
zH5B_Ls4?|vMlW~bU}fEh<%<uwl=a<Q_Kp`QE$bslZ?0@EIHvDBwE5rtxf8_Au8kip
z9;<Vcsh<Z=;OUMc4pZU3It7!V!o?oGl?jn9i*;X5ci@Z4+8n9Dh&autI8BHBjxW^9
zikGI#t3b={fAgAxpAtz0_6E?E5(8#$0LGWAk$?Bjolj0q$NT#wKz)vr6M(Qe7jAJI
zE0n6_Z%en@+H+~4A1Rrfre9+%`6fWE0xSHEGNC@~ak3%CowI0ro>E%bwmvnf1+Qm&
z51UF6-32Njr<2{%q*apPd+lzY**Pu@6aUt*PX)Jiv6S#~#S4}+L?SZ?h?KyXuMCLv
zX->Ce8<LY|Kr!Y#cS-M+Fygfc=nbyS^w!l-2Iml*&|OV-KVuKAG`S%KKG2_DUX)*~
zz!?8~%LZL3&Cu&hLDK>s8=Ihx1RMSTZ^z7hN;i3!SVeUg;$3c$u5h<8vR0_zh$q?I
zn8~^paYVL8x_1`O)<Ew~OkJ)TImj%SQ;!7}{PB`&7+4xkKW*Zyp^9=9W)}u69248C
zKa#t9_pYFzAli8~>(n89(kwIw#Nf5hnsr&fp@`EET_2x3h5FuTikwbxh9(u7*(;FE
z2_e#VCZi|56A!^u!eSObu#oIKBW(ZMO0(-663^?tr2lhTWb?6|y2!GeNI5`FA%H3h
zujSs<p+UY9+BGh3Uu_1Z4}JtCj!+H)?BF9u)8He*!f6y(JUV5vFjh^&^AnIdJq%_d
z0<H8?b@ixb#(o8gxtDm$D@0LY81B2sR`W3oXm=}5T=huFZp*aBecNYf0uLxYBkQRI
zxonFjsYN7K?OzZOM5>Y{z5D*X+i-|`kb4)Pp^ms86%4yxQ3UH+9f~=91}~9{*-HX1
zwbhA$S{TD!QL+2ti;6!HGU_gLHHn<8nUK}jeP*&4k$e(Z*|%1f_0WlLm^?T2Um?0P
zV@-iBs}^gFOWMyA_Os6}2U@HQPl=(yH&JG7IlYXbhhwh0A0^eT6kaov>|&>wy2dn(
zdx$(2XppE4S8p;MDpI@4tmL0{R5W<ylrFEywFQSp(}H;%By~)rJ?+8FqJuLnHpVjY
z<RDwe2vuN;^7qj%?$0!G+R;^#CZqzsp@tFGK2kg+=2q-Rxd@D(x(`YY<e|p#Ni9ij
zSwX#UrOHyyU^#{Mr>y3J&wU8!bVEAEou25?Jch)z!-q$D`o_g|kttf*z3ognf$dBS
z@IFSbomSAb$1jW88dy0f6v=UWJK6;3v)Jwzji}dL(Yh)M^+lXKVWjVW<VvQgp1FI!
za&uMsy*GQI+N)=>NKz~?xjfmq*!|wk{&mSSYX`yHs&LOYhH^?Mw^MLZL!5lkn!ykK
zP$ka180|fDe9u&Wps&Z4;C$-2Uj=`4ND@Ci82jKy*PWHnE-goLyB@N&$Md`%Vxwm%
zAzd#|KLHZ8loVz{H)z6^R#u+u?tbU7BnkFYzAH%bL=frzZxn%Y_bw5eqq?bNxywRB
zR>946uesflaIn;twV8>PwXu=GWNytAa`)E{`4eNOp}%vx^V1h$05O0HLkTS9l~fYg
zbu6LUHAk;0Kw;``f|(0qlw7Vg`{_Uf<ycycZ_hcGss!!cgs+w~NVgSRu#|6abN%C2
zJcn=Zz-iW9rl{)yl4^j>WMpIv;v*p;0l63zzwLWaEP*fgTE9Ox8ei1QR@D0bNQH^$
zGffFOX`l6bd+lg=ht)FltJ_&;eI)eW#;U^Zw{fy)52*9}8|uq8v&%HLGJT`f0b~z&
zNq|AQoL@VF-bW?N`Se<zR~do^&L=warX?{25}Hk8bY0`M2O358m`u*HlN8Z$?xdE4
z<fbtg#VJcSJ;XjkO?iMEZhQImc6xQ^$WD*@F)=QCU=W_L=}6UUsHuU;EiEZY{r5pz
zTN~KB0JItS3Q9h|fByzw_h<0R?>v#LPD#@Eq%C<F<Pe9#F<}3gDgoI*j0pugoUb8s
z4;{-M&x8Lc@u_wksc48>c%05?<1%Ly@so=coHmj5X1*?IY+CaSb=;GPeH3v`T|_2a
z*yOgbg1ha+M_U-uR6`P3*Vrqocy^Q~n_T`7L&9+5qYVdET2ZX8KlNj?m^ZM<BqAa<
zf3@yo63waka?=V1qKgbq|3hm2ym5-qSm*G#%FE@thyxZS-z)}GuG|d-X8gTmYX&ZN
z%3tm;W|zq!TZ=Ol91+Jo7gL^#ziCnM>S4ru2PV`vqOv{I{JvKD;o7zT$6u3;OEyV!
zV-v&gdxO(!0uYWARLGq7Il1FQZ_IjVq`7p3c`!i^;{W-`t;T$K(lngf^YlgP5_tY%
z*QfN*>gp;WA?IgjGqJO80|9eD4GazmVP|KbuCmVpPBTDfpRcf}s3?CQmfAmX*usPi
zp7$;o5A9xD*D(Kk-IWJH_C_LW>qwffKCz^B%P+0{^mTegKP}ZC4wdpzhaa3xC4S5(
zr1J@DiySme34^@G&?ue^*L9}nXL461vq82lpMQ06u6cE;YUf@DDX`^5xx9|kd&gtd
zifphJc1!Ba$yq6>y6lR8*K_}Ab2-n5*q-=0h-EbmkX#bl-Dr!J#&<8_>}!`ZHoZ36
ztc`rF@|3H%wJ^|htIC<3PNb&HuY0Mq>QdO$YImSZ+ie>|L?Qf)1W&){v@uOyH4aP{
zw2S^{vp+szPhh}#I=DM*FXeV&Z+X-BP%ibhUcV`;iXq;esDP~qaVsK^H^-V*4nX)*
z_^ztN{md@FXpZz=3#cxG+I$WG>l+sH+DuPkK277@pGquVTt8fDxO%w_ao8>X*%F#7
zez3PVc{G{IX?NjzADpiS&MFdi072flpkubFVxuEZ2#tDdK0y3M${5-$4}Vz!WiuI@
z^VJvm^%yWXGYD=nu}^G2YzLC#yU_=n1lD#Gj)FZ?M|~|fMJXZkm~wO~>SJNf0;|Ao
zb~@-%xg!-~Lbt07?2><8Z4O4Qp`%E@)Z2n9{qU99$L$y5RAg3jSd>qSZxC`bXP!#B
z-?%$11;mi9%*g%)P01S}DpmE?6lN8vYMp%Gzf|E`d+}X#*dhM2ip-_`%Gs9ByEW%4
zw-rgb7rs?V<0oluR9f7&6BIiydQ73lhuC~c!4Dk%=o^73Eu~jJD64O0vI;1|ZjYA-
zU2XvUKg#><$*Sq;>EOENfdTvJS_JT`sc}28&Lxzz>S>vD>BpW?8<F6&3qmEi<hF)C
zh>ck5GFw_ui+UUbM`fwB@weY=JGuy)@sBqV!l&$^t;w~Al^LaIlpnj8b2s!%mZ?Nx
zA)$s>(JAjn8re<Mr?G?g*Ou}!$LE88e0a0I?yI&^Fb)JGgb-_mk$xAyoGUd|e*CSI
z5wpLd_3uRUiq=;o(RTqUM1LUjIynEe!%cJ2#=9y<Ni8Ed?y{^uW~X#J?HGIQd~i?M
zu|-JUEyS-j$9k4og^4`tkT%E|+O>b4ylQgMRNNZ(Wavgjj&j-Jg;D?`Hq8ctOSfEC
z|Cq1;HSI!{%S6HTPWu3Nd!3ArgyBs73iP8#<<i5?Klf;3K>$oX#xI>K4wxx5HO78+
zyA#Z=!@3auJWl6^tJd0;z=XW|+EJ(+cq1g^|Ky*()4&zVEgK(}DUC}eyxl^Nb{|za
z4V2tJ1>j<Ui~)%we9+BjEbr+l3ITgL-YY2%4i434>-@ZI30ot*xVLJ?A%NJpmFcnJ
zL<JMXiDwO5zpH{9%O`{)hp-UGi9Eyj5HLB3xvX_wO;nGpzcQKXSrEX;&{v{39`PH|
zf1sZ~RsMdfXLxC0q}OVpUz?jOx^>~kdFl85%=H<FJwvf(Kmxg*!+gQZO3L+DzCQF3
z?ckaEustNMCTYhK%)r#2-4VzQW#>c+Q>xJrvE!^hg;nbwst;m`b)$RjCF#Bbbz&8H
zlV&ZxyBMdabIpu##tTgkF-V)$p&}P8XI0Rs=?E;l33gOd37h<|?PO@}WX=P1QLQoo
zSlm>X<1+3xCL~GhIAT=tWI9_s#a*mAE8CSZc+E<&p6UG%?-k;<wLFgf&*0jDmyKoi
zWE-9$r=pq$Su4mXsKh)epm>BZL^^EH^FX}t>C=9Ez|6VUuP2{h%E?NT%^J9ORMv`j
zR9^uPxZHU5?=`R|+aHEfyPxwXkgYTKuVwmG6Z&6F@*j5k*_h$&*<IPqKT~v62glaW
zaJ^i$c%k1nFZ!tMTky-vxm8s)3_<<vZ31ccJqo*()txj<x3;}vIv-JTbYOPPr=Rr{
zSmSb(<7Fqrh2$46W;+_o-B+eT_iCSZ>eqFw>?`N#NA>!<Inrt&EX#%o={hc5>plsM
z2XEhAQRF0}*i(NDfWIs&_Q=b)pkA|<6isr8X^QwYIHaKC`fW<);z$AgQW*w)sXqS~
z4v&yKVhtq9DJT2;fuW&5e32`*69ZCH8z?I)0}8v^rKUW41wa7?h}ZJjIHt-_fyaz6
zxZ!bf{AIH<S6`PgaijHoCzSZ$F&UOyHQup9^jI&L1M5?#Oqb4FA%C)(pBX<5jGk&I
zkXoHvI~m!$!cNIF|H%w56<e@uJZbGu09%XYm@tr~4|{h_Z%^eZno9yX3`GZ*tEU&D
zdX=Y0y;U_amE}1bLDisywAORw@hc*K-+$WNf0eNDlcjUqCsawTh_g?o%akNaUg@;5
zPct^8HFEKGn^H>SX>9}aoC2LnAiD(Z9vsNa%L~~pw&iKVrDO>*H-~EjFa7aQ8bPso
zv7Hzk6s{aE*`m|mktB3-uoiNePjkAcW?0dvH<DYlGyOZzb_8fn1GmEZzE_p*-bEoz
zYU-(x7*#2mn-==j$jzDb@GLaweO`XVcU+962(@{~q3DGF()n|7!G%6&tbZj>59a}O
z56XK7SH0HJNobn;MkR7y*l=xFIyV3OBd;5vx}`}4&)f0?VdVCRC;Q7mDni*m&4iTS
zE<HPz_AeI0yD`lbz6wJo)?cOL4!p!J|C-yqp)MT$vsqC(P*~BUF8CdX4qAYGAa{;~
zJAzH8T9rXm9~+`b`VsIbXRt_kuU_A^jqBAfSh)Sp4gYWm*iq<DM+sME&PS62Dn{b(
zyL<vYH$X(A5^`dVY=2i`+^#HkwYa+a!Nddu0<FaM_V(!L!cYoeN|sw!SC?4%GFxOa
zKw*CWbaA8WbAQw(9Tv+%Xu<=D>>=UJU?L_W+7}c!2%i4#*UPL9pN4QJBMp8+KS2HQ
zIQ(r0dt^X}g`KA)NJzc}B{&vk#1)KpUhem=yT)nC=a4;q+jTT3mrK^Rb!bB8+{rhf
zX2}ZO9_Pr{8`86qsAy;$y48!bvk(N>3d%{s0zeE3lv3#F>D$+*e*dQZU&m>G1}Ej%
zR)lp9+2F^>O%gz`e)_EpeuL@22RgC1LQLFCb$eLb^?BB_sEw=B!yOK!3tRndO>^K(
zAa^U_c#NGAMo=%{jLa*u@38PGnHIO^=jc(;#D(nQcs^>sJaP99a}UA*IR}6X6@=E`
z2yaWw{CaNGQtPre4@gdcKtauK`vYi`R8^gUN)`|<hEI3lIq*ABoTz9ICa|#2%Nb{s
zU4#01hhDslMzJH>wYZnjDm^SBTEcGo!%ZjiD}hA{2bS}btor18@<B=*R&@I>USDTq
zD?ln{Og)s7<#AU+@fK*}>XeL!lHP|yf_NOWBbHYqdXpe5a1kzzzv8cwq5D;qGq=am
z-m*<n4Z$kCbNV42QQJ3C@029ju%N^)$N0EsAO|}%xL+9i8$J|6hSaL9N@l-VPXzw7
zU2y{%lyrv%lkSO31b0#1U;dbOeQOabVR#_lk({w~RAjlAa);=yq_TVqNgs9!epiS1
z!o@!kxZzRf5{{?vvCA;jYA$;GAZ?@k<McnQcG!$63+K+~9Pu-ALk?Ui`_PaRC+=<2
zf&<P%cYE7JvM<xQofFXqsI<jrhVRrx9zs&bC3NJ{)jCx5HzhIM7Z=C&SRf_p9?@C)
z<nmdJqZP*|qCxNUq>TT5e1w{Y{o(K20#iW|R!7NL5S8OidZ0j3Z<&pZDLDaL0NcX>
zMmk<0r29-Uge8)*Ee-BKKT`9!i=6pPo}6cdX+giJlKL)D^<)T^)XS;s99D49g_Fd!
z1C)Mqr4gWFubMP3-*M8FU%cbvzuh2AoV0FWV{DY2lBrN=J)23aE}0b_MmT8x#<!f%
z6(Lmdeg)?)MhwO;)>FpxOz)aYnbY@XFc%|^E<IcH=lyY(e`AZ6$Fh1Asq|w*QEN9i
z4m!@~=1SV#pBrAHx@wP+S}s4g+!5(iX35z!%{<_2pnz5fM+WAz??yE@C2FEezTvZN
z5IOzeQ<G8Z%kC?I2-oyvV@T+H2sKLFY;*FB-WJPM82vDW&+~|5)E}n<8JA>O+S%_x
zoxZXg5Ao8t>v4B$0Luz`Tb@siq81?j-RKHAd%2BVsB&Bj+Vk;bxilw+m!h5nocfAv
zF7$rk&z{<)2sFEMaN>OJoV!L5(Q{oO179Gu-s*YG8n!L`9h5MA0zF#);Cfty6rG&t
zk+8F28~JIuF_!3&yX<mQGy%IGu9p}2jJxt8rOKc6J63S%3xE2WQJi$Rm8xevjV<5C
z#MIrzk#A-n?7-6<?9hV0&6QqASXIa)fowtVW!JsO<D?76XuQ$xofDJkfWsb>F1$}I
zw@a+43($}k+A%FCwJhl6SV@!5CEI?yJSE#jsGF+t-oa+$3BB9bR(yEdfZ_4i?uSO5
zIpiV2G9=}p>sC`kZxUE%qyKsm9d1}$>UwqR^F_L#&Mjr<fpjKSesRI;w@BC9c&^(;
z;_dKbZG-Z0Fz-MKYLYOJbjVapXnFtauTK10Cr86aRbh#(byqD^fjYCOP!@=(-yF@&
zvF4nqz(<`pA#o~K2UZ(U>zzcDTqg%v2?;ShW<U%jaS;zWrp9dkDFL5S=OlLI^39M7
z>4k1-RV^^Z1Y*|8*U_yHK~KwcF;>ql%J##RFus|NYPvYhyQ-D0=sy0vN;8rua18pj
z)UOZJb%x^}n7^Uudx;Fp4ZaSy!IaL36;@)G>oO>@7=?Loh1H+L*E2LVzwRArAFK=V
z%oTFpNu@3u+yUunThx5&zAGxWy)F+qefQI8r;3*N{yk6j<zY-R@X9N6ncBIa%A*DE
zQSs7G<%jszqC$5u2BT)qlo6@1`Tk%2Ree4De%K$8gGqz!fh)zK^M$k?PpU*I3*_NG
zg5P?ofd8CyUkR!w$P*TWyu6?iuZQK_ge1G!FKJhLH$G&8_8h&AA3{Lig9zBEH4(-$
z%_e89><Bq4YpPi-%h<>piSS9Y#2KhviTERq!>pBim@hpqV2RJRI~IyC*BxFkqjQD+
zC|+{v*c7wV1HU73euRCGC#QJO1UkaWzhLvcES~WZ#nfC%M(d^MqCAB|Gps}Vbwdqx
z9eB?3F#oHWyZ5X=WtIefUP^sjl-r@jBkj5b(sc!j{??9H$*5s!UDc;Fg1J9f*AXpZ
zE7a~>V}%&WqCTP+M0XF%xi%+i59a=goSRI#8G%#{$KOexzs!y6qKs!4<N{f*V0uJO
zZ9V2oSGn4%g7JM-k%WPRI$R^Nn)JfoE|(Lhrlw!PKGj}`mhZjuNV|+SQ6f}SpY$xF
zumlI?{i}wgROZTtgtWkt*CGk_8=vOK5k_PQ>enY#7uF#f7aBePRE)Sk;jlRpvpKeV
zm*k9}TQORLj)1bf$4DqwTfjLgoU+@yFoGifGPu}zDN$63*Le5zcU>#~cHL?vJvfeZ
zv#jw?oKhMtSUmOabk^k_U6<Pry@_fkq2_D7*#fQ&1|PL09bNSTRuwJ)4;^ak*V?%p
zwrRgQmgSOi!C;?>Q~T>45p``&yc?ixf^#wj=tP8D3!Q$Q-+&zGVmCa>3Otgn8ib2t
zyS}f%dcCYaQN6y%Um{r@Oj^pKUs(3?+MXWx3BqeDm(#Ovd3$%V?4cq<_hfPWiIbKz
zJeEz9x$7DZXtsB3+HVIhi4a}bQ~jQeQ_g=c*cUu*-}+MXu?Wp&a!kxZ=gEAl6sJ=H
z0*eewgT2Kdh*}d?O?!akEz;D4m<kK%0Y68Uh6q>zWCAoc1MLge8BflJ^hBwQ?dur@
z4=db~D~k{Au6lPM*7#>yZ|Wl7ZL-BEP~L^ko-D2GNXt5alp|07nFja?1OAqLecMyh
z2o4v(>7n87A4=ryQmb_ZK9MF*>`*22p66yz_p}bd-4dDsN*G?><~v;RluZ>J-abE@
zH@7J6@_Ne~I%)SUMaG`H>17oJ$uVyvo5YIr8hCEL62#Xr%I}I;2cL;MRS#mf8JCFF
z2T>Xuw==2fc|cr*?IHJ;7qu{#MSwRWCuiKLo#ya`F1|bjalNF}#FF1THkcM^OPjUv
zVT4_4MmyaLgOI*y(XuSo)dvkab9+yrqWK``q`)R?wY3fur_}xW-@xr_<<K)vl^x{G
zR8HQ{Q2Cx1>{SCZS~8=D&;P+9QH-vKvRSIsR9XXC8cRdTi%mR&>c7Y2#2E6mm>%mb
zm$;6R%Ay-bzj>ZZ-ZoNYGSp1$C^a4-2A6=J`JkgfB6&wY@BZ=@gBr_MzkP9z7<?4y
z3@7at+pj(TR<3^T$b+ZMF6bbpP4gJjiErSwIxA~okjzuj<6UA%i`?Sg_P_Zw7SQEO
z2?-GPxiT{QCoo6kdKl*#9-gfB7u`$sO7{u-*)IW53nU+%6bhFmuP2j^bySIjt5hxh
zEOtZ@^*4q<UvmEwlBWa+uyc~K5PJBPe2dH`UgOrWZ~7jq-T!5AIFqwt_O;85tR4D`
zDD}a`*ch|rSWzW~I?AWzd?$>5x?5-%@*hD-OV2WJPP3h*r_x>rNuO-d-KQrrb1HrA
zg(syR;OQ~N4?Kf1;mvnpwZn?=dfj4UTGXG8FDH6bwI{DHBLYI!cesT2&Ta?OgLYD6
z$!De3cyM+JFu0;FyYJh=Xt_+g2+Fb0(E%{`^x~p7j>|u=Mkg)AXR;m-x)z0Rz;SHS
z@@9m`hcP{ZYyHWijq=mJpCBeDu2CjLH?T3UHQ##`kXzLZx^7(X$HOYGpWaGcoq6wm
zqUtHvV660-EygBQ6W>4JWXb*Utvgy|<#J$tc<HFem~+J~@?ZrlO?l>xxFK9+wV-sI
z)AU5nfC53hyBH3G(KFQkZ!@aBbzW;LXT1iAuOwDcCAIpK^WTM7Q%Y@h#IXszIVM9z
zL!^i7Qt?+cao^Tq6JN$^dQi)LgH8MY5cSnjReWE(L!*G8bR*I!-60?#64KJ$B`H0K
zAR;Xy-6bvEp)?XocXxNkJN&-yt>@woVcm5v%*;7w@BJy9-x0~kc|6!p0Bx=cu;+mM
z7nozJs;bh{=K>lWu%3Gz0H(r4diA{g{Bhl;b)b;GxVVoIQgoPejOAsR2NUEVEYj~^
zzam0<^-Zdq_x_k|ZAn7!ch~=A(kmwYtf;Wficpf5lx*!(To{tHmx5p6NXeXUKSp-E
z=|eov(eohMS}>)ZDHe}9Fl$*k=q41V0PbLRB3}O-WHw3oH<@~1wq#@pUr8hbVL;^p
zp{qFDrx)oE0z^h(KrP-9Yl{H!Cf*3rvT3@Uj8}HnE6w4f2SVbRD&$3>tf4MN^2eT&
zIst7osr7`cYt4_~)PcMGyQa}gCOH?TAbR)jdKw0ERH8@XR5XiRuMt}Mq?@zKZWSf$
z)vT5Ed>YS>k^SBEE|`a`zSo6fe{VR}{49Kq_*98XEoW$!?;xYrU6?}Fr}`l>;vq^<
z>r*7{{JtaG%G?bzdW!5urb3#rIEJCTn|hMI(@MTU>w{tfn3bpHpP^sovMx44#MEDe
zl-#JhHd|hu^r$|efAk8+ti7=(&8VB)+Gj_^J`xE>_2A+@xuuEi1ilCsy>Tym`G8Fu
zfE$2DTEE_HCpXo1vc@r!frXvjVfY7cetZA_;}R$X*Gym`UqT69OhDiU<|@D^XmcH|
zSgMi(5;qXtpw?((W22*`6>NZoi3tLI{&aUgz{aK&a<i3}mtQ+K2EDO_5t>KVaqv_0
zjHaer6AOd~c^<SYBhdVkwb*3m%X@Cs=uUI*>b`{WSLXcunDwu7jNtxVA}t&k+<5}=
zKzqp(<Wf(AcWzIkCDIEM`Q((~k56y(Reg%UdGGt6pmYFRpmSEu;5DAOC~j1I@H)K^
zxQZnmjXg7thrG~u+$Jm8)>LkN*!gDzOIyyan-a;@qlG&ka)rt#<yx+``f~btfhh?;
zZp?g9YRi+O^FF25EhYy{YxC}DnU^+>Bg|YB)@k!%Lv;yOP6D>YZ>!;fBPIz0f|c@3
zFzm(>{*Y2v7Dd-O(B3=LFaE+h-&rIG=HcZN4bjq_cr(<(HEbNW`6GP8Zem8)$-wad
zt_-jpU0F}6Sj-pXoqH2vWRb;DnOYQlI>p*^fk^#siyS&Pj~n`#VxDpOU}~?l*1vK|
z&r5rQqTy~+*~w8PND0nmBC_Jx4&}-9e9tlZ-BH$ETUq^`TK`eQVFGGO%BQm2<X^~2
zHWUW@UJeY<b$wcDUijwh6uU#mTGWB7?QELsFX*|2-wiUh%|D~(fxY=hN@a`Xab_d7
z&|h^;r_y2a4SOTT)9zUMfl~s!6xUs=aCeME`kJdkD(dXz+!lCurdM0IHYfVA$gN~D
zene+0Nl7_1I^vy=kVyKQ^VzvdOYTXTl3Kxj4kuQHjEV{uAVC+VxF_5ID^M;Dj)kKO
zkkWm7&`Al<_T{ZDA<Ms1oK9e%51uzb%La=cOstfE@eU>?V5aI-*-n7?n8|CT#{lY{
zYldMltwu29_gv}-<DL{EF9w4y;H1!ed$|Q%>z0<5fRAYNy9Eh>z46DEHcLigtR2Rs
z=>Bisys;dkSLwt1P9)b9N6O!F)c0Wj2}N7Z>=*aZmIXSq@8@2xQdWuq-Ka4)(-K)9
z7~EWM&aLf<eNjU3dx?t{K`}0~x?{^43Y!?LEXSz*_+j<@8x8#|P?JtCI)QAG=L9G6
zDPo<5&$GA|nlDB*G&Si<L_gP~0i;X&FMG&I1&zhb(8d=Cs065H)Q?6+M-SZ_u<vI6
zjs4>0NJN1=V!SnjsxNZe2oU`^lUG-4Ri|2J+;)gxv=l6Q4(vMcUBF+}UpB47ONMlp
z<0uk~v0PEri)HP5UkCF4Sd6rra?aEd?Iwst((x``dOSM1YBd@;arkqmYi`F@&dXV8
zT9`X?*KtCy7j&%&lDBG!bl#osItSlU(h}|^iT#M(+3LO)zTH9#d}5$%O8=#Y$!4If
zv4<TdXTe!5-8_8j8u2yceJ?2kycyuK`LCe4xHB481WNmnaciu0+Pb_Lvt$6j0M0-g
zo7N1>f8t7&d@F(VDX`;M4=tPFGtd}yl+wiL*#ch<w5#ySoXIE^{RVc`OOL8*0GB-(
zsh1@Wx|@oL&5gw#QB%K+WOOAHrYOnJmjh2iY=f(uE?ae79SKAPV}AP5BjfJ&NfaUo
z$<E05ZCG(-N`0*SJYF73Q^`w}r9I4J^A~eje<LaR5m*?$d4r=hHE3A@NUng$YYU7i
zfGF7%fs&Y*801bB75N-2zC4WO223arPX<Eo&`?n9fI95=?~G*5Z)X#|VB176Z2SBe
zJYfY=CyE)cN*<u&lY$`|h>`JB@1J@~Blbbuo%xU$TDQ(W*<h=s);6s<+lso_ioji|
z)qCsRvm0HFkk(w#`jt_V6k}_{@Bsvr0bJG3vP<;6aa!~9Q6Rz~jawE5BXuNjvRApB
zDB`r0b9ZMJi9XG!+<LXg_K}{jrG9*UDDUB<HQeJyH%V1lG&4oMUDuF2;w`w15Bi(+
zAFO1nhg-eA&HdZemQqR>J|GYE4}-wzd>}(j!>7i>@yt!n;wQ<1{X_OQY9(_PCkEWp
z?c%B^kpHuxUHYE$>Dk}`QSY<bm=7HhOP-XMqi|=#hmW3bAwO~1Yi~i7@do<QyIDDJ
zIL5#}z2vrad=@SC+~Lj%ns{J9HSNwMLFGO5YCPn6Msq1ZLINPQ9e(zS=I-qKd;i+s
z7J%R2w9Vu|6egP5I{2RFd{c+ym%w=g{u@y(pUjaqwWVYwJu$Gh#Jt6?h9W2KkQjq|
zSs1_kc-`oAE+cwi+Z-P*ZHi|F_W+DN|F^!r^VJXg`IKPG<9*rt>Y0ZS00b3pY;9Q}
z`BYK8jH=}4Mu-roBqHI)cMFJ6D4BqY_vdy%2%MTqr20HRZzv%8e(&iXIxvgngn;!?
z)5g!1mKN}HPNMXPEk|SPm#*Xx=7~!VuV{1ESR=D!@NCP1NT{#Y7~G<4R$=~ztt6n*
z`@U^`&NVzy<*kjI*cx%*D)f$nraaAMuJ%e~n{}G3NGkU0;E`K$0~SPgIT;<~SwW2j
z{p9}Vl@ym?prWs#%qr+j0o4fKj+%^tSsX$O8e8_{Lk|++AbK1-88qmGY&Jxf4m$QN
z=kYP{S{nBMl9dFPfwp6%j|{DBRWn1iKB4($!HWaDZSV1Lg!fiShr1U;ixm6SgpdOm
zJ{b?bZaQ|RBYmci4@L*j{#NhptlnG-%le)<&q~CWErl2j<NKc>%O<hYWeF*~sIcz>
z^Q8W4>o51Eyw5t06K3is7ri8hMlXgQzO7^%y_{No1mn0aZde{3ZFD7;YV~e@D*Vc%
zW5NoOc+byHxelLAmK#Q{Q#CS*h8Khfi6}5W*;D->fzXqviphX?(2gka)bZGe1p2{+
zSKWKy=u>V5gIn!4tLH2HL|gff=)0>4LqZIo#Y%`@LyJ_^i0<?p2o8>j4k0)f;32<P
zEu_Qd=QMEXGJxaF&7TO>Blv;x=JO7{#OTF6sow~Y`Z2Ju!a_o9nlV?%Ka|`XE(5+!
zPU|qpKS=`x_{Q@9QRv_8(X-1FTk6nssWV4sWav@E9o6Ma+^DdH0~7-XQ4PJRH1YIn
zej8JDJ4t(v*Oy8PBBam{wLnUXFc|U!UAdOnP5JH~#;>Owv6LP5)0DzKC!*$NSdh1v
zbIq~!wK>WBY>d3sX&qc5`bDXm?Tcs1W0xDpmO;DDlAr<xW(0AqmnS=2S(BGawkRy%
zhC|EW)ip<&Qzpq%wd6!2){k=h_DmcwA;hPT6ZIvkhMnED0Hii=o6QdA3wR)=i!*Tl
zRU_WBqSps}Dal~z6g5o-4dfhxZ@^dg!{UY^3x(~EY(W#lF%a*WBlwlM6~Ry9!yb~?
zE8z52HnvrJbFs(F`RXBF54MTdYZuO(lASHZIG*K@$gJtm3_0lBQ%dO`yTg1r*>m8N
zHFm_e`s>JF*PIpVg-OOoPG+8+C9$}yVY{OG{XU$uQ6?_>bA-f@?|z4h6^QRjN%cR#
z)A^ZT=yy@+8UWAh>(BZjAZ1M<w2YjbTnzoEO0Z{p4DKN+1V|%A7lk|#4bS$0SZrcq
zf`fxY0sD7!G!`b7&@`jMx^wHo!JdfEBgj*Gvk|V-;v*OC{Szomb>sQ=Hv?t$_t}E{
zDpA^c5d4K)>SFW8#K@+JOoTa248q)8sxY68W0fgBIQQpzpORF2B!!n*oN;}W@!;8I
z_qanzy!Q-wF4mc#2NyMvK?RZ1zudulXr>_!vlG?_HdCPbIo$E@t{A^JZS~ifThF9Y
z0{UBY_mo!ZpkLj2huZmaRUIW`!L@HUK!Fcd6H{ePb2Ks3&}$YIJ<|S1Sx|jpFrlTx
zUQLVn!bMUu83Ei-#|x{=w*&7_RdK=wUvJsx(Ti-`p6Arrucoliu()IkJ&sw=5KYO#
z1LAem*3NEu!SSJA=Z;QB&h}^2C1emx^reDBMdkoOTzrr1Zq`#neZb?p9^HY5rFjZv
z%ht!y2Ywx?+T0upfNlcxbuV%HZUR?T-vNAzRi_GS!4|s~uToQ2NANqI)i=ecU6U08
zbrr}}+E14Z3a1dMb3ZWNpvTGj{rmSbnQ+Ljq@)BOMY4+6CSSKm_@U?1aDAs2tV%gv
zP*6sl+_w_go~r>U=szm2il3KDpB}k73;VJPQ*#u3R)2tOl$N=%Vq9%FEspEs!N;Q2
zY*<)QZ}PWtzt>;I+^i_&dqPn8PH)sr-2})KF7&$vQki%B!WRz53&F{#2GRuk%ZTUt
zRni%-si4BI%8(b!1;2CQypmkR;L+at30aCQS_zz%d@BZ)l0L?wa5>gTuX>`<iI$BW
zw||A8vfV|5{L{?j^sY~=cja*59!eidNSHHCr#JcJ{ra^`{Lpqb+Hm(%``12L@gH|a
zKWX=|%mXYRYEhF3OlMdUdjvjn@(tJ10yWP|ajv6lsMkduAt4_ILPiNCr)c#1bEk5+
zDYZ#OIjZs2WoXVKGkS1xUmugUolD5qcAF`gZmho$SegT&;UvJ^#Hr3N)>6?Pk;zs?
z2USH5^|o7CT8ay$_Ru&WG@EvR1@`6ygoMC)@&9IGJwX7jjX-SS3Nqk8Itqv-SZN@a
zaDordTofP7PWBppDl=Y;pQ9l$NSPaa+d!R6H(Q(X-ScRV*4V1>Z0K5A#;?6S^)kEk
zH4w{lI*7D1G7H4&PbRR-lw3Hua7Axh&a!<RFh_)H%ShS?TB0u}NnbWG^_q<0eIH9$
z{}C0kjBlkzk0BELsQh&SyL1+vY0S!Zb!XV!pU&Q{IJe0xeBe8VGD_j(^7Y~JwG))4
ziVt`Y*zb7H-j}Pu?<qvP-xOd~`HiZ_mYOthmu2u;E{FFG(4?xPL7x)BL_0V~l@+@D
z5%Un6;yrw!j>-H}I_<o`7=^ZT{&;T3drViNuk_ivoZhnn&T@>`I>IH5ugEAI;FECa
zwYuH67iI-s_a>g;x*Q!I4i67kIn06Yz5mey({6AkI0@A9J8d^grXuYvE|ysh<@`sr
z!zSiz14t;CN=#4dzIw$-Pk(Qv4)koM%cnb2db-+T2moLwk!1TA@rv0uzViwC1E+Ix
z$aA8x+%`xT6&r$;^V&a9R1G-4h5pW{C>d8=H~(Gs;d|1BJ7%ed;-zZmEaD0tDGTPC
zpM)2L&uCML&#JftPfWy3os7o+_|r26_HLr4h!Z6$KS@+}I-fmCrSozZlp77`Gw_%V
zFn9aq;arz4KWuJsID3I$s;S$wd=&nOht?+wKHG~BwYP;+VSqk8NC``k|8RdEn#=Vu
zF_c`2PfepE`)3VrcN9M<O6VRi<s@X+_uicrfa18V1}WidBP%OL3PW34T@@7-eSNlr
zjxS#x9$GWoPs)R49{OoL!UW7Mf&KUD+FCQfKie_7cbb407{1)deHlW~4YLQ}d=1t`
zor7CeatYNVQ5y1!6T0h3yR=BHNi6ffX^hBZ6Vy(Td=;7vrldlHN{MM<y%02rRI)Fy
zRqs-ei1|xjgXdRjFM35VX3VXRS9;F{0ftA1#rFWw0c=?7jq;z{jpoi=81wJYxn=HQ
z!bW~`p+$^(cS?`?plm(2D*4D@{-?ow_)`aBm%!;Nxz|HaN}<+0nX_p})+lxyIbF_S
zl?mQXomE7H|1HgN`g&whdiu_(srt0EmhNsVFu?%vXM-Go8VbF6%TW6$YPxc!Eejy8
z-SrqER-gj~=Mv~qNqKGG#<J;kNgjc0Ycx!vpTH3iICKF2*h+x;!V<DM)w#bcxVjPb
zjc<j6$pDRk{Z%g$tqu*ZS~GDp3gnXTd$$Dj>^-Nd_xqV|;tX0R_?pAuOMUDvwec$4
zX48`tx;1?PyuMFoNwqlfem(f2v-M+gRC#C02#75zn|CM8Y;^1@i;6O}i$>L}7p26d
zA{)=>#Jgy&==NYrLKDoVegVqDa}jWWGq9ffjpDH@u;yX9NkGqhIn7W~G@$6~(9I96
zf9o1yM*I{Qf`tTGYhMWFCLc7S`Yc#lJ%q5xse_9deNb}RlmZRxX>|DeozGsYE=eXS
z>}le<RB-EwjNA>3J(;%iDwz$scOR3b#Pqs3lVL#~fY*+2adAmWNym!y5@TYrxjfPQ
zK%NIw1LD{<gl3-*<dlq#K*YBB2y=_!drq1Rc%1Q}(p)WcO2D12>5aojv1H|menLrX
zglm4WtMd2EfyNX6zHvfnk-;X@6qC?g2O)wVQuM6PT9q8iOIO>Xj#xK6s|A<vGw1W$
z+B|JID8suRmFim;<qZbuPsrlW>oe=gi^Xpl(nV6oMZ(I~$|k7;t-`VD%-<Om=kh$Q
z<<ds~ImfG$r2{Rx%WH<`h!=v|-gVw&(7qJsGtS3-b<e=x_xZyD?ey(O@C>MGYXj%&
zB*4m)RH_DBgTqwS4sd}7w&3KaRn^sVaE)UtP`2ddrjRr|CNW3V$B367{W6P!_zPD_
z_@IN-0kdNTcx?x(WG`a0-~BU4yU8d<HJEKffT%y<EEHwhn|+HAN!cdfBUrXB;^_=8
zfgE;m7L^c&mmjWR=azrjoPg2<Vs^H`47~86hE?V%9fRvjmm&*fNn6Elg!+GB-?fV*
zjWCoW6@k;fgZj9#PA^HD+85qGxl-DF?oKiq5_tP0+fTOV*j5nyM21`z+#OoO;WzO^
zp>O_+F#BjYPCrZOh<bZA2c!96k@6afwA3r42zy+gAF#5rf-X4b&k98KB=|dv7Ilmq
zu<@`rO4kz~^ZT@p9XmV>4+Fc>f0h-1ziMPz{Fy3(D@M+hPT*?4eX4fFl_-ZhBeL~o
zCMBL+$izD$rdtNVn^gFAaxB=YxWGMFpl=-1^J`y%>`CVl3?=-A4l0}ka=cH*3I^yx
z4=SqXL$YF68!{QQv33w~P0HTpTdswatc}TBjVG!wVsv`Y6LP~{94pR3Yy)A3m`;_(
zg9?-G8Oy&vP0=bqIIBimYy3$u7wc+0Iqk}nWb#FJD`M44hYbu$lrrW%^;C?xRSm(o
z>&Ag%F;vkM!UMFAgF)09Q}ODk$7JGA9v-PPNwDl&t_kpn3wTFu#~2j%2Ona(l>?bt
zpzDqWXa4@QE4*b9NQ_nZ7UO@mu9?3w+xlCT*kVEG6C{2%Y0dB8uy=S^?A-wl`Z!?F
zcJGI+TZ4m#2ONW0qhvumNtON1d;EM~4sj^LXdIk54%0Qt>gpq^L4$)T&~x5VJxrVz
zI7nZ6@JryMwUb_B*AV&QnrH(L`evuuto;($Ku3`pt7kCLD0dbWTcj|x^W~Mm52W*V
zN;_JY4{pOl=+DWNm`kiyU>6338AYm=vS43j(Nog1X=J}Xc=pR->D=vg1rgjia_|Dy
zXI3-pQQg4L9c1PnSY0!{7)c05?G_Njd5~_;MC*Mpkj$H6IS_;BdUfrnkI=7&ARGf9
z&Nm~8<rdRFx)X{+gtjb#ijqjsp(StIRHgkn=SKP2oseu*#>DSZ^Z}yZ((vyM93wlU
zt(RHS_XZd?H#di!w;K;nPy4&OFV-_7z{QCKsT%9+HJ1M}Kymtadiq+qW5NOnYd%J9
z1!T^H7yOOw2dsF~`TKP>0(Ul(jfKwgB){sfGD@)1=6e4*h?^TxujmIJXNocFV0|a#
zn^e+zTM+11Qa4sv<FkaT%yFxDm-TJ5;!r5S_;uZ@tm>I)N8fYn3p3oCehZD)D{&%2
zczbt?y?@Vf#w7Vd3f>A6$k4PtN28mWND!qTbT&?NE*Tm7)N%Zd?>#J3J{NuHTripC
z)H~B}dw@JwtDJIGOn?Y|fLw6D)H#PXFTE~rRh(tBIBH+b>5#wtXi;)Rt(*JFgQ8i_
zvAQ8UhFK-ArDK|=r_*<3Sp0<ae-BcZiNmt|uQL-Oa75jB?#_b?0u+(A*jD1`NZ#B|
zhs)jI6h`o?uZORI9uN#NK%zPt1Qsg~C#MPnUVMB!pm+i98L0EYn4T^)Qp`N+X{UL)
zIum}BET9CE?Usg`uecU^IC$r-ttH4)>s_qqq1m+V^*#oQ_u8axvi_AtS=de}=0eFj
zJ{e_a_zL6X3=bVyB@Yi`VPV-<dDg}7EcTBFEKCvIO%wH3^s}-oUG%|)H<b3zlA&+k
zvfaGgiaQ~=`zJozOH_3)FCi#WHtVi!PJ02vz}z|L$D;;c%`=~krCsjn%;#AjZXbWL
zSUWm70W{9y_;_N71=t7x-w1G{6FE#!L9?B&Y2b6RmJ0Xx_TH1(T4X}0K2yBffG;jn
zdl;W@PtIiIT5PZde+U)r4#aBK??v*n+oGG&66StYtn$Lz5f?fa$n3NRdpo*6QGZvw
zAEhg59e!0t1|pQ=wJRVDj;*#fDM(g9p<Rpy2Yt@>swg*it$NN^=gphHLmoKD9G@q@
zpXy%?b61^Aq7n0Wlf|pEPUjQAyX}P}MUt9MHi?1(&~w&tS0d?8ROp-)0Z%U46oNXN
zUEf{npvC(5Xu}o_Qy**sCAq4G7p1Sigl#|OunML^$|B+Tbd~(P;REVEaHIlI$Wo^L
z19WuAudIyI6(dH@=jLSWK=(ZoupXwmG>R)56Mbm5=M$;i`T+=Zx`<)OSE*+%TJ6nH
z&zhSY#@2W}`<jn8KNcjs(~JB2&7lDD3+ZN34_Bv`nx6TS8on>NEZjRkQf1EkB#b0Y
z{sW=b1<Vc)SCp5lmbbUIB0<9&Cyqd{0lfoLbOilJFK}*>Y`x|PR%}B?UyJFBYEv8?
zvTMEu^G<{QDIWo)BRDb||E7Q^TDCQr9%QPRqU-h-|GR$9SU)<oAoeqs=mn0{khsPC
z&CLx$T3MOO%;NgGY_<xpFEQhcuX{Y8Gdevj-0}wiLUZ3fLpS8WHh=hs{v}mrC}T|7
z+DRLBY^T)a_!O4{BTWYFN0JqjyHVVzAl`n*Qgh5R2-sP}sN^yX-G!D2eh0qcn7Gd%
zSx^2H8JxCAwfv|<hhgG%%!cM>7=&B_=fzW?@aG}|ahgwpeSLiajSs{s0!G}gdVB&f
zWiKk)f!Kkqr0>G*E^h0My;<z(u+eVYJ0<$#==1TXPteP$h0R~O&r{~By{m3mvG@3+
z*~w`McbLK!e;7@4*Xg8Ng=E3z_RE<zS0hgip=H{V{Wa=|x4F7%(9a*?<m=Wv(~^>r
z%9hL9Vw8_RKL7EdzXi}9@gNYc0QBbXx5q#5b+or5w0P|`j!jNJgjfi+LfC;ZILJG@
zzY~%Ac0MM7sr4WGCBbi)+Eiirx$|Qv%ss*~yGhB`&~!EVi}WR7S$;%rjDvuu#RB17
zid+fNZaP+R{e@Yvsz$w{mdI8uw&&(i2VwJqkLaZD&s@<5HoQwdWg*O7n_S0u4_Qoa
zJ;xXekSv<1V!BV>n^!u|<P<A6<NGU?ZhafkA4(PSy*Q=9!fp-#-@?gFq7C%$3ksd{
zOAlNVC*8E6ipb*LWXmTMFCy=fz+XpCLv!y2v%k;^(op8SkNOHqnjS*zjmLxR&DUR3
zQ}=dvcQ-bGT8J@97MhRI)aPi4e2n}r#LC?BcuNx&@!jdlcHvd}%r~hOFUBS0w&VZG
zRBhqnKx3=}wpznak<A~Og&Ct@A7M!r)xx-CF%~3JM5EgjHIrw-Y8omdRCuq}xvkGc
z8MX9?bEEe**%GPO@id2d5TGz?vh6KV%a1#Q!;}Apd+447n;<nFpV36WDyNqr7j%|n
z_^R6+&F5?!kkTKkQnnu)Rjb-b>{Nr!sZ#k)5W{*3GVCGJ%zx~W1gCyvj2UE%GUzVy
zvGYYvPqh*+W7c{lP|3rhZ^vvv6FrqMlvGG`l+j2R;rP_;H6wWKU9^g4ZeeYyUJKzL
zvbY@kD)T>dmqA_=Co9^<@rx&O#jEPoj~@g?q<m2z+bXu8(Qq-LXYTk-WRU#qMdZ_L
z6@TjhQw0h;=Y4Q0!U%?-kpZqc0K8S1Rk9-oUNoRJc|Yf)$wIIcvW8#?k}B`1tLOK}
zCH~0g!^<yiW8Vh=SMo@xebptQKt!ARGX!}%U)!Ip>)DN6Bn5v^?9X)36b_UKM?X|-
z)C^#ZsijpfH9@Kr{5Hm2RaSex)pNNjtx+GHT&V%4CN0W&<J+tK+`|!mB;+J|apgnb
zw_f32Uhlcn@cilO-*}*zTD>H{LAVZFmcMSF&inU?h|WLM<<$hwxv{=5loR>N@k4*?
zu6p%>RP9KSQHsxu)+a{w)SF~CjsorS(yADPsU9U}d$KLF>YG4!&W&G}qt^*l)webM
zgd(yqhI-$kdbvB_;TEC7zK?|?yO<v*%IPU3-(mg~I`a7SvU}gU-kK~)))!Q_wdAtz
zyZ6_SSMaDg0uvQ|o)NDH5dOeKh9*YzJo6n;=|@FH0pocGOUwP!(-ba?evm+p>dT7!
z;xb&Esz$&vh_Y*`uvj5dy?b!zML6RCJX_@HsIeZ2W%ao|04-~l8PTuT|8N)Kx~Mnk
zp(wotL2(?HR)La+(U|nd1EXj5dB1CL#L!6ptX72=SP2*h_Xgw(<srn*dC*tmK@!)(
z(`0rCeu|m1LbZBqvCR05g^jT3IfyD$<RYGpri!5VzG*;7CRv%PY`4|^v6bxwj{pM|
z)uA)!c5m5O*xaomYk+cZ$G8uq++gkaJHKbP(YZKd(gGEUFyAxez3Hcs4fX8t?7(1F
z);Xg=V%qM9bk?NC?ZI%FM#{RBwpKu_p7(&d@JgK1&}^dnVq-Po%YiUoOXBeUsBa^s
z#rWJKl@BVU{i#|FmuVtNi8JBnR7lAC)?iYbK-tlG|BrP48l19S;`7Y1URy(=$j^w`
zIL<P6z2sI;?f)N{AW)F^daj2F*>4{z>CkXvPE7Q!PV<6b2V}H;WUz!Q9GfX>slLjV
zWCI(X3@@iWIqfRMR&kTdWO1suTzxA}0ui004W(LKeXz{$SCO4>utGFJPvy5^WWSjo
z&v6@0*I`<!=EzgMTNJ45qx;TRxDpdeDkAopi);TYdtJ*t>M&B}e6r(Tcu3$no}@9|
zMerG6K4c`?7%6Eu8Mo7u<!&`pDyQbV!I&x(!KhA2R^wWq^#GdHC91XbcQBBCvT1s0
znskRz4;Nd7fldD(GHVeiZ9R1NxxYJ}2233xUMWL!U_PW3h_5~2h)?83UyTgW0VCf~
zOhmtLg%aMC4GD7DJ#58@5N^KmkJQe;Vf6z9Z4sRYNwsddxuS#)W((}AC-6;!N}lx^
zrbZ-i3PYj$Dk~$qXly$CCKgV8yjyI~O$|pcA|@mFc{MlamMq6TQDw60vkEcsA;yl~
zdCoz%1af+8)9XqrN|7L0DT&b3OmhpboDD8jBnxSGeyu@Qec3z}A8&>eX<rejL73iq
zLTaZo@P_<>KjufH5{jUnu8rI`3?Tf;1<3V5G^Is~hxCl-`(0g<u<vnkHRgRJ@DsUY
zV8hQ@*lX2D_@z9J_-t~m`E?^!3Ok<qFz-c;SRC#j#^NVmXdoOy;^FBY0bj-#`t@Gs
zqDL6}YG--*4hs(i5vZI(+2gPO!q&#Y+bzwS>@*e{Bh|8IPK0QQr62USfZ@Hhu7tuk
z<E$6;hK9M3ohq{?l}lCifm5jgTyNTO{ud3-ahR>To8PcGw`qE_ENI39HJn{Xvy1S^
z3=r~njLrOMPcLh_m`rX^ENvT>5>if;rB>)?f&9SRZ<60~*8cXo6olOGE@%f>6N`4Y
ziJLV(7+p0@KDYB;Um5nA+Y?gPp4o3Ej4jdQl$@cj&bqA5B;-7ltG$L5!D4Q!Dcwp<
zVUFp}jBP-G#<0CtRu<Ddg4y{BJv?vKPIA$DJ=0ZLE`0uMIX-@ox+8qOH{xQH8z=F!
zUSRUmrB_3f^@sYQk5UFA$;h!OI!>3H$vt&BU+T=G-da1fx;R((IGtvJ&!C}BD0VMg
z$Xg|6-T1Z3o3L|`UJRe7GYeGyYpN@#w#<n9eCPYugqAS;O~#0^8JOk9skkyV>e@Wi
z8<?(9K`|gx=tSdXZF60zNc#|h{NfV@g$VW;gET!@!N4R1Y<3_m6vYp)c>5$-2$;FJ
zj)3XK1Te4u^$~4m+H-pCBPgCRt2pd=eb%{pj_8EPQ4Qd?>BKE_JuJ}uV=R@%H>WN|
zg`>u#?qvSK9D+;MZ!a(Y2IXfEXpWZ5T>Cr=DY9oQvS%o=Kfm%mEEwpWs<AIMUn~6m
z#bR!%W95CmPLB=A9~%<(*&a{x+X{BxLW+698Y-jfsz!QRm5G=NvXg*@$*@rGU^P_e
zDCgwa9ekb6#Wy=uRdUsKzMDzde7=J2CKhO?@N8@B8DY4A;+5x8`3*_A6qC;7B;kb_
z&v}qjZAL~_UPjf%T)_rf^8|moP8A~SL(Hl3kv;magWROdTM6I`>^vO-7k4K+LfY86
zk)1NHBxQW+A;?@fshJ!yzD|}8?E+`Dr#?J06_Uw#BY%;jf}lfXZ;WB1?LlnTH+HmH
zA=nmFx6`Zw!rPk!QXJ_rot%EY4v0-|E~kCDH*?;a*L&XwxO!SS3mMqjjlLd4JCS{u
zPja+XJ9b2Rc@jv~l_svWjoK%U<esONA3hr|pHp97zxpo_gRrWy5)ei}5}EJ;+=dI9
zJ*i_nGEjWS)l5F8ZrDCvF3pVm?3^j=1#B8D?LUr9FNRzE40&{#`KC)lzFgd444`z&
z^(ZAp-fzTSCjVliyZ$w_`6$J3BGF(G;i5PAF*xj5d)L9`f>Z0e3BOcP@)sexU-@Fr
z^)HCr{c>Y#j<&X~a52|yBZ4V$O)~Q@zl$m9+1YRLyI)QQ>_gr*mun7%m)l`6V!L-|
zK{vTXIdmQtZirAR_DC^7Tc_=m*Au8;w|+>(uQ@L10R*=E)8!wcO(wOpBf$N<s_eN=
zy?b4Kw4Uit?AerW2{#k{t&pLKRTB+|eQ1P^>$9!x6Cql4tSzw%xHf{m)92`_v@VBN
ziSVj2!~*%M7I=TluuluVe<qpUqCa(LbNSEi68HAbahsE(v7UdtFpg`Ybpo8{!`x-J
zwUH`XbKXL^rk%gI9E!EZOP)T>>{!_@7T&ok#@ntRl=P4--Bz@6)=djP;hY)y>E1ci
z<5Zvnj#U)_knRnK@GK=-(8I1QnNh3^#TJBP<p5iy{^UL`|G+Jocp<&=k~~&>T9yW9
zv32@unT?dn#p#e%swggBOB7YtJ0bzIW$8rF`+@F+l%mmLw&5aLN%UznrQpKVUh~~<
z{T?u>T*-_O?dk8o-Kn0__daMx(0_b?)R^wEgnQn?xmhVp-9@S7yxKNhd3goDfO4RK
zG?4n0q9Z=k_R_6LBN>k;=V!yIzEPlgySwm5@{#j&to<qMq}4;AV5#2&l*O8BVLCWs
zw-Yai#v1C2DnRnbf^&$ue2eB+gVzH#Tk{Al?5Pc$cGn+m?=WzUbBcU5S)!7annKX_
zrKmyvFknyuQV7+t&*Pe)a=jnTGWM@d!a{|?{$_t^xU0yfsoR*01Vx#6li9Ldx&J0d
zQ<4cixV<?V>m$zDl#Y?BJc%%Y=66lsyH5#ij+_s$DtIq^B<I^(Jv&&gsmrPQVK{eV
zVD~1%ZL=nluPH65CTX%~_4^n%ihJ$)#W4&XymlF<)3J!Yb#B+IP5LM1cXL4_ScMha
z)OamoDQZkLU-hP9Q$V%p^I-n2PWkw#)>c>u6)YLoGM7@Lnu)l;nkgDW-XvM(p-sw_
zt^Knc53BYt#z3s<*9nss#sWFL`G#idKTl~^;2L9z(B{n1TpaYDR^2ryON~0UaK`lJ
z=${E5jE*ZQcN@a6F@YyKg0$0rL>n6_uVitMiFr^-S|~+4@gVbolt!~Z1j*#GG=o8u
zzH!mf&R~oKHr+Ic0<2x4+R3;b-PHYqbZ%k#@&v}1!=9x5z@^3@8$(IyMy#g%rodK0
z&c1c%9jGLfdMOD1;kPJ-lfmU%e+5uJ&GfUSP>?=?B%-nka~?y~{tb-Zitj{7wI{K!
zy|5q+v8c3|lbP=R&4RE2HqRC3K*Xz~;SnW$A5`e!9~%Ted1{65N28n9i-!W0VJugo
zog<zspCh_i^S*cE$ExnJCOj^%SL$OUaQTv+y0yk3f6;9-U@5%^y@Q_$lxO85K(k&8
zuUwBGLEQNVnxtcGMeaI{#ZBy=rGA5rV9<LMf|Ldlofe&#Ah70+k@uZy5y3!>^p85m
z5>MuP=s8b%M>tl#y&<qD(uWa<W1RAJbUdh`IN=Vnrs6pVrtHm+9~q6~scV1gGj)bC
zUXFU2s{eFLszrmLlv4uszDJ~MYlc~+M52uquhuM-W!t(<bV0*~;3r)*uWz>1=YRR_
zH){Bs!)B!FaNC5EN{QPrlqif2FQgW*eQVeWm8j)6i9|Wh7YQW1obe)gkl2^=&+;r^
z#+l>sa<ehb&Iih;t<U4YI=zpz{Ru$vn;8N&UH8Lf_*DZJm??3Ff_ckXGsl*_$7^X1
zu%pX!<skFxC}O##Z)x`mFfB~Yt23#)DbtE#g|E4mb6X(KV&gsg=%esM+Bld+pd7-U
zcYd9}-_6#C=rlIyyyAocUC#~1&#>b?UM62(QF*Hz8#Ts5YHnin2zrPq<bB$~b(Cx6
z#ybz=9W>-EiO#vc%~MNma+BfmYj$VGg}{YR5-mwZ0|oYgIJ;(12PyFdFa|Z_cxGQX
z`jlR1tNY(C@Htn$!O!7=-KTYzwSKqF4Gefrm;d&>1Sfy3ocR!Stw+0v-Ccj>Z)TwM
zO7@{c=GW^h!hemQ*B&+`QQ_dKu*OvY!G02SJMJ8_U(y@d$yKfA@>!D<p)_LE-d%be
z*D_;%EsoWfUdc_W2krlk6@!H{3q9PHB}rq4F%TNa<6axzgwHbNnID8tGkJLSd^E-R
zuk=(-K=7p@S`T+ge}fLKt~L&vnLQ_hz%2NGqauGQBpE94ijo^<=ig$mKF31xOJ&es
zQ>3nwb65Q;IIGAt#YYggm7-d>rE`Avka*xhsyED2Q&(<Ye7=cF0FA#*5SrRjyj|N`
zK!&nM^aQe3<%4sS!w?LQs&1SO&d+l{T9iHPF>?w;Xch!i&MG2G0to7kM{%F@hTbr#
z$Eubg80vneyd_JI(1pqE>iI^we?l!&BQMq_eZ{cXe6hxd)MXI&MMqagGgR{4?o5M@
zuz6s{`(h}0Et-pRfu?KSk5PTh39P$Uxzg@d>@PGE2`2F?3~DKN&a%Sj5h0n^L%SU1
zawvhWlSi}cx85ELYxH8*%-L6Y!%Db<v`!+L-oEBPXg@xL;AVsV){<LG(A65Hk#O8e
zCxgL@5I;Xaz)j7~H&X)!dV7%}p?V6~iyE|LA76aRb0U3NsT+@|I>EuPpva98{Vz3~
zRFx^hK(3NImtG!y&sVJxOSB-33<AITVS@DEw_*{Acs#3@$qk#R8a~d6As?}t5N5|e
zKcQI9-Ua%%E=0eoM-=+2Mf49*J~_yoy{ozcmF}KTbxBnoPf3MfNd*Z6o~&V_)c9!X
zRN2q}wDWGe2s$E^>NT_87{;pXiU`<HL#71Z0Ood8=}Q5T*%j6CEgIB(EZRl66RAAQ
zGQh@j$^T&|xW88!+A!F2Ng&)O>Tmju$5|*F275=F`oK}zI7FZSSc_UU3y<eZmQw;p
zc5VtAZ)j-QDqYnZX08U^4?FycdTO1hP<F?+%+zQzj9jjQhU|*;)`5$3$}e2RG3OXJ
zfnoqFDmd!!bz)FpQNMN<+&YTEJ&BO|zz7SbPWjNFbCOA09Sb93x}XATG6e+S8d%f7
zU{3WuU=mEO5nDi>q}S$snM^dDt#UirY$K{*JqyR)j5UVa@o5d*>K*BL)?<?P-Q~(<
zm92@5Hyv!vjoIHmt<I|RLIcnkw2Y_8_?+cL|LH8n*}K}ZS5YSWgzo;A+kk~dM7nM!
z<*~m_iXD6<k+Ys)9ub0G2Up$wnMrM~vOT2{vcUkJCHPho?Uc9fhCTW3OXj+_2F4~c
zC!#a=!!9j~XMX9h&GsNdPons3RK!LHko-!c!ri5*hX`q+ue8ELpa{;1;?u1tjG<H?
zcrQ5X_ZLBwPbe2PeaR#@m-#5LZ@J@&zf~hF$|<MUr1`Nqjv}W74&@m69BT13n(?Pj
z;N`=g?Wp_+eq^dmhks%K63dhz0EQ0$Y$+c1JY6q(f@{JFRUEa0>UiJdB#=YY{mylY
z-PdQkKR&c)c(FruW5r>&sfwPF68bG{ca!AlG0DUHn`^k004iP{o@I6TC-#*Pu6JE@
zdwB+`r@y499`M;eAq25`;&T8GxB@mWfGh!V9QaGOEO{sZ03S>mot&KkC^-Y-Snub%
z=>Et5eiep46Cl?Pdzif`|Fk-R^~NRVSs-`X4=d9rE?>ayxK*@ggd#snk*J@ILf)he
zvec}3g-ZP_$Pk~54j~$8pQz}r;v?SgeRkw<_xmH0-$?x4PoeqOLnFH6c3>Zuyb<ZY
zK6IQzdzBuEyF~qDi5f}+Ye6J;S*53u*=oY}(cOyDd;PgPe2Q`G0ODj`JKgo{Z-v8N
zFUf^JWD6i+&x5xtWWju6*)r437Gy9o-SN+(B2WFQgH#yYULCg^EU0N$G4a-61J3%v
zk)T9oZ+^^9DavWxg*V>k?+FBUaRHv2<ER;oqQh7viz(mqqWMQNbM0{vlL$I5DU?_X
zj1EKx-x4X^<AJxzO^-<LY92v*J4)e)uHy$1-T<^pIOn)3!Do2B(ml08?+|N~(l;=1
zxWyBFRi-Y*l{jNSy7u*zl*lt!i)@J;cRhpWt3F6C`IPG)Rfgtk8v5H0ls!S8KKY(b
zSOAAf;9R6AD~k?{<FK*cJ34ak@)j2ru`)0KP>XHW8IUZ%0P}JH;3ob=Qm+b@Evwh{
ziqc+s3rCa-PeTs|WWKpQa##_XTZo@!#5;^gG}w0>(<77v+f+IHlbW^whx=idRRY+f
zx-OQ#dA(FUk~<*i!*qlE2!je7>^1Y$ko8_p#`F|(H=OGqNNWf|aS3s6<6K!|-aA*g
zI#=lNdJQf(^YPiuJln<thX<Qt=#Fpxtu`o%Rph=Swt_|T@Z`wS(kke-By}$sRJn?d
z>SlV;DtVg}@P}_>>x!KsMBxT<DA2Qq7yDHTGqH;%L`>l~adNEIc&}CfDl{DPW6MHS
z)iy1fBakM>WFxx|u~go|!)<yhB!eEApS)3BWva&lgGK|>I-vRnK(tMY-=Uwj8Fv@G
zOnGF^hHf{0&HO@O;y0C^_)?u!n4nrUr-dxu#li;eNMaZWe!<olRuEE+<ZM+q7o|g6
zS#;gn-OikgdcyrR{^_ZS@HQ~%th?{{N#u<AEWxcox<?t-*4NLQ?`}%K#RpzeKYzX{
zKmr!v^B=MGJvXvH7+`=dV<UtsQ1|A}E`@bfgxLH-*6O$Pk5AuEEoNu6)#ngFwFe)R
zq&5l)FxBMtZ`$q4^v~FuIU^Vv2NXl1PQ6h_-QMYzbHOyw^Nh$&Yg<1wKSP7}(E9uX
zw&|V(5k{4*p}^$DaePt%?}bTPE>YhVE9mXvWNcXBDdWFiV82l=DEEY}6(18x{0?}o
ziff(w)@gMznT~)I>Y^D0eJZK|?D09ZvpZ|5&}}znYJ$s{Qch}adh~;4A|(my4jJ^W
zw0eW0u|O?9>1+N|(g&Ow@>q@vcf0YTZoC}SbJx~^8YW7q6r}S3C5C&ueSuL9Y{R!A
z^IL7#=XtY~-C#4@ApY6U^_~(Ggrh^0V9sZh<~8qFpjpDk&VJvWnZx}_8!Ia>>7o(t
zIYmI#0-kn|>IJ|KJ}jrF_i?N2hl%Z%Li=rR$=@F@Uow?g>0WOV2zypP=epUde}62s
zmsX0^^fTiTsM#ds-Nn!B9iv{HA9K#K3jcj=BkBG-;O9QX4vkm1Ds47W_enm)^jez|
zIsbiSR!+ANW&(RpLsUHVHED&?TiP=H^h@mfAFO`=^c_~y%p@VDRSbpvj+4GF#STG+
z#-=#>XTI|(NjwkO>uy&{WD=(+3Qr<s%@V73G%*(%s)NeOq2bDk1#IJCQ>XS$v6~oG
zT@>hsNiO1k?vo^!P4@IodXB^sskdcmpLO?~?R}sG{4~1Tk~<+?5g4=hA16Qbqu&tm
zU;KjCLWVW6<n%gU)8VrCBRvG^Fo%59nL9n(ZVV-E!lHo5mv?fw%FGh}b)MHO)h;4X
z(a8&S(c_>~HlH5<K1n@5PG+9Fb=w_Y-xp`M5Zao|8Esv&U+mr)V~bB`bGjXFU<A0#
zvQs;JeUi$SD8kq~H`FoMkR9gCNQ<FlKuEqwxO_ylK#+n}tjZ$bCDJ<38vG)6LWk?>
zwx31*vw))jzADo}_LXz4^x2K`iAh|7cdb$1pVZyLLL2@4!Q4OQN)M0SimVKio4)J6
zRVE~u50>edHvM=L|1a8VN%BtERh><fWp4ijwqaho^5AzyNv2gl@rs}y;V#?fK`*l1
zy;$yj_4tL0>P~{rJ?MNm$H{P4{<YimaQ-ee3|y;qDd>4Rj*wYnui#)PW;wczd-g2z
zC0LFOc(0(AKO+nCqt(gVF5g79*rZMP>ie<wYhEdv3?p2hziHkj|Bck}2tMooR06y4
zRk|hBIb99MOl?YnK}!2{H)Sy_H^b?w+r>Q4dv#G#{t)GKY-0=Hb;Ta}98J6RVz5Ga
z*(H&|`|I{gC0|s(vlRF+J9qVXrC{r_jvBFug~+xT+O_E4x>`H6o~>|{i`h&rf-5SV
zeXoxUT(7<eL`&Y(Bf3XZd_^aL|5hBN7Z7yZd2Cw0v>UQXz00r%x6uZu!HFIpZ+p!l
zwS-Xaf)H!L;~AlkA{1anAVZ*?l49tU2HkOBLt*IAad8G8^<!Pu+oPeqJADQC=+7M9
zLrlouXXftO++<F*FzCW7;Z0cE6Ji%;s$46T;keXCDG1QRK;fz>c+qV-p!7rRneBw@
z`Si(gL)~}YH+j>^J2xW9sX7Tn>nr&q7s+eid*h!pUUdZyCUnpd=+J5MPwCVPO|T~8
z^Z7{BXwZQcN@CW>cjE(GQtkGRwQuA5jXaeSUDT=~dG*~gJ`_iWAIk2@mxgk%<TFj)
zJm<yqtVE4E+M`|&F7~Bf=+?r0PX;h|WW*0ff_OGS*J3gPE{j*VL2;pt{nq@%+XN+a
zY~6Zy!KUv<J70-A*|X)cI4!(<P^#|}nUPdpXg(S<;-f}2dV{PEQ-lGf$~-$@bXWkF
z;fX;RFoE4|#N{L@ov@9Odr9-;KnRW4CqN=ym(79`zA9}ZBo-cHMxL@SJ9dfN>>GbR
zQM^b0|EU_%T_goyNOJS;3L#qs1>!Keg>5~MSR*OWYjyjV4&vnLf~832^8g<FZZDWd
zM(59Q<{YW1ZC6XTDW_;EHpESWW#>djj%-XED3wJbavs?GAG71Zr2@ZTC0t#N1O-c7
zDqR=hKX*k*Y$e@QhM;M|gOJqR)7Im=irbsV-&0b;n<M|MXtUk}LtPy0A&y15kwaNR
zskO9W4TAYfTfH)cdO`g@7gNgDC!yn|VsM2%b0Zn&54POdsmm)zX#8vhwMYbmDQ2?=
zZ0hKzJEbYS{!|f#S?Tan%gOKKuQ`grWcL71u*r<3{G%+@e|zGlQp#Z+-FuJIVT5~D
z@Bty1N*AEKK#f6E)$h(zWVL3<w^>7{iBs7y?R!_nOS)uUPSc-iZtHhfkuf)p0hRQI
z=;F`CG-xwKDR`Z+L+)qY4?^gKE6ysD*)KQbH8ec9lxXx|4uSn_flt#~m*~fsu3g^V
zF?y41Xol?A)!v?kO|AsZv-UzdZAgAxTQ2Of6w=MP)pt5wD1NDgwRM{c`&A2bmxpeQ
zF%s7nF>7zaiYWnNnPhmkR5Utloj&@MFXvy5)u~2>@Aa+0rA3TXvF^H(-n`6CrE{))
z?X@^Q)IbK?a19yMmyNZ68S%jTo5rj4m)EWB!KFHLS~3*-Ssa7xAM}f}N3fuGoQyj$
z7OPeig3dRJKF*|2ZweWY@JGqWe_N^F1}Sg*+P+~gW_)t{MHzd<ZmMy&f6YDgFLE52
znDsBk`HP`u=oaAkja5{ie`UvV`#NvCN!E*(muA4hs{h(*qcpF^PW;y_0@=e)_BMvF
z6hAN;T&*|(q2yg8J&8zV4d&uoHjA(HGb_!_&6myn0TSNt6MQ&cI~B;iX^mq1#^i-c
z+tcpiU(dK9rgNwF!6>n6_hxO8^}5x9Q0eYGCb#l5?3rZC#H^>Mt-Ff#X_XH>J`wMG
zIqo4iNqKDAQ$z%N4%*w=<#tH;^>y|cIwAyy6jO>bBgD8ouIO{qK0CRtrIzoJIVI9A
z&VF%GpK?%trpO)o)5M?f$-&w6XI3YwpwROT>iCSNFwHkpvuy<L0=|~q)hbXwA?!GR
z)EDWL|JMCD1!wnQxecN(xXq3sIl+?~=^w6w*;o0~)2wOGe`cbNpNQJsrc!<H(!aJu
z)3*;3r)s#m9*osEenu-~{8r4~ccGiFnTerbqUvlJ02(~lwStq|x7V%}EQl{KZPq4i
zMYs~eY!_+^Cv$U8->M?L_%z%V|ACS?^Lg4W%6?}ACE-Ib{Ac6l+tbFch3|Z!w~d}~
z<wDRWDhL4Q1{94&YWebc->M115=nP5J@8bSlyYP5T<BD(zT1hoO?3tnFI$_m6eN<6
z>RsCcCb+2XgS_Ifh-P=67vGWt^AJkTyF5uY-!WP~Heg+%KTp6q6OS|!=c1-???J-U
z%L!-4EAn`+WBx(_^hRwU1GU_^-xNklq1<4v%j=#n5EYI8=C-XOr(j}ZgEp3Azqz-O
zM8xNIk<`sVi{MYBp-0+W7Ff`%q)m*0vX9QWdBW~zQ18Z?Ixx`FLjnYpvt<BJ6h{!|
z<Trh|4PQAJ%4I(ethi|>{<T}BO)e|7h>Lq|+iT@;<NGq4@r-uGPLI%`1xCO@H|+i!
z)q}6<7Bgl(DAUtst|UqB>4y15Eq(b*sR;45rKTzuQ-Z(OHV(uhk9qCmur*)geKnQQ
zBY0nyR>S*)+pnTNH6kCPgQ?kK+c;P~ts>xTx~brOvN_zfOXeR%C7T+WM0~Eyhcef^
z0T^<KP-i&#mga!~GVH-^!Nv4;S`m87{j0?5xR`)&`N^skHdGx#`cmcMZ<qrPgL?j>
z54)2y59JxZGfLkorQ|(6sC68Ls3MKMI46U=`0qH3YgIlGuX?dWF~>Z+wO7vbWA5<W
z(U_aAIv-(mKeo4)m;oCgUgI<L!<vmhd&iIObSOPSf7DT*FTi(v(|Z96Cl08hOiY`=
zD%QaFVr90`tEi}GWCswr05=A3otle33-^c904nw4_zEX&cD}d90T$16An1%qd(mH)
z8+)d}vNlOi@6RG&MYgE~7NPWYf8lUQ8u(_&XYAn7VET~7eknCAYSPC+Ww+S*Y?ypM
zGsO62PLBl6w6|RfzHLZ=b1Q<|A0zsO39$BzIc~H;N?sU6N1e9g5sWc&VyQT!>rPzF
zyJxQjV)~NCR%KXI$<QHHUUJYM>5R3@TnXH|Gc64^<|_s#Nl*)FC8E@6m%PSwE%p5=
z-gxx}>x53&zgs_)dBs0T`rn9?e9^a~|F{J33$<U|l;t=<zARXgOloT(9#%ZRrBign
z^R|gh#dJ3PsWAEi_EQb4npCH(q^QaCH>{*QpWjNST(_H4o^K&3c~~kUVr2K^KGMj8
zjHl)K==m|vUvIyB5yD@qGTsu+^UlmTuxIj{?Esb184u~6^q6P!!S?Rm>v!+CbFY&o
zVU4yY1#057xXHbYTsM9Dc)Kcrn^K}{8ZkfB89K51Fv>n3rP-vc=|0|zowDmEkgg}F
zpZ$SH*1f>}&I~e?rmfW`QaCV|KjYcDEh6ZSvB2<{2+JV&pi9+myMbrn+;h!AV6uJU
z^T#Q%eN8T4#qLRzyw#ty^S7knh;iZef`963hvyOO6a?67trPARwSCLSw&0hljhoHe
z5&kx#+!-Z8YR4nC{wqTIjI?DIS?K-pK!-E|1XvW<-bWgd5jxFH<ltD&^!rHLI40`0
z`plS>!XHXwar^iK9GoEN#AmIa{87*l$QK_R%&=!?V0eDq64|Los-!ma8^sR}c=w+4
zNdId-Tf+R9h;{q79wqE--LAr6+d`uQ>$eTV8K2M{uE^PGzrI2(uKB9+_HH8`&puab
zMrEv{#AYeALGN;LyVT~1gp2QJcg0}I?W-YydbSu7_HBLxm4ObiZzeq?1l6C~8@)@c
zuR^$}eLRC-2CXPSekZuL`U1-IaI2b->Xe)Cg2dC44UQKILEA%FIlSo-Gb(?gMx0Dv
zWna^`vZMyDGb!+m2D{}}sea0n)1m!EqJ;M4Ej9r=Ue#w49*Jp*??mwbq3J8b>S&r^
z7f5gm2~P0f4#C|$ArRc%g1a3Y65QQ_LvVL@cXxMpyZe6kUVa{U&cp7^^mJEOS2_1_
zqgjnY%N9OfC?DPY3BuxLw=qJ|R5Na&n(x@2N($e$igKV=sXd{PAwOG8CQ+cGU6{;(
zC@vk=Uw$FMI@2-j%(ew0X_(UA(dySs2War;7Avn!5RCfJ)jo=}zN0n$fL$FNBmF@1
zg_|jqh<BRGEY8ENL5<jx2}*k`ZpPlI&Vt9}A*C#0CQW|R@b|XEfuCIBeL}qEmuI#K
z=I)9z2&XpP#+mG|6R);A8yIiLWyM4>2u&f^TbcFRtwY7x)=!XJE&dTacyKaq_CZcY
z-{mG&`5070BT29bkNb8mg4{m=0EN=l2BEJH7=6(Y5UA<scs=fITwNQ4vjFck$lK@t
z=&%h!xSua^np75U{QIZUA52X+J+i_4^>Yah*yd_nOoW2MTf7Iup1FoAL=BVPJ?8S&
z+E}%1w1I-e!O^2%fVyWbtsVy8Xj3zjrRaFEq024z>N=$b1W7SWBwXJ&Eky_W`!3en
z<6^SBuf4L2PYXEsC&^f1k2O<qr#WggDHkt?=ulzo@3v@0xZ=oA97S(a!>qy2)xXyB
zhcP<6LDB1LzPJyy0B@uRWOpl<{BP^IJJYIvs|`~&9a0r<WRKPMK=_WWvi}h+QpA=_
z`N$srg$qmX2-|d2FB64wd2WKtLmJCYR4OWFfwH(Nl%T);986lX@kJNZv)>+tI+TZy
zy^3Yo>#~qtV`PxTnG^CLvH+_DD<V{85@XvB2Gi{Uu&96AK6K%LybkKq{zTss$~YWh
z%8#zZ7n88(vXgbpi%e0bHw^it*o+g?f|DoS9oje=^Ij9&U6cu|Q)|PQ8|tSe&<`DS
z?0|~~euVwk8iVgN3zM-lBpNwXzhXQbYmA00DVr%(LOpGB)rfmSCH>#&utGG7n3Kl)
zV=@?bYn*8qV$X#k1aj3B4^YG{LbIew+$oB?@72KXL&_P<VevArL02rQT`@e-I$cdL
zx57M`B;z8m|2Wx9j3mMOhg-z!4nMQD^mY7(m-P-k`ojzZr`7G|yZ>D?^Q7qO4@o2$
z1ky)^FfsCJL;+Y-BxF5nJax!-=;9?_2~qYoHuL3revn5IUhlOQ_c{`|0AhenUWF8^
zkN+R;)Pbrw0#L7-a^h}5*tOjOeVqFdM%l}~u{A71ZFPBVK28vT_Q}+Ck|*VN0n-+z
z)PZ*Z?1<^n;(f)cR#b1ZE&o;T1K2`El%RUJ5q1);huhh}J!_F_AR(+>D-AIjcOZ|n
zHbm-#@Se0nyc(%%EPh}51A7r~()^IR$~49zs|{D4L-%%O9BW}Bdc=x)1?qym!PkKu
z*-Jytq`T<oM@du(;vV>SE+$1S+zzSRJ&D++7Xv$a>fy(tgwDwvT?5Pg;X<}-*=!d(
zhvft+)1<IKAhQ&4528AGauA|=H?Xd+XD9+WJ7R3nm=UIpZ<V#w{2e~BwD-pqSl1PQ
z@bU>M6i2;n@YyZPD6-xhx=QY*?W_@yDfGp}V?Fe$RiX*F?@Mz6bVEMTlKo3ZK6+$;
zn^zbh?_68>jF?;0D3XWSzV)>C(PKnAyryICA>WNPK9H$S^B04$)MauHm)u__2|NH|
zIO}ZJjRQ2}TNCYE>i+Ef#Mmp-XOWBe)4ZnK9atk(dQ^)Nkab{cf-s74PJz2Ys|>wd
z><>lj@Idn&DZGR|Csgh9-A97`UbGit(-i>Zz_h;HItM_r(nz8u)*#B#F{nuH1zjIt
z{C7)C{(~PIG(PP(61&4!vVoxbK=kyh9Zx=SW6oHMMNU{}%9wx?LapG8w;%F*AvRlo
zEvQ7lnQXA(EBjkdS_5gS$M!YZhZYzQrB-MK=ZURTIATN8;_j2*YNkYHCU~quQ$~=g
zI-xRu({sjX=iU5H{Lx@L!I?I}1E3@LJI8kyWIHc+CYHWH-<&+oQlPv*QZj$blK+dS
zAp!cC!Gv7-B~2sX@~VdyNW?E_X<7=D_xds#4`hBX8pU)5^_N85q>4q;F2iVhlke>S
zRnA^$z?}0(R<(s#qh#qneHkd?$y}nTGl+>Zc)928W$a2@rwSK##?n|QUZw)}Q$e%^
z3e?@-!8czG@35`r0QoQExbF*vKFZm!<kO+LQn`1MkClqhY2{aD#eh#Vy6k^Fs^s3h
zGv0#Q`~e~Y((t;w+z~DJw|hv23s;fdSwEv`!UJRe<EF7et<^3WrJBei9w!{4l(L0x
z{JrY5ZgY)Lx9O6L=R4Mj?_q%`FZqnoyO~BPFxC@(D}xMwZgM<{6~v~6hC1I*GZ#(&
zxZ?`8u??30xj$Kg*MXncSi1$6$9=g_uMe^M^n$XKzl7_sD9ML0k7vVg5)Hyq-+~C$
ztQ`NhCSXq2M3lxeYBKlW-Yjos<yfp7M9hItluxLV2-m3QoME71QaPc6roX@Tj^5x{
z4#NX75<PhDI<jej>1(zcn(k&i8}dmS6QCn7SbZsIqo3&-tp7t#Z+VbxhvuA@M!iWL
zTcjUPK0Wd7)Y`j_{mZv2J%WDyVx{lH7!nQ1(1Oe|zH4#@aIP=BNs{wT%b@rUuNLcV
zJ+b@8$u9h@64@p%E;yd?YE-o~ft;(9HsXHB-zP$MaBY;V#ed+o`G1MhqanlS@D+yz
zWD6UW22*z4#@$Oz+WZo}3@pe#?J2YrWJUbe_WV^y7{Ayw!Dj9K|8R$9&)x3e41P^(
zQ4)963pQ)uzcHoRZsNV)LZi*{)FDe!6d$xKhpGG`hhT*wKUic!@%kcTAh|*2GVW1M
zJRsP`Q0a2HzD}(Lg(9j$o2%5n8sjxs$1fZe*LvXq6%E#pWS<3;#SNo_(!}I3i`_q~
zP!<z*sRK4uGatq=o+@YRQ%GDsHoCJf9?EM`>iE9c=JO2(<Ryv9DSrHh3dnwprq{ZY
zc2_~lrm3wCpNboz@AeTV`LGR1q$yLI#%!;X1c>`ryZ=j6GppgZALzm1tk%iZL|ORC
zBMG-v5N|NgYdy2-YP;Dv`N8*D2_T5_{qmjd+z`w*tCI;?(HzT?OJ8A^oHVY=-crsC
zL}nHUbtU9zwu%8&=!TF0g;cng41#dctmUM|c%?z2GkLUHj*D7$$ZUqy_^47}U0K`R
zUp|q(I#o{?kGCZq5N!<SDAi;-shU&SXi}=S`rivpZ78g0I53qN2_Tl#xn}f!+lml4
z$Nt856AL!&jAe*G9rse5m#Dt@IJG%LGm*7fm$l)tu2^XUL@xGmwH4K_O+p#^fkMNX
z+RL2F)gd`FQHTf+5s*sf9@_i4WS{Hjzd);y)sv*hTX<DwSal3zsx|*poZg(8iCF>i
zs(TDeYuKln-;E!)*%7vZ_74VEuLa}w>;XT%MAMpm`pcYzZYnYI3uA9>4K`w|9d1zp
zig>;wM~s-UxebNCO$-?z39sg-=pxE=D?JuwE{ptPwhkGD)t+u?8#6z~X@nFPckN=e
zF86fyPlR5+Iy|Kn{u=Rcb#3|t!4RUrfIcZ;KE0;aMlShs%1N6HMd9j@NAi6QAh@a@
zkF#|#AZ+(-LBsqGegZ=T`+ujvAPR5LE{?Si4fcDRP#%!0v21|wAiMl&V4gOGtHtRQ
zJs7g~x<v!{yUiED-22TTdU2=xX6&Ji3HBOQW?(nAs@)|ytsQvp*rzCi%%j^9vH%5~
zJ6x~?IKlrColT$)JC}?(F&BBJ*#C6aRj)oeW{ssoy#*f1{eR4(n0IDRYycG#=63)f
zO_j4x>E8ME<F_yZO~GL#lwf2F$lK#dgpkKw&@TM=t%>liz%l_^aGSU+qc3fE@3d(t
zylkHGBSN}zZ@BR%6C^T-hPk*FW$TE$Y60j8ekPy@Z=Z9bd7P}qI9PBrY>w_2{0T7^
z+{~$Sbgw-5TO(3VE3CK9W#n)0p$>fF2m8|%89J0VDBd7;4#Md;!ZQ0&9|mV2y~F8d
zi=79}@rqSm^8S9Sn%8$Ujpm~dT+Z~hV>H&QUXKib%c91=41DI*v4P-AVtIK+Gq!tg
z@Gz?`kLIN#RkK&Tcj55UX`~c+qqyGQQB6^ipQep10vB&7!{%VMBVCBvI_7a6$2z5K
zt3ZBZl&v4^3G@1Z<XSNf3dm;2>V?U>#2-H=X`rCjQ;NINVEMeqEU?Z7Ou=)XH&G;6
zY(trBT%No_jCTwL05X~7e-Dy3cb&!h!|@gl`b9fkIHoSF@__N-1h77svfyDo<aX6R
zPO+eGimz8c<c{mGeMtL>xzq0Ou_C32@#M*TtLocmGUvNFG4Kk2puc+<J6Iz!IXOAG
zN0J#TrIQfWi`YQ)MIPT&eajd#9$?-fO0avaBbEO)#)`XhA0pLJdLIG~20*n3%Raik
z$kMW&Q0vGRvfZ14uY_eN#X`dsBguF17k)FeqH>MRe-eg$>9;~G*Ve+{QisFh16j~a
zoTu%LC@<@|c>oCbK;Li5^VJ;%M3UVw=F`j!#mK6}{yZBq^S2P_n)nh`XT_Z0e2Nz=
zWuVwIl(!v^Q(h@AjY9%JmZS&&(C>v`8a%T{q$nZ<rU~yB4?s%s?Cg#5I+RE8cPN13
zlP@=|j~{MNM<~n6+_4=sL_)e)vUJ!C<7j}#lnQM6K?a>`?QpbjWjPFB?eF5}w)%pc
z&Sqa=1;KX%k|dFDtSPf*JP(ap@?O~Bb)Z?y>C<p@#aUs!!BO*TS!^)8G-)lv1M<qD
zlGbp|6=8D-0BZd8sZTLTzNtBZCHhs~1S)6cTz)jmU5Q8^DQHL93bU<azRz`MF|>wG
zVT3Rg{EF$nruWmq8|;?pAR0LMJ_cLts(}v!==iS8#vRC!tMI?5-1zcgx|liC*qGn}
z+h0Sza<p&xeg+lF%Edxeyoru@Kec(xlfDP$C#hGmoiPQ~Qd&cu123(iuocY(sQ1yS
zOJgsf+E*rnrAeUM-S_ZjKNmin!WKW_c0y7cPvhkmG6gmZ!OqqFWj;uY-3&Jdh;fq=
z=iMF8W~cuDiK!PV^p1=IiJcGo2fyDzQ^gg9wdwFVfYBu9;-M*y=(B9EP8*9rv<t)y
zRi_YyK&!8a)a&RJbwST~>VFn6VbYGU%(Rn3E29L!t>$Qt%wQqrObz7C?v)VWRGrMI
z*Xi_xNA5_mVQsUTan3jKyuVvK8;j?#hstaaQJu@nR^<o~RTb#K0t^!mwd;DeCrb0Y
z*p-0!sLAA|>+gk3vW1Jb#XpIA0Nutj5q$J>xka#IiB?7cOevE3R3;K?rOuP~PL~?s
zVE7wGF#35!(fEcUAqGV&2PgNE$Zk^y<1|Xg(iT&p8xI(kM~+zO@@GlGrF;}%1Og(L
z$V&|^1e7Idj->Gqtc!WFTC%niX6Rgh$&iJh0KI>eRbW3ny222Xbr(ahDXqpj-)0=I
zwKMGYM%*0VLITKkvXJ@Hv(@Ok_#)wyPlmoNr{Bofz1W~Z{@Z4_LgR!khN3Ar+n1Ni
z{TZW%gP-Qn@>x(dha!+1Y%7?_VuI=9O{DcaUtV%fhbZBCgBptknAyj7RHc6tD3-oG
zOg7?XX`{-fOR6t)b46<^t&Qg6DvJJa<<W7VuQN`~h@qYVLW-qCH)*d(IbOZu7^6$y
z4my`@`AHzBweO&_GHY+H)9jK6V7zba$=IB;&}uupK6-J-JnYSQl!*JN*4iDlX=H>%
z+&WHkf7&!<nMtv25tdMcf^L87QEl^V6r%d+<xogH>UU})Q<EE`aepb3`|MVV4D7<G
z)obu8&C!Dy+uBPwbw{vhd;W#(J#sUFA_yk__J%+>=l^QTj$vPcc=#Iw8$FUiBXijM
zAO8Z$coh48?G6=w`YT#`j7~ji?yo0bEhL?5B}%GEvoq(=-`kY5$V?SyX=WFfDOUh;
zh<Oe<LA3#R5Jm%|Ag1(8?-U9`t8puPFBS#46J@+MEqF`(ck%pAm?6fZYD`nCJbi%>
z`KRdvHn!ytM@x0N;=3;`P>~<@mEZ3jn8N*+<`l+p-%kAR%9zA{L0WHnI&Ff|J}aqP
z#aqXX>>DY%*M*aLtm~WN5=Ka2eC247E%y5maJ~7jVoVl4djX_eHIt9NsY*+sggkZl
zq#QayL&bkX4ZU?-E`LG<nw#z*Zq;R5-;O?_uuMxCXs)TVo%TV_2AOw2EGq`j;^06j
z`z~HCk?>S666t0lvlRiHb90PXwbf73pGvprV`n^Vb%3mJ9u2#F;7=*)84G-Tpvymt
zQ9DWdlM?5f5{I_w0WJ8qZNYx)Uk!hNp6q2!R(!2A&^{Mfqz2YC^0km$MfCtX0N#md
zBI#jyom9)T3T7Bwjl+gY>#xfBAOBlHf39SY<Y{C;NBb9oeU2x)%o9|QP;<t5`Hi@Y
zSxQdzN!IUigwyLeO1^>(><w9Wme_}H3=&yNT67y)9`yu132_%s2)BN9OV4(PEeQe(
z)Nun)?Q<&!>vh&k5DivP%#Z_Mf&P+khtZx;?NEzkB*Ol`J}<7(BX&EeXJG3+Cj!NL
z37;3_-u=EIb*Mm+(TvW*@QOP(@rRK(t5Y3o#HFJq&el-Qk?PUYXmsMd;9gryh=tlR
zs!w;cP}sGtCOb7Srqt$Px8Bf%`Gd>2jHAsHLZ9-9lLXTKAZe!{uJ$oxst~%IIbZCF
zdozLr*(Hk~ubD&dF2Iscg6GR+o7Jagb^+<)l~idr_C@n2kQJYSxoLViMLM_TnhL&r
z8DbSE&SZsU`Y^1}iu3YmK74`=*)Ev-)1)OZPM5|HLFpbxg5nA-Y#kJn0Zw_5bL}y<
zJ&6Gb5bACpc7c8Zol1Uz{!R7wI(gp@{smZ}``k{ag{*|{*qzQgSMp$r8~PRpDEOAC
z-SzCzl9lN5?dK!b*AS?sd?!&UI^H*gW0m%eP^(TkSm!&ZDk^w7O33(x0}E_GX+O2D
zwzC~Iz7TG}3W}qWVJHMrTg-2|g?9%mY`uoWLJ-3nuh7H6L2a<s1ARCIen-KnM&kjG
zj$DRpdQ}TFaepMqi)3(!-?VOiJ@;sYADwMknb=;pPzD<F_BW8OGS)2t>lriYN|%{*
zqo8Ht{~Pe{-(DpWw7Kt0!F~7|MhES)eqbrU#XTP?zhr-WNhv_o`9>BTfv|J~xfoJ1
zsHXAC0sFhWgb@RV(ULDq2Umw{Jw_#@)*5)<{+WL4R}OQYW<_Xo(-A__;qDkBEr~_~
ztb3bAknuiP&|=h?K(#lj=2Ma#2f5Q}fx*?}!swg-LMvxs+0ECR$N@tSej>n>4KIw{
zA>wOPR`>rC7WjUQClW{q(MAmwC1p+;-XKI`Nqyu%%K5%O#1cl_Zy@}w!M$b|^|Sj*
zRG*Kqk7wK4G&msiWfG1yS5B_Lz(ezF+wqr=tG?T?bINOkEMa3TdcQ6ibhoQoDmcqF
zb4By`hg|_l2NAH<9<J_KKq&hGr9x*i3EN_6s6!P?^FEIyFY_=6(jQ!PU@7uJ1{xUZ
zqR${Kwl?0GTl!@Z;sV$hr`@5WSn1L%=32jMpaC+%NyTI(f$p5}`YvE7dU{d<2c;E$
zF9!)oNQo+KZ`$T46=}aZ?15vlCAabwqMuvJTuC>!R5D2ZW*Sc?ZFlh|O{tGyaoCWL
zHY!hS9qq0BlhuUK-^_6KTYOrk+M2c}zxj;uo1iKWpDPic5>{RAypou|8N=rG+Z|Kx
zVb#Ox#OZEbaXs7_+gy3iSPz%GYKkSd0xXb%!I$@N058w}tf)RuS8asz@?dFG#-YE~
zS7oD8t6KW(QmRrPvS3|@SDN4$Y>1O2YQx5HTf^$fJ=g)65biUM48>bbYOcLhj{07t
zJ4t1rHG&ne(5RLrrDRA&Fa$zwIZdjWJ>TIkGnhry_CakN&~bNpdD-iJBPgqZ6l5j$
z_4I}c0x$Ubn!gl!%K9cIkOsg!NbV{>?Cts{4h$!TRPQ;*OAcb4vYwUSn~24ROWkW=
zYJ$!KkY5v1xZu+@(SB8t;by}>;)-?)C(jPV-W<NS%ZeC<3*JLotkWs}`6iEO`!^*e
z(p6tf`h7l#j8Q~leg!=kk~4B&_g^M@$$nCC5HiwRpA;n&wktV)j&4L92%8M2NAm1U
z#ZBH@i<xyfUiC{&D1lz+>AWreavV0ba)XKj5keNpQaX|S^`syE(Jki>@f)Vdr}q53
zymW5IKjPV0;B1FD6o`8fbENntv-5ef6+e<hMDX=79-WQP4-zIv5sO3s=zBf9aZAaX
zrDWb^Mj`l|7~q4Abj$EMIkV(rM#P_tgqpyr@u=Z^gdgvY(`GORw}2~(Z`)ZdFDpd;
zY1oKvBCBZxi*3;Ny9OoFmLME=lwjhzk!@BcUsXKgPpU8?d1Tw{R6iZ<z3u1rzN#}P
zMb>YJmO@%M#zzJ+A0;spLgoi{2m*fG+<<cyW3Vx&ea}2r(^^zu1=GlE_NOejt%X{G
ze5_z&hXb*WVkQzc)dgPY@h`~Gtbv4PjML9i`x2AIKta18cY(5NUZ56pQBhGkcWit-
zw08xwG+8Kdc6ad=sm|jA$cKL_vyq_fjfsIK+Nl1~uxTlMT5?>O$IcM&{C1q?tgG~M
zwtdWf%EEY(vV{W!UJCx}rvoZos1N0OJ+}v?M~#hxi*t^ST1`iY4D8*$#1NX6J1j|I
zNwfnaim9hz+BG6ZS&A4bmY7Jv_19PFQ=YR`R#tGV%Kn={ez*S$r;&rU+j6Eu3o1ND
zMUwX2J*Ma89PaQ=jhyFDYRmT%8w|}+re8#4fl0oG54)sBzk+JiOd^g68NBMn;Q$2|
zFU`pIowQqBiwZ=5)UX*X9+Dg4s-@;0((BJ(yCGaP>Qy>0KBc>AQ}k7Y#O)xD{7w-u
zdV6Wqc^=_VwnguUK1aset27Eg9ti)0qDYj;U*hI&?0-KaCM9gON4=#F!}+g~Z^t}r
zVIi}k2M!6y6f#&+C^BA+Nx=RhKEOSN7{40`@v1-Q>vVz+tUYCvf^<}1UH*$Mn)u-c
zC-(!LqO()XC7K`wzi&7Dtbl_~UbSWf7C;N%9sYxV@jF)!BRnnBLD_Ca%F!!Sa**hv
z4Bn@n5a;+L>2Co@EwEau9l_oJ<pWUcxZXkA#xb!odyHICTYEsJH?A!Y2|tj#o^32Z
zm#)<%AYp5yKh8OaK~A(LiUKfi)=s#&Or8rk2wQ%V3-Wlku#hOTktjRl)&MoD<7LQV
z6Xt0`AAB1XG*d!<1jC3Cge|z1OE-CF*4do~71aQs9iYWCuh1m999&2PQkC#>i4lNq
zbAD`URwX}EbT=uIp0h8J-{X=R{rOAO`GPpUZN2!ORXfKK<akCJ>3-ddsGJ`MHsV`^
zt=btf5{utmY%x@(n6(M=J!3@OcQzCNlKgqEauo`Iaj}t}ZlEx5WLzRh2q67ZQ`@~@
zby_PiuA7OEygi_NUlOPpJ($jVi89@hSv&bz8Ly96%_~5r5jl;5l)+2IYJKTxDa^@k
zWgt91FHmb0@=(TXig(TagQC8)4Li8lrvvX?qt;UI-(7uuJ?LQuQcpoG8598Iit;lg
z$e?$FpN*yQP-2C&-EW30Sv8%o6od*Am!e{Y$_xf0yxNUDkYVH1P~Zcia5B3QGV1)l
zVuR0|5%vR}Mm)>BvN+V-2^U@ePFq&J-95O2`g;Qy%sG<&{BbMM!QYXJEwGUNsHHA-
z{=^0qHQ}C*a6k2+(gLzlhc!<hQUW|%Xo~By-7LyHZ6-U#5IWN52$A|Idf|@g6f^1z
z$fy?M3P<bLn341HuMOJKG&7hKJkUHl$<xF-HD4TLG~_@%D)s2n^U9}ML(BOT9){bJ
z4T$%`lHhAy|A6TC0QicEESF&Pst@DldBruclfp+Q*Ngg;NPbBbb7du!Wj@bYzUa2)
zj#p*pw-=G-Ec6PMO@l6_3zhkUE+Ul<)twQCzny$vTVx^CqtpMKxuv*eJLOS!7zw|G
zL^;QGyZ>@N`t8P=5b&#>)@SJO@q_g&j|${Xb?o_N=8ubW@w{QavWJ!ojBF?-pFr_k
zdMr#Ab{ekL>)*hf{jc_)w*HDy9*(S~&#;+(F23}`3OigzujVY&k*VE>v-$tR1UDur
z^uNUOC*%YFa4A5giX~MJ+nk+cG0I@qslGLq-~#FCgzS6@llIYbXt>2+4GSQuZx5B6
z*OCq^EUGYG;FLT7dS2KaPE!{uAp&(PPD`M?L_EC)y9rn|XPQ0!rcA#FWv=8i_*+1Y
z&kl?KmY#64)6ha$778f#E7zcB6GN@%I1p%CJe_ef;ryZW@-z{RPZ>d&`8QaPgq(?(
zS9NtZ>qLZtS)Byy_NNbofO(QyGWz95&OZe;bq6Pv3CFxwkNVhzPn$=#C@7iCJWI&8
z;$bD_6Yaz6ap%C9vYb$*he^_+A!SHPC>L|uLrW->0L2dsx?Agv31`n{<@LLHNHw=)
z*a8AzwxM4ZV<_h4>eNQ3bg+I!N-5^&jdEvfn<wGSIw6I3`!y1>rcFF7@22&eRr${s
zEwi_)p8?)586%+BoVVwR>bD4aL{|?*KffFS{=bjw@usLl`3cJTw#}8rVn&CH14VP(
zhc)3^n1M@GFlWOj%fv&!KcyFN@JKcan+@45eXbv0y$R7-G)^N}yAh-MRmak#$)%9M
zRXHq?#y78Rb$=fWnIy}~)iiTAGAW)|#q^&quxC3Y=sO&_nt~=wBVZ}dWyOT1&M+#E
z>N_0fLvaa<E1HM|3Hb_kCz=rWH{5?vI$ByN02=Q(V9)~74l4VBE(8wS{U3V0`<RZw
zf!o#EUrS5YZJv8+&Y(Ims+|}Z7coiKE2ZF0b9lQBJRzdim8U$4t98S>FD6k?kT^fR
zd3yVNkIRu+enYDa9VW|1;!=3E4;T%R*)^PKKP?h=9Jx1e?du{QuNd&yYIS6k_B3SW
z)Kk(jJ=B<$Hd|nn<*SvtjgmfZN8x=oXxBL|XE(J3Y($dFqzGWf`w|$gN8EU>Bt#jT
z(1b`)@gaS|+J~XYIKI97d=W8hHa0F^q#U_HwxgZ8%5=1m5iO3LLf_BrF)w8K&6#H3
z!=kr0&B{^+#8%PqoENhm8n<lq>7QIyyg+!6*)Jc~QgE*v_5Yd5`hV~2DxSGb!-TE!
zTXc~Zf$>zaLI<|R9Zwf9>(ra@&)MhJ2w|TeW+C!RtB$8e9nNVhg{S4#y!j~T<yhg-
zVv<}3dLF>I+5m@B9o`9fLcoZ0R9?WPtK~nnv*#AsH9d;jxFZP(_`LcS(zS$M=0qIY
zMFl<1DMP17wz?<d!QaMk_7dlZsS=t3$g|$o=H}4e+uMe7l5TF?fReIuqxikYD>y0r
z0NIZD*82APl;L^1K&%3456ltHgq@HtjK$chX$_rI`QZSlzK;h=UAKQ{8=XF=(Ag(C
zX!lh9*gVx>F-de3L1x8)jrnNQ7GHp>&9WXRE|EJ44ZI(C5#4`Qoo~IBWR1Zc?XAc`
z*Yghr{a#F{y8Fx>+u~i$JUWbAoj+8mLFWDaz4>;MvuhgrOuBV}Cz`LJ9z;>kesg^Z
zni&jf?jL%Y(63Y*S7C18dM`lxHCTX&Cwykg%{l60VE+3JBl99jDK!Li%{*Tx^7rph
zG?lFx-O4Y6q3n-L7hdrUvPUmOqMXcq$yqWp<O;&}s>OMJ>z(i6ZZOC=n*TM_g9IM$
zq|x7~_vPuYph!jc16GOtA4ztqaZ(woQa6NCFadJj_zC%R>#7DK?lt50-ln)5q8?l_
z)Q_cAPiku!Q+Yv`Z`?8@yN;eZ@*K%E${4^a3#*5wgGyriwdequz|0cDW4~BT6V%z<
zWXx6%4-K-DFN0;Ji&7oe#P&OHh$?QANxQ&iqkg##a+IVYd&-K$=GGQeW>L|gClf6#
z65w<OdR6V9!*_rEJz##O(>vLQy!j`SnJazayJ|5T6fn9=WoJw+gQ{L5jaFf{Rr&fd
z7v!qhc0@)jPM{dG1FlR_T)c9N)`xb3<ueitD`{07s0uN;-Zw%kXzzIlaYxbb4-FMR
zoQIwrTD*YdPvir4u@-E`CThjhA|(9Z3*UQRTtJapy%nyAWJQFW0}QZ&Fx7|6YYWHk
zW+akdycgDbp#W6WkmyHO=v0h>AL^z6m7*e--ioYeN!aQSxlhFfs7NS)4>TPuP6qy1
z_~C~c_pvi}e5kQ)`zm3`IipOPM6x6;lqEr@sOn$jHu`|i44trKW4e@jDxYot2AF@H
zLXAen)F1G<4@1zt)-z?BTw=udLL+UE`(Q~Y94Z&wZqie!h268)Gpgm724!SS*>a4H
z5B%%bWt7+Fp{~E0(=Hyr<6#B%Pp(CHj)8MS29qB-d&qk{1u78`ZeTqPx0woIHi$x7
zXVF`4Yd$+hB>k;qL28nSNTIcrxDrh-ci_Q2i4d?FjBH3L1kg#11^B;Qk7=a9|9LGL
zJ|Fkj1^UA<YBk3bXn!#zRE|p5;-msL_Ibx(fZ#4!G0ew7TG-4npK${ELt0Z$ogt1+
zLO-rWKKiYcI70wkHp2~7t2^R1bmphpiA!ZaR>|obsE4L&(*W;<IZkyGJC1b?<WV<1
z1R&<Gg526sT8GygNGFSxHj<MIvmhrW^&Q&-^)fjLP#Nm%AyiCE=zzM023!LHv`;++
z0^)i4`$PF)g7=|XVc*$aIwQnFZ${g95k588|4UY`f>chJ%}TugO=0w9n6W=CmNjzJ
zH#kaCIY*U3_vl}Dmifs^!$wfCBf#`UVYZIQ(K!hZ;G{bysw2M(6+X+g_Ym0bm3~jK
ze3%nHy>EkT!8#A6n`eD-1@09j1|&2Ldv4j$O7<!hAN9DKN{*;0u5x80RNy0-s-2gB
z_KBJ;gs$w@B;rzG#29y2IKzp$oI~;NWvy=vG6rj@$2%R9LX(W2ZKn^pyZ*r=CWkpx
zMe$C`@(cK+;2z_`uP1o;`MmZaNCYi4&6wukhi$jIUB-X9tz`S==~YP4N=v3mlrbfw
zV!DrrQqvOeVMVI4@yoAWBG77ggM?L+iB_z!7&7^N*?dLgS2IcB+}iR~7#2mt*6y%(
zy2O<5!-Y_@dQD4p$<z)%yLA1HZ+hl_rPER(Jd~!iHyT!jwPd`9qqyAE_gSmY*ksV&
zl}`9EoYD4bTYE1v3u4k#0{mbl6OXU}2>!SZM|^p%w)%K0D|qf~h84ck7ksOPE6<*|
z0ztAN{KpIp1a|nFZhN(@;#%rOS68zE_s3OZq2dO6p`7W-$%i(aPA7{y7_|S%%Z!hY
z3o!(Nd5_fOW+zPdi&E=n0UAA*SfTr#4aA>nRr$0ST}*o4MI|5agPXbjULP>>*J7xh
z&0%r|%E`yVw0F>y^d2{ln?n<A?(qym<N^4YOE>i)HbF7QK&v$yH5V@r03axwVBR_E
z=AErqWsI2U%8ua4Ci?=|)K09=1g^Q{{ffN0%UK*Zi<FRpGnK`vRea;zVB$8#f}Ju@
zMGM>DjUGwD9NJtLoG8x<<(%Hm#Taxp5-kb;5nS;kVXw`&&d=e)L*n}$Ab$@b!s2$f
z3b<#mb{J<QI;md0!u<DC$@!Kr)Ic3Yzp2aCo+qtu;95r&C?dDbfejxGi-4^W+5Xz5
zIVxhnDeEEB6yWtz&?N;0ENApAlt;Fa$DshisFtezYvqw^&cnVDtWQG#8!j@KvL-5@
zTYPh<PcXU46gztdG<ah!%0RyK<KSWgc-3igSQ51UvU<&%v!kQI*o&3`6OAT1oqmuE
zXjf-9;9*31;%aHBlt^@Q{?wzBrp0vV%JfBh*klVhzxXgiRLAC}lS6M&=h_ep*Lc0{
z@&QQv9P7T?(V_kk$a!D2<DsN->1I-1=bbB)@ZV4;Xw(dPw?*(n9``|^V<<o|PZl)j
zb~|m@>kY%l#=~1&U3~@Bj{!k*^FyQWxITiK_T_5-;;tw>p0`?8->GtX&H4WBD#YlN
zK>g_wIOwih%{4Y-arKdp7KWts7h<$gTbjyGYI*^z+9|+?R9>8MXI4ve&?Nw6+_>Cg
zcvMhfE~YdlM?e6wtTbT-3;Y2cv$P_~1-=*lnpuJftbzYf%#P*SA4{|-eg@@mmE*GM
zA8aqTrHv}s6b&@DFzxZAI~D3@0pzfLy{iwMA^|YW>(qFy4eRfK4{%S4{FGz`Q-2Z(
zyOKY2xgF&wTIaVhQc{SQ&##<>I3cBB1-}_V&&A1PeEtK0D8a*}NS1_Iu}?CuaD4Je
zSW}WpBw!2dP;4OR?T7Ooh+-1(x!JXq$d3B&G7+sH1=(7h)c2*}WS-53+vj&y^IsC~
zhgE04#dY89_zy~-l~C^WRxu%$%gi7!wV%JP(yy7^Zk-q_-zmljk#1C+4xzo?bvXHj
zJ#XebzCjS1c~;d!<$-bafdGr*r^K*K(~uYpI~zj$A^0uq8dVvFr0h^QezH(diwsdH
zWkSu38YR7A8%(`C=>pHlUJsA)E|P<YNOqnbo$mW2luBSteg?G^b)f>51mHJ(3;p=*
zTWo#MGMer|MU?OdY!03RgPYxBNfKb%j3A5~0td*33ZFW6XY!p_8IP#9dT;@;2?=dC
zvl_$eWD&fb`W4dB(wm!`EUc_so0~Xj&miT4|9|;F&7eq~EPld=wWSGx6)uji?9FxA
z!L*W9wR5`=hGe#Qz43RpKba|{F^F|x+VkvB+PlZ@?%2|aumRwml)#hyPAlR_4+Pc1
zHpW&a>N3k(ngg5qJplp0X1upyUedE6(G~Js4-n$R0eOMX!)!Wt(Sedg*a|GK%b~=8
z>EvSiyz1?RPj3rmGQHR!Hl*<NUXTpGZ_|$<9}WYM9UaYYMeh)l&F+#jjEREv)=*Ss
zI7%at3TKovg7KadKC?LY;9x?AoDz*(6!>#o{qXQgS(xHD$!;DLKKn-orK`<5WYog?
z4yin-k_!ir>KJX5r+hBu_Pv-{OBp~oA{_U!U`ZLEraHk1&i2?hCiF6<kV#a(Xre{g
zbMKB}z1P;5h*N0a{-aWO%<fysRj3{Y1BeHJeRH)^@n3aZ)c8U(4)ahD;<C2;hfU?9
zITJRWj0vWO>lYRv4Y$7Kn9NWDUZ9nJx2)<@m12&MFgt+$No(2sWsG^|FynBH&GF?B
zMc0h|zgP~eyR;<Xa(}k~t#42p5S1ignN87nfqyz}Uc-sZ86KAiVvegR;$e6<k4I&1
zJB*MD^^(v7Q~;C=hrIn23xRF>-(9_7JD<S^xog6A@~pec=~}q4z$W~3zSp!nk`e3p
z-cjTl>ckIj5*xSsD{Q!442I#7G*2R0TYFH^*@G<<l0=Xtz4BL6OY776eY`%vY5?^!
zKD|~|4IZHV(&A)UeLbQP?VeGH&&$rwz~D`tgH5v${%0T3w%HqbxHXzRmF=j!sai5>
z2ns1u*L`vfuTq;QGU<(HOJOK_S|h&ttn#hVhs%zZ?Gpe$&%`r2G_`e&Guvy|xt~eP
zHhp|La3#gUn<p{eYVk2^Z>2-xp-TGC!I86pFSh58Z5K;GJ3oJLI?d!0H8t=(xcr#u
z+Z|?LM@tgq80#ZF{`Zjjtdxg8=kwc<eH>pgKD~e;^{Gx26iiOADN?%KMqh2ia^*UV
zt56W;M3BYf4TO@K&?g*?{V|3N?q;cp9cg^s_MDK2u;V^&#$Q!dy!oIZ3zL_Bb&8zD
z)$FM)z<q~=-$&8?Cz0E}1a|L%{piB&Y5=xsRT|P-Q+E>cg*jADpgfnd@AutL{I6Jl
zn^)oqdj*s%puL$l2uMu2H6??ywpegLm&UyXlmAL3lat$0>7p_ICri+a2Xx9yNluQA
zj<#yP=<Nx?M)N9ZY-|LrAi2rNwq{gxZaAlxmz!<3dXtm!a)00w5`tH{Ti<8rcW`j-
zkl+Q&+A`1$?fzm5){g}A>fzyWABg@!7r!PX2mfxfvTM&Sy~(8GP$Qp6KxgXjq^;_|
z*!lD2YE@KxPmwb1PMCK)^ntBLpDu@vB}GHw5u<m&Z2=GqwvUEh4_JD6xXF5Hw(r^x
zikHY)*bWAy9#ai$9*^gslK{)npTYb`_Jnw!sSqNRqZn-nA{k^~$IF7&NGAK<Cj_)P
zC~ijKt%M_ejyu81PJ>^s>Jya$0aVI&NYcr2FW<mHXZd{UC9ABDhF%$~#(7YJB)zVy
zRs#eCq+e7UlC3!0lecBxHNkm`8s1Z_ng4|n?ClbE8-Ndv)^&6<W&3gZ*f#`ew$^dE
zKY!8RH*O^rg|$ftkFw@0Hk;;R$8?w4(QQnt!AEFJU=L<nX&Bpdli1u|CMMDNMIx<U
zBKx!Ux_p#A37wfow$qG*TWe&S6owPgOu)-H!C+bUD)gRdkw?H5x9up6$a~Z?lkEdQ
zysjH9qXQcdn;mIqoG<CDjKdG%{0D+q_Vmhm=<DtAg!o^!?QK$GxRtFtZB042p>#3+
zWJmz*M$GG4=PoOM*g*3|Q1Lh^!PYCN>tdAWj5AGJ*a-Cbc^reh#*j;~r_jd?B!dat
z(GeN}=lP@zV;J7o!7sBD6BmcYW!lYPS^y0V4SMi_KS>Urzu#d(>cxpm5w|fmo`3hR
z@#`lh5;V8{Ye9qJ4~qM^r#NBV4Zq{ZGuaZp*DLA#tW!ECGLY4|in`PHo~h!9PmS-Z
z+s(8U!@o76u?sPgtz3y`{2%TupBO{<Q;2YvmYzHG<*j98<zwaKul2tPd}X~YOsZ&+
z;{>4V6Oiu-*=tqgA$1Dma`v{CmN=*h!_)kidm`^uxdfYf?Cl_FWBh}x71!MECiXR}
z#MI>&?M6O>!Z3p8-ji^T$r*%2NoB+`ef?;&cquodVS8*zh-07Ub94y-9Lwzr?x`Hk
zu@<h77&l#mmI-+Vyf5zKC0~~2sWwifMQn@X#(LL(K=X>kLxjhz9O7#oZ!!9R9~Q5X
z!zkXP0Eh0eX~jxT+GrJ4Ba?7>_SU?28Znci+u{g1$@m+#R6LbeCoDV#9dsi`dRu=3
zg3RSWbc*~vK)k!5rYG{V4V}3XN6H9!u1p8~S+mzHp7{^rAeUkQnblx(1vDI70@X}F
zHT+{I$P>?5^A;iYIS4R=>tj1zpaA~*30y7ym+Pth{e4jTw@j<a2}cJG+E&hXFZnHR
z?N#{kgwf;iDy#L2j>a7=85i##Ep21ITY2!~XU~^274@j%aD4HAzJTkS@)|okf(+J9
zg73i=p=sFM05ORwx^qL-paj<M=+UrHh`7NUZ1bx8+f~k%PJXI@isLTZjXP>M48Z9&
zy?I*q+|j;$bNgz91OJDjV0096JoK{7<G`vT1QuZvQdz({@j)lW7%9aA^WSN7e!#;<
zRc(5be#kFp9}Sf{3Go)Zx=w@avi*VYD|(D7i$HrzKkr8|ovg~AMV1T$tx^#9UFNxg
zTf3Io#T|cZW=>~iJivRX6T`QWHw0#?VQ$9H1^7O#U!vfy9neI>vt*YWa0%CNz0JuC
zRO6=c&|HIsq4qH0AxmKJht5oPyHxqTLdARvx#Vr;R#0&tv>;#t0~%<6W6*coh<}0|
z5D)+cI}1TUZ|$>%vVsCgLBD=QPHryq&NbjYwg*~~HiAMLg-W2|C#(GZ<>e*lJyfCk
z{Gamp>uqpw@K=l3f1q6{s8mC%Qi_a#&_Gv2iatlFW5vVI$|%NgCNPC*OW)C3F_&Qc
zwsFKN?T1Y{e=OVcN6NOV^b_=im{?PTv>6m+a89e(AF)7?5K>ZM(5KPQGr8cIP}!7u
zdQH>|=kSMGG;vXy)}3=+iz(>SrXGu7m#AtODu&TLzM|(JTGVX=Jp7f3rdJrhzQ$D6
z6ap2Qp2&$0zPM7+o-@7aK}_Col&}I<GBE@4FTW&)VqI*_sRWdUtynpom2eXE4?@HT
zHldq584O6@$wF><q@WT4n7zzk^TIz*1%?fER=V9kyJL_WuBB2xHHexxegF2lTu!B!
z7in`?9lkbnE&Kgta9OZW1OBDm;WW(E465&F<ek{tK|-Dy^<*r1x@`N8l!2`qiad{%
z0vn;t+!H@2#x!jD%?f57p(O-7AdgZSfL&_Au>Hl<o&q@rO<B%b?p8tlk~mtGTi&9%
zqdLol%gZ6gz3uJL&pfj&B+u<kILP4Nq{~uhROpM1Wsi)E0O+4r=tEKZE(!{$08?{w
zeQ_`fA_A_iuBQ{1dU|>X2M26&L;_wOk<#G$aLN06935#DXf8NP`fHv)_-!SGM?}Fl
zR$@Uu3ipgL65c#-D%7{Fa=hEL({+kehI3W;<q5Hd0c_OnhuLBWp%gm_0KxC)iedtB
z6?Zx`6dBm7X72++m;lqS3flPoF`wA*&A_}O@PKwKXONb@<@Io!?Yo1@L8}cjEP&D?
zO7yHQ!s|SKai}onVyROf%37CL`gd*osSg4t(~QLnNTO`Cu?w)Mc2iJ43Wi-&85GSq
z)p&XR%~sfJ2X=F%h;YtugBdm|eh71?O;VG1+%ec@>DLxZxWx;PYiu~9oerbmVbenc
zC2^V%>6hKr+Ott}oTHaO@_|i-bTew0@dgg?5Y|OR836CZtQqIpAd(M*qq7m=5Gvdw
zKIY)NIkk<`(Muot5d(vUgh~YA?&Hw&BW`c_jZs0~^EJ=Am6y9U>k3tXn7G|tKQY99
zZ*(Myb<Y~|v?RmW!scCr!~c9PQli`hjsORB4V^I3(IEpgw6tc)y$pQR+ytl;b9}tb
zRX0j<awx#|w#)9&&h~by03;p4^+%8xwyU8iFXJww#WC=kT4AsS0z$^s{=e86)|e#+
z_7Fe#_;W*HB?~y)FAIvL52g^mqc!|0-yPv;FN8emV7?OGs*k8_(IBQ{Al$-ZCts(q
z7yAkDN2pNfNMiv}yHmGG>Go6Z&0f(lxc-;0oK<-k!Ag%r-t!A6PSAhv^YwtTU32Hu
zXf?)pw^h@G5Tm5-O3VP1w-T3EmS547qLDdGi#t9qo~||Dg5}@7TPm_MB$$+ju0yk8
zMK%H}LP?oU5>oFm^*EyISCZVh>};}kM5}mLiWRpG*|{tGIZ&iIsr7=Vo3d&i#A8R)
z4GNuzxhRd4?W~5>@7y6psX*kG$TKfYLO@AUHmA1rs+N)IBp1){C=&vucyC$)786Bh
zQwa}{=Dt<hoH=l@IkM+?qaHLU%1O;FJFG0pRTZ84P@&CU0b2`(#fb^-kASR9Y-Wp%
zKT;dY{oBhKq*%E`|CJDQt=Dp1nNi%G)v%^#WE5~YU&q$*fY=_-GXpLUI|-PGJzZ!5
zyZ|QXWMB)L2Z85uZ)b;D+W`%feX6xuaRY-WC`<3WP-#R$LINJaqq(v<^BRNRP+UGY
zHnQ1C5P0W1nq(yl#V7I+L=G+Z-q5o<jx+rB?DiDe%+}C7N2J?b^DbQPL%}-l=AnVO
z`8tf^Ur1>+mdbk%tQSGp7q1WE6X#<w;X4P%t_YTBki8BHyI-~@9qX95T55Ig75?cI
z1si{CDf}ZFy;bO=g=H2NzNjh>2nGWwm>jZS8ST0*$hovja$oJk!8YaaI)kSq-mCqn
znZr21M3#4@=B$TdVe;9`Y0@GjZA@+*w!sMf8KN%2mSPIveBLbNJrGe%(qWZpz{x;i
zherJx#X(W7X65wqEl$+_YzM((qTBh^Hs;$}@B;NDKiGfOd_HZ?Y`XH0lUL(sw|P)K
z^I`fXhvI2&UJ!9rzh%C?OV)bS?D|uG4H`kHSwJ2M|4*(~J<Z{_Z=yZ5AGGdG+BN9n
z_dx$QLO_&s0`m5xqcYk6hZC3!YSHE8@!zle{Wl5@E{SE;Lb})E0SMf{!h(c6@I_aG
zXk34PKgfkF<G>}1xC5&?P=9s_)V&8Cwc9K_z$J;`e=HcHv}R9%$~38Qtxz3qW?QS7
zG_NVmyMrgvBQ2)>^P_l1MI9T`z2i>|53JiS+^ulF#xd3&g~qGAH9eHw4WCe(RXOfT
zMs!<vKMXEe)<zXP8+F7HOjMyGY^n?AUrl~Br`l)w2zC_2O4w_X*4NR=_T<AqehbIe
zDy5NoK%}#AatZn!jJ7R4Q14c<TPf=fPI%YSg>PAwZgErJSH{k39t=YeNk#5Ip;nJ}
z&`hk;XynVeiuki8nK-rW(T5sC)O%UE_Eo~UDf~G<=6nk55QFgofwFNNfG)HGw_0oE
z%v2rTTdhR7W|7tR2N)^+K{Ok0M$=CBvQB*KZ5nY{{xQrTJXM!eTloE@X&h!a4Oa<P
zsw)Di$(&3gQP0B#DWq;fQlM##ojEr6Lkub+?L_%lks5$bmUl_colrh92(`WlKd+V*
zf<6h&PQNwzZly4A!BV-n>5=r|t9`V;v-Xe`(>NnR^>DbuyQ{1b=zOJPeH8pI?}oWU
z2kwqsmR;?y7mpj0Y`@r3UflA;G)F8MSpmW7*e`jZrAI;J0}wAFv78^&L4Nw-rE1Tt
z>!4BD*6ge@-~tLzxmT_7)q{&%qI`5XQ#@jPwOC`3;kt{vmOv&IeSm)#B=FpkEs+KX
zY8lYd(!u}@nuG)d%d@k_Y<NgWNFjgzTwYz_;^w`j3a%e~#y|`J*0uk&>QMa={@n@#
z+u}HfQ;OyQxvj7M)i)#`H)GA?rMK#Tof^CE_03k#8C`&gj@za*Poc1eCpdWBJM{j1
z$s{x^Vt=vV^*rAX|LzlX`yNCs3UA9O<F)4PWZt$CxvQK=TMH9w{(zZNWu|)1!N{{I
zuT{>VhDT%vLoD(-Zy-rgoU+`8F)O-H=7{|>6-}8nghTB_2<q?h^u7_tcDKYkX6`Ar
zldR^ptGjK8L<)pf0jj+%K2HqJO&qiFDnTypD-BYHAH;FZRL7W*)Z-0NgJ7FaCl+WI
zc|TFb>idWzs>z2uM*paky)(6Co>Dc%UjdzJU6p2CbyCX0&qq)04Z+o{w?v*4c0uoi
zxo(<*d``|Yk>G@Gn7*sw{pf>xAsAfy|H%ZY^HJ<$`9kdCjG>Ohe{9^E?bYn|CS5EQ
z`4TeMZFu&|HQ=P6Qnb@`$52R{TNn5QpzolpB|M`eFk`-jR(ky$nOm$-q`9U3nScz-
z#5L?{@o(j&8y67{KO55A|CPHUZG^xZ!hl36EV-JX-&au@XV9HCx3`GM5733zjQu`#
z6XEqjeNKJ|fA^FkM)%QY2Jie0r*Y-hI=tphv+pW#t(an*XWzGty<i5zksAY`5VPmw
zSG@k9vj+wjA}+gk0EpW?BL+hDvw~hPA|RSzeF3_qd*Pmgo`?lSMIhMc3IWkQB7M-^
z42(Bu<QN0VUoF6-H?hO>>DKdlLIMm+YdWqwasU3&f&NeA+-alcHEyY+$BK_LaOpkz
zS=OL1kl~&eS4aS!w3Z^15Ccx29?oq(ChM7-Yl|Q8+Cur=5cb)w-8|oytiG?`S908T
zvqw(w{48>6bF*u6x9T9-xpxdWZ|i>thS!gu9eTY5l$2sVl%0R?+8{^8o&yA}%pF97
z2>{7xH_w;w6H12qkeNEO!zDIw2ig6|etX&eWyv8Y8U47uC$e_ReS&AEi0{Rt-gu<c
zgLF4+6x(OJocrHyg7q30HK|bbgz&hz!vw+2_2!-xH@#a~gK(7j1s6jl=TZN`3q;Em
zHhMajgcEXoz%5Z&TI0`jbJTu^<S5g)%#^_htaFCzse0k0ebyJ49c|3)!I!HaKw#4q
zW-mdXEM&2>!y(y$_D%@VGyar>Yq+%5$uPBv#{&!45#|i^b4`e%N8Gw02I}#Z(az?!
zzogeew{5dRLpFK{mPRQ?T~Ss+eM@2`Ad1x7nEb<+zix-De9_YJTuxuBXdUY(eoJvW
z)JF~?=HyYs6Lc-_RQc|wY^4UNFSt=s=ubiMruOJJY%;nRojeV92)l5N)wV<_mYPtP
z=u$ER>lOQ7F1r1h4)?fsZz?qz7m6wP)=U6msptm}NDHA@C_J4!UZ0>*Rx4dC6aW`Q
zZ414tfBvF|rv)>|{)&4pn_EY%n0JpE=c6|dqJ^uV^dMN3KGn=d2)rhOY?-H%x;y#@
z5NLvG7eV9O25DNl;0=TqX8iJa^Pg={ighaj;N#;1?jA!H*VNI_`<W+cimQK?zxS~r
zHR^IxY~fo4BlBNjQ$Vg8R!Gm*#b1$~2)d+;Rp%!B#To%ZDp$(`l+)-^PP{t{L5phM
zi&HeW7)-!;$3%8m8W|l)+Csx?*?rNp*7IV?mN4RN`*RhbTd7HiNn`!P91VN!)o4Z-
z?dxqi1IH2pP;R)(Y5eM_fyO%Ps8j&)7#MDzKPN`+!dsjcIAd&{pHxttxPNKYcSf0e
zHn~D`mxu8d?QS!>+@Bbj_{_cK(>XY0<XHA<%@Y#9%GmLLH&?{m0c*5$uAQsSZ;xR#
z_CP(AZ5cy0yd=plphUaJ{!3?#b9@69VgN&03o2<TY^r>|c3FaSuE?)0i1qx>|CmAr
zKbOAMDO`y^sH5U-kf8>3=`sO8#+Zpd8r1qOas9&oq3J4ss`|d}C8d#&kZuqG=>}0k
z8tD>{?(Vt>5`uKMlG5E>(%s!Yq`Tw0{{HhhZyazOc<-EZ_g;JLb@qBH*mC?3b}W55
zNaB@=Xs}8BXOxw&y6v>|ebK!I;_Jlsd3JO-34@<KBK`7LrjD&l`P&f^sr##WwC(b<
zY569lSN87pxc`3BE+Ifc6wa2%p$W+*JkKNt_t5CIqFBSwkL2DS?(6{2`Um<7mpMVH
z$jZtJj*^L~f7JvEBkiM*>retgXch$O?HaZTBHv3$Ko)i^@P6RKo<0N!WpZ%=sC5LR
zO32J?^L@D8-L=dU5H@nL;>ysnhkh&I@2^D50PjFt4}R<@3l6nRQOmRuaR)c!#vMX^
zGde{dEHYP&d)W8W^Am<slln1M_&m-(oiPaAUzHWUTohe(G#{VZzYZ|~l6cwo&RCvV
zrKl0=2qfVUE=I-!etx~_oS*SID?j(50wJi_w<kxDAH+W?zGH^T*I&#>?&?hUj?Ok>
z?5EIg=lKKz&+}0px7geor@&U|?JVjaY1Uno#~v4B?9blMGQ1=Anp9Cq7QH1G*A=ZM
zA>FolsODJFFOWbM+PD=P>D}_)Vk3)AB2$fB5&QU@{_`_5wx`I2YVrP-o4$8q?i;JQ
zzrT9Q<hCSa?_<($b(P8ce0lpr$z8I4!J4N;eQ?{6*ekVb9R(4dv=?Lwb(Dy&H?Xuq
z4SYxcotB<Hue|&i(lH?+Asre*YYaT6pg1}Tdq??Pc)eVUO}P5{FmXm8ty?zp;k2#o
zLfaP?NKsqB6jX$7AMn7_oe>P6*1R0(3e#x5*7v97-RC>GFC>98c+43Zts_PBTV`Mr
zaz9f|EMJ!17yg6=fKCkGagXbPHQEpUIh_GH%zq->>Pq3>_Q4c}<cjUoc$u1N&qS_$
zv86u($5*|!F00P%Z}yX${*a04d#xLW-O)adfIj|keG(bNTg_Nl82(vD_mu<Y3`nH+
zi(bmNwq(WZcy8b)2?Sci6H!fn+iQ*Q$R%dfsf5JxPaG(SAiM)of!bo5P>IN*wL>5t
zUGTB`cg%q>a%;kj`GSeLx1|%}v`8n0TfeB7JJF$F1Pwct%&sUXfQ}_}L9VTN>0ilD
zb%G%kN$vw-rpd`k$Y)S(*1*}>*~jMr^6UIFoC>4Duswl?`kl~lPZ6Kn>SbHEz|)r)
zK@hhCvL*`JEO3GP`0!tjWgu;F$)uC-g4-W)i-QCNXm}aPV(L~$r_ApvpHcZ(Px4`f
zZ_?qIY-Lu$pY(da8+S`}zw(F9R+B<pBgt{_s9JJ~D_~#J^SSEMXTqunGrVIGPE&In
zGFPd@tx_@c94OP~yGgzR?I}*&e7lv}e^-G0s4)j70!6U`lLje7s~SwPMlR{g$!NF_
zD*>$uO37)j@V*#D5xAQ8Ta$@Dy)Y)FE#wG*n(ulMGs!C}tB3o(`Gv~!`t$RZ)C|WF
zdfMAJD2ah4LD{q@;`$H@fGW7@sj2(NyQ8D#vxp<JkN7n%d+I>n$jGIl;HqY=H8`(h
z*HUA<EC8*8&#g0QFRK4FRO*BGCsL%niLbRfLrhFfp?XAJRka!7Fv0sx3a!@(%eEl2
z*69kcVy!B1YEjP*)Xzg&q4DTWR~f96FIk?h9T%(9RJGV_*7p{sfakCAq=MW>|HwpO
zv6NYrSSmG2{!tvdb)I}Ys{vVTEZ@5;<<r$3n4<!W{wAAJ9#_Q^TqP69DRC)b!(>vS
zMQ$04E~1vT3u@XlfH8wtS(+$oHbt<_>KQhI1m;3Z?g6HyQ^10?OB5Au>pND>H}r-w
zmMFdjF~jJh_+$mD4)fPD`T0)spI)4%awRm0@DJR#{4FRrJaot&ZUC>-5xCxpS?+K;
zZ9ZsY0co_{u;CWqxcBBSx)b#s5$T)BScKZdztBjK@Sk=Fk2?}HXHp#(%kpj-fsd(^
zuW%AC9tvBbb}Jdx#-r8FH8r$3&o#cP2b`so&j=!59?SovrwU1=2pB@^e!h_j*JjY+
zA5OxXkeK*%+=VTh2(n(B6hk_=uMP{N$(+4*NInM85L(F39GcvJ22B$^f^w`TYoeR8
za};gAVTF2$c6@a7dJMD;PuDQfr<<{#?;(3*==Oa<Nr?*|DdaH^Rr@c|dY*^u5hROW
z!<t~CSL2G>+1m$YZq1h=W_0L5K|I`m^N!r#y0)=iYY<U7rP#H??-_8$VB*tgAu2!V
zrSM6U-wqy_OYd&w+=ko|ENi9ar&CO)-2@fWSc}T6{+9E5+B8m%%@FUAll}3T6%uFt
zT%{=<DMUsAqfq`3`q(>z-ZWCzh|K{F0NIV$*Srzc*m@IQ(@Td4Z3jQf?z<`x!CI@{
zoScdx@W{;}z*gH+Bdj$Bne6OqOeWD(%~IKRlkDQ{CtLSERg?gUG_0yK<CcB}C864@
z86QbOtqDswzpdx3??S0@%V?#4b^>rZc67H<{b86yJU-7M4@x%QS2LGnbmuldk%^B(
zsCq;Z9wP(~HLFbJp_l;GQn4{HzatK~Ic9o5Xx3Wt%s<P_D+t%LSMMi6=*DfM``f=c
zz-497?CFmmKc1YN$R@IeM;!i9RAI#b$5ukdqFzuR2J*jcZ*P~5rC)aFB`b9-h8kv$
z4rkRB5Reek@^qMd>`cv%9F~pTW6GZKa1GqS)WB?rODgZ~i-MN(Fg6|e9{gDqPmqcL
z5Gd{uY*yLst%E*QvHjL&RhoZKLhFy$>O<Bl3fS)y64@KBFmx5nJR(*aOV@}9DE|~)
zEW>a5WI{yVi7|pv3$w95Nfiw>LLiaBu=e2PMuKF`wr<O1Dh=Tg(+7Z;=H9xcRs*5+
z<DPAMrqvh+xW#}B#|O6t*5SS|YPo7!I`)H;laH}|rM!r;s~TV-<9~I09eY6^leOL$
znc4QTo!wf?_S!7iYq$q5RNT;Gj5ikxl~5L|d(!{5*v9Yecg7DdWdwg|lbO)H9qHxp
zKPF;oY?mU+f63y5KK1M1B|1Ak_rY@%<fl>=*2KH@U<y{fX3gwzd=MXB=yJ|TQS`yt
z-29iuIuvSV^}KAB8(kEj%a8w{W=C*=VhY#aM19{APeyL;3#b8l;1&cedoKm(9c2j_
z<>|Zs?dS=6I{y+&sefxdIQn#c+7gC_{<^?oMCSYb<5A?b$U(ET7R5vP)l{HB`Z#qg
zN*$RXHeeK!y$ptprZbj2z%&*itRrxaJqVK{kL`A{k^$>~4ZO$1=rT{SwZ77G9WiyY
z#8~z%e@I!Cy8L90p5XphcI%y$>pAI?V;(9%z1^vJG)Q>X<Xlh3cKfr20n$Ho3LFR{
z0RUPu<twGRoQkxf*~Za!qGM&aUIP~b?wuuU_NU*-VA#7{@9c3@Ku|`|Ti13jDT8)d
zH3!)*fIDaf<GS?vMh_7gkPldq>_gP^l`K&R4g9Qo`NEx*fidNW`HH=y7-YC(^nP%0
z_*qdk`6gd5Gn(?6Ziiw#?ORyJr)qqC;e^E3SL8Zh-hQVHRht6C(qv?LVFx?U?|)h=
zgsfRUf&<izl#JR|f0_M4#7y2H3mKxE+$}Wz%}W)5n226-uOVxme_{~GAAB8}!ph0P
z0f9kCAnHI`R`Bl+KlpeL!M^0Y4x*rz%LOZD^<pCU2ON~{X<2XaV@M!!w-emLJzQAR
zMtb@R0OVy@Cig1x0f|X2c?1Cc$P@I=6lL1Wrnrp3uQ`&txtG?0&N@Q-@MIGbq7A=2
z7En)mNyxTB_!`ekPBtd)t_?D&RAsf8B^puV@${kEz9{+r5+QX}s#VB)d{qS{hMGx0
zd_X6SWZM?P?S6v;-_WfmG__qm>0sY{6O(^A7--tUyr664%4P8yD0lWA0v8^*Yh+@_
z|HK{h0|pg%(>9U-Q6|y8c0^Xk)pk&Dw}Efm_r)BiG|M+y>2ZFW7Juha3d^BkR~DcB
zm89#oECt0LT*cgSX2y!$_)rDp2Bgtnye20jVupFhRqnKH(@L~)iF)dOzjjYQS>z<U
zczNf8_m0pE3BEnX<%K%K1-P;FD+D+U4qjlb)-r#2Z7rnyI~|&cok?Xjb6fDi>DO<l
zKY2Q@8TY-8QFdi9BH*{&y}g1@?R+r@<h%Ewdn5vXIf?10Aj#_QKI1s7H||;C1;MkP
z-|P|qFWpyAO2J<It=T}kRPcgt+n5DsD%n=}6BB};+UEO2V{ErbYLK)mBQx>Mt9Lic
z1w19WoL66a)}>!K;JdFpdx<$E8(ci9xDkm)zHoSs(jDd49C&|4%&<&z+Ndef;@)?%
zi*Fn9eEygQAnkgC;prPeUS7H@ui71AMw$8^YQ5QOYcCAML++5NEChoPd3E<Zq2Cy4
zZ|=<xl0k|xY$4|}aB3AQ^J`qaZzmr!a@Qp3#NqvB%;#NBJJU|Gg(6alFXtoMvN^R6
z_1>H`c-c9JGe`#IbbnRjsXh|%5TjkwI>2HLZZS|!xNA6gi2#TP^?7Q~a>*yEdIcT*
zi40x!{I#PgE&ow%Gl@KRUu1Qo;co5d{yk!9pq;m&jGdGVNT3+`Q2+IZeEv(+m{&Fv
z+ucZ=bXQ)#;zSu_D_lWL`AHbb1zv0b+>ab-#c{6so|P)CYX)XcV-DifgYCL(m}A%P
z^Bli2#UfHfpUM?3mi|dvkx2v7SDA_j<f7n58_>fmdtgef`G=lS+3cKK!d}8Q;wm?5
z_n!9Rpfd$IXnk>etX@{rR#n!sv9r`#fukxPe9-mY22;CtV*dxsx5RG3rB%We{@f;E
zf@d&P_ijLFG*DqCb!B<gu-s)@*L+4W<s!nyWTEMI;EKphSPP`5Farz9)aR2x)KJ^0
z1(KRegy|p+0`bUi&AXNC>=?G>c#Y=HA4&!mY@N#?{=Z3Ep=$i@eMPK)<jCBUuxy_M
zy^x`?q5W9q3y+Ew;%c7TDr%TML)IxNw9g2Eag)VIMZ1{jz4*$i-|*PKV!UZITBx>A
zQ&aobxQ1AySFc_*4~R=hl<2kGp7c^+qr9Cqhu{c4Yw$zeGr|&2lk{cjbBx!S(WwFg
z{9s{y;olxbUfIBVRIP8N?`fq=Zk($|AU|jwvn|gwK08BollOPgS(g4ez*<l<SHUlx
z(RfXXvGI*v`dsx~?x4P=!U(~%Zm5J~EVYFPVs<@~Je~W$uSUj2;a@CCiTd0oHW_v{
zS?841k*zuv4!ws1VON1{<S9gi6Nx`-K+om=Q1$l|@LYbs&F!XjNCaNzvlU_|FbgC8
z{yo6|dP!Q@Y_Iwd*;~E(wd#OIA7Y7s)yIGY_O-Cx8lZi%T(Y+hxpt|XokjZg!r}!$
zTFidVwHB$z1ppA`_j50nW3CTYJ413n3(4#fjio|qh16>X8nbonT%%TBxQ+p!+g15`
zk!tM`njk#lzJ%5(3KK6)v(7_YnPLIsOElbE&KLyv>*6Z;KjCJAX|xZ69>dXL$Ti7H
z(WGx!q+e}RDol}ptTR#g4-Z-w_ZSUcIoti7lY@&SaDIVCytL190#P;j;_7s+RCm*9
zTOM;b#D_AX_iu_CALA;A*S=$bYpf$^a%k%$2V13%59`_pV&is4HaDvSv;+~t0(AG$
zOzW!|A8q(1sWvuNms8PNW_^gr?J;E0UeE`k$qM$FW|v_1V=?;_#wU>mzEj4bi3teL
zX&4pA?|D@?FD?}tz37VghF{^{C)x}-N^l+r9*5B6fifuS%M$?_qJJ-BYqVd{OKxRt
zE#!V|U}Qw2awp8kr}<M!E<=bPO7<MDp!T37BOV7Hf&B2}_F4nf-Nv}vYvEl!=>#!M
z_YmxgGez00<QyTAu(;yWSy#p+x4aBc@72*gHD@1*M<_So32xdp#qzn=;zA(n;`J|n
zVFs732s{OU-HU<7X|PUSF#>BcwKyL0K;K4iW-O23w1FV?eJ=N*&_@9a=NQfQFcrjA
z4jNdma6VhD^J|~ONePwS?+T$asl~w~YUDyX{iYST)B(&S3(tg}OneoaR8~Jf=9U-K
zi_QMxS^U<Tb>%^-u3lYSuCOT+;6{85M3FI@VYvT!ekP@mmIik1UQ$l{sJ&O8+8uQ4
zk`#x+wzyJ3WrhWp7HQnrLI1ZOJ1Us;KpzSra#*7-HFRi*4rqfq^i3riH|Fx`qzO%s
z9&L1tT7ExH@?T)~!+or@6{!dhB>Vf>pAaVb%|p+-10R#wG2j#VMfDpvh&%T!hFl^{
zHx3=B1|0=rK^%@wNOZZVG$*FyT(UsGh~H#-z3~+nfV9`>PT7_C=_H6%Dke@7#%0!a
z6I$QO6bhh2NiKb8gADs+zb<>0{i#ftP5XXC#M&-b&xq<}b;LC#c@P}3eZeS&s~VEK
z-p<a`yT&mwY6NojcT6f*U|?cF!3TCU4q~{a7`j#PlgZS14C?6T&u%bGH$B`9&rObd
z0!b;UGTjEp<1@<VAy)BbM;YLc@^^NXmqJ_u{FcjMMHzQH`8+j`vt1Hx<gAWloo}rF
z4mFz{X^ou~Yf312HfQY{G6RT6A!wDc!iaeh@>(Kp_d#R~5a%@7{1kAP=Ej!9$IkpW
z8uSK{0L9j-&xYz7jf5I-?dVs_IpeASn8KtIeLF!c`!xQ*F4LCU+XNL8mqX-x;12~g
z4|&;>wH-I!9(1v2%U(K1mk5uR^aY)ucCMk!QbTMek*a1slim1Y8m;_$mnIy5h{DUY
z*jwkLtXk^$E+sq0D;><W1Yq<S61}G4fhDbMyw(u3Q7<f3h!N2<Yo^K_6#Fz~%_GRT
zqX)Jw@7$NDq66qsNS?+&gG2bAc-2nwhg~4Og?gcnyJOPb$CxY)MXWQ+76iX{11&W9
zi;LWJVx2_C(Qvwvs%I9`6PSTy-#JP<eAI4{OUM8y`!p511x+b+3^cDrV%2~(B<X^K
zF6bOvhf?eq<jJ_K`-f-AiL8q{1<(A{1879d&E2Arzjcg*^`Rq&`XA8Y9WD&Yuh_Ds
zG_Fy^hPw1KNKQ^i60oZBF-Ku*U<yHkz00Z{&MaWgNdk3&`M)TSbGnUqtxsmQkf%o_
z9}g}ndOr^S3<!JyK|##tBp`VlE}kpItxS4euY7u)6xTy^%qvgaW$5uuO~>h})51ul
zb-}f`w&8sVFA`D+gE9RQH)!bD3=cqvz~StRlWQ<1bq=RUTJLA@946?NUSDqUlpM2b
z8ZiGV5vA9ReyQ#kB@55Js@5_ki~?<B#!7Tq1qpDV%6gX1uuX>uIxN5ZnOUfIC1j%e
zZY<xc5G7P(^;g%M?>XkEnRtZI(9pa(Qt9#BVNVRdALf$KUAY^ANsV5edu<4-r!YR;
zy9>&9b-WB{f*YO6?0u88K_Ch?c62v4rb<1#Onb#5_ne};Ol?tSQ3(AsHwbkj!1Dk<
zs^GaNJ}Vl5+wHv+C>81{xRpWR1@#UAc&t{+u0az{LAA?mrucD5J@YD0qT_;}{u%V1
zzv}2p6}$^(UT^5ivm>B<6-Dv+5F!a}0yS&YT?Y$$Tih;|J{R|2Qm(h<A?Xw}jW*Ro
zE}IgRL0HD^G|xlCPA-_SIlv@2A8wbngP^5F+haKsH!2+P<f+{Ik|JUB(r5V@fufJW
z$yBRJ-gb_{;7P{d*-{H{BjIl+b}#SW`zOT0z4|`ae>5msJ&w=x^;`++;DO-4wkm2~
zrF|`+-wN0F-7h}P&b>Da&`p5(*Q5YPHbPRbq{d6{lU&BzUvEEuZ7!;B|AH3YHi${`
zh5B*_<>SXH@WwiJ-k5=yfnUDs1d=WozxwLPOH!eA1ON?DWb`MGIWI{1iqsZ&xij>t
zj`g;Pjo}qXoA*4LVKnKpTW9J4cmTDDuE^qG27Yj0PoGC_NrHNSSxb8#z_wj>%$&M*
zi2cox;B6*TC-Ci8EU9643}x3S)7`f$#i;Nkw%70=6zIGq6xKr1-i9K_WbwZ~KcNc3
z{<mZ(-5+wh%n|(TiQDq|<q_g)c3LvHS*gr*J`T=%vSfDo%c34*Bu=ckjI}%pAxxey
zzxIR8{zHSb79XpUf9x2!{GKo10P4p~$F)#*K7twSel-OC7P&Jdh!S4LQ^6QhH5*2q
zphp;Vt9?C2FYIHmG+%DXnCWq~%Qj8(`f-v{LRzGSP0Jx1>fGGdkm!e(IJ2B8*_t{6
zVWpl7GX8XfV+7M7h`oT+jG&>r7%w{HUP4qLjtmVsRx?vcZM+Fhr@dGo2|Xo10F;Eu
ziHWP4v!<06il6z7kALczGcw!)lxB)SHmj2ufP%ao!H=h9xokTqydM=euX?c~WkmZy
zY7e^PVDzTJE30Yb-KEenv-njxd?|15wS8yj8^Hd&96|E7izX5Y`}_Ju#jt}L0k0qL
zgg)KItpwF?TlqtmXQ2UIT}FH~42H-M-2Xi0%L5M;qN?=tNOt^_;!#tRtN7pwH=T%x
zxOKUJReT-<!c%Jq;1W$-qAX({3>rAtM3nmWn%7~!kvQg$XeAuy-9xjO`O~BW++y$n
z`}ymnIooeD4eS$Sxnz&5yw)N0#SZKn(86b#hIL5F+!PxTh68T|-r=_&5iF6jZmSj+
zGg0}w*@css(%sQh`xSBUgNxxAIH#-&-?1dEiQ-?<|7Oxu4Bvc!6?g6L*IINMbqH%U
zCs*t^AhB0Au*45{t_PzjiZ-W33TJdlql%MrBf_sU?-1gAfl$h;&Deh_*M@&mb@j8a
z&6&q6AF{oUubi#!+SzI*f?wN%Y8nZ}^<^mOpI_qb(veYm<}(Q3TWxVauf+Otz}MaH
zhrAt<-@Yb)SyPvX^>^pPUu)0@^JU>pgYpA;5oZj(&v_O*|JPA`_soTCPVb!W=zxvn
zv*oa(l)NwO)WHq2DenfDEQ9FNyA+g~$ya>+vq);NcIY-AUK+eZ6lDJD@+%s;8cF|*
zWy^r+%k`8w&uI%tQwsOSPbX#`lGg-;8;3_!;!tiHW&0oZe=OF!)rQ!WuWnIrkr1g#
zla&7_SN;Rq<6l*y_rY3Lu6+_qY4z<oy%G$;vcOvHeNuW)iK{;b`i{Dwuqb;7``R+z
z8gZr3_F`BG5S6)>8+!oheb6Req@w>9J&7ooaqR6ui1}7Ef_QGAy|Y5v0T;Vs%RuCx
zf_Geda|>}ao8XztYLI1ixNh2u>#>|SM3NI-U!lTDouTHyztk-7e))wbeXb~r+}gpu
z6&%j9)Wr7WMEl%}x0FFa25_!cXr7mnl+Od;m&N}*&z-xJGP|k=0_ch13=(~KoyIi6
zpxtrBJX7Sb!imJ`<lSjYgyILZj41s5i?Pir74j>d8(}2WD|~m}n2b+JC;RW%T&_QT
zSHwt3+g&cc(2VHnyn1j+2>w3XF^H5HF8@7ozhsKAE=dOi3vzp88vM>B@c&wZm}B8(
zog0UkDJ8MMiQC5xbsX%`2>b>tFdF{>`Yt;h5y-<`vFD)Cy6tL+pikdy>Wqj}5|eaK
z{Ha$^>FWZS)BVI9X?FHFa=J!zxo}H(4{Z%}_%Q#A?$K~!%6qQy)<UYbwJz~%3K3CH
zCx4nNzIE-=Ly{9Y#hAmADHSfpa0P(z!}<5jX8^|9djDs|CtWkJGS<r1?)8T#s8D#<
zlcpZvB-KzVX7$XkN9&8QU=cR=Ytkk=?;U2rl<%&m4J8*<$_N~tfkbZ`CK`)$VU!j?
zJ+Ek-U`S~1P=Q?3J8MeCZWE+$W3kY1@P9`J`Me5pQIP~7`K9N<V{Cj0`KB7nLa)g3
zL)~^P>`(V6RIuf_2HAXc79b*PM+4^<ZiKhlqt~PS9;s{<|LpKlJ-)x8ZK1qvVNKvG
z+>m^tFyG0*K3aHWtdJ1D^#MDb*f?|;pjYX8YSzbV)gDzl9o*JfBFK67_|Ks!pvf)b
zEM1!j4LAMo{b(@s`AI=<%kRQV#!LGM17D&IJb9>KKcQN8jMjkbc0oe#h0twVf+ZX!
zuQu(*r8c>xT+U8*?ulrIN~YzTIubH2OdPnS`nM}A*$A5frE*xJU0Flz&E_X;FmAKd
zAu@V@WN0u=uViR~@v}e~<hhU)3BBs;m6M-s`eT+uUsc;<F8kP^m4UX-fjeNF5eLL#
zfBkFPG9}M}6dU8bc<5_A&zod9g?trO3TszeMW`JC4r9~T_1$E=&P;gR%32z1<?0L^
zH?Lki04QVo=d7!5yf8PM1ui^WAUmmwC^Zms!=4WV838~EeD_nyJ$G6R1J$xdn-L!k
zan?G0a4#()0t+I#G~b^T2NH?>0E6E^Z>2_M<=5e&=QBT*c$=<l4EQfQ;jl{`E9Z3i
z($#dPkht^7h1)R0-r9wcvdi6yW|Go2(<8Sh%L<(H39iG5>W<&)&_|+RvaKSlol`T7
z34dSxos%wsg0=O~RwM)mvr+FKAs8`Wa3S`4<MAk>6S`4Bdg1UG0JebvUBBTBHaJ(|
z+q2Otk5a}ph%e!cD*4UYt`PtsFP+C>yclnxy&2xxYRln4w22IDXC#3ln&$|cnw$-^
zC2R;8$b9oM7^xy5gAofdZkmjwT`>s4qB|m3RL?b@$0y&<;@Bh)tn-*Y)b`mM74BqI
zDD#MPHZ4$=l8lAF`0^IiomRs4srZCFrdS#27gkGy6H*jn#PUD{n|>){BO&cglhkt2
z9TseDM_P5`0!`ad-m#w2cXM&BgVHwR%i4rgQNFx0PXcJFgg?0l^fA?BQfHbkj4x*J
zxMf<9eq{Qb&<0*5ET1`UNcMOhMIZVH@+{+o3_c4+QowzQsR~v_@6v=mx}&e-LV+se
z=bt?%<lKhY#o;dkR$^(CY^O5nZzL$a5s8nX-U8%?Qt&=kUPnPSv)DY>yR?#@mtke^
zx_;{z$pbY}u}O?-!@*a_i5Uoi?z_Bj7{;aRz;xBN$&IXws>{8!Y|k{O0w@LJTnb-k
zK|jvW9NIsyqtSQb#8piP3i3oalnCTdqGsW8y%+U@k*!X-JU+oJ?fPqt2Txai&mkLO
zn7*g{jNC_CG;EogH(Vw!5L;`pkhfYbAIl@3l4FCMV>2Q54q4i-JdvGb(cMm`)>zAH
z<>qelrPfTP3C5f~<c0kFigelOFxuNi6s8rS7zx%`i>_F@ez$LaEf=1t@yp8#6ejOb
z)rijj#Lju<D0i~a{w5?bI~YOS-5mL{ZqAi=URz#|O}Z%ez7qfMZOpZ5{F;m+3lAR<
zFR*f{zG2ShAuD#>Qs_!z<ID*t6>L)dOJ1zp>FV(LlI*BlDgG^cI3W<m`L~{E@1ww}
zFO3;CYAg&XcknESjui$s-@oe+{(;f=!oAf}m?(Sl>f1d<YH6udy1KJ_-AP|TC^Y;=
zJs(@zO&njB@8Ucn<K=@rD`fCKyLg-)K6Au6R+>T}^3HfdIk|IrMaFsFq9t~wzmsUa
z$Wt2l5bKm>Rw!zhg>@C<`D}78r-c@z4XF@Eb^U}~M$Io*%Y@zUJuj0%@&ki);Tz{m
zR4In8nXTinsn3NQJyZNM$-vvX^vA+*KHj&fE!81mf49<P$gXbXS}7d^TT|e6!_>2S
z-Ze?nf9&j6cd%tyV|Z4aG2H8YC(8>6X;Y<T;@;n@bG{Y!f!xtX*$DHIiVL4WnY;8j
zJVLsVw+u$ML;2L73L^c}54zcf=V8Y7q0(`Vr<;6MjyGRm-Ur2Y8h1(&2!tI=t;0xw
zn<Alg-=>ts%hdsz=Xm~j`BDd)?H|>#x~BkCcPOOv>~JlJimu$wS0udEiukM)Q=r8u
z=;h*38zp-GftNw_rX{m%Rec(I{^n-w98lfUggz}5YH6W|DzgQ6^3DrC&3m;r7K|FW
zAoxi#WuzCin27nh#&G<Cb!I+AW+lVOH5(054&QznC26u;xV4EaW?8KM<wpr|iOuLZ
zm17hZV|Kz5s;fq>?#GA?_0)=9y&IilXLtO|rb{k)Rg>mNTnkHl$-lt3obJJy;eIU+
z;rM{8+kAYC_NRk=Ow}V_l+c^IFRgs+Znk;%r;64~TfI)35=rxesu#+<!Dm^bdHKK(
z4!SJUY$EsVxFdtB5IOf2th`yce9S3l?wg_;YTNd2sYB-!Ma?6Aup{zk%bR|?BD=S*
zSL)1A5JvG-V47y|Kqkn8_arE4eL%yQ=MZH{n$v1Y<IAoXM`|mz`G9=i6|`=c+=A6O
zD@O5`3gp*UEhcWO#B5M59>w&r<+C^s(qjUg7ygQpU8KOrpt~;qwH0j1k8N)7YqMDB
z5eVq6K1Jz9%z7S?QK!QnB!bdy{BLgz{cp}S2RV;4dd?Flz$H!J*$In8bCKBRX*pOy
zoDB?pJyx$OdDZ{q(SDtMB~x&`vfXpV%GN4}*V4XzyRzD*@uGfkjC*b8rM<P_%<0&`
zgq<F*@P1RRT8&Usjpyr+_xQujw>`;qSF?AK=Q_HXYIVLYg%ADmXPuW%4_;Dj0@FC|
z<^OF!B-ajTwoUtRusG<hdrrw=!Z)+dx40KNH2~c2{rapB<jv8PjQ9mg{OdmRGNjY7
zyut3{Zi415vB-WhWRW2TFYdD94Y-}M5Op&8XJyzi0(=>&%EG1u{|eyM)O<UhRj9j3
zf)%zXq~X|qwh(Be%%EBGU+LesW<VAmDs`>Y98L$_I*iHww-?Wv`k!Xj0ZKxHuu1m@
z2U0fF-n)L4hq#WO`^S*0z-`>e57NHBg7=*!m>frTt@Ua{_qvmX(~aXZa)_?)q*sfW
z={E0n0!q20#9~G2I=c9c*HT-ad*))L-5mU+auH@o*2!8QK-5f)UzwC{MER2~=?_;A
z8l?b0tJ*8?a@Fh4G>@MVsMd|_N}~Zlu%A)BqErZ;{flwowV}u#s;mcEMcFjN6PXYz
z0_SXKQchxIt?ynU!=h+`Wo=V1%`hL4dPe@TW^z*wpAGK7NY_P+sE3m%V9!6dds@7K
z!JXJuu*YE|IxZP?UZkHPuPUuEJb(duYP=s^`vX}y8D}jL;^!1@=O&MHtUr0|ohU}s
z0Df|>>pz?RK0p5|7JT&~;88C1<gZ(#BY%h0>0t-5dduF1$sTMg$6^Z@<J?*DZ7#8D
z_br@0p2=u@R>qcGPWM*V@Um_p+ij-~Zx|utUi@S`dGsq3m3`-BoELUP7|1|lCy^Oh
zxVM$xWAoOkwYr$E$!rh-^V~Gi*Ds`Rx^7+6c#dFq%{~f^u70P3e^ihMc1?gasTWyV
z3!fdD8N;vUW^E~u)F5RIhS4tRo*c@{M>le_yC=uiH_`&m>w}yg6x_H+)0+skriu&|
z9A4ddl9{m?3GF`K8A?MW0#D!&?V&8gx&g1nC5KPx@n@85T<2WAh{XdgpoRARaafn!
zPQ35b2R;YnuC?*`cTXos1?2+4Pr=!cFXV$XX9Zt)eyqbAj5h%EY6swjeY3cF8vi|Y
zoiIaYN7_to@ugQte$yT?gFF<78-{OrWuWJesekHz>rd#90QnX$K`|0)B!ngtb5`L6
zXv=a2MCzK<H+PqEl2K6D*zU7GBcElYD;lV>%}7inWor_DpJ~m#VcgwU=N#^>uAm%O
zH_3H&ddUeLTl7cg)s0H@%;m<=eU}&fqRY*z8|SUHn@gv}fA6DpuytH#<BcsdTI?!x
zg$5#gn>sG%Pt-wI4oCuL(P-Ltx0aDB!Dc7=aQW!zkCyQ?Lo*CS6(-m{cF1V8-`J&Q
zf=B53uBoI9+87_&6{$R5XPYVP)^4cCwnGfjrg&1%pz7a()Ke)mcW6>fd}Bz)IxQwX
z0;-|42@jhRCxTx{@Cz$GABBY-^2er~@(G~5$G~(V1avCwnY$GdYYS6&ZnU`v?VxKd
zUh&h-LoR%qr&$AkW?NNyxqGrUUW=o#2?oa67vg{au9Y%k#PWnpm`yT^$uSX6A3}L9
z^K5Bo#1$S;H#rMXt=9f7%%yU@1&1xs72$gN7~`PLWTfqZXxQ0MrRrO|I%`7LRlB0e
z$f5wg6C({NfP>&bH)3pj&Jfx+ofWUruouVM4GXD%wm*II_(c5Db7(Pw=ZJ6l@C5V8
z*pW}uqtw;N<ZvS9T=rAILl~ZoewumMRHLh_t212Nrn6L22QqNQlBvhz2s82Y?5wk6
z^l;sCXqmt9aIqTel^H*e<cW(cT!CA=NaqH*hJp6j)IT@Vf`9>$P-W@9c-YShx9-ws
zV&nMo46o?wh6ZE9h1xXxQq%Jfc=*o)Mn3jR$dMmM38yG|M_>J<XCOj$a%nDG2+3~Q
zlE3LfhKVpNTjaz2;HWH#@BN~oA;}duAAqOjFY^@+=xIO1I=wB%$;WiIrI3O$kRv$4
zx25|$=OY}@wf3m|Wu=2rqJy{P-rY8cdhT~=>krdjT{+IAq|)=aFUD<#dybv>SbdQ&
zV@7^5p2iNY#~3*F;b&ix>}kD6jx>Jh3!4fr2xm=%M+GnIy>4sRQ6|QH9)kMpA0S{q
z;S;((tV*XKv$dSjfsf2D+%7UTSqAxdd8N6zNvB6#FaSUCKJ2E0%2H}rrha-=gAYsq
z*<V0TQ3LjO<F`onzL1TBBoXgjQv$xo#Yl8JjW?^=A{(!yK<f10@Tsp!)#k?c<a>;g
z?%~$EO#gsU5l>c^<W-aT0g@9}YAfqUe-ozOU4jwFcJuqV#Fap@lVm72xb_UQx(3G^
zzu9%BE<ZLYd-FvB|NJXf(Ipgimk1Yn9|dHfGH#uhp4nxcqA|wgbEE$Kis@K<0>kIO
z)C!HS?|t94#livcD3c#LvD5O_oac5}d#`z8SWUNaLQuu1w(h+hyz}Gva<7O#;YT76
zzj{nl9oJM|qIw?tjCKtN1?(2@Oj&|*;$crPh0Tr7;l|$r5@she7@htfaa~2*LR|<o
zo%`bKu=4|ypJ)itJC|5)gcQ_cPUV^^7`it~NOY9^7NwH*$pJ<MMU=&#uW8q)QG&jR
z1xwkvy1qP^mVfZFYV74;HB|nXK`*$B$HJk~w;8vS_d7#_auT;5l%X;`V^?dL`t9B3
z$3Rthv#eMaurjYOgRShz5H3}7M?c6XT7>_whhb|!UZUPO9f|-%A|RbCH3%GLRpYE2
zL91jzkB@+0`x2PK;j*V?jox$bi--4%>iwci$0-_)(m3HuhzY14DXrIiP~vj`K;O3r
zHlGpuK<gtb76gW7{!H=I$wK)`0WC*Jzds>E9l6Dd&g3+=K`*KdAzNz3&sHOc&$X`g
zh}Drl4t)>L$YW{f#2!Z!XHlj8aI9!dUR)<20LPoDuWWTR*F=3Ar_1-`%z`^+EUKFv
zs6R4EOGmBpF_=j^Xq&gNCnjF6DQ#Bdwd%!o9d^WS=77ZCLgmBI2D2Jdumyw`v3Tu7
z=YR4>x?c}|P%SR%Yj9AAkl~>#`r($E_FOHTp;vaa`n_F<-)zKi&yQZN^WPu7UE$#y
zk|(|os@HTQI?##aO1Gfcd0BoJaw1qmO^?8gu11kTT<zH@WeVT^d%_*ZZkISjlkFD5
z*dK)(*fT+FfF66&l59Yf;-Mw8=DFSMIu`=azWp{KCOh82nWLC8!z(M(FuMkUY25y)
zQP+FnZTb$wC=p{=B9+B2&W|qiyV;T(UR7UB;6x7CUC8g`=KMT>km;PE2IISwFB+bW
znbo&`IX^q#F$sirWJ+2qJa_fw-bl$Y|Ks#R%0Wt~K9WPzzI`p~VWi(?qWMsJsiSjO
z0-?60x~#3AYMgTc??kQ=Q~&rJ#izFfM9jLS-90vPfM9;2yX~F>z&NPT?9vhCnQ#6F
zLI&P0mBRMLKRt~uTB;ml^t{bI3^uM(BEC?xRq(!%@NVjuu`NF+`rGQ^`#5Twj3;mg
z;;}qXkXs5lxAH0(haxp^b8_+w4YTx5=?+!O!ApAlGBLV#;Lv=Ber@E5v8}IU5>39w
zPef0M{+fntb9F`kX@JmduFr!!F*Xx7F)zW)9G@;ZCBQ{kCz2=9#Jqb;jC$?<G-yg6
z9N$LOP_=PQ*)KgcSHqy^JX5Y0T>Rz=GTYdXODkhp`6@m);&&M5vQ7VXtH^={SU8TP
zPa<)0Ew#7hSN%Q~R2RZhrFCMCX#foD$6@l$E)*V>nUyz5gb{aonp6?!C?XU_uq0FJ
zt#{2RoL<RrSMFUiZ2s*VyI-!$@OvwGva{!ZmsFgNV!R!Tr<dj{ArN4QI18o&6L~ZY
z<tXs}@l)@_;K{$#pK5+;J!+{8iyxMioiK%2yukeoAvCjIZ8CE2v(lGApIe(e=@%Pb
z3!Wt0AW@LpdV4+H><x09s%*?}f=~N#{f9W{?rn`mt@@a)UOzI_jvC$PZwHpRYDl}v
z$zzS~PG%IcA5@Dz4<0VqEh~Wot~`Rxk#_tBZkz8!+-xKiEb=62*PzG&1!|6kCFnrM
z!)RGDlX3XYpnZJ0iY!BLdi^5jw_Q?dlcRcqsx$X>er~}wa#7PLDLVt{o<md~mYpa?
zd>kTi5M0;SiuEdY)QR1fKUWlMUp_wGSO9$r%D!$F-y4wn;aAFYz+aa8B&%}IH4>;x
zu6uMU^~yNZ-30DIrC01ne~3?b<LPxbR3j6kKC^iW*HA^O<U&=KGH>X!%STpR;F{Fk
z@ymx7)UaRAe-5CTvSS<j{~VE4R)~SdV+PbU#oog#GJ}mrFqLE{+$&t6Xx`5eF)!;f
z-$ha&YDa~26=f&_UhW|gZb#$ul%%f^dvl%3OjAp;P$!z9Mqr`%m#Q-K0_{77M1JL&
zSX~%4W)XLbM3Ydxi-JgvyTkhJt*!^pQDH@s16GjmLwMkz83xl9$pD9)zUKAj>=pDN
zJvCk3@1E{|7e`t=f2c7D9=U=FqBgM1@+`Zkfpp(F^#qJH8U~JAYD}0=yB=4dHHWWV
znaeIh;LKXB{jrt`&ac=wrdc0ep<mNnv;<&b142$m8%UGN_vsoV1C5*Wqs@E``FYGA
z+Ba#3MPy_o?)PMr?HJ_?7zfghFB<$IJP+ZubObCmvC!)dUb&d8dD07u6bCeSxX2>)
z56ITw;evw792SeG)mYbUBEtfl-{@xZ_1Gz*M>)oi9j1gL(sikP{CiK6!Y<hwEYB9i
zl7!Q$>8t|gWi39UA?ZJdyOfcaF;KxIA5iQ^(IeWkZvwofS^sKjiu|Jo*ZRxTFx9AE
zYI1V-Lr?QsTZr=Tn5L~i@jvJhLor&w`4NP0gl|+zLObSeTT|@5%NXGioyZRA%q|x~
zu&4j~H-?Q9)-MSC?APRKGSWzJ*Vd?#rCdEp`KsuH!BzLcz8)@$f84IP;cm|q#P|{W
zn8Idg<N^D=&VM*@{I@hkn+@x~aXl1d7N-=1=pJqLEzpe#&*7teQN^e!3Q+WyAra%C
z^r_(JdxMq=q?^+yPooz%{QwO{n{@*R>$6miALuE+g$F~`<IQrve)Ug;(F<R#Rj#hI
z9G~Ue-mEtgH}rR0oaG5vo4VIU&be08RlD4M<C$q_;q$pCAE|Q0g<ErX0yQz8GFxB)
z5g2pJKolthy4n#20;~zS7Zk?pHh#;xF(MbOZaUz^t)kabRerO(P<(F2f{5^(q4RW8
z>SKAoGSpC5uI>?hH-f`uUC^QPnhO|%V@a050;aH(4rbOci-ptQv+q8N{Z|Cdz)t4n
ztjjWc0*o(VEaGJW$bQ7%Ht~<q;OI9#1NP;Dd@R3%!`<yZ_+EoarPaEF!(p31yQsn@
zHv$b0JtOj?!%Md{4B+)3%FN|YTa@S&!xZd)0gC#|YX~R5`17G9<A*s=@8s~=+)6r+
zUXgHi%8Gx|6hlaaLDb`6X$dN6np-z@z1m79%?Y7kw7|voCN5=k+xv8DIQn<Z*C`TN
zr4Ts3tYjNCaF};=6x;e#=kI$hO)ldiw+LlqE1615^%?Lo1LA0Sk4jl&6d5s5{+1n)
z-_J%k0YAcHZHtG<uAXBTGtRr@#pWgFuNHYeexWOXTe&VfDCSq1?P^_ds+H}7LH|!)
zWw0pq_HIZab~#%eL#hau_TlOvf7A^dx~zUUFg^<clFv@m0%~eU56+d9sq)`QS6ZgL
zlHga!rZ!Z~D{63r`_rC$P=W{yjh{%$&-2I*E{`ijaor!5UnRi%ruc7mC$kDZm@v;D
zWl0DP$#M{|Imsy|&^z`BSxeAAxl>3?9ETxf>`izyIOyg+hZ4i4&z1!O5&%K3$3ho&
z7Z8d{$|9~hRG=*8icLPnDh#S^2Gt$x(G<?Hr_-RX)J?-Vn^C$98dIDqGk2>e+xWLW
zW%eq@8F(`Il&1=fD^_&yeL?(Wk(aivqRSjHt)1*dPrW0<?sq2>*tT%wfHTA?)qJMx
z?r}mI{?*8Syh*T+{23--<#_$&Vz&V#0{jw<BQ9es&?3G@_UIq0pe*7;Auf+mai41%
zdkwJn`gUJBVcu#{mR@dS=M2|g+vM`G@snv`V+$x`fDGVZdzttcHuEh+#B@jtpC(8D
zAZA4dKcyy>rV**RWX;~)`IrLNy+Pt77);#aL}RniRM1Z6)O{d|l8`~_1L%?wpJTaO
zQ+=Tm5b<Q-a}#~57rT7c^9}OVHC`@$3%gMY+6~7P5d2=do7dU`YxEa@l!GTTvo+*t
z-j~_CqlG$&aXC<mdXe`V>bRCle;w>-Z8QBdT1#5qbRzXw;bkM?DjoE)tGtnQ(rnts
z`mph+B!j6wgidC*Z&uA11U-vutlA&j!vxib2huW1YrT#mv3Omu<{p-j_Yc9Ng1gnn
zTkaRo6iub`r#sl;od&*{m6~f^S<UL@<ImsZvF~4Rd3fjaAI|Gc3Nt8rxfoL?y{(<q
z(i60#gnu0XyAxg<!Cc|iae@^Wq%hjXba{>+9nxxf@E1M_0daJogNEseTQyz_)@=`-
z$ddzxfjm~5c-MnUJRy)#b{TJ8ugLm5p3cE;0A^=J&G^#QC8|k4k}uY*fKN1;`Mim<
zdx1YM8E8)-70C>+DY%&T9%2}#soc(I=8IHEX&|Kf*v9>S&iZ7|M=m@0_N)TO_0%g5
zea_p<*S{1plMVuT3;%s_y1@tio)+%`-kr}_7g52UL8)*2wv(H^!ld?{ObUWZx5|2b
z0Q)QFjvyi5xV$8Fa`x$YZJ2Hvl|nkDyx06MIXtzTkccbK7TU8JraL3_jVx#1z3TKI
z#6W$863@q7vM)sFJBvY(3SdrVGXP6*JMxJnmqdCTWM#wo)oV=ssA!jbb~bzm_k)pz
zs?YQ=zsulm=fmW@t5Lo~oTrh6<xSNU+s$V1Gz39!4?AuT#h{r5{JB~wxpaRacEtbw
zxD1CtUeU+;{f%s9COxC4j@n>t;b(o%<dCN`9au3D@%0-l-)4u}p1*>cvzy0j6r{Gd
zYxXK&P6na`s65TV%y6PwRMBM)?xxc`(qjZbMZv4n>i-zD`;FCS%%|0CzOcryw8iP}
zQpr8$;M@0+lNk3`*Fa{#vH<7$>FBQZ4~{?xbRYtO&(T9MBD(`P5ly6**cL<dp4ptS
z%TTEoua}OjR}n05^WQ&wqE+BbL4Pp|9=*mLl3n+id3Qc%5gSZpKK9{F3BFFsNM`yF
zrgGqaH7H>3Lr_Nf<xGq7qgmEmT}~hWw)P>W)_oZMFI2<*!0nC5)Q}Xo!qpykX=FfL
zXrH9VKlg7lB^-rkE?*CA2od!2GU(7+lmW&W`^+YVdkap7fw#n;FaUECO|a3wiZwVB
z$B=|gc4I}2c5Yfsv-!KO9~0u-FM0M7Aw!w>6GKJQGpqV#8IBZtZm;5C7$Ow?`$xyc
z1BnWgU%^v;yY5J54}el{XjzqGM2HWi`}*xa9NU>h)=|12P5CbLOWjWJ(CGoPPFoML
zWf$nRdXrd}iS>(@ATuu-gb;gfatgvf#EyvG6MDbp2KsJB->$rO5tbr3r*8`Wu@lP@
z1ig6Ortfb@ZqCpcqKg(yHZKE_&d^+rC@h6mmh7q<iIKTv93(O>Nd`l!9VWs=+x&Wc
zE?ax<EE}OFg!(Gyj{PRBgp`B7-EPhO;bGunWF4fVG|guK_2@IwQ=RWR{qN)WfVt{l
zYhic?BK|2+4WGE{9d5(3e>0M;h_IJ)DgI2L&I~*`3(j3<jvD^3)!N2>U&pS~^KFVt
zb}&5`v+VMS8Ru)B@}R2Yqke<^ca`u0r(#YxKW8M^oe|$t+pIJ-H=T4>*>^cVpG|+d
zr418n5qh-|@blispo60}WzyaQS3%$dE$%nk4W7`r$Ym|s`YQ9O1_Z!*2xd2g{l*O9
zXaGt;T3k%^i2<*GDKB11&SLlbb%y%YX<CT>Y5zCLftGCX#~BMjh&``AU@PMO`B4-D
z!7i=3qKtx3BjiTKHWi+|=D#ANa|hQL*7PiF0>oYMc6O1ov;JFU{n&^M@Dcs#=oj_2
z-|q=dc;?DCZ|MHFN(G22Q4?-9HT}@S?lU)e+K@)qV<8(l3H)#yb10myZ>zmi`V`so
z_hA~uYP~wU_~I}dn-E{c0U^}$Cqf2`qTo1`dXCRrd#mH3t1k+e<XN6he$N{b-;QM-
zg5RgQVGdas71gvh7Sb;PYvF}^SV?a-#`-y~h`;yx;$Ua3NMq~$_{FwS@-BDJ32S~A
z;V=SHe$xAri16H#L>rk{(x$}I=xXI5Jj%fxV>*SeKMVhcNPmd=zDGk`l(^!!pcnm0
z86{uSGT>s;*lTh5lX~UyyM1eik|Fe!m*C>^A6srbK&kKr>rfIrwnRt~#y=v08AFFu
z!OoQf<LmH&xISasAgX;UEz_;rlc0>U^GEH@h0runyOL7d-9_JK8ykt*Hf$1-QJ91N
z?I8r6kFn)`eXCpDfCs)sl|^AByD8&tF`=H4z_z?Si?@@nigfj|8Vl@(_pOsa0aBYf
zALj+l)g!|)UM;5P&)WAaoOnph{dYp+#dY5o@^kIZMMB;#AWONq5X)v3krOJ|6|o;#
z)ME0X0&5QC8^--MCa9b$QSYBY`Y$|O5!fMdWgBTNpZ9r!$|c2+%BR1?J;;Os!LVPx
z6_ee-{Z+A%Sdb<?k+11-%+Mleb^x&r2CD^0<r{(+2s{)exbiE(c->oY5NWr8rji7;
zC)5+|^yOzOC(eJUCnnb4B-`*L+D&G4_QJnzhd5=PixjAoAxwScRP7+BP?u>%alM$&
z@aM(u2Ae7&NqGa@TS7L>e{k33Q-{Q0>)q*yZ&(r3PVKA<K*(nXPpcnqbf>2EB#HbI
zFI~Wof*Txn)~$7gx0l>nkL6UnzLf&9UdL;mBR}h#v#-PuH|YFnl-`|u|2aAxJ}x2L
zR#dera&M+EF_f$~JFhnjt;N-dbhEz`q>oVv=}F!YwmIbVX@&kancs-XA=;q#rae=u
z+WU~GaYTt4Dh<XtpilY#5O)CQ>_lcZU)WbXt>j^z<4!F0ANla$v(#)8*!&=O-8knJ
z*)aoNO^Qddk@w77rbw1kYg88X92}1yrMV!YU0_?15ny8iBxtz>p*S)F`#)=E_(#w0
z<XF*w4Z5WX+VvE83-iz>ot1GFjC)Iq`+fZJ3i2Z_7)YAVbK1($6CQ|DCj%wVUL*0a
z_>MR3)~)6a^p*GeO+ReNp?#<-XLRvwd<kTDwC%c|g<$C54-1`%#Sd>I_>sW=Tu1Wm
zvFNmxBKg_r@(T0vV|A^a=)g|vNt;MO-^KDhmlR1X;8rZZ{}5G(MNAod*h18KBUHYk
zDJBC70Z@5q2lW}`Ppfv-FOhTzePoW;nchcvUQJqDJ|2c$O6yp%ICs(EhvPmt_DZH1
zpJVPJ{^f2dke?XR$6AD+ZlXY5dzcImB=NWU49P&mX<G%Q`epV&)x{8Vk57_gpXtbN
zmg;ZfTE@Q$dRHeJHx_S|5?o=6^0qq7tb<4Dd3f-E0=s-3P6Np(HpbIc^Ehu-e@Inq
zQZ8P&3H?<9lP2t`Eael~%ukZ+4rw3lBX~bV)T_(rhCPLIPl{!cV~^~jUP({s-?AK@
zf;Z6KjvxHS-b;&?oS)x#-1K!y4Gc-SMYa$1Fl}q-8885@2+yKD+kXEq4uS?m_d9v>
zM95g!k2`Iyq8FDxv9Y+xi}!9;zubCd+<9LmEBs3fZ+Ja^C7?==E4F!k&Dn7d-}tg8
zI~OpoJ<VI1iUl-{I+Hgm@q2ev;{%k8w~(&9Wm8p3Hgj9GXg%P4xZC5+*%dZVNT3k#
zW?tB9)TP4pBK+)h?h(5cyq?Y!5xUfJ*@JqkHfqC>Ew!-WPu##8PQmCGH?b6KVr@>;
za;*-cn8>rywbF%>15UB@IoQ-u#74(J=Fg_QmhLMq!CTVt3B-HYb$b#;C!|5CiBSLD
zRHBItw78yT_o4ATZ;E`JF71FLL;~V1vW`?4!DdR^F0N$N_Rk=&Qs&a^Bi5BjhyCsN
zWfuyZJ_Y64-glam`zIDz@>oGXx%LLT1bDz4Sx-<Tu5v_Q;815<2FGshbd<oZ?#`SQ
zdC-Y=HMQs#j4vc{r1Qp-<UC&GynHyw7^Bi2>J}|3k->pw5Qk>6eoxIL_ci;HlPpU9
zCUfr!^23!_k^n7c&-FOrqtJ!n*@=@fN`3z}F=ssfA5UK$mc{qPyEI5QNJ&a}HzFma
zba!`1$08^QC>_$$($Za0A`K$lB^@vQ5_kRn?!C|BAN#zldv?#6nRDhdpP5;~51WQ&
za3DsL>fc!m+Y=VtWG?)CE~Uyb!=Hl;RejNf{7cf&196{e&e2~%k?i|#E$q*gnAXQ~
zQapx2@BW?O$(+>NgP2b;Gf0SaHlIIkeulETdkxW8&k{}j?QYs$ECmfkugrs*${Obb
z3K>$8^OVAap}()CmKxvl$(5dG@4EY*yjjeRt(zm+RO#<(YrYwrGtf2!d+$Hkmz-M9
z;b*0_BWo{M^3Ar1bEZ_pZ!ryi|2{}u$@3(!cPgRxBAh$+HG2uGjGVk1RfJ>I9;fa(
zj*ZZq*$I(-{1zI-gN@9HvVwzR;u3;q`35`dSk-DgQAgKb|E%@eV}<3ge=>}dd|RfU
z#A2fREXaSwEH!!2f4P}oIWupSaj;2LME&@gBerCSxw3CNJ{m-p-Z+_%LG1XmiWI`n
zs7_C%$)gN3aT?aNP8-|y?ti+@Ra0;KGL#aNk;a}$`mDhob+-J17hwmCWIUZBj3B_^
z7>h6e8GpADTsxxN={ef*OtzQe2Jb?qqV_etuZ2>e*{2{pBBU{@n#@T!up0AXd6M5M
zQ0^Hf6rmRRl2vwpto~=IwM|%^^f~b<<BMP!hr9Wx1Mt!AaHDZ#%LvbG#D6GdsUG-l
z?qU-6p75k8b_MDTj3fwN4dvum;69vhi;RJebBm!?&Y4ML8x9)<j%*&w{!Sy76t-OK
zujcB$z2+oA_g&n?yy)*}wb?d|tuWTkPR=2yN_x`w!X!qgzCieJ4xZ}d_x7`fsI|l6
z`CL%`<-~&gv8UXs$M=;b0aH#K(*PYt;)m_Zn@qR#%!0H=oQ4(fc5*GJJz)EWi{jzj
z?_M?oYG`lZ<+ZTNn0-bBAYD)J8W->`Y9Q>`4r%&oT7i266SlaObdnvC5QVS%$A%g1
z*c)>Id8Ez2IeK~SZ(fI!2U6F3V2NpatzE(5@tC$bKr|9K$6q)7(|<xJ9X<1@2X>9W
zeR~*k$i)W@;4fH5Fp6n-;}UA$h{tX7=qCj|uUZN_Lw(Lv%-oV|Ppnp9vvOhO=%^tc
za5JnPu3FYVn+_FI-hyCg4b#S!Fg3NQ;YM4w{BJ(t??`bBA-5bj8R6}V!eUs&1AZdL
zN*ht9Bp2yQfa>Rx9@VRB0(>+ke0;BRAi7PiE&U&mmTgS+36(Wux&EX|CJhLV2D2CN
z8w9^vbUiY_3o1>_sqSr5w=rGKw3^P4(!1G-jnx&xMng0s8)h57Ki`h}j<T$jels1h
zQ)(Di@4xePKk{ny!Om-KKCMMrY*$IWy_--F2a2?v!=vk1ew~Oh&EO;ovO$>LmlhRu
zB6e^rK5%KTlIQxep*huBfgKa9nNgm^0?Knp`pHYBIevcHXf*r?L&!c^pfF|_;m2An
zG@v!&?kr<mw4O_791|r&eTsTA$3h3#k#)^2JhB@*6@AoVJT9V}L#vOG5fb=Y1gsX(
z>Bq4l>^E|4Tjrwd0A_wMn}@YH5wVptU7+sggSaXZtA4rRVr9OQE6YD&S=7Ln9L|nr
zMH}8+4Hk6ZUzn~}TdBt5+|+Jqlf<dS;RY)IYTP*ZW{Stm{bt=dZ&F3|)r*%j$5*jl
zA4+{lP*+{Wo+?ji*<;}6WLEl9O74x6=Kk(o$<8Uj^QfTKpj{_siqzCjyv=Z%1IJmB
zZ{5fb<j>>PC+D#tR^FtCYg%5*faBE!Bm}m_A2;!T{iyAS3G+PX35^bDNtkh=6?K-C
z(|eWh9x}G$Dxb|2yK-nw5TxJaNpq|*4K|@Z*LEZ$I4OT@i;Ey;><{eetkCoR+T-k|
z0ep@3aLuB=lU+P60HmMqpvd=u@HNlx{EWwK_{X?Pya^1^+ry9BO7O^K+P>hHF~`f7
zLvKQxivBrt(ncSKRg$BxGyf*{u+}n_2ksZrt8%Do4k;YU^9`d7wB)^gX20FDu@qau
z`lYEvm1nt}*Fb^+NMcYHOzJhA>NvyyrkI+!7fFIzMF4w}_!<yLeD@K=ZjSR@VegAp
zsPLw_|6>oQPhP7X`>w>#LG8b>aLa6)Z`ZHNO(4m%X+A}OvU3~s!0r2oVD!~_0z+>1
z?qx)GEtlR1&Hf#V(ZMzyIR)Z(^K=!myh%7E1HdpyS6cd#<w{(-ldtK)@!al?u-ybJ
z?8ELlw8XgGPLkH0SM%UKw!$aig9a1}E5qZ-sVxyV$i8fQIru{<t=lEA?rEmr7<0pv
zCRO(|LKW5C?PBZGa-eI+B9>iEeoumZ2u3t9B6BsY<5_jcj{q%RpiwFZiy8qUkU_}l
zM3doDPGWQRf}3a`cZFZ^yY<5S2r@uq{~qYVnX&1gD<a`T0D&!@+c7x)X#|m`9tHf{
zslq{4;@434Y4gO<H?poCPc11EG9guZAS8?`8x_Vp=Z8Oo{!gu+RjM5^&+Jun7-O4j
zIYzy@{+)jC-EXysIc#)7Gdit*e5dq%L0Tuxfo)EP;D`o&#El|))2t;dd5o$d!>HSa
z4w%(@+Svj!ADxXoffkoaTwLL%=MahEpD7D6|IAb;oA7ByTyc}}<gtxa#Zm89!Uscu
zIo+PDiwBDt)e2E9JqitQ;uVQ{m&=s?4%6?Li83qH)<d)3%}0&o&%LBRb;g1JS0g{!
zN@RdKJ^yZk6p0_{__wC|NKcccYF=!UgA9r9zxqF3q2?zW#N@G6!^7%H*UA2&NuceA
z%j1kE(=}zRNxDQL5fwB%k5BJM9#@CWQv1*VmRH>sd<2!}Z<z!JJ&smfMw?3bO&F>D
z$2ZMh7Z5+%c+EHN$a&h!Kx_K9fZ5mS+&A{~-F4uXniG~MP<8N3TJJqfZme-E<RZEe
z@bxJngAwpe;CdFz`@>TNcyv{~hkj}vOdg``A%&=QPsv|;bduHaasub`N;Kp)M?@oU
z7tp;Cd9oVlyzE+OHTz5xxln-i#G4jO@-_VbEsaCUb|1$`wC62zSq2fNFveurt48>s
zma+($yn@%pkWw2-bmqSEHN02_1VRu3!Fa)zPC7(j^Tko)Hlu_vm;Jb$F9ws$FGARa
zi4<fi0T-4Os=w!d4>!^3M|WulH$B#G0JEzJZR6!G)_r!DP`|%~VLb{fddxxJ1BE|~
z3|p-o>G*s`u$7%c^88JV`=J{?o71m9m|9BYk}Lncw(pSk3l&G@#fNTCs=jqWT#ySJ
zK-oKmgG&Ok!QZAVFD7#uUc)rVHnF`vhXp79V{gZ*{8ar+Xi~991!xg;9~@l&4bmYp
zjVAe>ihEhh7+S@pHKK^2xr&BQf+fwdRa}a!*bb$gMTU-bcNpd(1DO_TS2v*sY#MSm
z4hTS7;IMxs_>!%!3mh6K$DzRnhCJnyG|jF3hMq0k{XqDt$j6>AClSg-IN|hE9p$>Z
zmO(k8bQPfsE}$kDpPjQUvz}G!4U8)`mrLDU$DN2=jc6c-#D@CpD=7JeHSPVAEC^?Q
z>U-U?IPX$Mft!Q;jbGJEw)&D+eK)GMan8GF1_kiIzv=*+BX#{n|IzU)G$9CJgy7im
z>b;;ju)?Z)7Z@=ZIn}wu*Yr%*=qm}dod-W}F6&!zrbk%qe_>zb&}(w{$%>ZZUM>Dp
zNc0v1R$~eS20{n~dE&WMuNNpM10SNhg@%a64ur(qZwIT;s)k5N9gr!p9QoH;4VMm>
z&V+1PL1fY$#0GxlyBj~p=;V^4{Lu^e>@UBMj)Zl!F;IkF7m3_qm&PS0{uw4LL?fH{
z@w}pNRy(QAm0dXS_r3xbQE*Ob0g-V&3jhTrEi_K0n>9#iU6nfwP8PN>957NXR;43$
z>IoZcFEdue?T;Jc$-1zc6VPl>X24_&;TsH4Rw~dB<F1R<{VB_gsYZvel3V<(h^VKA
z+)8@IrU!R|jdw$nsklSX+Q+RSCQ>yX2w;GE6q8u1Nn?)&6v1Zq;WO?Oa<M6JYWIs!
zC_DtsWb*5TBL97{<f_1>yS`>gs4li|Z7pF6DWmpwNfpN-4gh2zJ?S(}rMVl90ROQN
z*g{|Ygq)1^W9>iRD+&$PR{l6hN*;^nUXf=pAbm}Q@oL>&e3%?=I&pWi7Ny;nlHy>f
zgx<8Ncmx};9!&iOE|0Y!?y_gO1?2BL<&13>k%b`hG-|Q6S+}tJw_jB!vi4Le?*uL>
zUCj=vQy2*m8%LKo;{hQ&`F}vpm(Q3&ySrt=&4nq|0`HjPTRyTxP3<y3d6hVYI?X?v
z#O7e@qfCIWuiufg=j8LfI;c3`4h$jWRpI(+<>vPBDC+*o<<$r0geOfH8YYoy3HV%r
z7iLmO&@REXl#om!6s+4*{|Kes!FQOJMPaOCX6ktFM{J^_qc`&C<Ums4GAKqow5(V`
zhU&n=r9V3M2Gf-#QZp)S=CA>5^uxHV)Iaq}F@8tWfwr>GFB*wGrM*(Bg#^TGc(+mk
z5MKC__d(AHa9ca4dbCFhiF&(RRi#e7qT~{oj{+Q3C^pr<t+{hRu9(%ba&vo47yq%3
z(Ck1?<%J3pA>h^w8l9IzI{v?f)1GjJ)o>*w%`Ea|Vip?-_mO<IY;IQx;(_4%kBB{8
zA&Hc32|Ksqrh>Db61qXM@77<bLwg4#r6P*|l1zutl=yr)^fyj4ED(0u(Ug%F?d*k7
zje8|B9LYl{>#w)Oo)`;z4Rwwpbm68HRj_n0Y=iL_<m<e{e4?nGA-YB$vE6J#7bvAh
zBPRkH{FWDS7x&rol0y3i{W}(AMJ1KL&+&8#J*Nt-S-N!`ag{QRH>){QCzSj#Zl~#y
znjOr&5$Cy~N_ptLjL83y&~G4sOiskz^mm%La8TZqq?A~CljyvSrS<H(_$Dg}5-J>w
z3z;dc^gxheQ*9A`KD%n~hKJiZvz>1T`SIKJ^x-*Gip>PIvPRlOOp$*`K8zKdbQxWE
z!m%P2vKU63Dyakg>yLQHu>VcM3=uQr#lCAQXCI=`*nnsBJpX>``OWKF&$^M-%g+dZ
z5V<xzlh?}{j)I6xBz}*T`lN+m0w9{fp%bGB@;X8HhS`eg*ELTbz>W%w(;3Q}g1)2o
zKb2aq(a5CJ9J*$D?q1%rO>a-jbq7(~l&Ak@L5X$DN;*#vr0G^4j^7agaq`Sko~J55
z@lwsR-*hR>@BKiK4ycEZMOd3g-}|N#{@Y?=E2r)+hg^VJ6Wg>jkgk1Fv-Jlz)@)nB
zt53n_To@j-LUM!0+#5JF))%i!<D9%d4k0P=F^r)5WjcO8LWMriTe@=#%ATkRMPMI^
zhM&|;D34E5UCQB??`u=Z3aF9p^xv$IJ#VgSEdSryK0$?o=&wC}1L=_*AN3z&7itw$
z*Dhmw!p9s&z=M4nPNj}qx6B6|&diP=PkKPSMzOW!?~|7jO>PkWj(T|!1|!%h81nd?
zr<2&&GPO%<#{>2TN5vL2<}^~f)LzZIg<2&L-TioZ8a{dbNso7V`n(ypZnm|egPQG#
z@h>u@RJ2VU(gIw{N2GY_$A3%q+@O<Cz!fR74($<GzqaDnm>hI}DFj0-te;X?sxFK~
z_>#)_IE`JxgFR65P>&jA6I?4=L1(Npz2>`9P%?E8@RiqHcF>LUe4O~FE@rBYTh*V%
z{w19|t*V7h1{862HWOhnZMt>!O%jUzQv@KsZiBoXJ|$`h3z_*m73mS2ZQ!e*-=>D&
zui6hDFiDBqN=)rjyCpy&(XG*Or9Wz#MNs>jpB5q$hYxwnf1J)*+(gxevvL!i*mdOg
zumH(~xi^*)lWPa{_{e<>gMKKHdDv6=?bZiP7`HIdg<TV$&M>#@Z!gKEv<FR1%US>Q
z$7$9!tayH%p0n%;C~u0e`mK?zGukR?`X&ve#WXMx9DA(NVHi`y?{s%MkuK{|9T&$*
zDJL^n+jL{2+>fG#1hD||KZR=6Pp7$Bv!@3RTpL~zoG9Eq!2EdjmBJCqBHG@vrWcSW
zSiHX*wA(yI>#C)6@LbfVR*mNc6c(sbeKk@eh03sOgfM`<!31IyR^Z!kQe!&U#xPR%
zrHyg0hy3R}T2SYGX+W3Kb?n>ppe&me$g~#e@40H($x}t=rF9<dFDy7=iqBg11(aW(
za@t0!ox$sO%Zc4|((6il-f61jcWdTRWXl@oEIhU}jJ0zc;?MTlfM2qhqP?1)b(A(f
z+)JC<)z2!jop=#0n)KgnXZC_`bj_Za!!)WWsfFj)u#C{s;yjO?E=qr3C_k-gBm|xj
zyAkDaZ(lX4r>#ghUp}VZAK=u9fQ=fur4+eNu|CZv!gh55ERXRf6ibe^LDOcf#o}uX
z<n))i%;|pny{iKPJ!^{c(79o-qiXu}?|Q*^*g^FaSSFZm)U$J3w+d#F1nTDBU#6X2
zC7Eh9x1#1>`G~<OJJu3?3{PFyg6pR0TVCI^qgdV!-7d&%ET`ZLF?u(vy=_%a=<75r
z#Jn$sGjB&8wsREaUZW(~%(~-3ZztyQVW8Rq+KPGR{^z;5cnBo?7B-LV(_OO175EHQ
zDgLSsMVr39W&>4|aP{TTqnb~r`S22!f)3einbU8US~5k_qVf+L9(As~aQr*#pIL8w
zT8rpo%7^Dq^&7NLtrh;}eYkVlI^0r1?pthcCIL)3TbTva$gDqRJox>Q9apay;Z#n3
z+;=<fV2Y+F_BB!|#@l|bbW}RNm()WiGO48J;GYFg=Kz-=6tQoIk-1*wt2Vk03FGjX
zj>mV5S-g|tueKtyHl4NN226MIVFR0Vd2he4@;{Xj`{#jen8mVB)+q0ep)HGYv#%bS
z>_v`=>VM|pQkLYCV<q)4L=_i%>cpApf0Y!8aJ_B!^da#nA6s$a=3VlX7RxIH;CwE!
zilRb72W*o+4t_jeHQsVS$6K?#sFX$kJ61M8<Aogzq7wnqx*)%F{y%Go_S7dG0lx}}
zV1H73cZovj4!`KEwoiqHnCDk^X?3sVXU{G#IQNCchy|SwMSK?IYPR1-*0PiOePksR
ztF!rG4u5uK?{7W6J!^Lhopq(b-;;^f^aE%U0f|QLtUJwNRA~=V_Z!BQD~EeX04JJZ
z3LBRSkvoqa<;jtYHy#n}m@v$N#Z|v@5!}{|#X7L7LkiLA(oditoB7C2eP-3z+OPtf
zxep&_rBTbgnKCaXEQUJRg})6~qa*b&Kn)V3IONO>KK_)3uuZXrv0;c@7U93WFBuUF
z61Y_p-$ea*86OvMG5?Wk&zXtz#yYG*>$%An#^~S#^?j#ufAs{IEA#8WM1h~~ONGSz
zoqbhvcEV~?#D+qpY92Gm4t(%fUOn9ImQvP_@8);9)X+HiB=E!Tf_pMDc$-N1!E&~6
zXg5p5`wvMWHq#Hj17j}M7=p@|8bE%jS=+0ZsZ`JH0X>4~oBL?dDdm7x+0UO1$H+{j
z%8S!-N!mvsle2Lo55r9@9N@RjEC6raJ!{<+jpB)>tWUl5+u-T!ou)Z_eA?9ahfrXo
zv(-^OU*yEBv2Vfol=`?XwdiN|TwK6ud`2+B*an~^TRb;o*{7+EZ!C)oHi`k#U43q>
zZ?`OD=~mnBN5p}+6D}Wsd!#EG4xd%OZb-v;`1bDQP25%ai1MviXTYz`lQ(3ju~mwv
zL)(naQ5TG->Opf$_Lo6#ZBAOmt&`-CLdN#lfv=iROqxVAS;;z<y{0yTPu{dHhvyu6
zRY!>`jn1&3_*0nV`=Dimxs@x}kS8*;R_Fki=dVVTlu-K?e`-m^8F!_YR>T;YW0b~9
z7<2uSuA(d5K58X!ddiZ1paxO%1DHdGrF|T{EMi@$m>AP~l9Boq&1f^%d*{znHmFQ0
z`tcLn?xTAAK7-RlAqHhmC!LhyR}VES-y8dQ{xV&7XD-F(lFtT$fp~z`i%~9J26)h@
z?k1bgCR_IMv@PtlYYs2huas(`C>OSlvOyfk>3&F@+1E&tT}d%~wIQgRAz4|S@b*MS
z()nD}{b|pt?ND7XN4u~>ux0#Squ2#x*7imAy_!9|C96I17OYQS11HxW&T{i^e<BM-
z&LX7t&f|yIym!EFh`|Bv*gyjDU8jp8rV4#crQ%JG^OC=qyV`McE&H|cJx(68()sUK
z#;Z?YR}_~G!ow~#3$LvDG0lwruUOU1eT3i;KpA$|{FfuQBEql&a;n*XkvodJBu?C)
zJy0jV_c1LzA}!;+&%V)yPSsR<h2%_XKRTUtCP6|rzT>Qn%(Q0>O@p=_dmrsb4P!0^
zD3(}Q*XWfeO`13!cPTt)-zcX0dFwu>sHXW4JBIUp&$tY4QMUxXRa$OZ`KtxYjuCjs
z+4t%Q_<=VS_+#BCS3L=)f8+Cggu3u?O;8L-D*k+?B|*XsB!4L;fFZXHQ5lZgUwVHD
z0zgPiTyg};E$>IE)K<pBn!9D3s%n9*&r=&npO+!P?iBnkOpf%-KKzgnBf-WF^|!{u
zd|s2Z$Vkjjrs4fYT)4pZp#`&()y?tL-aXn2K}x`dH4-fZr*_7#dw=%xlrs|-Fx>I%
zodwk>t{mTraU4;>vLLY()VA<HE&-^OUmjxctF==_FopOFt1r60x@R3ae)>Bu3a2-s
z*M9rgtDjx#kndX$SDaVi-}bXKZD?dxUVpvroMuUIbH5<#l7;chs19+>?!gaaTb1Dk
zR0k$KhkWZ{bp4q8d9Aj6NFm)a-wB9-Wl3wBW6@YnwKK;z-c4xR-#lM<Gope<o69(q
zfX>!X==(bP{s1fU^>GDLn!_0U4-GOjQ1Y6d-`>Z}_^0H@DU(gs%^uE8Zs*d9>A80D
z99tDxMOd#8S=T*n{xSgL$LKKa0d8O4V2kl5Rjib(@Zv_?!`0rF|Mo$V{}2Ip6va0f
zFhWLE`Kw6IR*GT<gc|UD!IwQ8F@T71=yy#u$%}ximtefk?eF;m>E<7RQwd~^vox*;
zJ~s!u-euS@N#0w-2P<c@HQ@&xyMr~wuEUpO4Ro@nJH}b96eBjI6V|6Tj4y{K&;3RK
zu|!f=F7(XJTa6L=V}WynAlm@StKo@${=(O{&L)@roSPo3lVD-a?=QIZ^T(OQ%h4Uv
zn*ze<fL<Jj)VzBd-5Qr%`rkU?wM8ub;=Zd5>z1yU?qp`9CRpxyLrM@v?!=C5$-<#R
zV9WS%9<{yPPn<v;r(x@8QL*CPpWu`;{$1O8n*^ItipBEf%6%9RBH!Q1X?k<Kfdo-c
z;S!#_R)E_wxN0LH@7LlK*mJ|82t5gzl>TD}Vl&3wu&;UR9BB|r(UTFVYOJyH-fjDY
z*e4WDG|kASHuTEpAC&0+AX0vm7d!|0yT-Bhv8e$(m`GXihtnR@_~Pp^jM^TBX^OTo
z%w*uT>j`2PR#p=?kIG<V#w)F{LyGF>zBL0q!RL-3CPVH$CZU`@#*YRWQ)6j21(k0M
z;=ors8-;x53zXAgg1vKU5^A)CKn0ru9Ybl8kY{HC<~qV}NbWvT{O@SG`9ELJS}&bu
z#_BC0k!RgfLz~(4>5)oR&j(Gs{6}#)K3pywZ(_;c7Kdc#Kvx*dEBFEjpp)E*7y>7c
z;8%W~ruTYX+`d-!rB($PZ&lrY9LYoN+ABJk5ux19DAf8-os}=C_?3A6?J>)a7Z-ES
zzlPRzZ==K^0I|^=zup9DSRQ>mkOP}W5CC%bAN<<<Q-m3JZI<Ipj`V>HiO=fyl75VD
z?jZL$1-2u+s2N|mwaJh83Ym6MkWc&e_RYovNd`r+A0@j!%g>pQM_&s6e1Oe-W>FT8
z4ZHW(Hu&o<#B<)uMU0cS?e8oM^eXu(ot*^8lAl6z=2|QgJ{F4bz3FE#+Jxmr){%R4
z*;B<5H7NfOxi53d6z}BT1^7})>}VBO5xU~~+~IbyJj7Pw@Pqj1TW7LlP#|UP&aTIJ
zLr>@H(m}L63KPd0h8&tQ+jB{PTgccTh2SnS4$JawoOTvxexmPVdEwG0DZd~OJ5J0j
zx?v&lw@A^dVX1kW(Lac1Ha)A)g=LI8i^tS}kch8y-{7Hc21GzqVwkH!d5I}YFaPAp
zSnw{Z%)-~IAp)NDXQG(3pcPGKwXf1`at~7L)5GMn8iER8U5-u9foA;yKyR_2Tjp~1
zWUBp$o@!K4LNnRCV_1f!S6BxNa=4kE{|7p-^(3@4&EQ<|VnIF$@{?=&3PAv>8Ibf<
zk{!jszCrZ*bRRli0PBT!esjOR7QJjX^A67xxQO-o$YB5NC3tm=`Emtg)xG+0QAk3g
zY=m&)<5NPS5iHRl@wvFI25zOyXIt1UKYN@nRo6b9L_lRXzCRPQI1irP+-Vogn~{I1
z9!&j2jaTZiuGZS@=e*LSN@L%Q^XZ{$^`ndG{HJ)Y`Fz{R*aC4`@}^biG9v8z*^4=M
z;Cs^`gbBr-$%6?ti^3fEFQa+)8?GC28vk502RKP~x(4d$2?Q6-Xy^MN*a547ofJS+
z0wcsyO1xaCfI=XbeV>Sw;tt5+Zo&L(lzL2VC`?kkYky%h$}|A83PA<Ola9~d11-Jp
zTz$@`F9TwHTKBNB$NlZo@ig7^^sK43_OqBbrV-9XKy=U+P>dOuG8x`*=jMG@S3IVK
z7&0>8bekjji(fI_o$&3fUdZ3SM0chpRY6-P(m7dnf;Z`^lK;3zc?=V6Dsz*RH)N?T
z3sIoHYZ;#}<eHIrdiF*tn69})rGVc_tEk2OE4rc^j?4neN8PE!ZfK?U?B?sW^7nL2
zK>0U7(qAk0gz7zhKoQ4wV{#!7h$DH9m9Z*8Z$g<Eg6mjBZMx@++I2v8d*~b67f`z3
z{%u<HHlROba1DwLfrbMBi4}`vg#e9{=ldtuYSC2#QAk}soKW!BSVXvQZYG_o<*@*>
zV&s7I!#4>z5nB}-{lQ}Kc|=akkz>;zK}tyn-LS*g8CeBsJC}bO7qj-&2_+m-uA@Fx
z%OBLrGy`$@)HTieOP0Kih|=df*A=#CLNt6e#=ey73LKLxC?&7D`75sVB0S+YtH2p#
zbkj_96DRf=?Ze5_^{RCc0mi|EEbKozi8Hd}WT*$!0|)W?7#379a(JV0VsF?$ZpNQV
zcplszgQ;%PPcCPx>DS$lSYpAz04^YNIs*U15{s04n&#s@1+5|~a0eHQKawUx)u0_9
zSvBP*0g*`(ES`n89&__~CFr^+NlhG99(iOPxMwUeDDgo^9rW)wn@6faxI|o_%``G6
z*0J0$CgaZZ;s7~Uty^6i1AIa#H*H+WwMWmuPS*1z*0S#(j_eVN3<c+!EZB!)#ill(
zu%2SZMYuf%@Ooyau1N{dOAz{V?#2F&`YiZdhX$~-TH|Als8^S#EyDir{jGl6f}={k
zzis1mglTBtfQ;NKf|+^4-IReR{Hv~OPnh_0Y#o%$D97^iBk3)Oy&;3lBvl;bk~4+|
zuqHbMDZNPA+%ey6B%G>GSSeS7QFz%F4`#1KAKKTXEm@_ajJ3qcIQ(WmEp9QgG>R)#
zLjz8P^^Q>Di?mNNP{nVao|?LjQf?@{5GnTQd}ECq@=18D-IL$izNV`Sp15RQB*dnA
zDW4o)kOsS_#zBSrub=N1&wXiu@AX?ybc@@U+euMy?1|#KM9f+qo;3=3==V7%cXxNA
zBfWj}I0%z`O@`{ejxplEEyMhBv3YkgKWif8Z}=SsGfGZA_3fnr*N+VO;>`@B0v@9T
zK@vgB4{C#GAq*lkNM?v0t0-&q<ZwlfbcT)kFDK=~M>L==qY^_t`mTG=hFLQC<S#6i
zf3E3#u)20}S<8?$?Dg2uE_m_h+*hgh`Iu)=wosWW5J&y|di&sg4Uy8S?)}>_xh-q5
zf&W07|GN|R>BZ&y+k-+%Jn)PN<hyu53c!uq$y^QV0S0v8>UNP9=S5n%d<xy58>$0A
z-}&q9_*@5X2=d9tFJ_i2asQELWv`ua2s4<H%c25aM0Fua$LkPqt)(q(b_>FbYVcoW
zT;$;XlBHIBedehkt-_yiXXM(`!z6`^8Q$H4Suevm2Z6Ph{Byw9F-eA1g%N_sL}#ks
zv%IAIWa0<?y_r~PQdB`vuK?g42h5B#tFz=RAc{UaoL#a)0VGM`onE`{v<?hT79WOv
zX(>iu=lfK@YE)$F=QPqy$aHnX3+B2dDttDcZ$;T7XU@*nH7}L_81(kbsn6IKb??@8
z25dkBp_o^;#oi|H%r32;ix0xfXDHg^5tkq@HffLP*_nRt#>Z<)E*bbfNc9ORD-={a
z-mS)$Ou!FM@_1%nGz2BNwh}&HwKyeM{+i*$AHEXgC<G7fAQwfU0Z@(bB?$V@wqipk
zAWQirc95&!Mp5S^k@#*OMhyAcTvWU*7`LC6J&j#@zfd(;0Xfn4Q2nk<VP8`FPvCX@
z|6407hloa6Jn$pSXO*8>iMc=sp9D^8Dso7Vvkvejo>eo3w#Z%U?JHW~OXix~c8F(Y
zb9#UG9QDqAM1Xzx;78JX8|TYE5wA>_wXUPJ(^fJw#Kny9y|2tE-#_uiqIlvi^;?u~
zI&C?;5%>0)10bGpcUKYM(Nn}0n{X%2juN?E9>=O0NeX5@yHnq`93+rRSrbRTzX}UF
z*!P->XcsyAO$U4rD4}2T=xlbs9R-D!uB&cmV&#?PCH~SOJX0FQO`gJ%X7J-?(?$h$
z&W|8%0?>!w=$0lTN;~IdQ~)$u)2@L%a&)&(e9^a|0wfB??}92~XoBX0nwECkR~6TP
z>@JPOfHW@ql78+slNit6^n*64uWXFIQ{bS5eeNqg9+^>@RfMv_ujs@bXMg{t4t-!X
zvTj`(!THnL`kL?DM=OCHHH6RKXZE-&VkZ=d&4U0LSnW;QJL2}+UP2DKS~#RwcJkYp
z6&51QnO%mh22hPfC~r#gS=*)deo!#Xk%6#CFp@X;o4+lPj>;)mZSjAB3c~+@iexMM
zU`pUMzh~3*AJJ8wS@(}ipd<kcD=qNWu+5x9$aournMOdq!-W$0pVV&jFaBaLG%K*^
zR2)jkd5wSzVAc==<AQ&!F=J~fU3>uPgv?`H92#QOuD#q0w85&31Ms#56nb?A`(E%~
zS%4;8#*AHL@=+?_ifTp2OA_m@c|O4Z*O@nE?5z7GwX`Cl6*p3D;W)_?4rdnQ&c;GQ
zW+H>HNey+-Pj>HWWG~fzE}KMV`3|MD)iu5yfDd>P874UEFK^#GbPpI&Ed?C8=lyNo
z5g8*tH>hanA~ZCdeNHq4!id`0Hb=%1)YS=)x@d4SIO{G3%Zu!`V<S}6XixwlH5z7w
z_gRR}KFD)<e`sZ(qBs*VLs=#t1$iW&8$SVbTLKSU%==1a5Pb!sI~t~jI(%H@a}t72
zjVq;)V!J4IGQ(n|jT;|vBgFFmsGgwit{gcY|CPcBG%6@V2HQBmjY#bhN6aM8^8~rH
zNv@GUZT_~~Go%by0@${%m(A|LRNd`gb4zr~t!CCj;+V~oTOhog97wQg6VUasYqd=Y
zKYUP*5K|5EG?#3(lOnHnGpjVoEZautT2Y((^*PAGH_FE3rY(^}+}*Ox@rg&>3xHY3
ze}7~x{xW+=f`$})bLEO@?ujg#`R3tUr*|IQBNf*yE(#TiV(%mE`|;lQEa&LCr*+)Q
zEZd%KGPmQzj>T*MV1!dGoM#spSj4DwIb$$#v>M3-&pVLZIq*hN$0-b5;!ZP`0Tgx6
z=}0H}fO;x=SYNe-1-3XatQh~TS3DzCeEx4+naU8g_&mt1_z<W_**BISl_?R!+W&O7
z99H4d_-Nq3-xo5X{O7GN+b8NM8{nCc%=zERq-9rHfZ8PCE-W^~D0q)_0ei?}6J>Cn
zfqZ2X{iw9MUhf|GUj33VjeUHf4QsB!flY}|YH8kM?BKmh{rFaKI(r9)wobl8*Mnb!
z@6UYfQGL+g-fmht&EyrH^aPe1A0lV`%>IOuQYkVp8>iRB_)4R&r_%^TC3}~ZT8?UH
zHjaONEP^==5^d4%UUBMmwzAu=x_sgZ4H^G&@P>}mJ&<#r2$q+iXoPdxC4M*+prR&q
zF?Hx%cD9A0WPyVqkk!$iO64SR3Ey#$9r`hqbO&nRL@m|F2nMXe4Y6jne!&u7s!0V^
zP6zd`uDPSPC#?K-R2N1x^Njo}6k&V>z=n+b3+HNfmGT(KtCsK=wfD`y3yLW1c#6^i
zJLh+2FKaUj(r8@ljgzZ>suPv=8r5cqP8?Xkc%w9Xn~@$zUZuKpT0OVlwiWy2lUl+z
z{Q@k`ai~F+ENe)kl<|%a?USW;^hhpLTK|?P^&4LnJvR~q$@l%ZFTsg$w5Jx_frHqY
zv%&7};BE5@eqI{YGR*o%k?Er|O;<`jHRU7*Bh^`)xN|l%Y$Rk_dbMJLY(C)yDuigU
zZ}BG4YhgsNSWD-6Dr&OjVFI>2EOD8x1SjyOA^OkQK9v)&vh~Dd_Y9$H@&UG^Z{9b?
zr`={nE+i&nP@ahM`cAx*?766ARXoHsc64fWybbcQk69NR(zN<<Vzjucw7yb4<h%Q|
zWQbT?Z%0_q=THO(;d{~Ri}Ae8cU59svfj$o{)Y9_W*3hxriRCjuIv*@g-T%i6aWHa
z2lgDwi4Qi$sPIywpIO-qmAfa1(!mK$P>*S)vux`V;B{W&liNx#=GX&?%okaQlo^c-
zfkr9Em4ZR(;bkqnbqMvaPb-mq!>3s>I-;E0f5Sw#^O@}J=MNaJnY@YowYB}#wcGPm
z@~o@kEe!VhYB^iPxLmL50!k7&T)@<vhl4yS&dbeZm~gL8Dx-FNK+lf}BY|UuXzMD|
zlXQvSx&Ax3CTx!AAp$5-Xr#SbiM*vE{)0tndsjmh3?m2S)yD^LA3j~SKnv|(UF5i)
zQ;b5Stc|8M6i)9}u3q&$2Ly@Ae!X3Pe)R=doeP4RZ4GJ)?XLgbe64^*`yOO^hP8_S
zV&kA?cQ2@(B;8yz3jxr}7UI2u{fqr{*7DxZ8LOg}9ampD&Jfvq681WKKjseoxpg@{
zx5A(hfi>luNCwEHcsN=^=dH)%#Fi15Mok@EajXM>fE33u@BF<);<uyKc>e`RXkOwu
zVxKINg>XjT0=`CcPHDT~q){Iv0k<wWgZPTt=(7h`FUdiJNojL=y;GPtaSwzrK;7uY
zsa?PIjad>YCtoiKG^FCD<#jONyYUm4l&H0Mq~~4H!MORR3cwCFco^pP6vq*cZHT#a
zF_D%1`^IDk3nR<Nib56`uazMy2dZJ$Hlw!1@-Y8kIQdg=CGbZ-!AQr@C7|^D!Wf<V
zxp6=0Nqi97gp#Pkp_{-|@JmjikZ<&Hx$7%&eQYDID8HKI&#B_Hv}aDKHzU=(O*p2N
z(z3Q+FZJdE?z}-b*qKRn<@blQY`Z{Ns;t47+rQPA0qK*@4>sk*#W$sDQM2kMU5vrr
zmV4DqGs|qFY@o4ZYL3pwroxi-1!e8Ir?d+>+#e#_R#r0W?xT)Bbx*Bf@dBbKQp)tR
z!m`O!J-12kmV|U$84NF#k}Kw({`H8LNmoia)gZppo+5(r6;=n1279`%FTNeLoxaow
z=-MvSvT%W59og!Hk+GH6<@2UH=y56*y{~(jWB%zk7>W8%!G$vlQ`!2&q4O6&ovO$G
z;s#buJoOt3iR;jnr#0+RVu@O`qd_qP6bK=eU*HU|8X3|mXH%Lrm|L>AkZ5dWd7B9|
zdD*O&zrEpVgy4rMv8T3$Lpe)dNCz0pk3w<^ptSm}_u4<Q<NXDqJG%XB(^4v=I)x1-
zPr$3)@9R5i#ZxPe#(S@$a|W1Gp6-?vmFcMJVtby}Xmz!N!ISqeG}q?CwIAOGcC3GH
zTfbb1f1rp??5r04#c<i~NpQR1b$puKGN6mYpD$c6<m~s&EJU-KP9bDT-ep`UICuDW
zq!ip=1*AFVoLV#{G4VUSui8AHk<{)!J_d(HQ6AUiWWQB53M)4Vv;E5Xe9W$TJ37hi
zNX`(r&gNNfnpXDtp1`~SQ4P!*n|RXMX%R<Bq66A~N#)EUJ>($)#Ot`%az6w;pkj0X
zFu<tlRL1h^#d~4*;>JSanF6}1v=I(|x>rs{?#E$2>Z0VEz0{*&)69IWkG?OcZ4JPL
zO}LO%;PEZiWjq4HGvb)O=WC6+1yaqNK5F6ON}Q{nkqpfc-D)J~h|||9Em$Vy>`z)G
zFRgCZ*2|@FNs`<<zu=cI5F$e4<~)Gwwr=y&xYYS}ydK2sc#dG|eEbrESov6;Uleb9
ze469nav8x2=(O+qu|~V--}kTNvz>#nmE*XJ;bV;6!2vp(r_L*^6z6v$80)5sl#*{O
zVR2AQp76cHk1L<ZSeop)m9%<|;90|4F$`dMOk;Q~ids;rM&<Emjf}jRSuT)0uqFMH
z<p7FrP?1)68F$3~8o>&Y|G}n0ZIxE3#v;73XA_;{=-+oD6Xkcm83(QQN3bH^WMTJy
z!#q*zC(FLJwB#l0sonWU-^?2RQ}Q0^p;d)?9m*HzwC5-h2;wj7eu*yIJzilzB#CCI
z-nC5zg7BObTFASZcj|OOwyQlGdm33<wCkdQzn>?Qn89|?lx1+)g0r7Jb3QUF3tL<;
z;}Z#${r~eaooeF<a@G044KcVN)Pa+tnwJEc9lR%D2!Z0mHhg@d`%=Z?=|!!rtply3
zc0{V*R;tov1=50m?-zj!;dzq8GwOA|^Hd1Hm$crD*FiCcp2eT<-yvZEaSY_%rnMT^
z5$XO7nOqaR?WXl}{yL4VMwLfL**Ue0ZnBlWrtLyC*EisvxVwA8YN4e!;A^Aq2we^g
zC2dZBxZ+GLYlR!|_+3s-?qA5q)J`wJsEeoOmX>OL>Qa!TOp5LX^i$_#<^Cv|#&fhd
zKlO3A`^Cgm<`p-Bq6L^{Fa~e5MHKXy5}bZ?9Z7V%zdSnm+%<u`e(EcpEz>|CotV0C
z=-ukB4#G=5<$aTR#Mfd?4h$~0R-GvL_w`VFAJh}sE$>$Ha9Z>itbGT+N(PUwG>I7F
z&kb(_4hF5brKbE!=WgC4XASLeji#{4wRzjFIz{i_W0C+rUt{VLGkQem<2qZQqx2wB
zlZI@%3G?dbu&NFfyW<MfL#-Nqrezf#1x*ovAkg&$c#-@4s<oqDtKr#D?R`T2G<6-E
z`9qxMm!Fu#Qk@oINM57Q`U;po@TRdV@&Ly1&uP*50EmRvbBhEZ;GjTR`0<JiiN^3m
zK2ub_No^Y4`O*3tQDd3#cSIswnjCtj1wWs9=Q{nSiOOp;sd`Db=7u_0+*A#mC2oN<
zS|XzqLBR<K07ed<l=0gEEV(GkPF<DP2gjtqztS3bwiVd?-OWob@1#TG7qqrKnt_m>
zHLP3X=pm{#4@l2xx$)#I!&D7d-Gra2`evE%3p5#3nsI;AQ7=ExR}X=D0C{;CzbVSl
zApTJ-v{_VAq$qv8Dy1gy7AmE)58~?dHsUXJK`n7tjw8wzety2bnP{>s8!wyQo9(=b
z{|C4rBRKH*IEbTX{6KsM4GjPg<t;b6E)U;IFoTgGQcqkiB|Mk=pjRw)D9di8ew?g_
z%N#woM1^j(qsIQW!(eFnY4pTVqi-lgZBk}4$;>y>EWl2#%0LvQlmvme=$m}}c6uV>
zSlQlodLc;H-mix|9?`uo5)KW~MoHI3(G<Dp*`i;9I<0CD<((d3{<{V5DTk)*B?j&L
zj_<R(KG0d;0*J+f{=jc-{_i5u#12>F-@yX9wTS!Qh2113$_Wb@WOtz3RE_^#j2A?(
zfam`n_QDkTe?Rzee4vh={rAiNzh@@K@qf?h0JuXG&~ufldX7*4{G2j8*gz}SfB)OA
z&g{QIGja+Jv!AV=i~SpRT<f33Kv#94@}@xo{@=h}|BbU5PJHn1ZQpl*&wMwQ5TNmT
z2<ByRgx=+Ri4kbdUYeo4VC0NYOgWv@3SGCFxV1<QQU2t<!W4Q@HG|M~53BP@DTPK~
zTSU0cZhQ)?PB2D7PTa;4)xS4~HdMUyP0{bj_XTnjpC<76MMH}X7sVVYV>#5*<LRi6
z)&6~;-TDuJ8XMfK_W4z2Kn1w1<b6U*vB4QXOAflo$>c-q>Y(AEfatI_M_ZQ5fY)Qf
za25L~S&3%?zvs82eXFqgq9uD6GE4eI$xG4XZ`|6|?mDG<T~sBQpOYwpuSB42NgSSy
z;(z&ZLEYtg!yWJq^jdC7E7_P6;ZNuO5L}~|LeHjl1QO=7hP}<;aBG`8#GX7HUG;=7
zJbEaE8wsyTn*5t0ei>+PEzU++dM-Xhk{b+Od*=`B9=lzOAipU{jNSX7CS>r~)HrGf
zyO5k6-*WluLDOsz{HU^jugoiyWUly$2|4p!4*M(3zBg0fJNF8+=WJVkuhsgM2#x7c
z8z%^ZmqD^`l#nwUk49`jvg3|t(9A%Kl}{E=1e-i+Ar(rEIYu0zN~U_0oM$IIle|C5
zn`nClk8a(w3r7ecDzuY{5!I(XL;a>xc9k$V=W;bC-ko~&+g+rWqmU(k2K&85Z{@ok
zcM*d$vo?6>dM<${*u;B7aoqTtU(@{jlMm=KE`S{J>JtV&Bn?b>SgUW!Sq%<bS;hXO
zFvR*O$m?onY!ohe@%JbqFS@S8wc5z%zzd<^ob`ZX#8czov)F)r55hu}lKc!zZb|~$
z<1kkavwz?5b>amS37NtLl%CMa1mY`yRmE1~?NI9$nngSuH4CwjFG#2Q$0`INS|Qn(
z+~lVmi!W=Yz8d6v%>RUA1GDo_es>|WTyR1|3lb>yyNOvnp9m6og?vWE33X6McH&dW
zfZtIixtyq+9ujASo)90S4Ab?T!ZbtQ-PQZG!G);)!m{dT!0$;VDaQTF#C=n7EFk)f
zAlOrPm)Ur0xs%ed5q|UqOnvxUYTwRvdp*CqZ(I0zQcmGz<J6c<^eW=@pc>~U0E~o8
zsoOKoJa}Cu4qqx?TB=qG25>9#40<e|GfebbXorj5xn3E&3SNhvkLeVSst|StPX;zE
z?_x}f<HVP)sw?9y^C>5mt`3YZPnZS(C&;@M)0Z&kY;W7~-Qon$DzH7F(>$D2E1>Gf
z-yjPG3gFJn`s~*t^Oz9WR)3)p!KKs>A6U|VW4S$}?hR^KBZZ_27VR*i(^Ithe%p}F
zN(6lGkIG-_k4ZzW6Yfl_!Qnd{!!BiBh>RSS>0{YpVP3?}&QCMRjKZ>g2~NvOY5%qC
zWd0}F8jw8!!xAx?O$}3ShW{8u2wl&FeSY-OeE4FVZr-(x@kyVkrK}W^ceD^MK27@k
ztGEx%=>+XHdDFqA6MGo16XMzIgb>M}H2lB91pQY-boSu|cC;Y7;P;w<Fc&FCVvphz
z%db4)n5)I8;qOQc0pMj|Op)&SvK5AS^S~^)MJKC}Uh&Sx(M-L<QZsFP$df<5_)K$H
zy`g^E3qJSAXy0#!N#^vf2fNa=Y9e{>N7>@Qa!1kB)uv10l=_F5?;Np0t*)%EgB_LV
zAbf?G?WIvftmyJ)4AyGvl9ejtXz6VJg2TJ?Q|K*`t~XRGe*%(<JkK+hAh(vVb^;WW
z<KqIC{l1RQwd$8~j5}av^7y{e=u@wpmK&|Y^-B8D>+|NHwYr8Ti>Eo4+g?|DXO3zA
zs5C~PPsy6Rbautj8)aF-UU<&z$QuRW*&EKxtde<Q<<w+669U%@Oa+KeH$v5i>B+CK
zKkQ2FY`^C%)y50I6tM?qHLvDC6vBn!t>^S?l@Xba4MDyQ6%9*Umx(M--ULS-utoDt
z{^mIxKA+I12TqzZ1GXb_1c$NZBAHro0amQb56wXxvZ)jYXbvp~!_;AY``8+hyym5^
z>KYjQcKJDqUBh9(6@{R`ZMt(7jugP~6Z!tLNx?I@s-RQmms+{D(U(%7pW#HN%5)jp
z(<{@xb7W@f)bWyD6qXt;t|&>P>J?pvC>zOkvw4qYyYIep^t(?_!}VJ}TbCH!#fXWM
zb*W?1=;#>-KLvQFL&r@qB4TfMF9ik7Op;i#N)ms5$P}#weJXmWzOJ;InTIsH#t0s0
zxn}V+ib4HlAHprPT|&pftzxLVDy99~cz)QA+!&#wQ7pJJsRO9|PF#s>72O)XZxny%
zY_gZu&QzR3;1&tNMWGDwaSC;FVGCSytxKCRPXkKS+V~CoY0cggz16MQa4Hk?;OJUI
zeV&m=<K+RRh%&!Qfr({yZ;Twlx5HPKngwH@D<XAaPx)t8yOe+pweQ~6WOn+uyeM8J
z1eg@uj)`qEBwpK}b(p#=iL>1)4ZK!hQ3)iKgY08CE|M=%+40hss|zq2JMNt3@^|}X
zmy@JWQ9|4EJ&v*CCZ{V@Sp|cNWt}n~M?bi3s?BmMx5Qp+LrUYL`Bk!Tip9ge9n5gR
z;Kk#D#c7?MI2!2ClHKaqU*1hE`nDBu#0D2AXICVvBb$Y`s(6apHxP|3W+;Hm`%?1f
z<(gHPwN1;}jVj}t9jn{cm1}t}d4z=EyR4`o2tWaqqU?DB42HAsQ(J-;e45rfxKf0q
zB~IUXHFqz=zjm+R!#Zy1Q(HmLfK%kfM0L8&pzEbVkemD$THQiGcBI*pd#g?C$%@O>
z80E`$UCM)afvKvkhc=(?y>m@u1gQ>AYx{;z5>4W|9UY1n`%C>cj;r}k7B2kG_^ors
z%DipAGwYuZ35a5J?r8eQnEa6X?f!Mt{xlBNkXxCsU4n+Wb*WM(8ggLi7IJu+DomaL
z4=dO#AZ)TQx3OwC8HDxNNDWdtzpc(PO5j+~-aknkZq^z5p<;jRNLA^BARVJamiD-^
zP~$X*MywOYF>ZtVerYKll@x%Sr9OQL+huQvrB?FfL78Hzu-&1ricLfEGP3(Gx(-=;
zhV2Jai*NZvOdZj5kS_NKO>}1DMamWv1o?q)`YCdR$k~^D)EfG0Hf|&RhX`r3k$_ZL
z_AyO0Wg?hxCJ9=7St+05L>0K4RDf5>`szu*%$E1e!hoTcM)zT@muS76d^!O@ZJ2zu
zq}}XF?dZ3!7p?UZzA1)@@T}ZwrXxD5LdP{+$78|CIJjnbm%dTAfu`~h!q^)Z9Ke+z
ztL`pXFR?EDFnh(@uip3Y6DJCats-W*#-yTxl0IbVjFHx|(jK$Z6pDjK3G<JXjo<$1
zv=GU2;$0%6&?YMnEOGzpfO{BI;z+S>v};acN)Dap?a!>_!w82~C5X4?cqkS-jH@c$
z=VMCCby$fRy)63+`!dyg=Kk#xc;0kuk))D$hLB5sO#~fPP96g5_Qw=jQSID}vjeq*
z)q$1WClF`~Z2@wnug^zcC7pJ(^Y0Gi)F<%Q9wuy1U=sRwzQQ{h=~K#WN~&*=9vmg3
zpnCUxod35ZSplluirMAA*&Lp@x_PgXbb7?Bok1y6QMW@OXJ4wvxe1Ei|E^}zJqDKl
zqW{Ub=Y|fij`lGJVNciqN>`5+5txB15|L+3A2y&|Q%FaR2K?iGL78!_g|Ar-Tc<KU
zj-{0(l2KiH+v3q-c_sNr!}B0T9Y29i;o+-Cez)zUVur)nkJ;QtreD6%aC2qZeCecd
zKn9vWUhNx3pA&!y=43b#`4e`3QVQ-4@*OgsUysp`uoXW)yWQ4%*zr}aRy*Bvmh+gX
zm&!p#K*n<_6-2-jux*{*<CtaxJr&&T-rTl1UDPf(T$vm|@IR#cZ%3r5Y&?G>Jd-0X
zbkbm#lyF51;jh2ux=J4OA6Z-$RlYjO*jXrFhK)4*lm><N?!!MX51(g$3wBmI>&hn;
zOB@lBt@6E$A<tb)&#8DgyquSq&a(0OI(IOrl!UX>p{Ke8>9}ZqNAEmX;t6{Z<ozNO
zSM^1|2P3#BpPzU@?GeIqo|yTf_kK_O5c?)-Jm)hXDy|YAR21<Ov*7?KLu5_;57gSu
zmL-Zy4KO&5nfF>H;Q~Gu^yyGLAsc0<)!|U+f@6myg>7^T55`vjuM}3RZ&FIfDBPi}
z^g#q<y3Q6Q$LQ&{96pGmto%i%cJy8n(WDX$uAJV$o`i~3nT%QDP)A(vNa){Z4IJg)
z2H1bg|82PV|4k@AvN-4I^<LX?rzSpF@D~!qb~m3v;4?IX7JQSPb=*NVP}N7l2oqM7
zWU22AFx-H>`NIyX<iq`;+s!;pe%K*}M}R^t4;|1^%PLe}u$WFhcyLfXA*cwrwx*Y4
z6X>$?D)A7FF#P&J@k_%x2v*sLcyyqmoiXj$BC0=c_#2K*(Z0-u4XF)1i0<p8Z8Gat
z!KS*Txr6~Y>u^zY4Fs8%TY7E3OMSr&4D7MHBcC+3x&5U0n%z?SRxRarI>-yo>Fs;R
z5Idx#8Ckr)dBrgp(~|IL)qXSOSy9haJfz@Y6)K{iF1Ax_D`zC3IN_Krg?_J<%jH^q
z*g0#?4OwXy*SJ?nn3~$)dWOwj%J%d3690#_w~neR`u0Z;f}nt)bT^`;(hVvKND9)8
zba%r>5$W#k?(PPWF6jp8?uOq&zxUmD$9Q*)H{N*ruPvT)_Fi+%HRor}`B`%*`xb?j
zB#{>OH)A4%ya+mK!6abmvC`@2T5PiPaSuzu0R=Oyo{oT1GFW!U43*NcZxBB18;{9(
z?I+)f$LZ3_Ny+YB{5%ya)7{f3z7)A+60SS_&5fhi^--qX@$pcSFhf`5Zrzdv(mHKe
z6@5XLX6GU>p5XioU+3|Oy@6q($vZ~Ht=j8ZGv>w3ncKAhMwm$W!N<#&%z%i+r{fi7
zQi0T8<p{#VACBWC-Z%ff&N4q2sj7t5r<E?h^)t1%UE5r8ldaC1-iDDqvq<|8Z7dc#
z&pEukkC-<_r1b>Z_HZWMVfBmOFjk7_axkRxUg|T;y0260u3@WBUOqvPSpC(n>s$2-
z=Szt!f2StPAoXoz*_BT}#PIo5E$4BK-6w99302Oqw70L{Y&x4K)y>SwqQs`J>y_0_
zZqI4V9?bCRmN6Ig%>v&!)pB{4P%lrB(zZLTZZ;!3dJAnKf2k*W&oW?MSDuyIa0b1p
zv}^8ivR3+WdC{l)X;Te7BkUbt6OAA^lHlzxzXJM)1i~OU{j1>D<rf=7xaQ4_u_kp6
zL6%TG-Z!=OwAISI<<R68lOoyur0UlzK<46s3lU~_5oQ8_mLl}2)ZkO3iEg3l^ZGF_
zB}0rH_hq!f&bV1n?a_IRX9DL<4nc_(^uo7IDDsF4LS@csAsDa+SfD5V6WI8Qr$R9Z
zmv{i1g2$zxQqbKu$E6#95gDTYUP(IhiXpD~{n}@UcTUM#T7k+#18$n8fRnR6s{Ksh
zqYpEGR9H^L=gO&o%6%A<9KLMFxyUEkOP}bh&*7nogs-Qd`im|P9S1aDz#x3^ylz1A
zAl|T)xN&AyqX)Q|v-J+iA|IJfx4}i<n?@%|JXwvUffvw?v`5$Ks4<m&uh@W82v0)`
zF?QxVHn1U%w=CmwRygXr{NbIJ!cHa12UxI=!#FNA?j{zF{en(HO{?bB&Y0M1oVL6D
z@~;7ffAwEKv-FEWxs$zoEUMmsC(afW$Tr)=%Cjr^)AClhaA4Z1(RAdx#aJe}t7DZK
z$s7R&qv{ac3tCLMb@i=rjgdTz;$iN`<IPf@(O2QaK?e*x&n^|M!$#+p!}G&GT$hPb
z*d$)jN#uOeYrI3-ZaS6}a2DAuk6{lOOE~GEOsR4D3Nkh7R}&99<J3%GQ03_IR-C}V
zW%clB(a7vJydsI$uOVbx<VJPs7a|x5MwVr^2tTaHKXo|%jSD6~T<|XQ=@F??%*{+-
zkDdAco)-aB&3;+AZ_3UDOV7TP5SIFcOsCR<_xdU}-J1oI)6n5%du-nO9{;iERA?#9
zR?k=OJ8%EDc;hVr;HwyL#Ul#r0GOxQQ0&J%aa$&7m@rI#H{rt<c%_|j3ZanK;!PcQ
z>KsL`%MF;(K8f&4O3HIRE%S5Eo$4zTw65Zz5k-Xc&F&JfJckV$Qw_|>5jZ(m4z%_*
z?G#<jci}Ks%H<^`8iPJD73ZT6(rTAefm5|@;O!!mXjb9S&!I#Ke~!n4KYw8!A7Zua
zA3pH`MdJBOtJ?_1yK<pLKf)gs^X<QuEWJHrWGYz`x}N15oh2IBB^g;xH~c=jn$~Dw
zp)B{g38lwZ8EVLEQ;4R&tZ#V;J7P}PlAA{?W>q=ucNB~0+B%lCB0m#>zyY3_G(N*l
zx=jkv_2%)Li0-Hr3!v;!HcZu^KoM_#;y>tm_L}jp$}57zh(ms2NPSsu+KlGB(qxI#
z2df<Zk1{V<6hDyjY~gu(BR;I>Wcj)*=8FDu=&W~>)fWBsq-PWDdTH4Y_KH&C8S=uN
zeeFrbI|<e{7%b@BABei^*vPW>gKB&xr@a20l`ZMjt516zyF-4&&yG+(yPS0>hi19n
zSd75z-@)i(a2iLgw;z3TFp+SZ&@<oC+n8|NX3vWX2)H|Oq@2FHt12wE--LtP|8?TH
z7FVE9xLSqoo^DV{)ckYODdt5zQFFeyZC(eaEtjBwg8$XE!AD5DN@~j@R9v8}19hbA
zx2{8yOUb592#jnM(@IN9SvvPyQtrZhbkT&`mcxoj7LAC;E2pA?t*+ptj3_{1(g^0C
z+g`FrZo@}#-_q_$(r4%iX|SGq%VLq+(LgWkmfHaj;}GAn{&AxKPFP&SkeUzA#PM3A
z`{dyk`cqex(H8<at`eZamQF4w<w$q7k?GO*{@Ctr`xl~9MKwrutwWE?yT-z(`F$IR
zkQ<ca)1TULUwbjyyUvD0@M=;BIvPv;(3u<6xk6#6Y$7&6aL-5R8jsTL>ytSP?3LnB
z_L_=$%<R-wUxZqqbK?I;O^*16O_7h1OO}v2M&JYF$?a%x$xT&G>9p=_BbWCCrl+6d
zH=N3AGQ}#e%%PHt)PNJz<9IgIanoRQ_I(_K(1>Q4Um!O05aT;+VgFNSoN9>SS|Zua
zf^(mwKt+n;iS)ebpATIiePCDn3)G2=#@1WaX~niK$gGeG`DCyxd>l7I2MAR=<7Tj)
zz{ns>tJQS$MJDIkkHHx@tV;0U@7nyG!ox0%CG>StF!NJT0-{<LMjS?MUSWi`qx@#8
zRk1*dN?U7)5-ZCl;C$}k25eCKwBEj!t@KT-G6D?IZ=BShSxkHwRYg}#KvVbpb!7@Y
zHcZ49#05vq4EFM;Lg`|MPI^K13ijaq*`%I}+04)m$Ejk+la&50@%SOaTUS22FLUwq
zw!foRb#h4SUYm2yij529*yG9P&SEC_7Q$dy8>!>yaWN^r%fqPdlPg6=73NGF{3oI(
zi&JiAXMq`EZCpS|tNK7?d|?X_lnE<8`D1Xw$xMfiStC+#%>fs_>wBbzT#zz3ES#`m
zTYa}-r)DOsStj_mK36YTys&Z}4}b(=V8P-jEm>-cS%d8YjeTWnrSO1oE8-m04R%I+
zLx&OfCwGiw=?DgEtMd%G*7P93jztE%gL>ZI=e}|Jo72u`>t4CuWfX!0<<2sLmP!Jg
zV2ET%MLie@KOF}&%f;bBSqQX(UFs<e+4jrv4hs&U{>B{7DA*xXm&>gnE)^d|%(u;}
ztgok^C@Ys*>YqrsXZP@)K#E$BfvyEJ68F$IIAgc`V19l^{l_w~&iZGi@2S)PAz|H&
z3BQG1s?9R}EXal)>`Y@}Oy-H0jwP}^t98S`DQAWYg5Cv}xGtNq*Hy(E=dn(p>Ph*p
zz_d61So8;FQaxBmdH+~oYIyT-q-jQ1HI^YRk?LdU!4z?eS_N8V>J>;~{Lx!Y)A?VC
zC*0GUaPCW3mW7p4OT#lZIl;o`qCcgWpIYBs%tfyJn&&r$9!vM)dmzEyUR;oaP#EMi
zKsTjHwB=ddO~BaM$coMXI*@DnTsXLIvC82WNIsUbitZ`jwEf>K;`u>M{9Zx-)uM9*
zTxg(phPqt%@ov``j^7%P_rsjS-|Cnw6+8V9i6Nr9j;Dp%M(~z-z#Q^BANMPW1CqMC
z5FTZ1yS`rbLuVYQHLYaK=f6nY?2kG--QGM7SeZ<5{M?{yQ~6ljE@_f@@DzFC8y27m
z2ACv^b@IN%WOrg2E_}GzM%0i(5OPKF2;^aPXM3@v+)JImeS9ona(8jw@$?P_aatpQ
z>`{!jxgWXk4Md}I>z2&_8BJ5U6ylYjlpGiFJ?Z|UPz&b3uP&^OTOc+ylg2l(@#RiN
z8}Y2!p>)HM^i9Q@E<$tcgANarpA?RXY{E>Ftg<(5H4*7C32dhcY_95kqsU2$5g6=E
zjDMb4AIsQj91GzWz0m!I{|@j6gy~FPUBL;yT<st8(q9;mh?-H&o^k6kgNh%zLHkoO
z#1AyzmE?-kxLE}~hJpHrdS)1}t*pZ#<nnXn9_c<sfp=h~FJt};;_jWCU4Iv!9&!@p
zkdvbzCT`b1=yQ3BF<HG*Sk^T&I9|HKx=1RnJ1r}d?`V}ee-4A{MR6q^&l_6oLkt~C
z-mCH#3P%fv+v8{)h%cebQTQi&2}Wwsc_5Z>R81XO6d2MpEIa*EI>%3AdU0$#pNRX%
z00fZ~FIT<x<lG!TIoW;ty+KJ>J42epQmNFS4L44h+`HSn*fVQg_S@q}kycuj-3O}@
z$v02islzCNCzIMflZ$?VcMSiYVj)m*ouuq@L^Mn@c|P^#mJGt~K|Vt18*ADvf(Mp+
z!-LLxP0#o;9NwT9wJa&k)r}QDVzV8p<y&q;-op3DYQxuo*&w?`ekiv5L94=&@5-@c
zF|WE4YBEV9SE*;5-18pM@VU}YF}ZyO3EAF(@DHw%6;ydN?iwM4jgHR#K<Ou7Zn}lM
zn|49-s_~p0klLYAMZ+jbvE$2#Xq8ynN$GqY>st=2ywh>62UKA%2)megm#yxR)+6Ah
z+RlD;$77!0WPz=T9ezjUL36=ZRq#)!I<^Zcb_yyy)kBdT=QR?lO35rm#$)X&v}a^h
zj9?2XE0bsRgXn2{L$hUqNvT+krA2eAkCkV>?qEYZT2Byo`$4A8EOvZt7{!W}P(Q=X
zY3{f7DQ{T?`#rA%oLL1jr1x2&<|2UxRvy`tItgkIyR@@)QwH;SWw!#tg8+3G?WUp_
zr;CAcy}&n7KRz}u8?3aNW!=fdQ5tz%N-EIjJgdvQKU`tu`gTq!MCssn@pygU>BSJ_
zpcGGiwY_yP?i7s73Ah$F-7_s5#twIMTsIBr5&INZuYR9`JqXxt(EGz=zQ>lA&XHeH
zg9QZk7iN@l8_QTPjWUBBH-;TIKG90Rn%|*dw)$Q_*(khFO>NqJvhg=a6KMg3wy2M-
zkb~~&p1r!y<tE5lp_Go2Ca@$zg&g*E9;mN;tnSmmd?>?6j7KD+*Cr<(J!SlLn#^_Z
zA75*{#BY^ceOC&ks#T~5^4~Gfjn%Sx7O7vxe)4tzHL-2+(Y}nAxdu_UySeojT(Lm|
z1S<3-c|6&Xhx)qNW=Uy1Vxe%&JJ!RepY6{@xj(zPNT&?(jO%RKTtfJo7&@)h!c1nb
z=M<Ji^xyd>m!h7`ZJw4%MEaQqrk3xw!tD`}rt6XC-IpP-BC?aLU_(i0?PlczL4y<1
z&K+>DNdm1*rzJQ3``TI9BgB0JGA8pxe6g~)k1}So8}_g5VwJBG2(9{bY44l71b#@{
zC|cPir6)dCj>Co0&DDcR>@2pv+?X|H$oib#7)&N-1yC(nbM$|l<)pI}*;9^AQ;NpC
zK_qS9UmM!G>b-8aYT91!u-f=F{l}W?)1IsQ;>X~}sg2Z;S5`HuEt;<)3j-U9==-Li
z5)ZxTwMsRY)xsFVhTKV4m&}v(h+Rd=R0Y2Q_HSLQEOiecWhF2mXj((ziwz9M=l|dd
zSaaWhPOLC$s8x}|3jJR2djZ`~Ilr&)#$Fy1bV+37Om_z$maVDWaW@!!_?RNOcpREV
z&`Yz`2MSIp5$rZKi2Plx_1~oiD!}2=ZFh-xhoSk~LOE@?CfhcU%}<L0B6m8A>Plm?
zNM#C$=rC)`Ac-`kWD#wU=mEFtpQIo=ICpok{B{@#Fna{#=h#olk3IJjMla;07A7Lh
zvVyM9#kV>Yf{v==NoQ51hqMV<-8F*qW;uCwMyyTy^v=vHKYImY!a#JGl#t=Jbb|~#
z9~F+UE@f?+R%QQ90@VH)0U^>>Jj6GL4PDO1A%`0sC`J_u0W+IPdtWQpc=Ui|ZG%Tl
zplD}GNh|j?S3JMW%hiGI&+`qhC^<ys!6UD+U_k{jjbt2)X)Hf*$~rgEvNOI-Ds4`$
zZe~&Uz%2wif<~eeZlRL$Kb2QC{eiCFsf5r$nzsupXZK}YEXr>!*bxTeP;}qwoAKr}
z(Ev60B6)NfCISf=z{c7;ga6y$XMb@I3kwwyP(_prR;|3)v`CVpd(;{wabvQu^)z>C
zAA_@CIEm0;!S{{3b?WOpK6p-=M2b*pmpJTa8enYh?jph%qx|M88W;A(@W6AwfADt~
z>lJA^hr1Upc06>j3%mdG{X2a}{9CdA_Yc;`e{1^x{!8M)f9m)5{}%>B-2bm{3{tt@
z7y19|D+}yafU@eRwII^41E|Vn9l}H6rEgy`$(B@#&^<#nPBQxUpy!nLV#W;b6NC60
z(n`A5k%?bV-K;(7M<1}~FJZ5HA;%r)`tI+~8f<GJ^Sw2opfFrrOBy{3oKpgLllb~V
z|HPF2uG$7rDDTzt;M+YriDJre$R%S0(=i?fgW0=Zb`Dj(-o^O$ddt9pejgOFNgiXi
z@EGYsi%KGh>nHnuAAgXuKKrZP-UNM(UkSk;n_{$K#DBk=gK+<??`^PiQKvz@xrGJO
zzk+%i%M851BMD9fK(8V#SNdkN&;B1t;onQ@d8*RJ;a5U~jBA?!0~2nKzh4>@x|fv!
zMQsB;aNdXp#fEl=f)LYRLIyeO%D-1AEl)LP967^HW$KYFV3NQaLHmzv=l<~|e9hEE
z-<H1JKch5l$m{EQKU1eiGtWim^KJp*-#lDu18YmpY6%#Ey?xTGHqL~7>b`oR*xypI
za@@Ug{t0#6@17(v`2E9EC^+s#zXTR2wf`F!fj(!5=f7LP-S~ew?f*kl6g|P?ZyC$E
z1&YeljO^{}z9o`nL>$Uuw|>shG%R!ir>fRFb3B-V+YD?{fI-dn3Yqh~lv+S{|33CP
z^}T8==L~JTYMG0vMP;3}pb00pJ*yr>=E?0F38Ow^`rRj)gsk-+wQKu&FVvZMGdtdz
z268}8B3rQRJ}!+%C{g`Y=H#(++B}Kl#UV0o=ipYEajVav%hdlV&>s5BjCo2R1>H~A
z&rVNy8o~xlSXBh|EDzKB=YLMk77k@qXlv)VbyM+ItLkDCC!=}Z(GV8Rp@j)GyRI2i
z19q<F%Co=fj74)z>ofR!4MipbaXKZoAK9aiF!^U44>(wr?B=SW&*b<vtU_sCF%}1>
z0p4BwSA)HJcaXg9GZ>t&8?;<)sanfT@d#G&Ju7k_f%p3NeWmZ|Tgz5_hwqupQ9&*{
znm2_XW!=*!?++Y2@elVZ?2+@nSe_`c?=J_^g%#I|)bUF33HEGd<)N!q0!gT{)=Zn+
zvPM|sU*bJ%12q0z>hmt2zvBG~=l0~iK{s>VgL&sGlekf6`bsl6wJL&efK^HT_HdiZ
z(?FFQ^wSJ9Vu1n51rp9CqUl}6#Y<5uIG7DLd<a~9QhJrV9cfNg#pE@dujHpk$e|fP
zY`R9N^9f*G&esGUG4x9+z+mMvxAYAzu7=t=S1PwwUsGdf##j0WWQfemn_Rnpo1F4h
zQ|*k!Kt3t#Q9OL4iF^6XwekG4Z+-%9<_s$hm&j|4OMPtE^|IJ|r;4YrZ&+owqbP$?
z*q7bx<1KXZ9;5ojm#%MU-#i!j^k@vr5O?hb7sBiH^LrosD23?}P@>q<#G8C>dMEUa
z;;Y2l^2?R0Rev!XMN=Cd7b;jjlc={YdmS01q#R5ia4C~EdDT<!Bq6qa{mp-o3Hk-s
zMZzY>bxwC`4I1^;yKhir+NP3BN}GzSJR7+4`c}=$E~*sb_B5iM&f@=MD@>}--gLvk
zV_IIPp(NaK8l}f1D)`&oA+7P8jVhDWF@710(*`-=*2c)2ipV*nC-O-xn>^6$_6vzH
zzOvUroh^9x4)5(FIIx@sKihV#bB^l}C%=))YpCx~j(jbBdHEveRqX_6_i@`4QU4+(
zbbWxG74*QpgDY2QgSjTyKBK2>D|K=uYmRFc4RxkzUl|oD?YafS0v-0Ba)^&(Ws1}g
zWgDJt=e$OGxti1&;KO_ios6()Db&ntIIfA;eQcdpj7^t)>?`Yxypdo~es?262+O9v
z1}wCCp^~9hfzSHJ%+5)ZlUU0PE~;}KANF1%w@jrXus1auobBbju2R0iHZ8OtuyJv8
z>$dm+*lA!m689F7l`X;SRp;POW=6RB4q_b$Z}r>hU80%&CG_H(ds1PSdSsS1dSS1E
zIlaVn$Jpibvb`H?V(`kvqT8%(^=`e7V9exaQ5<XG=_l=rVAn&QA8XUpl;%;Wc_J&r
zfb3DVt6u#)--voR#mx!>S3}SCYuFvP)`R=r3DYV^6Q~fZx`2Kohd4lAh3NHM>sUAA
zDHSxs3i9t49O2L1t{P(5<C^M6eSj^wUwzu3d^h;}_sTK)+|SnzaYLFX#|ZmYTH!%t
zd0sQuA9R0nfDQE=yL8&nt_4D+ye)k5kKy+F;$$canCRszQq|{%t7W-`z?8!`-zppX
zfv%x(Ynk6u?d9K;9ydPBDI#Jm=Xp@a7z3Ga)O|=ujWK5&_rmTHkhvEy+`^WhZOS?^
zV5;<g8P5>dU+s`qzX1uH6a$gqgYgKe?+F^cYvCI!rtD3I@qZb`fFSaq^+~|f*jVk!
zLX-MR)@60I-#c~6gaT+;al}*|u;v|>FbWL@@sL$2Z@A=5quSf|VT7>u2T(u5Gqyi4
zP1_WC@+COIiZ0^Q;p4ol%^h(+co?h9Z7PxOgqWyR%S>-{B;tN`y9zp1(@nGgbWiHc
zw^^|dB=?{R&<Axm;eo)ya{|;)3Z&u?y6}#vFbHYk-TMgr>9uK|S&*oupz}DC&$*?H
zPF{(`^`XOz9t$XuyBk2v6{(kI=4F$ew?#v{NU(;FR7=8+CM|0i_ynIKhIC841{|_J
z@CZ0p+ElzUiZ~Y!IR7l0uXId~>fS+YVvnF)&}fag26i_{54is|aGC7)OFuU956!ji
ze1zbg><lkjiHTFV&;0A)+;yc(x0<%Ff+tBcIES2DVi;`@8hfm(7920(TNS>R3W!^v
zxk4oN0O5s?drzg*2F(<byG~6qG8wb`UIIC^v&jxvEZwO|44;0ICW~n)=Qo;0%*uU;
zYFQ;+IQx&=MBK*=93?(KH!*h2>M=N|oCXoH0wwk;RaYV1$JnT}1Zl#LvxlwZ=I*tS
z(kXN!Rwld|Pt`|M|J$;3PSw~-c^1xB&&H@klA2Lmc7|Jp3C99Spkt!;cJc<cKWJMM
zhpSgAJmY3{<L3ppa8NNaKJbySD9}*1)frd!cI)WpErE%0a7ZdqlIy}dvB7~#POxp{
zi=&vckSejVh`@_bJ(%Vo+lGlY<c@6KxdwfRDM6Qpe?K)7eV6EF(RdW_g^zIPfK$jU
zl~dy7aGvM;oMFr-c6K2G|6mk7>H^KvSgl*f>VvisM2%o%;^VOLK1`E!tDOR~0nzZA
z>&hiXnKWVfthVuued|Z5FC<>);&RVwIy%}Hl<bYIJV_}0Avqq>qqdjC%3DkiIk#<F
zqWI7xTt({fxc(8&N;A=H)UulzSNP-qE(w@orce2i`}rh?iXMGSXh!KCy$W^LpCewo
z7&_(3IMj@UB~e{0XKs|JvPw{ITRr3^`g%)H&U5|LTFWItMse%NrY==LXx-|>IbyC`
zMeUAr3K14VFr<LHp{M`rVFL*c6mM#GdMAHxsK}4svvxf=&&<puy<R|i|Ngxc-^pi}
zJ{SJ^Y4ax6rS`{OZ6R@SD)RDnYYFyNHJIA!XB8eNdkF?`V9$oNnwE^4FYg;h8wZaa
z_W71r$*{iZ!IjIgrvB@lyrbjT*J`!(X|2b51BAL1bCI1<L347IQG^=M%%>j7_keAI
zVpY-gF@!^~La8CjA1$oU_z2;)I4`n?-0bB@fI3Ge#)E|+xe<9RDp#w@iri7NGem5P
zIn)ND<@oYpo6&_mQZ}$DFoo)1o<;(+P_~>oZ<C9GuezHK8T*BA-th4eY$p%~|J1fI
zPRkt*5jLT_$(^zPHg5#dJh~87#{1L|l#XBd!POtmCHAA@^x|tpS5)|ryoE{GCr0K4
z_^@-;rK+mc9O^}>*V%ZR6Mr#l3R8F&Rx>#D1y|jH9FK^~+|FHZc6Rn&YG1xk!d^sl
zZ~naw4-b>B|Ik=Dip!-54|F-P5H{!=Wek`ziiMx~KopIY6l)%`KbnZMj}Wu<`RmTt
zuGS=z#V<poM~xsueTKeQfYv#kms!9dv2#QFo`r{291@UTI}BmkYmAShq8kh_V{En%
z`CgCSR2P1uRe$vW_Uc_xd<uF|IREZg)N`|fP4+quXJh&eqoGW$uA7?>ei`#=>n;pO
z7v)d{2u8$zXwYD4Xm)<|NXp%5q0bME)N!k~T{v<BF8S5_+J@R`!sIGBTVZ+EdFw9i
zqfnT1G?DC&o^q<LTjjh-V%{Mgjqq80hdQ;EuoAgq768bC^PdWR*h3XG6?Pbg`@3u|
zzr~49|GK%FzcVv4BatkNQg_NJDJiM0K8FH0<{Y*Ps%|gGn=-Sq6y)R@ZZEeeXD7+8
zj#cTd+QRI^zrWwRci{F+y_#)P1ZipYliXuBgc5Ij-7ZIreDy%~=*WHm)o%{Yy@z}(
zV#BSp(Hlng@=~BP_GemL2?gx8%vWVv@A%dC<iLjtbLf>?kKgf{T_1EROi_kH(|mB7
z5gtG0w+#?rrb4kd-%NDhlw!edi<_wP8;ReufOBaq&P8C43QC&XYPr=I9=HS8>9vDx
zUo&#FDoP2uoIm8~^EQ;Y-F(YXXMIK3>ED340Z(<){RaMhC#z#&U96C%<(cCGUV3zs
z+nn<BBAoj~|KH{Iz47Vub_i9AzCVGd%63aWL&mUbdf8ZBR#sL?>19|K7+zB8(GGXQ
zE^gC#2W1ei)k5=ECf5xyAD>0fqE4B2S%k3Ah*TTXri}v*Eb6`dS8##=EYZvI*(DI_
z^3aJVppk+;{t@!wV|%IxsrBDOZR9*#>~dRL$FNCr&|aZ_IYJnPhTXQpf)GRXM0{}$
zYh>wTDC~5dE~LhE_?og}(xU;`^th{0%ng%}ilptRH&aMXnRc!jSw577?Ki(okfe6d
z)0f=?eHw%4rJS#CdKd*x+9NMl9CN>68J4PQmS5ScvzCc&SM`gd%GIn{c^5H_-59BO
zj0oS%(<sQ}8lXV38AJ~lE9&eu%8OrKYN(`R3ehRyiqbB)>FZvW)y1O9&wtq)FMK(?
z`rLbZ`}SQ_!O860Mi&PA_3D%JEw$i`8Rba{nclmrbhD%SaQcE@8)*c6b5cEZ{&r2#
z0|$tPemoAK;TDe?|2OC#jen4ejq8DfSKGm14^!S}DA#$nW^GhKBcN!q<Vz<^b9bhu
zp=>2AEDQw&B`-HuGM=-<?dI~+CPXjdiwJ`(l847eWU=FLbYOyRAH%;xWj%NO2diZP
zV`zZtit(ov<Umx`*2^$1(-Bl=zw+EVAi&(*j1O!*yR>My#&vEMdDh?-0=#f^s6{lU
zl&+x33-(j$s38CRY7(o;uEU{?F@mtR9U9s2)Que9t4N)WYWce~mI6>bwAxjQer?f1
z=o=;h3ftjTinuG($8I_G9hG<5UM<^(uj?A#(0*bRm-8`?%~6R>dRi%a7BE1ZY_J+r
zr^nbkLeemNv$_@#pjPNrG*I_N-6>nb-vh#z3r-#9g`QI64wk*YO8Va!Hj=Vl=GpA&
zSF!$)YdM&@-vo%6f7?5X`f}-8B(faGh)IQ_G*+lKy;8wv7w)4KiR5=So<ng~sa5?7
zA60<FsL>-%4qBs0Iye|6-TcATqid=WzndC@*h94|73|H68s(SOz^tFu9%GOE*jF{_
zU}bA-7Q0gE-CcmBs=NY=7=~9ho-82sw?;?hC@O>&SCrOuY&x6S%kK5hbF)~=*Iq}d
z5W*HVhy%)oR|d>_{Tj^wVl#t}u|#m!Nw>!|d83zp8rKw0@v<wAn)^1lw%{~Rq3ubn
zPW^#IH95JCM()62Rt!HL%_D?4t|20Pk}@5NIm&V2I}E74@am!<YJ1N3jbvWOh|z%w
zpE)>RCEgkLi42uV^zDh`a-49Xslu4{X?Zl?6_Xt=BsH1e!u+r{;da%G18_>khr1DT
zNGsjsmH+%Sx$lv<Gn|-vI)?;J^2g_MMCoQH&8y<rUNscp`ZMih`}9hajJ6yj<?J=M
z?pirZMT0}n$|h$`?DmRw9SZj>I$Slv#I$^mrQ(n!*`|YjPa93~;n=?tVNrf8AWNql
znWY?=2ewRu_H~D(YAHeSBupbc8ndeVw>5zyeSvnu?r_f~UUrXx-hz-=x2X4_pw4n{
zdrge56kTmUVZ2%)tWEHnzknFiJP1TL-S*&BVv70)y+H@LT;|#vs)Uqh1x%ufYT2%9
z;ykdYFquWNql<a<5PkvK3&ZzWEt^$N-;5qZnd)}NhB9WUcrdesnChyZ7-TV=mp}f2
z)I`yz&3jW7TkVh!CdaJJ$V`p(q|)(uEPlWTn23nTBc6T|Lt|sYA$9c0J{DQJC*U;F
z9l@Io#GYE_S&&-A3)2f4w-q>Wdk8wzwFj`s_>P3DqFI*oHRhE9y4M}2V$uuTIZjII
zKy_DhiQ6y6hx5F{$EzV9a)Qq`mV03C>pTZ_MYU<8OHkMEkLGcC_cR)p7UGqA8V#78
ztI3Wt5;bvD&QGdZ#U%nd)IV$Z=Df0=n1sjWV$AchL?oPYC^zuXNq{u)TfY<FYAEb;
zi%jk7UCTwOv1j-dk}@N|ch?8ZX_l%B{MIaWJR#^JJ{mrf{Ud2^A?&#!EkPs7GGLP9
zb_NYmr!cfH*A;bnlWBr&%x}iXB2LGmgf#w8U|KcRikywUFWHp{&ctvtu5sJ9gH1Qe
zu^Yt2BOEZ9q0I*mtdhf#y?OqK&l2p@Oqfr^K0!Sy!#XNU#npAuiAgMIzRz8Q07d*7
zHfO-v2t2Eyp`pgr)cB9d-W@x5L7fW7F-9B_MiNAGs_xh@*FyMYsFZhfqqQB7m|+WV
z%TC^wMf8v#%IdI;M6jIrzD<aq9~f2z__wkN4fL(t9T>w6Uw5YXiaqjq9M=H4F$|a`
zyk1-U;i+Ur5(N+A*BEHG0uAQtpEb>sUbbxQ)KG|)nrmG*zKHu$q0GxNj0H8#yt*+O
zvUL+RVvN!})!5v$yWH}ZZvZ``P}_>{p41uS)#x)S)2;sGC{BtueF#hVc<j|jEweZ(
zOcHa*_I|2!F2|(IX@EVNqwg_#&7F05&MUdwqbCnb6RQT*i|n&K@G(#k-zA`{&+g}+
zwLA^$Q^}XTHd-EClsOW!;cJ6*?H)2#DV1VSwbc|^3Vr(gXU{&Ddvxh?0GvB^KDl`;
zq?Gi)8MWEx10n2>!9!U?))&6o(Einp$1dWx=b+8-_R$3=5q&e)L(f~}Fe<JH{#a?e
zYnQ}B?TbQcEaZbQ7v#7>INtZ4wtrXz3EWmoPncWT`u9ypv6gW1=E+k)A%aECY&axA
zUih92v7}0|%Hi}=u3c6&ZBM|#MXJo)O!x_Vi3t2rFT_I_k}{RU(~$F=?AA*wI=125
zr@KR~eyg5xJrrZ2Veb=l+7VHyo2@!=*$VJfhV46up2{`N&)2=~v1b!HbKo^wtPl*p
zA<yUId4$jmk0OgVy!X5a3Rny^;ktU2nln;}Gu$wbVTVzcw#FjTl0|vN@**@<)OYOT
z1qY6S`*BSde-TcG$W*?1SRbJy;7sjmM87?D?U6=ep?do0J3*Mj)(@`$ivf#g?FzV_
z=;Pgn$ad8PQ>H%JdYor$yU#VFKD;aSHY-7dY3@-Uf!(V?55ewLh&%obEINTUJ~ZvC
z#!VLc^^K(CS3%8%hdu@}ZS#BK!XqA#*6M)+*z2H$$?Xn-(4$bbrquCh_4@jHkt3{U
zg*}epFz+$X8UOB8b6KR6EK^eEt`%ag#4c@GUTL512?89M$ZKrVeqZ)ANHWH)n3hi9
zLfMkjuFL=&Hv`UJ={{8$ml&^i&TQMDH0I&edc7tR6G`7S=-rC;Ts0w#==hDWK^udl
zYEO-$YQF|M?(tb}+w`9Gx?xvnueVuHi>9O)hA@8NAc>(u6k)rohJgu{NKM&D``T1J
zsodntA2CUc;2>b)Ld}|?vyJyOu=R**PyA0fi~sW85LT5BVe|O-IJDn_ba8PJ$8JjC
z#sV8OU}cZ%Te0#a)(^@aeV5J85Z3kK3u5E=1v!sgnHf7wFt5Y8vBNIEUuJSM8UN=d
zIaE=wXdkC1E)`M|P&b1G1Jre8xntdOd}UGFxU`Y?rG5`ZB&D`#FnlH`8+<zWUfO|^
z3!ZO5j^bob6^rchrJU_7o%O!aAM!GX;@V+XP4~S*ee+btFsWkqad%?g1bqCjBUEg7
zO<Sa?9XF2->&8j9(yo?Mmcw4|3aY~epanuNuzm!7msxiv{hrqrkrKeH>bCBvNOBUz
zoPI?h?ADZ&`hq=U#c#oj!yBqB$Y|NVDjvRtv~-hP>16+WODdF9pw0Ze!j&{297=V6
z5UWd}0QD!Vl@q6fGak8;CXM?OHXgKuU)sDSH>*-1d%Ayd-4d57>K&lONz|V_Us0_(
z<77u0R}EJ`WRPxndAQP3HEX9bA-1xzLdb3cbH~+i8PZ|y7^Y`plFE>QbMMf8Y^(=W
zD8mWn5x0Kdg!x=jGD^{OkK}-g<E-9EV{zIluzu<5O(Z5xCtqt!;KOFoVTJ@bB3S%I
zJ;OQFqo#p&bxEdV>4oy6OGlLR6i<(JUOe(_9n9k;L5Ds1=I!Kz6+v}aE&WTCFU8$8
zqVLp)SG2U0CUMUy>}K|@xp0w<EGtC$#}em{`c(v5+^kA}g&ctiVp*ru5AzP(inW~0
z7nJP3TTmAd98xG}^59&4Ka>3IorbCq?0p%@xob=mLM1wK{C#sc8^dQ|;0$XwuVTBP
z1aHU9kLCrgRayB`2Uj<+ka%ZdNkZRR$OxNJ5FVVez)8F|nM6yG_MX4aOqYfh<L2%(
zQ(peuv0L7=M+RH8d`?f$0~&m;sKx1VvNAJ&5FcN-$mhwCJuN<9PZ|g+!H4PS=*08b
z@3!8V@AWUo)(+o&6(ARR^pxxoMI1fFqp8|Obw6X30^6Bf86Mkr?GGV&eG}7y0(jaq
z+bX%vh@r8i>;*>Ov4~XLcZ*q2s!4b1_7%pXi(AVn@#l}pzXs!7*WX1eIPG@v4e)jL
z%+Q?8Snth~vKj=NtbG5DKw)J+zGgHhW!0^YCI58jmB!VNF%x@?!*TIVobt0Vzi-_%
zTp1k^6=QlvH?7_BY`8vh4O)#{vFs;mn+#C`V<b{NVM#<O+1YQ8`$CX}px>jDA7h(I
z8!xUknlgWjQ3>gov~TDzS(Gl;O+?#TR{A=}LQ92*XDHFxE^qALl@G=koym^`qS*Mf
zr7R|x{;;ncvL^%OHrW0L2Uko7fxlNxYjt(?{=Nk%Y0}`JwEWWMX09X~#pXzsl#R{t
z&UBTrkx_cppqQ8#Da7w$psl^Mw>PitHC#PEf_$njN}XDzmOrh;ex61!Fp=Gqwd40~
zO(<OLv!SWpdm+iUZf1rjJ)wwE#7;`85<>lND@ZyS^2ScuM*G?IcIYLn#qx@H`%#XI
zDyw8p;<Q_j1m?Kb4DUK)AD*%J+NnBy5HhP={p!JUSu_1hZxw3F>XxdOmZ(dZB%0!5
zydMb6;Gyy~l@xZZ-OlFDj@swX<+?pMr_2$tv9bI6`%_a>_Vca^yDcWf*2R~U<l^15
zpVZ2duY)pr7%re0H!u}G>+j#_N)#xTxT!Tz@il$(o=|Dmed+3Z_x>wl(a1<-)3FUj
zD!1iLlX-d$ROq<-2JPzAU{9ngXVQ77_mo9I<6U~EyJV%+*=PpcpPZbogdE%dezmz|
zghX!~CUp~CUE7_hiX7Q2Mzva-&EGjbK0bkgfh>BxgXWsEYM%7(rd_!zl$ADrVJVEw
zNitBS+C^qcLaV`zlJY{IUt%QHM-`k$L_<hi9T60UY)Xz9hvxlxQ6h}b8;mZsSVuLS
zxKPeby<e=hh+&^an)CKGMJNjjp^aUW6hide(F*eKgmwxfOnX<N{)9Z`>wh`=BJ{a^
z=9uAG^<Nn#V>^^#I3OW;^~G^Coy|GB@ZU|A>5q+#ed!FGE!Ogwn<pP1YyFc3kw^-K
z%c^3RpPi+PQ=cs!^%ynYO1%}8V|eCz8W?vhy7^3P+TOtUgAmfvFy`HvpOn?dGj$1S
z3TL{Jp;=<3lnN!n#>Y|Kj8qNzSM!85ca4dOp9UBFs8x0gi(($1^`N_Q3Y}nd9a5qP
z&N!t^jCr<4<w;SM47_>AO>k2q1Ht0@5@l7@q;Cbx`3kVd?<~kKEbU%C8wErdm;_JR
ziot&m$TjJ!f>E<O9F1hWfg~FtVcbI$90MLkhUk^O1F9_6*JV^?#a4?gb$n``{DG13
zdZwGyp{c%8{@;96UOuERxGAn%&NjP~AglQhpG=<E6U6%>FBej~K59UpGn((1?DlTE
zxKpCmC;j8Sx#?p7_Q4HvO%S1FIU_Ba4;;!`_Xyl=?EYyaCd$8HLZ<x^rv8(!Z_R?!
z*ZkKPx+*!J_9)Vo+Uc0MBY)F}%IaQC5Y2QmIkEjstPEY@-NDj`#uCe=^Az{gen0?c
zO$OV+z+sv1tpmj9`qmuxu74+8V|&fe`PxiAq59TbVE)P`Ij%0j(a_dZ*{q{vnyJ(`
z_x$+`ZAf*ps?CXLVp80o|M)KfhyN*|X4|{1*77GDE;3a+xJe0)!IU8)yXm-dqESiW
zW>Y_SYQ4S;E(BVwz;jPB;w@GzFO6_RmNz+9lUciW_Lo&$0}clwd*uA5r6;(q3pWPl
zqL#X*GJ0LUy<VkptEd~l#w5d5PS$)a@4O|cQYYVWBOY6u{w$%z9UQ1=h}alA>H>Dp
z6FYU}@nP76gnxiu<JRjObEL`EM&jw)jI5D^Iwkyv*m<}0@1=x@QXe;{In_Rfnr;W+
zDK2*>+SJY1w;t4T5d6qYn0#CArutWnHP?fin?Ce9{1p3OOEN#bvbj7nz2R9yJqr3n
zWtvu>%4eQNW~!rod(9W0Zam(FzqdBi*|s{d(_05VmBEHV!2?&6756GRDQ~HE*Y1J3
zkaSa$TTgYIdbHeUIcf~dXVCn2Lf+p;9#zCxnCJY&)BPkf2ki~{->gEgn*!K0ZgWgb
z@t3~L&yTRLF80jJq9iuZG8N=vkBiY?{3M==J+r}do0uJ%5*Ho0+jHG(BsLA2bvvj}
zF3o;R#mcd<@tkOFZKg|IPzckhE*^@Ol~pKma&=ZBu+{4sy0nCYf&N=C_cyZ48U%B!
z4_DaUV^^3}>iwL-iN{x$mL>}k@%Q&PG1(v99?Pk(udl3>e}T2h^VHy(uC6W{8=H`h
zsi~=nKPjY=0<2<aC`O9(uNA8kVibcU5h_{+25JfxaFyXeQOLj|zWEvxgP%0N#8mgN
z;n%<b4F%w(7#BbsM2xNOQ_A~O$_5uPc7KHq8!Y57#&-I54}<qaAo+LO3?uuB^zZf+
zOk@GP{r!as!*KX_2ZMM2fBdKeI`8@;ki5TTD*s4?hO|8GtK~9}fVVXEWO*9e)bdv0
zP!)r311(H01OM}zFoLfykN@+dh_J`J|8Cvk1RtRPyVZmRH^cwC#e-4p|9}7Jz5M@|
zpZ(&B1T&93K0c-vi#)%){PE+*CJzp1fm_scOMr<PmYCSTd@wURn<XAITw|^8iu4e+
z+1=gk>+dfvD%#c8XFA{5P*rsn{!-S&)bxZH7sk3ioFT?zw>?p+D|U6dgIy)Bpg;=o
zI=tfNZ!9k_zc<Wv_DP(U6j42+uJOsvA`5Tqg(!wDPqvJWjZcUl!Lr5Zaf;OHPOIkK
zC@H=QO3TUFp6$*$pY1ZJR$ELIYh4DS!3c11ajVQ`xl0sHz-4Nh5EBy<5@Na19kJJV
zRVWt4c>kgBI*Y}(wzd_+5*N_LMM_Evh{}a@baf-b|6akt#U_)?Uj>t=U%f$p<w;^g
zLqS17N%?O66U-ga?P4Wr-tDSDt<DahT26<3O&y&-+mmI~7+rzHJcma|C&a`s*3|55
zgPuqR?GjEhGBRdntG&4fFErxI1rKy1Bcrh*4Gme@v^^M{pgN?hyE50{92Oc%kXvlI
z{1XTQ_|xs#y0hl5@IY7=mX@O8;xlFXL5#}?u5TWob6bHi%VtR`DJgC6JcUUsNJ#W;
zjAYfQnkgzqXJyfNwJln3lDZxbHUH@C06Zk3Y*C>%pTEKb$g{MxbWBVJW2GP`Pbn=e
zwO;FogM;8qr^?AGWMxMd78dMwrdHP01e{OB=iweSLmfYV-uw2ca>Jpap(mq`*XIXq
zUg!+;^o^IBbf;wgX=%&8*mTvZyFepiqoT6f9Nq{7H+wLKM@E{>)lW`MU4}{V57~pG
zKni=Jn4+Shz?-Y=!qo#ue0P*|w>_E{)d)R+hp~a>{44|<2xpQK5)x8Y#_~*8Pj8ck
z9wz;qkB^UsC$4i@Nm<#7lPJS>qG-*{&8@U-WoKu{)y0JhL)68kK2ZDNO$*C@GRkzN
zDIG1Xa*4L52-Wc4=f-iEV|;3IyZO^?dan;A$IqU@41UhdC3=U3g@N8VtPtr_AEE6H
zr3}9>aPEx=i1%0NK$6=fP>CGvgz#`Iip|NgELF#WL_SPR%t?8J+2H1PetrP$1O<_D
zS*0f>k*>}>z<X+E-{0TQ$j5gxRbkxK)uqZf-qyC<a5mRwE8zYHkJVtXD~x7jc$nR6
zhSTv-hli&wKR@5w+Z#$vO*NaXZMZp|zC?QpV+~74kPz}U57DGH8Gcf3CbE#CtE;cS
zH(A!%)n&Na8*>@>5XRc(b_E^&5!KO2w<)2kOzZ&JJ1<oH)bVv#vL$GhH<~Lq58d6Y
z0HJF%(ld}U>-?&BJR*oqLxB|)`1&GK$f%mmwYGlqrlq5UPM%_}<0O9d!lA<Pn{;00
zzq__33Ch~DfC;kD8{VBY-R*2_;FFM$;N#c0UY&kQ`yi*FU_M)mEj91}?>krsJ40#0
zLR47DJhq$dg;zk&<AQ^e_Xc5tU$$L)UHCN$XL@`6Z?6T+{t6C-+h%<bz;GIg*gq>P
z8GDM*{RemB^YV;=`W6*2B@NhHSs|mLot~W)<>g&ntR<b?tL##D1f5Ei85=vhdWrUX
zA)$6)6`;$FtO0?$`rKTjzF0PNQvToe?yzM1Q$kfaxd>IRy0gHu=BCGBw!}n5MY*j0
zct=Hzr;COi6QjUbJ#%tiJ_{gV^M!g=x+Cvnf1&TV(s1>+XhfWXLT}!@0a`ZPO8)K#
z0s?|}#}d#7Q78pYP`3EJPEQZ`ijt4?i{J0WlT{oLNt8d@+1a_ex>{KsR$DIrVW*~|
z^2cZS;mvL`v9Pg`6Ldc-04dWkF)64P6#}~<6dn~tQ&-3at2g@yef(%;ZS9k7zc*(U
zcX_;l`Rv)j_9RPu2kb}3)KsiDG78E{x1X=Cced$trIYt#8<fYq<yibldkYI>3W{oK
zAUx2~)^5FREH@naK$-&QuFT8F2P#+^jV>=05~HGog#ak8-bHdqIJj`3q)fZKbShD#
zugWPb4D{~p?X`e#;PE2Sh`DDPoX>7=t~?N*gfi&oDOcCkZM|e1B&OOQlmUqz9h&Xe
zpa=Nk;vzpkKd_fhKFF}&99DmR9jy&?wzt>c-Clk8aOe7mmYqG)o7;9Xy=f2(`uX$c
zjEs!<`1l-VGn!J3A|+L|+89LcEx=s;UvU^@lrAO|6&cCzaz5DJ-abFC(X{v#Z2=#+
zROv$C62S9va&orD3ceJD=M@z20rlzd!*%3U7Zh}Fga`CT@kHJj?^ECgHik36;3#tQ
z@^~zl+CwOX0d{yU8J~S`W?6vLPfkx6X=$fQby4C}fq*bcNYqtSuAtqzz1X<8EgnGw
z%W8PwU7sL7etfay$7D8B{R|gZ-`H41QSk+ipCj+ww}&X(J3D{P`%;4aoa5=X(&x{=
zGbQ5Ot`CG$q@$CPm~i}RRCmGTySTdEtF3{_!BPkCISs(DCh*t;-4Bk9rN!_eA|}@F
zjZWDc6vTz@{kWh<d_n^7S_Js`j=(blBtt_(BPBKPQa1BNp$QVKC<}<}Y;RtGo12@G
zl9GvOU~&?4g1cD`60ozh9I`in4Pv38p#fue3l_}YOwAEcArK$H!6S;tFncy2lPL2A
z0#|B}`X8V$_=xv@tl)xR|Gz<r{~byEPk>S`2};i`EiJRPwq+Gi#ma#LFfWuJ{2!sA
zHCnCsHoo&f^y$zBr;ILI08d%8r>KrQU1bg;4O}Mm-@9fg$jFKcz#s}M8wlQ*>+5$e
zS~Om)#*G)xUL38JnW>iQ_{=<q-_&|Pk|}|S><LWZ@B0LSIV*UJ!Ks-hs7W~J{&5pE
z)@vR}5IQn4vJk~5Wo2a(lWkB5MnWgq-rIw2E^TL#B4c7eqO%X=TGks4Ur7U@1D{R^
zrm<M74VaN^YDPvgFkeFD2169@dWYrCfilt3(&8|xX735YK+6B;B{=L=FitXon@y_~
zAw)z-2nPoTJlo-Fr^@rKmZYR4SpVI?9|E^ksNTSBGEvN;-@nfJata_SU=RxmOx4vB
z8K)n%Rah*xhK7cY<|%CPJeBqnp#rZRDndn0PEN>S_S>FQd3AS}1Pkl94P79f*YPkd
zh=16g6O8)B{ivP5$|~MY=vldU=aZ$npsQWw+fSeFp%tW1q|Sih^D}_Z9$+LAKBsj$
zW!NZCe#_(ap~W_@vz=-1#Hx}KE$b8p28N2pBMx~72L~CM;p>Z|DwE07Jw@gAC=IuJ
z=Kye`KM7xK+u}M85zL+AS>_<TAl)Z<dG;?qQ6n2jdF)J1Hb<={d7eG?Y%MN+#okhp
zpAVUI_+nprV+k?<?R}mK^t`IQpNlB4;r`-F8Zh^{Y5EGd<!NbOfYl!>P&pwM6SK6l
z1MOngff)f_Id4x9h*}!p@#Kg{C@4K$UF50B$;q$3{LIeg((jM|3Xg;_1m;@3!6_r{
z+@`R;UX>oF$^a8!GY*a_=zZLvF91V;DYCkkRaseCS}H>k!U`z_bOXQT>swk}jK^b_
zzZpBl&28`;%?lF`?|f1}X=7!qykGL?0RI*6ZMKhLFdKA$(#sLGOQIPv8L*Jwy?aMk
zKnf<Ch=>ReFFG<Zv-$*}U!e6KNa$t&84Hr_pPbyRr$Ir1qyp+18byjV>FHFgtTy`#
zEdWTYuC9Wqig~5g8bHXQv(=lx6V<l(-P7~gCpR~Kl^@Ex$ebXwp`vQ<?Cfl91xLP~
zpPzsK{+)`3CT|j6;Cdh4tIZLxI=o*{P_*{+h><<TBP5)joP0veZF6^f^U-5GdzRgF
z%7PQbWYKsO+)l>^0GmljNTj8v{(dZ##u#z5IZ8M71*{ag8K>)W%StwYf0bX{ll}X_
z9Fg+7a9S(~($UfJCow&J<aD;%0NuICXO)-VtJmDzoSvQ@Fi7uypmJHS`p@)dT)kvs
zO5$_20|>3JuMa$^&Srzzcr5=II=0Dx<FI}a1MK72=x6{j&)CdNeM)w$wGB2lwlC;|
zDF5kG2|!4|IaQfX%ac8|*%~tiBmt>J-cQO(2}f^<vm_HP_uJ5cjW9Pd0y_@)_4|oN
z`Eou5E1M|Lc@*9n4zG7jSb4CyT|0ni5O^$OC$QBdnHU2GYqN(g4s}mCcH+z%4g1}I
z9S{^0d<VS}4kS`iP@qw{11schinQSh@Ht|650Q}Q<R{7v!^+><*pvqZ1ZXD!A^wt=
z)dIs3`dwi>Zr*ewZ^(OldrM7CZ8blft@h>17Yc6sT_R&Yz;5uusVfJ5Hv(X@v*EWJ
z#(9B6Tu^HD-TJygB^w`~nzDhG)*GQb`2uayKQ1Su3IKuvx{hyLrP(ZPo%QL-uV24*
zXY1(XK{*CE<Gu|UY(+m{eq?9TBhmo6(D$sf-;<xc>jkbH_!I22u0;!Zu(}nL`i|aK
zOid|wj2Az9mNXz>w+##n!!rhr#`-SkjZul(2~tA}$=ESd&}LXk29hyQL7{TX<qvFi
zSb#?T=&xeaDHQ4`!~;NS*6-HZ);8$|;&HGVaj~&&w<k(Ks+Ea}Nxo654bgBVK0Y4!
zyJsOaH5aX(DCj`tK;Xb;JjThzRnko(l^8yI0crz2zsAPo3x}ZP&Y-eOfD{1c0%MtO
zf*@(>U>~2ybV0tpMyytL!~<(blgDZT0!@R1gYZab^vdPGT3WvD=!3`h8Ah$30u0I!
z)njX8quStf0xVWpSy{zI#+1ZDjVj}_hX@Gq+_togjJx+UJd*iD?2BR&K>0u0+rh_e
z7h56E{CR{2Z<Lf`X54srdA)yicCK}jLjFNP0<M?G8n=hIW7&0e&epZ%<R-C*&-k2<
zMJ)5=10~ZC_&GQfVwWWnc^%jK6F}So-CfNW<mW3c#<CfwN3iiR5n$lq{n|wFYz009
zK-PTqhFHGq*0m$0sP5;(1P^mM?AQp7p&w3)X(K|ZhdRGoG+jY#DZ~mEqwOUokUW5T
zz@!YT%z#+0Z!#z^kM5Z*(ERf9&z1G*>ECCEiCFaf%cZ%`&ViguBV=V{CMvD~`q1f#
zc)o9IYYUc?-ilR~b&*C>zuDHs<z@FQBunK?X*QOxR8aw<U+}vS5Sn(}eYp0=PN`qh
z#Do|4prs}4(E|~~-}h4Vc|lJvURk=mjIB~>5;DtbKF2HXwl&7kaf3?2tH^kJu-sWD
z@$%)%;e`t3a15V6`-^S&c^Nfe(yHa?gtI<9AjQCF{lU-8tz5~=z|c269p`=S(pY{6
zBozKp91>?`W~QJpH{Z*WO8VH-q5T@=M~Fl^8?ubbjuH3zlat~PSw2E5UD+Nj+B6A*
zWWhH)An#H86(m}e;WyrDm;}7$dcv0{E_8Ew5+<KlSO@?Sa1W7UZg`b=4&#Mt8PwZv
zC)?YFUiT<Bx?CJ@i~!TTf-Fk@HHyESp2HD}Q)+VKn|4}gh8yK0lUU%fStkRdsC-UC
zQ$|=xDI92GRY$M?ikbPNxcFyc|L{z;+*yP<U{V=X+>Y0u<pJDmU-PNX>4bLL=6qkP
z=EkCC1{m@hyNo;wfQ*YVOrK;3QE16l&pWOWXQ)(I*=*A6gN02Wr2>G8_A3a08Xb*`
z%dH|MpUgE+8B7{Uow|Uaps4B38X9|BTRoNkr?fK-hk9@0_*5P`6|zKWQIdM1QP%3%
zPKIMivW$IgQB-z=ihq=7+DW7&6G>vmmVN8ws3u`78HJ+kOUY8QJm2Y@H_!E4&x_}J
z=9L#e=Kovn`~H4E_wRo(5_2~>+hkReAXLmjkg8l)IjuC{#xl3C;D(Bdii#I3B<Wmy
zDnldHOR?CtGLIfVK0dZDNTUOZ&bc&b!>J6Dt!TJqrJz7lP)KQ`9oq{2i9*v^9-?s2
znz)=Pe)z-*&0{SJ)tQJv1$zr(f~j{=B0Ie)2TFeYOO146($$qJY3(Saw0@3Wt&7UA
z`S94()Ddla1H;Yy!mHKxdZv5#EB`@XdR}nlcAbPG!)*Hp6xr{7W+%4^Z90_{o|u?O
z-`%#mSU+)>pT4Z9NCQLfqIdyZN1g?dwolg2!ETpgmUG2E3h=_{+XCi06xy#)vg*nP
zC-O65fAbo1X9V#k-MWV~-m7g%BGqr==^7~m#VxVr5?QcDzlx8Kb~;aJZ4x!KvTB(a
zd=og_rV1YbjliaU`__GxyW|6=ht%kl@6n8+3J#+}pWza@M<`7YTdB<MCoGnW<U1T?
zj}aE^m=T-%#fnPJUfpKm8K=3tL*+6poyL9pmL1FDHps?=%6pp2h6vdPgkX-VH>M6U
z4BZF&q$bAuEo6Lsea(x>QX<*T>Y`u<gxY-R7J2!Dn)OU)?bH<=Aye&c{OOsr{Lr~*
zLN4t>*Nca@t3(&`4QW;Q(;(6NE@hYFkk>B^3=Ew7Fx5#|m)d%`y6!+-D|)(AlDwsz
z3jGDz1T%T_&K=m@R7L|drrz(r3kV3LrKG$o$7T<|r9A8Y!Y?SL;`;QQsn+*r98ONd
zbK8aY{A<1ctgRgebsE@AA_RwxPG%-WM|Zt{U;mNj=-|+FiDEQMZeDd+)Q%RHV`jI;
zX!6$^1AI;qJt>3qRv)mM^w6?kSJw?c>Sw-&=Nkyl+chQYMq)<ow0xyLlbt5wruvD3
zR;Y=7zP>zr)3<J<a*@ul_oJhAT6eSwr&{KgkVP}#BqafZ#1m{D+KSmWU|e-zButxZ
z896yOpbq*(q_Fhe8#nx+M4>ryRhSPR$f>E(kOJWfE9Ry%5~>~+wz-)KSXx`F`&n~>
zCvr?zsU$nX-KlPLRT<{Ixi;WnF#B>qcvDFtyJyc}BFA<c9rw;sg@e?;mcIWIG^9)D
zCc5J&xVg2wTA$C!8kkO*yr({_Vy3YwK+R@|T*e6o@vUcI8al`7GprTW?tklRxL;IM
zbh(9qxx;z-xJ1#hc)VKD8Jqwcf_jjXmp2Y4Q86>d2Da!tHLBTfoc!@~P$_aHj<+Jr
z-L751$W2Huge+rdXvi=|ZiTq`!z<4k8)Kp@6>@7rkGhEYG&j^|r36ZH&v|DSWLP#9
zTW)+bnm!OMQF7yx+2y(J20^v5RAPSaGC5<TBRxDJ&<Vnvb)ztO^_$?}VBZW!m2J;^
z3J^YTy4)GJs~&-_kjZlSo6}&_BsPQzoN1|SZ6$){X1-G_EiArHym3~yBdWsx!O5>l
zX$|L`*3!meW9>O_eJ*ufE`goCbZHrvF9+~3EG!Jlnp+$oF9Kj_Y(4L}*q8+KkTSjw
z7x}2ZzDlZ}S3WTh9C>GE$14^xN;`JAHlE4v$TCl$rF+nI@oxj~YPjp4X(Ljq$sw7d
zL6aldO-+`mv!;{-;1d$YaW%*0f?Yj4UOF8nzV?`;t{Ca8=)@K^XIb$(UQD~-$)EA?
zVbt&fl*lt9yvbM{z@ra!fd;q35gqTFj*Xfe;`-s*eD&C?S9@i~K7Zctux_u+byk}0
zVZk*!>p`eNHVJx(6vXV0$`On7%9fk*Wb)1OYql*Fa}!Js#~h?gtYic3gqW9hnVOh@
zJ}iGLd`ib|aCmE~d)%X3bz7P1QDsWA0r#rOhwChl96s#6>zQ1CbB3e{>Z(z{Z497*
z_!3A(>&G@#MtQzYtjGgohr-Ixpy*tPBsYG7mj{&k>@-aLm2Ei0t*z_k1h75fy!!S_
z^8S?~B2_t!P@fe*EcEYAFm11HDKA>I$gn5T-h9Z-&d7-VPId<F<=VC7w67*uIrgc^
zAwvo*Hr?3%cM63iOlRA&N5Z0bH@tIhyFtj^V6W=x?nWKF@SIJsjr&(TF-Jv$mCVU3
zJvrflUM|!DW&A2Jb3dYuYSc)kzI^@qO4-H1_f=<eDohMyqk$IX9oagHd9A7NGjfcS
zloUFhE^&b>cJR+ju1vLf$o^P)>qF`nD2!XmM{dT(`s}>3O4*I|^g*JeZ~5$`{bY10
zV%oTq*!}vkV~LuY+LK$qd1N?dE-SX$C#J7{w)#p_d%L3#RU7Sacf8U`vb!p>QGbm_
zkT*6Bk0(iX_&hcywcGzFkR#ea%1TNohVThg2Z!T_4&|MAt0okw95~{@WnPt1^+2#S
zH#h4>3OSaaE@CphTwOol!PUGvh&u@h9-f|9*e@Sn#KgzPgPAPm3x|P>*=&ji#gFgb
zv*WhI`6KVGh57^OSGbGdHlC_mw|g6ze1%TO`vp1K**@Oh+tt*BgoUNmJsn`$D=G*V
zm-QuSyne)~DbFY7O-OiDjY9dzy57;&>~8oO;H?0{&#wd^9vt)4NIVZn9zL(Ia`;IG
zPiXjW*4NjEnSe4X$C&zL-TBFailX9Tt&`(0c3r@}7+?}q5Nuq3APx!Ndj0Ynv9ayt
zA@;>O_sik#(EJu&vy;VQc|VI1S+HQi?2oBHAD>`wMod9kJ9ICFa1NQE)?8U@8)8zo
zG_|y>jQ3@@j|=BF(cjqHXsSJo66)mCfo*)9q+Qqj1u7Yl-t_VBSd>pQY@AMZKf6L&
z+El5aZvjasP0F!{k5tKWD4`7x7j$~|T?L1<K7RZ$MaIy`NQQyy+%G}{p+V`+JCq0|
zyJG+JA<JQ0{?3_pKK(^ccj^AA!8gppLK|AIZ;YM#i8k4Go*PMx`-Wp;W@a|9W&%Z{
z)%x{a&0{A{WEN(aYR~2|7<)$SrSjPx=OZH0)rCy@&HDu>M=R!?Xq)Sjv<q@`E!yfh
zwzd8J{haBcE3+^lpTgluj}WnNm2w4!#X?-giWyE{sI2HO?1-p7tZmMxT4|#m02NsV
z+{+eE&dUZvd7L(TyUgUdI0IQ;h{ag7?mvI-+;eJ6U+GZ;DzzEKVa1e-hsVippFi;M
z9W7k{KOI&EnBaKDktHbg#4>abk47MX8TapVhpccX1tt}>9avxvJC0NPcwb1!?0Lb}
zq0{}Fi$?YJ^aQ1~-c?se)O4c^fqV~wDc~-df<)ad<r*;Kx?V+x!u@vh@QO&);Pa8l
zKM@ht_FiSG06W-DdBNl85oBd$X{;maUcBh%#r5rV+MyZ{_waFkKR=M<xL<I{&FxX}
za}Hm5X{iS6K8V8~f1Iu9=C-ze`|(41r}w>N#Ydf<jiNyp<QyN#geh#<Vv}>o5G?am
zZLRHEdHD?VIA&%}PbM)97b?0>?pFd|OX0sM_QrUj;&p>1ONpg%WXI_idoOf&mhgq&
z)jYK?NlSCb4m0-|sYJ=GGrijLt5iKKag_G%-3w=_^<zl6g~6fNHLlNZ%}O|%#fVg$
zW;gzH_|+XpT;1Jmq5oW_a1JkiFpzTU1)Et`Ev!Ama;aJ<10@{ypfHmPJb`{NlgS(>
zK22s`HIVfU2nc~I#oBO{IV9#;Gz&e!!!Al}Yb*4*Uw&`_DM-v!1aDMt-wrwha(S#8
z9P3tmJfDDoSOb(Xa9|~|RwJnX`SYk*hsCfck-T2;`vCDZYmU3RW&@90H%I}hii#pT
zmBvR$D?yd%$1NiG{?7IoddSPG;^f3YZ(S0uW*c{Ala}*5y2JYAuw=h=hLK36ML)Z~
zr4M@Qk~$Hn2<(?HjY4V`dKJSrfCVvabN}RWZSkvrkIm!9<3d&QfDPKJfeQCjkup#r
zR%xE}-?JyYoH!V!?g$1My~()ebz;U}*pioTe|{=S?}aq}`ti(v=@S1t+x<EAe;I1?
efB)Gt^NSu&XB(^*``F0)c0;{=l&n22;r{}{P6WRI

diff --git a/01-exploratorio_files/figure-html/unnamed-chunk-33-1.png b/01-exploratorio_files/figure-html/unnamed-chunk-33-1.png
index 10c2b8039221048b1b163a386846f5887e3db9e5..68505c3d4521ed172f1befd6061739771d858d13 100644
GIT binary patch
delta 152992
zcma&OWmHvd^e?&)MMO%JPD$zR4grx85s(&X>F$L}r-*>GNJ~j~NrQArcWjXEKF{_&
z|8wsBaL2f7_`%^Ave$lM&R<PN3NcbEF@FA$g-^(*$DA2LZamqABygLaK9=ykWAuRW
z^XGiF*A=;)_0}N|h3)Z*qTTp?NOu~PX4z|J<BgB4KNI)cEYfbqagk<khnfiXGj2b#
zdR}a-ppi+dO?~h97T=*h+xFR&JuT6#rdvPX51nvR-~@SYiw08NTkz}`oZESCfDu=-
z8ZisUbCsWRr<%PRFZeSs!1YGm<a6T%*UemipAZhbqKuusL>np6u9bh5Sl~TXY1^M<
zVQXWPl9Cc3L+unH^sZI_Iy^aXb#)znx`Wwb&O;KTkg1j{5RsUa^!vxdl!{%o1+<T8
z_4VE!9v+sKmZqlk1flmn%)8WDD4Z!J@|mo{smx9scl7Gqx{UYZ#2KmjZ$0k%c*Dx;
zvyw2K{vKBf{c%}9emK<d<A-I$j~9Dxm!p~n_`x!Bm#h54ui%~9orX|C_J)h?T5@vo
zlQ~!%a&mPwsIp7T&;M(1y<)KL<!NHFGTwN7#O|{f8i-Bqak>!bNb??#I{F!fJIkm$
zEOgYN88jZ9{a&7SvRt$5Mwxz2!LW>glPPA|Hbp*JB<1|z9W7C(ASP8*$nh+-O(Xtt
z&*!Op8RLRz!eE(1e%tN!b;#Fot#8Jmhc*6S$fojQDDye=O(QQcAt6lm?dtgf0%5c-
zxoY%!%!+T=HBp{+(WRDwfgw99YjN)sj^{R;oSH&hyt4oJdcKX2=4jiXGo$pUBcA>B
z%A7*Ot~kL2FIWXXT<x&W*z*sA{W_KgEz2EYLs73&yC;t(%in1Vd1O;~{w==PX%tWX
z8sf7!S7EKQYIznC60*9wT3A@P)Db!7I_KT9tHeSc@L=YrQkg|4hX|FE0c{~MjE-jB
zVBW6JZfS{<=};aR6id<*;<Ej~dWsSA@Lqjf(A1qYswZE<WP8T=OTv49u4~z_E(-M7
zRN`P`w;L6MC5I1IdlAw+79u?*wJuxTdibe*Pc_TTc`zWOVgc`QiClH2nC>4xendq@
z9jx^qi0n92Tk+{U_{g}v*!G+rK5K#57Dbu|Suu~_x+waA9Z!3uK7?{PGV<s#Ei+61
z(X8hVN>ia;ZAr?SK*UnjVcgZ#74*~Oxl3{Zt4@WJv-4p3JE*7R*keq;adJt5S)SI2
zbgUF>Q`3`-E+aSB!ox#|3_CJBJaa7%g^C~)#aAu&cWw5}+5v0HTtoyMVa{X8L*lwU
z9q;_4{#D@G*&^D@fwoZk!y^eE?d57hAx>Ypf6M=rj=X93=s>&RxQ)<2#Ori-WaLY;
z;m@fp%8Nf8V+9w6iTuf{TqNPr($aDfWQeUTa7;JH3Tm&8r?@}D*~9mUQ$G%`b1y9|
zG4GN;_KWqCr~OWhfbTsfWYe7&P-cp`<*SgH5FZ~bL%ruF!fi!xdjT9I9#j3_B6YIq
z!LsQ#l~X$>?|2e^OQ)x&AHG-4PRGF@SCk}VPZ*f0XIi8Fw{~qhs3@0vV8<!N*$6^=
zdU`tAk&lp&5QX>Qi<(k$=+=7(Iy^ii6LI69h#ZC)`jZ3?x|xeau1}XPj<?eL{aW5Y
zoxOYp8X9O&l)Ui*>VZf72ddvQrcO>@W?!3ISacPzCnK&-=5TJ`!faV)qt>c>$e^5|
z#+cX}n|Q~r+P`JtgZ?XZYIAvnORc6W5rPXr5ouN-6-f}<hHx6{UBJS^vdeU+r9zjB
zq$r0ctT6nkV-%iJM+FB5Yq8Md1*_$%i%{%rZH-!%AC??mWJVoWy+ud+c!>61BQH5A
zDPx>JUeUzFL@oD=&GDW`J@TqWchT0Xry-A&<C42ih3P&b_xs$*og%YQn6vQiAti6w
zYQ|2-ptx4qDz4P`!6$km{$n}S;$B>saCEf*VH&R`*6m&JGvS*PxY2rwuULx=J1`Z#
zJ`=xOz!ATg=;yQgqiSG)1M!&Oj28Kf4h`+c_j|S5RC<b}#2NUUm5=LtuEv{@kdUm$
z4jOaa(-c!PQ)7xjgtC01LcgQ5*gGTO%~SeXgd)4SMQyZF=W=_le27sSFY^E{rSbdC
zN0ivRdgPHlnA5W|gm5aeteLP%NlFTnO~($TDJ`C`8Z+k+6%~aBrl#@+%@Yz6ONF|(
zgf@#c@(S|vorIG8KII~xypWtc^&eojotC8Tf<wIKJ!mT2DihjR(7yXjo<6uL<7S}>
z?Mt5lB@_1b?Qx#Vy#i<W^r6gn-REmEm(7Aoo0*#Oni`?{ZM{r2@9Trq6Oa0xx`V#^
z>~*{BH~Z{%nb$9slzRCa&y<8RP0O_#Puti7S{52FSJ?YSPo^~+&tE~Rd81;f><Oi3
z(X%<sPn1nxyUY@wDmrTBC{XrV@o{r=^YMWeP-#2Yup1FV(J}-pG&&a=XPf6{yh|xK
zmSB5JWLnW2B=fzdrUulOR9r3|o{T|rTD;&Hn{kBG!-o$C&3U5a<ulbt<nz>-pr2YS
z;AiHlE3n0x>G4&@{Wcqld}nDHCcAWcDp21*h0YbP$QDN$-adVC?vI`IEp&}{Co=L$
zNWaUoR=iL03M(-G63J6^#^gpZb==EJ+XCit_*zwxVE=BAxUQ4+69hC1uHDx0^x96<
zQiSWQyP2V(uMcBfk969=jO$!-dEarL@XlqA!S&GixLo71K;PPdzf@X&-k=%A<m;MG
z?Fj!Yl5o>)m01MSYma)dO{ai2T&XO#yz{{E2(9b;1-BV-o`3$K{$u(GJLeGG9XFAN
zjd#C~|4x*CGWDo$sHosO>8h!@tR8c?!QHOeEEFt+cj`}8>pOp)WIW>eD#f1Sk1I7N
z)M{iu#&*s#s~)wn$+&&a(oyh7xa%l*%{{eWQ(r&jHQ9p#e4obD!lPk?Q#>!tp!UL%
z2k6uoaL-<$<mtjgJ<YLYaSnWZd}w~pZM~N-y6ZMn-L-)@0LzOgFFMT6H+3o^@nL$3
zu7~rdJ!`Sw`pLB}vf?~0bITM)r3{vt_1I*MHQ52ZI3J&#E$%+_BQFneaCCHpii+yg
z;*}L2PYmVc=V$bSe+M;Zcw)kRScxAt;S+Y<CWDrfH{{UMRwP7h78dVtiOGII!BtbN
zj^)((gJm6-YgAopx;g$)>V+<ypvw~0b1{nlKH4smw2Apq2%)}P^UevlLj^jOrW8U>
z>$p(90bI>;tmis&lr4URh=ebf$X%zSs?E20`R8s!35g_-%ilkH7jy8<^+@i^c)H|U
z@0<MZ6K1F(SSzPXZq1EI=~mgjadZ1Brc57QRJV@-@$mDT75;Lrol(x_euUvK^|<D}
zMy|TXt5@;8tAZ6LRl0@E&6278af<W+1IeaaSXs4Z8le}^av_!t;AiU%U)TO@vKJo<
z4bH%A8sltB=(KqD7uN9QCOf_d#neFSSYcu#`G1|BFSFJ!<;_>m?drQw!%6+l;4=p(
zsLcqVn4+7*Z1LRD4HcjL=tFQWYCwm7|Ni~WY2?kfQP<C1%kNR|A~zylG1g{ma<azf
z%A;ufBK77hHCG)20(j%4ifD`?y=}wUYB^p9o>U*4BI^0~Pw~RSo60Hj$PV`u9Dk|d
z;o)qRR572+Q>$uzPft(KOT>72@gqC#NTT~ofd&?cLz&e>j4Q=V<@{HE@!}>$+~Y0;
zz4<AA*0Pf|>qQ>R9bCBHu0vFs`O`6jYtNhVb;<(OzxL$MG(+qLot}1xQ-lNTTRR7P
zZ@n{knKz2z|G~ywGjG)3dOr@4;<6FUkaC(0pJxNyQkkCKV#whplA_K;7v0rxn8`kf
zbubbK%1)8bQ6A{Eg5Ku<1{^gtNl-_D6Xta|D5tq}(Nk28d`1`S4v&vR+7^Ykt5CUy
zOtCRB!y+S50Sd@3EX+EA$$u^E{`sTz<dH+|40hnhv4n0OF)<44z@F~zzb>`1>Cfn+
zX=rFb;qm*_*VlJ-aq?DM*)Nv#W^G)}wt*8CmO2%0RKDf2Cr7DWm?lsAS?-$t3~rGA
zHStXJ;4M+t$J94W7-44*kQ+Oo#*BdUY4JqQ*V79M3c#sSFn|pVsBq1}V`b&F+kCea
zk7dfL!Ysvob-wvywsP|R#g#^0c|*f>S|CMtx0PA3R>RHJj`+#6BiNzs&CMkNoBkA0
z9MMBu);k8&B}|$o{$ozg^$MAjwXU|N35%1HlQeP>pryz4#E>=xw!Dd00_Q%e&&`&H
z1pG%~F<6Wuw9VYAQFjhU3SC0LmHudql5ke+NdNB3HxFh6dd5;@$TCXW!U7i#0Ws)U
zX1@+qROua7XNwa#d|grP%sB&~DZp%WFNRP=%Dpd+{YRwM1h)4&C=KLG*}W0ZpK}(z
zqC&s*ScWZbd}PEJpBOK+4ejG&GVI}!<7#akSZ8=p_|>adPZ<tqqR`QH(OJ{AzTC9f
zV(8)bH(qm1)x=jjlztZIP)j!RO9^e=`e1xZhFVsyukf9r4u1*h<hR|YJYlm>u4;>R
z1Fw@jPB>_z&$KzLHtF~DPPjd!(+K~8H>sYqpU^d^1dPzM+fulal9r-5{!f-#_y)e7
zhya*l0vq8}v%kNe&rqyPFU3sbe8_Px@V*#pynOwY+@746q?i8n=-homp`w~QpA;UX
z@;k4}Z2Z30uEyA~lWwJVJd}9WwX#^d{cvW4tqzu4ub0<lL%$=LYuwFirAISW5|!J@
z)*p;4mwp;A-k@9(HtgPkPWK{VI1I%r7~JhBdD=%9)w+cM1WhEYBpY^k@1Pfisjndx
zl^^w`;`3QAz*mlV7=43biUQI7*Q#FQSAUf*Le~8mUxos>zBT=Fn(Dtp7V;YP`TC1k
z)h9eWJj;7-zZN)@vc=h}bsMJBrO;QNir8>*>n&GD+3jbG-x8aH&yQI?G&P=kK6pG(
zl3UgA?-f-MEs#$*VrOS(dpoOqkthH(T_z;6vnS&h6pUx1O1;A{^+21obxD8Z?0)cU
zk$nALiG0B`)_272{}pBr4qpQ{>Zl;GppGTS%CI&g=u^xaHulOXZ^RHC5+o=oDaocs
znM}Zkg9rq**%^U_vGMZd*}<LUk~b{@&YNSb!)~}ttd4~~21XyKJTJmKY2aE1(f>d$
zerw;{ZG5_9YBpT---9giibHjK_Zdz2`S{QvmY(g&^1ScgskuQ-QZtNr=Qp76smQ!S
ze(-l$RgoE;rwkI!OEIIF&Bui{8^VS}0k~X3LfSey%+V|ClhZRZN|av}v)qfGt!>08
zn9Sy#c{Mx@t8^1eE-XL#NHW-E%ta9?NBy9N`ddlK=cTIrmkz0t>)J7esG69!R8Nly
zgVO`&1~j~~|9wo3I{w2ZN@17$^$hb!k$ri4hC)HL+_IV)4>vafg3x3E*xqP3OX;=$
zv|Gx~=4N{BSeuc(wYBy3(-Qj8S4^Y7e=7p))3FrKSx1E)$ENok<+8iKB@aMj?c{4N
zf>6+gB0Ho}-RkuXUcM~-B`uZX#DPyB`%3m#+IMBPRocOIT!Y{4|Gmlxc$J%dWZh(k
zU+co-Di99T3-|Ww7xCkwpI%)9bkkwT@g**<XT_n$%E#0+K*Pq2$Q01K*U=k1TV(wQ
z%GoaqG@jGqRqNVt#Rmig+?Et@`Ku)@{ZYFBbjfBzJ3BiLGVIVcqm;F!z0=lKzd>`|
z&MXhVfs(hcU8Hq3mW%N>Kivz2_(%SA1HDj=c34N&sMzgf8(Dj0XV{RRZBo7<VAB$S
zrCq?c>E0V-QfbzggyIVt9e}3MUFteI!aO{n0r#bdCQbl!Mv55_&7g9t2{D|Fh1LRy
zV(ccVj<p}IB<_&Kt++Z-spro}eBOb}s--p9LH9|4v+I__IL*JJRsQ`e3|$Y(QO-2N
zH=l7<5@@630n#_DoC*vKlulC>AOkW7;2C2+c{w>>tv74CjDd(jLO~JHu>@EOQ_Rqs
zj*gBJ{qX#};gd(-?<26#q^+%4u}-&KE<jIViqV!=sCyNskQopd$UskTF6jv%*D&I*
zi|6VGuZ?j2*?D)^XR5+lHhpigEfh3Xk^WK^0Bg&&t-+!yWNIs3NOhn>BK?4D*XS17
z2_*dk7YdmrQ(GSOK2hQK$)7#g8pL>uDLWWmnKA3`X;MRDbgz0p;-7NjS4gKQsgs4M
z*`~ZX?SL;TwX)Dx;$~6jrl#&g;E=hWG~I=7y$7o&n?5)#4S3n+cu~Js0NS!ng%!B{
zCl^<*w}=BKsvWbN4WHC_r!4`x#h{!dznZw551944cg!{RX*7)Auj+m+*ojbd$b5XR
zi2GUmB5GdC&gZnm0loKClTl$|SXS#>6?1}U5u5_Nlr?X$RB-d<Y4b!0)HxdAIhDB$
z!H6EFhabmo7mYv9*3$5#$}v1_VSsYfe>jmp@hf~)<wsHQL5AA??;p-{$qp+%QcSd9
znZ=U}IXSuLu8fui$alK4w6<0lFL<E(U4aHrSY+koz<Q%UHtSERKN{0PU}aHH67r<D
ziTG@c<Qj7^;st}B(qT^>^#p!|0ls&e5fTA`OZ58nYndf#{rE-wXJr^#8v7IMv^dw5
z1I|9z6`Le~5>NITPm3077Pqk4l7s%0N(C~sU2}YJP>i{{IDW^n>aY;p_a3q=pWHeI
ztoZiF=7t8byisipjaYXLT0C&DJ0i#tmtOSIxuvC!R#q%9MdaO&&G{lX_iYUZ+_JK>
zm-Yl8tC<>J$Gd)`mPn0<DJCH)NwH)34lmnc4C8i;OrA7TwAcATZUjwkr9Q)g-8G*>
z^=Yim)5^bh1D;tpx2~~P5e6(I>ey}LAS>N^f*<9<fI_GYb(xz!*ma@Rj7J!4?ct;-
zzPM6Tu(1(QR(UxubFGiFv-5dLTRFRvh?A073x{kvXP&*DUh=4AucCyi>XMY4+97kW
z23AXuP2KDB(V|fCbKF`V)O5DH0Vq(GV(%Dl$j(L1VV(h@_qq#iu&>F}%DI;^I9=_l
zU%Ki>mSq1GA-XqD&E`vg#HDWzA0C!bp+hjZY>tr||C+A^72qBv*cRoJ`GPKx{Ekz1
z@9NK{WB&*_wz%P8Ia%3~iW5-dO_TNi&)dZ7KAx<WD|5twbd{A-i)22S_qe1Tlnr=V
zED)M@kL(>!#hQNlulyb6w_<v6#d}>Hjo+LX-PC;lZgi(c8m40-!$$R0q)%jhCDxlv
zfo^QJ+nuOM|8CCn!Pb59$Tp)id;}Zd0?VCIU@O51MMda>t#G1!+W0xgFG$8c-4%iY
zqP#YW;xV3Ir&rUe4Q7kWb#uD@!}0^Xc|VCy%~HJyAe*|qc7Gn_Of`U>#oTu1L)jaG
z9DU&@;e~pU9ZLWVc9Zp@KNjgXWAkNqI%~f*MJPlOqkx1|KsEynr*Te&unUxjVFU|7
zC~{LM_5NoW#xVV*&QEU?DNg+dxcfy!(no=~iOHg2SF34i{_oqHuiziY14e&yS!a#@
z7zi(-;0<hj4iaGm0`Xh=`viT$z>y!lvrbTc{DvU?)Pc6)1$Fyd4kDmUW`idy%hc4=
zsQ4=tI+6~%2$MzpQ*ZNh-2V5u7{l|I+FvJpX{M5SeaN)8TKkYGD#*r)aHkf3Ipjn4
zhyNNkuNfo53J~<_Ki<V@*!^{XX$fA^JbFy9GgZ0s{w~F7r<_=c-%J{>LazF`mso*D
zt?SP0D8+q=;-xo8^)#{96bh1|2C|ETgM+0d6GddIpkr2jee$*sdk(S^m22X$5(Yf5
zo?x4-dqqU?sAi6Vg5U20RWvsVtIl{+;T-f;=k~u#asKl*%H>3<8H`kUz}uM8)yB%x
zRs!yXiX*bC2_I>K0=q|^_70WRA~umnj8!yLYyHbbcNhan_$^;87Ud!zO(Y~L0mAfu
z$VH^TR+lZYSHm2+B9=w8+WIU#lbc$eGnSF4|1fev!XoM^kYmB8QHZz=EZ7-~u}8u3
zpSx(GI<OX$*Smp)!V*rs0Bk{@<Z0>Y=}SUVQK0?<+r}#w>L{Qzva+($38X|s{9ny-
z^p7D&M*nvl3`^a0d4;VQ8FYL<zYncTD5-&NGp(C+`Z&{W_w52kDF~6!Q$e#C#mm-q
zr}^Vu<ktquxoI?u0WlmBf>EHcd2rU3D$X7yKV;7W;YxwOr^?ep-+*$<(~&-C{(Ue9
z=pYwYJbucg<m8z+>!}iH$o`XC|Jht0i0?e<X429v=SQ1s<`bi%ygWR&p_5rpTYGO%
zUO{C+X##RdZwd1Yy9{%{FF_Z9d~v+jrAuSfwcufS4v9L7U60PYH`T%8s9nYy&np@p
zpt>QObT1OQ2awvV|NZ;hOEyh)+I=k<>`;^U$&B1PjmJDiV`F3Ka*_1WUEt%2N=nqS
zhZ!G**a?UT36Wr;p)}c4uEpB1NA0O=YN{wJV|=hXTp!%nuq@DEj_xul{<FMH+Y3*2
z+Uwefhq{>{SMdJG+?;K>Hi91AX(Vio4T7kkXEKuNy;RN6URCqvV=<gP-LuAVR4<or
zVw?E;tBn5uWzF}WkQjMA(BFV1aRJV9)mn<@zQ&ZSIkt9z%Xp#QT3>RgY&z)iuV3Tj
z<>m1ML9t|sGhPu0B(S%)x8J9Qg@ybP93U0}K!6Zk`>)27$<ayS`hfzl1I}vs-47UY
zohuSo3SKGL;|tv`Ul-3f{;AUgx*^l3jX+cjr1Uv0tlj>skojDg$|;$H2$O%#%Mpfe
z4i9;hTC&t~M=XzOoHsR0+N-|p@$mA#($F|QflmUZ(g4Co{~(p7nyb#27@n2D{$FPo
zWqJU_TPLSS`~qLtq~x1a;%ZTol}|i^$CbQh>Ci{aHitL$vY3%@YJIZlkvhK?b<{2~
zrEZz_7V;kDs(5e2>ezHmzkO!*$iRIi21pua_c}tx{IpmIF#I8jOf|4s@F)c~lNRWq
z9ZO^#Km;t5bi%1q%T>%&+jG-pim9lqT&u^@o~Ce<3YywNgFx>;^03Fb{DA%|Mq&n4
z|K#q}h_}9criKScR>(XJ3?x#H;NRNICl?lY1`$gNmyy^jD@e7Ak&Q(!bhm|}rtlFT
zsyHX-ULvrOn%<+rpdQlsoq9EZI(<B0mI_2wul2M*v7=lS>zNwoc(f2)sYmw${h#X@
z=cL}g#0`D_BaAcQ&rxrFJ;m6sK+|HaI-jfaQnP+QB*C4snXPj#m*ZiUlBbUPq(|z!
zdguMAw^OySZr}Mfy*|#~g@%x-84=*Hb{+%vVG(c3U@~dSbte&y`GbOWLnNvw%X&o6
z1aNH!C432(G$kpiFeukR+EGpw!-ABv4-na@KIe#RQW6phpVNgEVkW6|4>_skkwn`J
zWv69X-Q32yHgo6iQ#+s(V$&I%%vzR5gopS3{v8$`4#+m*=-fXw@qb|HFL(if44_8}
zB?DilKS~w|tcOQOfT2B2e-S0GuAu?6-gocbl_u1mZ{+!XA{TZ61gWd*SzyZp^xOa8
z<zee-o62DX=tF6$Phu3hRva?bUcY)}!j~AI6qmeq2BfjL9(?RTfY$(U&Q*7*^?pBR
z%M{}Qv`i700;h!oIB6By1@Urb<iUgJC^swkI}|5(i)-fYe)##xCl0fPV`U|5Tj5*b
zFt-qS`Y^XPd63yzhwB8v+J9$Hb(z8~vG+Kp$ZLh!;4B2_sMlv}IAj8LNWN(&K!$Aj
z3R9PtYvSz8%(aNDM7zht+S0e)`*ti90}o<xLhMcYD{9Q+gPW_Dc-XyTzNdAU_ir=>
zTxIuj<s|esnAd@nf=wyLg8mE0gf>7wyVIb8_CMhHkh|Fs+_i|XFhII_`1n>+@G#L-
zk^~+^b|B(DPyu>YG%n1`>)e9HzpU3J&c)?$eR-D8aPM()a`Nk77|-HTL{zk3eA9Jn
zGKRRtvwjv}M_O*C%)vBeW#zGvk+Smg-m7F#y!Mx$X*=64wTH(KUQE*rp4IErDswXv
z=asU&EPPW|lSe-)7%}KTg;y}mfjme~<XToiLjwc5<wZ!H?P9XpaSa<R%I>^mjqO87
z-)j$eSzhp<Q8B2fs5oM$^FR)CSTn<AV$4<lk)Hl>?M#B|r~OhpxS{911CESmJvZ8&
zYG!F`MvEPDaXx$S2i=BSsLTm!xmM=p>j=a>e<@JR+~4l7fLr>f_$?1DXxeBWC8?tT
z!5ka}-Uz}@0HShrxB#jvxZQc98tUpWI5DaH{G6N*LfZh$h@cerIq2m#D?d5&!VZ+C
zkDl0a$yAH)3*g}3ka%kf4wjytQe`l39*T>LRi7-o)RJS%o!0`T0SIOa^wHE&ou;Ms
zWDI!0QfUJN1F9?pq!bimn@2G*F{R$wAE=UqoV!;XP$YDohbrabo_h?#QzLgj=h_#b
zK#z)ZoGo>U!=c9ysvUk=VKINv54UMDEh0r23`C|;@u-g5hbiOedd$?O44QMQBDJM~
zZmkl*c7TSgC(97I*(pLmUjsK)-)+G^H<uCm`t_?DaE5@OM+}Z95PUAd-PQj=@j%<5
z+3-`A#qT6O88|QEw<1I8McXfars#WS`)39KzU8-$4~>tGhDSyYL*V-MB?@*e*wIEm
z3wTpuHDOe&wH~dJH)>OvDqv4dgn4{)#2C|^e<?a>4zz`uf`af_sdDWC&|8oiqDTFE
z2}3V0k#`m~Eeit(C!~+xoh}<n1#E|U&l6La;zvP&sKc_n0GUg*E&}rL@JQ%8vz@6K
zHRj4B)hQTj3B<`(%kApwnub!`M1P&;%pWBs+p9c;eCgMM&el^OYBK1M&-grqp1h4B
zH;a7EHI}3)C^tJ*BK%#|qip*7Ew@+`9_^Y%awm^~fA@DqgIYCGR?{O{90^O{u`ihh
zv~`|91*mmJio$^Q?1DA)xtN;ssMm*xNJ%fYT^mDDM3_v<Fi9apyvfhM8V_%%2$CE1
z&=%C<?9!rV+*-J>ES>>P51iJB2yATvAMoQGc>sau@hdDSP?FpEB9->1%6<v-=jZ70
z2?^bk(}n6xKz*cxqr1SGI;;;omc;e&^enyQCM6H*oyH;C|J5zY|3|jWzVU*Zc1}>C
ze+=nej`Q*4&vAyxr!RM$aRKNO^<u$ByARRA`>@QQyykE_V>&TJ5<vDU{p0JRk4+qY
zZ;bc8_6jmpdvsK3$=i#$%A|$CY<8&r@~!4&h#ADUVS^O)adS;B((fExv`NA=P6in<
z%cCsJUvkgRWh+b`P7P@+@~8dl62D^K#T9|r0uV2*uP)%z7RqaIz(nlko1U}RKf(@-
z%N6#El@Yr2+Fx74oZy#X>cEs<Yw#tdK~#k`+umv5QWXGqgM%;N$|+Gmt4b$K-wBF}
zvZCLDjI+Kx2S(A#%F25QDk4l%a_veRz3pQ<f2lX66SA_hKve8Mh$p68&n%fNMarnx
z6YT&#m!w4XtUwaz#A5E>Lv~x209JAnZCxt7hi0DWw5udQgEfDzsjWTs8CVl0-IwKp
zgTl~HziHTMF?mAv^Oxm`Asp%l>7#nHafM4_=p#M6c{&xoReF~_jreTNdPn(G5>2T>
zj$F>Gv~P1C&7E*o6uyZ;L-O#n&!vAYYJhnIP$6&(PJo*nr|)4=T`O(QIdt-s!<U_0
zYAt?I83Z1<#Jd0dh;tEnURYi2ZfO}3`-K~hS9CA!ZjLcfj#uQHV$aqtz`lJeet7<i
zY<ky}e)e#9Sl9rwS;gylb{J4E8R|6EPF_(_PQiQ7r=WwPdKpw_X<BMatLMQodc)i_
zl8s&5EoUMds=mpN-fsOhM{L{sVf8v@LC;M}WV}-BITUrN!@Kd{_;;uCD<hPkCV4n>
zacT!8Lb<?Eu{!Ja&JwzJeV4%vrh|`ue5W7x*<YMi*PEF2EN^B$4kRLuDOtgoe_Q-#
ze*T+&k;;w_(4l|<Fu>&;IPkfN?vPG?y;|$CdWP^R9c9z=%pojNJ|`C!08O69v4I-J
zXv=-xLmb(GL}P-5vgpq;)!_KNX|wyC!~JG=c3<nPi+_^8&CnhwD=V62M;pWN7$sQt
zrRHioK9u+}nQ${99m7PT;9d@|Dk{rklE%tkqSh%XpLX4c6srViP*pj(1ub|EsH5J_
zoR4k)H9mzPBm0c+4rRYKUOl0qw%^zw--jTKO!J$9v&PUNDoh-D_!?figCG9z_I1o0
zN=wjj1^dNM<ndPD55buSHq^}EV6nw;Rz7XJo}M1k=sG=~vK<~x0c6PF=6qayFIa`e
zB?FyRT&Sz7B)_ig;a3B>+W>JrO;xJfrX%aw2mZMe{?+=rIY2%_d!AWN<W=rb)4qJE
zhc)N<A;ByU;LefTdsP-=C(PAzq93)rQ|qHRFd>c)suWnv=?_e3GqBFH1(&{NGHsqG
zjn}Uo*0S{pv=GlEJ!Xr!hq*NBpwG&$T3?u%-81jJf?Z#|^V!*fg+BVCyZoVSDJ#Kh
z#t`$_EW7PX%C9MiZA97}J?ghXbFgU&nTrB84i2A9VT>mxv}B9$k(DIf+?giauayUA
z<9;{>k~Y4R;zYSzUtgz*6hT%f=yVOfUi%GQ|BmMd1+zE*#pU(Ij`->PM+TJ?uZ`Sn
z87=d>F&vV(=3yDCPkt`H38sqLHTvujt$G}=`+0P14Cz-7Sh-YZ&Op8zzQdN~o+t2j
z@)qFdmBJS(PigU@x{RyCY2NpE4z7q2o<)<lZlZ)gAPRsl`h$}mZR@kHX;o4{-PCF4
zX=3%`y%?0gn`_G^$~T|H;zoN;E~lXR=13t%*RzhYi2J^Nl4s|!o_0K+>0w@Oi@UKf
zC-}_EZ}rBhvZ6a?=lJNMR1LpX!x_Aq$bCaGV{?+ZZS#M<f!;qIVj%P|75Y+0&9?r<
zu5A_-NFivPx85){*F#8EL;KmCySo5x2)q3K7Q9jlgz`wQWf~rjqJo0@gFfM8<+U{F
zzz@yym5o<hAXIa;1H<<M!Upz%gy5u%Gu|JEEeph}_x`XFqd_>tL1n;QjoIDZy|K9o
zOiv*H!f*<nIY@<kD(;~ntft;Pj-Fc?YX9DvnDWB)VNsmhOkp?4PjyZ|d4mh@kXgD3
z&nYWH*90aMUwpp;LgnX<$C=SdIb8IbknfYW=Y{IV>M}GKDV3i`s$?!3v^zrEu4~1u
z$|s2;;Aolec8j+mEt}ex8S1xu3BJXuU}0mUeQXBSZcPo2X$lDOK)xmudbKVqvtPC5
zH$J|Sp<Wo-a}5==_flW4Z!Mhrxh#!rr2`~tT0k+lWDq}lMp6%4Z_1lx266YLaQ>{w
z>*LDCOvWgr4mEB7>lJ9A`VJ%N0BeSU6L|5F@^YT?dVz8bc<32Ra^_uzXtipzu^b9?
zj9I>SP@yo{+GQGTF#st5hh4DyvtYM=G~qbul;}Yn1wc+G6U|*9R5xCY=>!{>sIld#
z(8)AQFg=xcg$-RLoAqYTnG{A?Epb(se#e8HeAr8Ud9m%|F`-9$S~l<@Hs(WBgJcF3
zCgEy#Gev4%^xQ3N-t<w<B#JVjEJZ8`KdSS>wh;?r42>x$M1g{qb01UzYdd=w6f$z`
zz)?%39m`-~GJR8LI$&eoJqww$E9UZhz9XG-`__Ho%1t2BA{WQG5NHYg@by9N%WP;K
zzS+6CIx2!i;mviB4!jQ;zCH%)b}xJu1z0=#nQ{667Aa+_A#L}@<MJZb=9~xGXkVKP
zc{rWJmoFz4`}A~<Dz##*-@Oy|xvZi>2T_UFa?DIjW!eS(Yr^n$lEC~hLp0)<bMspD
zw>)_}G$?0VfcAbM<g4}LTLT&N9mB*`@8wGvb`=UxQOH<J<ip5E2VjBxK5-%!iT%5y
z7uD2{W4v)&dW`+J1+2<_sxcsm+%MNxH+QuoUgVZ#D=R50d$fKY$SQbku?y621mNcU
za*;;GRn^tirKQnpNQuX0qQqaUNG%s2Yw9R@5sLn>g*%c}P6#}#`RDEwDVT@&uy^6l
zAV;Ko2of(WzCamTL-|%ybKb>}`djFWJngLe8aY%2I1@-7apO!_RS1Jr5XfPQg4Eoi
zZ0G>jFXGc7TFV*QM}$qxU*0kM-i1Ck8<tM2GubOt4LtMGUvz#LY39vyoYI@JeD<*3
z+jKmGem$<k=t1=_i(D*57{rD|#qzNO9oU6Ub1{C5(4T~fP+-Nzs1Wtq$CDwU!V1dG
zWy@#gFbP?d+W~_0o(A>$*<+jTCa*c7h{(t=16zPLxpwBJ@YT3YHeL}^RZmY;P>>P*
z*I>FF5L*BgR-^~ERCIK-Q=(9Zef8*S>43r!1z=}R8k(9gFcct;ctw!h`#?no%(}ER
z$pQ_A=q`{t7?oe#3l6@QKKL~zwdm0&|GN3o<W2VfqAz-+7#DVq$A4c#Za^;JdiUrB
z5ckt-=fuwkBUe-YfRNZmUQw)%gQ1h?^Zur{kb&Btir%o|55E=hJ-%0Yi+|C-j1BdZ
z{sc$u)Hm-W4iSC+-b15uJ(rrP6N_qf<S}b7M-tt|5EgsyZM&3*RDSeQII8GtLe0Pr
z^d}<{qe5`?n1Rn|z^(f6nW3o|VTyKhyjPlf&n^qioP<=LFr$d*>4C5GSaLndzUBAb
z+<lN|Xf_mZ-=~F=#JndBIsQ(T_ZV}Hj>F?8M@J8&fJs$4d&x`ETVXv#@R_Nky!@y$
zipK8~kSuj<s{P*({XfjNb@6A8k;b!Z8rmWnm(^Z)YK%=zKmT2^{GTe7AT0qQb$y(A
z5Pm#twys=x_EE;w$_R>7AV%Az3o&Us@#Z-@pd|d#9l@C(1M^p~PkjxMl#&eUK%icq
zG*;I^vrF%dsY~ODzNP*0C@!Ykc`DFoHXa}GCUK^X_&e7fjxa6~4S#9grIro5l-5<u
z2V>!wMAO_f71~9zai8R$GB${i<c0+>Wg3~DC;^Q{Ks0@ci1jY0A$~!8w)p&ftNK4c
zHpG|<)Y$+SWO=-13^_)|#)k0sfiwYp8PJ_?yb=-;n3$O_E-pYb;32`kEt#g8qs|0W
zXud?k)9*2HapT~BC=XBk#UOuX@gfEvE|hvdD2%*E$;>RiAV~um@{oX402UNv$d_0;
z)^D&a%Yx=>XxID2Not%_3LR^H#X%4r6K3K4dv3Z%ezI>{Ct~|tDtesp5lE7tIS^nJ
zVB%p8{6Y{rj{rUQy|Ke)44)$dXQ^_yPIkMQFw?@>`D3}O0TIvfJbKx@D~bVo#ZCI;
zh$Sx6Bj$SxCDV@mqK*c`#vlflRTrp@d`fJEfX6#$*Mj9}5Uqa<)Oa{~%Y2rZQ0Vhe
z^Ey^rY7ailhU_JUSpIi*d3gcU6$O!}%`?>}PK&Mgpwodsu`eew37;5kmm^gfL3ZH-
zRa3d}mCthLgPRzvvaPMHGWrQJKtq8;yuK7p$`7vCa0loOd+nsqz}T1uaH@4%dLK`^
zlKy15Ie+d5_qT9EVG+?@$DHk7#cP;5r{`Gip#^nd0kO<IXzimN`RVdvEgO0|)5jj>
z+3KG%3Ow3TvpI!@Xcj3VTjtKioBSCOWK8?hSwr=lNar&x=7knyps6_yl;(+v_@m1n
z_C{cxq8I|6<gSvDh?n={M+nJsLYeGR=DXTjSWu|GP;Edsxv$;pT?RKOs)fG|36!0B
zeA+eI8ZU*pID7u`^n_Gw<IcEKvT9cN1x?YpnFV!7dm$udYBAn>#0!z<m|bSy%{|-~
zxc97P24H$xqWU;`7d_FJg^ctu<UJvL)~^E|An}|zD<g~VT$#o2yQ2DIusz<JI`CZg
zbn9t!>ZPApr!XqKzuW~PYbuc4#a1P|QO>(RB;j#|%;T>x&Otb&8h-q-nW~^d&+<==
zEJEDt)w++tx)3|2fIJoCh4)B7*48jlTPCrRqdkP0T?zf(?Vxd`I-6Y&#P<{9CLCb;
zQ60rqevW>7p#L_PW5Yo<8#VbF{`f?%hcn5iobS%JOn{<nY7^l-36yMH2u(((f;-Cy
zuK(8@3XsiVF_-}vNqd?;i0lG5@F{9?{biqPgCi20{;BtL1s3BN<DmIrlpF=%7A!qJ
zf8SE-e1B8KdC>k`zo?RDDRZ;KnG3bQB%cZ$sAi8adf{TG)}_4+QZppU=6SL51ApC!
zsG>T?+NS=JuAea0QXf~o33e~Ih!NTVo`x7sMu}@yDA6iG{~HK5^OTO*li;74T1A#>
zaMl=s%89ly&NWz({HuCc%3w!I#n5-q{Hg0qW75Ze5NT5NkDOg-+kWdqmCK99n~Q4L
zaSd4VZ{NNZAXAQG@d1)6H>rxEqLY)8$k}o<b|5Mwc2eD6UQw}Hz0Pa>eh)9%huz-|
zF}ZCMp<EPA>;|$JAj>_t-r6c7-IC8cG-JaL0fn^gzBE8W+KHl5wh>ggjgi~RcZ(8r
z<JB2_3~auMF-v7=W(TpOwdi8QBf~GwSKOV+!Fw~t#)w>2w?;NR&mz<ydjVor`+q*$
zVoD-wtG2q`$YFu4cj(#~(0>zfhlUR%dl!(BKX>Yx;&~5}4y9fnW)B-|&(;I&<ocMp
zqlSQ$L;w5HnF6hTO+{^D!SJRr8U^&*^Z>BOVMNjvAy>T*JZn^(|2~<VM>YF#O~;jF
z{`QGv-`3AN)X=lwviX7JH(&R`K{0dk7(jeIyyaG4qx2<Rj#&X9kw{ok?Bsr0rSQr8
ziV+3`HLYWTvK5f4WxgPaq^f#T7}r{i1fh)w$vc06$0r9`<JrNQdy1{O`6<w1ZzlSE
z_-tX%c*SFXcCQ5>pC#~zPL*5Ex^?ZG46oZ;w?5B%zvdn#-^%g(QYq9DNAs2&$j4WY
zRT67zXkcSw_Z2hwjfH_ck$WV@huz6{MUoA5AOAn{5@IJpcX-i!wFIW;W<-}<<f5NZ
z$<SvUMybN1tzRuYsJT^D<TK!jQEUxQofG#1#bbkLxt?U3DKw~7#r4ZmslzV*lUZ4!
z)qs<J5)tm<B>SHgVmYe3*n&Sg0`J6nZ?@~8e4DTzjGaa8=TNK;h*fF)tft(*?i3c=
zLNb#BNbj;^WOiSY;?)`N;*l{$^<^Jjbg=C#dkuUhjiB`hJB%y!wD0~5uty!gY~Ara
z+8l?MNauj>2Of|DtkN&U0(Z#iyk2H8GjoG2MgcUh^Q&v>8sGzLY;1s5_0HCol9JL@
zfGkZF9Rf<bb^*}T0qu=%c2C=K6jq{-E(`lrUhk6vQsxfTA_Sqm6-<6;@Ge1U=C>Y?
zf3ovHToeXi9)@-f4kcMxSp@~wY2kT!It4M5UyKiGf;&W$@oMsqjoa&E(3wNNShR2S
zY|^S-tQ$&7C-+vhxfB-?4>dBe^q4!vQSwy=+18vna0wP*qdJ%I5XqJ~2d)QL|LXiS
z;27m5Li<R*ggUtSi4CmP&zFDqUcfeT32M3>SpQA~2m=1ZA{O8E)PeivY*6mTn7y}H
zdmrs1Xh$Ob>=}DxWY}`6eBpoyxzqyg*xA_`K`xvjSgPh<q4M0;w{dm6|0O}F35X=^
zup4s_DHh%B>h0u6{Qr{yiJ$puQm|r~O!adRWPpfG>FkT$+>A3WBiR%6DF0%mEQBSq
z@<WhxkjZ2&)+w=+mu0Wa)6Aw4?{cgVw`q|r-Q{<9IX}psjCT>`Fuu&~?y0ucF_DdK
z;s7TUYT~O5(fGM8TGQ~8MW;5RK<_RLh?AVRSgve~{^s+4$sKjF>G*2A)rh!tr#*$%
z=l9L)PPg%bWo&KP^;-3HUDbI=G6$<bKqIkNA#-Tw1avMfmeqqZU=IO3ce|~u|H~`{
zi~eqEi81Jupu)utEZu%Ra75)$ty`dRadn^d6<9=&_C+xS11j{<FymF-|H+!z2{5$>
z1dU6Jqy*Dat|>x4_kLM?Gz+NUfz&&f$){`-i@y`63z9cI3_z6f-}3cSv8nl!d*F#U
zh<UkGf!Hf0wUL+4@aA>SioX~8$z7Hak8{J}T(!=#NXzyc<I1-sB3!O-3r<2a$6HfH
zlhQB#vU&KyNV&%^78Pxm`OT~wXd0~6bvz(vi2eD#CHUgHY=p@<w!nXyuzKnw2W&hj
z=f{u9Dtj6|2>7&{8ym{WLcrgfv8n7?;UdC(TG{WEobgQs=$BmaEEJKTl8=m!n^An<
zlr)_R5+gm&>){BST~fdjLUwTAI?KQmu2^Pao%s930^;jc({6eXAr8Ow`oiC#olC_D
zgK0rhY4WY&M3G&1YCoVRF`=&!v$^}ejW$&`YTl0Og?0(n6~TkQ)2;lUs_XQuSK<7~
zjeFNhfkP&a4gK9tl-bC0@OUT%0)|CrnCRwR<bn>gS7S4?vtqUtaR6JVrNNKS-GHSt
zIy!pD3Ecj=Jv8XgmP@$o%W1<|w<WBUHBwU2H8$OZ(YM<HjL5PY!-6@l`KxYEEp2mc
zc<=L|`EM`%G}V#OQU8`V{vag3)aFWviDvOOVku;k?cj?Ok8+oB2^f|l6LONr^#jv;
z`Mz}n#}?x%7#<;Sh(h5uFlDjqB*YCrZdtIS!u732J!PJH;@2#8y(qYN<U39i#nhzs
z6rI>(X*6G=hK;DR#y3m_1iruSCFRuGWQgY#YZlyr?3rRl@gND7iv2wwYFeVPMjz=d
z)~zsQ6?K!{*?SP~R-<&C?}dy7NK)3-c`a-`=w5LWS_@C&gkLfu4*&rw>S+vzZJoL~
z#2(iJ(n(xgT%UOzKVR$>e`^ojiO?{o@iX==y}!fO3ru~}&zesj@g-Ut8@FyAVSH#R
zu^QYOF02IU@^6iwUgrTr6c;Q77-H(Wy{DT3hOC_EBkP{NJyNH|<e$ZWeoYMLG6zp6
zVXXM(!d!HCdOvI_igI||O$Acr$T5DN#eUQLt%yDO_r4X`D<%|)_^{_2By4M+mIMYI
zFtg1@R+>0W$jK=p8MF)WXMU1HU)@AbsPEffU9J!A#{)U!La%T0Xow{IUr}#shJ5^U
zjmv}nCrj}A0Xwj3qzDzYSqK7KqF{wgkTBxl;sUnLp+`MXok^1kM%?yh0BC>z9z_zj
zrnVNu*+SOYl=4Pbz5i3BAqiT%F5REn<`+Cd08Fvezp@vd7ueqa@u_%FC_Dg}{;>DB
zgu*P06rB%y-Ds0om;*^&P=JZ|>?=v5muW+4Iv2@=SMGt49_*-|hMB{QQjTWnODD5*
zqEVIIJNyt&Ob}l3?I#P>()wh{Qx+t&US)1e*}BB!nQXktu<t_yN)JG7<EoFW7w${r
zcDEq`oufKw+WMieb&|6@3&EVrl(k5I_6`FZ|0Ip~Pp*IIUGu%&)Ghpd$+|DaAt`9P
z{>qK0Fc^HBZ1BDS$bcp)Xl^I_`Sk19QG^p17_%%lwOL`!0hRwHQw&HaQW6mXpRdh1
z&ONs}6(_JIJ~igPwaUE@60WDrG2I^&I*K@XI&@ePJWj}g=#V<zsvz6r+*fM$qmSaQ
z?^i{NKla>^I_8mN#yswagPM~hpYb+gGgu%!k-IgM>;0QL?Y_Jfv>)T;hfc&p!n1UE
z^)hw*>-NuNxE(S>C0Jtch}!c8-Fa(J;;lexmoh_po%Cz1!8uJjF)l%R5UC=1CYa5T
zp?(l9YnI0+HF)9`-Sx%walIn_!(f>~b0^}v3ByYvM_^zF1m@qPTw=T-YPaIUfpKuM
zQBTcz6h2N--*SV1kqjC>pe`?8z66p|v6dq4ZEfD(1v_2*Tljy#v9RDt0DdDF94Z>u
z(RFov9yoFT_msGf=Yo;ueloxGAW?2U3iOmkVVX@g6;8cOco2$L6xRnB${KyR^|vwH
zam=|h3Tv&j71^nZJKf#sFh(XdNj#T2<SRlk>{sGiz>fPos7ut^K?PMpJ?+q;9f8pk
z_Cw@xl)m|tgJb6FLVhl?jxTzB6lQmR-a?*3``Q^Gt$^1B^SMBm25;}g>!*+zBAf20
z_;XnLzM}r8*LNSJWr8@s-Me>zE^CESH@BlQZr+-_YvG3rOXA8<4_e|NlUdrNAUQ4w
ztlE~AmP76&kOJA^PqAJ&snd$(1;o3Qcop+~h<P3D3~{(@f?;*l_Ij{87(8^edB5d9
zYjd7CaMi#eNU$0vfqEL%+?qB%v8n+k0XXs7reD2&G5BQ9BdA|mWE<zZQxW0!`+ACn
zN1mBou#6OSGMD;D1hz6M6V$Yv@WFkv*d$6}iT-~2Xh;vQTEdQA$Fp)F>-b4*a7%%A
z@1Ek@VV?Es-umryWE8AT^?ygfHXPojxK`_yaq5Xazz%$wtG+GgC52w_OTc!+HiS>s
zTo8mQ{-M^8l!mpu(TzH}JCbf&aQV1lEXyIlT;}FMvpD#nZ9sGekPM{6BdW$%9FQ>2
ztHGG-8NVej@aY=NvOIq$`p>Q%U}0i{u^6Nw_2jpyBgkq}p@YdD`sJR=9hcf<WwR&9
zIi-KYal_)4r4vczCBa$UU^+N)(7euVZxM{h^&hbGFu^ej1sYQW19r)@51)dH74i|J
z%)4}-0tPK-W@s*=oAW{H+XoW%hFXx?iR|DO*C3eGOPJf~S1?fl>G;E#`|sG}4jONi
z_Um0ONGf{fk{9=%J{m-a(Bo?G>f>4|&-~0U@|~A7)oG<@4)UKl2eG;&vr2;_CTd!t
zM9c&;WhaT1qd`oCj|}^z>EAke#MUN#(=T(3EUO`L@{r9GIv>*+G#@eN*`BWE-pp`X
zg;UlLA3+YzpFivDzFP_<1%*B^C^asCnOSr7_=HVsTgX_d6MR3aG7&Hzsmlg(oTUZd
z*$Y?E|M$FlG_&g%9&(r(%-X^qKYj#5XunCfkN-VcjT!dk!-#1Rmau9Q-~y|xTjRty
z1L)!Q&JGCUOjX&xwRtNgZ)Ro&hGn+wJv_9F#&;&mnO$8!Q=x+_90J4Z1J%IG_Kprh
z0eR4{*U!t0Q7*xJl7RtbyzTaOgz&c34~p@mLna_$Ta>~RA{01}YJD2M;pQMa-8RTZ
zfLotFXf;()P_;dThK6Pd;E&iDxbg{6pm~ZgpA6M)bu}B@(_1SY`uTJ@t~8Eu_^~}x
z+f5@X?!8c(YJ3fhdx6by#p_r5bv2|RN_h>q_=UNgrL3wNV#axQ0;1>_6MQ2(mu1o)
zY=*<sS1Ui>f4_3=Bq5Q^y<LN~P(BX%7He=;^w8z=k|^<zfT=g1M)7z1c20*5XZ*>^
znlf(_d&lPHC^7uOpsYq-Z=K}Q!5NtGha}QeGuCiG%pCwycONe=FJyqK5fhMTv-!c2
z5-k?6*6yfoLfk4-EOt9MykcTL2pB|kJ}G2s<h{xp1rarnGXSAfd0L0d(>;)WH8HUx
zqNL0|6I(a-;fg0CCI<cWThuXdyw{YEIYFpqZ}BPgsmquzQ7L^8OgaHm0Az1c#geCd
zz^tb14#;TM)FjNRq6fABil3$$w4Kzg)HgRXLmn&xTsSaS3R{*3%NT+2X6$`1c8Sb@
zmM2C1D@J~F=<~utLSjIU0TFz*jJ>)C*X6xN#RIPnkyGeY(OpCD6(nQX^|ifREz5-n
zLhk_P#PKy|!t?!1CXh((6S9HmPNsMl9bPcES>HfpBj{o5Io{p^Ae`4f+%w9(YjL>=
zvvbpPPfgc+N@F9t7dZc81`+ob^v}27Qc|Uw8JE=nm~v^=K)uaWz<1UthED)Q{51jP
z<&%~u8o;;(zLs`tgYmcJ<&1wL)dhuxW0vIr`Vxc=a-c#25zjSWv*&QOaP;=}0*_FU
zJ{k;%S58>*1b}=E7;QUb0Qr}ZFW6uZW^nTe6!G^G-@t%*VWD!a`jACE2(bjbDLkm8
zD$6&AQeRp7ecxPr1tuE`lvRZLLt_uVF(5Bx{c284%_JDLJ3N$09|WWRAT9N~6vPBN
z?14KcKexz35)Q)n<dl@)Cb~Pin<|`Cy&Eq)^gIP=L>K^gUT*H_DIYLzSuuy4T#Fd`
ze^0K-QAg!VCMafs5j+t3aTB4yL>sofg+y1~Wq%+aJG)(J%5zmOL56!pbRaUq6hn`9
z&!Z8z0+0B+etPW<E!LK57l398(mx{ZCE!khhSS+8Pld|?C=~o3hf=QLNm!}^zcu03
zWM0pq?WLG4yAxUpifU_7VZS2!O)X)@(dL_Gi-RH|;YYL-{jZ`ooMc6#o-A61eplrZ
z!d~~rC{W(-d<z_gPJGBv4E8ZL_LD7-J@-(3lD5mF0V4km4Jn1q#1s@+l<ULm4q=By
z9+KfG-@e1&zh`EyfC-QfRAC@$aP;Vl7RxNCF`x^q;)8Xl)3*hwN{@Pw<>?j~85xmb
z;C`TuEtJewIf?#i`uq%xE}UbJ4Gn!Rf1A_L;A3w7gd(zbW;-9EfB@?#C3pf9fI;{D
zi60hfK$&kg#0B%OiQGF)Au`{AI@@psA5XomY`QH;<=45g4>a$53>uJ(HRSMXS*UyV
zggBs!+!&m%KYxhGY5)0Xs`{Z=^EX>_*Wt1|lvnF=l;D?7BqWgJxAWYs2|<2qA2cSL
zD-jmQy0Xq~Of+c|lS~j=uc<8qq&`{Dv3>0zlfM|r%>jnGQ$9BUdsFPQ>d--)@gRA!
zN5pI#M@{2nuyOigQ0Zcoh!#k}<EXD0=jti6_Wahu)b40HbaDzsEF6Z#qHQ=ph7n{a
zjj4<;OGziW`ayx=p^a6@bmvBAQ3Yq&#Dq?78;qLXkyL&12>c!j)6x}sk0&{tNP-pJ
z#UmiF?oe$Gi_paNfYBXbCEHwq1XKeE(t;;GK3<+y>f~Ke+aee~`tgGabAE6iAU307
zOe`$H<i6jK7#a56TPRO1MN|9SK=OCGo16#2_Lr*8oOcq^vpGz>h~tur1UK;nJ69@=
zgCzV=%!;I^<l<x}b;~k%Vn-8<i3p+vGs7lGKsdp|=ZhTehUP(zXBSKdR6Sa%J}wA$
zvxu5`rKt(xpr9(6^Jwe3KI*Us5DUC5_=2EM(BkEeS|;|&Mo$qA%<4Y4*>ZGO3GP|G
z)pN_&C^pG1w{MuomjM}Geoya{&A1{$v3LYEF21k%@%M*>UI(X;DBNOCm7JGxpu|7(
z<-|RG&<#0YE(*r(9dL?Sgmo^!TqLPY)f<#K$>wkr5ZNuSU~>}wC_`;wgM5IZ#sacG
zz&Ky8Z<}tO1CB$fo>_z~L0TOs808@;g{j%O^Y?4C1<25RC;Ae$TqI`i9p9Pd%M^``
ziCM8O(tz_yZ=LEEYq5ZaEcH09=V{|!-Tvxq3mAa}s%ISme$zv#b^YxQ3Qw!xcE>HP
z^t~Z9=WKN*9c}H^<b!yAiO*feAPoYZ`+xauJr~pncc0yo5GO79U^f*dJ}6x$PYb4a
z@q+u-x;!Ph#6F*%z%)LTEG&Doct+hLg70CWj`c&Tn8IE8F&?n{pA!7hS%ZID(jcC>
zk#rCUT`8ldh4vE)xTjQnHE<(o>K|LO>IaGGv^3=G9kUo)M*6frDzX-72R*&ByI}W~
z?Ndt#d(M{!k#`Gac3!vUjue)5{QAMwxAedGdJ}Lg*Y15-r6^@6A@fwJOd({fClVPl
zBxESblrqo5U7`$`lgzt>N~KW99Fa^(C7DTxLgw-L&ZqspzyI%l9N+id$3FHx_TIvC
zUDv&?wa#^(=Smu~;{9z*FTw=8dPElN-#Zs%BcBnzHnRTCz7tHO93ivIf?Snq8M;X?
zpM9W8yCOz&l_ZlTV$JRQds2QPUNWca;8u|F{||da;u7oU*sa^~!(XR~?7s@F8KTC$
zQzN*@5O|FqALJ{rqP|AG@@8R#Q|tTGU4;Yw-G9z~8rqso^E9$`bXRKRGhf2SjWYyn
z#*uHj#>VP7fYg{W_0+R_cWwzb%n=!pgN~EsPPR9X;SG(<KZ|}ZjMfTkx?+mDuOyuz
zJvEd$bZA#ZZDZ%Q>1Tx$LgcA?EF6^-LJFmcQ9)5FuS{0Ep2<3DTfF($hQ8^*c7N}v
z*X_qvl#IUmH1Dn!KSQjay*28_RX5_s_E5l0@!h|f=X49}?KJ`P-zixSl^Cl?rD)6v
zk6+(jK4h}!&S)8?3nm3ccd5IWyjR+x#om%P4(XrEnYR~2<&bgN){97oc6(MVcSFkV
z>EY2skXMx#su*m)2nF2<=VeTvcVRyqm0bIC-*M`A;xkqB$|!^C0hVWyTlLMGM`oh6
z1=RL_eksQ&HrKL+ufQYoOe^RlFc_Hf`sZts0Ep7RJn-g6)x{;xxLH|2x;wPzp~Y5+
zBOBi8XlY&ScKb)i`enB3tVX89MB3bwK;FN{qx9`@c8Z}!oQ@J23p#KrcnEy81seQ_
zhzRR^?^nMr_b$}RuyhE#{s9WLV%@ryxT>|=EyqgDCyOYQd-H+$A;ar&Kegme@1hcK
zIM`7w4rd-)_pP404J#yG((|{HTY6G&7V4(bZBWkefAQpPtb=#V#k%F#&+44pW1}~&
zr#^A-!=@rWhkF<Ib&5FY+fIlCJta<UQNOqQ)N5TIF{xAHg14i^>_P@>v<JRMdY<hO
zJo}2J^IK+X<9yw<qbKr&&l;u-tt$HezF^&)ayTPd)@L|nx_omH4^>}jcB}2=exH=#
z_?F>@p+9yr$aqqEddw?&d%u1yvZ{~m<8&$7Lra2rN7JY0OeARS5XJ&3Du|aaPw$C4
zxH!4~+7YB^3%6by3D?fDmzg^{ZPn2=IC+wezVAFYHx(7t-J?^RCad<<Y`9k04^k?R
z5@)6&^(3o&!hsK8G#PfaF34Ikoru*GYzSfAsCXq??dd!FdzTG^oA&eY95XO7p4<FT
zq}skg^H;$GCN1jWk_#5EZ@A976V4v?(#bb(+<0P`(>^x4ylN_XhL`aH9suAE9;lRB
zd3h}?ObycBJ9Oy_yh+p@bx)rTX523-nud>9P*?2Os}w8fE@$TBvs56zjtYNp(n}$|
zCA)o$i3CB1&*9D@(<?`hR>A9O+++ESg@1fv;=Z_gu+QtQ-kcpHkS=RyCs)O|%BCRJ
zLXRoO8JSzY1VH~Jg_W;;y-Hk3Ni?qBNwe9(4JHXdZ<$rO6n&(Wl`SaD>VT5&XYFID
z+Iw$S^Qmv2*aHy<r%@_37OY03htN9TD<7iHdt}xvW?4CVHgzN=En-o7CUU`V?(v~1
zH=CD)Npf=Xu`e-5oIq2Tmj2S+{q&TOhK5E~Quh`rC(-Jhk>#bO$?wxMUs^64K73f1
zWeiomc`j?qup~ez5aMkVQhnjGq}2wo4|h~Hh~tN!eYQ`eN?UDO5&-+6Z4|YElU!$X
z3{aYH-%Nr&gzngTTVrVJ@6RxfAgDx5u&-#mAFo{}$|xAxC@_9hph4+)ky*g9Y@V@Q
zaxgcKa;LL=vet=<nu~6NI-5UnXU>|%S+E7=|JLAFm-xj8!NS3CZE#aGGGk8&nu@j$
zPlGxQ#31bc6T?eLUrN2FfhT79BV{gz+!bHX7v!a2zA!htkEm~MHWXq8y$}g}u7M6d
zIc#p4o3&3Je0A&A)=-s?Jv~=mnrygsL{BePI~&drmIK~bPQ^6j<w@YfwU2Sp({QIj
zTQlmW_lTMAc3hmKxHuy%wWYB-We={ZGd(}NO_OFc`|_7QGGB81D-&Da?+*2wYpi!j
zEVeIyB!;#K>MY9H>fE^>Jy36+J5gAXrFH$FM`Kb&$a1xa?d#qsnoXNPOJgEYAN@Ww
z)RZW>*J|-$@5IE&$cM4ng^6>js;W-!9~xVbcx>LSv)-uHh7ZD5GvwmDK;0}RsE$e*
zU|k>WVpl8D$F?>B<}FYp$#-5j%!<IXNUR0~MU(EG3trhk-An`dPu*bX+kp;l+Pe}_
zb=TVhkpT6%@llyZZ-2=bA*hbjiEnc7NcW%ABi+`i!ADZ7`K(+9lBHZuT-#A`AjHb<
z$wBG#mTnK_#e~8~X}{Gw^Ab$obhr%`WO!z`b$tCA!U&ub#H7f(7sDEc?%cjjAnnAV
z|0w0ZG&9<FTbGK;X$}Vo<{f-NB$prOi=N4sa;(QnTDRVM`?ajB46;Ba(AD+z<38JG
z3Wa6ah20N;xUn~2`PWqAmcaa~VY<kVcF8*<Ozd*vJsN9SHnSwhC#wnR(H0Mk)l7sj
zlVTNqIVBwxR4?-E<&h<d(iLWQ|M3<mjH&P7cy?awVaV`tgr;jpsQ9E1!$uC@Kg1qk
znDaI?OI}_c+doh|u^23<9?1JOx{>Fq>v0zsiJo0H;%s<%19bpQB{T#Nbbe%@?7eK0
zCKZ{&!dRVYyHY6sT~@0N_p%pcM({vdYm{thd{;&sG4(TN&vj=oUj6OXIQK5_)H*}w
zeZFsh2Dvr&KX1Q$?zCvOcKUr2cDpvlU(W?t!lMOu7<W90b@5nKs0qC^^|<qwGzDr_
zO6O_F$q5Wxf4%BRtQA2J=g*&?$;`iYqz_6A!<=~D5Fo5|d9G;pvPB%f39MfhA?OG%
z!5FH7V66Kd0BxgnzIiSW7Z;)+*kb$RS(uqsOpQjSyujxG@q5yd>e`X3o|PJc_eM8N
zZeN9R@P$#=*RQtu9kxkekgKb!L$)t3&!Kz{^b7Qu8;0IwL#(-H3tQH_E!EXfgwIc(
z4C3m+wK=Y9fS3_g=Smct0`2qKP^!XzZU0W}zVycR>q?(3f`R7)FBcnMSyl;HF-Rw;
z`~6Wjl|lUt!AQ`DvyNr%!>OLlb0g)4V0xC{qO_c!zPgp1lAdm1YDy7KN!p<9?!MCd
z;=;}DE5z*QGlLaeA))Qdv$tM<)`*h0_Bz-@&c@e@IJi0GiIxFX6eB6*MC|i~Tq`jP
zcy&l|k-Gw($HXv|K2=m!R=yJ<=%=JCBg3lm?X2*fzSK!Rp)e2x5ddiK0V&dEFKd%B
zv#@X!XA5K)aPow|!|vUgh!B?lboNpn9+yzqApl5PQ>_Kr4E`;1i++tv*=?hQn;hcn
z7CO3P){}R8@Sdy3){uA6Qp3rMor8mmzAojkD|`ngrKSDe!}#Ipm%zWzE^jHSBW`po
z!m>UyBd{9)X%)}pK1jB7b=ML3l;da2>`2x7?b{Nv4gMhfgkHFxG>DdL<*Qj4Yqo}H
zz68EjL?rSND?|j?%IIJRQ#`4A`<>gQ&wlvnNw=o+-6i!`jFY!)1ZTYP^p@ZPgXUw@
zV}tYA?Yk~>{!>>M@;yJV%_Xbi)_|}jE#FHMuEE{=<F^pHk=rZvnNhgdNGCIqpd`Qv
zi}UF0?v6NS8gt^oXZxv@;<iyvHZ~_&f)ik8Muvc(_;K1q%{_&k5!vOq5)t3V#yB`R
zYYtAs<uI|LZ%^Fj_7Hwreo0Av(S#k;kRkvX6AH7xd>NT)zvC{H%C<?`Iyx=y-a$wX
zXanjhW>tvYpgu#Y22>&@rmEjbbC15Kr;Jt`{%=P|2lkX`R@*?O3C-9OqZe<^iErJy
zRW?;Sw;fg#ySJ;Gm?RnI1a~=LH7m!%WUdlJ@n>w4P-4O?4=u+2ZMep0c>Kg2drF~1
zrI^9}D(sD%j=Xv_VC87vyNZ@MGw+pLnaqaojbv4c!}?d6V2!~^AH<lOqiC<D+bVT=
z7jgLHIt9|Z!K}Jk+o)v5(9MZkoxP{rZWM;*b&P*)*mF?*p6QNK?|XJj-|ORKuN&Ce
z?K8iu^a!a{P_&V-`3p<c$@hM#+MWxO0Ke>{ldCgK3p?SI=U2ZwLA-&+VE<?EUAJz!
zUN`70mx}%MIsx|~T0YT)v%0!5v9W;!ht0){P~GEV&Ta4f@+GU)mQ+5~CdH+Y0Yi?%
zoZ7CglKAzem|0kU3=a?L%f`8?snMwFB3G*tQ~CQ|8}*%WLA{)>$cS}1_eF>~4AoK4
zDHoSN4pBq$1Kaw*!ek;*&R5L{7cf_J_kF*|sKGFAptPGf{Kh(8m-$d<#t{V%ziZY8
zQc-FrFB;_Cx>G-5wJTmY^WvYYDq;MaGR=AZn2;hDpRxo5WxdGb*lLxh5J|sB&8w_y
zNzUV}WT(`^<wPHr1JQ|EED1{uKQiow&QzbRI?PEgxs!*EC0teaHp0=Feg!?+dq9d4
z8a}tSUW>A^A#7mBsq;zF^)M;q{=*}@m4j2*)CDiw&sQHj5X^21-J6qmT~k*b<XUNI
zmM)-61G^mK>c7wWFp+S|yrR7#6|GL}+O?~@8xAHpa8tljL@1@_o)QY|lq{a#xc-`{
zD1*}5@z*lRO`5Mu>=eGey&+g1zBtw5iA5xyYp(j$tZaGN;gB!CU0hL0%U1#1RLRZ_
z>)M9h<n|`f9i?TWUU%B%=s0cQx$>GceJtp#k0Qyo^}cFra{r^}(cNytxvmesureJg
z^S)(UPE~JpBI}66;n?~Z&1+T?CVSLKOe9<OOS=(S)^W?1%cZP*w=!>({MzRH_|7cx
z)`n@0{S<SFPRsAM>8LaMX9$KNEpHZ99oZe!Ov_spmaGRiyY7tJ9@{cJKQ}jUPV=du
zTG#33K9}Mu8M7CbZ&gI5Sev4c4jdHRyEpvpK<K8H;pwS;D-41+t6TAkB!17&&j*Wg
zHK3wl$ny1@)$9ikf&{BCum`0NOnFKEA#7P0u*#IQVEA3`KJnems>R(`aQlAV)vN}8
zjCLNTEy0spe)vJ^*Wn;NJwY!UC2pbQtzbT5RoT~NRoGc<D(dGRs$05F0oRFKDqoy?
z|I*UZsSiC1(`H`oUB*W<?N?qDvTWJhA+x)~>9TZjVDtf}f0V9+4c)%3ln{Ql=vuC)
zB{<@mW%p;0pnkl!)Bjw4-r5}f@J}s$t?attX*1D;-ifHymd#pi3zr~_e{<%b1uyQz
z$@g$P(pKUbD&$CLh;Y$|m4iz$F!2tDTL10TOg-l4@bcJL-vMN)6uhGyB=yq!N+~M*
zFHD}ocjv)@AG-w1iR~_12w}-tTULdntVF6ce!mquD)rbag~=_$)vO<;WAKlPJnHr9
z1nFdXycJ+hwxI%yk;a6UkzSVOqezutyt?$KK`i2Uh6?{aQPFZREek>7T@HV+frfsi
zm%M5LJ=>*A@Q!GF|2_#t3yksHn+~-Hk@<1M5Wc@&wwJH`X+X-G+?3Te3JjFn`+Lk#
zLxS3P*Qi_h+R~Ws+_;^iBM@US%M(mp>W?uf8W<U|wO<f`XAYP+r%%Tl=B&0x1-Q+O
z=q6>w#m3^B{D#&JasvJ!bVAZ0{Z{mygIh2(N%!yHrx4Z7n7N-`8GJ%Ke(VE<f_94R
zK{>g{t+t{SD}U5@{H6THcIP`vS+Xqc^Zz4xpodCHS(%J7jE0`G2!_8IQaO(v9i>$@
zwr0q>7OiHoQE<CI*F(vjpDxOtQtqs1wtX6T?!||hId<A$m7hWNUk6owFI)W?;mQ#D
z6Q_7W$WqK?yDu3EQR{pH?`L2A@W@^FzW%Dk$w{bZ*duCfrD6p;K7PzQR~7h`X3$9%
z6%6Xbwo$9V{*ge}x2o4RycJ>)+8&DX;&!}4f7VQ-BYV^)gg+{$NQ(HFlC1oYzg0Jg
zz_2)9K`sN{-J(qP`xvAaP(gmxc?CSL`9OwO1qDPwL4m>P(`Ha)QU1`R&og+onYuO|
zs`B*%Qj2GPykq*EKvXIAh;OyMy*=`n;l=-w2`uo1uvl!20o@i)FL`Yq8RkMR6wjlI
z?$3dLkSA{{TfJvE%o&LT+v!+QmXR??*_!dZ27IW=eH8tll$6_KUsbBSCj}{K<KC+t
ztD`WbGFd%0GgG?wBbiuiu;_ufj_s!6XPtoBw;Yue0x=p1<mFy+x%c!Buvw)olaJk`
z@81h2lLLdLT(rke#2ODxZ%$n4^mb2NpDG`&6U1-mL7zF1RGT?|KX3Gxays+pkIb?~
z1fh`-?Yq|e{!{H5iEsDrjqTTX;ITB7`T5;D+BLohCpZkv%~4M(az}l(Kg71x<msJy
zHdeDUXU@QZDE|R00>WV6V|2^arBA^r^m@h=b?}$`Wn0QtP)#kgrLP5DXuuMK03&Q(
z>gC`tPO(GkDf8|8_U&2GKv))$l{NA?gjgN37^Kbl=bt{TO>;v-hTZG3vQ-dVkNNs8
z<Et(3mVj%?Cd19n{u*z@W$9!i;d1DV5=72IU5AZ=di$a{TL8X~g-6Fl4w)-fay+|s
z8RuJXi<~1Suk$=H&Og2<4m6)KtSI6>bhvwI!KWqCLZ*)`EVdBbkkdQ%_%(<W;sOPe
zONqQPjk8h=D+wgJn`mezU;xVDSgI?;900RG^ei9P`5+HDTX^N-9YWhZMHvFnUf4ed
zMVpHDqT+>tFg&q97WILB4M_aEYaO5JFMZ#^*F&`iqs~s`t(BFP#5S3($q#s=XFeRO
zI-Oyg1S{Yu<c0JVz3GXGi(fB9z4ex#<f5)Bn~m(?R*@S2JeIwqatF2V=Wo<oY52nU
zl~;#v=klw&Ow@Hf?X+}wW|*@%@E^O^Na;XH{3zoDj3+0Dj<3c#|MYq9Z5k&}zBCab
z!Krb4vcFmi9-xN~JvA+M`B-ls{`TX?kAaM+;8D@;kKga5$dE5ge5<j1(fKP<qAF4+
zc-$_XDKtQevb9%h(7P?7_jGoi0(oRoMlt9^s%K9zT83CJNPd4g63d-%WV-M6r<TK!
zHwXa<(sF=^cLyo!G~AR7ez+^7@Wh^5E93bsldKP-l$LDKuw3Y)82ZAUbSEq8vR^sE
z7B$VJj`C_jlQjYX4~!b!+s$q=^tiQYK|VEcV;cu@rvSVnuzomP`})L=&D85#6s8+n
z0_G%F(i`rGAmh<doMqN(L-!N3?{DzNxT<63Gny%=j^Ij+d6bl8W?qB|SR30gKq5H7
zDDzU%b@0^h*62kkuPtDA3Scsh_b*dNiD|&>A2F72{O1eBrrNRJA@{wVm{`BURQf&k
zq!fEomR9m)Z8zmZx_Dd0g<W^umh&*^^eKhUIbToSVM>}&xpmWC@mSQ|EZ6fVcPMTr
zF2VkC%OUn{q<3Sp9v&Xr-^!mpZJL-D2r4kS>{ZZWRiUY^T|2#q-?F5{_v4i*J${z(
z)5gXSQL~#jh}^LWT`!iFZo3Hl$Lr`P75VRjF(b&R@tjD3(u|8pgoW7N9dhR>)^pA^
zZhe*4gcRY45O!ZeQW8Os=;>qI&UZT{rPOEEKIqhB``e`@pCHDly0ay7@^MzUaiSFG
zzCxJh>gt+E;eS>DRc0c=u7G#P&I@v>zh~Pm)M$&lDe`fhp?d;CLPE5OwTOciQN&$k
zBH*3gBE?*%aWYG$nY=bZu63`$<)geYI~dj-z}X0UB*nrP(thqur<HtO9Z(gx5eYKv
zOQ^I9KlJuSMe!*5e6o@NMN!yT&&-VLPu)gSLD}R&#<06zd@rbE-EnQ&wwZmadzSN3
zyR^<dTFZa{rO5pyuV3FCnqmTZV_>$&RMhk1>!IM_V6W>wYHD>4+^;xoh&jRRtg{SV
zDlyZrH<;|sr!FzzGrP7HkfMz+@zy)T>3<IT;y}0<BPla8vww=|W;I35mqx64Tc;`T
zE2a11KIrXjY1E~mBmYUfbLR#lFnX_b-p?r>A$TFG4y=EU(s=9e{s5?y-pE_zTWjGL
zB#l7aqkEh!{B#as7~E)f@MJM6K#sg`Qj(H8$d$~~X+LEvmVO=x`hX7Y9Z@s(k9T=P
z*oGUkAqyjwo*X&&(j=mNX`;IX_N*EC`H+|legBRMLDZyR1U&4rbAMW{KjK%h{M|jj
zOkUV?nLVZ8;K_uB4i{Dpqos?FD{Pk9=Voil_W{5pZpX*NFz6KX>9c2(6B1+qxkUG#
z5>$ur%h;?>Qr6}`?Bd!3?`)JJ&d$#M{)+0cXlA3ie(NKJk#g%+k*R2II|t{89xjxR
zff-$1dveh1#SS1dQB%9Axra{l2`tdRefe^Oar>5F=VwpDgj)Z-tfOVQ2re$#d!Zk<
zlly3t@7mv#(&C6XUS{44{T6O<0zNBz+bE(*xWjxIb-U8}isB*M1qH{?&pA;0@izze
z^<T8I5<ln~5*r%}tPQB~qew{=uXuNB<BJQd?<PO})@RS|zQV)KPNIe5&T)rQm!bEM
zBDndOt0_m0*ilJ=RVkrls|T&<)*$|1&t8G=**u_q?9tR{>CZd!<cDd%=|GqE^D{E_
zJ=quhVogH%3QR;)b+K0WgdS5cv68cB+6ZnJO&FVBuB)qqUE1V;D`(kF@|JKRNy+@&
z+!p(n>=DPG8ivR9v>vml;M~l)`o3Z!T}YiE*-98Bb|}*7GP}baX=TOluf5(cj}r4-
zPq|byIa#&1I5}b4(whH3MTJoXn{fH^)82{PM~~_o8*2_ipKJ?iOTb|h1B2TkBzF6k
zud}oFI!Cd{C$w})&)ChsyXHK9wN81@kDQDAmjiy@P?#MuoqKB(Ou_W2T8V=>toeC)
z1LHpdE2CvI0iTtwo}S+xB^d>cEiL4cjHrN_3?8^h;qJu42d|zr3OEEil`g;OCJ;t6
zFO+WhbWU6)JrMYabEI;ZQTY*VyOhRPU;3KXSCQr<R9Kg>)SFlA=_(5s{F-N#tt@<u
z|9N#ouI)%NQDJ6|B{14OG`_nbWZ`FRUC<1%EP2%EV%VD7ue(NipSmAQ>*T#nor}{i
zOV-YgJ0XZ=-^IhTSKS4cmrmkty8;dNBX@$IL_}HVgjIe+!#2MqoyR2`%zRK&a=#2H
zd;d?d?u}@Vo`G_!;_7}`_n!t%4A^s!Q(nCySokQ)s!YAZtqt!+^z`*lo;-=9Z(kUo
zAUqeeon=#SQ)gxG6->AW_-E7&F)-A(K)u-S$Fa-uk?s}T|M>A`+s{_8{$bA9@=7e#
zRr%&leJv%*>Y4(z;!%$k`thc0hd1nA06Rw7T=!Dw0c<}#JxxbPmq3VFtW}VgD`d}}
zeGcbUwCI1d;kE;21d_H4sKK*7^Ao)kmkCNp$wi`cF~RJ{*#WH&arcG71nYyEZ>z{;
zI!mO_skD}ZZXT-g;*&C>sq!H2RT>+Hdm2Ng?p&{b%Bd`kPC3_DGR1xm(k{M9e<W7(
ztcYN^;%y=ch7ayx<=W9o(RU8nQRxWk@YPzQi}D0n0r<UE62?yrbK7B129xb0yy)cH
z+1W9T)vDgi?--r+L6JGt%56pt@|M`5)1D<{u$noYN@6m<jO}oj;|j2d{vSYI*pNq8
zTRReKI5rC|VZ~KuT563<LPAVTM~GRwq*EQl!~SJ-U#JyLm^Yp6E;-cRp;#NrUgZD9
zUyIyfyRh7JQdxt=TJzIZ%G$DW#GJUDmUcyoeejWwFRnEhHgWQS<3|y8J+-J%T5zwV
zq>NMhmC_0L4qD)EwD%ApkSG6wq1QwY$i7oqMDi72SZ{;FT;XbW=@RluL+Dq3F{N$P
zXSca>MfBPB;7iKE)t!=if7$q?uQUcyaW`(F_tZ%IWgWIxB;!n$)Pt<J6i3fw*VQ%Q
z3l_I?#7SIj+vh_J_Mi6hFB<t|@9uOkiV?R*aGBY)W4$Z|7Q})D)7kw`1>RR%JpXZq
za3U6XMoJtxaqjS3k8=3>d>tPM;m&Kt=lEl6?@yrvke8SDP-i-IO(5x^t8}tr>pr|_
zhB*RhTZn>PpKPA44W1cWxmKTWa+ia9g|dNn%VYMC{&iTRO5DbvKJ56#2N<>vW!>$h
zB$Lyp59u;c)agtl(mf}NeMx_(_|#THd?gPa-s0-ve{eLXrzz<!H1CS=CbG_7q*ONZ
z<@1U`r_P_}BRwoB;TX2tQX!weS&B)1_XQllE!68Hj$e>skK&SNc_EYXkkHd#{l4S8
zwDL^&7WG(od%=+k&u7@*WysVv?eS4&4o^$t?jklHYE7FCzl93Jd7!5Mc0tcAhdn;5
zC1%0hmlM9MoxPZ<a6vFfHjC|$?)r7#&a3q@JJ6-ba&wmHggwrs<4hs>ipMz5SNa}m
zuiIWG`HSq~utPn#obm6cSH0AccaioU^25nT7sBp)O!S~`VfEk1_l%T~oXo*}?4wi!
zQCLKCw%o5+JG<o9tIjtimLIz%-ThcZOoAAp+k^+0vzKpvhcE?-Yo@5zg#;AnvgDdi
z=NR$@l`jm0W7@~fLIUshh2l{I<wcdNRBOV!I618>?QT{d<kpcs#3&gn2!xrQOU?iX
zOI9{yI(7%`K8HJtHri$`10lqh4-Jt9wF=ZR_brDS)n{yvwMu>drddvYHqnt?+Q~Dg
zxGFg@chkQ`LA`vLez9df``fN!LhSNOLCd?W@hvS1zb9)8`oqpY|I9VEb~&(?lKNep
zZiY_<F(6k0M;40{fe+vs*4F`yqC!D~q3D)3x&xDwp3>KR6w+u2ZZ9gGJv4#k0|Osb
z!H!`_pCePMHsajJO=hUN&?`Dl+wEM(h50>Ou8dDF{zh#AAAZIfK&{-<JNDKX)GdZv
ztgLSj`u5uW(uqfNFtu_fuigFeIt$s#df|J@CozfV9kFSzFSTUzkjpG+p0l%ytG#=k
zA)ikqLf$`75}MK0Q&FV0sXp%J(N5dC!T-Cm@;^&Hq^Qoi4coH$%?|F0-Rie^c(^us
zkf&4ilGlpnU}bN%bN{_pTqj)usJ%xi+kKT$Ply?r{-=y6B@+va3IKFaNxA~2qrW*=
z&7j<sZ^J;cUHmiDT0NNa?6o~CieRrtG$kFPAn@wyp`Yz9+uC}eQvqZiHEr9d!=+2h
zE7L7v#owX-$pe83UkI_Jfq?<Yp_frHfYfwOoZ|EFwZv15WV^aFi_)pcHIUcbfA%vF
z9N0HONCtG;8wU{>Jm{&E^clXzOe7Sb>5{`w2%Rhb?FMzmMn**l|AoB)B`8MfV|P=S
zkg>Z@qV1!!Jf2UfkN-A63NtfzFKJQoz=0~htI8au=D8gX$7x|022i!4*HMZ+u|n{i
zHO3<J^z?uw)2Gpg{2EX@>UJ&twP#-@D`mHQSy6rRXHio5L#Xa7;l4KCyR6;3lE`Q)
z*HpYlV<UN~Z(~;dBq61ks<`h;`b*b^aLXolNhLw7F;z+wz$qrBbV^<TGPTllqhwX~
zkCt>Y#yrJy&C+$E{P_)*1wP0!5(*g0Sx4p@Jpp~_$^H8SS;<%K3)#b78jf8{lhq6|
zSGtG=)VmdUk&y1drjf4<Ia)b+ZMmxR9bz$M+)o#j44ba^Q(it;s1Em??Sjh?--R?0
zeXe0dD0CEne}UMDjY}K^m_+$PIH~Q~)`w+k`EWge24|211-!F(Vtjn?{7EFN*6Ew?
zGBw&SCM6N<KQ`>8<v(=jWnP}k?9b203t&&|T$Cc>$@KVLeao8{F4IN56}wbL#17{_
z%xc4BeX%yfGl=m|(IfNKMo-_?5%FKi@s}tk(5b%|{*zF`QRcazSp*QeZ4@4TlaPbH
z0;e;(P0MEyowHf!MxUK66g_Z&&VmR4y6~mFy#g+zGF?v9nB^i6R}U(de$NUrU}XeF
z%214RRqKOEF}H6Un47PaP54hBM^N;cvGr0Pe0ix|T3T9Z?&}5gd612554Gxl6AV;>
za+rZFTCcR-eYGz_pO?|^PyODB3i29!3c&XTf_waU?HxkoEaZ!n6q{{AmI5y^RpsC2
zJ&M5&miZmBoXs#5ZL*dW)k{rIKC7p<{e@!lVPjKMg<qNbkOz5%H{hGOd;B>TE2z#$
z4U;u51puUcf-K=XOitW7*lNG{<jIq+uC8|%oJv*+@{Z5nIG<Vso)aq$a_ydP`EmQu
zWrBOIY>EF*@EsP0OE*6hnD2AC@;biER((cVmQFG50kgRIg7D?X9DTdIAD!v)P=9m5
zD=ABag^m=`Fm!`_a%;2ye>&`Xq6zZemz*MZad1?Vx5C)6qp5LgR-JV|aEcd33V`$V
zJ?1L>){BQ*J8bsnWu)-z@F71Ef3w=5DKFf5bX&H7_EG52@uuIj^wKOuZB5q4XmNML
z#VB6mZNJlrSWszqY@F0Z<&)7m0&7EhHC{%r4Pa({;J^Wp;4BxZgQ8o8$HvE{5;XxL
zH(Br9y&DCJ=8UD-+}ixFPPJGoRn=f4;VTXf1>gZ|5m45lC1<dIigf)5?+3WI{^dnZ
zK1DtPG(l7Gn_$jR26YdKZ+?zPw<+TLnt#DP<r(3l&t0ZFIHdCmVw5v$_~rv7yk{b_
zc9p?0qFSl@fS8BWeKV!$0x?hTy}53b7ncg{O358(VEM0x$ivVf33mhrF1Z2%nnjFo
z>Boy!`ulab8+mVBxu}4^20q|E$cDF3I#<BUU$nS>{D1ARDtvg^`3qR2&3hkx=8O0+
z(#iE!CeP1-pB0`@+^p8MuxR<Du$6i}MJQSbW^8F`>4mz;e%AX7w<%8g#M6y`A7`a+
zlarIcUsAO9x(4xv05eQUN%^X*XQ+NS_d&`llZVedeB79eQ-fqT@6$6{R8`R<xoE4X
zMcsWUxh=HpCgY!nT05@MDxLW`XPoU>eDy$b$JF1CeKq{UR_e)$#U@ACb)H{w)zi}p
z2q@Rirl87Y<z+;fzn@=eSs9n(0E_|jR|@4P`igGX4epEA@B^E=?>dy8bGB&Ka#*0B
zjk3>t0LnQ~LeOV;YdJbHLc94E6A9&!dG47#ahX|JAojv*@o(2tQbGbtn2&l>b{QJV
z3_CfzGXerwQMj7H7fg@&kQ5<hmEQr9F#`SZ<C~B(Bx<6QC8Y)4oVAV3^x}%wmF^E8
z`lkF+)zj(*(I^}rA0HJ-;%DCH@HWrRRCe%T)~4HiA|gRs-@F?=$zr7`e>GuCzO{$=
z-w&S;5DR_s%Nhxjiz}2Fbw>w_Pt<T!7QT8#L18d_iC_*(Umdr>eRfhq!?O8x_ivci
zQanyy8406L(T+RlV_Vx`>EL4UzD1}Pn;ZEs0l|8wzn$Hv%2@TWzNBc(5O^E5v#L|T
zU9KC<uvG2r4hMRM=h$vA7{$cuJnNwkZU8$X2gd^jMw_m$%cZ;PL6*{@X?aMX(;IZk
z*H@l`f3f_h$?2tY?!Qe=``JRf?^({5P@sv6^Tum{>4mTr{Gv1Z=cvZ`Hd)g<qB1WE
zi~bpU{N-*fN4Ed3M*6>ffO8&Y=u;5n<xl&^-&;x0_6;&Se$59}AXd;@p@(Ab3QV6`
zZGxcXyLSqNNrnJKN3^8i#tGopx#xFjwJ@E{fK?D<_}e%AIp_in4ee3Y1&p}<8ZVeA
zg2e!Xs$7>v6ArA9ppjRZU`ubo@I$^xFXsnbUCPN#Ye*fi{C>UTHE9LJ9+kSg?@R@^
zBS>#dw^X9W#Gi%?zMx5h-+N`d;MbMK>GoU;Vl*?s`r<{^&>r>JC)l~ikKattHapLJ
zH6ktB>}2NSCjNMwTc5d0MstEhfA-YfSI*Wnf3ZIC65tn#K4DJw;*3#V8y3VWR<`rG
z)p2<WTa^3Vmj()2GWM*5k!0aFrJ6!H;@c+qV`z1>B<~XOw7^AQ%z&={1-B`tkacy@
zN<J(PZA%;a@#99zBwg?u48jPCZX2!XY)@P}W?rTFjQ~sQ1R2BZ$h-*+d*ybfxQZ@i
zs4|t(lGpdo3O;uY?0Ztie{I!*I!|A;N%vV=ZSYCgi5-7<bCmy{%8)aq6h#gd$+jbp
zP=Cc};Zna_b1=LPEcdh{^l1Tg2K#q5$0*6lp3wayC0BQN68%K&G=*J5B9$tc5AL#V
zsq*d%^e*c$T8i4PD%me5P}{V=F(XyUU+&Hv>sC^1N>Ou1_<_(ymH3QOFVn;GJL%<T
z7(Cqc&VPCQCVjW9!rxz_@Ba1)?gciY;oYD$>)9r6>s3F%Rd{No11bnBD{5A=<lEK#
z@Kr_lg9i&EZH-Il3FgCNvaV>$NgOS&?q^%J-Z>AOGs{usFt%#<A76C2MNEzsJ^c6D
zoMLDj>v!O1*7%g8qtYV$G!j?dj@Ie7#|2U=?5fU^A1*VnznvjEZ#A%Vn=ZygA4o6V
z_xCrWiVKx_jCTgca};(71x1&V_n8qQ2^|<4cwgPf9GzTXEWFO|-yTfBW|&lC!C<lU
z!sy;9(Brk-qhxlNNyc1uBGcac4r$7R2X^zyyW^C;-R{{KBpYS;Ek8_SeYU+#dFLBT
z;`fiIAv=;Rs5DzjpPZ#t@$>qrEg@cihbHt$zV~RFTgRJx$$w*sQ&qUkRf~!Me_*Ic
ztn)86+zDm*i1{kxs05DFqSq+e+?lHO#PGl~*E@IaWPLtv^jfoRrao@(1`5{_1!C*o
zL=hGgxpWkFX6Btck+t>_jhDec+F=GQ0RQpln#bi#^CL!ivg_0+V4cM2fs7fR3OnU*
z!uRcx?mx6c+@Q^Q$h<jNNT}KZl%?jR1791?+@s?^NiM!{<xA7!@3CbOQ6p+=7B!ow
ztCAAj8-_AlPExa;>E8BtX`@ep>xkxB4f=mJ`5n*7=7?_=D+8>EbY~`;ZKCR?;crN8
zA+JQWj}5sL1BO0=*Ck4PY#3mM{yL=~lc(IAoPz2uUS38SttzA2vKh|3Frok#>%4E5
zZAM>-Z*f{YWgVnwrFYUsEgl)rW57j^GPukCa<~H98g6EnW2$50!#6xn|M}-1;?*k_
zz5@Ht`$JU%8Se{e<YGL^q<ru>LG_=rKKUKvZgS|{pAdvO3o_Gv`}S2e356}8VZunE
z^&qe3Oa1@MMXZ{M-g4weaQVeJj~_p*DbzH<@V6*tpo>se=Bc1b^dQI<J7{Hd+T(nc
zBc1l|zyBbm&&l?fvZB%r+ayrk>_-1m-TVZZwO`6G(nYwfaUGzaF)6o~Ud!<PUU}EI
zaxg!1<0Fal9zYiJ8k0Z8#&&QE{9Ws!54dPRZdvU|AFpS=99$$GufA*9X+Do1K0J}#
z#uo&?*RzC)2?mBZH(0FxJkx)=M$AA;Yp})<wT|F}t#>f$HW~c(k#!1bjTra}zE&c`
zVN*=JMWf_I5r}b1%gef_=)JSZA-g;7&qPs5c5u7s8QQ^ks6J31eBlb0`LQef(#gNj
zyT%B<!<Li@3d=Ai#&_`cUVNZPe4wN?Eu~gfP7f3n6jEYi^^A>oabEy^%FL{()c+r$
zEf$vIY!ncFCCS=nRUqi=0mF?++HI@)U!Nbb{pg)sc*MWRI3_iZU;g4-hbRhF62PCZ
zHE(`9Y9S?z8^dK&iaXfNtpMP3R@R=Pqvy}}Pq!#nz*%#Fm~by=+qJ7|-LD{5@PdLq
z*+Xhun=_&wdn?vf>wg36)%whDOLvFL>il*ohENEakGtpLn-22kz^8fCoqVe|Z^tAT
zVA?5;MJVHcxmt089bH|wI&tG~34SvvxWdFEBFkG>Si2eoRg=3r;f`f#vbM3d-p{-R
z^i$V9Z-um;o}SycnX@!{RiJ)%b`$xV_@pTVk4Cebo?`e0f1aQUM<8eJ#im}5E%wAg
zi@LV9hO{RsOD9$C<&|!^AIH+n$+Ubm{Vt9ro#al7Ht9<<YLy{Y`{rL=W-R-AZBaQL
zB<?1D>S_li0z*+K1ClT*0J}R7A}U6C98ouur+CQD=HIE(k7_Qx9*GhByFB~Cu+_Tj
z&0d;xx3wM9{RFQ$U{=V`dMB{Ma8ns|8;BS~P3h!xDpN1m?;B8hq#nR7fq6MNG@cXh
zQ+JFwE)Yl5la@dXTA=NzR<<Mmhq{Z-!lcOt@NYLb`ce8K7y&E$g2(039#gP0Um6K7
zd|9AXk8OO4Nkvn7E2`lv5CLVSLqCS`67VHQ-L1aCD<w`xm?hWf1;5u1i|jToCBLtu
zZb?l)CaF4yTu`6vK6ytxeQlL6KJZ^M!Bvx|&vGd$du)tH2|V-;r>z|%#Wt3ab^pM^
zRI$4?J@DNpm%=VB%~SgrHCvEtC91OIoXMs4&V<19(WC5a8nOa>0Yf&6{XK<LcbOzv
z&LD^&HsfM<&yrIhbeV$-W2P7Vf=NteFOFxNCk$KF82XS^>F7lLi;Yh{N=f0&)C={D
zg1<Es%!Y=B1NtZ|*&{IRquV;aUu$A(yv8A7ZDn#TvHJ&7-?+J`@{W<T^mLwb{Uf3Y
zgOAWiTbo=R?k}{SYmpr4opN50biKrB%0zun?@{0~2t}?x8bOD9Q}5wV=rgOf6&&Gx
z*W65RT-eg@bXx<0HLKW_jRLgypF9!%z8V7ZZ|8!UC@Xzmb~m|DMkCf~Vo|-X=0)Y2
z|I9-X(gt-2-|kIw0~c+xc8p#-cRa)SV}x@5VS#3L%DY~&IWXkz@PFP)FVSbAQ~4Wm
zZ>rHk$FDw?o(cZJ!08#>K#<tJSDO4^q;m5Uxql(-X9H>fg|GwZDt1uEq^C#aLh5Fo
zn^4_<BJ*QoW8-AP#v(Gf9)pXQq2JVdbbmeyU4xJ*Z&2{Qe!bf=4E>|-daoZ}V9zBh
zJ3PH8`#9l*pzKxLKvIO(;MqapP$@0g00z5pT!`Js68;iOo6zPPjz4VPSxHz&rdE}T
zoqNTmSwfE1&0`h2aK=Ed%;#~&PUEKq4M#(2PVBK6V$E}N<$bYYEG{d{w#pFSWL-UG
zY^@3wG7{^AhO&XM1J;-RqUKb64uMnwKk(_HKCJzEw~4T(m=W{Q?0pS;8X7Tn21D4?
znV7s^$=1%Uenibg(t9N3m8ZA`K3iZbwv%&Bi(~f*u=XgEY=Pr|YiFHr44M_QLACx2
zgp2sMr-s(A675(dydxdK7IvBodjDQy51pbC;Da|cKXHmcO<<&jQ9ocnJ1HUwk>-&B
z`^EUR2O!>|8j$5gcR(y8M9A{-qLKZlx~~K^b=AkAJhx)~BI-u!zf{8REAUy;HEbOm
z_3T;@j->xL=sw6|sT_3qO?xldv3!?~f`po-8^9IqM~bou{fa;PUH@!F+n2Mlg%(32
zAi=+JV^cx{0agEvy}yiw??AU;D+RvLLF@cF?}@*3Bb2Q#Ot-(FuG~7u(9E?xv}Sl3
zo$jits+E<M(4_wMgZ3|dSXe*)jb&HE>prbtDEa-GjttZO<D8X}YL*ytqq?z0yTxna
zvBCZ`x#qZF93&D^=s|O?s*7M_<4L5;IQFQFF|$DdfyTF2tNc+$Olg3C_^o9y>7bNU
zkp2g)8QHk|zXnsYwO>YcPR07H=X<2Wf9J%WUpy)L;B3IQq}4V>=ql9l)neIF>{>9g
zNXpt5W|=-KMU(Gt@po~=_Zg8GvJjC_f(z{!UofbNAMPi;e@Cd+S-!wG!82(l%+m1T
z1LWLrt%q9Kw9MyF&wiemx0VQ-M`_gsV=>t;USntJ3bewqbLWHp`F3{w^1N|vUFUli
zzMxps$kMcC=Hy^iUc;jYZC2xhG<NEc`y4cMbRg{cce6rOwkztF)eVmR{K<|>&yIZ~
zv3^~%`L+2_YM>Hmm0<|({`kkm#U1CwqXjV|Muk5{b5BD{i_O}8Y?MUqwUq$k^y!_X
zhebuQGV`6?ELbO7DLEA5)tHAty&jX~aWQVmy}WDL|2h({zn}f?K7G2rMD`1sF77C>
zs|D5%Gn9I9i!m{z=VlOg?BuDt1&pjLgXz0DH#+0=J%VQOP0TS13yWjNHmJv1=O31o
zbiL2*^wH}t^I})|ekIG|Wg}{O|50UkPChGPp<}c!gzV1`ZZ2;LZ_yLcD*qxmK9pnI
z9W?f$>aZ%t@lvk`qqKFukHXk94D)fblt_eI6jw?y<^d4UvqA6r=FJ-bBa<gVYk&64
zTB8GD`nuKis^@N_dZ>Iy-`OQyxpaxn`*e^zB+T0C>W`j0*&6lYLWP)C8(e`mT|cj}
z`tK=R&=R@IAhwfb@caHE-uj_Gn41+V=<DlyJ$vmrZs(K_X^*5^&h4)~Xq!YIO0!3g
zFf&7t#I!@um5ym2Jh(}Js~CTZG%8iqt~Mpi@o5l$qq}O+65_My7Zm+r?EL-t+*Y@~
zcYGY7j0vm(hh=H(q@Wh#Ie4{Ba=NHVW!E?iSDsazU{tZT&N3C<OAC52OdL)zfByCj
z8AkQk(Sld6UYm-(FiMGw1JX(ODOnH%xv!5`s|^}!c-?b}+87xcegTNm)6?AV<iz3=
z$U8ZfjbEO#3<mJ%p%W)zx&?Edo7LAmlY(s8qlU_hE`nCR)+v{{Mgd}}j;x&9ta!H^
zvH96EQS>NVTlwPZ1H*eTI}86IDEgF;)%JJO6tI#w7W<&;+`FcxZD|=}BO@Wsn?Nys
z-F+qetpozY$*zM)F{c<>8!YiYn*PImX7EC!{$HZQhJ}5N)i-}lvKE`M7>AwZotauF
z{f8~h@`aqsmfE7yN3#sp4D@{<1HcR9U4uqx0NAJ)-)Q|sj!moW|8T&Gs?t^PLNT=t
z4=fttdJHE8=2K3^rdxt7(FovdZm&rHZ13*xzgk&&wS5e58!qE2!jm?THu1M~qNbO_
z0mCP>&7%+7M$x-R!O&#w>UEW?!GFrR)Q5Qdzs>~#i}HceUSP*uHVl2^Z1D4mCEjX4
z!)&lWkT)zm95mI{00P5juy>?;lgq1kyR0Po*<L<;7!s;O8+bxp9rzUTD1fH8!kI!<
zyds{}TUq{K$bgwS{BD_5(fBxn=H#BZsYO2=dc-dquAN&{y<X0p`^!wYdH*3MMuEzX
zwLrh3km#KCDFe2U%T{VW0nLg=re$DaD0O7OP8StBuF-ZYU25Gk<}q#Uso(SBIMW37
zJrXk;+a4loaMkL?g$bTU{$f*%{hlJ{lkSx0ybkC&Ck}~J(5ZlcfOp?^Cu$m=I`!e}
zR}`1n#zYxYwzP8$i`VZ#0P%$ZGjUd<WC(M5P|ox6^N;P90G^wvH|i!QfRcEsoLRM1
zN`#R%>)U@oi{4LxH!(|OPxJCoDx%!>%9ThLX%Ufa?Jf|V0{Ojs<w};&<6O!UWmj#T
z4~km?8k~OTy}gT@tdVg^C&L{=bOHJhre+b=RFkLRIgX7J;}7g!7Z%D0W|SE0e~NW3
zCmUiOnA>ao*_+kQ7@63o|Cg(<UCdX>c_2vdQO%)Wdr2iJ*Dk1iDBWE}6L0AH<s4D5
zr(xfAnVs=8FSB0WFn{~-Wx3nYTdUVQR2hSgXF$7tkUgSmT`4plJUnW;8@zHphcON~
z$?{DdCZGd{Gd&FKz@<I0AyJcs`~Ki#9-$jAIfPj*fsJJ<Dr4opg{723S>@LJ#TQ0k
z(xyI)6Z&7rDqV>9xUk^$ZeAMKtHNe|v<J}gvb)8QvJCanCwm-gWY}J=^7PygX}w9v
zqoWRl4(3UI{thuqqAvOKztke+nZl4_Jbd_22nKn%J)R}q%&M4agGiB>7&`q?=P@Pb
zb@V&+1<%MHl#N7o<pK=~bgQ?w&$E3tiYmRgj9JgBsUjLV#W*C)q5h`aEfe~;O!Bvu
z)oaURuE+?NrWS*fcibF<H5Yx~G2{=kJ1{Rn$TIrY!QE{9&#cbBFhaX5PO*`lXe)QT
z^}jwO$fTQ<i(sPxCg(;z13kSz=wBfj(!=;{>OhEFfKgF!bTnDVfB#;?nAb&$=e_mu
zZQHHUj{#pQ<>zUbgX1%h$EUm|ugryRDRwUI8lXIuU0;W%ckkZae{l}X)vZ@>@e!m*
z@z}UHR$(LY(aQTX{}|@%-Op<AH#cMVvqI{{bXKoP%P-e+wVw!1*E3PO{e_mWU?ib!
zj=#p^FUh(NALme%r2(6D^7M?XW<(l+qUgTk{zEFTndRR!$>e%Kd#8B{Z2X*KLX1Qc
zKl)>o4+K27mmmv;E|h92+>?Cy?AcEApCA?BM(qyu>>;n$$20Z7ufWu;Sx~B1h}CxD
zaz|b!P$mdAn&kuOLXYr<=sjQUQe$AyBAc|$<ZiUtc9ao01Zpr%D!3y#e|bH!&QDlT
zAm~j-?*iR(l=xuYm5#td36Y4B`Me+z_t@Ckw{*BmL@ax_yTkug`z3xQJd72<H^D!u
zADn_@Xg~(YzDh4S&1)bOpr;Gi3>BD<3D@*g*$LXp;_I(E2aBXhKV6*Yk%vtD`o+b;
z;aY4;0Y58%Y)&qPRQ*eEjctxFzM!0+!NtQ<_ee-^JE0DJjz;E|W)~hk&H#2y7VY^N
z`_F%F7v?AR$ttsGneYZlU+H4g+q|y^eXw9gOoJg(3ODq1y-5_;C4c`KjUwslnbr(P
z!{))2={W1NmEpsCP0nk#N%h#p-o7oBpYfIh_>CcK1E4W`=lEk`jCJ6_Oz0R7W&NqB
zRG1uT588?B-s&iQqkfmuH9o=m0(sK*lDp4RWy!l_v&C~pK)SHD-lHw1JSwJGY}(V_
zjs}yXqa(Z_-gRJ6bKAoi!;~VFw*^%2@Ch=~^8dC&2^*15ptxlO5VZLx59J3j6`5X>
z%zpG^&o>EDXWeFLn%%@nV-v2Nf|U3(Ap<=GO|!sMbLC*(F?4#OdN8y80xcgZ$l+g9
zvf0n`_zh^o7freww^0kS(6?h=0<bPd5-^<7H>41N#|kAq5c^yRs0CH+>{$-d2_fd=
z!V)sO`QmzHAHfEyN{p6`Sryc$O*CJp8si2Jkwp`T14sY9eSPvLB%JyzdRW(PuJkEB
zE{>7j#+MNODyr?qT4ihXTA}%T+1!E0IsUKIHaDl?WS|)53!+S0a0i$6Q&vvS7>3eZ
zWZ&RabuM00Bw}SWcG1GrXOT6*T-&ASN$j7H%R*ro8)fvDopKt|<KUzLUoS5<LixgN
zvw`7AR43d~HyJ;U&C>Ek9GA3vFB8b?^k`#6FJ?ZsefWT}Sa?1_R%vDyk<kln#Yl40
zp0bUk5J7c#MHF^Ub;+iE%oh4N(wc!R&CjpP@&$Z)C}xnk?XI$%(62m>3YW!;s2yMQ
z_jFbM!{tH#ZAN?_KK)LzKUU|PY+vtcXXQ2GUQNXjM|=C%Ff>Q2MulHDAYe_0SkgE0
z^PGEI610X6KsPuAP1Yqpk}j<+)G3Ray_R}VH8og=;y-NfKOk;I$~w|i9u7Wjsv&f`
z;IK<^kuB>FR>q&7CI_Xo(iOb7J6?bko1zXjQYHLqp#v)H)MBi8u%m%>Z)sS0ZZ1QM
zNZnmR3C(h}=ZcFH@5Ay&Ddj9ZdTHQCYXL&AG0@z(1W1ZM2<ZeU!xtrtO-#O_sQ3@>
zZuu{^V0`v$IHTO(+YJ5Tl%HDY1ZroGjgG3HB2$jzLQ*I3i`;!_CdBT~9Vx8<bJgY2
zlx^*TS7c$(SqnP$_vytT(s{fyHa5c`V9#I-3lFS4AqaDx-^d4rR&Mc+B(T30;?J1d
zfiC&vi77nF@S1_qb(9gM77AC*f!7op@TIZbDBnt|l#^c_n&aD-$baSxatEK35OOQX
zNReC1<_4LHTI5;PcZBvZk>ET9ak8oCeAt0!vhtn^6Sh+9oY$tkO22w;00;NsXL~p&
zNGHREngoL*q4@VO47+!VZ8#c4FkM|;M@LZU&}{aKpdtYRMbO;Jz#z{_55eQ`G;Hrt
zbFU+`wVR;}0GBUBFGJ}yAc>Vz|9MSm`N+OR*ghZf@xHzQL!UjDeg~>-&ANB5pE5e*
zC?hlpD;Por8F9q%K`zg(DG&&Hfpmj4p}fOcpOJ})Y^@qANPy2tnbE?ms_IZ_olk*=
zaH&tf9*_x%!OJ-M5@XO;2)_5;*H)}_`f{$nC{QqyAY@e|7{5I~zroyLB{e@=XMP?O
zbqsyVyV&rCLr@zy!;VVWbmX=tkKEb`JVn;CHtEnM-jB)qq_i-XIfyZ#A)a)nW|~nw
z7Mx|w^==qyo-?Yv4Gnz1aO{>~{<Q9c=U#nvkmin(8T_nRMHo9B4Oo28EVV#h`C^dE
z|3I%tC!J*4b1TXQcCC~Nj5~nigX(vX24Da#7X-aCXA)rLFqIq_V%^8XMt^hq%UMTo
zo58BsR+Sl#^i@Zv0_u;Jl*5$&b99DY)F35C=0$>V_|soO9NEc9Nlr6qDJiDL#@^Bd
zWwbCJCvCp@iPYND=)v_F&E8_z<R6#6f+hij5@`EaIt^V6ek@N6q5M$c6G9>)tLP;%
zl0etT43L)L;U7OpALvE52(t{|Xj1Xnv*^7uspv5$U)+vw4&QG@=RXCz0I(e&gj}E=
z*!ca4pa|-lzXcM0e}BCAzQ-kT+@YtelRe~YGtT??Z+gt2BekJrzv%AR?EE8+m=0=V
z`@XSp>)}H-&791~)Fs5PTL;|H;g1nEdVW<<{Vs``b?02($&P837&Urbt=X9meT92z
z1LNbHk_w~C+zvZ0xp`6b_Vh%C#~}fs^jzc<tD(svfB~cAf;4cR4Td6kZBe^_bou{n
zD5M+nf5p?;8BWYCuOgz--Wwlx2qyK<`hej>7xV#nP<7i=kdq)@2>S5&bPj0~MpeMQ
z;$KMQX0^J8#z)vJJ(a#g$k6`Z+IbtWyk$TvI9kKJ59J38>x7so1jUd4-pS2OpXvv%
zeM?IG75C~E1Wnvi%=2k9S*(=}R(a@wO-)TfvTJH}vlSv}#y2~3`^N9SU@Ivs^aKq|
zEt1pD-agOdDYI(7dPO>3m%?5mKt4cbL*+(o!GR|g+n-j%dVFnt7~xS8`aqDhNA1n#
zc`vjf3JVJbf@&<C;dGVWlKyZASulc)yzPx%%B1Ew)b#+0P}={euIJ|9xUkj+<v%C=
z|CIKtdsB`6ZREw7Ru4}DJ<Iw0xP2^vK*Pb`kxx|@Wnq!+TlgV57e$^d0?ZEhX_Bhv
zm&f2-^cpkeknkR55jDk((ke(W;6h(vO%D<+_+xZrao=MlBaND|@f1-`QIS_&Bp2o%
zOF;)KEuWO-IsO%3CK*)Y=64ZPvSN(t+xhhy-irxJ?b5wZ%;Nx{XB32K$Vj|z<qZM+
zg6#>j+5}!UL{ezqQ8m8}{{@deFf8BXq8WhXYR6v#O;n^tc~+aB5y#tRpnrC{WoBsT
zfy$7TnVB~T?NJP{Df5~AX-8AN>H0lzW3%kIqY~oc@}c(w*?Qo8zbKKAgSprD4H9Vi
z!tR5i)Mzb9dPs3!JV3`&x4VmxgjkRHc;EMDw1yB@sRFW!{C><wc8fbWrv17`S(QH1
zDL6bZQcpn;p$NwxpE5Wiu3+M9j^WcchkV{t+`Cqjc#UceGiSTY{pIOJKmel1gZlEf
zhf?31BA{?Nw=%!JFYqAlcNk|7duaLJLvVTunugwqy|k7f76Pq;6P$oR#)Ah|L+|6g
z7JXn#xlp2$(2#jbNVIH554JF{Td<$KuNpr2V|Q-*B_sq-HwT`(!L*XDuc~VR4IK$)
zz339$FDQ@_5IC0eS=2a>T8z<&DCw?x`Qdu2ls`sNz+^7y6qX%Lt4ZUP9zI6ezST2T
zPE_f5Ez6^*G_O0)e4kkBjkO+3o3}3?-dzXV2Fe=9?nVE9WhlV{?hxqHV!v?5Al6^D
zw_mE)2?%5i{IktQzjU1+evdRWF=1iMFFXRU$Z6LR0+(2Kw?A~Myg`7YgZFZ*qXJ)g
zX0#1+(u044pFF~WfrD>4KcO>9e{mPuD)1tt<%5X!>G;+_UZ8NKJYCzDCRv@kykdEg
z`Y))O<0G87ScF|Hm8W7oOn2=JHQLqiJVmPD;zDAFv&72Le(9|UIx+-p<;?P}hra?Y
z%vW8MPT=OV@+g>CoVM(xBM*|RIj&{gw7ZF<k5Z~!?U+F=hH)@T(9mIbP4caqa5;c^
zRN>Ge5p*h0+Bz^~pnC)P{ej#MQX=shpkcATSW4G>w1Qau15z}YG*H;F7?|VS!*rDn
zMo0pLRvYja>@k9abFryUnG$MzM99)9XjIL;xw$*|gTA{Q);<te7)VpSON~S?F{_(+
z`KRPYohTGArPfMp0r7oKvTtM-+oZtYxw`iw+&pXic6sgeCy0uAD>Sr?1u(dF(iq$n
zl}!riE6ZQvGRU+_Uxgi8*vN#7U8mpUryKp}bmqc7ryNSyd%r*!AJ)m)^31byJDG3t
zWT$HX$si}2r#~rhoBMWvKaF4S5N6(9!8Tx^y~ZR{3R%`tVox+LTHXklwD%wa-fKA%
zIptLfDy!n3cLESEJNM9ErQ4v@=2GlO!J6o8z*WUWW>s1ObP6Aw>HZ-OFtSJ?EyFi8
zFHsza7VvnD1P+HZBxaS70`$C6^z6QmXYpO4VS_NE>cB7)N!HoF#*_+3e?>*_#5_1~
z@fyJl1J9MAF5PcfTm5rnq4U}S&2|0@kE#>venc{FR$upLaXrnN>dTi-jwQ*_@0-_n
z3E+5F=@?=j?Us1Sq@$Z{|EPzSETt3n(r-7io`u?-M}2+4C~LuyZWyYnn;IC4R{X?!
z@BbF(GaJix&Z(|XctxVsy3LmY1D>FhVh&wJQ{%TpJYZpE?Q%EQyi5*lbsyg!#+U%1
zs_O0Sy_t*qcr5W9y2igvc0S+i7Gu3fZlfPFWes7`0aZO{T<{twF5V>;%5bCowQrgE
zN~9~ll-AALj<;gl0%}{OBI@cJC+lqCsmV+Gmlv-d>tJVR{G9hqg&bUxs{5k_4+9(9
zurLLb?iWv1?_jI6!E97Ua<>rak5RGO60tg$6ZE}MOwIOK1=%y#WrD1;{?%&=xEWeg
zzG81b|Ll>u>Vt=0!|&{|RDdT>0H{OU5JzIFwoGFczzE2gZKaaNax0XWWHvIrEurDI
z>9Mpd4ee_@9h$r(pMbs{%qN@WpFjQidwJMg<dSjqE8&hjhqt~LeXL^E5$c<xNxbXU
zGj9>hbP$?2W_<C`Wu7I4%!~T}P@5dze%zSv1GR+HiE-Hmffrnk9kC)Ozx2EwTpg_R
z4-#z1yD0DMr$ltjr}+M!cpvLmF<jUj)x1Tsc!Dh`Q%K<WeI&I8cC4(d$0lBDXNNy&
z`!T;fXe#}rcG)0c;N0*DDym2#YVm9)tm-kV8Jd2of7-_~e$)@jf`LX)>bWqvEp%|<
zM7h6;gHTeJ>(7{ZdJFE;WvnLFm+imZ{k>VoE<>cc#`kQM%CH45=$Y%U(Qa%xP&sUP
zYl<_a?8=67FZNWY{XaaN2RN61-~KNnAt59cQiPBlGD8_<q$CY8$|yo)6fRUGD<nIi
zjO+>-84*cHQb~zaGRn$c|MTm2KYzz_+{gXgM@R8>U7yc;oagHVR?+{Z)pHo8aWoPH
zbO5JLT?Ms+AmBk(B~5earS=F6`M0&Tsa^W<3MOQ7D(B40IFnPS))6;1Ah=i4$k^Jt
zAL!7wlI_9Ll9B}fF0B{mi|4(oc0I8@$<8SRG1;%y<aO*vfZDh#afWwZMVOTy0!d>L
zvvRjV<E@I)0M`CH&KVMO=MFj=4W-UTP4Go>bC9n>BB@EU!_C!o5*79ik8Pk1nitG0
zETQfZj8>;g!FVv<&5vuNVa=lY(p0kM)g3|3TTVSHsJvgzNYgIwm(}Y2Xq;Bnc3GhN
zCRz#l`q029edRTU!GG5k98qQ)IaF{n!ANcUn`EUK$TRAYaQPVFy)dRHX4wCBDO)Dz
zhbdlL-E!d_BEeUBAx>i<KH-&sDNprX637OSz(Z3bZ*Ri~fJ{NbsKY5K-}{Cj$2TzV
z^viE|QLtdY-eQB-3-^3yq55Z8EJEtGTvkmP1z|Z=Q_^+2R=elCwqW<G!#ln)N&Zxo
z-ni{uXXmk_M`>mzH2g{5%fvpgoxb#gDS?J9QWW)eRwoeVj&fVQ)ekQ&D!e_x{JhfR
zqVbeC{e+X<iPGHaD*{u@+owkoWQSLO>HF}u^g2JQF5fD#Ynu;3=z5%ah67@v&$q%3
zu!tHY&$K8Y!#+wNR^m(*&P131orbn8+bRTCeIX$U+GMN`WH1T0x22ICT|$N$l=yHC
zh;T7I>KsDCPbq`>#m~5mEqE_M)x^iD(ETfvy~vb_JrVj)E4^#!M#EW#aul3Mf03Zj
zS-aQ5A|e4>o>f<?$!?~)djo4Bqcdax-!uRNxT6is6B~3g;p4U3=8pNB*m<^MUR7cr
z<Q#vdKb;wee|2mUFf`uHE&lx$QFU1N@DaNE)r}Q$oz(E~6<}AgL$rtA%Ju2MRNfv@
zo}MwaZ3JIk-lzB8jwROr!X!V-?q_M3m!mulwA4~mW1dNga!(4Xc%xryB?N5ZbRG4%
zeXPum@eNG#Emfq?0{Y0#^IxmU6-YBOd(&=8vTYp0RshPC*rv0z>hKejMme2&v4bx|
zHi5tj`vNw>Om#gVffc6RkfD-GG&^#Hn!rC6omr4UJUqf+l~Ie8{O=9txF6jtdV#m&
zO+1ZeBl@m5o|J$8+_H@s2|EV|)TgPLWDT!_vkhj#0d&{r#=sc)=Q%tzUziH_hz^T4
z+W62#3_}<@7WH=0XZU#=e{yhlagieY_-<yUY>NB)cihwB#CqI2az6b+`<|(0RsVg$
zUM-j{oEUuUnyYV@cX7Q$xa*!ru5=rB9-w?6-*f#fb8p@FvUc0RmrF_)`p6^RCcJm=
zhG$EUryZGY6YpRCW!akNVevZjiQ`VIp*X)U?I15;F1ZYO6GdkaH6*(zvywFx0of)T
z!gHAU3i{>^>LB}Z^}-(-9c_F4dQ0@R1c4^h*4STRx)SwZA0HO3*qDedR|kz?H3C<~
zcZDh&_)-o_^}AP~R#{8pCAw^<K4Bp_i{@l4C23Ps<hw`&rp{<*)~EkY|Bx2cirF=H
z@#U_#zc&S%U_$^U6LWT$P=e+tZ%qWR3L^oL`n3%oISv5ixi0kZ4a|dqn5bDk>mYY(
zbqN%v#olQ)IytAPq(maG=;QDj+2rfiy9m2i_f20lK9CkGU}b+P>eR57>A~pG(8R(L
zoWPR<E4OV<qr680Vj+oMCN?q>%Qh&Jy8rwg9>&{C5P2qByCzh8d-IW}L3hm`@&qeS
zu0TvbR#9j!zu%^Q`EnZV-lekDU3R^sm>$^)AR|y)M>7q8`-QINQRjKk^rJw=z3#*U
z1>5=UOmM-0ZOj=`Yw^}jHalBwS6aqlVaVZh;YN7T)L->8KCtVu&DF10($n`|x_QV&
ze7LgEO~K?wh>l(2<p4d+|6YVG4szzk0h?ly9F<sYQ-XIbBfHn!lkB@exk}IYOdUf(
zjG7+|{PV?l3DNgF0}8~}X2~lM>x`v^8SDe!zklk`^;p3tkU1zFBeD0<*djp(ZIHRn
zQ9t(XT{()B6>|6WUwP=)TSHl$8dDyhI$IZUPI|MLk({LDZD?oqQU!W9=C9=);8EB9
zmr7~|L5&PO{QX0m8m#+4fg&KSX*|ka1fEZEaZG!>|GpY1KE>|-{reaFQQ>pC^!870
z-?AnQ_Gb8P^;6x_?0E6R^ko{RI@_N`zV4gv$y|7UC?XNvDCkhJ&46XxId~3|b7G%J
z3=&WZQ&Ve)r-tx~ja?4YqM;#G8rFAIE<rX&_@RmCay2qHU*$0Rb)LVmAjpJL2?ep`
zFe-fRMkv01&MZI`tr=8rH4ZIv|C|VYo=vIu+{D|X_!c!%{ko}Oh2_lG4t>3ANjrtR
zy3<gBKj?_h<0;4346nxJgIBc0UIT}>dTz%F2tFpSJTb|kOsRyNYbWXa_u&)zhBeit
zGFZv2K3)pKoel_ToyIp%Uia*Ysq6c++HYh1*VDsH_}z*P*5$>QZVd|yOVGg!8T!XZ
zJby{Tm5H|9n5yl4DFPz+H>O3bgx5sOS7v5r=H!S%njaba9~-Wav{}fag9AQGvh4Il
zOxXX?0YVQkr<d|Qbh|O4$E6c@X{8rcRq+yjR432L>hQRSW;j|k?YlyEqn~H*oOkt^
zy=U($c4r??u|8(}q@a?nQ0Sp0-4l{(qr<_2jG^~*^!3e398TEU3O&U!&B6H6Y8;Fb
zw(;kh9E=Byj7Z9?@Z0Esvx46GMQm2^FS@(A#pwwGd(28ZEA>_dhLa;TGz3<cKQDP<
z4mvhgh(AB>iG3F8Q~WYo!<0>1@+&H!H$$;UQFDhQDnq!}Xs+DzzTNl4j%p>(U!)!D
z!v@RL%OBd5OoSr8jEn6~`eMgHOHHD=W~JXRkhS!VBQh}&Y}Qo2wl{A=d2E>pJ0~X=
zx@%m4Eh%a#?L9z`oy?mC(7+&*J)a-VGqpd`Dl4EA+v|3LVDyCk;cIJZh$dCros1B6
z;a+40cmH92vW-zBOGuk-Z2a2>ZV?d?=`CT{eis$Fpv>k9JRwGM#?8HP(~qD!V1#A)
z+ebq~=WVLA(kWxU3m2fDRZM0sAGc2kjNZ$!H*@aUn`KQ0=Rl(Z-gha&@tb;`l^(lp
z%%Dojpdx;>D}I)(@A4Mvok%nH5%{iWit{$s%>Bd<7o>#1%V%sXJH&Y5L+)mK;a$7z
zaWqCYPEJgac((}x)X1lk%0YGDIt9A0s-{S=(j&n7+4zRbJO~mN<GmboAoXyln<8-^
z195clL#!QD%nY+eF7tHm++`&m&jc^gH-6d_yX>{BHpR6$lsjxK(+#!%p7MQqd)fJW
z&BTwRb}%$#(Mt}i1^Gm2`|%|AUIkSW&?h*RYfHvLIA4lz432w3ElWjU6>O<Kh}s%O
z!#8q8MZW``_2C~69zMjh9vt#00lhz%y<IHbO5E=}56DiE0oa(=FCGivvq&$HD*Hx7
zNXLVVG`F;j3n=R!msVKHwqYJB;*$gZ#wQ5$LK-8^aB#sL9UN#WINf@7<!{~H>#V2n
zeb&AAH8(fMod4S>C_#aT!1voG-E>uH^AxkL`zP0{N~b1nJ=gWVb|jB?HnpCnvzzD3
zGV?%#zJ=h!AOUpGJ9j3VZMBFA!$j_FIm1nzD?u0VIei!mIU;KQ>gegCWsWAFZ`4+P
z%)UzspBNvP^h&mYtOqMzK30AyD(Y460MK<-9`v?xMbx|m><54spxZ%y?!v-?_4{uq
zwF6FVFO>4+g2ox2*3%QQIDIOFOma4IE(K^pZIWkFT3ihDUy-SWk7vkBYS$NypUHJ~
zk}(H6J36c-xir>PcK-<^m8hj?7fna$qabBRe30T-oW&M*RGB0q8c#+y(GskTHzG@e
zElq)90a%T3p?3JOZdCg~>RGNSp1oe`a*YXM49WYF&kIC+C=#)?yWAkvKXp4Vb5LgY
z(@u@6U%1W?52Td~2nq3!YT5e;*XjO4lzp?-peZ+>vC<#tAUX8+hyQvYUQu!>NLZJF
zsIhI^n&Q{EXKx&nSMUEmlZ=G@YZgI0^lDR?(Yv`z#vhyH^mh7FuXk}UipyUa+U0f+
zn~6y3msqIQ_4z&COx%WsuV3MS5eD?B)aK%`kR{;S9rxs-V(A|^A@)Im(NCRf8|ML*
zesTaD(CF-YvpiCF?%ctcW>w(p%X((a`}am|>0J&5wf&c<2wHRb5CMLnevX}P6%K6i
z^1Bw_{kUhA>$OgB*XHHL+EF29Z_N)QG~3ruTWZK%p5TI7K1AHY%uPy=OR9kOSI^bw
zJy1lW-|Np>Ueux0`6zGu_kujhdzpN?W1yti)!eF2t<w60>!#ZaGjAJei_C9KJrUjo
zR@3S;akEL#h-#%8_ymIBZgj!`2dYLC3h(dkr8_NN`ryG2KpP>+gr!!%R$9ymZ6@Tg
zo2~ayxF!h6J{*SOM1~>{e=ik*Apj!99E1^S!cFcfBTw^gKESfZdt`pn`Z@nE`otF4
zHLp#zJv?HsTs!H^<M~n{dhx5f65%&G87}KTI@3jrP1LPXb}TirJvkmV;h8TjliHHe
z+g;F}k-EY5wt$zjTIZLVGJBc3o1e#E1f({iN`QUL>=<&>LQ*lIH09arp+7gdzZXRp
zoGvVuC9RhX!@nzB{ko@?tYk`aGz+6uEue0b(0gnA-_K58ub^XGQ}^x+z>aNT+{$m5
zPFlhKiR%EMG_f%;8#~wYp41SN(`)y~!35`3@k!mN2liXaA8)3@pRg_MW(cT}M2+uX
zBvhj*$%Krr!bl~KF`%UbWg9ukvHoZLbPc;l-4s=L4V8I4Z>3oz>`O}IkljFQnj6oh
zPeo+exn|w95$M$pkYibMW_jx$mk8<ggvArO<z?lWk26;1^+%d;IWSDWQ95Hk@@=|n
z{LGqH!UgUL&+M!_wQv0pzDzZK!fQlSjy1)b_$_>Hh`;vn<GsTLhu)j~-~Chn3_qjh
zc!_0QV6dc3#$mW$vCgeM?+!;(ND|J`|LdmhlP-vG+`4<0Gj2HC5&|o5Q%vRpp8vE<
zKnSNZeW(i6z3rIOmX?0JSgN_Z#q9?mqEf9}8hO<5Cr2cMEV*=}pnJlcEd3td7K@MM
z{M&nzMQ<j(e7sX*<*mHroY-(pj~%f0)@Rw=UskVIKd`Q9X+CR+hZxh`T(ggFwm@EC
zW?4>pr#U9lFRcnbibm=j{!^}_BQkcZV<l(!%C4c`Be}x`aWjgsd#nPfbFMJlt!MF_
z<)sfl63Hu(+|uRA-+Mo4!~dRa@`+|z-(PsffiXVEM!kbA`2NCYT;%TVTFHbjlTrlZ
z691bGXf~U#Q7W8<HX6cWi!y<dux8&CY#eO`y#yq0@h)=hWf^V*`1_M>SbunLEi|tQ
z8HjPrHE@!Sw~$`Us^M6A_27yQRuK5{9UXTs9_w|dB9y7g4Z%;+J_Yt)^rt8I{%Ns~
zF<(`lezWf;&A0knNfC68rO_+FQD^=8Jr5qdbuv$DNQEuvcy+POIpe8t>yeZgRoBZe
zuS&gNV38SkGtjfbYA$E{AfGW#ueEAgt2HdK;j`j~)dB;JsK0<nZ$bGG!oRSvRoKw*
zh7VWoee#8alT*;o4;uatDXkSplO-!{5#%=zc}!pko!V$=?ES!Rt(syVMEI$G+e#OB
z7P$_vPy*Z_bubroUtD}Pngzl7#Zqf7Ib5a~m3wMKVR%c34|WAAVuxtACW`d07j3CW
z5Nk~MSrpA5Gdh^QnPW2*^Cp6<3BF=BUH{KY`P-Q1tAwR~g$U)7+AN7epA#nUig%P2
zj3j10;l9Ov@K=t!L{8cL-c*Iy7e_WGZIyS~=AGozrC8}8cQ5bd9g)iWamAjSZ6}k_
zTH3Z0BpK<m|L^#2RYyZjp`Kyk3$WbyyF(J~;|~w-z}!AOmW++1A%t%QtZ&oMH8c#`
zFo;E|&7@&HlYR&OoBuvyk!!hzSW}6c<76IRf*(SKB&k$4P_g*l0nv;nkOdU{th?5d
z;YwBk1E^=9C88o=8R7~=OcI)^u=8ipM2}qk_WnHw=k|A9T?`ex(l-d{T^heH(-7rP
zNSm%GGHRX0rkx%?77ynaNF0^9Pyg*vO_X--bG7L0rxMg9CFd)3Kmb<V+x{yl=&hgX
zm$^l?>NEY*K_nJ1@89k8$jxetiYRN`emCm%N%!po4J}DZWZhB=f0DjOVZkTAflX8<
zcY0B4u09F%9MM^ml6&`h0nBpft{6!H@0DEqdZq8iCKzEGv$);i64~kbaxxHNi{g_i
zNA`K1IMD^12vVOgAJWna)DuJw7CgqFEe*O+6Y%QZDQ3dTLGop?fG&5Vrc#zV70WP$
z2&&<9mr_yLIG25a<r*C{0cgJ|w{Mx$*A4m|P(DAfKS>I6K%`iA&8kR2LZSfAk%i=0
z8=DvTp8TK~Gcqz#a!=h9S$LJROmgr(=|*X#Z_uKGQ7a=uCI7>Z(vl>x_^m&j!&~x=
z*AJaD*m~do>SN=Sd+JME#|^DbUDZ$AuIv1LHGZ9CLQ>IlZyA~)z2Vt5#|xIb&+PGU
zNFry+fRc!arN@`~4i+ddS!_HpV|IJynJ~kYJq+iR+fU(JoZs(n@;#n3f*SB?e6J5?
zFB0OX+IA2ii*`FEEm7{BF$Xt^pGHdvTrD(~<^?P~op!R3H=cn-X)JQ%uLo53<zX+}
zuC1k1d4+7Q8MVS!k`2~ci&JVsvXQ5_w6yNu6o5q6dvsu#Mnqo4R~!<!C`aP@hEL{k
zHo9<K+GcWLpBQHwI)N^annTSURcePnbzRXqwXg4)`NI!E_3xepTexV4oVcra`yQ2K
zuA@n6N?psSeNcMtyRPvs%b|6gJ6*RM*x`pyW?=ffwWKqz^cm@SD1<DEVw+B(A}*Ha
z@F-V3>Yb~+Q>-u@-8Kt9VV}9Bfq?<e#`ot~xD|-fcNBM4h2YOJ8g*7+e)bUev0_qE
zT5w2&rG9nTHHb3?Y{onaf$kpbQ5!x~2hW~8D=Ly2vO9l%2qRXEJP<;0f0`3ppCZo>
zx#S7UDvv<a^si>U=tt;S*uqnSvH+SgSnIUim-J2YsHbhaAe&J5(uF$(j6!&Ed{ocC
z+Nh^_Q{COY<h1zc$OtztFKTgRUPA#kEcvk-4ANTP`(&bsskq_(q2h$go416By=>H{
z=IWhWy3}1*QZ3Xpa4aE=dq-CD8<LLrch;Ukt|BY_#={n#;y+z)uDh2tkoi{^2ky-z
zy<M6NiT0pvT4`o<X}1+-t)V)8)847#J}-;boU*i+q9ubE6RmnpB^zhR#x0fq`xyKc
zGHclMP8VclDY%btfn^8&CGZrt<Ks71Ov>ec;Z(yUoB~EjV}4G~59Q$Uep7&65^4*3
z@cvtD4q^LYCrgP@6BlPzf@Fo_vBrX$If!3g9+g$uRyhZqxBOxBF9#;occQ7C^Iq<f
z6Ee>Vmq8Vyd|q_qo|xm+R+jBcB2QfY-pDacm~0sxaIb*qIKYduHa6v<+w=6c5IVsa
z7jm1IVq{KAEbAT8W73Jmql$H}F`7JSxwXV$Gqd#`8CasH*CEc*pJr`b(<U=FqaSO*
z!NUbkMZnAOeek_!xbw1CFGE0Kmjkkmu3eeZ*Bo(IV#<zg{#lfAgLOa<RQ~3l;ZwRB
zUBo5zx2bNpIUbp{7WBpS#MoFJ$+Gg-qdJlP>I?tcx%5jIZUI4eS3sBW(?~5`P11ul
z2%yS8AWOCB$t1R%Dsk|S?`8}Bf$=u3ne(7;pi?<~_0eYLd&g_cFEbR`9v`va>+@AC
z%At988Woq0zR=pRb<&-7G{TH*n~DBSW9%22z5H4_DwbY#ULh~|{9xTWa8B5^EEH7l
z;*<Oy<}WJ$3P$UuGq7y+k5uYyyt27tja(w;>*r_dx=Pmkcb{5`==@GavSQ7j{6pep
zX9GqU)-ulGTbIlIu1FB>{A-cEMPYP?3pA}0e`bLcuuEl00!<QN8ReDsd@i!q#;H2e
zrAD6JwE~Fe@ZZ1U(6o>%qbpz=Y+(wdfGsigL)7K3rI~MPx}|<`;)UYnanlR0MyDI@
z-97i#MJKK7ekh%Vo!xIBMxg=$fAM^1l6bGz8siRykS1%d8|)X}xMg-fb<$qaDkX(4
z>AYzBY`T={*Q$0_nU=usi=V-h%r=GrcFA{TPM}x)l^XT`%A)vv8r#kBU_0yo>T`PG
zPpyh6$S!KZAg{K2Dk$`yJ)8Ak%j?Hd0PQjiZ^uu&AddsGsOudFW5b*o$#b9s!9B@H
zfJp$&sF11H2W&NVTrJyf!_C!D<-l{bUq4RRM1iCej69oXe$?m$1J9^j;`(1d3}a$C
zXeu9gCxtaih8|P3O>2qIe=A#DFf~|Ob!OwS=L4AqWs@)1&z~s9={nl*s=z8C1GNbi
zfh(xcDXq&P{7q88dO>tqCw86DW~B{HNf5Ab3U4~K?SH?y*E+9gu+gt{?t4)eM6y(a
zf&-oF;Qp?9oM9Mq8jI+o*pkXiPp2UYtzO<L)cJ+UJB5+x%t1K+e!HF6hpC+1Q*VHU
zK?#KV&e8D?g@(gWy{7)@wLJv^$4(mX=5OC>_-?*xrTJE)q_XJN6qSf#XNF$q4k`qD
zL%t5f#-JO*kw|0;ZrRx?WzocD@^i=I>0VRA`sQC4e>X|mf9_jvs;WF6;@*EeQj@FZ
zF6W~A<)#!#rnN+`0+MnNhbqbS&r=!s7V-AQCm8|=0%1N_i!CoNTS!6^sd=LuEi;=(
zUqe=x17NXhh&PGy5>MaL*TGB=rEN-E^=~pQB<>mjI;(U`ujl3j@?VZX@P~u{Y`{Cv
z-1J%Z1Eg|T*kM(<Bdh<!R16%Ib~_6>CEX~*3&5!gd^t0S8HM*Y{WtR7|D#4`IP>8{
zthNE$XQ%b=e}?J%!H08mb4F^9pwG|AX|^f?>l^<rX)J;daM!K*Ux`ix0X+-IdhIL5
z|7$#t9!T@LMuNW_D|TW6!b>dhlp#J~Ph@hC-D00$A`(-S32OY+M?pM62mH7i*EtBt
zK?_`HKRM6_Z7%Z#nZUSqXRq9!Ii}6WsS1Khy6s}uu?I^9b1Xg;16<*C8hIcDc$)7s
z3o5)U@5lJ}(8u`~nw-%{KK~#&{FzM$VL=Ftcg3DoZF1LVVJfli9+LZ{tP-i|o$syr
zvSQ7kn(k>-0{*EJ;8UPWyaw+HQl5B3^#zY+`PnW`>&^OL3W?-BT55byk<>lIQ^><6
z_Wi!yZs%ZUM^8&qzuB<rpM*6OQ6`xB{0IX95Yo;UOKS)a(xGqzf%4}i%oUNnQ0_Z&
zZeP>`v`(gC{{db2D{cA?b6mz_S@E@Xap)3OII-ph1^M}bE~ZQ<*f2fTa|pnW06>$`
zQCsQ4)v{(EU+WrF#Xdfy4_0(zl&~%!kU_s%If-E@9_-LL9JIlt3zwQi%t1hS`!Rsu
z8rFcCFwdj{nsh9$q&l=7rK}ARf%fjjne}uu1m8$%*j}~|cezoQ5>-V1y28lDoor1I
z<zR4b;V$(iD7NYFdO7g&_FGNsCxQ}8f0MM}enM5_(mGXLf8>Q)s?6=^a?;0lV*0JQ
z3zNuT<?tibC<fabJhqoU!rx<;qa)mFsqcURLZr@!dqvaP*RZVp+AmmR&6PoMktR2t
zMN*Q3EvIj604o(JT|i6r1p|L07;VXAQ+yJ$t6jGsp!zvE7rB)RHc|LOA$UW@iO&rR
zqV{%mBV!Mb890ED?*&`an>WD5VE&ww6Ou6qs3AT?g~=X(#lY!4ER0PBY@4j~2wH;o
z>aDB~%0m$Aq`XZ3lkgA(f}%`vV3$UE4HuI^<|Gu|<EPjC?;tvlrkgA<tGGK?hTD+y
zVMMJ0tv9|Yw1<f;cma{bi!U{|-)~_JxHR{(Ncl3PBtWW25xvmZT-W|`QJ&zFj$_w4
z@S^003m#&2hiVVrI?uU)S_`{^smYC44H_AtXl39EI(_Pt5(RL7oI#8ufX^=BUR<v4
z4Yoyc;`pa&(%<M*h0yrm)W<cCZT;iT^!aqIthmlE2QHHbc#>l;54YV)Irk)L$BH5a
zBFq^gp!>TxL-_T;B;LK3WjKa2k#`1_NbyP7hmV%=ly{sEGdc2ZB@Z_Lw2ndGoVb`I
zz3!z}SEILQ3s&F`ZyNx5{Nckj=>K5hc@tmfU;%|D0^{-E!)o&3g9#imA58I~Qy#R6
zqQa*=m<tjFM#J**NC$hdJ<nv_4V42%g3-ZUK@Ww8f<Er%3iZ~<5_pwTDqh)vV*A~@
zBLvvre&0r6jCoT{V^_V8513umH=m-Kf5L)_4v#Fvu~{=ZW2l~Y)l(b&-p)UU+Hvv^
zzKy0ZtSq;x*(C_lQ*SJe50_*~Nnc)Lo%wSIYe&HTZQ6o+gjg9pBX!(R$ASES+qAU%
zIk?=w5IYQi$ioQMs7@O9DllcsaHt&RxD1l(o!HpA{;?-KUEFMeiStX#v+GH?By?&b
z4x)Yz%T%e}B_aYkCYE^ux^jS${`zHbVLc0BkrvB-;NGpTvpEkEo{w~dSzI?<Z*=&_
znW|e+uhp$KJ+y!Hv814*@<NT#03#FIouD?|m<iY7$DMQb$8UNWT`Ftal~lDX0g<kv
z;eO^K=bqPJY!Aj%k9jc=X=L@m9g(jtmO@)IYc_DH%;Mr+0?&S(n($(Iy8RUk3qr&(
zN}%sohK>N^*Tx^0$}kDyqrX<mI$u0EIS0uSWnEUcM_-VAaXaE|>-w%&xjJ)Wg8_HW
z^p)q(w&{6uKE2i0k6-VNR@);TmdoD1S)-Exh=^~L_;r>1G&&Px5ZEZi)L|!UX?O_&
zDg2w1LM*i&?(ebF#V2tC{Z>R+0Y)yJ*wx+f`BS%p`sU!eMZN7A{2T|Qj<8Dcj4v!n
z@eWEcX;SmL@ZQey3o^VET@T?ar<m#eb~`RI<ghSk#)#5dqKHnwA);h2jNBQ;)_obI
zW30mQWP_Gh4nki1_99b<CtW+Xh4#vb4$(yD(Bd^RHhzzQv?pD{w}yWI=F^pyjq(!I
zgwo+qLCq%dpGo$Xmpyvjw~l!dDTYFa0=iB<zJJ}g{&hugv}(7<9y7y)FD+g(ickMf
zOOU$OY?&&Sy#LLe96Q{$&DQqI$vm&yjjGxq9B!xFM=w@1Yk%4i6czpW<K4<+L9vkM
z8jxtbnE~5N@+HqQ5?<qTXMg-S%iXx^*5mF!hihcIBQ1Y!m1K%;I*S0Iztoa{_j5%@
zSaMAkGp(#EG|kPLJIZ*2+N2c^mL)qz6m^w%^c=u-Vc1GqDeh?~NIS5@$ZoG+;bkt1
zHNMP2?D31#(agQHtHogJ4%)$CuFqfPo^||ixV#W#svsX<&>D4oQ+EDg-Cy?{hgpxl
zez2BSFu)||oQAMw89u>kv@7}f`ERd*2nqSe5^Muh#GzPmN+v+vk7zW7fU)@=@I0&M
zMK;4D<khRagNy0Kb3?%8I?B`Q0&c?ND63yLV{2CwI2f=h1N|;K6taI`G|v@+#v-j}
zTTuD(<CQnjvif1N{tsw`{*j|kMyqXVwWZbn(wezj-KR3^p8Duxd++9Xvg&(O^MHfC
zRYcF%-6|^2VXq*NU3>J%5w3!(2Z30!%en*eU1AG}dc0>9?J_bmgAN#<v$09g6Aa*^
z$PJ9BttYTSK03~9lFCj`%YV!nne%dqySbR8hc#WqIT*EX^#L4Xsi~*4H=@xxFz)8h
zx7XgxV?lVy$Mg#aar@34TJ6+QeRzGpHk_la#32+Q$Ifv8F(;~IidxY|%MC}beH^~w
zWQ^o)jo+F4g#!nYj>)kOaJ;D#!e3Z5(6&`LLAj^mvQzU>gOu(6p0&5XzqoiF|6kdy
zW7v8{98?^2IB_%1s@0xKjqPn14gc1@l~=%H1o2*hHVz1#FUYk((Hd#19?S7&nrV${
zMEI`W%|2y~B%N>2C55ETT$#pbNZ!AH&@o65g(IAL70Ft+1$3i+$j*2Fxv;cURU*_M
zZ|`1Qb+mMTdZTvKT%W#ZE9b@P!MgO19?U-Md#_vPt_w*qH<=Zt`+kHPPg_PUd<t=n
z%xfrh^6xV;Dk@(c<PiFED-D(3zm$H3{KLX8E|Kp+Z$qvG`bSf-LfT~PPl{?Wa(I&<
znG4c)*uQ&lzse=H3{2k8$0mV4nQiV*E&c4<?OHY3OC^}_N%b_m+4a^Jqu9&R=|f<S
zad->U2Vg1(^T^^=BAH=!$#+%@UJ%O^!TP5)!&42j<{C-;H9>8YqY}S`NbYSP&U}-b
zJr}w+S@T<^O@5WbbV$tSGlnIUyiy7e6v_}=*)2e=V=uSu%I{PFMfSIP19)H#(lM4Q
zswrw{$pN472B|dKv4_d>#d7U!fjJyBJ>a>jEp)YAf~-JB=6J5DOSfaG<Ja<-K{i=I
z9A!YblH-shY0Z~gGG%I|`wxBXA6JHm8haxMZ0EtV<UIEL+BQ*vWMwbu5l{H!3BE@;
z+x)piu@%0jk~rvtu^N=ix*P>kUQe05s|tQ*&kiX0(2RaG({1mcJCfjxqLutc>6_L)
z%a#|PzvR_t?efTD%v#=a%-UFak-SH{Ax}Q;OdsuU^YC2BwVSS2JS^m9m@QgX4)3>I
z`eaR0ot0T7aw<vu?We%)Y9f>&Vk@WmyU7o=Ayzpa^tx-*)VvmIr>pfIsX)l4e&E1M
zwUs$Th%$^t5X(V<)Yh)qgrCW9QJ7Z!^ySWD@?YILFir#sqWR}Mav=(0#m(m7^e}wn
zB@-09VT>{b0Wv5|zGoJRGCpWHQN;pG3Q0M|l?2!i!Zj!ov7-E@fk{dTFJ(Lc0fnU|
zVnS#L@cCbyf&XOdbMrGOesDcEjGaXE3>6X1tFq<fmlu7`vU~Qx5(>>uL?@|1Yb5|-
zbbQzYv`tVwJ-N^}$Edr;ea})LoptxN_EQnO!(OgYdbd*e%?c`gEnMCkjGmad;bHEn
zEV1<cRqCgjLibx~sKnk6fLJHSa$1a2bM$)C*-I5)3*Y_F*{*XqYhlTkE1P!pJ+M=8
z=Q8C@k6g>05HLGAPPLxgntfjCkx&o@Xg8m{6l!$pJQdNPM$TjhwQ^M_2v{TS<|1YQ
z7Lv~Zy8y5sj6gR<3g(%qt^`sTk5O}*Rgig;P|8_bTf`(_(^m17!o=&4^}!^!_O#mm
z06jlH@^9HdKIF=u0<{I$S`y_)Q9xE=_C$igV{z~YiYsJK%z#H2ZjZlOykbQ9nW|e9
z4dJayaW<7+ow-<K>J3g|?KOyC>4@A@rQuts6kzzuEi8O{zPJil)we%Ufu?x<ipMvL
zZ_RbOk4;LyskvEHLL%sIMf}k)@|OMYs~?B{Zt*dvUFcqB<+t>GOm5ftCe$eJ)9$=U
z_kfqR@sBShg8?*bz8@4!Y}}sVG-&!`$<xa>cE8KP=EyBAwlSP6=cnM9No=8e=xM6$
z@l3rJ4E^%Q&?rPMMt^=Ru%S10Md$ee!CW4i)rq%ts;iyh(UowiWQ~sZt$`Kks%8SM
z@DhQlj;@lNXjK|T2S89Z^N1{QURj#ykX=F+#S3^P2JmwJa#jFj*YUuCHG!52;gnAU
z(>I(ig-!B1)Y)46og5u~AYG&)E|o17HS$vM)*31A;IfA%xC-#LzP>(u-p7v}tF87(
zIt|}G^rh&xyb#`q)MqlKO+H>vRv(dKYk`RV^)y<5A(4$T6ALN7dGO&CDBgP8z(=cb
zLeL8aDqwm7Duc@v9d%@+3u1Ia7dGzW)7qh=1Z~-gybno9NlzS`=bG&6svAAh>bk!<
z<uyl1B$!_PUOS@r<oU&G?d=?U>(4Prq|AIy%H2vfJb#7){=hG2*DSM)Oc#{}cJIzB
zD3IV{a`*7KcKg$Uj>W~wS_?o7{1Q{SnT2vyv-AWJ3P`K|Z+cqG#BpSJ_}bkXrx90H
zVCFs!`>3X@Mt7%bj`em)Gf_GnnRusD_qZBQn0R~~?<p1J-@J@2i133hc)Khk$wCxj
z>sY-!RiGWQ%(`m~$0pGA5IIWCJ-7`O>@tdnh@s>ALSwLmM`FAVD{XA^5}t4VbzBtK
z=HH5fM~}_}jEOUol1H;nqLOvxJK}rqp1Qb8R21<)wosMT^}T9qlMohuHvRd+PBw-l
zYw38dZ3|gh#^fW)zd;$X!JLXe2oEBdHBe&mpIX12m0bff0yigmvK4}X=BG|yGCks_
z4W9cs@hpDtPnVwyF!~(z@a9B8&a#3^nR4Wm3-_jwmsktkO{kL;8|s$Wvu9(SWI_hA
zb7zgFBc5hJM8(B~amig91Wj0?9KB51+GaNYJn!U$RCB${R|vlr8_*$W2>fV}gE<>S
z#l&2Od=q*ra6MuU#t&(uGuX2ym70?>B8gM}_aWWH(Bp?o#J4ZD*;~IQ6q`wmmrZ18
zety4Eqeo$<n<STUPUp*)FD)A_9i0RjS$^pk03hga5^&GB#b-CUNKEXp&p$=fU@W<%
zc#-ccycobq`yv7oN)?&#LliKTiAlID&^l^BcFrEhU>9WQ8HG3a-41Z{`0dmQkMmFw
zaUZ>7V`(H?H1IzCTj|Qq^9{bb4Kc@edtTMJ@?Pd^Z(p0yDOh2@ytOxa{*Ot|<?3wp
zi9BWrvNyugbT#nAN0KWyFYj1R=i$0;2RSSN<V{0v-D>Sh`{E#1QCYc&6CSxAy|vd+
z9RLyZ#~rFMWI@Lqv`nsc{oqTfRrSt<yH#JQ^cb$s-}}R^$K^*tmS(AWp<(kWsj73@
zkL>6o+7f3p*(=aq5`GXlIXHYF0ql?D*T&0_C@BO}Df1+XrX=D+v7DVxJCLsMm{Y=h
z5ocvh6F%=q1*4@;<NfPRg<eM)xy*%a<L>(8*s~3_&q(J(@?_M?nepR^5&CvA(_F8W
zckd3{FqW!1sk17~!}~)}eyLf}9UUF0Bn(qj^^NWrWRlmau0ZqvkoNT6s=<4ak?I|8
z*4C8h_uYH5(Ya7XY<U8V6D{?I1`05CKy76+F?MEU>dY8@ZtAJy$4k#Ob7s{1p~Nn{
zz@v{79lNUjF6k-n9Y(PlNqsk*x7<3gKj|`kz$Xx_TM^)a^#Ph5mjW<DXg39uM&4Vc
zvg_Gt?)n!<oLe!JmJ|9sc<a@*X#b0|3&*EhBOdkKsC=5pNfyhl)CM0JC5Y6+uV!yq
zuk)PM45js?6O>WT^)?eendgpI5IUe8kyy%OIF))hE;;!*go&=FpKBrl2}f<3u?U7!
zAuo}91*l29%2m0JVp?*K&FZEtpca)6yfP4op_i)$z5RIp=_c;A1m%B(7Bnk68_Xb#
zMm&1FGQ0o4J|pw5jh{`xh{YSiU!<2|Gvri0n8DQnI13Z;>h%^wvvago)RyO6s0fV2
z>iX!XOm}2#P<EG2;h&__c6-q8mt(WUM$(vWUkH}_zH9ALbh(Rks{4c^ZIDvRis9rn
zfsLWuuMJLIaqMJvc}E*^in|e*S*$~j3JZu}IR}4Nm}`s0*xChadQZadu&|h@Xozhs
z;?qNAWuuOXaf(8dHl8(Uotpv#!MmMkyKvXOVAh~Qa-<Qx%If8Go5W)JH)iR-hlkmN
z_fzUkiv2-KbL<;H_FcfcubZiYWD%|@w4^=c%92ou_oc@0qL>fb`=frdHw|^>P!T|#
zbKd!Fl%11vb+SdCia_mnaO@KO{(u^-5By8F;>>?!Mr3sm1PA@6Y->n(oY#Gd2q}Cq
z+oqpH5`FXcef#qp7SbQhy!4v+_SkLodd`e4wY#IffdLp1w%vDwZJnLRP<|nl@aE?U
zAB-&t)~oWgl(YTj=727TDj4DOtLBH`!@Q7RmQhx=*BC`>;*%aEh*?Hnpq%Zs5f4W;
zFhpgoT?UiAr^GQ425zW(Q@qKECftrPJH4qe+ZjbG?u|M7lCN4&9c3}iR>@5rNe`ic
z%`ZsrkbI80wj>uq&RR>%Zn_-Fs46KUPlYo4`5|+oH~Va5=$hFZxtXV$8o#&+5q#h@
zY!{DLu6vVT^5^o_(C}Zcwu_&7da+0Cb>~Neo=*1KKk~)Qq;JpOr;UFHUDxVqTuiv5
z66;h=8~V#Pr3x{h=wApw^MbbK=5;&)KWC^2k(eRu^C&;|WP(>h;^QBkk&P<8llM9z
z!kk)+jh3SpSGHJYS$JR2fQXRt#D5&dt4>Ko!2&_M6*45KWFn`AEDX;j8)$dP2I`xV
zzk1&Y7*VCf#1!MsLrwP1LGBx5M#w|+1i@x;9daBHpI9@YM)<8PeOuWrtn*v{+CG6-
z;QETy-;RlS**^~E4#tr!HcGZi#{5=CAC#ZooikL*yV;(4&7$qA-3NMEb||NWF|0db
zSPE*f`nN|>YI^sM=`Sg%suYoX_dVMY^sfHp$75uYAhT2LCGp<%Ct02@73B|~3-|}Y
zV%+VFT@svk3iH=+-QU+6IvEyP<D@;LC*y<W3&a}?NAvQ+(y~HBnQ%Ohd!k^qP*(y9
ze)qdsiU33E&nsQCd(q{}Lx-RArBihWyX2lw8NPQbshjPIC?hi&)Q?&VQ8knx2K@Nc
z6^#duplpX56Gcv8e_@Ryg1nB52Fp<C`e3fM@UO!Vc|KGG)(9mf?x^?yG*k1Ql{6lO
zXdKa!P3UxEXNy^H?ni7qfj^cEnD(>x4IBpyJ7kZ(KQEGMpI~1YY*9m$xVJVtkS0e$
z_X><XjWqQ7ZB7ttqdAI7?>t<2taEwp>?h6k`oBk6$ivlAkFH3V<W)G%^|>B>zLiJ6
zU?h3PXW_}==QPBT!i^h~SYCHYFczL52>!Xw)W2%gl}<hPpZDdSRNRG)3ryk|4dcYa
z2pyt}1zcg2L@WD~e5}UCr1AYsEC+CRNp0d(*%SUGYHn!FUdUfKQIR6(16&vc5lv$|
z68@lvitza2g6$7>p)6OOv3L_D__QuK?_Y#}D(JM!%Z)gbe9)O91O=mPd{Geqg5&Gd
zQ@eIFRIh9d-`rrm>C!)SCC@~Z*q3emm^96WY)vjvpY?jJh48LmUi~H1ykO8DI|gqp
z(j2<GyTPl0+QCH0Y6%%Dkgj5N6le($#rhCMNWU@IMuPIsOOXQcckTdvf)E`j-s3c8
zL405c0A@(%&><zC`KRXCc~LjI;M9mY2!^QTEh999Zls3?Iu=m6ndsn#UmdCAY&Fv7
zX^PA8YPB}=e*NmsvD16WVQp<vz2m)(D}Arsc3x3SwYtXlr^R-5Ev#E*ItYDW%WNxL
zdbOv#Z<HS+ssY=FD^C<gmxG^w-*4$A``pDMJH<Iv)-q1<tzq?gqVTGllXt1x<eG0}
z_Jhfc)~cqU78?n(a^r%J04`vX9XGiXcCmvGwbIppu9}>rrZ}?2C5-bOw|?2YvWu%m
zJb8&bXiu%7!fi^#58;R6-9jLiZKPBY`zMSDywOKcaI1KvqY@H^{%eeIoS))jr8M}_
zmfd<Znt5MYN%oMo0nC6Y_o_nQmR%blkT`YnBtc-qUKsQtm|f_;p)UptMir-IsCxz`
zLFapM4YO2SxLEu`4G#CFMCD;V)*AS?vB{$D7{NH5{j<A&#vo;h)O-MWZI7|)JiMM|
z%_1xV5Uj`her4hcL~MDG90Lhi6RfwEk8T8En`dTj&U;0mMCwTPQ-J@B<0I_HpJ;UV
zwz3R6-Q2E@i&X3~^XQr+4jSURMjsq`)+5^@VlGZDvxGai&4op3et!^pA;<k}rdV>;
zg_nLAgD_MRHCe}gZ?2?j7G2=P>ddmyB~+Aljj~=}hEnBIaW_(D$J;Mu6LEe@RuPKz
z&$d%Ic9{FTFTSXV5pmzlgVW;L3#uI(aynoCyu?4#xM4tbV5H7AU^ZY{o>`3d(!Frk
z<%t)Kg1dB~NSl~j<Z8bUYpLWY6Y=)*XSn~~;agy;!!rql)&0={yadmXYbsP%8Ly=g
zKAxINregI)OE}N{j7{I7cGHzR>o|&(^sC22HY($)Nn`Z9iHh9G34(u^hxk}SWusTg
zY8q2m)$x9i#b#)oIXr3LlgxUf6UBzfOdPR4&TzAwjl79UGoq2k=j;EiiE8zh*@X;p
zPZl$t_pSa${`*a~TS&gLD#9k|w76}NDeSJ9lqhDL`WX)%To)W~kZZ7cGBq<JL+zUz
zlCqI8=E7)I);K%`7e+|6N22C0+RC%e9;%|Ojgs*oReXg^1Y{Y5(}L}O%T-GfR}EAS
zw(tLuC3pGdo@3nyD*w?INPCSqb+cvTG~wWJCIf72pRPBr&@L_gnGGFqf<yP}f-tU0
zz(oMHGr17CUgg)QRa?EJ_pzpJu^W|@b?^RmE1z}Ui|QP~%82oJjAIAGM!1M@Gh+~@
z43SFIn?!p7fnH2m|5hzr4B$gR1H6rDMsjGXm>uSrDY^g)?yrpGjey`ZiQT(-cC{2g
ze{Loi4Kzjr>s49(ZEC*U?nxrMceg;ioA?&Wc#0GpL*(^lNHWjN$Y}WEp4jqBhiY`(
z^T#DGz$&A!ryUjszrO1m+|Ua^0ytQyp~jV&1l;TXVd1zA%;1pmW*GIOx7Q-8fwwd6
zaLNj4x=HPn&YG^RB0VR}<}*L|w_Odoz-{j1=N1xee97`+<ix<<?B){+n~1_yy#Pzy
zaqLzv0yP#aFgwf8SRL`z`}3IY4_3OuhjT3XB1`<?$rTLy4ZhVx<{xi#`QCo-qr|I~
zy#4p%oumKTyh_r0*~?3Gxjvhm8~waYc5i*El^ne*$Mop5gTc__CK-yvN6)Am>#2e&
z#d~X+_A3kgI}1`7*46m9E#E0oyR^Qz9ug$@aCL;rL<fJBWMYeCzA9TnM)T{}Ryi|x
z#F&_wsR_)r_+@_Vx|L>6@m}N2ci_Q&h~vn;!m{*{q$HE5C@H$`XxyrWd>a9Q`hCBk
zD_iRd3s3b<&LNW&z6{wUF<xtUid>!bLQ*aZEancK@Hp01e&GSbj!0-1=Ak`-FU;nN
z`g)cpB@MUQ+n?rLTq?{_xg;FSUv|E>`{kp{{ig0B_M2z#+;2E~C-Ol+=33PgcR~(E
z@1T4hq*Pg<^WS%iYeg^lc1n+(=&so$cs!@s_H5s2hBL`K&zL?a*GMd3qsjU-#I(cc
z<68g+iZCz0TyZ>H>95~ODD>9)Km3Ke0|A&S=%1*9@TOx<1Y1JG)x?a!h6b%?7^K|V
z!gIZU_P{1H*>{Zo`z#)8hbqs@UQ3dcIhBgl&)D%~gH68{BmNCVbjC^TR^6>*V2}@?
zMotbtJ-o)eDp#H&w<wn(XT8s;kMH-7RPT89-Vqcp&!7J`U{gFNsTf+(NdA3$QQ$w%
zK-`ung?C;FXIeWJCmgm0C@ct^JbMlxH;VkHx3`xP<}>eBxs#LiCYfCiDdpFTV@GF{
zFnL*j<qSNn!Xj5TyJl#a8AJWp`SIgg68Oo~s$s5KZ`F{OSg0c6R~0y{cTck(y6w<h
z!dX=r0^&KwL*e1fIh~5{{)BeM(QZ<ck8C;%cE;hSiFz3bP;1=AMR)b3kK4aZMx6_o
z{aifb<wl;&^1#rcTWQ(#s|vd<{c2Z&pdE3xshSmZn@w$P<VaB?CB-BrHdf-q<Fg-f
z(SV<Ldtp<Pb%e^@-i;ezF|BU{qD5w-bR9k2HNj{^79A`q`n3H!wm4jYK+OS&&$|E6
zq2@}jb|r_}7fxs^{<t@upciPBCI_MO!tq>G)ZRZ1q$JnB@cw7Wn#wkYDlT$-Wqhkl
z_{+>Spl*Zy4=EVO0ff!+s`*k3>**y6C07?pKAL@X$_wGP>icr2dBtSUaBEYM$C0tw
zmVVWbJX|Yfc*s^e6{ybqD7mn|3jn3Um((wQs)#u9iiP)8<&HGQS)*$?sFaWp{LdPb
z1z0q@c)%8d2XJG#mlPB7Ff~$<%J}-{1cr~M$Hb$IrUW#hX}q;j6eVw{7C#g1ec|&U
zntdouk+C(fTPiEsB9rkla|2{aP@`bH1i3skAlNFP`wf<=cKw~m5@Ak`PABdN!fO@Q
zx$V)p&S;yF2Q>q)|MXOiCyVxme|knm`Q9mIv@T{6?7MnH_vujqXdGB98gnpA@L;=x
z>Argu1e#9N57XiJn7emhoWmIa0gqh**mUL)b7W1UNDJKk8?-5qu|PKbok8T9r6?$b
z<F9x~9$_Fb%*5~v@g!R5?V~d|c)iifON!fgH`JHSKGn=NZjG18Oie{a0na6Np(jsn
z*&9ap&%=@Y+&v+1Ks(sv)v?^FjESYePn*nM9B28wwZY?}wAzJ#y8rjtGwd{)Lxm?)
z&rE9-W$}Fg7?L>VF-tHLN!X(T`0xoDi*&|)uuYl<bQNO+?2QdFpTH~lG=WEKOGN)H
zkVgFJ-y7fG<rjf-f`Dou-p+mXv%$mTrf3F~h&t;k6+5Di@&fa~GEJ*4D=!a`bwvfp
zm!>YNPH*SO+Y>+2d1&1Wp>n_?L^LFYd{H$1zDRBO&B0Tha<e~YTU(ChCYW0PcU1GS
zrbpRrP@?^Qnl4a_hVWE2KWnN))6t;ZYOxS0EGBk6>e=(>D)<%`Pw=tsZ?VT(Gx+qX
zMl6+LlEK6(QuEEdHR`PWGB$c4hDSlt<Z4{NtX*Tliusd4WqH94^3jz6*R%3s^NVRw
zyE}~0P#l#05Z30&RCh=wKf(Fx|9vAx@fK7pQ}3`i@fhs^i(lsJp=10RO4#;JAAqGq
zYNBZ)<C@SU`7L3=c1HLeoj%P(@X_yhx{F3k5I5xFa>~Ci8N$Gv9z850^gRe%bixm%
zktt~@&XJ*^+M(u@+Yo9$J{V$>)6&)!q=p;=@F1xOgP|U;{)&et9>;%p$DIB9M|5~y
z%!SPNuY<KjUd1cr2g<rW_T9eem;N)A|NC8k(le!7z)0=hm7#2OJavAj=wq9KWbJ|a
zcj9NW(2xfG{zXaN%MFdZ8#zPBHR-<*GO#5q^6Rhf+7u$IBCsx!T(Wo%jYK&ZpQT)$
z9qU6}FI?A5bkG2S2G#usoE=(qEcQNy_;O#~iTo~bjSm=sx%A#n7R)pF?kM~w^|%gn
zCh@)#<6P`abepXkp3XkaC@j34mety7K-3I&WF|6^quchV@LvA4ICAPkt5<y8t<HNE
z<%N&V9_sEjC?%P0aM6Bw-?@`aQvToV{hvXur_cEdg3&2^m>;OUF`PP4Wc3npz4I7v
zH2lpB+ova1sKQ0~HTF0COgDaGl5Gsbtc9e&j^0|soUYwxs<`VU5i}PaO=29_f2qtM
zNnwNHiJVTH;s~juA)p$$n~+eBlB3AnDE07<pFc0VxxJPB+TFd?M9X|f<i?M2phEly
zCV;b2l8WAX59JZ_z?FZk_P#+T4bfY2N&GJ7ee#Iwy|77Nlj(s2Wi!b;X@q!Pjx=2E
z+*a-5yit8ln@m|xtQ&C1U6zbjGdsllpXhy`-UW!mI1c2QRdAMsTNjxghd&!KE}NDF
zq_e>pj7DPub9r{!B9wmJc-GGVh_(My=)U?j>?{RGvbB!IVKhKd3N@iHw2+L`adOQ_
zNQN5mQ5Vi}_~ty{j?Ta|_nv(J9*z;N`}v*cQ9A2JanA@I$1w#^+3H{%gmjJrKpdwY
zhW_7jaQ3j()^wAc)aGQV)t}ar@%b?i9XieqKk8l~-F?#j?$%S<o8QgZHkarkCuFeY
z(2PUQscprY=EBq$UpG8rDb5zns2Bh5MxfuXU>ZPfETn+>>^lT0s?L`sj@>}#KG}g-
z?fmNH`SNl_q6Q&6lStm&5!t-G)yG>krTHgd3pX@W$pEWR`-JmO=?Po2{xW~QNaIfk
za#&7qRA$kpsBAlwcvt^b+len1OX1<%W7D^IEhF0}uln1!5>UAImb_X^r|c##bCSGY
z=~fruF+1Qs>H0q%E1Xgn8|x$Lx~mu7^M@^scDUX7ypyZu^^^RC6HVkHTB9pcI|La6
z_@KnuzeD#d%LbQ{!tOtXdm>JYvpBW|$BCQfnZOJz`jRhWQ}o1G@V=xR+#MLFV(#mU
zWG`SvBj=rPjf$HV){qvGxkw<!Tq5{1v&&zZzHh0<hMUG;RuB4`{j=n@-JNBXH@8=i
zO~`jQZCo~W4?EZL@81%i!_H_kTN0;Ej&<YZcv0xO6s!zSj8w&EgF&^wT=Nkv>63I~
z-uZeWuUe1gino}i2#VjcySkx|k4i01o!Ynj8tvl0H4L8hq2!Ks)_ICGu?9OAYqvH?
zIw`a4Gs_d36gK<c+&1C7etfQxPfLmB){fx|+0$~Z5>9o}Y_H9<-<UF$^mh--nW}t^
z)WmN$@RrA9$49;A$82vH8}&`R(2m`tdZlE&`u`s3{o7M{tfB-`OKkyLF0y^W;OxZY
zf0Lb@SfHuT+}3bpR86Bh*Z-sW{h-jJ78}1Sr&lc$qr(h;)FX2)#v@l*&h@?0|B9pU
z+(O$~DjWO#)qohSzsDey)tTiiwZl9spV?|a$XcOxVEyvs<-a~GRN@}*LN)X@AZ?u~
z5I6wzD>DoB3GhDPZxW0?XKNclrgxYNU?@*Z9FFRk`W2LB5SLRnK2cVXtn~c*^xv<T
zs69RDh!8j^&*NHt;ZVb-aAv`>y%G90I7aau;6Ec<=U1mK$RWlT*AyGAx;L_z2|rl8
zAXyB4%KKj#Vf8uctt*5UflCi5ai@@nkaiC;&qoPZMwfx_DT_z^g$1uO&Lrr)`;|pq
zHcMt?j=5-b)NhjFA4tD<RxWbS)f1bxlY{>n&VmQ9M{403E0$}{3YZk(oq=HoHOXM!
zROH&dV#{p<Rf>j_%?xGrUV$&I_F@LqVzW-=H|RRP)n6}yYylM>38CDP?0-<N{8#6=
z@J8E3#vsR(i8y@b4BRmuHJ^9Gljef47WA(NAHJiw*J_)$+t3>yl5u<+(ZBzD#dE&T
z3#?LR!2DyB;Xh=?#|q;!$EHmK0|QWZ)|9yWR##_dGnAa)ChvGnNqb8a<70U<XCiwh
z>o@6E4!(irRI61HrdGh2qZu$azXvb#u?oaAKq&ohe%#xaGRdTTghjftY*QG<{T&>n
zaHm%y$VF?A4QIZQBv^e02F!t$C|$t+0|aaUx|dn+WgICO{R5u)8wa8&`Y`patOMJO
zkJc%){hP}or%j-B*Jx4K+W@{#c6M-v{c44k0B5P<!45a1BVdFq$>nhIA{3##L5Z<x
zO-_bt=B~fuMOp_U(+@mq{wOhP%eK}lBK_^RZ{I$Bq9<x3#gOSEFE1}P1cput;ENs}
z&>ununl;wwv-f{q2F-xGSS*p0pDRgTVBP|er_6-T!Gkz*6_fvBC2>=+Wq8UQnQ@Vg
z$Ma4W!xTvCxG#x3oNUXcP|CwFJ4kQrgA^$RbBK(in_dW7aKXTh8p+28x=cjyqVupW
zzb<64T(}$<o{E=Lq2Z+YY(>lg!=#1h-&+}Bo-f%z*1A-JQU#UrC1gwB(?QDu_;IFv
zd1|V<NQ~g)1I_*84Ng-3y?XU?I%(8ZWo2F(Bm4N4S#G5zhr6WccRQvVY9w8S3U^h&
z4tn2%E4F};Vk*fQa?Z}~Cg083uTOmL)^$gm{xhb<@P~7P<D>D##2EL!SyGJUM#kVi
zKIi3KZ8Em4w*?po;QPWJ^=t+S3P|@qe-O=y9|phOqv!bl@4GNeTlL5v{NZLC%e&!*
zADR1e*f-##%sh8}+1im~uxq0D9zh2aFR4oj@qoF<?bW~}T%{0|J=+997!tx0qN6+H
z$Vx}sIu(J9Hy<bDTEVV+M149d1X1LLt}tL8O31*<3Cs!%EP!FF>jOEV{e1D8BvfwK
zYI`YK?@hW<cw(VE5fwd@ZEVHGY2Sv_Esz7gUIaTrnHQrK{hUrM^D6=u;70?^IH!{>
zSYW!DOU=CCqlF~+*PhBfStJj!PhH)gGNMfbSRXY~UElszsl*mM%&S5-q{;>FKXUjm
zyhia1{~v6lx?H{C|2~ra_(<|A{wA<0v9h<Co~toEM<+0`%lpAZueE3IubotKTUqNp
zFK#9HGz`Z&SW?<%R=Tc^)|8NsDppopi*qg4AG)jjb3=sZPJ^2=IvVr>tXh7Bj88d^
z9Ncsx=x)$mBgWl#2AYNMto3z!ODDYR;knWqL3e0%UZGIx7VRk!YkFK<>NYwWp(&ve
zb#C`)+pxr3k89hKj>l|e+Z<EHS7rk419n9-z|gdxw=mJwf!FL)C2@=_hn7Jx*~jNO
zO|6+xXN5`G{JYKq*S!G@N`s@L2%1E$fR}MJRxG;Sp2AFZ+)D0ZV#!;*>O-~C4?a9u
zeE!()x|foAP=q7$yyC`nLCU>);<3Px7k?g}95AWuWE%?QC<P5yAeA2mWS*yarKP1I
zOx?%%2SaUbe*Qk7JrxzJn4-5=2ODIbD!X*f-F^D{l8{FjE3G>I?-vorZO0sy#iV0w
zQy41G2*Cs6sii{>HP|3)G{uDm`#4Vj7CU!o?T!wF84Azc{>S(T0&ogf;#L~tx^+M|
z2;bTwH@ziLdsnZ0S?O%e*)+9vN{<vrUrUlxyj{Uv4oKTWJ_1591lTrma9qEP6)pzR
zkmf;ag?;XkxtBbjzJAr`6)iG7ZEU>%ZY&&Oy}kF5&B(*^XJmva@8v6-k4Rj?7ls3?
zk5*_{7!sPrCN_sY<s|-HQEBh)Kaof~(g|mq3vZf9PE+&tBQ-4K`t{%rL1z!GcG>B_
zGbjQV;O1!xp~N0Z*)VPIx^SJ3K7fy2T7Dmu+FD&>Isp+GdMT|wh{7-oJa<Y^s`r;z
zF0-I$@7{G6qLf(UwGjF<2bFo0LelSJ1m;-K%!|Y@5(dLPbEY0!)#6HT&$#lwT37as
z>*0(v$x0B#PZ^b6E>jaH^GdL;z$9XL3Ww&v&!3T3?8RVl!B5m2E5S0kaddc?hCoC)
zK%H;RaB(5i7oG!1(0v|{xG!KMXJ2jglJP@)C1g>>#e3XT#RH3(>aY+G)9Pq%@0*-U
zJDEr;VBz2VGwPu4c#T0{q&2CDfUbiNFQMZev8p<mchTN{^Oh|Wp2|P+q|Hlp&+xB4
z0k_Mjm|cEtNtLz$HzkB~w}_Y+w#}!2t3rsLeun`2G2z@d3aC?GH#ftCj4%9`kNIYY
zjG;ukrMzMD4cH7yo;^cwWme}9L`fTX{!Y%Bs&Fcjz2<m^eSCa?;Wx=KgZ2d*W!?OO
z>1p@*9QDzpdDp0g`8zv8j-FO@Q92t4#Bfv4>`p6TjCSg2s?*Xm9v*4e>+SEDSy>vJ
z>hWu-eab?6X00F7D<fBNYQLe%oT{tKsSHXCU$ITP`OdxU^p82l8<ue(*Go=;&tFGM
zV*Z0&gTtnXM*i^csozV|)2}_t#-K?+mV=w14B%)!Tbm}v$DaT|=3cSxK5tuE>e}nO
zPhb-dyi&yabfkT}*_g6>GuN(JP2Kq|+wru#$_=#X_3K=Ia<A%wL2;h_UYEY2qu(6<
zc)Hkm_Nws<pI!E&_RWvP8$}x1TxYZ9XdNp0LH8x;`w@d<((4Algox9e_DZ~QJCT#H
zsxXY8B&K3YQVH%yh3#yN&(>*o<))fV|BtFS0jF|(!@ifPkRnN@By$o%k};B@%#@Uf
zj3p#x%u@<c%8;ZC$y{VAGAo2k6*7i0C1lE!={;Bb|9$WAuA_bI_CEI3TF-i(`@Zh$
zI?vyE{mIX>6xmca9=<5F{w#Gc(@U@RL3sbZ*Aca)h=CVA^+og{lGR{j$PcUh?%0uS
z38FTn40p!&8Ga3SLT@S2wx4(~&Y%AV9ghK1d7IGVb>^mz(v(eN;&%M+tB)=%vE#_R
z<Y?aF`h)npp-(D=JG)Cc@6@^#TVY;a@#47_9lNgLZ(HyC*Y?mPH<|Gnu!p8KoC~_b
z68GoEQ}^mOq0bH;NG%C3c08&Y$RkC4?P~j|o8V>H=<|?a+<HVslrKA)W{u8%DwSUR
zrmak}x7qP(PkEr<&nwi9Le`_Q3|j(^m=*Jl<o)==fj$wEH-Tn{F_BC*DTX*YdX^*0
z*SurUMC7X=MWNjO-VI?A3=m1fmm*+k%(x^g<diaVhL8N6ydFpxJ#$(<>rfl`QW}>k
zNGpx@r->f7@rkg{buz=y&Lw_GAoo*-F!h2Z{j`~xC<{HhS<o$hL?=Nn-*a~3y_@ky
z{JW`m5jT#r{dbqg+9?vzTaU#gPI+wT!Mh8iauPYyxc;|bC2Ng>*JmwJHT}SfbI%5?
z%s$?hEqiBwJT&(zjk9DEi#q=0#n#SAB4wjRulJ@#wRDS$9i+vZ^BU=PA5`Un5<Z|`
z*6MxW!!ICRe*TwdAr~9=d9P*a{mb(hb_YLKHOY&}m!U&+7IK(3+=2tFsgoXTASE^W
zMY@*^Rp&&?4`AW)(-!v)C|ntg)@%GqcSertT~@5q^1%BU=aC`p4q3th?Wq9=@_@J3
zXSNYi2XfaQW-z#rHbPK48^691RlYJb7=4hy0+_0)x`FA(L-vHKL-cw!6^-c`KZu1_
z6&DGsnEM@e-yfcsaw^*#P9LvsOX)a#joD=@XI6W6YpdFQUdlL+`ms!*=Pwol!Y5gc
zb^x<g?dkl0*ccrjJZ@d$<A4PfL!`mWiNeQlagbf22Eoppmtvjb*5bfVMxZD<^Imo$
zui+w*!L2vZX~GIP&4s;8RY~oU|8cQnyt-c&Gq+;Fvi2%`nJ3RZw}o_=g33k1u3a_9
zK2xFRh|Jm0heasSQHwLW*#!PYh9H5NzHj;Z2To_7;b=Rsp{Z_GfPA<f5x+8sJ3=0c
zGx3tC_ga$9<VRLFEQaBZh3)ODk*rV8Grm0}uxW8YfvM*WpU2?g)?|-kA_L}bd)$`_
z+5RfM4AbO?vt{BMupJOzR^Pv4`|?Q`&z8iA*VW#~Xoo0i#a$o0iy+RHjt_OMTm)f2
zvx&KT118burypGv3L4=Ixqm-DKcDgj@`7HxK>kd^gVfK9k-VE6;|F3D*%$&8^GDyv
z$Z&)NXs@8KyYQwftyek$O{p2>xcVWGD_b|8i~i^wLf*YvYAIf>GRj?74P3MzUfy}O
zhVzk&klm}8j0&Q?wY6$dx%g7I4E#fAut3icVQi|vJsIcy8#{s5e5aj-$9Ec*sq|aq
z&wraiQ`45;&M+AFxzWdNS4H;$0k^R|DaChlp0H;{eSBzmY<ud9M@A<p6iPxyfF!EM
zmx`EkH@DXI_8R-T6orPyM#cS$ACym*ovG*%AqF4bmGzG~w_&qI&-V)sZ6SMmzt&!0
z4}jQ8Lbqq*rf3C0UEwf}15LW8aKxjyQ3+MUn_%NJv}61J=)sT9LJ#Q)Jhxq(V_`~K
zUq7h-i{3^_mDLWnStsC(ADsGndv&=+Q2GcOFVpRh?BgI8H7Q|%UeCo7i0a|jEptVd
zgO7Tq1VUQzScCM90pGmn60$}q`Z@a+)lWSiL#p;~5X1$zA8H0C=`6XqyMqQG^W~uY
zJ)NgA$z%64`L$|q+_qL{PP$m3V{@#qzS)0hB*mUl@;r_dNWq&~UPXT!)yY6+5B$zQ
z;`XuLz5mnqYBn)OP*mu?agdF2QAzr4_B?f+f^>J2bJ4FDW%{?wB*^x7W=|qEA%Fr^
z0G2+OpCOKwOg$eHn;m*A7}$!L97+<xe)Eqe8=}&P))vg`<chB3&E4YTcyaTFBdT9@
zHDRDN&%%-B<xAA)A5NL<;^r1~l1D2k)0K!9`{Mnt&a?I!o%PW1B?jjW%83Wn^rVUk
z*+Tc4A>1~Z{9eWm$*1^ru344BonrYoO)KjNWO7ivKqg?>UFSl3zUPZ72lWa>T4cF6
z@Vy=!sbaKY9sD9x%#rLtepF8Ux~TLKGmjx6djO=52tLPjmDqf@m5qx_;p#7zC55V$
zH;H35GoBASW1G!{P&&#UeEaO;iMhs+6TKSOKphQdKLHcVKK}~CQ>QxG+J;<YQ6-}B
z%=su@Ur3ZW2tw*AQ1VK9KLx_NhvD~(wSO5SdI^A+S4!YqhnRHFpMS1_xo5uXx7v=%
z(?npU3cuURSKT{!%wE2fzx;#xA9)1`6&$%Nja+fTQ4y@%N)^5Ua<*mU9wVXBcG`b;
zP?W~<26JwistaLn#aAevmVW;^omPze%L4PQ$>B~4Ok2@kodtHk3=@AGWKiCf!5h6i
ziIfq#?(tpgiyEz!MBRMy@hCy}Yy1z5X?zgN<iErGvHk@CYs3FaO=g;nJ(??MCm=8v
zJ@GI9ggmG?exx5YcnDV~M~LH-4+a<7O}p#IZ!%DC{Asecka1erVe>j&p~i8u<1hq+
zL8a3|(#M2aUW@vu0lqQJfUxF$N;u;0PxudB(@ERLiVCs2`*_PAH29$sOfpRW7qE+S
zqi$+M%|O_bt=Mocym?YH0C7a-$!?6eU1XtsGSSzE1`JTh5X^4ulY~Vu<Cd<pFuuqh
zu%C?q3{Zu<=`);+u>K%<M(O&K+lTPzyd)z&;od|W2Dzl1?CiUx#?(qj29uPxI(cPn
zCGS~0msKhI0ltGUngH4v)&juX@FJ5U^^{ltJn;RK$X4^a>&+O92h=wpigyogh;2y{
z>ECtzkHPa^axToI&xkzMA6eHY=FE{4K_Bm_O}wKx7|w+6l{f4sCJ{>U&11ye{K7(D
z!L!FV`!5#tcUwMf{9{=>XcX;Ebwmn}Tf)W}j}nOI(8<yBn2(8x0TdF)^iR^UZbH-{
z>W~+F>FyqLk;Pa`)g99>AZ!lG$WT^b?Ld8nwgn9dOvZobk;T=uUh>QQOI+o5QS%ZH
zM@LanWVbS01bGq#@XFT@I7tT?izxorsG_FCB@c@f9CBTP(q(9dUuaQ?Kjgn#FI0B-
z-YF~yWJ(g`S%OTo=oNu@@b@WTrtqmNq7+A-JgX<-SG+JjjYEzKsIos3ADFwmqcL(?
z<l;X;K`c&Ll86qBMbKf6*y0^39U97@scOqlzvXS%ZRpjT>+0CC0I{>@bUG@<!_Eiy
zYaPpG?`yHS3OjC0g^RH0<H0OQ7&dW=0$H=hS`u(VvPuHM)Odb!TDk%aPoIAA+1_yc
zw!VppMv<#5S5>8eynHdFC3m`k?|Ao)08F;_q&YfxIN~5b6cb8JXD1i1-A5jS7DF(i
z?!$)<KXM+y4?Q%40@(1J_yl6kAyy2zcMrQD*e=N0nOy>iR$mxB6Q~Blt`2<B<@n!?
zH0=V<)7fjCHs}_jDt0swW`RXE!#s-^2s!9GN4&Hs1rORXumq_9WW-uU#bI09;XM2|
zv6DV_A2Z{gs_miUR%WO~%N<<<%va3u!yKeH1wn|JeZB@tlDaxmEGr$o)7BNJ9C1`f
z#|$F+&JU<B6rwNx{o}wZg=6pzadQR|$SY{&VFnS#{sKv~sYJXlJqhp(nCTO4IUo^%
z>bL35mQnC=6W(o+y1CVRj59E6XHJ3o?~na;uaxZ0TzYra$hl4XmL;3v`3JR&qcR({
z9v|ibZx14ru9l5FMgX@8%<Xu1OrR*TQ61d3!#ovo5A%>1a3q{J;>GKhOS+Uk#jE4g
z`==ZPBH4(=%gZa{dC+j-6!iG4BsD3b&JA7962woDrru;-pGZ%QdvD(<9UgweU6#u*
zorMH}a|y5%e@y)wtZXOcS#~!ZrXVub-w@o(H<j6v*(_g`tffRc?-dxr8)cxTS#jJ*
zvy!plzrT;$%x_Ur=8%eak>xAkTwNpVHoDc#r8hFIWb=ssdi7NL7q8{RjJEVo?d>Lp
zhLC9BG+n3I)81Zy+61#4Jb!>?IDj=is>l}}zNr+E(hUv#uiAzCw^D1#Q*<p)r^j=T
zdnx_u^_plj*|$m+&slmB(+ctlKzrEG=4<lDspkf_cb<PF)g*j;2eHl7#BxpVHBEZw
z_WO;t-mo2Dn?k74)|0#K$xdL5vTGsq$w!Dg{`hXE;)#Ot)l01+CMX1?kBBnSI=%hA
zfwb7SzG@*k>CZzOE|&3XP5HX7x$#`nfVElSpnLlPM~%7r|K32)K*Eprw;xy@ZX6-d
zA)WZ3NhE)PVrc~?ApXzA366bbJ4sE73uZGTP+a%)_WD8><_rRR+cJOHi@WX4zOzpu
z?nAVyIntanGY#Gwq23LDW;dAEH)+1#+UZ`8z_#tZ*Lr^sCHKX9z5K6#15WUPUsvf(
zt!nhAtWAUN)#w!OXF2)}*?A-vVl1=3ccxCrI<aNcXYqUBIs(N+B0p+yd){_jX^CH>
zCgGvbbK&~p)UCju%;rH_L<R?%FI0>Qu>Lg;xgzfPEZCS&pK=bHy_*5VA=y>VdRWT4
zGL-54>zWEFhSJp4X}xaWcM@)cy{?Vb?c&5)(Z_#`NAkAfEHy!#W5B&}zTZxsjY`=&
z`sFRX#OCtnOCIqxn<_|Z?I#m@3_=&Is{OPivB<%OL`7C}Q$RWT(OS}w>2$K;rM|LN
z5Y=c%<$dDJK~`yh(0piWYSPMeSqpt}W0zjq1C8L+BeqIc=61>vkIgA2H+S?Z7R}ug
zD`BQ1=^rLCGgn4GJgGTk{5SQ{$Od7&5&U)Fe}5Oh;xW}ZGHrhK#hk}$HFwLEg8cqE
zn(7|vWYg!M=5Hh)4Y1UMoCpNV$H=|#XYS5-y#&ghr)NgVd*S{T)fZR$u_v}zZS>>0
z)AVJfpt3#stYTQP8s*T{`h1gN!m9Ate@(Vc)$i}E9^awZb#Q8>KRfXLk7!!b5byUE
z8-X*qCydTDBfta^(h9x3XjhwM#8Mh=pF(>4zH4YrH3u2&71PW4ld|2}PD`+UF9W$e
zVY0U6{q7oXraxmrvqVVUU%BFzQ8~1K(T8%2kOhydiP*RB%${%}^M~HddwI!3%@Y+k
zB`0a}{9J0E-X*h8%5`-zH@y3xxbMxH?YcQ9@ciJH)NMb8cf`d_YiG`oNGbu*jwYVg
zg10ESRYFIre*IeS4jbFpsKM(#{q}wR4&^@mnb`CK`v1KdzhBqmB4j9bF>L7b7zn|6
z+8sONRpP3rLL5qS<j8<r@y)e<p8|Ii3lA-`Zt+^B`;)TK`KNsDHbKXoPcP}lS&g=y
z!6WwHJBcvP`?QVhNWd}Gr9^HDYpR~&Tr0Fqk$;0iqno*Ospa$@KN0=IAh9eL*X23<
zF73?^kt^cGPMz1n&UGE@Y&4YEv`RQ`%=^@Q0l)kA8*`VbpoYC)vL)8A6|CY}5RAvS
zhl&X5H;fka^z`7-YP3ayFnQl3I}_2&YSjzmsdcOwk}7Rm49WZOfY1H1@Nb&7a#`Yg
zbDHe0Sl<8Macv224*L)MKL;I;LsqTlJF|_XPl!a4U1JbzA_4-j49hr(%gD$8&<*>6
zZn7ao#Y92htCrC(HV7olV~I^(T8PiQ^P;z!vh85=s>|y8K>bp3cLTpP`Ge>dkz5;a
z)=>!eB`6;l7_Hy{UxFX?{^R?*YR}1k07m6ehk4L9y$&qW7vh{Ge(oba#q#BWxHf~m
zr8{gh-wwQgG!vs^<h=dg&!zQipl81S*$WdTvX*kb`JZAH3t>LlYvNu@HcBi{6i`*b
z-4h|ntP*Of$Me&(d}*H8S<H9Xn{Pqpq0hqnRqr9*iB`^l|9yRV{#Dc~SY`<ogVs`7
z{hI$zOn|dDfl*udJ{*@wF8CuqD7j#<cRY_aG-u-dCgpKc&)3dp2E~+2d?s?+U!DHD
zQA|T(E`Z#1!>?LV-u)EoQ1t)qF33tr8Af)e4=ju%o+keJB-7K9y?nHi#r7@hV++M5
zKQo1R-HAN{Lw$<t{o&7@bxZ$ulhaC$Y07R|MI#bQXe5r(#W!yxVS`3177~w=p+{$|
zxUNq1tB^ij_^iyTGQ3rKTgTPEcFS_{wmcsP&tC0)bn#KC$l&=mQv9(Xd6$|i%g79q
zAy>-r%{%D*MIuFngaWONkCGo!f;zgC-$-2-vh0<qO6x<b&+m%;8Xxz;KIBw5^k|xE
zlo8u4PUobKx0lb0<nz)>?jv`$WnD-lYst={t7Fge4?)`%6o!K)3^-JzRQJKmlq`34
z_nb?tv*s)YM@gSJRz`C_W}VfalQ^n#GJOzjaZKIt#NkEIac-o%iGM??fWQK+AE7#a
z>s)qDPWdat_o}VvzWw~i6l}XN@%uMs<0zs4&MOjC`?s@bg$<5X7_rchP^GT}ZD}d<
z6p=*a&0`RTYJ5~j&IdJlNZi0U9Z;aC=8-;FavUVsReoUf=K}{S_%~GKdh74pSj2C;
zxR>(&s_%wep&-Hdh?-Kt{Ncif+FIJBbCS`dXN22&AE$G6VPMAwg=4(WUy}~sS}3|=
zA=+jpA&}o}52Zu*M2axt^)41&ik|zkuz;w`@CH$c+uxi<8Wbcs%_HFBqFtWiAi4bn
zCY$>D62!Op{~k~NBQun7U};-m%YB(u)@3aTxA9r!^`&Nv!0_~waPH0mkpC=$Cmccp
zru0Wb-S#UeXt&Cf&<TFph(a}`J;{_lu(IysN6BbyhD!MEF#FEl0BXkADsMpvh?G5r
z6Nysp)D9N*ocoPQ`lM%uMEdaqpQENnI8pdw%m?Cki=b#+6f<srA>9}f1}NyHqX+pu
zC3?jdW=%UF#C2LQt^_gyLft3#{@q2?T5-lS3M5FOhj3yl$5Y<b-M#PM_!wRcmacQq
z3nNCc3TFaC<sE)i50F@}yp6`^OB(oBxv!mMp7PNsbXL1AHaxWi={CuD`%-9PQ@pfP
z1Y%pyx_v%6Sn1a7$A7P7RKatGpX4GN1M~XJwqLa}K1$_KPwd@`l%#0O#}Ik$WeL_N
z_W#?`LQPe?)L%kt+J?QWz`4)2zv5;d#<+^!tR(WMs;_TWBo=zeMUAYedJ0=s`*C9!
z3y_=OTW*osf~r{H^Y$0=Kn)Yf609v9Ol^8dzwa6!qqXaWQCc6auEQ1eqEX-abBN@Z
z>__1!b$f~^L=6oM2{!5*kmaTQiWLCV6EhW{LMIo{v3bOQk&Q9H5~lKuxBJ-wV8tQo
zhja`6TRgPo9JcWTVuz04nF1#rlR(;VdK|&{IS|!nYi$ihGxF?VA2eHgVY>fLym5!b
zq9#SNL<2@%B)87K{x36*P5!P;%(RA6@eQ-+W$EngyQK7qu^(8Qyo3H7j$v9=O{4i%
z!Qg9FwF?4MoRQ&3yr0L$g=q#v!Fb_&LR-dh9Yq#<05o0xI989%Vc~(@2->XIuOBMH
z?=lbG6Y@l3o6boHv);vc4O21zt6Q%CYJivs9Eswag_whYeZZY#|D^ZL#T}#!i*hAw
zWv@$X21dunr8z>9Y5}EbAex%Q_Us87j~^uS!1Q0X!5Xx??@r}+OpW|ip@iBUrpdgx
zmE2XOj#g~w+kVvAiR|1YVYG`?43i}4QFP3F=VXt3?X$gOZ(Z5{PhG)yf+aDL@8)3=
zNu5eVGHu-PZRp16Xk-6dqof-}eq#kOd5x0^D{Eo|Nna$Af8D0`d!k${qHVxt!iOTq
zB;-H|*x0!FA{_}Wn183sL$SIj@>uT*Fl#~7I)36hh*>QyN4WM^S>>3CWg&9=gsimA
zI!f_fYJ+>Ubp@;Wy2*Ncge7+^udX5Kk3=5m91%#vHTd<5gPr|yG4YKlKBm5@Nw8c`
zf-OeK)ydlWwszYIW8<m;zKF1}oOACebM2Mo<T%*rpl;B6KUxXW68a~Dr<ni%gNkHw
z5;xc>%xO<8<I<z$WW`C6jctN?3;~X0+uebNa}r$2zA>8a^Gpc&>FX<pst!eJ?GgJY
zN-hM(L>Fw<`I|p~`t%7EB;S#<|CZKhs%JMehIyO{o{F*na2D><iVMy}3+Dp{_i2Gu
zr--}8?zk_IQk!$>?RajzY>fEDB!S%@n3+T9>iP$1H{Y^r%=dOHK5brm`+&5^PVJ{p
z!z9cCD39sxyK1Dkqh0Au(TWD?93jmZI*+8XiBQYO@b-tLv$D=?|Nm#pY16iOOECHP
zmdAfnWqD{$27;Quwl^@&`!_O{d;b^HoawzWx2+AI=U!3>kGO&jGRE2Uz6U`zI0wx#
z0iWa)q)FlY|IBmudl&!pDMQ#xxHm!8iO9=8Z5}z3`4w8+aBfDD>H!%3Wv>h~kkHQq
zP6N)H)Ry#?QOr0|Z1d^B!;Wu&blAWc1XlIL@6#T(d7dsVp(&&J`Q&W+=tJe1nK+t+
z9TN3d#ncmm{5!J-(0@bX)iO8!uUwM6qN!aG;rg2(P0eKyXZe}^oq{6Bj03ti8ln8>
zTH3>Qi(@HA2jt&06qB6<4$1K8uYHD=hV5UOBj0Il<Je$fY)nlmO;IGAOLAmg81>Jk
z>Pa!t-e_z5b0?kVSntQ<Cf`gmPS<lYf2AX#@3k%Op9Xqs^kWyfSS;{c0?)e<Zm;w5
zBKDodU=!TR6<_73`hB9J=wqNc@ZAN7122U*sGcK_MaJ)n<94>UuN`%R;rP*`N4P3b
z^WmOkX@Z9wd{?Gs@onYW!4~=G@Zp7-0Fp#7GYj`udcEsp({@%?TN@h#mhGuA+vxq#
zCZYFN<zXTkHXg$IM7mKw6kPp7mJwNx2{i%WL`TCB>a*nf^-bJ2OvU#E%Z<cbA>Xwh
zV6#Ez2C;JhjAAusWnsw{n?{qL#1p28P{yNcCX^L^G@EdncKF>DWvC3D$EJp~53Vg5
z|AGiIPo4qHs-pvi1-yL(#>J{5)IqEN&7d>JZmK7up6S4FzkTdY_r&A;ZRGRAFgI#j
z2t8wcI^eu!f0%vd2TWX&G)$(5T052SG%89%=9gO$8>Mf_T!942m`)WRVXT>eep#|5
z8`f=kW3=`B^qmj4tzATJ2fUN27@u63YieOMBp58-XHk>(UKzjJ7wW`%Wryao+{0zT
zijQa)vq>I5j);Y6etmV8^F<U&0J0p;O%ABBi-n)<`6GXkJ8VT?PwVSvV-SB%BCUOl
zkEWz$l)*^$!W+Ql{2PXuE-8QC@xVf`TD)8Jjr3821R~&LA=vEKKJw`vLi*c>A;J}P
ziJ4eDqwD>KyL8lki;P$U2X(i^8u5YE4^2l(FPL_PCWr!m&+QYY`PyUp2hL7X3@vDo
zkR$4_Ljm|eUe`LiNu;-CztJp2gVgA&$}I_LX|*mBV`JoP=Ff|epms;W2UA{=*{hp3
zZ+=?bLV6}lyg)Nq)a6-P+8s59LbF$gJgfgr(Bdki-_rRth-lWilh6vZySnwV77Ym|
z6&bHB{u$+&7UgAtFDok(B_c*p|48}WyLS&yUw|d%akwt(rvrj>aG-;GWXJaHfOBFc
zL`K{b>JAt+G*Y<utj0@bo1jJ`(2wH?vF*&Y!7zbzxQeYiJn|hGZUos29Bz1l#nbiX
zV$fs8m(!1=`5CxLTw7|M20AS*q{n~JK76drpn`_Pd4-E4{EU9)*lzy3oh0y`0xKy<
z=#OR$CpkQ^WJ}OLWoE{e66K-59qmbXmPeq8mgT5_GW!lV-q8(4X+N;9LS2L_^W<?Q
zHL(hH8--|=uUd|stn0tr(_%K=oe*-pALDlJ9M6{~^q;cb=)$fwSv%hB*Em0{yqPnO
zo6Sqdd8Nl{_QCq=6v{%buq;0aFf>13l;)`0#_{Bn*i0z4<XnI4uUOB}%D;VpHApSY
zbEfX50yiP99ls3-^XRy9%XYXHVPA6>%<t{k`R_E77gcq#-&CODm3H{$Lw*fhr|f1v
z*<3>Q6gj8_b~~wP?SZx@!9{DVj0~0V8Wfv}1wY5t|BzDlRR3P`+kBY>hHG-M4#rN7
zY%8JP@3N3wn+!>`HyW$|SWj8Ei+U%E5bpKmfE}EZki{I88WsQU)4YEWtRoA^4BMny
z*zrVbN$4fOCd%#;2_RIA6<J@r8rZEHEuG*f#St<c<D`11y|))?4zy|Cb1=zwdPYo-
z$3%0z?j+g{X4gATf40amdN*)3@uZ?z#y(bcJ)u2Jv`DKN+`Nq{z*3Xws;*@(N^3cc
zR3Q@Xoexj>!;dA)JPm2!%$-2JlGxOxnfxVN{-#0#=+5;+%$2HmKp<=*&)OMXTiw{M
zNa|DmYcV{kqV}<VF>)xThb=lel;mOfeUF<r%-fv@6ciNVZajFg@B`!h*q;Shn)GXy
zS*lN+TGk}?KT2cJ-Fjt8_^=?!+VF&}JvHf&9UbY^=m3Z5M9AL0x~2?jQsglkX%5rW
z=`@O|Fx*5C_93rHumz{~j^2`z67-wSNqE3Aj?09nBJ!9(RDESnZRMR`yW`W|2N?+p
zh;)D8Gj2NiJJ#E^X?s$6-{&{qm{x1Rpe4i^In58=Q#ft-r{(#S<~9=evq)58W57cy
zTiI`#ymL=~qVDg|_^z?Ja4U)dm0=pZfx>Mco(cu`=yugen2WM%l>TTj=K~a-S|v0c
z?F_yv^e@|vsaJ*>3}IwLUABdhvAwPB=yu=fk$3b*v)WMzv#v$7inz76g1_D&DeGV_
zm_62T1Z@(B{`FWYKkYP1l9KSzXEdaA-hq?@fz&<Xj)E^6t*0$SS+`L%BY^PYVwfm6
zA?K5_uQ5;|Ef&TKy%^c=u7w#V_fDq?nXlsU#baf#D)?cOpXMD|8`KB;t|`x|HH9ou
zJ=KkI%GK+y8%iEsCW^e9-<pYwXY~8b?FOFT+N#N3`q9IORaXqJUDzWjMYOh4YLziD
zR<$V2$Hr7Wq(8);+a0Yee~U+I+pb2g{B8}OwTrDeDUn#UYz0TRPTf4*p9PFXYP@d|
zf-cFK?cd5gXKiy%m7s%*!d};V_eE=^ioeb7*RF+k=@Ma>n13ynzo!=alA_|f3-omi
zq$EQJ>-6k&>=^a^gV}`}*RKu!y_s`_)PL@@jdAI1BQ2UjS6xL17p=-w%HL6aVQohR
zBhs;g@(<B(z9Hmx^=eoK1<Dg({or$fkl`QWI$zt%z5q;h2p2Zovc#ZrVEU7^A|WSt
zjCl^rqy{VV@w@MX@*-Z>=XacKq&PHU_js)NwN6^E=~~7?R%=>r@zvn0VD-cV;b6<h
zFpy-Pl>e^GffAstjfRv|VXZK-{cbhuT?AESoGq4#n+Kx?gnNa!H18vO>+&Wb0KTWc
z;~KX_Tpk}Sb0n#f%U#wzdLOZzG2e_Wndh+(6Dm&I9dY|53jFL?e6ICRsqpw~?-Lfe
zA-^m*74}f`){V)DiPFLF<C>=90P21M6fGe=eH7dtFGPN?&qS50&GE*CsNJt%VwDMy
zdTqY&aO+8(OA08zo)>*YenaiudfUgd%d3{&R5uSB8Eu;eiJ5qL_erZ)iPI7J2aZg0
zOxFZHjm+kFFaL%0q39I5Xyh?4s`hYzqzh_4TY%Qj`+yZB1pc}2JHIx4=4_exj77)K
z2uUZ!9k;Ds@INisz=r1sYkQnu&47)ywMVIWskwNVW^FTXM4f6k1nel#Nr*bfcq+kj
zxiemXcro-3S&YrN=o|Q&3X+mmlF=pR&S=Mh-}zopiC3JRl>ISymHoV#)y%D`EBhe0
z>6>&fG&3MB$xJ@08L&Jm=&Am`41`iJ*HF`yyi}Kb5Nd0AE!QF>#Uc0LFGZ;%9;Z{+
zTOD<NhOW0C^th>3qW(9smK+!?L9pS^LN^c<lhwi}2o-$iktJWdQRdg?!z;2ex+5OS
zJe*%thxIiPgRC2Fb^Czr#Ur`ylywFpZV2Sw=rB3w)wyF&rNrQHkXQA9&v70js%61x
z4~0A@)l0g}){nP|Eu^08rYODw$@?Z>0gLhMW_POwv)zBc7~G|P-%s3<;tV3DqN@xW
zwJUlzuI{>3s{B!G3#0UA{cP?<*+zM1G=>3ZJy~jJW8?2B;e?&q+BzvK%ek-YI&3*t
zjf{-OZ?tt%n*M$2wAWTnDc(8wK*={Jp<5=&hU`~w##%mxC-g~*hPpZmz`<{t)Tagw
z25*5c)ZL9w`<3X>+P9FlX*9{K)1&^P+CBX6?9H{K&riSqwHf_@`7TG9AEkO1YWJ`N
ze=^|Pdu!Xr^BN2;g=P#AH38;b-K(^XKGCP@Kg96wdi|sRM8mHe%EA$KyvtajF?@4$
z)E0>h-v^PzYxcSi8V1H!uO5hRW<2xCI&z;kZDB5YD1>xQylMhZ9VKK?R5w9ev1>Oz
zzj80*y(RHsTI{lFyI-vY3<hp;d^Nuvd#?Vhjk|tWby>(Fv`R&-_c=pl)s^N?(mF~S
zU5AdaZK96<Ji1LWn8M?k`#FtHlVLxg#Y5EjDkVxwOH~sjBn@xbxVb%Sm_{^rPft(n
zF8*I{2tkvtlM9Rt3~IxdVO~XJ$G2QrT>QbxMt(A^8%c!zk<a<Hr0eYFb$Jr&kYj=q
z@3Gwhz5MElVP<N;9QB-`jF#*;NGDR^z+Wqos?d8xw|#5Wh*Th-_Upi$i*6Z*mxeeo
zE|jSgkUSowxR@k-;VT_5DQgcbJ=Wh6n}3piGTt1OHSVR&r4!6CQq4SBBj(r&RnQ|#
zSGib5#Xq0WvbK|qZaOp)b62eBCj<<7yvL=PBaLZR;Mq`XUA~2rb-l%!q^QGuv+sP~
z@?`n?G=^G;MZj1bWcy?o6_)qJT^3||v~x2ZfH6!r(N#!U+^3~#Nl-k)JVUZ562x8T
zZg0Ii2f9thkQnP-zKuV}twX?w{np*tspZ;=GdUB*t{}i$PiLZ+$YQ|QT`$3~HG<yX
zZz6hUi^l2qFU*h}I?v4B@ZAL_;ccT=^{o2gyc{3DY$+S7i=?JUuDzf>>#S#Gmu^Y9
zwPDvr>eIyc12Ru{k}#g-TVC)gRZrZ{L>uaH9+82l7)3PdN^Ys~)<KZ}`le`8;ORA}
z4c?MD)1|IMAG1HefnHq92^NKAM>E@;U+#;5RrZ?VD!Q|qz!Z^;MhG3uKg)*P4+0wp
zQ}u;ATD|zsE8rBcib5^%GxQM?*9h-HFIOlu*q@*<RO^+(QT$Kcz%+)r#`1Qw#4e8e
zirM!msSc||)GI#TozcAFnxK+wxS!z)+!O(p=<<=<XlM^|7vR<aaiyiDQi9>A0i?+x
zay2{V0u0iPy-Itna^TYuj{Nw^9RYZ1rBtw*(M*rwtapSsa>Tcsu(#KI*_Io`25vV{
zp)aR9<W=BLXV?%}nb0zdHj<Vs9s3XxS7+{Fi16eI$2{G!pG|5h3@9_40&4H<wl<`b
zY}g-oCB7}c??RtL{;3})`$N_$6n8&>wi~24wpWsz(I4K848A`+@>JrQ*N*^-gRj@v
zMani1AG_)bE^wW?4nw(1^>!0<O1Ma^7v_+|4PVPLu6GF#dzQ0ZnyyKI8HWLAqP|fN
zJS1)~Mx#k;v8ZLV%HR%cBTZ!y{&y*;wRSL6GL=3F3)_0W;~48(9<>Rfn%Y`T4ULT?
zNRK;g8c6s~)YRRmDxjC%+fPJD-D3Dk)>=OHWcG=pR<Eh+yZ)64bc(|j+m0tGka&J6
zYWw6ufU+$_<z(D;`lB5#kM4z<iJ6=V09lWWQ4?j%^id!Di8kQa&4%|U)#7`j58x!R
zb`^RRe>s-WhNb;4-uh2KGtFOgKXx~y?V{H2jX}Yjzmr4|_XA80#6Mntdq9mF=VGXQ
zGz?gDTF?rHDNDiYkFdcwevddF{3DRog1@dqg=Em?&43-4%njeA<ntCd76hK1lR%3B
zdBh`wI#6AFR3Cz1_~>0$vShBKz}W)+4f8B`p1R73@^|OX?V8s=p!>Xeh43X%#=+F8
zo67l5hSdrQS1py{i4*>QOe(7QS{+iy1Y*}Nrgpb2Kal29TuaPTbPjp^!(@^)m0m5E
zxXF=sHf}BUa=JU^kxFC(Mdz<ne@EX!?)mg*ja?KKX#I8?>yzn9a6GhUo@wvv!)}iD
z5eDlJ=n=C6yVgB&?>)6kDx%b5{rIYMzN@1VXi;+9Lf8jdNp>78nfeTJZyzw13Q<4e
z`k9kq19Nd)71?`Ys<rKIHTn{|_gc{1fL|CE6I4e2s$1nIAG7Wk4AX?|;SR%=wkc^H
zh^N#)_It=Gt`t1b?c4q=@$HS_x~LxF?aYM1P94#Wq-bq*u6vd6J0m!Adw1BDt)NmA
z6hx(rqE7|DItcUSea;2l-d<jyjz6e@Kn1H%>&n2i2h>f_^rno0z8-dtf;kSI3oTEX
zWW#7Px(YWH>oa(y!Y+Z{0=jf;(?1&?@{4PSXd^8Ln+Q5;N<;{YM9z>+rkaChfbaE@
znfv;dmb(it{U_hL%=(zh^>>n#=j88*`@%ISk8vhyN!9Ci%s<Q{p`gHJ>aQXeC2n5w
zSbKxOZPkQRZ!WaqYJYQDfEVY~tB~i<@z6f`v=92jsy!70)3<F~_zT1ex2ayICg2GN
z@x3|b(2<tc2$af{tsR&KbFiUFfw$HtAmjC9{wen}j^3FkFY?qSWnxA>N54cvzvSi5
zb!9t`6zhn@&v}7sv=f29iHhfw*AETFsy~-Y^(gIyysZW&aAeN!s(pk@=nEk!{u!eQ
zG1CDQ%i!iRVpc<7F0cmG0L)s+eQ!P0Ur@#^zSNfG=DnQlAbP={$1;2GuN)0D-c1re
zGcYo?NMDj*<9*qvL#xw*hXQ>`+I{7HyAE^#V2}H~j`e)d$Nr8|8vDt>n)R3I3(0BT
zQLL9rt?HbY%(kYLnXPk)8&hC?Ik2~%PccQBLx<4PqKj^X!IAFUiPNW_#KwLjJJ>WH
za0B}SX96r(d&I?+z2`+MC7{u=<lgLBSdx%n*_Ee`GhN3)J91z%9$Ex3zB3szBk)>}
zGQBt)5R=F*@i$yAG~uqMk_OH-7%D*M22C8EHpsz7X2Tq|;tqLAJmnm?78{B|5fn|B
zhqH&M%j_#L7CD=JPNg+E(a(=E-MH)X=lwD=tPC4)9!4Js-2(6PxBV&61H3>+v5H+2
zGZbcl@+>MUs(FvQKDAbiQ-NJnXKuaYYLylHtPuONI3cEt$a)#uy7pR|H{sUDKJeDE
zNM&#0xx9neI>K}>Zo`Rh{A?^C7OTHZ(8<dxo#&93ggnHZ)IU=%&N%l;bl<+SNQ=kh
z22W)vrpVsG_nsC*ZHMk1P;=;9T8Y0uc;EoV;pgUeX_<wjb%3xSBX%B36bYi5X%HB7
zE8?dv0ZHBIJ4{T9UT!6}SRRuEiIt?_95ox4d#^-am6h)-bgJ!PDZE>=-MnmZ2}1Eu
zWi18eJ?B}^R74zhH_TMbRXKY$7Ph$2Q3!;qZt?u5C9CcFrzImx$ymQ*T!&N+A@0rS
zou@cJg&n4e-92%`7rjk%bu!TGoj8Hx*2tU}4W&w&I6)P51u91j1^KI<9Cy9{-0(DK
z{|5zFm}^XTi9Li0$LBqH$ORS)^r}rtHj<_o)^w=<V20X+H+6!aSxZMJq}s~ze1FYA
z_;E3I=P025OFXZ{u=3V6+_vpcYANitT5YiUcJYnCzOGG@)Yj40OI6KT!!L2G*lzJ9
zgg$*1Ouprm9jBA4H(*f9W#>`fOs)Huxemx}=yKvqaNsIiYIM%!tn2VVjFV-Ec;R3;
zaGjysH}M~Z2n3Cq4tP8<v-E;)%N{cDOKx&x?iVEqXjxx~UqaKE2oExe?Wqg942Cv8
z^%6hB1)ItvFq}mEyZXj*o8^0*iaLilLwb^#qI%x`$n?t)ze)=V;-4?ag`~BU4RwV1
z@wiw|UsJ^$a_N$XAZxUCKKNF2dYkuezkf<}i_|@M_6{646#X#N5y<6T;#AMocy}kX
zZ+qXx&MvlJYKV&_bT1Qw&w*8y6&6}sx9T&c=6Y6y_Q{M>TbFKZTUg}pS=?&vR7r|)
zEmD#?<9;S1{!C_KUGV%0!#gZEOXM{t!)jj7o%c%I%?IxuJ84~1+_bM*z3;>o&e~%w
z*T~`qd8T_MTL3!!4@^LdiRZQi&br{ENo2zkH7Wcsy;c~?R&dege!a|s$U=vBDpLxr
z1TG@&MHkE`cdbnYqNS>uFa`a)*0dXr%4qN`o<N39NNfE5uYK=ZkwK`=94t(?pS8}z
zO4NXzi@x$x{2Rrh;??~DLE+3Lyd&Xb9BaI+=fuW>R5~9qjc5vJZCbpNBqgWuT3e7W
zvKvR)_qKL8zFjVVK#1PF$uqm{2X}e%h4oe|_8IjD_oG9(D2Q0IGIg=Dd!u6c8mqFe
zQAbl=``xny3Hmpe%FW==By?*nS|=%scD_H^cGPh92?`4fi*>A{ida_>q|JHI2{7BU
zMQ-VsqSe-d?kNw2^>jbY4RS`<I3-#D%4^TOTR#()R~SDtJ+vLPIljYoDrwLB=uAl2
zH@7RPmKuX%!9OxA7H2oT;7&>Fj1|4~@i0D(iOtJ3%VP>#S9jDj{p6L#*-A*L=A6Xq
zf`S(5*m!bsw&vBtxnkhr+Ih25QET~2>F&@{E2z1kBezri{*CfA<=@jMtjruVLeq!e
zZNw?TwXl4#TKs)}gsKAns;Mx)1;d$-JGA`C^K7!-3wsv%3<7Eqco}-}Tg2TVIG!Wh
zQZ_0ts{9&J4fWJqs1-sULph{3tMA17yv*)DllESn`7(K8#3I)0@N{<aP)KF_lYg=+
zpP{*Wi>zAq_8p`kYt_yV#E@&w%AWeU3vE>2N;hOFe}1~aZ1c6j{VuNED8$g%_b^a?
zT>R{mXdE5gj;+?gmL^@eH?_R`(w0^tB}Y#ex}ASSY0|aTn!^)vWh{kca6@g01XOhA
zBtoS}5<=g&bI~UIOM}w-7OG0;x8DQ$*5(Vy1>N{K0e&?~3G~Xa1HLrLDuWm`wLnSA
zufZ_Jd_Z2K;2MWgaL@Ln>vkc#;t%*{oQU&2TX`)xNi-qusn`1A_n>upYRJiE21m?J
zzRBdbLqKKod8=9(5)PAqQ=TdFIdq23M=7FiXzh8@&y73J((SBHOcP?|H@$LM1d}rY
z0F9jp)8^trNBN&b<9PBH98OU(bedB#sr*Q)91`3+QM~W#_shW=+L|))M477ki?TM&
zDq(zP!6|Y^010j>jzT>24_f``x}heZIJ>n)&5K%r6w2hzJcJl$9EHY{Olke9KG-Yg
ztyb8unplk6Qk~zpPGoxzwVRZhWXOrfIjyQJl;8jM`OEr+*3!)aeP=D!*8S>C7N4}=
z6Y9oD7)d%%2z!-c#+Z?nrNMZCv}o@8dxW){^F!Do<C2c2!}Z~NTW;7T2)}@XcyL@L
zbg9y`z^<VB7<27dc!-0^wCC9147q+WosnaisEtuKpd(7`*T1glvT4Z_?o{5^I>5iu
zs98g=YGKg_&}}$WjnNmlNt%{C`9)Hj(n<aU@vlQ=DApr&Jxj8KTaai-rhJQu`9i29
zIB<_vvc_den>gfAef!s{a)?I3u8x&&vyArEIWuh(6_4}AoCC8s^SST#RF4F|IyKXs
zeDn+fqzqH|A7`}>5tra+!d&y@%L8l-G6xT09*P>@A;)F-FvW_4W#Iv~#d3pLY901x
z!i%Ok^G%d7g&LvkflXuG&Uq7(XOy|67QDh&7Is6D`7wzzZ;&&O<+zm8p|q~BrJK2|
z4+rjjb5!h){n>Zk*wsY!-Cq{RCwENDJqftaoqYy4(F$>O*3oH%pBe7btI$RVIQ5Zr
z1EQvWTjCsU1dGZ`vxD3*0IenZjJl~;=ye0N`EazYD%Io<Kct`SRpc>IVPl{;az1pg
z#mcQsPYxMBjM|fc)3*B)r|HFgVZ{%_b2&}QVudUyN!Fumw;q1cm0%+r9UP!=7;MHW
z3uUX?>PVC{L*>2u_bEy7%`jC?-eQ=&+j}#0V?c@9LpfUN1M%ys)qb4Pokk8FuTiZ3
zNT+(xpq+8@W#8mRQnKNp5h?*MMf2Q-vTmyhy_2u^dX&r#=5ERz9NRwQ6=dCYwD{k7
zl(}x)WmZKsI^q_AP4}m-^7lj!)l)O)`*l({-B4Fvyf}th2Re`)h3?R+pljRQ*f{a)
zm({l%B&(q^53Kw=>c*z|SkB?pP59hl7+d-Mtr)MGi-%wAwXtsa`J<Fp^2krg-^ajI
zy3y<PJbAs%rn>Kb&rzv&dQB@n&NbT@)S|Or5+6%{bOp5E_VJ0=6+>?Wn5)eX;Ln-&
zP=K3fcBQ&DIj!c{-LlxIS?l%w<&^!aCR8>rtDZwBy=(h$y~k+G=wT*)y1z%%yua+7
zkmPp9zG1%mOjf9P-9wlC34DI{Zh7rtyMIHfXZv+Zdhr;uIVEZ7)~+ttFQUXPxZe=(
zmPX>sFSNZl61e5tweCxx*}_H*l5Wx1t68tozQ11!^h;C8tFcZCViv14c2x_=K=K^G
z0@35_%_HB$`906I&mvjw|H>b`uL%EM+O+Ha>B~l1Ha5wc{BjTt>9k--56lZ>{_*=>
z4n=4LNesA0+7RL=AHFFvTQ$#8WHR2gDDeKCVS-DM&XG{hklvB*i*ImeFPL7ki_!W@
zgu4`u{JXO+4(3Km?%8!z-mi+H`sT;1%JjZl$_fd~Z>_}xdQL<*Y^%Vti|PjlLk7J-
z7}_v-{7ddZA#}+Yy;j7APJ4)mh>%?oMBM?WUI>NPYn0biTD`Tj>L5i!vr=hg>ImsA
zvJ|K(L2yBMtQ`3K5UQcBa)qEU<()Nkbj#wcS7HON2Jg3>(R&>G%`}{5{1L_kguH8M
zqas`9sqlodvhP_oPeQafg82cqiqz!C!2$~yf*k<iMf&k>Ey7~>5EuOgfgh*o1P>n#
z9ks3<<EtOXax3Qk&`MxBv~K8CWPR<iWl0A?Mtop?-2ko&Wi%66w}cc@y;)k1C;U04
z`6AbaGE{#aYRoEoBXe_KJf>s`9{eD3UVrR;eb)}Oo6Vw4GT?f&2;E$=o}7fE7bhpD
zjU>TNGV`8zgJaq6%?Ep2rmJ^Qkt!wJp?yQvK}Tz=8-ADUwBN7c(!G4Cgt^{rN9`PJ
zX?%<(kMH6t->a!nM`ae;<}wgm!15zJN$+-&pJBQgAa^*$q6fS!7TPAfS}yJI<C$f-
z)x2KVR@8i<+xgAG?wxhX8q8M?5if0W54d6%b&NF=dp!RPmCouNY5fJopUVdy-=~n$
zeZ}T0Xn*iVilxZ7CoVjqR^HG(9U2)~pMF5ZP<jswrPD7$hCoTy$?8Ik=h*dn6xs3L
z$_eGIKuE3rZ7*Mg@5L8nbMWgMam3maw^e0W=Gvr-NMN@DAxD~`iCuSG>YcvXo|tc1
z*xv*VcI6PQ;i0YJycnP~J~;px5>lT@U?lW7EkGrydc`8vwW#-C$nTsEH>LP}OrL0}
zZB)M}zsT5o;@Eb=@WLHvvWZX2K1rI(PIupY8@fL$OefL3lQrDta%knMbGp@=@w;*p
z^)}I#zlmv>Y$^VCgiP-IYtI()0on>u1$<z4s_%szqS9lqF*&lv!FFiF@U^LXl-sCD
z{q#IWn@roDnN#+7V*BU~`^Dtq;=)A2E#0v%Nrt)ZiyBNg<ctJd2y#?KIg0rIKDG2t
z%?X$Fw8U8yL>!Jnd<(=(;q%5h@i?QEOk>@fNt?JI6abnJixV1#Z~#`^Xc>hr3=8cg
zv4!j3O4p^YSKB##8oQV;-^;Pz#<g(a!Y1G6`<Slm_XI|d?%bax_w^dZvI9<^xjz>A
z?mEMl5V~bv{Dr}uPy1L^1b2pQ2o?)?ewTu6O8#c0#%Co$i#SnnoIgDKqYIi?q)(ll
z>_wLm05KJxMn|n;#j`!dc(m(#+q&;UFvQ!&(c2GZ`Uq*{3rDVn{ZTbAzH})Qb}1|_
zl@d5squj}RXyTjaT(O<2Ama%i+saYr_2rb$MC*wibkn<J{djWNrk5Ez$Coqb&)OdU
z@_Sk#_yy7Y&=I9p7AJ#O`Mp_=&y0yTG(z|IQ~D0Ob>Gn5i<=9_zQoTk8&*4kMF~wS
za>M^D`B1*wp;VNV@@LBBSlK&TN%T0x2K5!5Xu0n6B}Sw^U#uQF>p79Dd_MDSv=-0?
zn*0}>oDdQXleWi<hfY9J&_NY}sq<S#h+mJkX*8ubH&-BxQ>DzWZgRcHitj|9uKd}$
zpm~NZc9Gge=DUjz@3z<zL4SxVtG)Q;91A*`7ye<G;M0am(klcpvnp1hFO~&7FOt7&
z)FpSh55>Uz3Yirm7jheQ<CQj(I6?%gTXxsRXjC=hBr85z$=4=N1pc`6!u#*o%U+E+
zr3lbjP!A3TtC4uD9tSGi-@CFc%9MNNZ+^^$IjOqMO<Sp>jXG|<J8N6N`tAB=jRzy+
z%B~Zm`8xsv^aRp+Jz3spKk>9M_}=vQ^lVpTaa8LnYEB=2v)b>Zim54yN(mAKwqP%k
zDu!H5MGbt)n5YSrP%U0~(Ljr$I;^2l1DvI~$M?HG$nS%OOtrqjZwI~C`mFLg6=~BB
zF>M^lBzM^hFX<K)g<GdNpUpdesk*e)x_!Hfp*nAvLZW`$;0&15Q)jcr0=b3UpC@um
z+>dhmyGMdiC6xURpW5ra^V~&07e|GbH$LdvIY6lQtKHf@v2JFpSY&I{lKWSPmBqQi
z@Q&>F#45Dv$V^(m5j;IDsD04F_~XM4Ate76!>b8>2t?93`0Wpz<FDVmf$_iyi{=i}
zN<WSzATtqA1Au%@vxybsU<9p^-om-q0<nF=Ro3l2+2ynCa>uU(ud)y?zxHPTm^8HX
zJ)NN{_;ly_VZHHJiq9M5OMh?n6(^Z7PYu0|7COpocW-IO2JSJ=MJHdeUG<`TwJQ}I
zw7yOPK|dX9m|K5|`UIaAt2fIE%`&NNueF}Gu5v5>W}C-OLh~M~p3|Ab(;n~xq_(uR
zw(?)7`1vyp6&!TSfdn5L1B75Xr!w%(T)C2drvNC)pQEGuqqTKYTP7C(Lj-sSE<VHb
zABRjN*c5ygr$D2?gyE6s9%11-7lImX6CF<TGgoKh%2;=49ohYyESkm<wS4*8r2XH!
z1TPvrh!a?ZGI{b^anP&+$!EE{uWaVyUL)n5gkOWFaPXD0m1E))rbE+bB7P?&3>H5f
zv#2<IIqyz^hB^fgsr3r`50xzotXWSfZ5=+XKdfIf3Tr*oD;rKp-chQHc8}d<qQi95
zyhJkE5mb5uMr_t>>AR66ENOnVncOSsNJILF2#b`!nxExkycwiE&U%z!iz6Oy%TZ-R
zKIHB?qdGlcwxB(O0~?4FF|jy?*)Oiahk2$@JIGh-h8_x#VpFisoBUUjXgJOfQrf*`
zdBbm{`s=ktOr*ew&d(pS+dCq!@!r7HNmsY&)2Gmy0Z8fjI`99u`JcS9+K2V|AiT2w
zp?lI?Wr(MZ0207UJ<bX>X+ig#*Y(Sn_iIIPQ9Vh)23_;v1L+*_>j}RU4UXd>x21Go
zn(|Rv$97Hja+LIn^T|x&cL11@uR279m~2Nu!~V|(Q^;f;_w3QhZ?9)!%ejlXIJ1Sk
z95?6{4aHo#O4F_vU|BnuL$s{eaE*4a+*1jj?c)w)dUj{a*Nv+iJ#vPvu01JdnS$u1
z^7|+0O9kD)3qIH}q^L~hEr5EFwsXUOGAp!Z#p3_4w2<ZZ*Nj-O-ES`zIBEa^$UnP!
zg}wCf@bCxwjCNwHS!nzNU*X#EN^XYVt9r5Xwt-T_<A?Sm2Cat-=}aAz^~CB21qHi~
z>{d1T=lPA;ho54dsGEA%q5=%Dh{P%~LpDr}mx7eqa!6M8r;Dt*h6cFOaCmE-|3{h_
zTv&#k*3vS;Fddx*NTo0t8JGr|3+Yg7jI#f_#)C%->_78np5&}#CJ{VJGzR;~;^y5+
zDiwLw);`HkJ_l?jH`PKvtZlV6&pBNBU4CTO;&RxS#`vY_U6HYcQz?;?_SSDA?ARpP
zso8C7$2NU)y;D=JCtBP+@W-%~?l(32=(3_kQ#|~x(b|xnumxb|Gp_%%+NmIKHWQq&
ziEEgnWy=5}OC-H86p1v-CLNHF!2Isr$n3#OUDB2VP+6|6`I3N?6YdW?6tGuwS9a&T
z=_PDkP;2ap9Xz&7W?Go#?>}C)|9IH;ORVAscLEQQ21iEnCSgpj^!nWX_3MYFSLpow
z4;ge?X}vz`3<_L~))0?ZXKiJZ_t~D{pL_`mlmD=)n6Pk1skgfXTkF`|fMWkYL0N>4
zHXh!vHBrk)j~i_PpbOJ{ej3s=$s?MnH#C^|O9yLz`Fea0O1w_Uwa&f2Q+oCN{9StI
zp4>kVD7hbPjJ<w}@?GuF8>cOZZxfBH)Y;vXOyY^E*D|*Uq+m~f<;iaZYVF!DM#B7G
zwv&D^?2kPL>8rjlQpGpi+Ret`T4QSPXL&U-F_HYgK?*{q;9v-S$LBGBH=fNs+1sKG
z-uk)aRp#_v=(frK{idtLP10b3>xM{>Z0U1FmaokUtuzRA1Rfuz#n~lo!HCnDaoM|`
zq_78UrT4F{Ch%zOB$IU^B?gJBuq;E$`ORsH8_(0Sn-EG0XJgCgy%>UDf}!Qh7ggJ7
z=q%yxHF`~6>;gjL7QA_w7Li&FOGvy^&)aFZw`>cXdrQD7aPE5TfUf^QNDdQq5Jgm_
zg~8uj+mn?g4#^cU$S=4np1(G)CNn_jmxLTBm&=+dOD|y3+C<MCb=;t?u$J=hyop9U
zM%x*pIdTdty#3qh{kta+Xq%^7fnl(U3M__XNNtFt$hZW<2CB=vv}a>C&9XdY_}^7A
zuRSKOr55F;Ex+-Zl96kfs_#wJGTq%q+Qau4>24q;%<{4rT5Hw>IV(Vc5|;ET7w`am
zEYI9l-A%%$?d#i>f1CUKr5MY{z_~M3Li-E#sA0i>A3VVQy0i17K$O{+-rg<btcZkS
zfZ&acTFxK~2(8t%A#viJwE}IpG=#*r&Idk!;U1mVA239S8)im(y$p4Gmz?Iuppqs;
z_^7hbl|NrVL=&D{3ATGIQ;-Z_MV|b`wc>SRzE`0vC$DUQX!=~nt@X(I_)JQU;`s*?
z(+yQ|m&5DswM2)KDBjaJ(I$V|<+A6~a6XN%&7S0;6bFLk%RD8GuR}|yg~P00?!#SL
zY_*~Lqpy#5IzBk)*;aO3quA|!Qz1bpL{|pM9G{WoZk}B`t|V8_x1{N%)C@+X4@zqZ
zHfz&-o`vljrf4YN&F5qdoR@hDf*S=1%=DNW7J6$6^|^T4v&Pc(xeS%hu1*w(NLQK#
zWgYyoh14Ms5%=kWu6<D7XL}B{Ghrzz;ntGT)XrPVkKhYi$%jS<6D1<@y1?=9I%~sp
zV+7HXfo-Y%rijta>qMBu;9$}>C&cri6F{wET=oeLP|IPAq?u@~CE4FH=|@9grki>s
zLBB2x|E@~Lvb1lx)A5lJ672G!bX*@tKN!>ZlO`r;Vk{%m&)dhR<i0c9Ulg5ve-k1<
zMT5V;{=To{mGa-kPmB-09E+^DBkvdZ`pJkw+28xu@6PK;a@JhNO3AQc526(YK;H2d
zVGjBvy_tk`n`rG1qi%*QN5lCo(uA1l>=(_<(6kxBtqDV{WOT#GtOS?co6}2C{b?z)
zErWyn+6a}=XPH9_iGl<;<uUaJS%TIN29QsCijoX0l;Q)`ufwBXpKZf79BE=sarF0Z
z5Dp~ka|5^Rl;jA(+{_jd`W#3DX0AA4h5=@qELYQzt7LQr0PVnZShD?e#Rdg%&|b42
zgLR1K1!jnYAz*aWA*%jpi}Zb<j?EH3qa;(=CGHe6bzK<zcMGH0eN~r)DkXNDpZ+ZD
zDR;~~Sm4*T2{zJo`lVZ^t<;Wi3EautI^pwn#S^lQyv}34j~E!ECjf-W@a<3jVq9A)
zox|z9MOcDi5Dy9h!2wsKAuZ72GCW~cIV{%YN1*GME<r^mt%GE0E?Jjzv3%$%+`07L
ziy)}tKygm%B{#u+hxS_Rd3<zrDSmwk?E58%82^f|&xiw_Hb0)%?E7O|{HYND?~9`~
z7OuB!E2TeEb)F=jyWFfhex0L6pKJ1dVAG8PevUXQB}7=O_?J!kXFoIG^*U^HN~G`8
zCz8<+Ul^^$*~$<=6JNHuUW%r?Nsufmu3TQGJTrPq6u#3(x~YZxvZH|uB})bJ>==oW
ztJi!_Z-X#S8ScfA>q-`vje$<CEcDR%?q2*&Y5#tMlIw8e*xCk^Mh?bk3*zy>N?Ue)
z`2vSjrNl^cd>}pb8r{{m^+R$#i!M25R4CGh+F~gib0em!(luF#YuD>D#J~8KpT27L
zGDp+NPEY^U&7LR4C;0mDli+vzzLCCC!Vla>f7Jx+hG<cr`8<G|kB<+ol)jPKcP?_3
z5()aWMO6?{O*Kx)g1-P7qUBxrsQR|9uG3~r#uCiQhI^vw>HLHbGQ;@Tu_eQ#Vtp!5
znVM8GL0oe7#^QbaHy)s}l_7D9a^KZd<<-$<V(kjT6z<INuxKi~ET;A;UDu1ND_ia_
zpFiE}U955Yl(-xHTVJY@sKc8QCyF%=9Id*jk90bQm7{0wrfKG1Hfq!4e=mMHkkUN0
z<-06w1aPNnNixym-n+c~Q1nLgkn0XHvm1n8NqV|gR{LEw-f({C`HmV$2-i3){*qTl
z1&tRn$Cj2n@%hgFd})vXoT#raH~CN!OrsE`Y$5ip?6JM{g}hFvOxrF*)(`dk@NXV@
z8W*>Lgr@B9u&p0YcwJKW#QmB9@R!qP219}7FczU{VZtm=ED9$80%?#%mttivKcD%2
z;6R_M<NiYdrH{2k!gH?FwJyHicQ(tu?`7x%`#W1h{<5jCRbPA^-@)<zyNh511qm^>
z+m~%_oUxsBIgsg>#m~<|-`LPVA{mREfS}~vx&9GbSI<eBe%apER_bD1KxOoW2T|ZE
zS^Pu5{(>w=g$VF*b{6NlfQCD`^rs#n9Ry7);q~HV4}N|v)g6E7|2-+)O<pt#E5E&)
zy?c#33xMYx$CtFHhyCW^jE$<jf7gyE8`nSKe0cWaWJpJE)y{Kk%A`mVzKt8r4HJKF
zy{qFguWThWlULprs!_mGw@2(&ssqPb9gQ~u!kAXLL{{AK1ixh#mI_2a20xczkoTD4
zNeu?12|00a!6<NBoH_+=u|!14-zL44CFuamwvD0Kli_hvPpmfm!%&HA4AKqV)R!q!
zk={4|r6Z;B@IhnDYVPFW@wom2n(6I}i?R)0vf_+Il3PaM%d86dYEMX;m-~MEKXko!
zJeU37H~z6oR+40|Bq4<CJt841JA3a@b`B#XA%yIetgP&0lk8+?XOz9O`Mo}Uuj{_A
z`+nR%f1Kx|^U)a{pW}ERuh(;er>AVpLxgB@yGs>)8j(qqt}bZ%@KRERNB=f0)u>nQ
zpSly@B(`>gGSk`nrJI=k4>>6KsS&vXr{I={RbtT$+a(Bcgl@fdP|=0iGmkRzh`-y0
zc&r+4185)obmbkl^Z>MRLXPhuK)9Hog@Ur-g#~26L!`dEytTL|?->&$lawlHJIZ=7
zMxju|*s7lV?;1m?C5c}WHraR(=k3#=YH%&i_6YZgmLk|11O`G&Q5yWucptWkudUoB
zY6#T3mx=O|VQd<^9U{dbz_!qxIg8ah1}SU<yeB3jP%&f%f?mnJQ3%1|5{~aM<BCZ}
z%yl^C%3Xfn3Tso-$<w@~k7-0699HGZd3E@3<gIw$Cp#cE{k`@*M7BJrQ6PS7+0eX6
zjEjyQvaJ^iHilOyPKAZOe3?s2{yjZ|LeC4W!EY?b*tS}=)_UC)d&zsB2)WpAKkO>V
zVeHo(I_5~&-KPvc-&l_rnr%vuzKArG)y_F*`BLS($C=O0D=|iC{CUvye0W%~GNj3;
zM&IBE6Uk4@Z9!<&Ld(j5p(ESqE@W=(^A-g45>~N!+1Sj`3e_bzQSGh{Q~AT&8Xa9d
zHy0bY3IFt~qK5^N*j+Gzxa1VYB|Fd?Qg`7>m}tdFT+{G(EGz}2t?$T-RpYCR`DVWr
zX>ikQ8tB6ceX}fRnX9C`_z$kpgx}Bj_WRK=xW@rM@%v@ai9!`>7HAegEWQABu-@{@
zi3S^N`f(p%fAl>eTqd+KetbGMQ@}EcYr0Aj`u<@5A#)4`L2%->-91XBt((=){5!9R
z4{k5>NtLa7AR>1uZ^~uO3v#FCjx2vvXAUy{d-2|j!^+`utw;34e(qZ4@iJNcxD5}0
z!h2x&fC$7@xY#hQ@Fgv<Sq2+}7cJdx5h243@ax@VdO-Ff4C3Q`RKAZDL7~{_z<}%N
z(X&LTw9kQnJMC$C01{1-N(?X%s8fa5fOT(V`gkfwY%dSQEYZi+lj|T^phY<Wg8j^Z
z2=)!P3v8TNLQjtJrni=aNnSZV`<Q(1#l!V<Qe3eCle4loi<)#bvTF?lshl&14@v|g
z_<4Cj&(A^i#SBAamHuXwe5Y*hV^%1`1RP_t??^~|=#>U*P#CGdgkL5LdAx-V64#Yh
zoOfBjBmRP6wOStg?I&=0DPg8M-s@AX>d0UA{%<kCn*PF~-pgmv=zEKabj`9u`-?$;
z{R&}!k*`XYb-lEVFdavxfnb^}RH3nZX*a6bJ!~@Or376$d2o%>tQ2U36tV+)xZ~pE
zV>;feu)MbxO6qAYLbf4)w4!2YWsjS{749g|r-NcRN*+XLBP&5^Vkyd6UpO=+TwTk+
z5g*b5b0-wS_+BK7e8JSA&jHRog^2fy`{+x>7qaI|Ckowea7Fi_{<-C{_(_kWZ<&CY
z6Li&xpR|a>Rs~V#5bClk#EEW^6!tq{MPA0Ox9(oBe=s(}>_rxvH%%@zQf?-Z92REv
zM0p}EUhMAs=iT2`l<0035zt=v%9B9QeStIAB*=IkhFOK|c^5&_O@W9V{25~hH@>8x
zwVuwVPwWKsLnDWIMBNrVtEQk&pJjAmBJp3?uc$<RTYhtuz%bm-!yE4^J^Q+yi!2O%
zfNMIYLHXBO{2H`+Fk5IfGLg|S4Bjell@<!Gu(=*=kE@l&2r3>gHB#tJo{@?kUn^pt
z0rMbSTyQy{57tLIg_{3z7A~NwT^3#$6a$3@*=S&pg|fyH`J6_IRQi`-$hFaII&g4<
zeDoHskHb+PJujj2$44oL*=C^j2;CUM>AMvMGAr=jL+GZc(~QK!hwy6F0o74rVt^sI
zS{u%f8|cW%wfFttIrtTEOF(DHf##B@oD@;q^xKI+e%!Z1p@v0NBHgb-X5CeoW@^Mi
z(`;+YrbxYTy%7tQ&Qx?^G49aqc}kj1n;EwfqNH?}V9u>J1B$P*hEF_E5*@34os2aR
zln04_In#&Xp0hF1b`p*9yZ^K}hUvjb4f_w)*a^<~$y@Q0#cwY2Qu6bS&AVaj%wnLJ
zW2}2>t4d3+6p<gnLvZ5*6P%79gZK0ljFNBMeO6}xNr*%|^Kg6&EWvcHucqjZxI<}-
zfxf;X6Uow3u*xMUf;KTRDJg^V`8_8Gw@{e?*D^NpP<f^wOHbiyF5C}*8)w{~@Eh(?
z^YwZozNCiTY)(i}MrYG}@hmd}!Nl=(Wa~P_XWkk??x3$OhRpAgXCFV=ITaAUGfwS;
z_Zd0q=J&>ghDZ=QBMR9NSxAh_%gfv7HG)|M6}`0K_wP~mK!Zy@olRj%+Oh-_38><V
zM_>tYtx1OR0L*&GW@n%Bduk4cQTYACuUbCo;(v!#40rCx2*&-O7|l5y>9=~9&+}u7
ziA-D~_mNfT@ug?QQp(Xxt7X-uTnUPhlx7txjE@~qh1BmA;!t(bIPg+--I;~jJNP2A
zgU)Iwo7HwDfOgSlzlW&_;*eI`ToJ`qB@PGugd5+7_rF^}Dl`XB;u^SyhqMeU#_S3T
z3V4<D*yF(ajfUsS(oXHnkt=Wss=bXU-rQIb7opPerX%vD5c%lBSrQZi0`L%|<qE+C
z!RwzfF)k;yJmwPo{V~<(T~U4Sqdf@I0Pjk7CbU)(eHuPFbdkVFlV{S7yAvTHReM&5
zQSn9h7CR~?bl!^BO-P|$Ir>=`ph_n*g>jBFpM(WO&m5ThKJnz0^CT(FC)_X1eZp$3
z<L-%%eo;|SEp}i$oG^|RRze19^oh)=89)<#_bFw`v&ygws4sbI6Nygd5HCdYp();O
zjsOQl>mW1t8wrTOVt%{mA{g1Oo}8#Wuxndf7uU4_o-YFJ6mbL&E9jYkI)m%0`sTlL
zmLcTvg_}CW>~l~Xk!wCTu?G`fc-I0XH;JVc&m=p|1?y)clqy{XLPi+X35qEZ(OIh7
zdsgh)jvf>6!Q}>JKAC)+wS{4#cQ;J|>6ty!;iC4~=_z@X=h2*dE*iAjTSg%RFIT!G
zlYFRjiF$Gi{&H_+=%IkPa=<&Z#WwWN*?jXm9c@cVbV@P|dcnr*UCom)`by$B0z-)&
z-jJ-TSFb{8bOT=Fr`6V0Y{f6Xy`l!;^s49jBtZ|nFL)qe`u6IDHz;98+C@}(d3kqV
zpE~J1Vc~*)tLkLLs-z*-*>P|?-^<ABM2GuKmv^~yc+9Aeg^cFL%i~K!SEQb+Wb4O!
zW0+9ERQEexvwmq;C0mU7YUsQS;~^3$HGA&h==nvX3SIFqDn_3W4o{dK9tSp~V0!d3
zS`OynsgTPfgtVx1&4P!ldZQGyUTuNgsz=av47{fr>9eIDH_OHVz6WU{cfwNv>hBX%
zHWe;C$u?^z*kuF8q))QdAWJu~a}LJYfh98?vimc(6~zV67`?i&fi!sS*)T>4c>lF~
z_z+)`xWnwnf(rm`)wvj*<TLi{6@1e_SGVl=E!3<>x9xcqj$EJJ{~2!5|E+NMfL8ET
z`3Ci`!;{(^3tt+YERPPk0Xl8%%Cgn-Nwtc8beEf*V8!XzY0!*rHi49qfa;Z#Uj96E
zY290kM8AM8S0LA3d{`EVD;cg77oRs!OF&~Iw`Lyx4f|om3pzZ0hY9%{bc@o0%T^l5
z4kv5fIrlB$RH$Y-9V*bmK!JAi`;E^Eke3GsgAVGuVDXXZNUI%R3wxj<vuW9B<uRT9
zJ3mYVdSbhoYYzi-4X%~F?M=RE`rBm7@C9Q;K0<Yq)QC8gsNk_2?$D*e3$j1mnF#R3
z1b7-gcp;{is53ifSvFGN?!oFYEGRmBd6cWvX!ynBir9>gX3ZV+N0v)s|0o5%0nYN1
z3YzBFroam9ljdejmdB{RjFFW{TG7wd)u6V_&`O2^w>*tnx3ym(0}S_``2wObIf+cd
zsWE1!$xb$sh>A3R2Zynajz6baR${j+7%pNJ?um=j5rym=eo@P_b98jHww~QR2{tTc
zLVqw(PMN1>0w#Tfw|_pyY%;D`ei0{-F+{Dx>iqtvJ1?^F^z^*RNh#zc%HCQ=`YB_K
z6wLQ`nrv7TydnB&CUC=dzZP-&+Gw8_4KYcjA5O2h173YWhzEMFD7cTurjo@G+RqV?
z)3r8McF_T_Q`ko7F9_eehcl$QcPFUwHJtiVD;1C%0q8oOcdDRkFw=t!m4k0LBJGC$
zD70`J{Ys*z`FPkEceT;w<B7EnA1NLNN`|o?m`R8wL0deK$=VT-J4yq<s$!!cvj<{6
zheSTsUOT9H^RI|w`pwlJyf6b_3K&hi<}V!7_VmKy*R+V}XCvh+GR;}oTj{~%{kg}n
zuX?_(>Qg~NNEOdf;|8jkbkcK4;k{c|_ddn0`08XCg|w921*`4Q>Vn)mn1<{PQS@^3
zZm6&@Bn;Z@D)FT4Fvy91PT^TJ7$(K0Xk1j>xBvN$ID6Bp4oESSABbbFJ;E~<Mf@Mq
znMgZXy8bpPRll;WHTzAWF3K%ig6lQonmhmWKF%^=0Dg2h0qY-!*d;4>VcH#c9yy$V
z8^?q^IN{y9>J72?L{&sR4&Pm<1(YAxHXst1e|?B&+~xJ-%yr_--6e?C3N@v}!~nb)
z8|y$i*W6l!0tk1zYrAwh%xdJlCzrNfs5rg#$XXcpW%{|{<x7S)<W7&@*e5+J&gJ+%
z;{y7j!>zb<fY#rK%KU;1>p4s!$f?6b&7j{{UoR8FZ(x{p8l81=vTr0t^!yfBP~=X@
zy23=9nt`GX_-4U9l%XmtK7TSoW&-!p?obmM*0|hE>q9@q?p{xIIpQJV6nJH*Sr0ab
zd`WF~h$(~IaI(e%4c4+TpV|=5i>7^M876e$(a+?54oNj(y*QF2bXJsx8xxycZS5t0
zHRU5r6bfHrjlT_v^)P8^4}`;DC8x6Hr<~Y!ls3EDLT4;k2Ac1nhpo$R{N(7@&A<@_
zd94;@Fr0;le{*-oZ-O6OMYnO8Q=(JLjjjb{^_=`AUroLLw62dRq{wu1JP0;Mp`zsH
zm9KB2{GdEuzuL*4sRKZ7094M@LC4>yOs9$vPU)``PzYaR=!-60GRv=zVZjdY9%6+V
zf<^MBYA%hcOrmin7gAqT+oJ9$4o{H(sD{ne&F`AO<;0Kr!7TIrEsp;v1#VC?@QnJk
zZg=9llpZ~TXJ&H!%H)IZyy3>9UY}oHp-Deir{~5G;CbpI`zFlPjuy97;wOr?7V-C`
zdqctR^kpT;d<>{GwY4&KVSfILg{v6Q2Qh3icB$?ygqXlI1X%CD(hIUkn1r*{Acw1>
zvU1&z+rO|Wr6;9+DCaptO5O&Og8vL|hS&#H0Bj^<)m|<#ynHopyJgK0Y%o2G-tHd^
zGS7hg7*cT#tB~U=2wOWBX`oR>5G2;}U2^irrB`sv;ses5oC8TF;F=3;Qi%z)6-zUG
z@?_6Lq{ED)vgC(N1!$QZamB^N*7v-esnJHigMeEJim77HDaCx@zB%{=9e3!=FO(mI
z7QxuPSKnL#`KeIn3Q7o)P#-bQSOv&w1||go|7erRLPyLk>qIC4(O@BsiHV7le_^fz
zsXrR5faGNadBeBY*N5#TPXkZPyh0bBH5(>>_><o%EPVU+a(wR7q5Y)?n5!7DI%jL#
z<Ac;T-B%+ItP>$U0)p{7%%U?{fCN`zfkCBV%=>PWkrRx>l;O?NxB&=oUR6>VGj2xz
z1vzu<{{KMEx`hq)_xmPMOb?nHg;9-0To4H5AnCqah|y943@3bzi@p&iFpTg@VS;<f
z)$GJZL^X{(8h;Z@VF88!sfQvGxi_uauvPCV_)FX*ZrwRNB9cRh+~A6Pt0noJFKab;
zK~9Q1(=U*S8E{rTx5+Px_=-M4pjZhah&s8tL6{53YQ6|7Sr{lqvj?ml9*RK43+xXm
z(~0N|f;xj>W8e`CZJL?gCQR#F7gD(?`T0>jpH}x)Mj%K8xuLNQWmMo`@$jhDmcK=U
zgo3znpsG83=TPJU^hCwLsF0O42O%OQX@owkpb%)8b)lKG<D=RY3;{+}MSv791cG>=
znCgJ0|3=g|2m&)sW5xBUszdvj^=g6!16myqet=SHwk3qEJczT=q3V&MBB>*1JqA65
zTp1ht*J2R6U#_RV=f(Cz3`t`BayK7BH#XD@GOkQAR4BY)Tzl!Yr2dVw_17NFc<~#@
zRU<TNAPq0xjq$1t@RLD{ZDx-}AkPSL7>{6}d{#~A#^3fc5Wg&x;)sVQ5A<690N=;q
z<uEs(QC6={{};&m{TIH1K;SIlF;W4k!OD$*SJnGnp8Q0#7%|BNSVX!Pj2xh&=i*3F
zWCkJ#u4#}IUs9pA2M9q+*=?beh%5{y4zF0_caRhw*$$M=Kac@PMX*3yTZfvDL7)lb
z_Xo`HJQX4w5|$(eU8ze=GSh>q@stj;%>F;IY+CDe$i6|fB{)#&=yF)V!vrB75HaH~
z0Ty*lO-<lXmX~e7Z&FJ0xobgzIfe_zj~@3`CL~Wm`&$<o1ezCrAZMiv2bXlHj7fzq
zc-fV60>=;eQP^NQHn=Q(dP0RNE!+B?JN*Xs1U|}}D@E4vvj=rAY-FTGav6y75RKdq
zG^Gq*)Qb!B8XL+at4agW%miVEMT>Dv$L`W2B=Ptv-T`v<8=3kh3D<aB5)v@shhZWq
zF(ngk`^*Gc9hp!}L_{m{?SUM?EET$(P%jzTew~cW$lLSX5?sK*CVE2xX$*+&a;>2@
z><-k!G{{zaoYn_&5Cn3|f@px1+4?#(^cdwQl_B6Vk6V7|2Zn+)LDvr+-NbR`hgY5z
zYQj(gFcJp%^elkQN{$7vwZ(bDqU<h#bhMeIl%&{yhV#|&s>?RkVQoLaG$kfR8!7`+
z7Z>@9u-o9PPU^u#fl{0H>;c4O|M_DjmbQ<0)wN-YB`8985WM7_|88FhoQ1d<&}qeV
zNZ1?B(gcp$@j=uAFK;V?G!|0%9jmjG0jUJPA6j3XumD1kp-KKX_#AjZyi3o%Cm0(c
zQ$}nGOqZY|_9~)%pNn65Ajs{@N>V5+YVgunJUj#gmHGFtsrSADQYGhgf?lH(vOj7<
z&-=z2Ext+Ji9yyDQYi4FD&IT*_Y2Lkhnh<laRB3SxW2MRzI(RcfJlEu|58{W0{u*!
z9PX6d5xIVR=K*VMz>g|$n@jL_ntcL4B8a~t%*Wu$3S5HRxDGRzITv?4wJB42EMhvC
zP>nW7MQ=|}+nn>O%yb!mq{zbjDO)zU=Jl=;w3M(zLGSp_pLm0S+k+?o4*$PL94Kb?
z%_>N>WC2QAR<gw@tPQhWGuej7q3+3ve6<XtpvNZKsbMUOMY^eyZ%nSanflG_Q$$3B
z8Mlr@b{2#vrVHxBs-Y+Zud$(A#5T?x(jM@p(ll7x@+47y3fbUBfMUM%ZbH9NjR^<$
z4opI^ljf=;T-DtbCPB)ZkL14PndOW1b2M$W*enE60zpSm@Hof769@JEc8vYiy@*j`
zZKhx14t(e0Zk9@NvnZcuTH`;DVa07-GpfTpgG|tgH_mXAufZ*ZK!y%&<SkM>z%cNN
zp1>R<6t{KSEl{pos9PA(W^$qi`Fz*3FI>KM+Xo)Nd;I(@uXmjU2#!aCT8!h&n-;G(
z&I$Men(5LP_j*(ses;X*KEczEBP;l+zqT<))l$D|7E}mC($wF9v|f@>E`q>00g+2x
zpWw*e;JJuGX5Q*;9nOkp9?UNoXgzrK2$N^CrVEx$DDVquP*m~3qUXB?>$Xm#9AlK?
z2ilE&F4;5&)9(_#+)*C|BQBv(60c1|G+0@FZrs%1em`_Y@0(#|RdPghyv>96=POU_
zFFs~@h6^K~I$&H^a9fE$O97HEHPU;cqV~L$hX}KckI|++V^m|h?OjMFB@1hk5JIJA
zWMH&A2OCRe5F0kafdmjue3yD2>8%;#vo8|#Go6J6v!kV^J}bv22e;c7zL|07X|P7g
zV+TsYkLpl`_hcY!!FfRl!VlD4YT@rczBZ$oJOO#7c9lM2;(~(KRpX$emzKtoM2KU@
z-^CJ?jQqqdpD|Qa9fVU+z<df~mZql-eR>1n_<GP7`B;SZ5j2Bs-!?j+S^u&qc=hn4
zP5VUSg~+u~gRomb)=Cjjsjw<r-P^yKZ;G8kw*Dp#`=v0^?TGX8;^Pb|en*!Jgm&4z
zn`ylK&+D&MB%?8`_@CG(u6>9o#0Sq{(fP$_I6RE?+0CLWv<6JD^q&4Brdu%&9>rzm
z>KF9qKC5x5nShosNQdt@H8%0&OM1SP9nYE!Lwn+fASnioxhA7Gt~qmHd4X#O=BH~S
z*&{1eA3FHT1MovyO0G@`^;If+>VKip%!h=`|4kGlO!}dX&iVyIP|ODu$eUk8d3Jaz
zH&1Q?jr{Cz+q~D54rNp|J}V{tpqk3)v<2{uY&AIMp9-yQdEj*l9W)NHj1ra16EUTT
zOGoEqq0yilQ6G!_D~VfzH9D^l-pyoKzoqdi<%b-xbB-xV7$^0_ew5C5Bqjr@GvA#8
zi4dJMIaFs;NR@+YlK7MPaq{>HD97<~o>m<7K|{HO9lwuYr<b$O&6owp6Dt1_@khec
z5zJxPrEhPD@@f=Z@O9iY;vx@~***j@1Q%b|B@d|7hz7ksd9ZAjYRTc*&Kqc97#q`h
zM;KvDE(ywHIFbLL3tI43P8<>Z2`SY61+e0QW4~c~W9JYS80Hv%3c~NKh>=nH0ZqZP
zdl^g<+zJBJv~+YKu{h+xu^(|+J|w>3G!X>`@(w3w?c}-(CbW=O|ANqk131n53}3cA
zlKoa@w(7#JJaqFX4tb?vUbSJNj|79bF|So+lL;YOuYvo08v_XlQJ_EQn_BN@Io!>6
z<yH9MSU>)ZaTekKlQx(;uCgKkH^W-Zsdzb{y>3^sc2KBC997Z1WHL-=4;ariSvjy*
z``W)=T(2#zwt(G8E=x81zU%zAC{|8OUP?;~3!_5ew+M_6%ylhOF6>?R8Luc)IJjYe
zisK(!iu@0{wEg7N_uS$#3iU=ITj6Vm;Esr|Fc5lRDTA*Wq70R9D|@hb+3|TpA~G!9
z_EQ2hMD3uf(3k^4CR+^$(?8Jx)yfRWAy|Rg%=%k4YZbD8rJkzN9|RiB+lX|;-+k_R
z#FAobvE8F+a6y8&r4AuK%a`|2W#P~cJS>)xK{fWoRh}&_rrQ0isV3hMf((X-a#@^D
zzTlz+oM!ymOm5S59Pa#XnJrOdE|?K*f^i3qXjn;@NDh4r!1A`R4bu~h<^i#JkTmy$
zS5!d2ii;cwAQ(QC1_UBtAcLwAcp-ht37+0C)IoF@9>91G(+d|lyZ{%tsasQs6fcK<
z-xcIu>|XqJ_(hfJ=gpX@0?kWzq2Og=9RoT6T7SYC14D&R<FyalS8v0yM$;OJnGtK2
zPj#kB`r+AZz&Yu|v~Sr-+q>uWB7dvnSVnPJ{iBK2h&e>Mlqw#x#%i$km~jKHw~Uyr
zC%rJdaJdT7@my*as9p{%fxuosOhiOqf7rnMg=fkeCY?`%bRQV}^lEm$HyN;yy5w!U
z)KD)|{8K+jqDj?{L+M2di_j7B`}eFV3=0AR)t^68;$qu;#vo35p;-`=_R&RW*8zIt
zxo*XX3peayqfmY^vwzI{9T7(l3>1?5Kuo@&qoOpShd=PdC^Zo@Bkby><dt$9!cqWV
zUz%Nb$li5k_~EM}Hz+QeXnug0qgD!{Ax$vSlyhb;etBQ{8|w*)aWOVg?)<4QgnG1q
ztZVCe8&@}cT7qf#!NxNTVCvbNZSgW-zS8vx%1Oe)!cI!ZkTtiQ0VrlZ>ok>KMU_j<
zX63d+&v3PAo>F5A3q{&JVc8oXRQZ;zeX?({*nr7;X%Mb$q?!(X*GZL~7o+_3x6N0v
zF;IRyd%yn9@hUa^{Rw?0Md|UMV9OGqhD&TwwH;(DNpV8<Vj2VWJzkF^KAtywj%oM9
zr}FZ_DP+nZ{XTaB0Xw@^!BC0c35+sJPk%yUD&$(s5F~SFSbOo@$IVLLgbor~a&p)N
z_9js;m72}C;lba*@nP*t4r{xgHLpQ`AP#~4jnhm&7*VM#&klrLDRQTt3!Je<wT55P
zZcHR!#7-&WIy@6Oy&Ar4Wk~1T^>W=D1^uJ_R)ZF0&}vgK#2ViPJwd_7FsQ@3ni$uK
z@P^=o1HVtn2R}ny-GJpM9PyAM(VPH3FPJ4mWiW?dlh&^H_4a;HyB#EDT%im6>OWBh
zxZ=#mK)?>Nih$#{lDFhhe)3G*Nj<uSd`Ug<1Um`%04-WR|3ZQufiZ+YAp;N&As2#U
z|AfPC-PjIN)O1{Dn)0ny6v20mrZQOdK)O#EuB@o|QYJIt&oJQBAKyY#<18Hx4u}Nq
zylf2AGv7?Mn4-ieR$(zB7fkBez9-Vc1z3p_R}xLgqnrWQ?0jzQL%Q+a^5b_O>n;0e
ztE1>&J{3Y}@Pg!{U;Ijd66MB=pc`W*OcUo<kZqw)u^6;Tx5h^Ksd0j$3_?f%YTa~x
z&W|#Q@xv4&3rlhF_KnF_mg|6!vGx*_-~R6(DAe`f;$H#EAIjzC#+TSpje+=Q8Xpu)
zN5!&XN`rl4UA^-br@lOM3}GIVs@V<4vN5DwckTOk;%SL~m=4w;DwGZv+o<s3Mbb^1
z3a=jCXMaR9R9K+yy;<DT+xr+KBA}JyB4;Ptgs7jK0UDvFFL{)JRTwEYbQek)v*V+U
zY;kxAQUPBxFpR>IS}h<0u7d7v0&#XU;tV1=>`NZlzl!E+9X8ZH=SLvluVVIFM;%D!
zKNa5npO89v8+<1~!)KjZ=bTXx(tGC*9q1E{LII6vcTVP%F%Gz_Q78~GGAL2xP`$nj
zYv!NjWjKdPWg+1RRM=+Rz)4sq_F9y|hHk`V7ajW<3S-9CKxz-mC`Wt@(~?`QG1B}6
zHn!(`K<XEwV~vd80jLjaqYW=*s0{oBUko@u4nfE+`fAj#*fFM_?-lz=V4{o`f4_qW
z_{JH5>V_*QiSzC3lNpAlCI&y_Qtj7zUmBf#;N?m0q}vWtw39ZfsP$^juMRi;bffKa
zAJ?OhOnx2KiAU-vKg6~9$0;x-QjK`1m%lu!CC^VdRyM472{Wgsmc8m=Gm85dYot?&
zp4(5Mpcf)DNoAqtxZdv(tfMv0dp3nzxY)a(7aP6n#^47v<IOd3fjzf)#ubJ*JT`Nv
zjQK)^J{iO7;Ft+uZ-#O(Bd%(A<^W5RK<FKARP@NDc`hbAGuYrsJM0$f^OK+-?r88(
z*pFiEa5Xukm(5eSId03FI6FHFzV`P6pyhk4QUZVaXZMY%0BOi$yPxHMUlN+|Hpts!
zo3KzYg;#!+{AtAtHL4H`4Tuvwj9Y1+bPK)eeC_NAP)-Y-*MKe-5&42pm_q&77a&34
zIr=sV@{JQ(Uts;p%3J}dK(9_inBG1C^iV!GqIhZ^cYO)*=IG-xDtW!5&nkPqdEDJP
z^gXodNz&&_RS@5bCcilI2AyGQe?icq|EjHSEE&_UV5>FF{$ub63iY300)rY1)B%?N
zrYbWeKgG`Bd+@ZB^LU~f1&xj4B?<ftf=(pSb6bSX;E|d=E(VG#CX)#sO%MMI7X8B1
zL}`8PpV-|WHIHMpl-$D+iI}^_zRvnuE~}q3GQ>4#Ml-9LBNl%90TXZytOu;<+aB9Q
zPZc1hRZH>O-W3Lz>Lv{0coI6ybdb4H6L|t}1IFXDF@7RvOIM2yM`=&5y0bt1S@4E{
zeaOZU1~#=ka0Nf+*tOvzPxs>_dr?9_Yxnp}1ajc|tav#%IYIvUP9A!gfRs0OgNFrb
zX%BsTcc}LuiUMLlnwtZlzsZanK5)P;`T5}(C0iYvoU{l31IYZYhRO(~Sdiamj)9mr
zP_2<Ke@tHM>+VtE6v5y2q}Q9P_}Ch=!_Qpp3ZQ4)!<V^p7bFJd;*2@kujbsCw|e`9
zILwlKal^HSV{<&a!c}T3OT+DY4F6rv%V%GYMxIx(j>lt5HM$m@ysufBc6X=feEAp1
z425inW<&rRAo_dr_I?J?!bBl~nYj=mQo;A)a0f)UH@!JDK6Gxq<z2X~Jrp2#59JrV
zO!Id82oMtV74=v9b0rDC>k~L1HumB2+wpaMVt;xagigRzQANj`{I}2*oP#rD5Ehjx
zatZZkW5dzbc5WlN#{#B9DI|lqd}Mij-QXNGEFF~LK*I02eCqlspOU$bHTsKJSILo9
z9;qP+2cS9|ED&454}SBo8_5JT-#!DwaafoZ0~80GymO6($R$ndn7@r!m)}nOBeN+;
z(@vt~Xm7ipf$g<?-pyY^W&QR`8OFTr#~}U!Jp5w~6$>SDi)vH#d`t9A9~PLh2c<22
z+YmBK^6H?rl7wE`oTc#QXMs&leWLpsXUTB@+^vLIj8TQo=gIG~Qy?i!#?$RC6QsDA
z<VD5_$JWbozxXd5B&i*?_bBZbJl@CKta%7x>HXZdILo3@@(_v(W)jei^l5!2maB=o
z2u@feE)76SmKhKQJXO=?IjK+GdRv~!PM@IId~rdOj)~;bNC%1Ybr*PTX-oHm3s2|G
zTeJTlEN{;jrybrfc`kNSpd>6UZykNTPrq(u6Z0s^%=z_b9-J-^-yLAm1~gh&8wg?`
zBL*&6)%^-7?YO$9OKR+7fruKSk#To?((jhmL4FWwckwnI$EJx#hx9>pnfVUTi+TEg
zg6F0!MIpJgI6Ds)CBdk{f93eI(gwBFVY-6upT`lS+<($e$<2Pm?q|qkQcYrA>7i<J
z3u~__9%39#?nGn+S>i<L!qSx9!;o^Q%^qvmViz5VGwPY~$6n(vZVGZ7rTtls!@EFg
zs)8D@%)=Sq@cURO`UfsWgQf5Q!l0&7VM2bN7-F4NSV$owe`Sws(AH`5p%Gps5HXO9
z1iqK)>B|m5)6UeK4%#ZwM_<*$y?tu#?mS{73p0L8>&#zPU2RpWZPcQHW-l}c`FO??
z^EFy~!G+g@^z<}QyaV??WjIKVJ}iN-90X?@`><668@J^(C&{po1}p>q3%SjXl1W?k
zFW>ODPqxiRs*joS%9bcz@QY;QjjFI{%;Xm4vOAk-<|*zwwGZ}EXK;^;C=Bi`3hqr!
zJ?wm*f_QyAekkR$#zJ9YY8*#KK5n=0Tuto|*{jETkbsHuD_u=ps3D6Nr6a^W+Q%Av
zY#Y;2wz?;?%}yaq6@Ch$^T&{$qpijw$vfUOZ;w@%qT2%>C@%_T$NYU4xH<_56lnAa
zo+UuKd}8tMkl2F224+#fu0v%YG>LQ*2B{&lC-z@G1gd$Dqxn2gem;m;<e)Gns*$U^
zwY3#(n614S00(A}ovPJM+DPP&G##X49k(MV?S(arzy6X$wr6;O2^1hUoWl3oanA&(
zVaOttg^@9DXt{HzOBXgG$SWcCX`Jr8`RMQN?%RiF6DFO6&vv*%Wp(33?^{Ny*SHxx
zx_znfz2CyM4kYdaJwh9uZ7NRfZ8CMN5QOD=IR9tDkQ3(-GPC)JG`f(@Sq_DIN2SHK
zA(4M?#Otz(6uj@{)sHQ1Q8F^1P*emZ92bYb$Ud}lzsB2T5PA#lGr$1iUf;~`BH_Fr
zFh_UqUUQM7MJc-R5W)}eL;N)7_Pj7qICT@R{F#w8fe2`*y>6K?60Zp7Mh&|c7#Vbn
zxDj8TuPy#_)&Dl7@1a@b9J2WS5C}vRg0<HSs*CaPJ3k11_of^dpC577=tp3mOrp9z
zeP^R;-$K6xy>XgyYh@V0{bovTVQg#+24&NBl~3f+Nr#WsS!>Mg#z5z}b?AdA)Xp86
zy-j=Ag1P!x;=r}ERh=Oy6$gdI2#5~ZPniXZ<>Tv9H6$lW_e@JddY^Ah=$*Bfx|&E{
zs=eLH+|;e4HD#RLhBs8p9y5jP<a5W))3Hfl`UZY0TOE>d;Z=yZ-@()WPpbEjnWmpu
zQ;X<*^eNt9&fljlkAOrgM2d<&Pp-o|$EvE{V%Q5UYyP&d9X&0J({m*syWJV|wXh4&
zUS1`<O%j^Vn=|oxYt!1w3KB@0jEF5Rl8S^KHQqp>U?E&I0epy;kMC8Rc`X-uNgXc3
zzy!(?nWLJ9wxuV!x~x?#6BE><7WtP-Jr7QO-vx{g4<nkrOV?4~M@Mr9mYQcvrZ;k|
zy}G2nCt;wCB}EhY7%f~?RVOc)9^^m}V~x>8W(O8!eW&{@VmC<-w<>}GNN!XO*VWr~
zJX#pEEYR2>=JcGi91%OKZ$gUC-p~Mw6y!THM7mxCD_~Z3U^w;Hbk5u%+i&0M$1Yvj
zV@^V_+zP*LzwDgXapQ{>ll!VH`%OK{>b(IQMU1x33zf9*B$t<XHKFy->FdVrbu9<(
z1aOl)h<+gjb`^~Sl3-YyCFoIIi3f%z1TAm@B%1W)CdScCkn0+(<^Wu!*$$^UE_(>L
zZA^iqM<=chQkl^Mf2}Wo<R9(;{$XU>(NktP>1kJYYh{sv6mNfZ%oS?p;Y<=0MURZ*
zqM7zeLAQzF>>?OGzR}9nEj+aqVR-WE`q1jh!CYkVj>}77Oms(6qjPdu5NAl6iGN-!
zG|iWmMx6tp2&ufGFeA#(?|1@A8Q%r;AmAy(4sJbM#b1zZycwk%a75_K1~gg%+MV7+
z!W7fN&cOjpIuJK#ET$yoox)?Qp)<%rD`WTm^PK^1iajrTK0S7_@mlxdAsZZ2GE@zH
zSZZpL3~b^O6<u?wu}b{v42&&Itq{<5=2PmCOSQ+`s5K9(36Y$!TCR!*$xM!HGzH>c
z@<C_wy!;)0CH60#__+EfU4n2)nz!ge3e#-ts~d)~+)2=hB#V5GxOS|UhSh|b)5hmr
z<Hba-Ouvv^b0R03kY@$~?-H(O@N3zXYiosSm%790`ia+m<TJO!5CaKK&?&*miG_m1
z^2F(DAP9scU*mURrUowp0+-}f=C7@T?QO8pyhzv_vm2h<f?mz#3XnIE;$k2A(0YA?
zjt(tk#^QAdc#S{+1<emKjLv^ob;8;*hiq=Yf_7sV6kE9RG!m6_z$bd1J8sqRnGk)?
zKBa^iH9lm00%(1*oB;}4Y{-5G-dE`qXXo%_c^SzECQFob=11g^&BCKwU)$S(`iz*1
z6xqf_?AiV&r=0)pe{jl#eR>})y5DrF)1y#I1E5WV?6(9(HBty(2UR{SYx}gCu26Xj
zZ3(idMqud`$mU*0Mglq)D&qp^T)3wkpc>s?ILHY^*7D>|!Qcu`L`?riv`h(g@J&t(
zUY<Zn)6sz;<g;3y4+LvY@1*YAHNLqro@B>|f6vp@bT0AK7jBtXIg;#MXc<3<l=!8t
ztn43CQBnqngz$@4E$tkFEf521@C}STip*<n$B-)noy8An;n}#w%&dTj{Rg8l=BGTk
z<#Xx3Fq*phZI9s{=0*)qxP<%1ht8}7EwI#pE;C8-Lqol@tB__A7gay;#^7~6EODOm
z$H9ei$*+4wPcx2Lhq`a@y)q}$7_TW{8gIE{a@<m4MDpq{%b^#qru+Lzb`r7N{*wr0
zo=gU}4jTs~A;MT&T^$TRA!(;J`qPj}4SRNHC*Etpo&VRuq`x2!LqxyT3;S_%?gVA6
zgSFDOxmhTqi0`p@E=DJ-3hMU~3W#89cV-!%NgHqVeMa@mNdUkxS;l^lB!3B%`|;mw
zHYDEmBUy3<iVUG`CJ^AjMS==#hdfPa^4ifGpdh$VJkrGl!qi!|aW2lK+tfrm;>{E4
zk2SM29>tn+q|cMGkc8^j`y@;+TukkRu}DVA<}+^<?-JS_<0F(3Tq9_!TtxgbrO68i
zkbcpNDB$|{M@Qf37J|m{x9G-U$0J{48%~ghg*%QIs1o*LC~H0&5I-F_I5_YI8TDQy
zd1=pVvj2heVxl%6K+TW;g8=O?#lXFtAkQR~@utUuZAAVHHMG}*JQn`)po0Zp(O#kB
z!a2#YkC*_QS08I?bP9${$jL%wl9j<{1~VNv)($S-0OA(;U(Bs%)9zC@ID{3N9$SXr
zkLg+f?j6F#b^Zb!33BAzFBep&lTbA7K!r%Q%T?%3PE6R9YFAgB(5+6o{#hnOR3Fy3
z@P@XX5373!2KV%N8s={k;IQK;o9E_glt(QFy)@~KE^Yr{!7NKc^EjuJ?|%BZ>P;l>
zM}m6PpFG?{U0%w}Wzsd!_NfL`r>RU9Xsl-54xpEDbogO#l{w~;MIl5^u$sFX8m9GI
z=^Gd*Gm&sP(=5_m0ram#w-EXp?titRd;q!bq?<!k$n@Zr6X2-4yx{nuA(G>z%w|zw
zCj)){pS`2d&`Z^SRlf*<r5P!37t-74_&f%0UC_-!EsYFgX-SDnPwS2Nt|!$_7@Z@~
zF9v4<NvM1@h#N^|0pvlUNV6{lV$hP{1$ACJP+@7kz6n61*i6JEaXuYSTC{q)9iOn6
zn6m0ykZ?LU3@<%}ijsf(2Rwza(L)1E*_eKNc1?B%B0>KK!pJ*GjMMN@5=x;YxUDH&
zUJYizUZYTS$D{4(UKdg5YCdteTcF_ZB92I~G5CE)kgYI79|)B>hrowXsHg2^W2_X^
z$4b4|va#SN>U#~Z4KzCkWuu@chBOs;r5JH(7x-O78Igarwm375GY4-Q28t1hvV!KK
z<G&you(r3Sp`z-tC<Dn^8%{D`l2XAC<bTH|+vKR-PEmG;7mow0DKBNQVZ-`UA!xD_
zq5Un!4X+i5o4m^0;{q5w(B2ah`aXqxdGiZrbsO)Jw-#CVo`s+tIvdmD`t#x&H*Trb
zd`}FCXX;bw-OTR&ii}g&jg7Hea&M$25~}W61#UHX((I*RZ~XZ=syew;f>OMtj%JnK
zo+$B>z0ZGw!v^a7e|9i08R^xyhNwk}fv0=S_LoIAB|6rCJcrQVJBOr_#AxM7_(HQ6
zOv><eatBn>;{_B9*Mkb^XZX!T;DDE}*?E|a{BVh9VKFr(U))_9`0l%h&KJG&7j6pS
zV!Y2;SuG1LlIZfN;n#}5!~ZX5R^9(_W`+D#WsT)3{c<oujpspbPt-6LfQ{!7t58vD
z|1_V@wMn&2oxT@y3Ct$5h%a>F{fGDWs`J4x9LSzUewc}+LDe7*j+IIyG)&i)iE{no
zc%T0?9j{Tbtf*+ZcN3hdDAdl*&IDPW#u0iG_`c$h5-)9s2w)gaL-ET0g1<10*-7z?
zT{LRW>h_=t<JXn|Djn3FMr)c`{f%iu8mtRWq|0>w`0h@ePA-C#l@&GwSH|gfe5|3=
zb4N`6jdo`sN#UIhz_AHrWi4ppU3lEtwZH1EBCSDw2{mJ!vbZgZbsX0H8v|APy1>J9
z?+wLoFCsaOVB;q54;C3;PIOEw7TSZp&3nl8gqaIRXFpxksoIz3sT{^PeQr9ET7KtL
zfH?ng9aUQ<2qVu&tr40A%f@HZ-;veHQ)kz!IaBVmo3`qOn%Cp|9i+j^(>+)C)%O|-
zwWh_tyv*LUXuG7KvU2wF2mnSX0GQF#JLnXK+q3#$3QS!pbx^Y~<Eb4NbnK_+ziY+Y
zdCFEc#=h!~zXF-&0#$hT2YAcN#zr^23r!bU1aSh)0}=AzYXoY{G0&bpg#n?yv1soH
zm)K?0OfoYA=nV|6FZA75rv+bCNLR?ooy~_A7weoFGvn2BlnY_%Tf40F^oe1DPz3Z2
z(sAzF6}@tDE^NH!0DV6{-sWa!n>#oValE*`3aK+}GW}@xQ`to3eRKSUqp25w^@4Dx
z5YYrJPeFHseU!Yx`1HO4%Qj8b{z#NnqH%T3Jmv{Q;i9_5fZkJcSDCWG&y>u?J5Dm@
zXOkNZaq?&dqpzLK;Z9KAWJluKpBFc7HQL*!J9#H4WgB-@^|LG$e4cub%F=%N+T+HL
zdM)ScvLTuMEXj{Y62*~Pzu3){a>E|ze{!hu20TNFXE1S9d1aJNt&WF+i8c~oOL)&h
zTqr4CtU`8&nT@^u0(ac4Tek!yWmN|ucOM%CDJ@G7<g(iFc`*S4HD>o0TnIQQY7vh&
zbU8szd=}txQP0!3c3*8fdnjBv$JW{!GIN1MTzaZmVhX7WCDx3VzmcyOdnrsx$s69%
z9NhzQl0|G*9bQf~HJ%O*pd10E!$v9m!;*+f4`sOW4nLK36jz;7Ck`lmoP7!BUuK2z
z!8*I35`)?f_AVuE<2nXt^=U+ZXlyf(hj;2L-r$%IrgVjDpdesRTuHtZUtoNP+QhQu
zbby}~QBM8e=JzCL{3d7cQAejtIZ<n|#?CrbA1r9}4zdcy=Z||AC-VE#zdX{N{8A`+
zTzSEsJFP04r`L`zse=+FJUYm9{bN){)>}T!DdX^#n=3-pstJxY4wJXY(=q;zr3Rkg
zI};+Zu@d&9pv2h+y;zgI>7v@wG$PZ{Z)L};;I;hZF7HbjxqCKN^bJ}J)AM5fdP-1i
zW5&I(x@Dvyl%o7dSs4@ML|r_EobdrLZ^TNL+@dj6^Zu0eAC=qJx6`t1o0u_d<E}e!
zRv{E~hl2xpfm~L)oEC(ri8vt<7bq<U!z3lqoni1X&g~o;JrvAB`n8Mhh!9vkV2XAb
zz=~DC8K-Czhw;&g+LD}^6mJU$=mF4{xzs?bkTWLjVjq}RmQU#mGKTNEzMYu<e@QBc
zZ=RknSC@B~1uSfVDgl4H<-`*jUe?b8BvwWEBpM-}bOq||S0VAvxf)3in>&D+6dhgx
zxeYGnF5d<1TI_kQj1w1OM~GYJLgc2o4%yK)SE`JK2Vgjq3{;}{b7$s}JkC9k<v+wb
z=)8xy!ih7{$&Z0b2V*ANH4tONYqrN?L<9WLg<o3pJ@>4x)egG~WMUe;q7A!wpyz9D
zb=0|DU)=o&6G3UCSmitUQ)4@18qUQVg+30h)5A;x+`T^gS(lU9J@uUNGzEHrkD4jp
zNp6)VVkz9LZ_zlu12i*u4~_oXfe&_z2-1vfY+fCg6&9{V<c2M-HZAx5-9Pjh7#IK(
za0>u{;LQC>P@!2nu)JS9l+X>x7uu=ZYJ!xune<!nmY9rf?B}TEg<a<qg9KfHi2eW5
zeO))rR;;Auxu`q6DRf_rA9CT~cm~o5T-<>F00eu}J#gm*tOWxYXak(m04Qhjw*X3>
z^i77BKDe8}{Tz_9>-C&w<1vm64Nq0}JgIz!3IRe$sHCX)UhQ@?)2zbPh&j;wf>4T`
z%-73^CVqHjuO#X{_yEy-+f6jRIb3coOJ8|u<=W;345*MCFa~O75UxF6R!{SSxM&O%
zqzTo)ihaF)GYEtzu&M4Go-I0_U3kk~Dzhpgpfwfn^HsD&iyGRj4006&$)o*0{q<GV
z#I|MrR|8BgJvo$nnlkWm?Jx6NnSxPsm1h@!l|7>_ctg(qGlP(h1d}^T;d;o=hM%SJ
zUrTc=Zn3q`-bH<Saewa^@mq4=j$Sbl!g8H^B=<n8vRA%VnaRgbrA2J*6t7MB+8Q-{
zGOz^>Z6NS+hhDu}&$%%Pfbx7)(!B94gIP)OfZCY^w;Dn}%qv|uv|fI6YWwM8;6nlw
zDoD!tVAI$^7Mzen7G-F$=YVxB{NAbFbL+YBrnsylo{`<)K>qV@nmp0ZU?3tW`vVgw
ze7g=Vf!`SA%-w|n_X3-U2`+%o5)n(Y1V>DgAjXE1fPD0OYswZbE7vj$3?C>Vy%6Zs
zIP{)x@UDe?&Fwc}Dx{!*q$ZdbG!+|3u|`ELjg0(~R)?IesoNo|3Fu+?sM>0{P9o;2
z`>p21+HhMVEJ1MA8hRC7z6)tn0>Z+_?#RBoN+YrzlXe1$9Eh->;MmkuJrGoFZL$X$
zM_J-WS%l==Cf+drLeVK-(%$hYB#&*Iz0}iPTtpM<976>f`RnTrflZC)QJcr7RANR!
zQc=n|Y)mJvf`6%emt5>llT`SR6U8@I5;#YkjnWpJscXH%3izoEDI5zeKZ<JMPL6yi
z-fQoO7Dndpu1>RSo@PbXwfk5O-Wh3iVgC22$;%D7wqx>2&8qfLWj>#NLjRM2pk<=k
z8EZHZ69v5^4;SBHp>*~1z+Eep!i@VXVzVgFNjFpMx@|AP5TstnJ6Ht^2?}27QNf?8
za&gP~&L9A)sN}-V(~<IvkbO*DelgJZ4Emo6xU7o+T%MuvqtS+;mW4E5+QM-vp(^$4
z>}<pkq4_*UBkEP|aq)i|14VLR1HFA`)``4@E6!Y&@yf}Jgw~TdbBMR_!PAK9#Cyh#
zg0Qn|-FgZiwukO6-fMDBw%tO-;C?YE1|rI$Osm*{hJwQXF)On71gf6@Z<JX0gC_%~
zDj}6qZ3BdZ!{a@D*y~<>?GxN&=u8espGNPO9Xlw$NZ;M84WHk_BZy$@-*zirIO>Cq
zVcYF8Pm~m}b?4Z>)GOsgO-D=1q5S3J>ZvB3lLE%TU|83TPF&EK4h|ur!53SITwPwA
z>J_i4-t95tQyEY-$#602OLP43h8TtVaH5{|`D@F*_(`STaf;YJ7DL+Yq$uvgm-1hj
zw9cB1&K|jTB}<*2uca%Cr0TB&Tz-w=k!<F}C80SU1%D&$LZOVC>$e6&hFJTV+m9-#
zc$Lt04q}fv(|WMq;3456&eRQ8_3joA_+I<K4{sODN5b+rJcGJx$^{b*n%-OoeUDj4
z+m47iD1b*jQefcQ=+=u=%Y*Fd<g_$U<@s#fIwMNY#8BC?|Dcu!)oHLR21z+(r-rTV
zbj{3?1xb}IVqWDBzx}s%p>or<ZH3$wXT~HPq9D{3jBSysxHw2Y169jECC!U^a`-^3
zctI{h9^BNjL7Jn8ZIU&L!hroxY`>=O`MPV`+^rTL_!rcb_{;`f2RO;n@;9VV|8=ih
zd9u<l3@JsHVmk@2>We`rAI>BAZ0R7C?;n}jke6o-;$d_aY#5;t$gB5}hVxT6^uL;v
zf<zUHH5>NpHtsAL1m`Bf+8Nmnr5FQCyh%L^-1Wf3p8)f`F!ACtvuz`n?-s+5yL=HX
z1cX<OgVo^4lgWyL64r5$QBWnZ5sqS3o+;&0g{tgD_&c7@Y$KMo&T`)GH*6JdvucgY
zPzJ@;50DdhcOX`t9S2`u%QbV2@!yh8+;C!}x8!pdq6}*K`&qVDfgWtXHz<GksWDV{
z$q37VCDiATdxR#xp&`vSq$XyDNr^jw>E*pdTQaQ2FBF;C614Q2{=Mc{;WY;-YliDV
z@jUDwgL{N`>@&*B1{Pdy#ym*s(fp!<V`u?&tl<A|;|%|>1i7kUTmcjHy93l+asn<v
zx?b7Cjzb*t%^`sY&P~DE?3(np=%=gkQFxwX_ksX&22TJGEg_fbI6Ph@y>aD;+<QW4
z(FGv{k?mMoc>uWfTTQ&?ki~7SG{13G8L#y#G?K8DH|GcHbL@BX^r>VYt*443$eFF-
zv94Gea4MAH5GMo}z}X^y!(l<g5!+e27&!s@Lj-&eg$kf|0Eh9;qLOwIa?$+A9#onj
zMZdV}$IZt#3{*1sLM+IAva(WBY1eiBM@qr8>VJUgwa=vtf54>yVni86<FYYcO2YFh
z9)vO+1ke~Lc<I*o^!FI8OuyLsZtJ<Rv3hFQka`xzx=*+6;qvV$REdVf#PQCvFF4CI
z7t<ZblccWC^)4z;+Z<MwM93dwwNhoCrYkUf(UTRHV~vIEc#=UQO#d0!I>+)X%9D3|
zVo%laR<&mJG@9z`lPPQpn|{Mb)Ip0V(i1!36W=4WP|}W@e9tq@;Y1Cc(hVM4j4fRH
z%6M}B<x~AHn<|~OiH@6zN!8UX(f<|`V7}^xN9asdr4cCT09=vk#~1nve`@V4-N40M
z9EZ@bC$(<Q{E7hX3Pxi3`!%GhO}}H)7fn$H&qFj5g0E>`{rX{{W^M1oBt>TMCMaaD
zx(T1Snuec(G^sB|<L-7dKIhxr^;a^C(+6MqA(jgLHk;b7*J2_82oO0PK_G&4o~KV_
zCuPGY<oN6;1UU!+0Sp|LeLjmkP(QWi292}>;Pe3=c_QtkBh7BDU~CquDtdd7-k>0y
zU}F#nf(`&)qSx@?SD*mf*mos!2b)pDc4vib-GApNWU{c*M#3l!%G`JSu{{=|w2@6l
zqi>>b5wyUT4h^I%uReQhnLsJa$sv?jgMqnd!_D2@aiOn;tl<(mjEKVPCKdix0+{&=
zqv@XSsv>89qeKnW9whqu&RyP7%)aSF&_cI-I)3r<nOzE2Q6&1oh*fw@ald-N`dZk9
zx38!|E5ez=1V3utd$mEZ<N1h#->YC@+$TT3@_HLs$8Nm9d-Yc`c6<HGng*uIgGO$N
zkJyanp}pAsR=ZG==|lv`1sJfZ6ZU%p)s;%#B6Zw1*R<HwdkHFrU!VTFC0-X4Pi>}w
zxsGqCyXp#<_F;lyn|!qd9S+b#BdZ$E@#LSDOp`fAa*M$6PQj3_3brv9`Ah?STHl;N
z??@fjx68j71R}JcRh2bXYUlQ=Oalm5h5MjmVOyKM^7Cgr@vd6;4Gya+O{D!@SO3qS
z%X?lF!Gd4{wzVy|p9OMG@PYEtpirRGJ<D6Jg=V1tPyxQ_>o4!e;<kSMRFo$!s|uhk
zP^N}We_<hizM?k`<x=OhQ+q{db-`~EYIw_{Wzx6-Ge6T@V0O3>tMJU-T^w3DoWe9%
zPg`$^IU_*qH%vE_l%RgE04@qUqC)m|nBfWI>#C%lMNZ_fEA6257R1+{);1j1+Io32
z+GG|MDivxzQc?mLG)&yr6%HFC0zsd40=b%{b#>3p%^~*#oOt2qZ-;CkSOb1YlpmW`
z5de>%AR?r&f&TzTi``w`7A`lWHbq<`L#5hrTC8mB3&ZvN=<#i46pHNn^*dw_pWjU1
zFGoHqoi`TnHwes~S$8}=Ql8VB`mz7`>J~EUz4NECXk`kSzO*9}g_3_mFB|nXx9Jn}
zcr<&v%Y~5#hDQ$*t2??L{u~j=EBg^^R&iQawEyFSk}f|dP4d>w!!i3G7KqJ2d`84d
zdBjb<9d^}E0urPb-?U0}H?}#oPt(Z-&?{m$N-{nZ>*@;oygv~Zxb(}nZ|b;KZ%HH4
z+CNfgJazA$F}cA!#fa#WN#b7zH>w!dvI<|h5z300R8v?_n}46wy0=!th21CNLs}g@
zt#%)O(bs7ed$Le_q`A2nw4sRVC-}g1uJ(M656~=lTrdQ^TEw^n1O&Xc4X?UYR^AyE
zf1<D{`)ia-P}{Uv94@Uo$emhagSq&`7|LjXWLx$DS_!PYVERe!T7dKQRedBbxB$lD
z{f)_EI$ms}mHta6a;4DR3<BT4y;O(=U3v<0(seu>vzLh8dvGLt(kKZM9D!J=^cliW
z^6X)sDZ;;kbtP`fwEXXPmK_NELHX$xI+#7d3e>SIXPH3<x@U7fUxJV{sN$BPSjq-~
zHSHASO~+$>=sCfDy8aaIqF-=SmA=)1xdd`$rp&nOpz8SiD}TLe<@M*YT`qR=(a7qb
zKfr3ktta1@Cg6;Tf|?bb#5>c&!+Zh)9-y;^JG%}JHvLYfyRp^!MbL5m?{)=RP#}U$
z=;KejO4#!A$Na#h1>K)6F5EF4>C`SsJ@85jkISwz*jQct$nvVt6KAA(WiRt06dL}~
zl+IQICj^9yPQ)S;4Yk{h^sbXf^?}bymX6Y<c8c>RCagAp?}Xwf_((+GO!?Gf{Zs7w
znexQm^Vq{DTF28p7krea_#KC=&uzy=<8RO`9*o5B-&h+N4dzZ}Q;nH#fxBQo;|dMx
zCADG$0seK#A}+ys9b7}dP1zDV%ne}%(lqoK@hSyW_lT(X-@yk&hWx+<hDStHS~R@k
zXsgjzj(*>1wX>X*!DoB<!;sC+p-(AO{cBL)b@%k#B^YS-+Tu>7s0-f5#0xS3Y9Gvm
z%LYk3ZZnM-W*r*()%`1bpcbw8H3}#2r2~aKuR3QkN>4noA4J2k1C~s<-2W^k*1QNM
zTp~#0Psz|tb-e>su~6%luP8H(TcKMxWCJyE`IR<~YZ*}^Bu2+V-KPp{m}ep4C$k$T
zy&LJpW0?k=a2k5S$#ZlI*}jDw<sd(sO&RKshKj$=ZK2C3AB(cSSz0+bsAlhK3lfc(
z``#(F<>k*V?Y=sy>UI!8g9sNH)aeeXkd{4Vc?H@Lhe~PaG0NT3tAJ7C6X~B%PaQyv
z^U37~YA`KSvjD16Y|OkG{@ONZT5h0+PTg9_3mR*+QM#7sb0rQefsnTt_bJt0rPkeB
zy#BR%bNK~nV)2P`=kshf9m)Leh@w~iAAGs0w`iBbDjRnkk@Psn(D{3O<B1D_ubB2J
zcOw&CM<#^V?g=djb<Ww7aa_te#~w49l;m!mdZFUWXkAND0b3F*P+uM$@xW=C)jCJE
zsue|%&+-bM#vc*3dp#;i6&HDudf?1H{GnxV4l*;~9&P_#EKI54HsS1Y(eZE#0{4#G
zjZhL=Br+!=2^vaa$AK?GF{_kbqVY<5Tl>d#qu&Z8Dbrj}_2^Mt<OE9D(|pG+k`II~
zkZry@j`^@*aW+eLHVe6@_m`i;|M6I__WY~k$yew>tb$CH5Lz!>`@@OfV*BU}qKeyK
z25t~}wmgWY?%~PSq$xqpeTzqg5mC`ov=bRP5BbmMZH6bZBDB*`X9e9G{ak}=pIr8?
z*z2UwT_9LY&9?H2bTHod`;@Dzk*IDxB<bkhQ|S>;QQ1?Eokr%hNjv)M3qffznQ6O|
zqh5ba#<zGjJY9DST>bfzG&U5XMU1#l9=QLWU6bB=?oH%KcVy@O+c)d~W9q%*xo*Sv
z|F^v(Nj6Eccd}O~B74td@0sl-BO$Zw?Cd0jkd^GcvnjHY?Cjs^{(L{*-}n9Kj~@5^
z$aszGI<NCQj^}|a254`Yp~*MaC&Qperr+I!wb-*}y1{B7;{<{-VYm#57&s{NCZC_}
zQHo_isGo#v&FrQ}n>z4tst(qFCu!_7;=M0-V2Zj-;tQ8m^wg@t+1T8a^KbfKSgX#Q
zp~i9@#ik=bwXN>3-isfK?mq{sWcOo;OkbB>K(=UgbRf32!W6(4)miN`nF5EcpbfnE
z7jr>@`lP>E({#3J*5?h}aiAl>=i<-SIrI%qX00;3nL%Chi$BvFdO_3n=1!3G0I31a
z^|MW9(@jQpuhff9fzk8#a-r;mVN^5?t0VHeydz-iFre(;Nrd4DThn^dP#4;D<iH(h
zq!O7iD@VZ5EcV=N!yhoiTctwxw^FO0trt(a4uJVR0eWn&u8pJglv;%kb2+{8c|WjX
zd{s$xF50>t4j<Y7*;hK)C-l9Y<h1!!@!B`bgdMf?1GRL^80}n{B!4y2pWE7gr+);i
zcA^ZJ|83P7dhs5*sw?grF1<%X_PCAmODbFS?4|rOVV<O;GXcAEg~|SrAJ7>Rh@EL}
z#B@YXz=<B5h>A_zu-U_=5ySw|kdW(@^?P24`5Jy`>=#(eSq`>?3{tP@Jxfx~O}0-4
zbacSezGzs4VUFBM*K1n^l}?^|O3$)t<h4CAuMF;tJ_uojVgitS_vX9+umyHEg?O0V
zPz6?`Mh3@`-N1YTIl{R^g94?hED>y}7L&sYf*zI)#`Lhu6^&9meSujIECS$zm1Trv
zU+ZcJ{f<znT~{La&9r0earNiMP^!e`NJ-1XN_-;}G~WG6O|u>=NPjp8z--b1x-3md
zuD#P)no1CZbDx+c?cm!$m@hU(+dhWVL#_UL!z*I{)%&%Ef1c8xQ1ny)%i$~iX$!r;
zxpNLnPNPfa@o2$%&0}GRLEu~GN$Ue^G6iA&$QtN7U^wm#4jyYld^YWI{@mfJc20xr
zZ@Skt=ug>7@gkgFkJ%dO>ON#jn(aZuMUM2AMcEY~N-(0$49q-aaxi!pS-WiYl<a}?
zb_mP7Cy`UjjSS|G(Lr*T1HWzdOvSvMA#1Lr`(y6l>nhTOZ5<}7V@{B{4Q9YZ3oL06
zEkbyPGDe?X;&MgTy?}Fg!c2K^Q3f@RrCSo0LhM8f)(Dt!BRc6%*dMI9&E%-lO)s1|
zG9D3^f75L|mAO}_eqP<S_et7wd#3g^3N+}~Z=(gWM09eMyZ@|+C2M<a|A!(?yx+2H
znV>ZM*$W0b)e^d)V)`N5XLbVX@uL@20n2)GM=QbZK7LpCxkmhKf9e_2W1wQCH|RC7
zd?x>VdUpR>(<@{H<6SQdSg@oQyd_}jLd1ZZ^WTX%p+Onq-67~V{K8r2l~~PHGP(V|
zJ>Amr;w}5-_F>Nb!*NOw0pZ}_IL&-kV<7{nksi(5qBCT#4`YC?1TQE#HT8!#odpLu
z<SPJ@%i_)h<PsP8*Gnvq`{ATTbrHXB!yMu+!ig&*k}jV@AO{MOxHkK>qmq8i#CH9r
zz?oYg?~!d+rux0)s;s(wy`s9tbTZ9?u+QiuZbhua1t>X}=VN<8gW!2qLd*1_w}FxY
zr$Q(Qgq7Uw<&UD@AA&wr+4*`I^PM}$eu>H5+Y89wVETW4jnK}j)3zb`u|I*BTy*O+
z=1go{)==alzlrH3POHlNH!lIS7rzJ2?cwgDAlekf+7?f)>lfOw;f3X}_<0nDa`do_
zDv*5hS*NLtmgY;ibg29I`f?k7X6V@gjR&wAd*)hlvN(6YTIfB4@NoC*9RGoIorPw=
z7RW)yG1U&cHPStXMh5abs2P`Flmv*V_Y+*|Q{l2-tp(uZU#t1_)-*7`lcBSIB@8;Z
zCMNV$3i0MHJAD~V|7=yswC0IEj}uhkXBoV$aLP;#vidy^f5K%J^)u2|yim@(d$Ib<
zT5MVG>b2H3G~TVZTRyqHW+>>%)R$pad80xYY0P`XB1d8{TKvFcN?d=xRloGPzLO4x
z`x%o$MU^FqmV)u9<_3Wab$C|+c0X<^8H$9jDDC@)S9K_c0BMKS8;N#Qc2G(}YXPLX
zjB%!V<fZ2tWO-w@a4vz8;`&bqCnxY1{HIjP*#JpB_*P&gs|q4n{c2EM&h|NPr(|N(
zYqY}7(eXpFXrN)A5tq)`ZOx<}IA+6^5#Pl$uuIs=5b2Z@Fo4E|ZYh%VVE~O9$?ogd
zbb7$%BI_gDcC1<b!1WhS+QVn?lRT`u1>LctTFdQUmFN4{nrf5xrQOi!gc~3diC9bE
zF0M4mmd`|e280F!*HUm_SMq#22A-vVX-!9I`wz~4N}k_BU0Px}IiXH2?=UO>?%4>n
zu>Y1X9OP+Gx*Un1N=NkWhgJ)W%{?3QvmR#zxmd$HFqwb&SMxL#%c4hvEVL18c6JsH
zb}VX@_M7j$i1m;C7lr=xrtq)Q)ig9fN(C?BYb9nF6b|eLG@0b7XCLeU4RvJ~(3-N0
z&ux_-K|68PxD;N%bi~Xo$C`Y3+iQ^aPESRr9$Zd&Fjo#!ULhvFwcpP}xXxn0LZ)gQ
zIIwbGa#}I3Lg>hHz~mJzr@m*{9|pW2-Q~u~3XWVqnbiCekyn2(3vF+!bL0+M3$zLA
z%zh`cQG;OU2c%t*l-l+RpGGjtHOfS-`6O)le)Chgmoze0q~ZnvnE|Fa8j=+rv6!DF
zj65+Bv@Yd4a@Pq^AyU+S^~s~kPpkc6$JH(6+{i~a!g8dIerA1|gTgYhtzz|f|FZrO
z@n!E*PNOw>S43WGSy?Ns!&5o=D(4o_<J@ij`(1B=#L)pP*DDPfS_38?%%xXjIzUZw
z{3rSIFd3=yW|-<rZayVahSNGKU5|nKEJQkOY*=s@#>fzWlV|(IYZo#fxBKo`&W^kd
zeDKHwq71ALp2Y5ZYK$d-ilH#Ar+)tIPlXZ0e+5j`&tajE-FwK>nrXEheT(9C7{xTf
zEzWxO22+}0%t&u2EGU5URP!m%21uW;T@v_afFuIR!Z|g3h4uhdgbMl@*=96kS4?b(
zyZusv1#Csy@*2p2C~v@XTM@1JkWPb4OAOEUX33as4>=JGg#n7yU149;3<)EhWAY_q
zf94wMg*KsI6c-m)?7*BVa&i13-)8i8<>*P)db?cL%O68a%0mG+`|wNUr|3uR)Ou?C
z<%0u^A`h<LzJXwuqZlR{<(SRGSy8C262jd2HT!Vmq?`OWi6>gkPyb+PvW=Y$es!K@
zAUFM;Nv^Xc_jN_e2K_LXhzL^Fz~hwE1Df~WMq#xBHjkTguMb-K%6u(7?akjw9kuna
z*m~_J5i|ZEZcbiLmq|e}UQ-E;GwACBswiyEA#bq^0zBAf>)dTF&d-umM&NZc`F&e0
z!Co-Z+!_2DlQ*<WwVh_`_Wy1FWT~XR%PzC|1dV}$Fp&!@htvn<ERwz|^ZYF{ln+2n
z*As}(w}IxYrDiad0($f1x=Bt8#v@ql)_(XCte6~XDsPuMNvJciE8bBrdfGL*ZGIeO
z9{;-d<uj^<`jim5B^qhOPt~sYH?HDU*(Ujh0c#<1LD&xa>4edD14>)6B%QO??gZCj
z#Sji{V18!$%2StiPfqg*aaRKSZLV@xjaE~A=^r;;XyYfcV=-@He8MKf<~ax|B4&BN
zw<XAbfj8+o@lx*Z1-=rsv;n-VSn*6QABlpc1@0`sgiRzpBRw5)ICwJ1>8oM#<+98_
zh!S=A@eRax_k@K0tpG)&?UKm}!@o<x#zvDrq}a1`-MG}{sUXKAZY%wu-c(24L$?OX
zkO%Y*_Gl!*wZw4zfPV!54`|l*_dj!YAElza5~en$H+CCd9eW-MRUFe#Z2-cq(oaMz
zdNBQjXyn-)+d3E(b|5FdZ2z+y!4JmOo3KBQnQwvp3>yB4jy4CU$$YwMVsVPO4^JjF
zEr(K{p)CmYeKA@XtvJnU&U{CSVnD0j@W4@i5zPD(EqCbQyx|g1KxGh)*%({pU>eFq
z^~jrdRa^VbIh6A2I4du7r1(td5wveUZq&{1#hK`W3QW&gaF>ncNhFAwF#5w<s^|+}
zXw&lcs7PQhy7kdBxAr2<3y8t_ktxqc@-NI&(3gQG)HvX1x#@hP$?=p$A^v=G{v6WU
zN{o|2VKU)=sd_f?+m?eF=-&l!*<(q29W5szr$6amGaY3de!(F^Ks>O%wM9=&eF8_W
z(@TI$P9enE&bSS1iy@#PGu#PjdBZ0pM1dbF6b%kBecBP`(R8&U&2d)?h}Q&?+b2$d
zI_@0#b!HVNQ-}9%j@nlBS(e<*^OMNzvjkoS6RGsiNC5tF<l3&imDeWTc#BX12K#bx
zd5J<M_d)aww@ID90PZJNF#MEoPOeGP9<Jc<g~9FKAv9&wsOVO_&*A60eYuMaf4Y#T
zy8>h-ANXa^Org+d(;iEpuFP@gOAv{FNc)&+;N5T^)lfq}3-rfPe;>ylxPPYOr#ws}
z{AGJ|)VWFj#}{qu@T8-ogA#e+Jv}`)K1E3}`{U_41NlPx0D}+K@f@+h2e!5uaeM`e
z3%ZY0+`Xj4Ze{n)DX=AQZ9CZd3Ed(?BN>ubz6#1hG$?7-w?FC>O0Do%XZr1Q(Vwk8
zfb+&>9kc^|@0ym}#_cGltgJI=JB;Vw`Hq5xw?5$PSS4WRbrW&|jY=rrAqDkT+)I`S
zGCBXHTlABkY&2Y!F2VAbAYL(_y;bPFiPH^4QQs4acB?g{Kk9s-ZDvN%J0+K`dKqCi
zocD;PBkI*F`|0Y^b)Glo;5#?=S<4f#dTpUcf~lz(1*8e6u$Q9MdNa!i_pSI+psv>C
zXs<x*EmD-Oeu_rb)X8W=5JIGQh=I^W1(afWkCGIXcZM*b86ZSctzQb`Ke?PK#~Peq
zY4aYFx;<6f#jBBRigx4?^%3N0QIaz`y-Ng=Z&Up5W{4#!E$5A=HcH3n@4uX6Qc)(M
zsY&PlGPY5qrK$B>89*W%H5;uQ5#88Qh?M`0V>yMnz<W(ZX_EK?kJZ{|i16_~AD0^J
zGweTTUtF&--9>PePr6kI-0o}--KG)6-t<j$Vx#EwDG)}x0JzNskW+s{{#=K|K{+(Z
zE>nOb>2{{<%Cg5@1k*MFW9MvFY50a1vgI$MfiG7MkU{}&DLVQ}88QL*<Hj23Fu5m>
ztX;~sVZyhSDT$mzup@co@?sc$z{SaY%+^)jWJNIMsggI$!?fr3C*5ME$;9gD5|#2+
z81;kY7Erh>3+%fqSAh@;xBsEXHO%M3ThNzkGsD9UE<B(sFI?om<XPm{S4U+cdJqVP
z9Y2C-DR1Qt6+!{hvYUh9CO}nsdm(qox+suaKZLw*g&O&+h^g23_6c%5GIx2PDLKVI
z)aEv4)q_$Ib3s}@YI=T32SN<gEyjDiyf-7vglImv3rj;J%U=G>%HpBL^P!**7<}h>
z8Fy+>jr1)GVO0B(?WTzY;bG-lci+ujHA6$%8*vOIt_U<%8Yr80a&kx%lh-n{uka?H
zy#0mwMJD?nV8)I&V0|gqwTC_HQV0a&Qp&JgpAW+@)4hzHfEDKIev%Jdx}=Q!a|n*S
zI9-t}t1D@0x`;VnjUg8gh65rr@5{g0nDQLr8C2VQLtbob+U#>%V>1}2F~>cGfGD63
z-3`!B>bWZ+0YqQnXbl#!PGrtrFv*w^;8&fUxF~4IKYxx^KWkh(thN=-*oI*@*&m*x
zn?IK(f>6S0vL6*!k~nzJ(Wh+4I#xp}g-C+*FGS+}{9N&Z>R~gmFwBho?dVRLUbS7^
zjuXyMT4oAI`-e1nWSA4hS|i`GkQsYd_8xX_$Q*>P23StTrGcJ`UKc|x6$xpQ;)8Kx
zEY+Vs5(2DGFvc+bXibi(m@bzD`!P2cp5z2ZXPSzZUsqU}nlIS@Ho93bhZMbtwc|zT
zTJC<40;Y489vU%91%lroU~(T)wG%BSFX7RM$=SLNXav@j(*&VC=i6W$&jgD)e4x>I
zoJ)U(JIi|wbNdJaCTapUYP3zyDF`jGAWatPGzxn90dqbr(ei9<XOEewNWG1`$*G6U
zJry+gp}+<lJN4G|ydS>=-F9{l9X&n(wy`DPdlVIlL=7{ZlJ}cnv!Pid5thnZo6kgh
zDgaQRg@GoOttyp(Zd~SRZ_knbcvmdu@oijmpP;ru|6t+{g`E`lV`oNhfsB7lDAXwv
zsc+Sh=}DG+lzt<v03~gN4^c1B$h(`QY;GhFt;?zz>gu9UaKIn$(IIPtcTS2SaGh4w
z)2D30U{YoEM$^Ag4&-vO^h}7R#dL6GH&49?F2l`|Hb@$lBME_uz_w}%Hg~Yp>m{;A
z08<*`C$vr?bA)(#0jFIt$uYIO?_*agw6uE!Nh<J48*<^G04riy3D;8R@%w*+kZlOf
z0%#WoF~l+z4OuHOYd}K4cCHsNBh9Yh4j{XQvKG$HA3;*m-(zsSBbY=??<$QQYwRE7
zy1ju530<YnOYdaK@$Tn~y9-k>J;gdPw~q};KWnjIj}R&oc%i(0rC329TH^0>VrX{5
zA>@X!S9ZTCz|ej~a%&&Bf|uVe2$m)IpXD4IyGLxL3#wlQJ6wQQs6Gck!N0IjxzMca
z`>0V1hd_n9u@BP@3c_Il&Im?@c#FkxKhU>QMdjO$3dFHEIayTcul)K2Jp(bivQHHy
zhW^$Er;pz0jrfMO^4`50+Px50PLE$(GPVx-gz|D;y+q*t8yHYeg=*78pf~onhF8#Z
zTcR<1-95Ngpb><NLI|7hr1{B<+m7ABZk5PAr`2||dOG`y#QI2{tPMmL&+2;)UbH?j
zmsdUT1_BrKIpy#vK|M6Gw%6OM1eX$6kYgRGDJhh6bRX(~f(FGc-k2i_8ZKStJKYK3
z$;q#&IRrBy{N*$IC0M^4b}{W=9N@u(a9<>;F_;5<(4zeOkRfF<Wt;tyNdf!|EC+Yr
zFyVP$gUhx#Q|mT~v4e@%u=T=@TX5;2Nu~_?aKNB%;!}MJC5&$!()OPoZ(K3IYNpdN
zDQih#si4ZnKpFdwoLFa}`VWN?|0l);v;`2FLw^uhfIoB^cM{6W%Zs1FaxQ@!l*jl(
zP!7wuxfxILT^xmYSf7MF*55EGun6cY8W^N?nFG8I*usRZZ+tp?D#XOZUWF*9P-CHB
zyaQhb=F#uOz}3@@-UcxqkoBC_mw4chiGpNxdHL98uBg}0=5YxKFir&YA@|l)RfzyK
zVryHRoG`m{^vE088+yPVT#`ovqNVqnez*6Pw}4%z5FgJZmpi0>sbz<9N<O#CvQiE$
zW@m*pqC%HL%Z&t6Pl7f)BT_d12cELX8Houh@C#ERqcR{h{s$RwYkyA2-}bKU(LwU$
z>ieN&Q#BwgLY@c;RKkjACVmBq6QEiY_1;04ira0SoPg?1*a|k4rul(gF~ygKqko~1
zva-vXpjb8m#;$H9nnA{w0OP5jg$d4+CyTH-0<-DyV{pCsNdT7*a;KC>fYa0-MP0!G
zp*Rg7WZxSB!$p6FH~|V%+TGn965)avz&Qp3E+z%oAZ@Gac7&sm=xEDI5RyPpyE5~&
z4i~7}NpQ2u2yG|$Km`eQkliCNdD*1ObL|nOUeV6k2$`D?+_<QFN24SAH4ID4a{r+~
z`lfqFxo&eA3n{S--9G_mR+62p!TS=`|J1aG#_Y8tTTF`wb3j&ErGEyoNU(!d1~gqn
z-J1h3Tqo5DbQA#Q1_{XTw4n*v@}3jweh`Xu3n%R?pq1dBym~?W%!5`SDQ^)?47t32
z_71o=dt7Z~!R0(6;J-O_7uC743sXvH)mbOCncY+vUfETMe*=K7YVHb-n9FSPRvwtZ
zaQrF&#|}_<0NFu~9|~`0!HAw^1zZ=f8C<&a2WA(YFVBXMEeCI&tMKbC&Ehe<AZxw4
z8|98@^2Ckw2>Jlmnpo(%9F#v1AKwfZ&>`9eGQGjE7!m|B<ROOp7rS~SyFCXlG4b-(
zUlc=QSER>TXKHft=sB^U>Dmf3lMXr31RbFH%7Ur?qRV^=R$q87;g}9a5WRUm=uuB~
zm%oHid8#NDC5X#qZ{UHWny2UC=%^<1Yor$B5^G=~x2v2~UD^fuj$oPtNLx`n0s>&2
z926@6mQ#NIpjSLojpHm;WLr0;U*(DkV_)hVq&;e0Wr)b(=qc=OU|9g_5MPQcwmpjp
z06!xu{DhMv0ZiL>T_7&>a9QsiD^?&inYc=9C6llsZ#~u@g+^ojVvTI%yOn_06TarQ
zUDV$eL0=ufs+CDyP3nQDY$gTJU*_4}gE7|fr=(v4R+V~tKcJ(Tlwbq=f>MJTI>`XT
z#AVophWWszY2j^cU04h=9AM79S;Z|F3CH}RUBdauA5yAV`b|<>CGXt}lTf>CzmDC8
zqy!k}KqL|Z?;%(h$Hs2l#5IF2V6WBrb35xUpXN+CDMj6rlLFHN9?a;PfcV?%2w^^J
zNa{Ho5ly6I$?b>exWXiE3|SXDHRtlK;zWkW`9dZ>8ILQ!*I<+y`%z#AvsQ>bxlM>&
z*$6FR$mOi+4tEfo1}7p^0$&$-T+K!R>C%8SLT8=t!1&v;>rFIx?MkpIKpjGJN%b;^
z;XwGGPHIA}Wv_-gXq1{;zG#2VR0~Q`Qyu_{Ukk?^*u@Y1YFh;sBgi=8n?4}#Zz9v@
zY48-Hfpo0GaoW4(RutHIe{R*X^gx7Bl2hLC#emoGyyZaF`+uc5e?!FxZld^^_Kot6
z8Qb2FPf9Xq@^(G&`#ihd*C-LAh=+=2YVb=sTS9w+WCS*ckBhec)c%`17Gj%+ESEmr
z3LYrt7$b;Ja<r)x@z2X98&n&i#}01M#3Aq^{1vbQg|?%4@}-~8mb{6Nke}!BGx+Z(
z{Ip!QMVrn-ws(ucx#Xtz?fMY3e!xCb>koN}Fc&a7J~GI~8iCR&=KX%|5R3W=(#GbP
zrwD#18Lq=k-g!-l3$8%npi|?W>XCB;%qu9bz>A{s4tDYf5zIC$!vUT4elyk-!lzWI
z-*;Ezd`#3JR1(^-vWyT0BFF7Nnt&4T8&1OY-eOPa)@Nsh77$2$1gQ#}cGSndkf;^q
zm2mM~pC~S#>W^b88d)<dM|3S>UKG=6C1o(pGhkx9`VjtM%7W!%0g>f(H43!i8%~_j
zuAGRrj%3W)A1SGUdTBMc?CbyMQeX$`x#;sZKsxGw4UD<pOa&w2lA?%EoZqdu<5&h-
zbOwzgW59Ex`#<og(@I1!D5)o`H)s@jB@duU)d!^@UN?{_7KP!?;SAGC*rOgY!BB~Z
zLY6KJ%)}maoV%G1&_AoPWNF)0#ah(9&uLsveip1xRmw4{#|*7aEiAwEs^GDf)zF9g
z=BY2D+f22If^V2PRr}ZvR6Mp2jt#MOaormbt?3z_&DiL^A3JzPrW^L>Kn!+IT=KtA
zohFbV=110i9s+TxE1y5Hh*0FKl==+Yt{ET2ZW$Ha9q~<@{d0KoG&}@X3d2u;gbr_3
ziL6O+DD0oiMQOk3%9<!!!wR0Os8kx0p1H6*30K|^-q+h%UH=ND2b7f`%imWe4QpZ-
zO7Djzv|_w~q+|@H*iI&Q5Aa;yty7~}Ox<8bDx(ej2W6!43>J&ArtO|gj^+dwXqsC0
z;a&~4!RujAYMRMOt<rQc(Yw07RGuU^v^6PqfPM2})?F#PKy8fD#<P1GGlgjSL~rna
z%?CDb`;@Y4k*^+cn>G^!Q7An!MFSa`_S?5i{@%~dk&~4izqYN~t@c~;ttJQVoA*Af
zARKl1N?TT5i(_6t#=Lgy#kMUNaP{)hgaS8~HrWrZM_{AgO$i6GFVXs(a`cIH-LQ^r
z;#{H?pW}R1A4vD7w<=-xyA%G4C*b{th)Y>Unhsz?F)ct4%a49+kN^L~^rtjY6(C{2
zH^TlNN#^!j$3>m!2_2>~LT!G(kLDHz$-)#4Q=?t;dq#YW@AQFxkMfBH_JhZ({(U@t
ziACbd@UdBkZFg95+=osSRm!IJ%d>b}J$&x-ZRwSp&c8#Nd0$4MboA$B0~;~kPiHyY
zA@TIQP8qD%vyf>93sr^T)5@j*k0qZEzr_2u&DWdT*y%m1B<CkuXOPW>CH$e!YiD=m
zw;yy*$2+~vIhI31wJX#KcwJ-e<WUf2x{CUsD)4{=HJEQorw;QsTe`bcqXdgZMXYi+
zGbt#AMflV=7884(Heta@2A*EMJs1mi%~o7S-t;Z5uP&MW`pWj8T5U&c9amQm8)OPV
zfi&1{DqwiH8i0_>H;yQozD6Ywy(OBoFy-k{==e!OkCh=8`H)DOEo;N!?%{aX(}6s9
z^^7;QY4zIkG5+2+Q>Gc;77Z@tUEI-TFjBAy`csf~GQmKPBS9o{g+jcrSBg8BlLOnW
zDHsdVX7!{W7YsX52k^6wHWJr)XdFZxEkya9&sW>)mr)^&!+RN!9x6W~>2ewuzUIl>
zU4#2pCUND@mj~i1Ll2uxa=W=FbXbmE6fjX;=I5y$ZiD7DqlId2UN4@!Dp@U`rvfAp
z%MH_j0J<uxTevyHfh5YwKR-(`s3sZ_!|3h{VDo4?yc^~$1ZeFsD<{SH`ETKV-P?0l
zQSq)VokzAFbX~xvb=PW);RW5_FGTql4MW?X@Q)=?=KG0$6utO4gRSiai&OI$0xSi1
z*7F!CdJkXK#C9}dOZ>ppG790p9=GeuXz~J;fTfvOFHS^g>F>e~j2MdEWAp0fhLd$`
z@%<~FA({b{kNk!`e^OI#4Y5tP`>!2t&6{}Cw<3$3t3;RmX3gUST|egZ;Jv=aAM@I_
zjiv8%@{lam3n?|nc1x>dacuwk@*W$(&B2X~x~hiv5sfY}{_#wi5Uf<et96g-7MWaC
zO$`w9<a5=`ajj~mdv;dPvPw$<#(0}Fd~xx~iziU5xg>G)Kz&ZrfresxXQB*MJp!91
z;AsX{p!?WHhlYY742V5ZQl;8D4tx`2T4`h)v@5MrBs^!`KU}=>t_lXAMyussJmFz~
zu8RsqtIG&Uxj+^5lH}T|8oxNoEfJ_$4ppCJ{ko<3^A(jf=ivJ7s_|B6!eVeNz$H=K
zKACXB{G7_!QL=;l3f?L=K<MwyM@si3ZZEq&AT7ETz7!|x+Bi5vp#H6UFeddCj&gg4
zQhT3%-*Klp$S<Jx1VbiVKZ`(GUI793-`{yNR#Y-2fC|xTS*c%YsW%Q@qY)bvq)T|{
zH<m%HH=yLfgO#{)1Zp#ILNX9R(hs<sVsq!3o14|Y>>;XZC`ENhy(DZU6CLIz*d%^F
zj-_#N<+ouJUaTpC63ImU%3xpWc<;bO7h>4ZxJ-0ZUKYB+&^1%kTB4O)W=QATgRA#4
z3RHkQybJtN>RCL2{TPYTJnm&tI&{H$9sjcehE%Tgs*vY!(){aqJyI~y7?{cJ3>pWW
zpGcQU`;U<$dH>O#L5)$m;Pco?uTFa(se)S*v*NG&ETOl<67o3S+~Qme$c&q_^@pH!
zg<#yv&z~!`;K&zr(C2Jg6Sc9kn+6d#M0PPLz!?)htbcNBd|=;CxP7Mlyc-9#0O>F&
z9wDLc)AhYatjAp6cL<q2AoqV$y(@r`L^y^#fg*qMNs9<38;w_oK=0XdhjH$^FG@LO
zwO_9(1UB;~OJe^>mozk_;&USr!xOpZHFxF3pL%|Akk{xoKjlfj{v%Unahwzxc*pm2
z(y%GNjd{*9uFLQIA)|*-hFJW=V`|3$?HA$s#L0w(8`IjQ&(6XR)FGwTxz-Q_%$F=)
z%KTrYANnjsofn$FeUlZU;i{{LUKETK#$p<hfZZQA0f|-8-kATjEgZgo%wCq?lY^2D
zZY8|X*{Yz3NZ7~f!9dHG;Mt!&w)$!OB>4op+}UbmXMdyPF`GqbV&Qcr?z$GWo)mmH
z;+BFZzmb#Jl+t_5{5ZD|;&he3jX7Jwpq7~jB<z}oVYN<OE4!V(Kp5aUJHK9e{E-7j
zpo#=O22WepcCStP*(okOYM9MjbLVzJFKKmebY_YZd;5@1fyp5H6Sa+P)6l=<F0VoJ
z@A)I@rBna9D*UFu{n7qQUipFq^Jd&bil}xGJIZ^>zKQ01UAtnJ-TJ&YkKa?pt8#rc
z=z3~GAduXV(ch84)+YY_*u{gS!g)U}Hs-EK<4?}P?e9s?x6mjM)dh*yR*SBuc|A97
z(4GQgA*X&R$=C-^o4JMJChk}EvM=Z;IHa$wuj4z<l$)=^4y()O5mh=l_dC?S`)0W0
z(>E6;cRfWo<;2Yg>7hr;OLai8w=w36D7RKNTRHkzv+iwV+ief!J^-ls@yS9<aAfu8
zdY^r^?if^HDPm9qXSD0-bK;%wtr&5?!kHGP++H;+!#u80sz;pRT)!Nz#Ok{*eb&)E
z5PN18)x_{f275jx6~bJ2d?`LX<{^In{S6O=64A#%=>GH!pbsQr+v5gC#eYs)VG8<Z
z6>7UlE_wdcdXYXTE__OSb0asm@Jzl{`m)OTd><*xE4OaC!^(8i9pCI$EcdoE!*$as
zRQ0AQrS2WdU(lM2j2H$YqR8tOfgs(_v^-_EhC*jbw<F%Z>qH|_YBvmE`}9c7VxmE+
zmj1*)u_Hj|rE%y|6mv}6lXVLo$IthU$DZ&UUi&8+)?%7S&#3@tCmyC9V!Tu1=QVSR
zEriImZwOyxO^ZwYtmQ7FXW}<}f(4R}rZ^?KNBRAC)GxRpHB=>^nJU?jRY01CFO80O
z{g9Ff=aqchF_*2$&rZ_bM(pQ58Ss|tnXZ^kFgzpLOCQ`dkuqu!pVtS@(ddo8ARfy!
zIl0?+C|0xkJ@Ufp`M~pJtn>a&*Wjo@CpF~bNhR&ikWSr!hSF^#_71B3r%fu!;xpZ9
zp+X#nA=O4b;t=hoapS+Wm+w}D85h+N2G=2KU~P*doh&dpk4jOtdU_M2672jfJ|V*A
z`{tL=hgeEuY(GN`3Hm~lZ4@fjf#aY!yA#J<<W>jzoq(CKMvubmKikRmk-A9WHocUq
zS=<vFNy9Vpy@s-H`wpej@$Z-vpfUHlYqTrp!J?;t*LABJ<IiO~-|cX^CO*-hoZaUl
ztPLfPiE0Rk@}P3+zZEhON16o09Waj!Yt3PyW0|~-!K<it$EJ-pRqykm8T-v>L_AK`
zf9gR+Z8sX9lv%lr9m@`KGnj7QX5L|tA9H)kjzBNkZZhslWojqszkpAleI76rHLZC{
z)d;>N?*Z8?O$Vsjr5FY$blhm+;-Ugu7n{TP@@9IUTJ^O4lp)_?Ese$CeW2v>PIXzP
zqMc>n=>tSFWU%O$Ydp^(Ba!iOaeOMeo*^;$O^RMBbU2A&S0YS8(ZC@8FMA`KL;Vig
zE44jdeV133W{1g+CJRRzMSCzu9`nkfu;C6yO|tZOyc1O>nRcJLZ#i<a(j{{!OP`=|
zB)Pb^;R>F~wV$`?+M9@{1LzAA)_zwHLreo0qx%O2CP@E+VZUm_O(gv=IIt|CBRYch
z>kWp(9(iQv`$V(&_3o!D9w-zOo<qGeohUYb(RfUP+yZ}Avx5pxvX(i~>336}WZzhj
z|3Z*_wkps#H{=!6-az1XbyX8&PzRN;FdG|z`vf>`U%x`o$7?j%7uzMH7mB=L1b!*j
z7dFg~(N0tJJrm6iaA%yhv0_+G8`O8*<h!n+PUn?qP>uPcG-}0%6nlls)5b{*duUn0
zMkpp{yw1?)_vBrvnXw#Xa=ePOvh{K;9d9Kx&AKX?;g{21;GvE;MBk|COcHwEXA<_p
zM=7<&3@07!p-JA@B}Li)rkoj^IQg}ju(p;~FB^os_QFe2*+*kUUdUtvM=mnKGwlgQ
zyJck+d}?Y&AUOTvQTFLyYyKz7DE@}GH}X_j+}Cz2GU|<<D!$CQlX>BfDv>$kj#wve
zce?R;zNa$p$9-@1Pv=EfCRC2^k>A>c2GpxpN=h^R5@2Em?=ch`pinIntK{ebDKrp*
z*k@ldJrF|v&xEgyDqtTk-J$%7uovkJ&baxXLI@ix(me8O)Fu~?s1r@^*SE4hXob9v
zfw#iJQ}1p;Pad=<*uem_fi;;`d>|+I(eC%7_rD2@BCZz?ged4mJ@u8Y`)QZp_V{CZ
zuGEa-x7QCJ54U+4aZCu>P|Ru+cSV|uwj-5)wtq;<dB|&8;g5T_uF}bMYj0@jlS$uD
zImnD#n;y_qEs#$C#UZWr`O&@FP<6D@#`JD3Yju9-#Me%DRwGkC6)LQ3@<Y1@8gbUQ
zJh6U^EXT;P-YAoCT6g&Ro_dvy{%A~;Cb&g_;J-7f3;f&+BvNP}DvY2fvgnLJc2WK8
z<GGak=krsv1sZEkz4ua)66P+I+UeJ~x^DLi-ijztUnAf@S3b8vk)U1S?^cT^MLU*T
zbb0<1CGl;Uk;yxmOC@lQ0e|b~jzZ9Bs7#(}NsTZ>+u2T*P0*jjn4n8F$s48KL16Fz
z-SSJY%d%@q-!IUoK}!`W{|@t3w!D9^*TgP>!<;VWLCF|8AoAdbf%cs$3OLYJqmDie
zB-A$BQa5pmY<LbhHte+t3Pyy5h4r+Wb34VTG<sS-6cM{+s|tZ(c3_uG?td^D5jkL8
zg#MjGW@bhPOmexB9+66Q$_Q!+`R>zj|7DWaTScx^*~bxR8h3tMV~y)pQBmnIsbpiU
z`8ln(67}Bd3aT(jzG@+RM_&(Ht&RM5We!^_b&g(v3cu%xoz};`@{Fq}2CPuCmcEb7
zDWt+0vHW3j7J|ag-F&AM4?PXL@yCVL`qYKfMV9jPa+P+yPn23nK|H-@%xB{F*1DpK
zp&NO6XD{^g##^NwCRNMg)~pI1wqjI@scmud%5L;pOO4*_pXq@&KR>sIsW5tgdE2pa
zlH!E|^Xq`7%oXoTmfOQtVgR8K5y6wXz3q~kKYvhN(^6VZ>+$UFKf$~EO1ySKdEW*U
zs3>6DNFKu?@apyd2K+g|o@<1`039hRuyUBo;HK0|gbvgtuN6IE_8mk&!T{huN2f<n
zyUCx8I8>uBJ?4@2p?$3nwFANZ3u_HTC9we)1;JsmWOn3fkn_^**B@tpU^j6mhAAG*
zlqkQuY3PB3(tgr;^@_7!#v8g>OcFFN*pqm>hjBj7sHI;0^S18<`7Bm4^vP-RIH)Zr
z+i~uj&KS>st~QK7?ULCB-+QDysKM!J1M)=fAU>Tc{Y{8<qWJZU<9&Vo36OXH%L(9E
z(4h<u`%S9&h5x*Ya{=NvdMesfxY3sf?5gw|j5C$AO;qiEv~t{c7BplbLuEB>N;j;M
z<Ub!kOTf}W=ERFc7_a^4+-RBH;Ch1-R})s7fsR^jrwjR<OJOI-HnZ*Uz+9apIXDl!
zaZr#5r73M!R}n+vG$2d+`?0++Td{6GNiFAsLii?)+|Q_7-_9@C4%(sOdNthByv(d_
z!Y8jpMvacm<?cQz;Ar2T4e{Dl{L1uu(6`()?KPRV0@9}LP20Sg<5G3~)q!hM%`3uI
z4HOCn)AIbNz<K(NdP7uTdOf|nkfSH_OspHdu&eClcuJRT0VrYvyO`1t0iBynv-qqY
z{2_q-9$wq~Pj^-WKKBOZYnW;W1N!w}7t@?Zf{EvN>1$4Y`31EwkpJC?&)Xa2IWR-&
zF$pcJTF~i&TV|d!DZsEB*ji-x*=e=vkQLdwXl-W)$#qe0qnKY_RF-?jNMA@gDuoKs
zW!3p0C2f0EjdMh{V}^-x29{tiwz9HIuLdo!ZB@?0lalQv2^8wzW9P+=US;--8eGEy
zCiG@W_EEY0Xo_1!TdbN*I-S+z3d7gL4(aj7o~0jL&CQ^C6v&~0QrS$3K0y=2N?Lg%
zfRogDf+j#zujiJrrYIvCs1$<OCgGi;Ig->@I&6x_W0bqI`B0&n&&ZR)PSMnn?Bxt1
z=u!~E<9>M6veMhS@U(zAK{1=d8oEWSiqvn4Hc)D0V9|}JNeXSm3xXO2S`J)EC5^dS
zoD#!-yvk;e%@|Hc9`0H5-JaoVsO0a}(B8$Frg7rEm(=XZm3}n4o5XFdtComnYSI=c
z{TwHdM*37#Bd=rVYUuhc#7~+te1PiH#TU1N=p59-ji<Nx0w?VqFFg{jXdfr-LCEOp
zq@ps=7J1`QO)S$9VAUo>s5?rxEeGw$!%h5@XqW1L4}TZIl!i_pXjOn)p$_2%<&f!y
zBCOzlMigJ8%|gy4b*@`pBV5!Y&J{_5!E&z^)w%Ctil!HW`0f27S5!qqk+{xD<l=iK
zT;Wo7bb7*b^YyRDyG`1BM(0&O2!Y_WTov&SW4SyDvkR}m^<^$!H(j3HVc%w1CWbV$
zRlb}Q?ew7c6n79~pb8zbWF*Pf^sf~~E9Rl2^b)}*Q&;!LD?cGc2*3h)=+KSn*2pFR
z4o$K99Yint$aI`CJ*IV4@~;1zsVn29>^%QlwjN&fX=og#79+uBKy#|n?{U=ogR?I1
z96C6_AxnqZYPjR?#GvRLiOI!N^DvG0o8R0h)!f<09t!JNfOZHf+tsaCEKrU|GpU38
zba(9)Lk|P$-G<y_qY6Tl8>XJp8B&027WUY8&|J)#@6SET4{s%CBiodMR0VNh-r-Wg
z26C+7YsJN$DC=+K!kI8EnS|J_#K%<LhP@T7qxLsGQiYY#@!{5_=7f9ee#Vp`XBo%i
z0fGOErXEtt(pW4B97+R}$a6Mp%SI<mlr#YkCMq@qEg-wS3omx{OJOe(u}<jpii~5K
z-coKf#pBpqvDd3?X_LP@<h6pH&A6_%?cCRe?v&05F^Y<KiPp7SSXw9)3zj=wT}nrQ
zOElJ%-|6kNO~zwGrHpHVC{$scbFF*RB_MM5NLEeOkqR!7Z536Ne1-mf#YBrI8_ik&
zJNZPD0%=+uBEPnO3@}r26YVDw$sR9?=sD6Do|T5*$0hx;_3dbvqSpxZ-`wF{z%?PD
zo@{#oAHJKfAqguURbaL~H~E&SiAseAF7cfaC#KhLHLEmymEXqWqMpQ0y&D64@0FCO
zu-<(2NZ%;=Ymsei8spTxSLN=Fs2co*6s#Zyq}$j4B^9j-bbX6S5xMhZtzd18Ts!eN
zV+Rwz-U!!`7;)c9QL4<)?nIl_3smehVOFpo8S@;3xwOdqQB_pB`4wB>d789WKMe{c
z_ulgSbTdrb$~BtRnEv(Xd!_Lc9u4o(mKye1agL&jATHL`c$8EpQ@SAg)5jL0%DBDA
zk6^t@!3_ai$IuAY4c-2c6jL9zl*r{Btw!xP_%ku+sLqPFnce7t<5|X|F%56>%1pK8
zKv@e~2`?|9v_9t73U68<09#+!qkc9|$JD(X7%SWjd+O<K-4z_T=(FmgG5NoO^^m=D
zN!6G4;vUZLGf@=33x=Zn`sF|e`6ZDP$Lx$9EKL$puPz#UJ+I`qvfSDBcKAh>L9HGW
zTJ_qe7c$shYes@SCcpi%)_KUpOTJ7v@r-L)>1w|j&at+l2$fN@WaQ!H<rNW$&=i*@
zfhHhGf~)i++eitqBl}XeteCFIUZHUCC=0p%b%uCxx0L2#tf_uVubDz2!u<7-)^X|P
z4ypEYTn4i81HWkjD?IC8_fUcFI&ZeTH&kl2O20N~c1JKgt5Tk=EYDS+`6+91&{6dc
zG5_mNDPhDr`OYmgKa>ebUnk*(-;?#eFEFuIDO<DF@;qqi6Sj3#lqS7W&!VgZnbi$^
zLt|zp;~UDWIl_5P;BCtzw?m?kQq?sHfz!K9XELR_ZV$li_;;ZNB2Mr_$;rugj=nG{
zz@&2WUy{U#;)8hoQCpgiUv^vGOR``Qo(mdYSUuzzKTmkqa>Dgd=fb1Sl!yHa)UCZg
zUgijii__v_s*Y;r4#DIYlr)fR6!8OWMJ*NY<#WA3PEaIztr7`tS*UeeL7Bpk5){;*
z#i^ol=IlZK{+~@}$Qm3YU^QWPqQsJh98VZAaP;hmfo=+N%qBLB1Ptnnn4LL#@V~B)
z?f+t0^Am9sOYs3+U#gP$w`EBT6pB&chAU3sEcUF^!QPiP^9sJbB8>uwEz0Z}qk8pK
z&jR`77NsFZl75F<VV;2VRDe_z0;{mDq6AQq9AvnGBptUN6aoE&j}IN?(+~jJ6Yn>S
zg=icdm^V=6D1syg33N)ym+<IIKkr8Wh4#@2XTN&`>jZ-n-)k2&s9fs<&h(S$3+U#s
zW>mezs;Fw*<IH+hyW=ntB*<D+VQ!2BSIo0MR<6t!d9eM$dYm5}<^H&Ao0uj3VEFVP
z=q4dbD)9L6ooD*W!>P~oXZx)K3o43a+;7qma)-zklXx0^hr5;j+_8da3T+CFA*~Od
zy*HN=ZZ1fL{X4#p->6L{2hvxNxR&@^=n5;Ql_bf;n*&?Q_k?i(ZVdt?`qy2boJ{i8
zI$Vbr{6G({k99^L5=yXu5dEXdxRM{R>a$^Mn8Hh;#Ch2!B8L#LbaCJ6y%*e<JHQ&W
zIvX1$M%{0qSR$st;23`GPQ~PJ0F5pzn7Vvnzy+^7777x;G`B!+0g{`$>Rdr@K*;<*
zLQPI&47x0*V4BFQ3ACnaG5uIlVh1|e1pXMvzK{Y%xwhbQSV33Gp_N{0=v+h|#9`9^
zv8?@CS^Mjb_#(fO?HszQ_1H=GPWzCN0@T7i`*GuXqD`5UP3xA>;=)f!7L=tHpMG`W
zfC`uKHc3neLJ@ts`0dR_|G(r1WP?4(aq!=-6Z14@(?oOMe{+a9gGfNpqPEJ|qT<Y7
zW|gD&9N+68I;sM<7ORu9`IPsgQ`m<ZjM><`(vhq6&0?tSp<mIA4_C$9cejSc4pNU#
zwds7%DnHo%0mDyqY>;{m0oEfa@%bLTU{tg0v>j)bnqt=7*sWwN6nZDodP6a?*)6JB
zhuwa*^NOC#=leld+Jo$kJIW)sg}+m=TYni0uk*zZA_<p`-IQB4jgk-67G%d65Im}a
zu-ePBZnluE;{Z=DDCg3=u2Sc^o_#8^ir~`rpwc4CwC_pUyIlIpNSK741v!X<2+)tv
zm7uv&YB-B0sPP&{ux{N`gvXQq@}%M8Oyjw;q`;I0Z<~W4bt?}<#B3ezia~H0DywiY
zP9ifkwa(?|i=JgZDkqT=69@#(gtYi7P(%Jm8A@kFSqo6bbObZjarA&j+LDj*O_{F1
zIVTxDpwxeQXXF%uhBdNHMvbgEzkg*Hgz}J{_g=hg09gSz2D`5ac5pU$W&q)rdQ>Wa
zr67`^NU!kj%AI&P;f~9SqJ2~-@$S8>5x}(a-Er1JJ251+?-v|TrAYv>;p@|V;ts6l
zH!-$+V+8!;9k!S0k8+H0k`@F3#i#nmbm!F>7ZahY&<_&j!?{$csarzkL7cTW!fZ#T
z5ZIO3@-?q6*j)EBZV@ZRzlH9(yFT-HDYdCr3yN7@@tn6>bUtnB=?q~U&+t@oJMiWo
zx;oT`%9>pics5VA<c&`nfC~KJDRo%-VxT}}g`Xg6%tnyr{mzreM(h_nV83Tf>VXg-
zu%y4)teAA<s=Iy-Nu_$FeTVu2(TAh8wLrwo6gxpEj7H!9JY4yW{nlv(M-u0R%<BQN
zvn;TQ;$pIip5g`lUk#pt<E@ENcv-3Av3F6)IG0mx3Z=W3@8|mB?vYo<9b7ZOB)0hi
z_+)e62U;Mn9~WHmRiD*IKA(C1{JAR2o1SGrIUgdM(+@~wImqvG9fFu6?}b$%p9XBN
z(=}zIl&~QG@1ZR&Cy*G7SI(CZxx9Nxjx{IXyv<<&Sp*-NiM!r?%N>@vxvaKhS2Zdd
zNl>gNlG|JgRU;=yvkhlZ#3-0`{Xe6wJkNCyadJ9<-UZ3qqx}vF<KL%ZH+wSK%q0tk
zF*~P|Mm%>y#lO-D4@yU$U}Ylcu(E*?Nxd92OQ5h-r#e)lF@8ZU`MlHl{!;>%&j*J*
zd%UWG6AmAL-Ou$B%EaI$I4%ADx$)YMr^*kvW&DJhE8|Jb^{_=h;*+4rm{)yG%BN`J
zg;&<xkdZHJMM6Ai_n;bclrZPQc6qIUl0>2MzBt?iqdwqrsV;b^quQZ=2k_!9a~F$i
zi-#ve^ldM$C=)q8V{T|@fQMOZ0N_&=U`fr(PgspWLs=LNQinaEh{%yu1DMp-<|b!U
z$cbNLH?E(?z^DIbEC@L@>NK#U{r$^hD+d;Ay#p4H4{R3osj0+pX-)r-UygR}f|lo@
zC75oK<itj8_d!Y$LodnIvjamH-XGE<d+;A%3I}ntpx;~E_X!INLu<CJN=$6L#eF+L
z5wO5ozcwC?eCo4&s#h(bMOupnr_5(oV5WjiKo3bQ-Fr+A0o4!_mQW18jjg?H=z_(O
z5}UtJ9E^=Jl=3RgB6DW*RQ<j*_&0Tbj^t0(#oozM4a|SyB1iIVVPRpi(#ktS(qo-W
z(Et@F%gD>mAFPxCO4Gz)y0sPi1Ex|hb`$4EKjxSC!fe`v<Tfn4!<x6qywOY(-__+I
z)BF<B>MAIJAnz=7!A{a(Jg^H0ri{05&Giz&ER_n_4&btnE4`Z_dHy=8*b_7J`Sw=p
zApgETdH76vfKx47qSL+I?-Tz-@VcK`2`nFf5r|12M;WOViy>sL3Z|b-Y3mw>EI&eB
z82uA`qG-))?p&$CAk+Z?7!`8C<0}TaO`t<5f7$_|(?IQHOE~faLQ@W(2Z%s<X6!%B
zjV@SPedXze0n5DKZlAk|Kb+!WL4&DG1f%OACMtJG*QeAGKZpT(yuh)}QwL$_>gMK{
zwE(30Iy!QZ;cHdrS+A#{HxCXD!uRpYkn}(P1Z!y;(q~(hYN6LO4|#WJp#4saez29x
zfD6MDpN5Pl$!14BV@7@bR<v&5C^FyI*Jsg506EEHWo5Kx4HfY!5jVS#oXsoBBau_&
z=9QCRhJP3@OY#lmSc&Y?+z_cG?1qh++exf>@A?-Ntatc+*H@O%k~O6D$-BKTb8PAA
z2#ju9Or?sF^Fa3o>C3R4Q_|Bzuo{4J5;~HiZ~m{<h$U7X|9C&a!rRg^pTpYl)o@a_
zCfmmRGmcw1946hDJ90<-T27<^!Cs-($;0bQdZ>T&nn~{OZiy3IrDUwZCSh8w&+;eM
z734H#of#jm1`Gu#z7&vn0eD_^7)Y$s^Ej-sSzY7z!mepN$!lj(@Rc>b)3dci;LPq|
zjQY`|*cZ$62LGQM*)CtMUpzOUI4|u^8GRf3z$yWyB6>4ePe6*M_-0{F6ju)hF?(vi
z#gxv_KYd6&_9I#;hiXs(qZAvDDdkSvSK1OnOte(NkB9@#w02fi&x}_J!%L8G!LQH6
z%|24e4oulX!@?Hg%GSm2AD^^BlV&KwO0N|wcjz<Ne|{88R|co0;!Qn{UrKXo)MFuo
z<xP<o1|6c$e(OeXzMR-lE*J@7073Y>tO&r)Ff=tVDvN8NfaVYW6u^|YawP%vo*n7w
zI`k3)r6Zdt#_`YpJ0qSpu+pjAt))*+PG(WAFkpvqL`_u{1hT^ELSCNED=Wj~6c)m|
zF^~!Bo$ZJaVvnHzw^l%l6zrknKgx~E;4w@Hcz7wI47)%M<^(#+*?YD`iBD3$KmV2*
zK69wAtH)6RoI*RLiZp&@8z%NqURQQ|)~`?daC{k=+bb+YPo=Lu^_EGzF?o~KUf4WL
zfIpQo^j%!AeYtO)f}q~}P(3?KZO6j*z26d1q@sU6HiR9XF2=_chTTS6($`u!%tjCO
zoUcDL5@Xn^U3q(<pviWf@goy{hLZxsm?tQ<nenD5A*2sfZ#^#Hhk4t^enLV5NC3ev
z4r#rnLJ=58FF%SQO{%9HX(#Z9Plx5eVA#*Fw9$s<vyH%v0F|Ml922QG**s3*Hw;y}
zm(;s7OP&l(7(U#u&T!v^-TnwA0rbk>ohWD*!$r|A1wfL~xUW-3)63gm$EZ-J8TyCF
zvvlvIqUebYk&FS2(F)B2C$eJNaQ6>vkIB#ODE8H<?;mPLkQy;vZPG6B@bXIL$Ol2F
z5Dl}%m1w@|U6tQ-qwLLxTodFM>$2@`IRYK!1Q+<yznA9quT-ChaJT9E6?xOGQVi1H
zx5a7?z%2gBwFAFj*Oerh?A3~prI{Io=E(JWq>*?p4B@$wB14M66&?K(q}19a`(9$Q
zf4{_KP#ubpq9~}S_|p2II4jbOc$+h_26J~<O0E)Vf@5y`M;pvGl$amok#<oeb(zD4
zWL&0W-hAj&fHJMphiw7yQRxqvLe#0?3<E~>$$Jp(1qxVr-jU+5Vwn7~kcH8$dx({Z
z(R4thM$_5Bw3T>lMoL;5pd9vnAm2F_r_rL)T|_yNhAqL?&+pW|<0Zqx%$FE-Bug*h
zzQ(T=jM>Im-%7u7a&kgUCx!|<I`WQl^R=;IrHk=&i#wc0OvrXQFh6<}9^cno9bOZ-
z`sfC&R*@#M)p+vZFYtbAzkHdY&;h;zB$z?GM*VD#&}l9(Gph6*jyf`ykHWB0DK<@(
zcKoR~b8WAee@EJ^otu(JGNxaYB6#~;_;fv0JVz7Kp7O4pTilXoHgzhKwJ|p=5z{{=
zVr3X&0|4_z#r7=ckd15PrU@O`QvEw;*UG$s97jKOQV6(FeO^eN;uC9Qy3>($NbB;i
za~qNhUj8VWv}dj_fqs&93BXGEXYA$b?!4{-geqv?Lk6$GDQl2}OG+s7=3dC=rbC65
z01waK*{y&YZ?}k*BanQ?(KLXVp8hslX^O^`Y-Ei*6W5h&5F>`cp-8*_<1I845@;81
zb!}z`vgkC~?q2LqfnfCecLq^wGC2r}03qF&ZJ(tZ2l=%&E_BGBYw+H+WQ=>+@e`;F
z35ubYF-CHG>wqM1c3)mf9Q6*!)nxXVLNdB6j}sW4v5`#>VQi%m&fiOWp;hX((}VZ@
z_~(cBo^aa!{_h6Qg+Z1WLZF;yP;0!`ZL+j$_$P$$-|OI~*>i7|&K29sd)o3pcdBrm
z`o41cy=-1HE{ghi`}-+I@=58>!`53J&|dsnyvEH9)igYAC1d+Q<Zm6K`1)6!hmVi+
zN*F<?j6|kWe-lQavyjqz&yq2pFXN&?40rIU{^AakLp23jlh4TDblkv-La1;=&cm7d
zLN*n)1jdse@R-`#f;r~yvDHgMResX&ke6K;P@yU2sh8_=(8a)a13)7ko92kPM+TzH
z5+CpMpv#6{oi#k|EFbtf?bbIIgufp@EsSIzuW=%7PF$F5dNFzNscC3mCDCF{BA3_;
z#im}0CY6q?H1%GW==t?^%r>!9l*~0#@u<Xd<n003&z0%b@+Q-qq(IL5MjSnVE2OJj
z67Hq0QpFZYyV{p?7`5JCP^qh$!?UN%IAiTSE7kZw8N*O^gD=J2)pZ9X8}Q&dIFy!?
z{C<Qcrw03Zjnh1KU|*kdQqLI0`;Lm=?d{k$g7=Vn69;VKUw8dVwMCWJ3^Bclq%r8E
zr)1pisB*{Ua4A`$;%vkQB)xY+R=odf|3OR<@s%qOsnoIS@abMx6hK)fJ{Z}Zi%z`#
zF0)e;9>V7hyhJ#Sl}$h!7xI$z5&`oOWwh6_>$<#>^1B14XaGM{sY$+m!R8|R31V7(
z!$4pg)vc_sMa#f%{5^P&D%t9ajo%lt<33u3+<w0O3-K=(he6q1QCDft%c?zBKBdm@
z5}(!WKb+Hw&MuyG>T}8;?M95Tt2cAch8eVil!*8^J4L^hmq<U0#$Y3+yU#kltJj9h
zq|;pIjE*5gvP3=O2vTll<XeMr8PaC9c0L1bX4v4Gnp7&503HE;oRhOC85a=g#k84`
zhA*8R&O6O?6xbY(s5B)W`)xHGQcd6yww~@4NMZ%Hw6uWp%AGD4Ue)3{xaw|hl|!n?
z!EZ`Dei};3t3`WSYzdp2n{d35-%JKlo@ID_jKuNFl}k9p6ta2nAKMBB@L&^qv}W=|
z9~=%zj|*5YiHkYYn}7TWaDxPbvBiQug_>WTu&si*t)+!!qJ`wq(rf+FUxz25e6A>%
z!$DM*oEiwR0XGGgKm9ROco4vghXk`H;KS)SHL-s5-7}Y4@E8DJ96~UbpeK<{h44)4
zIE(ULzOw%Defh}}tv_+DM{FHfe7Z;4G%@c@a<Tr2$gAG)F>N;KWGU!2^VBy5X0+Fa
zP4tZ}h#e9UiMt;X6sd_luKs^)y>}qh{rf+DL{_p=Ms`*uN%r2^BnjDjuaNzc?3EEh
z$O>7volSO<5JFZIviIh99rtIvzu)ig{BaMRj@NmO=kvNA*JE_53|ZcP5>pQk>)S$2
zH<7q_oYy$1yBURB)9zbOucXGhSspomtL8E|zRf*NGS`-rp3eFha=J7%!6Ust=dFZj
zx`B=kBQ_^`H^2`x<ouz94tb3k@Bu-T)AM)tFvU_;_=?2?r1bvuD#PH_xD>)TtE=a?
z4nlqy<!gAJ9t)uIG;Te-lj1IXI@cZ7Va?C2JjhK$fDP_5ogE!IyOQ2`!GuAxA9He;
zo#9MfvTV44$RY1AU^#?xOE&9XN-xLVyP)1_(ZEKf&BM*xZSYan?mB}zq?<+@CqXZD
zcD&U`8RT)ey$EU$`TIqhUXG4xx8g%kWsZWKPLMp&=I5~8dYvH1x#q1x4*WyzEreIZ
zkh45`QZp^w{HRv?Z^ao`%F?*y^P5+tLspUZPKL)_{`ZCmUf0H6vX$^9|DjN=Zq(_o
z&bYxApHa+ADxFP%J^gek^;b*tXB>)A)Q$U`Wf+>GYH^flJ*WK*UktwKyi?kt==6=M
zENcukDan4PEn2x?P@H7a{F2=vL37+5laf{9uF9NuX#UlA6aw|yieU_fmeDq^T*X0F
zS05sq1r;}#VS<}Gc=*~psjZ9wf!6$2noLW6NExrx`R(k%kqE0PL2I<kh0+;-X9S?J
ztGo}RQ;jSP)z=Ew`&J@!IntAoJdXG8^=*NqN&2Ay8x5b3&~3XPH={advo{%K9ztBN
zDc3`?_-<3sX>>XP?gGjtWb`jw<QL+wxE`KBMB|{!M%wJhqrEk-B1OGPPj|?@6hhyX
zfxbqmZ&Fg~T$ir3azdjE+HQc}i!|Z-;s=8H7i#~+H^ZH%Z)?sWTeINm=TZ>Cg6IsW
zB9XG-q0u^nLe)()*46bm{RFinD5apPXypu)M2svfg5O&uDRq?DR_aVlO@Y7*Fp&AJ
z80ACQKcNu=(J`od;vr<`2HqA-<7$G`ZnO>h-hQ&rlmaN3kKvQ9CJRpQ&Ua1r6dwym
zXiN_8G!UUqzF);eH4^@AvkG@UOK#srWQjt(G)HnU3%|gXqo{ciLm6WB^)ohtzM_2I
zcBMU+mBy!LemW5y1uXDH`E0$S-!c}sV^070ekaQB*Boh85u|Wo%oaSJ&m&tBOL#f|
zjtr-@tCSE|BAP|^!2=jj-kXLms_E5LrA<!^A%K~Au=8K40sv<B=)}Nh=W&J_{Rrq`
zVHi{}G{6n@+VyJe5ColU4YPu*dqh(dK%8y#Dslxw62LmZbSgDUHtSY=*y|`TgNS?z
zMNC|wdyTFycuN^cwE#EMyD=vx2QFew4Uk)8jBsCwH$#3n2ZtM5NbA-y#1(djnSn-2
z`yr7;^eTE^5j28*L9cEXo=#Y<(CKjq`CBvCrOp40FWAMNTttEChvo4WuP9pE4DQd`
z+OXfk^{}soFvYdiRggtKcty*jKe)E{v`Dk;;^}k?CT#4WEd#G>Uf$j1(feuhd%L^R
zR6CHF1FFj2tH#3%0~Vt{Ek=*;KC6*_RO_0^PqvvN+qxGK1ph=T{$z87CNVaXVAWV3
z%wGR&wC-xCxISVQ<uw{>7<|3^lBRiHH_nd|pGD_YYN6ITaYmZ~laVTpq+uKCI@g(t
zm}xj^yG1c0$zA-99gc##IkC$gecu<l_~x{L5{2T--JiSBWfLyVsI8^-(dUTw=B2aY
zrw*=`y0_xHrWR&q5Tl2isX9gaT3Y)MgV&wz9P14;MoaX{x8s@0-wLpZ@*rw@?+Bve
zE(Km_<!s|M*G^!+U+dCzDcVpxYIb;dIOJ`kG$V=C2>7-CN&>Ajx!DmQMxgVBeg;ya
z@dKs*@bzthv^TYA56&L+J*YQnX{*r;uju6vKJ=p%{GVld$h%Z1`<8Y<>lCTtz~DX!
zmYKggyQWS-ng@a(2ylSf8U+ty`S^;EkWjmE4A#Jv7z$DbaEfT1;f!fuAO{~{=Net;
z7C^+R%khjl7tmInqRv@{+U<Zy^`!4sXY469!*WZzU@J0Ua4*T83df%6^8H-Kut{n)
zpTm&y&+=6sk3I;5nk2qhO>aKdCaSqRwpd!wRBX-h7?}XpA#)siIk77S&B<XC&(k+U
ze^BB&LwU2UG_tEcd)YxQn$#QJVmMRRu{`d8bsOY=@0ZGas9+=w#R^q%uzH`Y^d3M+
zZ_%kY$Y&MCO$bQn{(d7Ex}MJ2(Vsu(JVaAf;?^6W-56N%yhB|uxCl=eCOTu8;g&Ec
zGBedK43sFb*eufSc{V^7&BQ4F-S4MHU9tRUxWO-?fNCTtI5yZtf@Ot$*?rEnbJ4|j
zy#*(A9KyI@L=P`QI{>JB_gm?ZDk1Aqtz@s#@s*WRY+&&7;?Lsgfw(chW&;W)O^!3O
z#&svbD}~!hZ;BEXrl`L)*>HJ3$ax#9P?Yef=w1=A;yCU2;!g5M#^m*qJh!nPdx98#
zH?=Z#X4Vn)zZeGxLngU~b0ag0XP2ol>jQE`uFu)6OcGQOJ$2N>Dbhj|Hgo3Z(!8-G
zq7tE(cNbPs_zY07o7qh0mM(Azj@B5#mPs_7HOd9`j}|*r!swcR$g&xn#^93nTPozl
z8aBu6Lj>d)M_)ZAw74;H0s9Q#@Ka8mI-MD<@KpM-JQHcE3e+?yidz+Rk3J3Nss{f0
zc-}$$2HS>7#x9#nP?K+}Z`SJ`FrD|>+$rI^DTDSgJ@ym?e33y0_bC5VP?5H%KtE>S
zQPtQ_Ks)*V4k+W;0+N|WGxV{GRbv{rB$Yo6J+7CzYv&SM9d0u3AsP>JTL4Z&Y0~&A
zI4F*86YoD_tEm={xr&<e2xlDJ72xbc{MH&CQBPTAP+~3AWn)~V>b&;~M<Ut{N{~!!
zDD#I$MxeX-Rz_Ls09tYIxcD(+E!<bm4g&5Fo-`QMxeB&`VipEWx8kLP!H74}ct27u
zXQ02oKR*5{ih`2Ts%oHnSPI8RH><ybds@h17BaoVGQmKMn+j~(5Fek0oZA=vYkkyL
zIIT-<5L00~;(y2*uYVnVvFmNP^amE1I`gETV`-SHjc2@FSTdz`L!7_N3Rv;`)%$n%
z{CNAew<qas{Y+p)?`;@U_QFs6^W{D^S@N9K)hw3NZf=;Q1;)@lX&iGWM3JeD8IBwK
zeO>cgJ9>b8ljHAX!xmpU=1qCWx~(6*MCcGi-{r)R*8*q{L?91bo;)Gs53RsKfnHmQ
z1%x6E`tu%X{PP!!JzSgrn@oTS&VIk01<;m*j55}q%6{b0-HrOY)!QJlj~UA!X?Im1
znox%Q9KvJb6_`_19x#%m_pyF1MII1mvwrh?5S<vCA}h6rU2j-{baJl$@zbAVdj1^W
zk^ME14u5yxNbXJP=gzN9cj_a0yORpx8;-tx`_Lx;%+PU~j#|3xvbl%;u#};A;jl$O
z`vI#`$yTf&iY<<oa)QxRd<}DI&mh0nsh7X_)0P;NXzjDKfRzTrcf5n#y*u^MDXZqi
zxpZ7MYXbz^disO*)g>h*ak|gLGTqZ4n*ai}Q6U*UTiz^j48iaAu=Y9ntt)|0JleuR
z-}YnA1r42}$#t;Cf{W0_eDmiZgEjxHhb2*qb=e)u<7SPOFq8P9FvUvbYa+wnbm=R^
zhq>CXPL^IH&Y8W%7)ewV%zIYMTORl5iiS6LYNa^RzvrEqlv_#UUi52PmkG1q)LD4Z
z^offoXH&OR^}$GD!Q3(QukzYBNlLrB21&QpBp7g&F#Zy-ZNIDXw<6UTE~ow7ZBmDI
zRP9A5)-TvHCg?{l5q(8EcM!swp)tG~-VQxU!ewOo`qOIWW4^I?|BD49OOQH}qYm&`
zTgsjN-RpzX-9k`fKl|!272N|y3kx&hL3-@hW>T&%)5b!P1FZ(=iC_;1l$-(ZGtWGj
z-Ur>O)W24;{%)TmsnM@@JHF91En-*ptI5vYzfZU70-28Zm<Q)3U@@22T*B@_N`!QR
zFV;fZZOyTySK1j}Pas2aBXi72@%y-INl!DDe;4ah4G%YR(C~q0asO?eXNE~j-W>&a
z%r7?RS_TpPo#Gr;(_0ir{RK&n{OITDGO-^F0%O&Z@FMHs4d})Y#6Uxq5T%uH0s;xZ
zx~JA0UVo?@wPoT@*ZLaSf*fRW?_Y5Skf7TXbkXcRt1ByD4-^s=1TOwGhTK#~tHxkH
zO%(}SqB#AuTiwH6LWx66^hMP^Nb6sJuz_g|ckFrbXT-_&RbJieO72ZxmF*v)!S$sa
zI_2@c*4`uh-z=LS37^#s4G=%UK^++3oAX83^=#GGOCzvnt3j>Rd%Jz)awH~;FJ{&0
z$J^KSKlq0>VgDVW=f#lt%1Rn)3Ldo37gBf9P*aCQfg9*KsR0@kc<4wHSfl-+gu9HY
z*LRrJZu9FiFJ~kY_Kw3=Et{HQy6g^d8wx|%7r<)k2Y-+)Z(U6d9tx^9vzQu=7~~wT
zgMp5@wws;Z3doB(Iqk7LPQ;;F)7Lz$8aYX?OAYr-T^9Bj8)tazC+%jNyw1f#AuC~Z
z*rDJsU?S#l%v^AgzS0?ZcKRm2zIzz;PWIthaM>eA|JRwK|GZK0_RhV8uJ#QQqf(&i
zP*qvH+RBMt-h}rQ`SgSk5%fvrdC7TtLv1hLFp$1$11b>_T2-Jpc({PZ31&->j#Kv>
z@}Pj+TdAWw=)y(@lN5ACBojCYt%$r+$;c9EXm)7?uG8j~cA-$|fd?2h92soZkX&G8
z6>9Vbl>45-v;$9gQ-J7aiUOe12X~t9*M_@)i=iNbu5rN=)9EFqL^KDwM%CBk`0IHk
zR^VB-^)$ZnT2+`y;;oFg0mir=w4F>+uVYVPEd?gB`|gi?dGAqV9>{^wJns7EK_xzF
z9d)LbfhQ+)LU<M_ab%cc-Fo*flt90oDN&G_Fuo$XH1n36oAAv^k>`R+&c4ReF=XQa
zc`v>c(8AZ^Ns@QtP8tc804)q*!6mnqvo|<Gb`}I(X~G`r9EnihlN=GRA9&xt|10yl
zbIngnZWwYqduGM3XSZPbwHpic)2M>AiI7lTRTV@e15FY8<tK(DrK*4anRc#;npDh6
zp4e;L6VWJ_iq=Pe&*(>=PDD^haW#OE*A(vMUG<Cglf<DSXA?9FPdCUW+Yx(pFokTo
zgcO{q)W~VxK2s<@)Sc|+=j%2_p8t4$3sYDGu-kXX-|NGCdNY_fIiGGc7R@PasuR3{
z*m+$d9ac8|7HNDek5CZm10)P#BX?dw3>RpGY#kh66O^deD;Yt6bsyrrc7Zn~l48AA
z)GZd9h?P?8E?i#7^!@r(1!7o)ia_yYT^SV_=~}!uDdL~3SE?n=BZRexsX*K+9c==R
zpTw!^(246d?Q~g5q9I$68v2Hjq^YUNO8dggo75=b*PqXdbvk+YeqpeoXiZ7~W*sE0
zrFcbO`h|F;;nDy}rUUoOEHec`)zk9Nzn$Y1%u2ODPFwe2p*xNhqM=Y=3Ieoq!j;ul
zA~7sKD3nHwVmiwpBaerIz`5qQu+!NbobusLs}fRp_*x-nz>*vCpiNDs^T`y2eUCRf
z*?Ux!(66`aa_w`@H6@LQA9RYqT@&f+r735+(Nji;9N$CnQ%AM8iz?!#wK@#4rhB7n
ziZFW%8J0LWg$)VvQeQDY?n~?6MgzD&rS988Tgp-d#VF(e%Vp)~M=+)aM=2aXsBPga
z<KFH8GH6!UxXZLwEq|bso8SDw8uRWWVH^}m==NGc<(<Ok`-mWq`0cpUt^AF&!o4l<
z-vA{fe&ED}9vDoCiQVOr2G`7PDzJ@!$HszNX^uH){s2mQjzq9%2B`SU7lNDjBI$Ac
z=}Yh|n;{tmCI0aB;3C|6&|k-Ob0>CKF<lFfA@7VO+)Mtaw)W7KhCXTejE1wnvdZ5V
z&|&1v-PlFni}3bZ10>?vZEtM}8x(0ml_8k=$3v7VasX`{QYm}g*Um07zz3Z4Gvns2
z2HlZxyW~8s)_dV8Y@M%VIU`;;T`<)fiFD?t=1tKmDS?8&=eN7CJ_n$WdfM9es3KR4
zV@a?ii06jTOSeR0$BNsP_GD8o0R%#BGMK5>8^D4E#t|sL%)IuVM8vA{#OT-b6m_;}
z@1F(F)QZyX-)kw#O-9wHV~UJI4&|ddqL7z11JOI^bJ0^&Scn_wZ(ORym4t=LQTOm=
zbdQ-0EE3;-7BX?MSbltInnaFl`jC*n<YmPrGg;6f3k(0;{06*w$Z3AHySL|&X7KbW
zm<EnL>^wi4chAq}CeO8SnH+V%4_yS|Ss9m~;3AL`?gp}8RzJ#5i3KuU_qxKbP>q+m
zIN{IQ$_u7K!^}_u))%4p{ZrHxXT1_asBB8!o-)Ef-J=g0kH9p^p<!n@(ZClbh5?ci
z)SXBX)H&F2adVGJ=yRoR5~d)yLrUY=x|cttOIlxKK?pwF(0COq#{_WMhWp#ukGdQq
zXMrHCC7v_G?CGb+-Mo|ip1jj3R60`+a5t6Ach?(A04C4ZNVD`vF<JaY)^RgamqW~b
z-B05iP$Z$L%)~aYv6U|JG%I}WuY|Cq01j9J?1JtlZe>i+a{6`73bdA{;xRQ%7d~(*
z`*yUxo$gjPdGWKn)=AU5x5Dbr;r`1V#DIIX<oIWrh1}w1IqgM%`Ys!+LvdoPgKLRB
zG78=@Sr2muN6UI2{(@o^T%>zH^><$E+oJAFeeW}w1o-~hsgJQ91pyKSW>{c$B<V;2
zDM)cz^fX#}dd%3IC@?_ZE*fz?yx<F_xrUtJG80=&jf>TF1lk}n8EB7zs{mFu$?unT
zY5_S+QUcq~lC=yKLjX)<v!L(QVjo^|WmH++@q|t%@YARN15IWxQX1R_k7xx%ziDsM
zE3%KNl@m-1RWS@59oGT25OIC5zmWCHJ*~-e=L1APwHpH&ExoVr%NI1xrQAd5?y>WO
zf~xA2_^;l0zbz&F#~<HI*$?*b5zk#68;Vh3iG#8X;z5llAbFgV0;a<{))7Ft{uhYu
zsMUtTZH2ZA>W|X8x|E*Z6hV;o%bQZP(Q5Zci+11bA9lqs-+pf8RUl>@o}7MFXJM>G
z>O;St#ovBn%L_J6I!vTS+M0|p(F3U2+1Wc5AAd*C^sL;#^{=-_BMM#B(Ruso`LE@e
zv=Z#1o)wQQ!C62-G#C~MSJW&RI;M-qIsJ&FM4*ZV6TvjdQLRkYDWavOrlz6!%^#Kb
z^Ad}F9fzlmLRB4Z#?g+=o-oWH-Yfy8>g4pWq8wm7q9PY&+r;_^STuDxI=7CYt)vXX
z06bozU?{n7J~1I7Bu5$zB}+0;5JbIt1$sJKPqVanIx$~>o~Wjn()9p;f~FSc7dO(A
zd(DAuLERaeO$p<_*`<z^TQkvW&NWA?(iOLBMvSmw3+RqV2@8Bm7q9G;+jS<*On<r8
za4N)nj|OhY49ghxCbSr>lKD8%TOU%1dika*bPRB(UUkl@rU~bRPs)?`R$y4BqM{Op
zCE=bn1!FdFUN)GLtDz>qZk=(gE7|I&+b0BK0qhQ8-c3|@X~Wxtf`S6X-w<x^xj&<R
zeza?6BosT#?+iv;B)yKoly%@TL{}XFPhm2;?xw#BGHQ5~z#~eSmX%~D63CULcShrm
z8d53I9z-x9BaJ3PaVtAGI0U!Oz@49lbo@L~FJtyeu}JmUUDigU7opP5fbhjD6gD<A
zHBF8Bfps1XAV+-ri2Ql^M}yv>E2~;i+tfVE?MF99umY0d-y@Ks-U&|T?n@VAO;3bA
z>cHDHCFuR4Bfl>DHNc7KiRqu+?s@}Jpl;aSFv8`YZsNH1&Cwp&o8XD<HOGA6r=Abl
zxsX+6%Nw^R)CW>;uug@tK~H7V<6ZL0wwJytKbtULI=rke`7qvZ`Py4Z$J`WPujKcx
zf~iw$d=vx(tzg{A=(z`$z|SW=9_-74M-CwUsCR6iXE)v`v7psm;L!Pp-`>PTsz{UX
zl(e|85FsvwB)e7T8t}WT)QRkzg)ysBllV0H8Gu_Kfati*bU?0cc@=X8j7vY()jiGE
zfQ^FvT2DsP(22+-z}PwJp&t+7F)PvpklZwgPQ-J2wQddn{!gP0@KXA@x66iJozM`6
ze|_`xG=1pk<kXY{@ASoB$2tUcWImR3Z^PZ&&h9b_Du>Ql$cf`th8Y@|%T7x;42$n6
zb|-9C`>uR}Nz<P{EFL0!d=^{{^z1OJhNv7ed{}Fw$V8It20rFarU+cY5X*F8gZimb
z=gvmQBQn>6O|zMs;78`$bXHK(2(>eS{LktpQsXX#WH)Id1r5F@o+e4PFqnd6(Et6S
zDVMw`7pcn6eZm(MU<;Xvwz%3_kHmm7JCGu7wa$i#7#D0PA?~f?QP<Im`81F}+l@D%
z|I;to`?ejnC3Sdn-G^D?8G$Kpi*HTvBdr98z=N?o+!{Ejmo4804s)!=Y<UeDJRiGH
zBfyw}Ef9SA09>mQ@t&o|M*SDW9NzhVL(KB0;NprKD2amLX^1Mig|#0M6;)hP!Y7z?
zQPBJ6gxLBDgtaqC&4Xtd@deqg9Wk&lgWf&7osehx$g5s4C)P3u>Pm>UhUX1V5E(vV
zdRF|3QVKRS!GE8icfhj{q;V1r&s-3Q3h$gn8S<v2oOG{PT69FamTML?xUD1`-h>wM
zXA_bxz4%-;QL)<%eR;%7wY=A-F+$nVn2(ArWX!_-^n(Ew$Ccu87%i^;*1E8$s^a;M
zCX(otWCq~0cjFaIkt*QJ$#{u_X%)D|_1tZm^YDeI6HxOBGl8gC5B(f<5bK4~-U1?5
zrH)lHN4&zP6eyP8FWU0n{;rv)S-^A%p>7WL_Set}Z(TWB3BB`M-==-nDupo8=D`9{
zI@cxbMNLf&6sNR;nDddk({l4w_oqijMgl)yyaO!=I?Z=wHyfPrrGp`iar6TWJK7AR
zq!|Mx!HTA{Q-*RDYFT4rV>7cHP>Mib>;ke|nZ^<FyZU5io@fo^>;`z)@Bb+kUZ<fq
zdO#+F_o~2gqrd-Q^83K_VYlrQ_eW$y!mXm|zz9Q+(HWa+Q!Ut&l#EF*5G)C3N3MEe
zNwh`3c@{M*?PhW--dOvZXGFS#Rmbw$TAb?^5DjNZAKi-QfGwFyW7PhpyncnQ4?cCR
zugEq&BSnxlj18}V-{W|@B#fpnph8O><e_6aaid#ybWxF0OZFCroaY$1XiNWuFawsj
zQ1(L^4kr>A?4~|PCOsf&Ert06<K4ZuByr(5rpsydQpF{j1@hZGeOr>R7G08mT=y3%
zA6ZRy$c_1WA0PqwACmKw^cw2X69H{favV~=<bUmOoPTX6pPYWG(aRgIhPejVY|9#b
zlGRc1E$31Eqs<amso@3!Ii@;LU7}Ec>WAesRD2N-L=G{hFv?;+ZwZG!bTBnG{sjib
zP>rRQeA}A^`(L3i|3p&^d35Nwg+Y<+f-M4TNC?b5fjTMN2Wv#fvHCCK_iP_n<H~+3
zB=7%#g}QZym`1%|2=;O*nmY4ZU4^QELW6Odm4AHg@bJc1C-ov_pB-%ec1t1QNVKs)
zEOO5m4~WAPsM5@CU<HpjbUX<)1%>7Ep_T?p#uic&ii_Q%zq8a)pg08u6?lERzOcnB
zJalRBL#MuOwc$*Pz~2EDPIx;k9)F6!?O1Yv7$y&+F$}VSpo=aTlCHc5kotp|wmwC8
z&PTFFR<U#@XEW}LBOQZA*dM8ur(V45RP!h$NRh_I%$lwx#B5zr5xGfmDE-xypPI3g
z3$-q6a4Q~~b{O=+J_WO4U0vJLgv#S#ZoV5{rhrnJE#gBFI=Bej9;=Qr_RuE|R^b=?
zKs(a#Op>VH51JCJ3oufdO9JH>npa6NwqoNVg(SXu-eZ8>Pe3PVBC<FK^$R@~jHj_k
zveekvuPJgR3G?$4ONM)-k^h?a!tYo$>{xKAP1r5cFInsMTQk2oDxz0x@U*P4QAL&`
zt{dX*o`#uR0Sb{q&IgTKE13)O`%=LTu?b^dOj2mI5NYUY<({WH23cE0cQEb>b0rd(
z@JX0hDo>}#cYh|BrlWPsyN3}sk~pUrrVoFeT`lDcH%z&1<_+RupPPiJ=Q9bKR_e&G
zsO?SQRYd5E{vt4~HawcT@W|E)zf;ofN?D3w+Kc`sf&u$yx=|QO_&;d<HS~q6eL`Tc
zNlQy>?_p_e{TG&OnApY0nQ|dXe<l(*o_Pl}RJmOS?gKb?5qa0`!r(eHuY;!5#C!T_
zCeV<{$pLGEO9zjYfl;WmF;~)CmE7@_Ycf|WvtD)9r8Hd#{`v;4crS%NJ+3nwx)mI_
z0JoA7i$@w1uaGT+f+5RB&bH2JLk^|XHMf<``$WCtWKPKT@~Y@?NiQ%$84J(HfBY3~
zrGtx*U}9y}e5Fs=YEY%6SlbIClz>11X`&p7UIUX$`WOGvEWq0XnhAhKIP!Kz55)7r
zWoa|)7YfHA{Gy?>RATt#|DrJ6(C&=<VOs_DRk!Ki9l^e>wwaq0$U4vs5_?M76(4Hw
zrUV9I?Vr}(sMz+WHN!UEGhAbBrem>uqyKUfL+OO|op6!03qGdCVFU(_jKNFTtl>#L
zr`5ornPZ-Qtmy4_?Q-~(VQ&|f35RMTQ$6cBErgz}P1YyLF@dKi6sK_ix3J8Sqy2ay
zf4L!d7S}db&4cQa?gdKrpRhB)4d{n1<?cXK1q{-uJE4;?ZoU-X?5W&sUlX)s%d{7R
zIrYek<fO<z29N}pGHiM}JC~(sGh&0vCf7=bOj=y{Os7b*WTg6vhxPCzY$6bdPDlt(
zG5F8J{a5=0gK497ZU+dQqe!n6Yre81Z`bSCN1);cK!Yk0^4HhQ4etO9EIqDNG6Eq^
zkgC^VQE0XtG-1zr@e=5~H}<=KJD0Ue`n!9iL4D|d|DXOpi&=GqE@>DEVC`GjVrc31
z_Wg+A-VMK1-T2>_=d>;5=!(l20xASIqNf(meu}Ct`#i}*D2p_oawH0c=;Gen{YJbv
zSS-@%6gezJJ&1|00q*JO7~N$Z?HCIrnOB`Kfq|hioXk-M+8K8`2a@2gS`GRSmS`l1
zv!AH75lTc~>-z0%jP_@Z#Ze(}4{sk}V6XS;`c$kJ^NeBpkc@umNg6l^pZxmu3!U2I
zyH|%y0T&k1bh0h)jK2U-w4y?qeH;oc!cUu#P&1MWeWT7pERVf$Rz!gQ<?K{9GP$IF
zQ7!treUO|rCMxDTHuT|ARkYNE!H#CeKb})ioL#1oRWt?SMmK<W@Z*L1)*i+BC9J$k
zQ1(Wov)&+gnU$k&&@8Zv*k8#)>Ik>_FqJT>&Rl`umO=KiA$^y-;`Cl0YP&L6R=$~g
z&;n}a*TPR2Q)*bda3(lbgXTR*aiv6gLZv0s7$q)043P7ujaq!MCVq}YMBabbZ&|Ou
z_UjkpfoK^!M9+>JOuzZ}R84iCp@EwF=T9fqFJ0z`+x`fnXUWFa3O%+rNV#joZS`@|
zsUH|LIa5D}2Dyy#9w(<Gc;<7|UHe2St+UOo9Or6>gV|W_=2F6<oOsL6Kif@_in&rd
z;p#6Es=+s*BNZ^|B_lV~-t9X(?h`v)$bycDFpK#SObT`0t3*@mo6Bi<mtzo3@z2PA
zzgNG2_^#E}xls#ZqGORp`+QlFU##bzgnVILm8z-skMPN?dRcQ<=%?GFoD*<*gcT;R
z7>j|b(US^^Wzxs3{(>|f8ghcN&%8EG2D;Go2MpPF(SG5;&*P;d&AyFwr#pdqZFWLw
zp0ZUCc29br^T(@*gc`EWHA0Yd)`!Utg4siWB^6p`pZz<9wX=Vu1fD!oyANt?kTf@-
z{<W5-O`7%L_NNrnB9Q(C(9hTdB}97gHZiWtZVk_14n94lDNg<!(${Hy^Pt~S+@%Nb
zFbg}ab%vhqWOK93GFrF~(skn;iDE1JJt|<rLgo!b6lhqxkq5)|TapC(&|ioanq!Kl
z^u|7u!S+TbDzml-gnbTQ+ct84MLl`>vQtn0^rAZ<Dbb-oDUtWonl`U$@kQ0~Ln4uH
zqv9(OFk4sI6sM>E@H?T`9Jm4);Nf;D@H+JUU3<R(#EhKMM*WzWxSCCTG3xh=HPk^U
z6x_X`yhw&xNWEn)N|k@Ivo7iDeU=Xz=amNLAM_GV1tbtiv>*%ylLT#dSOrW%({WHG
zBXyOP?I1_!2!icZom*ZcdytHI*t!y0eAAOL%hqSq)s2n5kjev1LH33FG7lZ<H)L@H
zmJ*2N&)^wrY7*n+m7Qe5_T~d!Y-eTzj$Q!@zrCUhp%eA~F|gCQC7i_fvV{}yM?pb8
zzUNEBcY=&y`Z160(tz^@Z8RA^1RA5jm$-KZ6$$bT04$?sSAW*r7ZnAlfieheAOZj2
zR_-z7Qm?m0kSSoojbDbkIs4<tk}J#zJ_<+@xRdJABj867W_jW|@<Pf2tpRnyE*o#&
zxz3cP9h|XuBwWp^VNP*Y@XtYTu+1_F+fAio9yRLBEq7dKjGkd1&9uJD127o6OgO;F
zhNTVnu3X6Sc-cbl!364#fM)7*!)B1ulxo$xE+vew@FppNe>5pjUnI~tF)3n(qB(i7
z2Ej__=)**<YtQ@^#OCCvYv*eimr}?Z@TK&!ZrJ@zld@3C9gOh+i0j{>qQuxQw>sck
z^lh2Sj}>_9;h1ub9v`2xZa#cn+My!APJRywH=^l$RFQ2LgcW#ZIRri*@+2vdp0J|X
z8%tU5@8MY^KgP$e5bVR5eQOo`)lG3lBrFrC31y?Usu$O5H87Ut?LUKFg`J%{ycEJ?
zK{5fM)Y-ZmFu#3BDV*NNcmJ6jch!VF9WCwT#KfIz;gCUOEj+h(1Ydk$fD|R{z7Ap_
zeMk+aj_vsk5y&5>pi`dBV%B(xo|-cvVHpc0=le$8#H3aj5lW#Up3D1%0@dHg(X^Ec
zQcoyYff7x;|3uB<j@`x`uFFct*bBi?V_9f-7ARGjla#{3XxyNmdy|@~_w?y3KL-Ut
z?8(QH&YN*}*x5rigP^bZFk~%EYe9k5FoBAx9vWwGIV>A3?D7*kufP9?JCrHHpnwIV
z6L2tIJYEsqg-3$$K1@)eRdW0NAggf^4_zBtg%4RjW+P+jio>O$n{C(QAje|_2k7i$
ziWmDaAsADS_&t&B24BdxhDD8Ow|r#js=uvQ`)_9e-$>#yB4worz|G)J_C!^74&uh$
zx5X4vo$7c(!Vo*=Uo;*8NJPE{5S>=s_@3;N{-aX_rfNSXCsmk8{XbXhm(WpCE)HB3
z01yDM9yhnJc1FBZcoX1xZ6#nA3yp1qKKVsysC549dtAuJCQ{zmdaziGxnhpKzx7MP
z*^9^e3sZD3zULY@V<LV|7yclZ{KW#dJyWicY5i&5^HFp2tgUMY>gC5h93tA|X&kYI
zBSA2b@ovn<x`C!x77PgjY_2<?&KSea!O>w{3Ua<)bIKr?Qd|o+b8*oK5g~Ef#%aWa
z`Usi~FRwbVxBE|Bf!?Vx-O=ow%+?6xG66vd7df%~Waj^PvO`NKbv~i0QX_*Qdq`^=
zFtNQJy$Ts$fv)=L)2D#Z=$9<H0A^uSGIIOJR_CON;-OYk8G9XU-&MR{%%44b=f1}+
z+Yg@NvDx3Go=irLdFf#{)6vA<`BJuwL+Aai>_xPVW1gY`!jTB|4ZLW^tm+z<-Ms29
zjDV96yn5GEm6R$93ZBnErAE%-LLsX<ak3sPw(NLQvWJr1Y>rT5PdKZrxedYE7R^qV
zT&2`_)|?>z{$dVEgo$C=#mL#{+Yn~xMpqlMv9a+djA*TCiq}FMSRi#}ngyxw0KR|Z
zuZC0=#prVIaiL!W^-|wYFsBHUl!oosCu}Ted#dwW2?{hmxan2Ay@%4$(xUf2fE`RO
z5sk3QvBI5As5hd(6q^uUW&h<3Onrce95Tt2F$UA%kZ1NIGOB=0kiVcEDFF(%=XOYd
z&xfJEkj6R7mT@Xc=EF_3<BkZo(YZo>W3ke31z+Co^>=$(hMqD==K*fz*3rqG6^m@K
zsAM+ZioqkPC9I+=^=oMoBXA73o7OLHZR7gC#VFNUhP<KQ=t%Jd=BdzQbyce6-0tDw
zYyI7c3OTUycl6Utw^4lk?VOz#tDe8iPJMZ$GRq@vYGQ&J`xq&;!APsf<{cpICE<Fa
z@#m2O`Z<H?f{Qp2RKXT75J27Ig2bP9jVi!YWm4Xz;Mt0b(PC$j-Z%K@nDOkrXCl;w
znLPiLmt&$-a<{$_Z$r0r;Oz@{oq9enaRBQlTUl^>n}_}-vlMc3KKT79?_YUwqrZhe
z^7f`>^PbeLc)?z-CwFghD~E?QGTtR=xJJ5$8Q1-~+mx4&FT6byNNzp9VYNx=C5<<S
z+(9)q6(b4}fXsmo-)&PpKGBIQf0;MIOokN@Ojz4}wV;@ueK_N#!G0}pFZ8`3`!$>4
z$?MeA+!O@a0Xr{ohye24-mX|j;)dKw_B#{}K#t`Q@Np31of;c=7>SaqDkSUFLtCq%
z1`N*ml}BESmc#VlmH%!Qh$Lv^yknxL>t>2O86lTG#0(tO*N3TP&qvNVvy81@ahZ9d
zf7|K1JTNcs`+i$=v23n>X+%LfSfIuhRbbBVoUBpsIKfOlU^mox<`WVKk2qlDAdz4i
z#R7RJ_{<z#6A@*g2=dsR!@GIeu2c(lvf9GNVchNnAp1)bt2D2;#kFBTxw+YVsfibq
zO&f)PFG0}^CE1`$ktVp7!py|D6sW`b8af<_U^R{o-UGFMj1iwffD02$Xl7poWjcZ4
z6WPd9$$iF84c-f&EGl#X-wL>h2i~41+H+5vBps<?W&ymsyi@<bG}b<s{#H&Fo7ZUZ
z9knt3?dpEWk+`%Yh>G@)vJj!pwg4)c5=(1NPfy4{`#@C8y@7<&Ti^x;u{qbn{~DL#
zqTqbipopALGpuTX;+%)!7J;k_(?Np64CkT-`oj*PK}+v@j(>I9%YXg=AhTFTJHgad
zAO&?&B>L*&n$68QAJjZ178exPSh$QWwRM5}gskcm>)le*k7S8AC(6;8%+6sjEDAD0
zHQVy0R~v9<Uk8U>$qfF$6PvjoE$ytEI!>$C-T@RW3ai@&;x0_vr%g2H2shHjad#)D
z4QidWu7XI9YeC4>T~X{bq59G>>^L!5bd!T6p)cWh8R|1<T(pCNmlsT1zr$Smo7l^@
zcrP7Zd?px2u$4+5-QOaJI5msjeai)NV!)miQE2R$K2<Jv5R|&i=mFt)eE5UtN6q{Q
z9yx+zTs<>=Pec7Kb!?YmGg0w%PEK!L6Er4DJV%Bmf)Hol`R#nr@)AcH<Lm4OqyxX*
z35gz|$-nTEuf^xeXZbv)7a0<anC&P_J*~}ukJAS@UBp;c$_@HW$eAcf70k&&gJL1V
z2m`vxtKWg1goS?m=#kt@AF>&7{lOw3jy0RHT0LJ^SC{PZt|a=m!$$B32$b-zYY@j_
z$&HVKju6H_Ig-kp=YLE*7lGa=h^5ikI4n8m*f=*;pPTUc!%dHS0p08Php=4>2T|Cx
z2Q3QJi$MF-egzqw=bwO}{yoqDqc>2+C77Y5%POj>={q6+{yFiS&G{P+s9<m;;we4I
zjn+~Cvl@`~;g|n?lqKg!F=qp0T%eeux=hR3+tFEa!#Do>9S$w$%?AqAZ|P!Tfdgh0
zwJwXXpg$2?B54=~oD(Lb?|w6cWda5Ra>SHg7$-7%&^*r+J6@MU({*WSXhfDuLyWX{
zXJ3h7gS_n^BO{}nmJodw>($X40u0~5U!praG!!qm)v4;GgkbZQFbE)&0iFd&1k|cv
zSx`<(c7EQwto*Q#3;I@&@oH%2Q4#0kF*nPa+S=N>x=ScJ@Aa0&h@zNRVyAD+MuAEU
zui09bzAccJ0Blq~YF+tx3vwVpVGQ=!mX>lXLeKk<@+X&lv~xSUy5wqfc~g1;`nC|c
zHun@J<Xxf`+EoyvW_&Nf3<fE%{z3HY&BUHQ#5&;LOO@MEbKfvJ&cw?Lgut-3m($3e
zH?(9jApf|svM@buc#c2wG$sM&c;L-`&Fl2|psfunxlN=6Q~2gUUA#hOhh>g>iFytH
zyGKp8Funi-e5HCybm<|zF+h1eYgS<I2I=T5aiBeiu?^ghxnc-!HZxb!a>L%_jk{kx
z+0n<c`QZhkYjA@<>Z!iI=(8)#Wca|B1BwV8R6?vHikuX>Fl5iE2cTra9vGtRz>G`n
zH3x94QU|yO8Ep6k-jF6`<SRyTFxw`Ptz+YDa3lV>+4RsMSoq1dMg6^idV99`&x(m0
zK8K4)u5ceH`rtxtUQo!{EDZKTKNTe`md#E2*@KG^ZvyrSp7gAWf~hRSbxJHLsi{{d
z?)GhEs9E~>q#C^ec|Bx(W%avjN{PJQn(jHYINF#Ro$M~(`6HDhxaSa;FQ)kZwBy|g
zopMAY*QEPgj-TA<ReYfLw>kYR%^O8Fg!`P{YTOdZ*e$^>nb$)Dpj{rQ%lji2!2k;~
zZHE3H9@e(bxCbtoI8pB{^xsIE2WSH{$D`DFs`aZdj|Gy|>_^Bu(yMX)trRrIY_ZgH
zSdzYWdSaOU>^klAStv8eg_;cBRCYZ*{(zfwOc-ne%bYYe{(3iSI`Ue9xoJvFNazw9
z-MXi^`qcTf{#Lw#dlNvcP%y8p4RQnbA#^hkn0p)?;@|n8CpRu1g+~o+bb895mCEV|
zH)Sw01PajWT~u?~@;PrP+bS77{LUJ(KJjgWw3*AX-E$Ll&oK;*jDSADJ;f5k1vW{@
zTK$$vj|W6Q!R;h~^4_bPf~BhWGOdUwS*SE>I`#P^ylbu((CtWBXwgcw3P+aU@ZgL$
zOu}XnmU-A{^Irq)loI{DR9MUCaR&&f)=Mt^pHjS)U+v+!!#vtg)03}K;Rk;H&4nVk
zd>L_-D+e(eI$B2sI8g~^ulk#OKJVRj@gyZIJ-b6K^4*hM>lWGB-@(Xq{WtNmx2OfK
zLb^II;OOJJsfzZzMVp}J^~Hgy7yy|N%j2e#M%_8l=p&TagLlS!%^U>ETfKZmnl@7_
z_mjRnG4N^UjIiVcu*cFeBzyHK0`-X`e8XQGAE(yHcOG*jh7#wGuh{cl1vN0mI*jdl
zu^~1$yUTybnt52LZwsh`EG!MEnm}nrqf)J6y}Dsrl|4a4#=|AD^95V>n;#MhX6-YM
z`!MRHCWN$+mV?2|K&b-DpXq78!cd^I3|Z3v6zyhl@(Uqvn)U#(r9#0F^bz9p(S7sa
z7kBJ?_WASY$w^wP4o;<D!H#ZIXA28fD83xxTMR*yZn<ky8GyT}$P_Epa|1G!7@eTh
zls0Rv-6Qb<V<)h~Xz(0@xSZAMSEDxXQ|}fK@foV_T}1WB95(%&6mbcN^iJ+Q@#Q?8
zLWmlE2~c$j7<qZ6P;^8Z<oqjr+B}_}h9P78|4g^9;9{ZPjRwKc9!4K+_w2c;z$}uD
zY;0-@@-{(jiIx#BTpBR*%B<+*jL^6R;%SI$x)l$79HcZujZ|H2f##Hv%RB_I_&lqc
zx2_QUkJvo=Tl}C!2NkQvEx5lR*rC7w<x5QTRy4a2{X9(!Q6aF3l%3r==<2(uJUHX{
zQH%Y(dOmm=>k?;=Xp{CE<f3mme{O&5<E3^$03b83!|5lSKd?xxRicEM_BsEc@$uao
zGH<}{3Zlb*{LtzthqS=ML&THr<xjA58yh1%RV?ha<Q|I*2EdW{)!Uxm*4tf`rqDBb
z6#cAz#YRB@+Bm+Zsi|dG`8~o?(bc6j;i-Zx?-HrQbpM#MF9Hi{Z+(<Yw&>~n-eO5d
zIy*SPT^jS8Du)g81#or9Wb_)nM;BdU`9P&KwRZNmpacwM%Mpvd<RvFq1j<Lfcm6(}
zu>v6k++u{mtxyT7v4L~JcE4zh9F(jmcL+mVcB^x&9bY+xc&fiUp6u5^28Nn;5rh*8
zz8&V}nco<Un~4YrTqZZcFB!;}bBcFrg`6+uYur-89aV&n1Z>-(u$fO-__&?i;2U=;
z0!ymhqP~T^H1c1ME^e`n!Ch{uUa*k*^T!;9x-f3yQ_OJ}Zmg)lLXpXM2-C)OOFx8=
zC)j4GJMC<3!A8csiMSMusdA>)Ml89rBeozV3UZ?8iGPfZ#de!Y2kM)0%l<!Hg<mrs
z$H@MM&|3ZAXe&&Ny29<>j^A1AdjQQe5F!?QkJ1`e7bh8pmwZ+pe>f9OT{Mvk^gJQ@
z6*{%Ei#+PYRWNG0P<Z}wu>@-J#f;DD`O<}L&=$97F~d=zsZCACvGWHF4I+%M>7iKs
z`SY!1N$31lV5^a{c8t}S8RAt|Ba60!zH-^z#pU&TD+ZcWX?#usu$rNoeQue|LN2f!
z{OPwdWUOO~Llg6plI+(!ELGiQGyGBW9`Oo1+}z(S<uCyShX@8}pYR=b`4l6!CvKjX
z8IkWW&c?u(jw)0-1f9bq)S0s-+*DeMfdbYf@SG%(?+qhMI|ts1ig0yYpFIQaP1xIm
z7`Zk>&iaEtDUp{dWye!8^Ag_kD-SXfz3Mf;v8f2c^B=o3*dR~ox@4sN^Y+O>$b}cc
z-`0VVpm5IG=u+A$jd8=|$XTMo9v_plJ%W9#$Y6J(PO!-%k)Z~y6<-2OV@M$fL;M-S
zB7g!m!?_wrRsHjIpbGYO<gVZU#21+9GMd(gu^~)HfGx8jnwd^QDno}K2pkfV=cZGU
znR!x|-CwrAB4L$$jdET34<^+w*cs)8?g^tT3CKzNm)Lm*p8xNoX%qbaXuyDlB!@^J
z?(ctOvzWBx27)MDO8^Dc*qC%uHig9m1a3N)yU3RZ0}&$jv2DJ_U)$z5%f~3o{5fwn
zpQ8}>wDgpeE2#r}%dV=4=TD@3%&LQ&Nb0x&3#7LL+}r!xnV=OGtd!nib-m+{+5f0Y
z*rOKxdVZ5i>R`d}NF^766P}q$mi!)V2U&G=5wb8q^O)WjKIw44;I^M#YrEzm2-2qp
zeFJT6h-ZZ+4p2#gpad^BC{xMsOW9Ktast*dNKfobwFrChE}#;7uA&nsZFAH~#@5!>
zF#U9jR-^qj{!2U$0B*LAqGxK(SC>)rErR&dqwf(xW8RboeFq3w$Wfnh)N<_MmO9qp
zN-E3=T-@;lI2#sQ!PIL!aCE76R@{imlNNecCl>uL<u&n`O^IeEw61#X*5)b(Ytijh
zB3Qv%QALA0wN*BBuBu7LQ0g(MyiwywgjIr{dbRPNPx4C+S>Jbov!GRrocTxl<XKlL
zaC42fY}M!ON|%gY)2jlVb$EC<P?o_dB(W!E{sChQOoPB~&aLh`-}gez0wop`ZQd9;
zVoW$!<l!<4jJ|2&1n#Jic4GkIz7dN}9Sf;eY{M)%zhb0!=T%zVz=Qv|7d``}0T^1$
zzn2j}oXXZsc?yTDpJEdz{<lynb-<@&ODFkc;;GQ5$Zy;eWfaaOE{iv;7gv|iU(_M^
zN5-CO@~+ul(s5fP9z0Ax(FFPTVD<T`6A~wZ;8N$UqN0Ob&#CGoX270Tlgn-pFL`=v
zqPiOJM2AVpng-iD*y9lOd`D{7JiFo?XFgq&_^AT-&B`vcIC4y&p+DgOc>Lgag<dGP
z-S{Erbjs{J9;~WxMonM6%48$;MKdeU?IAt~Zu^YfSuA;U-c?aaV4m^-R|g2Lz<&Oy
zSpY0kG+JYJpY+uj1GQdOKXnPll=2%(J|y?K)zwa8NOH+hx5^TCp@WNg?}AUC{@yls
zpU6aFd2iZRhIJGfg<09^#M(Dk78qF2f0)%N+Wk9J@aa$c!UAk!o>@;D@3R&)MXHZm
z07wLWRsRZkEJP9<V&s@$!-W?LWZFPW1(kE>?2z@&OlvTl2QtL>BZNmmny_X8Xn;+*
zNH8a799<0LTVTnvvjZM0j2qP0x~>8!J8KRM1V(1bZO@bM-(khHAy_1@B8p19zj$8x
z=!-z*;{%ZmOkp5~0+ZW6?1T37gh}Q$|2RD8B@q$+q!&H6^);HbcaFWqZ>{ga_jG){
zI%ITd2Y=$*l8@duS9LyHF?y6=xJcT#8Os^8vy^9m#>iyg06_<Ap+X$in*Y6XM<pfQ
z1;4D5l-qjd=yIUf-Yc}&04skdmYb1JFpKNXo~DIyKg<nZCuXsQ{ghJ;ycL20zE;At
z&uet=%+7a9-Hwq%*KYL)-Vg2YVu!EU=NHanM;!CC(j3GlTYX}z)^312TiU}C2a@>U
zp?`>(ef_hRFT$zPEPx>M)6-M%N}r$e$oTViT1Z5nuKysZ0m_2AI$xS-+$X1!4@Zg`
z?66&vx>dO^!XuNXz-we`ItRs4@UH8?;W%5RPgOM1mIR;-@C>QlzLqVcIqwBS$G%N@
z5kK*qz{1zL#<qNP(tAs}pFx7}MZm`^Fgk&3+8lK#Oqsd>G~?^Qp$UUUX=w+DN!P-=
zP0KtR;HH(!_rOpuk^-=mU<Pi7iRU9C76a#7QoZN4c0Ila4*+A?V``8<XG{4)=~L6V
zjP91ODptbbV1W=r$mG>dC%~CgpT@e|qUbY<)-zvq%{Yb}|7|!7`P#R<<76NjzZsn1
z>(7O4IzT668(eMkE=9=~P&LnIic2&oX~DAz0P(>)pOeCSe|<)N!Qr`w&zu(6nnF4k
zDj_iu_SabkuoytVV}Mq{<af=LKwGwK_545U#4~#jw&mhjYcy_=)C7xuRw8|N0(G;?
zj3d8v3;moCFu)kQQBkvLqCN_?izd%Q)8)@smd(E}X!l1_LG&)i1R@BKK=HnSyOxu}
zRND<0WuQfH;f8`NheG>a`W3L=1H%i-AUq_v73L`BHg-x-enZ)bf@Hf-k#r@m7LLE@
zBYQhXH>But@7umo{M$6U^}~~$?uRmq7zvr*ZZfp1v6*=sm;K^~r)2Y2?wkIHT;~-V
z4a{7U-%Di>cNcEJl&*!0HQ@hu#D&0a66y`c3z;p@tdYv7roAP6EIBsgTm!(XK-3LA
zZi5#4YEXW|L>jCdq056pk5ef_?Z@b7hE+1#40PZ?!-fsc=g`OEe{SRErkE0;|By|Q
zysRR5)mh%q0P-~`t@<r(zr;bkn395|pn&mrl;`i}oOUfq$ql@W00)8#laK(gC)|l$
zkEjTPp8(3cw76(B1AIFXetvq<Oxc%_vmZeUt?whv%IDQEFy+&0&2!fC=BygCuWoHY
zFK)xZ%d2aI*=4p<^gA8|Zh+yEY`}5$PW40UBU`94fOPByhL=c<E->vfJ$=z4c=JEs
zu}!MS!K(sNt2(xhCnwVu27%cM<0ujl9n#P?>&n=lu3%FTEEo#3ObUXw`4Wv=DL2-E
z)40CAzPbu~Fd1_iP{a#{fXRS`0?-RaL-4DXTv2zxgYkbQ5b!+0$oSzf!pIhuoR+4q
zqhn0*2WG6auYwC%K<>{;!E`ej#y27|?y*C+ddgr{UNQm?{f%3Fe6wI(mzxU&LV+lF
zNX928A(mRFC?PWsZFj+h^}HB5b*0zn1wkI}p=}p~NY+zwdy9I_Bf4E3;p@HiJnzm1
zN99MmEnngxB+Dz@7onEt8$Oa(G=go}0}!~$k4^vt3B}0aDqI2l9#)O-pb<KsCWSBm
zj{nb5lH_I-Zg8sv!G7A@5|vYpZkrzjlit}&gW<^Xxq_%vi|NSoU-PWC<D-*PU?i7}
z2$A(KyhJbD2*7;$_W<HUq5fBR1?E-|**-Q#)#=n?7@ciK3ZvCepZw)$7u5nbmtm-P
zE&&3G?~RBCp7VrWPghrP%z6w>tPw0x-M}ZwTQb=rp*dRKd7!!nN}8!3(|gBKgH*i^
zVqR(HTI$Tjlz7ck&X<wX5)<Fwp?0!RkV05yJ`_DlJ=kEanzHn^f{a(ej*cAg@RvlQ
z3r8+Bx}rSjm#bwD_*pegK*KR49Hc*vR!uac<w-Q5e(R5Z!sa3@07n&kT090y?oD?X
zW--PzrLF2P##CI_SN&gz0%C9w$qWD&_4W1bJ-)B)U8$A_!PVm!fXkspzkYp*eQS~w
zrlUMO5VG>5NE3jg%qGuSJy4XvuE|CQ<N%2D1)l4?M~yBF-eGG15l_;P_p)f_Y-&TJ
zz-OP6A4W$!;fr(ujAvYFKl3%Hh!%4O=ya6MBKyj}8H^$|`oO@4O*%;_mUJ{TOK<pv
znAJ_&*JH9#gm_1E6ckYM0f=~BReU7}RP3-G%)utO0RaTn-o!(&E#lidtA0Mep&Wm3
z+MK@JJIrENTk>M{PPp1XVO^s41^>3|b<|`ra!vb<^m`@sFTK-Oqa-J8Cn?T!Hj@5>
zFk^DCMSP@J3sZY1o>7r~l*}PN&eZGciwHme+IttzYT&){pzkEIAJ*sbRbXl>`{v1?
zQT&sdul^S=E&Tc`YHDsiFtnz3urPxzG?v046oc~og^_wyjjkU(6~nl@fGl~4{;51H
z!|uPlF#oJ{>yqxxgX6^Si+A&p!Ml{oJm$~phOTwb=Y&WPnhwk?xjv6eO`RAR_{6E?
z;c*Z_d0#g%5GW}&Ha1)Hj4gV94ChLSOww8m-?hqPi}=!_P^b=0rgaptpLaf!0aY^P
z2laezxMUd_fHP1Fxv@j;3CQzpdExR5Sqo4{DY0brL>zJcR5UU{^6e<7Yp>V080Hlf
zF$Xx-{FEpZBZ5yfp*{7+^84qZmX>9y)`I6{*Yj5ezyd01!Zmeu<LOwwT8W=J)W<pb
zi=UbhOAZVUVvf}iqYnlTeK5Qqbywlg0a>8BetpNY0a{G_Krku4!m0ujgyq%M-Fe2=
z284_^VJwJ2K}m_ayI~dnGJS+2=k+W8J&+kKu&$o(!bEQM(h$~57%UUlsY-R-BN!YS
zlEH;k`(-yG;%h)dgg%6VfSQo?-o1zP!~uegxLT^JL=zJ9f`Xn{yEuXGpu>^O_~!cJ
zDlYl{?3IqI+s)v_7(kKO6A7R;5*GGJ^CJA-Y0O3jZ1^|4a4{DkrNO9^KcrU<m*E2C
z%(RUZ`saQU5*LV2t?fT$Dl9^ZW4CS2wA&bB-7~*1JIP1~2#lfMgL)wj?`fI;CpJh*
z`l&2M=NaP1@SfM$`bJTwoa8-uZSjx$%vk6*PSYK)5TFnV^F@XyZX}OVN(5DetOR5a
z|H|){goyL1$6%n}6>$HH>_tDrvK0B>4=_#F3L9nBYvTCrTB)y~|Frpy9~LdjU%otX
z`#HAS)DKDPCW<Oc^q-(Qh|w2;TZBjY3FW#r&3s6y8Aay#M^ZeEZpIY<R&mUqH%NgD
zICFQ!od1mo5xHp$dj4hefv0?3&vqWSSrM6Yp8tsn5ve;0mryFJ^SzVrGp!T7Jwo))
z4S!0azpK&G^?}9VN&5ElPv?|nfRi`vs3qkgh<;P^n_%=IO3}lcBt4>kp?`*=$U+~2
zqjXyU2u#sG@ViHZCyc@h=6->D1IMG7KPrSwBJ%ul7=A&Ybcs(4A8x)B=5e!D1JgS6
zCxoAW38r2aZ>MGpyx&Y2?*H@k#6x(8{@a+~{9Wtq(mp~AZ&4~Uj@<!5cuzQO@11{0
z#8o+D8iW3gF-y$%1K*K9(c4c$7;PwK|NRa3_cP#Yb|lZ$NK?pLH_@xql~@ZiB)OA^
za!};|gFXn<Cu;PGY5ZZuq^?Xw6jvM`;8aWZdtGi>BbEzUFBpa!4TXx3g7a#q74n*C
zOF!1>djAs=lk25BB%c^01!7nbQW@TvP)Qm(^gG2rp5I_0cUj=ePU)YYWElIm{rev%
zdV+U2wDfPcg;J;fZYM^Z+|exXYn8u-m)m2;C;Qpst$3CFr8IEA9v;?ARk7erS#k4{
zmJR?ix*QYS=};Il;NAQPn<5PI@EGD)lojAPL0tLKznQZbt{N^;@3p0H*UQL|+Z*cD
z@K5ddSOu|H(p8jWH{EYsQ-vbZ!^1;W^_5LVH;`zR0-}S0umdGy;^OY}4!EN~#jhWQ
zCoE3!W3*WBVa|rqL_T54%0X?=nZ4an1${NYy6A-s%gQmJy#WPV68BXnjH-~(hzJ-z
z159el1tCm7R@@r&&jwY+l+#5;7#NJOBrl*pm|_dAV}#q?AwH2OPsI!?&BI+MYr>J+
zT{R*so?5eu%gd(adw5q^OwG*jur7??1Nj};t{*;x8NIo0@s*66{J{ev2f5bZzWKX4
z6(NnduI%Vf<(xDkL4^DsuFUR6i5<i=JTed7dXo6z>-C6Mav&6;v!Qw?bCpvPz(P5p
z;ff^I1qn^f;qUp6@EBlE2B~~2c^@u#qk|zLpa0<b1Bt4KM}pJS^@r{40-lI7R>zQ*
znSEum%KLuLE<DoHv*S&{V*vXXd3kwJYw!Z@Y)9HjB*8qhpFd*i=01$+A+uqO0)23(
zAl#Rd7_udae^WeHqQv`d@llBNyB*&W-7n~bHSpUvaa?fPhPgZDFu2CR`3s0XJ@o~Q
z$nE=#<M)w~>Z&SmdJ1lZtVw{BpgikoP@M%X8Q{!N8#Fbg6O)pVWT+$gKs*s%qT_e;
zGkODgKppMvz}S5J<O!H4{h6PK{h}&LzoaBTe{ppcEc1;^`QfVkY72!Eus+io5DnQ0
zpG~<WLk1U*0nzQ^(ac!;``6ma3PeoUFI#3Dzsc-F*>;7Ll%HXY{QP^JZ&g{ZQ;poJ
zY;c|<tHE|-Gjh{bwTigiU0rZCF^8X^MSGu?GC*Gpxvl*gZ~ua?+R~DZrDYeO+TGnw
z1yhrgP_n43sQ^<BTn)Z`i;jznOG;YY;Q+DBd|&z=I~;M1_W@Tc=W-4(F&;e{1j{KL
zZ#K9gz&)3TkFW09fRa9xYe-ug+#WZAap2herqLHqA%1(?b$t1y&|XMm?#OD$9FE7B
z+mk2cKA<xXa;5>dDX>C@paWcaIXSg_jSDZ{!8pCNwBNiO&e|18-DmqC4};?-k>saV
zQBVj|&!2PT7fdbA&+i6%c}>l*_FoV}UtI?VWpJi|e+YhM%%pgA(fTkTfhrD{0n{VR
z%*?lLeYWJTs;JoBLHp7k9N<Fb$L$3S3@a-u7w;F)!dP2#Q4m0$9Lxa0fJDlQ`TYFy
z!D=I$f19IW@I~1@rN~j?$A3hveL{X%&85xGV5-ASSU$>6{k6hsko3wGE5r!6<2psB
zi@E8Chlj=b4aslcD&dsgCwQc-Esx%1Wn^Sv)2@IKjkS)>7=%}V@(nbg@OPo&8`pcS
z@EbPyp9!N`PXNINQJ71B7=!7j0K8n)^wouh^(_*(c#AUjwTmnxOBSZ4)FPhWExFy?
z+!$Y+pK|2ib6%-a`2ThGCD2s9ZTB=eMVvCE0i8;d43(jbbyA5mC_`pxFhw$tk4h*-
zlE|z{B_zW!lOamxIGG}ndB~Xg+sE%)|Mwf$`quwh?|PT_?QqU{p8LM;>)Lx?`&z0w
zVx3th9mJ-&WGG`qR&E*%l#UFEO+}cLe=c;B;xpx?rH084aAE_YF@-iiwn8lQbzEFz
zD(3UeOiz<27cN||m^ihI>mKrVh_-<2e)@EbcCZ&+?L0i0caV%IC!+c;xiaKQ=J%I!
z8X<mxx$)!E9J*cGbBTBRmMvT4<XQ%%UF8j4s$p@Vg}n{a6>%-OrQXeaY}Z~ZC4owk
z<*-6a{|ig+t!*#IwXnAasa=^bHmSTh4|%~_Dj0Orm2MxqM&;-aw|c&K^$IPh^y$dp
z8%JlpAKh}+)it|`lBBh1V;{YMna~w+_=sF`ok+Nt(XFnt^t;oGZ+F;}-_hgj?z1~L
zed#%IM5aaPr$VA>l9uecBoINs^)2rJo{ETwND#DDIfwuspp^aNs7Nq&Ev}k{mwF9q
zZ@%@N>W2_{d!U_zgUf`YOOAXF8068+f^al(o3SKck8ZFaaAPt103||dsvTK#@x6J;
zYw_N_UP?n*Z>F}M+F8nVeA!;!d#ZPl#a-=^^e-tZExj((-`8iO(q{S!0-@a55fy>S
zi@cvKX~@e%Zx{$Q#d>FNA5Pf%>%y3sKF8F~$uEXt-|Fh{L29oJ&nm28ypQ4ahxr_Y
zjBl};rY60q$sArQGc!)IkxJX{VG<*Dj<4@V^^K~0vW``jG|UQPvHk-cu(q}~X0PZc
z(zN!l6+*+D+2P>qEFGoL>%PQP<7sfPly$qJ_C;C~N?)afB+kFXPj~w5e5kzp{_UsA
zZaOE$t$5v2zVGVBt;0T=4=386Y9@2`uqYmH<nVF#kv(UUKs&K4cmK5o1%HIf4<8<#
z3hBCj;OyD6zz%l@!3PXJ*wrWI7jM)@iQbynBuGb~yZcsOpZS{&yp>t%#P~QO42tER
z+SuBcX~)$F#--q0h{as!{2a#ZnfW2+6h6L|diY<SJBMUn?$Ivm?@rPZPNtXbqOblr
z5WeHwBd@KIuV3$k+9F9yTSq6y@0o*fa>MtCE-g^rkUetU3$x7p90mBj27zC|ydcTA
ze=a*zKM_G*-+y9;HbUUO!liO01>JAJ&>`6zFT3?|bpd1JG4Y-W5gfL&BkWEq%-I6)
z%e3oTp;>MrIn<aMmy!a%ZwEwqgt45iNSEx2c)rXpUz7ncl7P(M^uqmEvt|v-iA(!Z
z)6!t4mYbVvT;f*_+`=d9#`3pFHr~Ef-b`lrQ3&K_iAzagK*1gRvk)t$ryl_OOlv~q
z#%Cp-p5*cEgu)UvcV%v~b+^=?BrB=Za0%X}(&nbr=50L^Ea4o>y$f+TL|gUY>%^Xl
z=*^l(pp<o~?vc`vUwgxz&#_V96ia2TQE#Mxx|b0~QoA%TS1<cvC?>|%(a?HIR3iG-
ztL;=$a2d=ACuVsHqwM+@bG#K4B7ChQI@~{Pw1Qyvux4L>zb?&E@7gtC%C_CR-#%Ns
zva@HOUX2VWV?h-!6tAI@?C|uYhOX|WHGdXVl`!Zde(|jN&|<0JGVv}-FTm(UVxs>m
zCs6az(VHbC>c)2#iota$Skxe`!2(T_D1XlH<ce8kcgR(qAgZ4o9V;fahL-&77ycS?
z!9Ld}TEL}VJfpw-q^Qbs78QgPc!XgjoS7M@MEK*=-!>qQegEElfuEcEC<)zwxGbAW
z*HDq$IkdW6P1UY5lV~}1>dkpcadDve0LUxVTLIelgt#gYo?NSr!lD*`(8}tx5}|Zm
zPLXvkgEWwjqeZ`DbN9G2A+nfskzZeLJZ6b89`j%KjGw($`$`p=!SdzH8QP$V2pO7X
zOK51QGXS%JCK`*37IyJ*11@gvjLV3v?B4sa4@2t#>{J?kGJ0SX0<+tFY=xaS67%Mr
zY5fsos!NOMLVPXJlkGfiwzGiE0|3Zf#X@nlLMc1NB6{FxprX6*E`=i^GdI*CFPY3#
zPHiWXUY&|io=FfA5;`4z;AaEIfHc9J?Tm)TG18yepZ@;-v7aZ<a?nywGv@t`es|lA
zk`pD{1qJU+Bmlw78sLtl_a*9PP)X=ebgw8@QdNCP$j~%3G&0#=K!gGB$>HSDaQ3G$
z&y{q`kU}3lx?HubnuAdeGaX(hxpEc>5+l|%I!L9P;ZVjeJp6pAQhw&0^2;rr+XMyU
z^|GFnbhs_dP1-tiZ^p^D)f0@6z*d*QNAnKPK6LTa8>zy4vuHaGnG*3CP|QSkw~2PE
zC5>e-lb6w~!7gDRDZW==f`5HDSNp1&htuuBMouazs&Gp0MteZt_SM#-2Z`6`-}WVB
z`Bh90_V!Ny^cFxtej?Z<9eqsD(7`)hXIfAj1LED?bGx{^Ep1Q;Ao+nnCms-jl@&6h
zsXrwGI@;PKidiZVs@I7=m0z8m)zdKZMDApTlith>=EE0HoLK4h9F}G|A{V;uR(72Z
z6~9YH-=rU(j6G(vrlm=ulXUj)pNR?kw2iHkb!e(0F%sjVyPW0a<xrd<H34XF+;7vH
z`1qGwCAyWFKhAw*6m=ffBC=jpIIApw5ZM=E&d3huFU2uD*!DK)@nbUe-o1Oe$ID7e
zO1e5b^|Bq4kONILS#ezaYTKJQbU6i`A9}sbeNyZYSUj(`wY9CflZtcu{BHGQ$Gn<M
z2;brnqz+{2E7h=1?pyPF`}_9@3k!N;#?x_;V6ad(`63PBz@n+^AeWRzq@ho)$wKn3
z;#zL*pooZpE?qZi_NTCX)2~;B@z+4!KyU9oe}4x#9A9-cHTeEWYzu$-`1m|j+J;nq
z<;s<o=Ylyl9mp5^inr)!Z^!0*8y5$4CIn{Kh&m;epT^Kb;XYocS!B%ZoM}QJdRkcY
zBZP$le;%-hR0jq-tL|(O5-KydnOe|Ic%)b$y1kRHdeXwGj=vk3_nd)t$abDNm6Kyb
zk2r0%F<ab}R74oQnDiH_A~`v^Q?bd<pBD}YLD-|KrG?(|q46*-2zwx=)YpgG0PKBn
z^*f6T3zrKIV5f-g+!?*UcjE?P1Fr8!5Qs4TBvN52_jat_99(WQrx)r;CskCS<xN>=
zo;2}IPfthO%FKk;;yM%&^dr!izf_Bmw!24G)6jq*1;_QA|7ylo$@d;sYNJ^+pc%J5
zK~JTmo;_hhM1cIYMseDRY3(aK(g6O|Ti%?9@#P+Z$M@Lt3=MwtQkl@7#Ugnb8M#4c
z0pN4&c~VRa1vOGhQAhOaw~XuJLhzL^VPRpq0U^hi-FCb9V$4A&q*GnPrc!9biQF2s
z-s_;lV7QLjapNkpROGElNcC-6V4^YFRq67C!`xUpqT<n`N03ua5T>uOObN?!iSDk8
z<!2S~pN93XTFlHpXesJi-a7M8Y*ueje^<~K9J43l>e5nE(K4Id@Kx--(l)fRFcgt@
z?~1eV+*L$67RlGRJE-?}m{q-rIPyWgW3V1N9U}J8#@mVNqaE+(>l+szKhWQAJIAvf
z1~Sp*O>a;HWa#P>>gshZEf@xaeo;hvGqc@cIL9XBk*?M_;vmy|809Xgn0FsTk<g@o
z^rB>9oN<qtZmzn$Y`!Yy2sSOP`GyM4<AI|Bx)Rdj^!*Dj1T9VLs_-w>H~-+&L1z|n
z9W)`hC)}z>BZ<rKzbX^rzm+Dx4lAmo)c_avuey41*J#h~G0PLi3|n&(2;$9BrBRxo
zBSTy~B`vLH)&}WoZs18+IvTJ!o6d!XZkV>m9ZE?NcjiNF_U+pxF<z)ju_fdj-a_kv
zZ3ILD$LO}T(g&glwyVZ!2&FJ{h8n!i><-l^RU1c@sSSp_*`xKs!e;eghU}k%;g<r$
z8uBjuDZE$sv#6vbYjwo%<A8vlsSw%x0@nZg_q3A?GzKL<IJz0_2p)D@wr=&kcdvO>
zk0V+XzPfkhciynd8jwuX^zrq5Zrmfs&Bay4{g+t0nYT<*b_`=-Vd6JjV)P-+CC4HA
zr}}g~>|vHl9r;yxr5dXV4;)9W=R;r%2?-+X-kd;s`S{TK^ZePf1c}d2EUS*rACwG<
zIFg|DroE%X^z;o#R;e-3nuan@fW2Nd>}P_>Oo6*wzDV_uos<QvrHL~;U!8nP?ekT9
zHaqL;WrW~?_7zRd;GwVK*G}9#)Yg^IXxTkYfXn)Hf4#+5L<xEMd0-H@VJJmK?rZAl
z@o{q(dzFiwM3Nc(;4`Eb7Q_9rBs!JeHEtn7CAI9zNlp&bTV`~<QevdvUglcP8J2x}
zFcGW!de|G{T&%MnIVM>F>n~j)7hUp`$9bc=tv31=->Ty*STo~~2L6dTRjK=@*b0Hp
z%#N@f?;rZgEETt*=(U_j^_BBGhS|MuTaqScS{Huw*{d?EKNm2qtQ6+!*iaREb$j6F
zD8X0Yufgtv(9x^%GXmYfxZofZ`nBW}@##}hQTp8e*6QlRQqSRygE3oUE-d#-Is|v@
zpx4*K8;f37R~>nJLNloNqILYqq+OYPQe6GJEXNzXl~q)pcPum<fxIq0{sAn0d1gcf
z{7XjfUp)j$xW^VK0+envUJ^3kW(H)ny;5jDTgQ5P#%<D$x44{kfoRqac~Z+g-ZHq6
zk54r(3kKO^b9`eqcfNGr-Io4cpo~#JJ{`qY7>kwKy?b}FK<m%zo-<qTU)9gkz9Y8w
zOfb{^-RNM*XCoim53#Iqt}xiBB^<{2uTQl(4E?|_=3M%`wzcEl+`H|;WvHO&{O8-_
zugdR4HG$7fXzjsCi3pDW9;2@c`wCjE>({UJg!8A9-T3{x>@M9v3`dPnegzQ7#_Qc9
zZKap?Ro=~#&s}d__$upm9|1CkuIJE_>NoxOD)gH!UUEww^rU8Z@{iWA%;}PKvD*L(
z*=FLK#MZ9o<lIUoVd#>6;?*4oii(ILPwA73dqDJ<)e>~K%Fm6}qIjMe%(CjPBT+zd
zVGz_rN#-*_d3%{rnA|OC`~^FUq+LKD0OF&@2a}GaVUY$wGdDMP601Y?*+bwe`_%M3
z$=H$WykkujW{w&IitKt~*MAn>7W|iMRcO=VoqqT46Jy!zSvzU;tn`cCf>3_9O3b0&
znVy+3&8oh<9M(=<K@eeFHNq1^dT6myW_L~M!`b8vJk2h*2Ei&<@!%*d>+a!}!=grF
z)2dJRy@ZGg`M2fqE@vem1rgE`?=&Ny=T)K6LVA5c0Mza<9!dH>DNhoU9-ky8Ys(cO
z^6{$j-WvG1sN>J<P?AS=;g#g-qz799O!6<AFc-9RtF|Fsf@Lp)H3Y#u_ODSA_{j`z
z_O!rko7B|0C^O{cw%%7O`RF~*%$dhCR(A#idsT$tpN7VhY7<#v-W3qigN0XNme07m
zd6~g9#|t)J-uIMPtk+VO%xK;^w0T(ePI(<%yj>oj%Kv@+fW8$NKDn}nZ5lhdruB6q
zRo&ld^m)WpaZ(W_c)fo9aZKpDxt`4n(G}D$rgLOuWYSA>(zI~1+DE4>R?hV6>u7BW
z`0IF*c5cNV-K~|t<K_mae5GQU(AF`oP$CrGl*JbT^q@$jP$<X`D(8M3-{ECsvIM4^
zB@;3SG#b$F^LZQBsy{3y2R>-+8C#cyY^AStTZaZg<n`svll4uO_&bn4-L4MH_bqNO
zDpI6T;Q>(F<4L8I)0Qu}8HI&}6X<G1I8Nv0{)S{dA=pU(E1P5^1Tt|LPhYiFQ>#hM
z^@0)!fUT_a*z1pmVE#~G+BBzXj@^pvF?i3G)av1WmO2}z2*}}7l)NhGuAko(PDkj}
zFI@OsA?APY-Xn(doVc`fCc;5W$&d23dF>6(#?!63w_QYd*=<#BJ%0T749h>&jR!vM
zI&Om7$#Uepn%W1WLa*#>^yt=CKY}=mc)Nyrr6N3oSeOS~U8@>qYGh?Kgg!t|>d)|l
zHn(`u`;c>^a}{M9jJ`__X-G?jy_!aM>vm99DNwVNtUb{Gs$pJn-$)?9y}L@=aNqvX
zxlA7TfwN=!7GL`h@K|)>0mDHnBLf9<!qW_mxYavD+9ZFUSvHxmxG4=cZ8&};dlf6I
z&5awj(k(hxIY(V|g?Y-nR&Rt08`9iO&0Kzb$0{uExeNE36<SQn$kF@A=I#3w>nlIk
zMB2<BXBtq6J-5`AZed~ZLUirgwcV`JJCNSOsRRjj);JNzk}}4$(%zW;==CVp#rOOa
zWBoXvlN#yV`{woQ*Dz;d3|%y@f9C`Xb5Z^5kjIb5O1Jtin>+@myDKI*yXC#E>crOr
z)}Z=m?YWY7IJ&F{OD56yEu-YzCSd<Uy3FRLo0m~{LF3jv6w1w0$*3`&adfNe=xik9
zdNT*5=8oOA?{JCS*S6i>az#tX;dRYjix0`e+Qe0G7{_jyl+`@L9e~FdzE9}O=fMZD
z?)VfvM1eUjn8%fr?SVZd0>oG&hm~X)G2OExQ`uERgGVRw(IcviC))X2o-{)wWP2$W
z>S_88Pw3@Cg#n5XS$4nBdNtF1Qy^F00FWTcR%c(}9(CWoIsS+P?%kYWGiIl+&o2`g
z?jB&Y?`gBWvV(L4BAs~kxWdSxjW&P(!^6t4Y0Js{^J$|3=7hwY_x1DBPc>72-!Za9
zN{SX6>zyE_ucij*^!a&)k5Zu+k1tVdD26_{(D%sJa3$@yaUdM)5FY^D_BAy%Sg@!N
zsXqFNRwgFVLeF66^uhT3!A%S|QIr_qWKNv;jEW1344xaP`mC$~S4Emq%AbJ#QBm>L
z6`9OTpC%)Zg%}S-I1*q4YE$vhzkeMwlTRUMz+|z@!Rbga6UxYd|B;^^jHm!h3o->J
zADO3JFh;1_<k#qweOAlGB;<T(p02SHaV4ef^*;Uo77gXs%ECJ(9p&K%M`2+ExPz_G
z7BH{9y>tBujIS>*w-frO>_|u7Q*9604{#ZBz<+8MRqGBx!HoJSannaAWVGWkctWJN
z;!An?OY_(aRD;h%5jl>KK9`j}=llIJ2mxdLdZ$0_n8S?{LO1GG`;Cl)2M=1(kY)p#
zd>t1DLKyX~ogAQb_1$#<tx1&2moFp1Oiy>r7h7`etI^uAe^wab@n^-#S+QqRG1~E#
zFhGW$aA*h`NdUl53Kq9x2aJu4nP20>2m{o1X6EMb73ldj;p0OJQVs&;kdc5uUzF`M
z3hcbGx*7`vDOQdP=m|5kw^x@4z&R%@Ovk{$DoM+e8WYt-Urkk0Q*(87g;W0V<HrdU
z<aK73dA$ZR;?>@$ojaFviX^&c&x=Y+G2ZSc%)5hjZ{2#^eFz$*u+Y#=>?~PoZ5~U^
zzW(I=_s{QPU0n9pjO2iZOSEKN`mly~{oweZZF|o7-F%<NcyAS>9@)bqO=l`#Efljg
zKZX$<>?t?J0FaJ5yPZG(m^ic<Y#8}H?rEl-If`ke`MEn32-m1F*O%Epxf5L|5G9rn
z&En2>4_|pY+<q}cFp-2^kFfpoBY6e5xqp82WCrK_&t(KA@SmR!@iG|k#p{1~DI@<j
zVw%Z6GK>j46M1AviP(Q+4c|F!kY30l=K3>-)xmM5VBJhlVHl}{+1n`OOvyigi-~Dz
zKH2%7pFPAFH2C7RnQikTv4H<US_&rQpZ~Xj2|VrM^<Q85PyYXZd+k4e;XnOZg>TIH
zAq=)`toC$QS68AAP2^h=(|y)fVqq@kh<9&qFW5nNJGI?pxnW~-gpVu#JH5F%`)Vb|
zo1im^b1kv<#*G{Nj&00A%G|x_OrPEX2eIs{kB3dG29*jzAh{tMrVEG!V8bxm`c#DU
zc%;i9&i<_eJMypN))p3+rRE~!NQMc2yNS1E_wK0EG?24yUq(>`Auu_K8Jxdy=AM@Q
zQv{-c%Ey<lrJ<%a(rBJ|a1%QL70sEQ0P+A<`ENQnQot(8#l__*E$xxigC;Ks^Yl6F
z?CleDXbEX)PsjjC)PY{$($}B=y1x<wEdo|^8fqDm<k)<H3xSXEmR{xYr#<kr{6U0L
zUEO93Ep>o}FZN>zzsr52ka!&$qJkIRVXsPTEde`P9dv|!ekABJ(LdF8o5OzS$DFUh
zY{A7Ow2p7&@fG~WyyU(F!aGDz*p2BLcT~e9>!6P1;sOxKf}teMSh+B@f>su0O!AMc
zm#_(J8z#FV-I&5m71U-_K2m-<EI8PK(scZ(Pc*=&qR-G%ZmJj{&>(1DzXv1)0&_1D
z^QU*fAJF(67M6vy9dn67llJs7F>Nq9p?LT_>8_8DR?@|z$jGl=?W&8@Ji@m*|GPB=
zo)YHuvSU?8U>BLE5do4Cd=8pG(TVfql<^5=Wo?YB_Ai;4pBa*omOdt?@A322GX7id
z0J0OKzspzMY1hn~R)1~G%)1UQU@WjcG5*9xd*k6RH}M@4r@Fu+c<&bT*qfM<@s#`s
z&q^CMN9WYkNEBcKF)LXO_n^8qH#J>zS1}+U07$wy-*klC*|PzCCR4x|v$JKrj8H2b
z->Kp8)AQgk`I1hW1qvOkt5iZeKE`jkq3!{vk!_S0>FM9&Wl<U8;#$Lee;qIb&xdc{
z?)EbJejm%MW{RdCOkd*h@j{wK!-vad-F*fN({xlWk#^!6R#PrDrkGlG0w;zo=@z+F
z<iUseN5uG~ZC?Yxd!l2xYxnMgd_9kV!!A=7jg9+yd;R$3w1IyVcl_CYGGsV8YL0Dc
zY5x0AjHzB{;CdE*$N-CGh8SisX&D(#0HUQMEWbS76}aOxyPXG92U<p0DMdCMByFnq
zS0+yxJafQX|El3l88`H}iD(h%N(=@n6bU^V{7ef?iU`)!4cG;6W78*`dtL%gn!{ud
z_^%0kAHiH$x%s~x^bMV?Ooc@z;%lk0nqe4f=}8X!%ylb->DFqSHVT3iON-RtWZps!
zQej9aI!LOG3thzgj{W;>3cXfiDKl&Z8Jjq{VHV(}F^@?8dui+TqD4MN`Q<(#A>S+P
zSfRu&QBe*%6Q&OKmVEdaWCZ~h8L1>hm<#yUI+Fa*un6S36{tQKUdDrRjmdEZ)YV5%
ztgNhD=Ue>w^JgKAcThR_wOB6v?eJ%;$Pl|<r2({7_pr(?3|q3tjbpelFdkLw!u%xW
z$0cTud_gWt#cPSb)V#xeLqiTiv2H`(S@=J#z-?T%;lXccFv=?`auJTGe60EDoNLw`
zCS_o+I>|<wPr+b`l}eKQu!L70`)`})9)sNM*ROqq!y&1EB&dQF$auG^J1fNat6%EC
zc3<b9Fvu^8&&ti;#aQ@nMZbJm3^H6@y@hb<im>$lQdTyU-oZ{SMX&%oc4Q4-J}*4y
zu8Q%>(&NyGvDFhNQ+fj6#g9n#5I9#-_FwsWT}IsI_g6$6-5h7`D0z>77iV^oC}_8{
z>_i4ZMxMpF_`p`mF1x9<2N{#Cka$=*j0kpe)ffaXxw)AuhMBnUhSCkxI_Ol(PCh<9
z&xbEwZ2JY7*WtzPGQbGe4>YqQpQZXJZuuC?VYm%xS_DEW@<L4gwHJ;xS3|26xp+DG
zq*5D&v<O~vsDZ%WC}~kzR)!&BdP>ih@b0k8=zK!Vb_jU0^G#1x5*3~Etwpa332zQg
z>W2>>sOvUhL2|8vSORrEI^a%SefPfm*jrwQ6=DVdszh}5@Yt(%;wV6XA>Ap!$MBSt
z^@z^j<jCYPEQ$4n+!4W?uJ7`|au)_kF@h<2pzkR@6I6NPDSgOeZy$ot4o^npY;~KA
z0ROVh6BJP;5>`BMxUEb!+nj#AbQ!O~e=0h>Kf%Fe^+rjQ`HXSHyMw&&w>HC~gXz$B
zogC+Jf%#!?G@S%LehhK~#VB+qYwv2rYjX<-2=sR2SObXKgN_YI`d!eDx-b%)hZbn`
z=|X@0vEvWF)2{$@E}mf_i4Dq*e>;b2<Dm1H3D|sXZEaQ62X$tu%F52a0*+JpHXORn
zIN?byyLRoe0@(q^5w&1TZEaw9c=*$&Q~66d91|>9m_CjV7jm>-FI<h7j9D&2-Q7x{
zE#vCow#{HTULRS?>yH&eV8z0oUm!RrNH51(LRy+Z1-jz6?$#J)GG_WO2?uWT|CmMm
z51GOLmT>%+mm>f8-;<O7_}Udc(AzM!`o7Ee;2U*Uk#qY=trU~$d)z{Pn+~q+t!iVW
zhj4Yiet_Y&_1bv7@DI01J$CO=JpUDw9{@z9yutZc!Q<yTqkyw&5j49-T(SfmEcT9(
z!!7H_nr1*r&yM=0U2C#Oe?3aoxH_=UE;!?^HKmnfb{<l3M3jwF`B-OQo-Sx;NRH6F
z#K4V=AR+h)Gj9OGF@Be|vWE*;(T8()M<mH^Grr2<d=7?`!ss_#ZnJ?eVT3J3k=Pz6
ztes*4@&TU*!lUQau5;sb{2Gcm-S>tc(XcF0aH1L?2gi1z+8dlq$3}9=xsi_TcYz|p
z$^PKCn{4aQ*$?^)w$EL@qPgi%2p{2tXfFTZ;!?>+CT21L=ZgAw7e*J8mK*d{aP@#o
zup=?m84Rl`pgK+nnxNH0&Q~w_YY$p0$7K@Zb%&?>5;>-eA(`sQhhbuAw1Mnq#xBO@
zX@P(HbXt=o<nS@8L2DWA6!YX~$~SOxbMu!SA(j6?u+7Vx9VxSHkfA8PIiFyVH{asW
z`f?n4{EhdNm}fs=m@Yvr#YhR+P3us#;k3ndG0E{j!cZOxKTw$n{sitv%mYY94Sg67
z>96_R^!OqQjiaR1x`a>OY<_~47a?KJL;2PA@itoKw{o@mpaMcxrDcWXkfJ^C{>PS`
zkoSDOGFp`g5dr#z3dN4(ee(zf{xdPq1OoyT)csjk-badfx{rpnJ|x_YkUEod7vHfM
zuu*YJka8zOBk(TIeQugtNMnyq+Pk#0`q6HjO|*-}_aFq@zHJ-wLG0ruY|P|Jd9>h0
z2DvypV~C3svb4vKxnKu#=FD}RjPPJ@POU%#C~<I81>99uQXt5%jj^KHzI^k>biB6?
zN<-kaBEg^l<Mneg-@g4)$>Z1+0hYGn#fd-?1!K-jD=Oe;=_-%>fU#5s+{OavZA&u~
z4{dI{egIK+i)TNM&)#oupj1HI8QqcX5&BIPDD<@?JX>e}IW!S>6V+hH)6g7fb0B;v
z3p)lG8)20ozeRJPNN{Dy-j3(~^3Uv?U0kA^em@NlN6+S?kdOoLpoB^Lmx1<vI~e&G
z`^)hFR}E6$x@F6%Rjc^XgASrBk(uL9{0pqyE3_uC{(FLy6S?coL{^xHJ10<w!m3SN
z2U6*TW%^IAkPu#I`!9caj7&v8<3_RhWkP=4yR5-Xl`R2SYYU{5AZ(mt>#)dX5!Do`
z48n-%+Neb}TIQwCkId&z<#s;NL36bE98?&XZJrnu<=nd~tr5{dR*2C^YjJ08W)k(L
zOFf{zggV#nA1<>}Uc7%#3{Q1MO>U?JsD*hs%@wP<b0MrJ-+=YFv#1#gk;4{hm|<Rq
z6v-~_VT*A;PKojC`T4n-9>#Y}JF`7-4OQQ{M{v5J9FH9RAYou-oVpQo>~jvtxxs84
zSjPA}Opa4lR`#lSAzvlK$h0E8Z;jZlxP6o0$z7PA$!&^~cA2n%=BG;n3jzV6U!$S2
z@=hF9)mp0LwZ>lrC)VSANk&CCJJRkeS~J#$l#Ta`outM0Ly1t(%%Prv5MgMDSeOAi
z_8~dhbPCx-!x!uoYinx;6g^$UdAk2c&I~@MOJ|VgqrgDZP6DRq1>^0Na+8}Vz=1Pb
z#_+Gt(M<Le`c8PS6k=4b3G9-oF%O>%fgZ{)HC2PlUC%y42;*>jn(zrx>pTq!QA$j9
z{L>!O9K3xPr>U%ei3*M^M3)8Z&QvtxB%Qb&mOtt!OD#o)1QkIUcu0jgqUSVvmw$(+
zs6G+avG1;m!5kPy$c!EUTL4L0bHaY3f@M@Cu($^EBf+Qxu}Wvl@@83c(3;4v9H`IA
z_X?Sr>(qy?>S$TLLo0zyUgjJ@-b^&5U1QW#{tJKoAtciN`$Z`6ebCZ5-7|~;wvs|?
zu@6%q22i_;Nkje1G|bvm+|t}E`?&8#Y%C`g^(p`+y)tZIli=Tc4$U8VJ{bGwP5s_T
z|FYqM*LDj8IUn;3`m!IEG@w#7H)w;+SbWvF%^<R9k;LhQoOay2*X37?IUt9E@lXW6
zG(4rEhg3Kn<)iWpKQ779A0lM<H4dq6XkHdc{@_06O#hNlEA+#;PndoL)-wl$7oba&
zP2BG=s7w1hl-VcZtn|G4{YRG0_)EOMt%#Y86vdb`2O>^YK0oY~>9O`NO0rrSMtjU9
zbV92FOvsZA?k|mqj$*x+>kkrW{5V9Yo3jnsF(KaI`=-i1BEJOzphBV<nSzA`03Tav
z6juDitTm0hAHzy|$gN;B(}r;}dXi(#bf-nldamST93<F?wUIilnAxJvkV-nuPj%_t
zrg(l0mFQDg7*$wF2^P_7M4<5qe#WTYaTbSKoC9YMQ>j;cX`(nC$wPg%gftS6ul709
zJSKrFH0|vr2k4huY%1l(iox+|#%p&vqZffcT2%IG+NfhTPN@>%c}5+bo9;Zm+igt|
z4JT)y=fi>mszB4*<$XOX&TDGYt~E(gOY`QM@{*2lyasE}uxDe)gXfX_`17z2+U#{2
zxkObmFlc>As01&0fTj(@#|)IFZJne693pS$`Y3HP13m}Lc<zBE&fI2YK*uDCM%*}N
zVLQmAShi$xD9k>z&zLzDJg&Bri}9els&OLK4%VB;&v*8~+U3QOH+n$bS}>*NSJXL7
zL~gy1d<CrQ-}NV>RX6A&b&oO7A>MyK+sw+d=X%+LhLt}e`+8eo^8j-r%R9E0m?@r9
zUC+i2y9)KVuL3|W-@6Q7a17|ig?J5WDYjjYlPw}Y(!`?9j1`j^kmHUmcLPM_aorPu
zP`+rt!Gb>R4rJyHy&vQKzv0va<&W%zF*s*1qbqVL3&EbQo{rnnv+5^Iouc*{e^L?6
zv7P;E4I%cKUnvXdLZtWEH^ai~%5hk`RkBuu>@74u==KAeJ85mBs>$)IS|3x{zQ6d0
zRCh?DpmTmm;hn|wfKw@_j9#n5+~(Y%F}aaUUVb61(^An8{8E)5xTPQMCsD?l9{8>!
zL~9oKn!BhX!L`g2{U#gm+20omq_4u^i^&T;3?$mg{t1A+R-E#dcUoOXAJA3Q&!4ww
z$sF&ley*;rj(SesGBhYCl~|a^xs29rFcwkrZlPq1JH1jFiA2R=p;t2xqRLHTH$umq
z^!A5;DJ?w!t=huMSt3=Wc>9kH54P{!Yh5)7IAMJM+9wOD(ifK~o0tZVc-;D{Q8N_d
zx?@<LO;B%AF)A)2qoKsnovP&EAmz5VyYeV-XmfKRMz%{bz*1!zF2hmM?aK%;?vd?w
zEPs)~*`aC&<bYBa=Qjf%K*(qGKO)=_P{BCYLzQE6^k^|`$x((Ih5o*<=Lw9SA=TO^
zDmsKvHo2cy;LVx{KFnh9cHD(}uW!%o!_87#wiHw<y*?CX%QX^_K5fvjKzlUtv-V~m
zrOWQT?~d5o&j~lB(kqt_yquJF$MLaw4<j8ME*tBq?eS^H)-+3PB$)dkj6@MpA3vOh
zVz*FW9@ztGpS-{AzNoWMnw^%P`USHPhCA`6!DYXXZCl%7N7rL~P&8zL5yWD0C=;fU
zsi;*j_ahfMb87(l!q!Hq_*RQ<=<YTrbo<o$wasl0IaUqz)wP9+Z?-70DqWs*!nB5v
zY2&CYjV{z-<N?gVKkW3R_f}QjsqOjJH(oUKhtc`*9CBPHro7>)+19y12TwBG^@A}9
zrtYnJf1~j`>^vXx7WkXNJGbc?4c!Rz=N3a-;vmS(&HW*9^Xr!{+CXdY6vlx=a8jjZ
zWN^Y9Zw;t=H-saO(3iOa*=^k0x1|!Vj9k<;G=8VfBC*{9hNJ*G(4Xmvnl~liSVdZ`
zr~E>`c7!Ara_jqx=euvlZY*YSB-aS452NVV)IZ{Qdq%h0R@!+v0|ox~+|e8#6a0<m
zLv$<>c=U8-NfHI;C}WPgTp5FZBvDYw4D0gWH{z8JqSw8<kc^H5e&=pVrHZg_`orAl
zh<ox?shwlaa#=2EaSQW93mDNT=&6gmI5)A-8P%j5X9r?*rc1*(-Qe^NC6>8d{pF1>
z=5@=_#Q0L@mH!IlFwz0J85svlDePbH^jA`lCF?6lAm=2|#k>fUJbA)4X!BA2o&Qlr
zch<3kjPEdy0c=P#qZ@k8ot7IPjaq2l-wIg)_!GS}i}ltGHoiw)&cFw(X|pR}TId)v
zvr-Fx8kV7F<QVCo>tC^~c}a7Go3W$T)=5nix)7N<KW?xPG;uu_t1BfiTwR;c&a`Zi
zrYd3J2i>+b4#Y0bu&f_1F<j~+i$F?)h3CVC>BNN%uV$RNO#+>{H*QpsGHgmU5AJi|
zWGj?KwXRR!N4mRNlEG80q=1&ftPZV<tBY`8AErraLCO-a^|&s_h4^~xG$@m;4}>4*
zck*PM0*7@@#uFN~YqsDE|0Km#p#n678Wm&6oV#PZ!7oa+x-zuM#Qf~~{Z~-fql6e4
z89~)^B9U&B(JZW!cKtv&a@Nd9_r7>qF2d-QWG4h<WK+8C?(Pl_lPF<uzCpi$<QN`C
zcr7~sK^Xab>BQm-CjJn`>!3CRb~G4*@74rt9HI$mL_<FXhJ}^sM^%tv`HJkltfIot
z&kq<5cqjlX>7x}gfKq>deVT01u&*&d2Y@z!joQmO`j~W1Zz#^aTigz$Ia0>K5(-u5
zbZ`<eM(hpz$OedIdPV8>%0aiBRVl=Y9DUES&Wfk|yYg=?#fu;mTbLvkCc*mPKp6U-
z)PbT3B7L$;7>74*GuD{sVrB-^p<JYjclRdttH5xRuax%=4(g*IM|Q{9%iq3z1Nx7e
z8Cz;CHPUUO0qA|r^i6k>*EKb2?KMZiiIAVHAauo$)i<Lp!WJhveL5#JBEs$Ou#uM$
z#3m#P^lJ%S@~}~lj74iEQ08>@I{j?Nz3~g)3bQ}8<NSb<VcHYjH|xP6cWi}*BVm%)
zMndUFj~_?2S{W7l19`q&XJ29-Gz5{s)8mN*ri!#M()^U)sJrF}7l9SEFDzJQ@RC92
z04U|4m`NO9H1pJ~gL>1#t<iX)*`_jMIi`r2bmEbx^IcpQHG^TQE3(-O#l)=zJvTw2
zpX%CeB{*JYF#m`=f;`iaf7O@Tq#)2BZfT9FjZsD+i03Zs+A&t0K!I9zvy5d#qf%<#
zKnNf%_k%uQI+PEH=oQ%i{?@!Gc2j6WFgy%|M>dLbz*ufrSIb#>$W*LANy}!WKW20r
z2F(J$cf<)X2?_OWYcF5AB#WB`qktLgR#(Q!hwK|(%@>nG_&T$|@gttY{&zUDF9GE2
z6O6J%A6zCD69k;sA$UigO!sYdpBWI%8B@cUE2VA9r%q|SIo}@xYl;zQEs>QvVy`ZW
zcaUGkX~auod*C0II?pyFU)4N+{+)h~Gx}f9#DfM@fF%rq0;94Nw^+w*Dw@$(iB|1Z
z2LN^XawkeDIpXpcRK?1yU@7eTzwd3Fc*d#`D01P0C%N6O(Mf5K*84Vy<AH}g6*pD8
zbP3MeQ7DH{1G?gV(0(|V#oss-#Y%<i3CMb%`}bcM=wLT{?qatYI2wYVwBT=7@m4*K
zL_x8oYF+Xct_BdVAw{>hnMo=!BT~`IkCp~-ShX*geR{u6-|;67JJ*rVJEX2Pa`kHA
zZ>->uW6*koph%!q+vk<+K=nDZQ#a}2a>^y#YG6IzLGbvNb=Fu#<8=k=80VFTwOR2!
zZm^-KpP{wFi+J5;253N*aYkMV{)n=JbWW*_z`6YC4M++gIT<lAKVihkYPYAb8_8K<
zStq7gvMpjqNA55PKEv7RmLYd3WRBx~^stk!&XnxiBaf2=`L(9L6ChVXkZ(*(Y}||w
zFpT=^%c6C&ZfVCKfY{4c2baMZZ{`}UMFV_(ua?)HW6uev>7H2lE3EH5gx|cy#E626
zmnZXLpD^R+{|{Odzx97(K-CuC?Em)K|Mt?qgnaAHlhd4Wi~Tewj-FCXpc>x#H~Wjb
A%K!iX

delta 151851
zcmZs@1yEF9_&<CJ0g)0FkOmPX1f(0KrA1mo>F%zJ(p>@q0s;ck($Xj;ARS91-QE43
z_51tR%)2v=j>_)6=brOCpL#aE8hf}H`}>#_>`FN`>P!`S<H_dYOYIvL0HLrF5XaCm
zc1}F*=6T>go7@(X$``UhXS0C#;jJ*kRwNiFKS{@##$z9QZk*IUJ|R%{d4**rZY;r|
z^J8Irtl3MtA8daEz4ckQ&#vt0NKl^>Ld5R}PdF*@{XMsZgQ#WZJ^T1)cN+Dvi9MF1
z;F%=s%2UpC(+x!7SbsmqJJkoVO=lc8vwgk%`0$w`Ue*F#lu)~7;aN(N&zowSp3mmC
z*4C*>Ng)!nPLTqZb$k#G4vtQ}TMVm*DT!3pz&BN<c-c1lso`N&_0-*pUvdFtIDx&r
zy-7()Yin!c<Kvnv^jIGzw{jSlc&)~w;Uu_t?B{*wy}`$Eo%+{bgN5$+)1BN@51+<c
zIz(ZJVbS>aGnEOTo+c+JyXQ-_E$nINd#lkAdLN|B%4=LT>@;5Vri<Qe*X_7CJ5R$m
zSCHM_e4FD+ubQgr<&fyry;R$h;^OObMALN}o6lbJdo+B}OFJ7nWho<$&PP$%aII3#
z^)B|F$37Q9h=ia^*zru~;+xyy>W`$dV=qVU<t<3RhCL$C(fET<=u`%_-?XMyfAG?B
zNQn<Abl#_pF<WSl(ACw2?3Vu|y|Jlz?5H=f<)Z(?4h0&vaF&w`8!-E$1B-CJmKFJ}
z!_uAHV{$%8PZ;uJdOI&K53Gk9EabC3oY&i{cz%&z@$<ghUYktAg(=4p%l2o9&+r*<
zEZb#?q)P+=av_Ma3eip_Xy^9aDzvvQsWkB^g-EaiTsFsIB4_eCC_Oi?FLsm|ax&9T
ztF6XkYqT5l^Ybe!D|2&KRtK_9rd?+_yza^cy_4Xzx@$c~7mkBX+;<N~u%M%H>2n#{
z-mk29ui{=A6r4cX9a^<RXg$t~b9~2}IC%0_2KAGV;Zof^Jmoz_-&ZwlxaawLZL0Bc
zaoY?^PiA2ltihT%m$^`PnV`$&Z`}v!y^<PbW?a~iK`Ea%;&~p}5?$c=4GavFoA&Js
z?Ksp}@w~b1$GAS94a?D=`N2$cLO<<ak&rfTeggMNAg1w33v+j9Xvjah@VgMgbEg3l
zLcX&{+5%tX!r%=#2?^w@kW#i`Wj#@mmzx_zse4N^p^~@kxV=Caui}FQZI$`c$%<o(
zsI2gcii$sf{`j}d?d|OydT~Ro2G1a6rg-<0iVLqs$zynOWaL4FRD{&R@z!KHUUv2R
zu8{u3+t6CKPXBAYU@1AF6CGZR-T!k{PFMBn1>SpY#9;OLgoUlyNj6<BIz*03xh+M;
zmKAbq!lf2=cXu<ygU-*-C1?kdxh-v)u8#lwon3PhYhU@xb^fQdwH4b-5-Z~|4Pgju
zCgmaayT);aFEUOx8~UP>XJTslC`LwzvJWTa){7r@bEpwX#GLpg17?{6-zK+?FRn-<
z=nwkG#z?%s46UvD8DPF1Zj+-w)U9V)p#@8amC^R*7Ru#7mUI-QFcVQ2Gc$9?zgJyt
zEv5J2tJ-o32=zS#9UdQV&o+s|lu<(rN~r=zUCbpy*QX1f$D0{_{w?pI-#t9~uU?@*
zF*09w-@7;Hy&zb&<>cX!+F_}vtXvvR>^%g#4HFe5@SOwkU6z?r!w!|ak3A;cJ*YpX
ze$CmvDq&)}$Mw~zZd${YWQ}8uGRm}AEQ%<sZOv)0XHG#u!LG)kjv6Zn#;2+px55sf
zjhB_Ajp^*{1m%G+L?usEh;nCRW5}ZNu<URnJLW*g5DUZa5aYdiL27bxCW1Ff{=tI>
z;1g|*_dFV)FX{zxopcSVrN}*Xu;A`fY_gxU77Op((d$!#4R!EWskkfFzwUGlP-#}I
z6NqIFJkb>jP|rIPg}pd%;%fMaGt?{|pznhBgfFV$I^%AlQcZHapmg~9Z1(2*MD$|3
zZ?xzQVr&cx;xfA_xj8SXs;cVqTIO`An_k$fzufOGX*izX>=POq9+s#Mr;mI2<1$7d
zP9;xX|LgM))c3hVc;SV^fj=gDZl5vm<<w?v_Ud2_xm+{&<;T90NCFPTTPvOv#72Zv
z7Hgth?$AK~nn}64g#{C7gaCQupLUaSt*im?$WWi_>!(X)nOj@;tUsyVDD2QKs;aG>
zSaTv2%gF_=k|&wPDtXB%DMJxq6!Nc#^>-yxHVE!Ly{aN@VW3H(n=iX=vB!Oqo-U$F
zdWlzTaSCs}+1eEhE_OG*;@=#7V>?~%?r3do?J->^n|*${BD!<rG3&mPO04HLmwvsL
z{@39ux3;#{*5@F_Nw&wL(q}JBA19~<zCK8Yj|%j?!LM6TrS{grStfsq=vqW==DbmY
zG4bd3eOz!rK5_xd%gYxQ77R+=Hb#pATvg$Hk_7d0jTKp!&;;4q<?#{9-;Z%x{&17T
z$PD-Q?}9ycPb@Gfh?_J*g0^&WOCZ(8!GQyu0K(gOs<LS2@v^rdqA+6@7oK?8&ZYgr
z?&agyGx>9SUES0Hv(KMCeL)bTzo3unTyn@7_~fbI^sm*_6I~wPm0TNV4ben}318OK
za~8kykKR996PNti<Xxa!b9E*z$+(C~&{VTJp*)IGKW$SzxmD9QYu`b!Q-9F=_b>HT
zt_~}HfSBv7x9h0Dq#*d)KKR-ZhDEO!I_yQ#+3sT->qdyY9_}3!Bh4jFw%|%W6Lrrm
zFj_l0FC2Op*y&Y<ey9<zlBWYNI%5-uwXMqKtK>N#1i$PP;{`43ovP^@R#_s7#gf4f
zfY%fQ8R`CtcK6kb9~;GW%Qe>xCo?H0v;jLW87<h`^n~uv_+PIRgtckatq!@pR?Bay
zuGlCiuQq6US*#?~2MS$RSJ$8UV6)2Iiz}z1bj956!iOR1cS;b<aJ;OVW~A$bU-;>(
zwPj2<&8li@D83yF8N8G7^!K1e`hwiPzTVy(MB`w2S@J-G#cAGs8YkhmDQ{#&tMd*W
z@BZu1^YJ@4UUY=+8%5Qc2@Mz9b>-^yKXh23Z$(AfeY`w8czAe&xCcp~w4r>dtxe`B
zC@K<mURQN>J%JT?r$3Hw{z>A5qFwW;X#5k<_I?B$to&}}yN&T&=2^UUyVW~{GkZ?{
z(sr`7&K+z?^Z3QFBVLv>w|_lRd9I7sSL4Zjw)))<vpc*gJt@4ly+TfZn^7Rmx?>fs
z&ARQ{jVSnXf$DlUNO1edBgcV>6x+tz5Em~uWHng@Z@D&IOo!ZK;E)rzh58-<hnLxd
zpphHjYWyblS<uzS$jDzrkv^`ZejgjEsH!q4{^eXZt(ePsA3H#druMyhp6clE@TZ<-
z_o|a0I>i|oH0gbb^7Qet+2EAg*tBOGU=`7EEG_E8FMqg+^pzs?4YqJI>ux?273sI%
zm7#;@dnZp3`b@_{Re1qL`sH?l$Qn*lVb$e&TeyY&WXcgA^4js7&5B!uzynV6+^!M5
z-ixxRJdaDir1LJz$jGRybhfkGoxQo5H9tx<;;6BmZLFU0kfV>wPN}S}F43(U-8j1L
zfG?$);!%Cwgsz8QD*>zp7Qwx;wG{S`9?Dgs;&(7IGHTs&=~%E&TM^Ib2Q9NRjsbKK
z0gdgR#>Pg_O%#=sz`wpA5X<NX4eX~M=9&YBe?PVi1^>s>>!zis85HO_38~EWZe1R)
ze$bM9CUfl`QM(o6HhG@B*Q|7cDc`>)`b~LB5Dg!qx;gF08d$Z3;fu)<?=4G<O6?5g
ze%vi^W8M1K9MeOJ!h5avZg%l+cK^e64{<P}IzWpw;uueCJbYaHF713jvFUQ8#OJ6#
zL!3b=t+1$wf}C8Qj&NjTBueCbb>#GX-_jQOa^)lU_VzX?J&hFH_6@H{^7He%ckdqb
zMU~0Y61H6P4zlAD5)q-q4w$m3ZkhX1nZuzVFAoqyXef5wbn$>$Z0F+Y>gwVm9nIIe
zI?tV%hEECiGM3tJ)+&cnY+0rv39gk&;(YZl2Z9nTpAd#D#x%)Z&+bIqxv2DAl1T0U
zHT{%NYv}Xea~<vy@FfnB02gTH>3P9i0@n}D&Qf8a19$|gHkH@z&qDNWo|LU5)3os>
z_y%muOP{6@3u`MYuhop8GLA_tmV_Pn=A!9*q(m`IXnc0I&wi>{w~o?dl|~!8QQk<c
zxC^^zZQUbYw$k%t`wv&!(a}*59tCz#3u_`XuP=U33wek2WYyEuy%{sEG!fX_!@z)+
zC~RQNr>2q8pTO5{S#`Zg^!|*<ksk4D<Ga&*9iwRy<X_9$!h_}x0X4W-WV??0BhS{a
z%9<#2s8v<z{cIWu*#Mk*HHsos>AW#2W|84Yesa1HMLf7Er++@E628Gifc~C<HnDqS
zeZ4FcC#225?<WCXkh#_cth2SXwYHj?pO=>rcNQaqj^^f{@QS7D<{Hsf9J-<pCt1(_
z)@jsw`FSC4`?c7P`@G8w4iwtES;a;OuF|cIac2tUOIlJX+mF{h8mDNmY=b93Y^T>#
zk}7Y%)8dPV4H;u$kN#cWlClnTGn9UPAF`?6Yfm6zp?7-lodvFQ-I=M01el`+Z_TL|
z)X74IQbl?(=10z_)VG7eM4l$eG+bfsNsGMj(z_m+y}eRf^5a&t?A>%;=XK5X;b-kC
zPaAi#ti+E8Qvy1d=IgfaO%F5H!-}hdN;<4qw_fBKF>zaob52&J<#n<4h5Rd8#hLS{
zr)*wxcc5K(f$D|??UdtgSQ~AO%#-^LH9B!|agN;Gf<GfKX6|Y~%-fF3PJQxO_^H)c
ztLTX%A$IRjxV%7I-?g$A;%ZaDQ+72VPdN;U*JsrxsQ)X+p<1`(R$nD3Kj9J*+T3&d
zHOH=y+eBTX(>RqShP5OqWWyo&dcG#cZa-HPMirSoKW6#R+;lE@m!`ieqoVP@x_Z_E
z`NTgYA^G99M_crugFRd->lG5nhc5g8)Wr$qjD>GJvVTNOXj?K>U*^z9Yds=S(n7&P
ztNZtV>s^3}cDh7M7_#c;CL{uPYqDTKS`U}^9qOiuS#=-)5~c?S4Y*)Ztpo+C)Fejx
z1lt%XY3aT+;WxvK02{n1H|gPsz|7d^*jYb$OpjGoJsT{vk6=bTLUtSi?H8NB6qH{M
zGR~vEZx-r_?WF4=)0bha+uO5;-fFpYL{g}!sU-~vLeaF2I*<I-KX^%&ONN@5wq7+b
zkN?kGJmW)af^~2(&|292qf1Lm&(A%QSm1%>Fk-erO!_$UpO%9)9@#2MCl(^<ws#kv
zrYh!I+TMJTa^gq|?^uYD@h`^uR$RQy$GY)x%9Nm1J$s`${sVNs=}K~bbqV48afb2V
zr7~$`qwh*bk~8u%PCSYdQke)Th*V`N0_E!A!w|U509@<j_jlszG`wjm7Z(?6#m=Rg
zIsoQ*#uUDEj+f;Y6eJ6gnBQ|V!C}Mt6>x{x3~Jgx*aZc%NIm!FVOX13Ml|TxDD`?J
zB?Sd<=4fbCMLsdz$;zjmQ+~;i=n!jW^@(js=wJC!-Il%m7d`_O7Sv9bqNWW7*ca1x
zbX2_rL578X>2m{+j3GM-6EpLY1HUehsGy)6b2TSUx&1;rxRzv>$QAe8p*;11$Alp|
ztN??zw6uJnCg*qfnwR&oNSz_B^DR3c7uS8fpf-bNeFuAPUJed}W?V#Nd6YC2o0<u_
zh!_DMqc#DfUo_3Q-2d7EjGztQONh~AnW=Hi1;zJOb@viD16tO6kw9%oRWIoA&~tPj
z-^zQP3w;4(HY4Mw5yuDufsc<LB$feMmuVHiW*B~Rr+W*K-TvZBzn|X?J0h3Pe--nH
z1yf_KalMzDq0GpByk5`kyiRgo{m}mQR~7lP;HjhkKHseJ$zxyma6tNGd%A8>yoUZI
zT`Z_!<6FluF)_+dz`Z<*307swuABvIdU@j9h9gN^TU$y>YJTqrQtHS)JUc#iiR<J7
z_l+H}D(2z$Q(0eM3C*W&T0}%7O*R|*Wr*uYwn`p&yUxy++WmTYs_RYyb<=4z)af1z
za0IVg+H9FI5DJ9g>&w<w^u7a@Zt%WE>XXC6Usdvmh8oBAAgJ#^TwEMfaB@6`xR>DF
zRPr`m>hwzVjTQ`~Rm40iev$aSa$0zr=Q^X8v>mJ|#(LBK{ND{C2y?zY9dgn=L4u&B
zYfF~b{`+z}?*X|&^Noy*gzG*3qP=h4TL9hIY&#<-C#TX9VIm@qwy%)FgNuW6dUaiR
zjMHK@+c^D}{o@q47Pp2Ou)vQG6dXoJMx33TK2H4@i}~2^^J319k21<I^Cvj0o#|IH
zcK1J1HD(fu9X$B{sf0O;ac}HiavNQrxW?_i1Cey4tN|Ii*aBggNR_K;RtX+)Bu`YM
z$0)+qD@jx=)!BOgC365j9m$6WoxgyyHAm>-^ni{q<dJk_KfZuT`S@FQ89KtISy3LI
z1bKQ2JWOa&Dvg*;FC55H+u8*qprC^j4HjJN_u5ddZNrhehhLumv19}TWmGz!J#CVF
zs03|FPdjY?S%wZkIk1yUHLZ+{BuWq(8XAfHKm0XvDl`anZ?Q0+Z4$>`+`L)kud|!v
zR@m(EJk`H6HP_uc&KXpc__xN!cxY};5BSI<fu_X%Cr2TLh3Jm=phe}x)-hPBTYf*U
zuYC$ev|qi7aaX4!oa3{Ztg>Fa^rDZ;tAp!2y}X1dqcHt`77E=2RI^zKadL9*xsgLW
zwlkk=(0*A|QVO}T$IAlBoNc(DNt(FZ_3>CXvp}3F&g=XjuUI^<T90AQ?wZG;<}~4r
zWcB8)z?bIEtt(G!i38`}OxkVZBfIlzvOm>9zigNUZH1d207C-HU+-hLogZz$D82+@
zlST$4ISmcM%ymA1`<<6%SF*VXIVp&=uuEk<E3g+A_ZzY3k$<kNirj4b%pt|3fp@3f
z$`1?liQz*Umn{J)-P<^5kbFS0wHeXu$*VPuCf4)*EowrP?ogYiTz|^go+8FL_q>k(
zbgZslp}ozfYHZSo+$1>Z;R^z&*`|7ha%9aZK0f~a^X7#;BMPv3T8^7g^vv4Fl%9;{
z<F-g?*2JM9X=&;5suNybUei>)|9v$aHDVRsiDnn9(2tMFBKjfd135)vCoeO0#O8=i
zn1=UyClgGf|2?R0mxZl3UK~kYS4W7OeYTsjd^q3YR;{=(D>)vTuRMJUvzkaxDkYYY
z>26nw2EFC1=k2X9il{b&j0bDIse-Pn{Tbj<mLTNN)^sGfSzYKJH^qJn50-Gxa)qG4
z7_W7<Bqq7*tQw#c=88%)J#(e_M~}D*emp!iP4_1H2R#5IRLS$h#0T=rO&d(C-^%~|
zqZFdJ14x^C?#5{*{sd_&oW#w<bzWKj?mJt>db;*#bG$r8rgi6JNOGU08x)YRHiJrS
z&LL^y5#;f^l+9EV&xjVeZi>6`p*dt_$L}+9j+LfF3wtoN9siK@zn<yd=ACl`NPW{=
zGO^!tEkP$cv+&J^{*8x;3H?E|S#b!xlW@k%Y3vy8*TQ{9o_e>c*xm_Y_+^xrJDQl7
zczAdKo=p^X`@>(LdLW_l7-(B2jkUkXhy`O~^<P#bKe-<(-SkPooA#(fBZPN0RXIe$
z$uQTguZUXXz8cZ}e>ky7K#rgBWX5x63*3JIOsx0m)2APq_@!>3_2UceHcN}_$2{J|
zD^iio9$Yyvubh~!bJgD}g+2;UCK&gClW&FLD?>N$QkM&gRc<aI%b+(12nw!kpJbAP
z3vuv#JmBz-1nhy#h?IaBnz{Ll)`dW20wV{hEMv0dvC0j|F^n474t1##s9|4G@AJch
zK2ZR+uM3S30KNki#QX1i%<Fl!Yq1gpTjL`q>o1Rc9OBw@vrX@Yr8?**?MY<4Lb9>7
zksVo3i;)a4JTfv(*fU|pJ+a%kd_2@zXH%wYD_){I%vj44&C>SA#KM2&SVLhA#Y;)4
z@A~Qjtp1K2%*T<QC_~q2L<j+iB>*2jZ3&~Gu3a|cB85PK16WY2tq<~DIjf%a5sZ-;
z+d2lq69>5xC{kRcKoyy+w(Ya9Vka3i9l-pzBbSb4Y5haR#9qwr`Syv|+sq#kmSEb{
z-{NZ-p(I9bgD#B$6fb@KS#SkF-h$5#R{Bgu*y1{oDkUbR+g~8LAl96q=%my;85Q~Y
zOhxJ&2WP!$!pt!;gZ3;Cff)F6iVPj}1uVZ1vTOV=U6RXRk&KKi3x750c_~nzUe+|k
zn!NzpI9HbG>(@gQm6j{!<3mH7LP9tYc>K2J-njVqC-C7R)C{EQo-*cFc3+J^?E`QK
z%@^N$RopRNRdYm!{dqm*uE%f8o0<>=TBnh^Tvh2qz!AXX>&f~7T?dlctZr}jdP!v{
zPXV0@xq-Y-rll>_X)3-PA0M{_;R93xKCYy;Ry%iy@qQ@adOSQtI2b6%nv-A$v&HQV
zkB%xTDBS*Fce*<_F#%u|b6lrE>Dch_{ho>6_B!Sfa&E@Rhk1*Xgyg&S^wZm$5JU~V
zlz5?9<a7_c)+G9^uiMn$BZBlGH`bfk4HOYU<@5t=_k;WY^jKYW9Vk6=MvZ9%+~!p?
z8vc}HpRjv@^&>9)BK0~LGzd=9UWu%MkdTnV!XN-~0QhB|-1{V-)85_=s*Tw1-@j$l
z*D)ic%8k1d4$}_)?Q+72EDjp1U*qMbu3Hn%m6#twrGIBXzjE+O!$ZvNyL=ivWA~K?
zPrKhsb_%V@v@b1)%ICDWZaY>sTTYPLDV3cBhj-TNpc!s%{__V%p(RHpZ`k4ph}Fei
z;r6c<%+1a7ii;^pNU#FL2Fz3m&;!IWRPwBN$P~Es9sYwHs`Nmk1~Rmb-4lKPXy0^J
zS4@G_ZLr%?PV40ezV(gdBI@Kdq>$H8`JOS~<~wjMI7NP#RZbvx+3WQ4;iuJ8v2aaC
zZ10oC{Vrx8b@UNNJ}AO2)nvJkjp7SV)|2}ho_ycT{s9X|a@R^s2i#20zwxrhfQED|
zsN_|Ymvin|_YWR7-xHg%;YA*p5=5Bi-#fHeiD=IFZqrm&uTR`*;<eX)e#-w4by+Eu
z^^=aHAZB(a_oWwqB=Z540Q@l#u>nxrkMB<0|1>a_DV)p2#l;f%YQy{Dm_o$s6T$q!
zS(^9pI4IyiYdy~hqC6kY6{&PTST-hoB8K&fj+}_~FnSJ!a95lKBOkln%m7hUb}Xfd
zOO@Paum|?NQo2a03>~nkngEV25YFVkD@GXNzd7-ZlMM@Q-O1{no>Lih<=yiLo~<5y
zFVMoiE*m)H7is83p0z}B{ycZ#JxaQ*B&HwO^BHQ{KcJbn>`19TO7!{^ypWY-Ytj`!
zU&u}%RY~;fa7amXdU*J7w*{ZAPx-DGLXZ$UXU6N1b0IVDyzJH4Q>lV`(>_sVL+Vd1
zYT*<00%tETFBcaVFR%KX9H|lnv5yT?Jn6%SNuNIhX&1<>{{vV>lt>4Fp)orvDlAM9
zAtg!sN>Val#G<IAq|f4by`Y3ptjc=Qgd+*m1#&!TY64Aw-=}Sqs_p-`UY4QC^f<8_
zJU<16h;b_u2R)ZGcqJ$>P={56j~s|>;E`vm0OXC@G^5IN1X_f<5M|&y9-ApM(CVUk
z`3bY`qF&1}&j@eh8#p)Nqi00>B;vHN$H#|nJA5e~;uN6B8sgNZ2wpi|sIv+?K$2l6
z>nHa3^*tsRT^f!1@g-hub9O>Emj@TeTXn!%f%stZNcegM&aw6Z61Ik_>Xg%v!jEX2
zB3eLRTwF$tINtdLIXuF_-MzVbO^DYs;(J<udFSTQXa2`$!;h%hP98W>P*7mvQ{G3P
zo2qdHN6PVi6T|-o)aOZW!9l|Ss3ql}zT>EL8^aG2xUi0e!}Jf-ZVh^H2|`d<*t-Rn
zchNgN>R%J#*?1`?<Lc_Fo+&{p+V|(r>d`rcpo>xau3pK>SgB!WxU9L;x7V*<J3Bjr
z&pv>&0R<EF`K&(urpd?K<WS^B#>r=AIW6vwZ2LM(x2gq6(RS!d=Saf}@?Z&OUYviO
z4^tNB?-018M)<fADZl9pIxhcd1RH*5R_VkVwK<Hf@pL|nO`9bFd02m9QcYD_YHVNP
zh%GGxq9u|ZwZQj>Ku_I&<_<7zO=;v(17;3Fs9^o`wi{fBGi;nfBiZaSpc7PA^WMik
zIyqU1Xej?5^d3D1$_NeyptL6!9qn*?Z0FaOImmaabL$?1V(Wr^k@_1owQN-;Z;{!>
zMI#oXl+;vVkHgg--r3J1Gb;!D<akN)Ce@Q*7YUI40#bZL68NLF_0ws3QoJDGbO9JN
zx#hw~zE(o}|3dx$U!)KB9mQjRu$ynao!ET~qOi8+$rfAPk+XeGC7(lIfdMJEW{$4i
zHs(V0mssq2S<YRk?7?08BBkM-)0Z!{;;+oU6fcmZQvNusbrsGUF!&OQJfrcYV~~h-
zOCT;8kCoecUT!2ge;E*Efrs+W!-Eu}XJANoUj_o7JFs<NZy-&>=WDr%`O^LgpQ*8Y
zny8?`q4;m)(c?*rB|fiCrHnb&<V9}u0h(hzs652Q#RcLemv6Op(H#>-2R`ynqu6(!
z09n&wAv*s$Vo^D{1vc0n48N<ZE7@!gW>|@f6le`wtE*qCm5bCZ4GkB7s93LOOHBZP
z*Kb%mdE1W9NnjeOxa@x$pbch<w*XW)DQJCT1D`?w@Lye44FTWuq@)Z{2_UZXB+0L}
zs}~&o{ppvdY5^Q7US9^iDE4RcZo;{y)12tgmo|P2k;~lH_)hhXsu^^sOVem_Eh{#h
zdxg7q;Dl!@{K-ekr`6w}G~<U63B|A*W@_5pVYp(4Sxr3FaMnP)-b&QJLTB8GdRhT6
z_x4oHLEY@tHc*^FojXaHy=>n&LTv_>4=DG#IDcR9-wA4|U`3<Pc*=Y!cY|**5HM*>
zz<P+aQacWOf}!XH^zO0^M9I$jdZzIlcsqWj1sOUq@m1ifI32=@023T-Aew+_;U5hg
z`NEzpuxu0!uQ`&ehH?}z@yH3#GfGO>n0*=iE2^taC3ikh6NoK3xC3M}N&`3}I|<O&
ziVS21$ll?!kVLc>X7G${wCJS^5h*kvB)ur-k9JG>NLZlg%g&nzl8=O+ZsbOoKxU*a
z!{NBb4KN;R3qfE^n<3B2r^POc1p}J4v#GlQ9NK>|uw#M;%!cZQUgS1$<rNPU8}UfI
zbJFB<l>OnBrXH(NUZ|Y0OqMcUg=_UjI%(GbAJf2oa}K2uhDZSE7r^EnB%BnzoWlp@
z&T}n0@z>lghjUPWtW1;SGha5UU&>AWuxqLRy))5z+O09o<vsikdv0!SLqkKGYCub;
zOv-yk986542+*P5@{EZ&Lu)`8LVN$5o#c-|g-K8H)Rr7}z%P^Xh{(vmmUn3@Z4dcJ
zR<p|{%31!Ys_h9fbY*EV-AmAOtcUzQg@L=R3p675Zo;h##dk2wQk-@b_~1vle{VOR
zEmtvSJ%sXghK~4wX7}BH%4svTMk@B>qMR-b;WO>&Z0|LJZNKLYDvr+GL<H4g@EPx~
zfHdZ`2k2O*uX7`F6)8+vM&>U>+IhHivr&+^>)R%a29yqrn4LP15!tR|BMhlO9F*&x
zrVtB@NfT>@8ojWlb7f({m>(vctt5s8x)XaWHKU`WYis7(DmH1GA&1+sfDsm`)M^*u
zp`#`Z{rxDF)j6q`I|OcbklD0~@jIa>3lX5y8Op!8IXU^cZ$Z()2PF*>XwKqvv=-Yk
zAriuz<ykM9Iyt+~gw~aPQ(e8?c9JVuw+F);-k1i<8JD$+!6{lK^sF&bOQGMzmIlaj
zkLaPdUsd+DP9W*rQT`m2*MTSY{@0C*j-q~_cQojcrQiZ>ZyKhlHSY_>#X&NvC94|L
zQ%x2gF0QYLjiEZKc-bX!eBlcDX_^YULC)P<?d)c^uo_n6L{H}3UyWn#s^mFM*E&y6
zPlFQt3l6vgs?_hvaH-`zx6619y`&Wj|7aj=)pU6m^E#fN6+$AGcdjrAL;fzd?yi1f
zxF7j-pN@`tz-)bg>Idpu!~6>`aZA+<oNwGHl^E;3k0lmP)N4%taK;geBsrqk*-K%3
zv2d~IGFz6U?MSZNA})`cLxY0bV)EBZnF=&A;PBwqJZ9J5XwZ|7nGMy;H%wiA>>M92
z&!X-ykuqRiGB@1{yt}onbNmi@tKT-`DVx)U+{+3IEJ1zMdKvEN;c<Pvb922_J)2V?
zJUKDZczsj?&)-Ye+?2Vi?Zbzr?}GmR)cbL|LjCq*^6vf8aVw_>S7T$EKu;0a^UQ&N
z-D8fO^5scCto7K7=kC;^jb3{>Z$DJHljzvw8CY@d<H`As1bT~}!9l7n^{7imhU~D<
zuH9ZvVf2W<#1ql!s{7eKfwPuhT!-l%NqfOky|&GF7lk*rQ9n1jC_F50Zfy*^$<v>x
zSujMHykYtCGCT30L4P7ggpbp+v^voTPQvm1zm;{yWX3O_W9))&N=(C;&U%3=->%&<
zL6%^viWD%k9{~NfqrIKdeK7_N6Ax+@y<QMC!Pw1JOr2=(bTx^X2XUMm(5{7iE>B-8
z2|xZNCity>;GtXKyXDXR63k%|zrtR2DTs}e{pwbimhKSTf$PP~$CnaZ&>(FT#e$LT
znFi0|Faoiu?M+Xj7#RurxFw!Qbk;4&#FMm@DA&6sm|RkCNv^j<A3Lc0@T~l7*rN@7
z3-%`sx;WC~W%gpn+!-`M6lZCQ$!|Z5^TdEMv9mR@QU+2?_zlHqg%M>wV8PgmqKWm-
zhx(c5Hf+Y=)>RwpMCREoDr$cno$9^i6x-gZmK~QmHl2Db$fB|~JG;${tL)*$n5l`$
zvWI<_^m-9*sE;W&>dn6GSEb8zj;T99!2(TJ)d`xNoo%w7<Vn^Pk)w?{{VgpbmURNE
zJP@^q)|@E$?5Ig_d@d$u73m0(p^n72zp<Xe_%8@n>ohWyziV(;K9DtBba+eh#7g!|
z7Tp)+EPovP)vH$nV`F0o1Tl612FCu!=LO%s`G!l<*8DcsV%)f7P2?eZJaZUco8s2c
zy5XXRc%L0eBXK^lqjJdFxjngD`(sEI7b>K-5^+^YE@w+!l!$*;g9=G7>c!SqsOsH~
zfS<U{$g12U+`F)#U?aNNx&A(4VaTUQ81k#s5PDm)r5K}f8`Adm5a9;;G7bhNCMFrF
z$R~G7Dk@AU!PpgKzdQ{yn@2D;?uPikmY#jO)Ee?EV+Xa)WjYrBHuA_6$k?L@^&O;N
zE;5)O0_PaM1jg}V1n<lzk#o>X;lLb;+P!u%efY|`QX3ai%sl|OTNr2}_&&RSO&4`d
zNFydTk}Bizy%34-fjZ-N<w&w)KdLl;K!K#PbocK*O$R&`kXJjtO*_8B^G$y1S_R=4
z8F@uTllT}kC~aldXQN1hL+fcGcMf6ZTk3e)FZt-c@3)7R<90YSV5utmq$w>mTu7iP
zk@0PWxDGuUBq@|Y)_~T6KCH3a8P^qQJ4G2MgDYlHpx$HVjFXF56q%=t_FRo_|9a9E
z?Rl4hvE>E?(Yjlp10QsB&8dCCKB(ngrM9`YCnz)qNT2tBm3j9}cGj+x!~f2Xcp5rt
zK<cC;22w_V#a0D0fH9aWdbQRXau3>N1G=F8%{ihx@_L@Q2?C1@@5iP0uTnty1qJ#Z
zYuR8!0dfM!DiSuo`S9vd1rAsicLUPo&heq_)VkVr+ftUvfdTu&)mJYFK;)r3Qe92W
zyiz;8&-M7vL;L1jCHkSb^$&fHiW>2MVH`+#()l?N9tu=C{M_&{B=Mjt4EO0%G~%IW
zUMjT0lVy5uC82(<`6otJU1qICBhpKu{;J0>aM<GR_Wu1oa<Rk&LD?$j)i+dT{8HJ2
zX?)MPxtH$BiIMS`X8)u@>M;P^2?>X`&4b>yy={}W(g%pB`4Ihiy=-=JDx3;}HG6JC
zoPosXbHH8V@?~3bcjFk-->>;a+bj8xy=MTO<!mJ#(|2NGf+5|DV?&uX#%(3l75bZ#
zVq*cKNJ2*C<J%w@1EQHnAjuI%_EfAJi*qcV;nf93Mg^pop`(i;1OXzH(j%o2^7>be
z&HMq-3EYG%v*6Mbt7E!8s*Mfax*V<-O+rLrzaJ>ddCTOX`c8!0b>{o%ISz`gayCzm
z>C&Fr_e@|YMRqSj%1g<`IH$yv2|Dl4IWVZ%|8%B$k{$_E>h&=*e+X&q?^os{2Wc~*
zzK++w?lbdr85tn2J}-~)1pywFh|Zfg1tS*599k?e5nyJSlf(Rx-9<d<0qXfDZbFK=
z<yBRqZ`mQ=P9suEN+R^R-!YFszp$tjNRr3!`^hXt7^S4BND8ptrPq@0{rwMp&c#jt
zWnlo{+-~UsmIKOPAQW^JzLKEeckth)x?YtQP4+Z+^JcQ3q*432^ZmdPj^*!o2Dr-b
zkzpZi`A}QhlkKTHOT<k}v~vVTi=MXo7dhO4Usnfz$K4#1J9W$Y_UQFpSWX&poP8bO
zF#n?|qZI6Hk?H=OcVm8CxqmT@vQTTGWcZx9WK8+XN$uphheJmTi9+T?a(Bq=QMe)|
zwlTMFZmy}D8y_9kbh*u^D{8g0X&acA1g9g43ez`^L5UP65D?A(S9ls_gZ~LnL^-}*
zR=}zUQ36{3!^6WVhwi61E#Uux<1jZjr~E|XhOS36_Ew#y`JU$5%Kl54Y-s2w5{My6
z=S8L!6%|o^r)~O#5058phcK(B6Buv_r%OXnU?C>cAO4N<rG=-BQS|WjLZ-_2GpV@M
z_FiMM*Niexu=e=8n{T!eiA2qwLecU@JonJJ)U^6|tCt$ER!%ngMz{^i>-fY~bD+Ri
zQZ4y*-)sydD0<_@KGXb49(dvB=`yUoGqOrZnD?YjL+a7UfO4hkpUwJ^0ehTWQ>4BE
z|J|y7PcYElGrzn%B?s~{De^fWQv>cNu>=H{fhAlkHe*vA6%_?U`@^e}goK3k_4W7f
z-vetfLh9BF@EZuXnv^H?ETigIfcOCTi3xLo6y4HCjBNR#gMDgG*gcR!T(M)@t=$Y|
zEGS&ca61D8bnZZZOdLny^E0{W;5LW=r#LxQaaG`BF^xt}`bCN1=7jnl#mT;Hef?WG
zu{Z=HP}Z-p9f+`ta0qewf2}=q9-eO9cyHuzu*7q~@T^=>Ni&yWS%7Koto@zz!Y-;O
z!jitj;)=50Uj70jHFALi?Wp_qTp3b(JkTF9;z*LusSg4t0-l&CvlUQPSD&$K!9CEA
z(>n&`KXt=wh8ald`#)E4v(vldeOQs?q8Jz40T3M;8X7?UY2!@!iPPrTGw8HGNaW**
zL~=Bv?IK(x;ItZjv+vUe63yDE7L>xm!oXQO+yhRNrKRPXA?OBf3%C%7(|iM7b@7}Y
z>|*W8tCZ^>ZCMvM#pErX6LLI-!P4DyJ$QMrivxl5dAfJ_v*oDmP>vg7xeqAw_I?O)
zK&sBDbudM0PK(2SePjAN80yN(5<?iF{*^c1((*+zgOXrvEfD%ug77}N*~p#Ed_s!v
z{asI3eQnLmR>3v~lCj72Qlh>hOdkh*2N95ZhCzi8z<nO$I*d39r1qR8d^G|3h<4J?
zFZF}mi^V}}Cu-+`_ovGeOGiMDNRaP+CfHM!*)u8OIwRd3|CDCl<@Su${2qb6fbxl6
z!1D&NRJ>u%>mff?5A)CdbkjD_TROt<neKaaIWc*|A7i^BjU~x=p0>cj0KCvtB;O})
zwjKLESBcQOKt`I1Y`vKs?=2|#?B6@5-od#9{wxSgiJ~?GoZA?jQEUK=frAb?V3rc`
zN@;XESEBg*<TYX^Vd>+go9t(~sw=@Jo+b9^#rDG^#Ddb%17Zir#u9i2+7-e}^IPTa
ziJ_v*JKrlItB3V!SA`$$iXV7DGFN9si^HcM5SmQkwXp+h&vS36Wf2ZIxlP(BsQoH{
zRaQ^;H2o+EevJj8Wh%UR_&U$l794OaUFOg`f79LOLw3A(JO|7Udj)qcuXbQ1#7CbM
z!C8L0zYw%JRa4&BD6;X<jO+aOqo}Vcj0wh~$E8tE;eqcOxCzF$UVwOShO$|jWVcgQ
z98a$R9Z{IMUUPh2kkh~Yfq02qgE|`xeA2E3+$7AF??qcymKWLZzL$O8@~!vimx-gb
z`o&p_^?LIM&zU?;YlGj}?2)5GuR`@>OHJPm$K&J%$?csj-E8m?7=H12dJm-AKl3vl
zp1==0P|?Drm}@t#m&orUx4C-IyvT<)EC~#0Zu7I(TeH2N`3<WlQ&Lh|=Im;X_JHh~
zHty!&04T^tCquduA33yJ)8}0b?9)EU5_8Za0Bp`007>SY9DX^Kp~GwXIL4h?<0LD|
z#9w-o(>)_=&MM-?&AeYHw*t%DVR6>C#jCmm!L!n?xqb)*R;B!VO>Df%>Gx;`nP0ai
zet5@vYklp>)nI)s9@^2y{cc&xYn^%N*~gp#52QbKKaLWrS|k(yw}(g{2nQX5jT5$V
zmYeRAOfWck@uFuAsrz%LgG_9CpYX{P66PF=0ukhF%NF*GG1P^#|Ck)~_eT${eP-6%
z2kFprghA=Yx%LHz=7V|}I;O7{PUc&49@=6}f5|u+e0-eiY0SSUiaSMJ_-Rh>_<deN
zyG0AV45?V{Pkr~POI8Y0RgeC)Plu>(MK)?5v!t12^FCgeG$~!L6N}BZP5C8;GC~lP
z#VbR3#+;Mi{|oj6OHNQ*_28;hg@p(P`vq7UeNjG#S^q`_W`T}<2BNj)=~q@W#~ewU
zcl1wx0b*16c67s1&Dr^=zrSB7EMe`iPP?erQ<cCUq34Oazw~q2-N(%A_mV?E%bXDh
zh%Kox#ZTKHHjZi#jNkWEoD*fU0|Nsuec;>hG}47SJXBRxb#if0l$Ylt$J1^C0|NOY
z7QnK9{W|X9XW~f_+gwB^j?n$xz^~kox)yPhi0%Yovg&<yuFLAABcQ#faxU}D#6gFf
zGwI<s2D(AfomKRLhJ|^p1FPd{%%}vx@zPt6OWJ0Bu>WeD?RH}X`ybq$bRIVoQ$Bb>
z+1pS#!-CRfSTP)n<zh6V8ufX>LjBVQ%(Qo(WO1LS`|xLfP8(Pwc^brB`|9p}+0GS8
zkk?N;AF^!S=oE5TtB#kaZ*(tNUB=Ij`_FNT=VNo)X*lV9T}2Fy7VBO_*m||tRJ*#k
zXcvtRderc&ItkS1P=>WFYgfp_4E#@t<X$Bnc{BjoFst(>@1J4~yj9TV1Ox<RZR$_A
zKm2@aYs=Pm5Wtld)H3&5XDLR84rC=Kw<swnrmL(|xK_liOSgtNEPsa-ACXx7<fX&?
zTQWgbR#i2vef8oc4}f}8oQxSA7D^^x?ZUoi$^@9ymzKStt&GBjA8-%X$ZJ}ykZIF+
zmjnS#aC@+2lxI(YS6`MscjVod$OMd*U>@6h)88dYD2P1z?nMRpqrOz9dq&cxU^$*s
zN&T~*0?BhUG>ZdL-&9$Rml=0+=z5I+7YQV0+!r1T+?AbNq{m5-6QUHgpO@fBx*sCJ
zlM=DO7b61#J1_{B8<vj!{v8bBQ^m!s8GG|}a$vm3+uP}`r}^}&0ke-%3jpa^@i52B
zHqGuh2^f}s>RS<X-(LiQ0OBE$4W09O{ez8_ytCQpHC&4B>p&wd%WQ21EGv+Ml9tkZ
zsl!6~kfP|arP~C7>FH<6Z|FLS3pEoMB-vY_`2@lQ>DruHYEQpg!QOGSl)Qa>#AnIq
z7VDVI&Kwx)Ose@^vNqi&Ndt83IQ7S*5U6{Z4_G{+irHAYM@!yqAdtzRSLsUm1K&06
zA>=xo{q6k||9m0)IW-7amWxT$`se|Rdc>lsslTiF|3piyIFbhY-V}|(Acz3GKV+n@
zUR8CW&2`10X7!JSv~*C933G-Z2qK|8S5W8y{d*17tzcxLzaOcUmmoG=>U3EXf4+TE
z<DIi%dL>l(mrR8<k!_32VQ}snDHXjQ{+=6soY!Ygl;(3s2HZ5Qi9tyDah$7l&9k;p
zlZa&}NQoq8b_NHA2t{h<ysCS5vimfYl*UZx6-(-9Xkv~L+@DMsw8V%=vSnJoF|Ud5
z^-dc!ywCaQSN!Pt?=OLX(1Uzfyb*?C_I4%`UlE9z@xO(d2%Q8{<>`UH0vcpQL_|`M
z>>E*~N6dlm0PI>mzz)a+32kyb=KG<8q->3KZ`olPGDmU^ZP%pSuUYo8^l@OIXblXB
zeV_)z7AChA+p3i6d^pBl-D#v*nFtn1JQS4JvCj@9{~sZfHmWX?Q)&_W%<VIkgN{Xg
z&Mk;U>`+f*S#ohtb9&59B14&Oo+`Y<=r*Go^p=WDR(eO(xOtYH!!Dzgn@NMOhX<Zq
z)Led5)kT|MV(K)7IWo-5Z4iYSNeSx3iURG?1OA5FGyty<em<6v1lUUXTwTM|bmhai
z9!O(=t-H1Wv}?rZ=wOFXUlXA5(A(-sZc+tvKlcS>u*}v5`O^DtU)wR|MTA@>rZ*(b
zCFp_gSctlJcfSv<!GFSmENn+j;5(1zR-=Ol`A35>twMagn)Jy~3vl+$BA$uKj1s?b
zb2WFV6XuPp7t8oQZk3glrB|XsfSzD1XOr%ddjGxBn_XOfeF%16sAV4p3{xEG+S94r
z(zL*$X|b(Oejvk(j;2K%y>}a;#^H>)=gb?9XT*J+4_89eirHn_sDP@gi-<h;M{hYt
z6c;pQ#|^B7<fG0A0u8CbHTaI^YRrG~^BvnLW!BuGn!mO+YRcm%$OmtXgLe8rY$3Z4
zfQOngl{^-gMlhHL;#jIo&cHnw7<d)dmWkK^>oT)i@vS{<r`L!B7Y`5EK6~?U8|c)8
zwQ^w6$UPn89>9%3U=WaFr7qMu-KC+YxuxD#CO{0xKq#H)dpgN9#J$c8_zjYVN2+u<
zyffI)ueG?m?vQZ>?4>Vx0*?uG&8$+t<@30k@TW^tV1GYL_@dFHh&#S{izkvP9{M8R
z{uoyw(c!l}xfy@p5a$pITxI^h!UXHGZ$E_=vqp8?D;f^98N-I;h0;7@#B><y82ZbS
zKY#u#>U-kRP$Bg%y{c@^?yFBj-m%20owKJ>I<$j~%2nA+dmL|$gLKTv#Tu$&g0Zte
zDp(2aq9o?`VaSL1QFaOV(Lh96*mE<VHrBJahF1WrF?p_Mby2ZljW1->NYl|l%0-F<
zhaGEsXzV$i_9Jt=3o&+OnS?fDD3I9Bt#=;tgGpyLtxvLIB8Y4Icq^nHbopxEQ?~qq
zrzyteV;k?;F2u<=7k7sj@AOY;aKNoRDe<sxevx6vN8&=oWxO&xFzpU;rYe&#rIH7X
zE7^%XJN{mo6$t`A<qABK%$+q~C=>^A{i;cu1$U9F4YQg7$~C|EwY$iN9To&vX<)Um
zgYhUMBRYy^>8w|zQwapkXudz+u5NCA26LuqX=$mc5p4z^sQdHH@q!G2Qk+!N`foRL
zk@f&v&Y<)YV&iF1-B~sQ?j-(1?|psjPe6JN28b+c0?B*!h{AGayo21PTUWSCV@qXI
z4-WH^hon)c>Eke~rg0#(mMa2A0sp`C+qw+1&i3tEy{pSO_IdJeNlFq9ENXsEa9cMB
z2OdVchDfm8-NEzFAsi^wS)(P8%SI2}vvAUjuu-P3o2M8Q<%65x!xXvKB&<@mnI#Bh
zUl)vkv0BoIUbWPiOkQgR`qwi7hk&Zl5e~#AF#B!0{quO3KA>E7A=4QTAMRaT0koN?
znv=-fSGnu?;@^a6`8W=k+5^KJN<gzf4sGnc&Q)}FaT#pbc+902yB#j|sHaa=+x4)u
zzGoSpDATj@obyuJPvtcn_@=Cj56OzWoMM$q7af?+zZ*talGyvr^m2v1hvLg90fJ@r
ziqcv$d#PQZ3Z0253p<;ZG>JzQ#aD=O$iK|Bh&cFh>tfoo8dHcmU)h6yK|bDCT`ehE
zrq9``wq}9_LH01xkcrhesvK1l<?W1XuxvIQDwTC0|9wdO*0%TWLZPVsL5`Fj(6|+I
z9gCr~>wQcptnGu~+z=Mc2Wp*eUM1k^XFLu73(%<}Iosd=T(o(>?T4mQs1yrPZekYb
z(b+0muU~Tt2!Mm2mA0bGN?bIrj_hk-$O>ermvc)E>f{bCuC($t8NsAZBkXW{E1)s#
zF#T8CNkHCSbb{Zy2c#??PLEgLTU8adQyZ1XWUSR18Y&g?k9XR`1lgVq1W%ow)r{`M
z)8D%t;xwVGCd$KQ_ze|`vm_-d>`0Jgib!DzS7$CBmN9R1rC=CYuXtb~*qm)EfD$#J
z^NTFjryG=9ARAl-E5TT|1H_>3Z`W-@d8EwvRW&rgSZz;vXxdDLTEU2W6!wSR3Coh7
z<XWqKKmYFTvg`H(X2i_I#6l!zZM`LAU(&rqN=mwa<^^nE6KybY)LTCizx~XX$RoXv
zmyd6C&kaOP#4^}ZdVq2Kfx7-|IVGBHwh=hw0;xddxDR7zt1#B+D8k=*OdQp1UL`ZZ
z-3d>nvi2^|4uH=r)K|_@pvqLKZ4Dw^M{!~3$QFWvvCGWR5IVEwG-1UP)q$2>oDR%Y
z@F1nK?19s$6*L$>=$6KYgi?Y4aj_5$nQ;MR##mHW@$=_TAQ%<m!=;+xCSMB*3*kQ#
zwKaC02}|(F#-iuWAeWV@g#B7eCM_unG$*E%x%1)lQ96P3wAOFxL|r<uM+YE>9{kiL
z^{IGpr!c2Yl>%9Qer~ErB?gqwO8mA@oZisMBcbl&7d?;&t*9b$@{r0FxEy`ceHi%l
zHkNxD)x(Ec<t7U17p0}ov|$m~hp^nXfRWg5kh`X?5CYgihZUqf9g8yAip-w;ugP|u
z#&Y0ff{<><!b4KhiAHbF8N2Ed<csY}iBv-{wRdZ>x=&4kbLj8izaXGLC>+=_2P7ge
zaSXci3J<-%ldG%gmE8a(@GuTNt|BP}9X9;1KJ<@_#KXnS3|Oeb#s&tN%`g7RWx0><
z`~?*nOagR@Il8(|*>|WM{U8PIer>4F+qVz|v>U}-<~ac{1W6iEJ8jc6o6fDl$>WkY
zp?=;a$*t$*3OFcH7)TwzW&adUQrNsbm^rWtCX4}^AEde_3~2><$yN1Zkn<)&G{QCf
zXH$LtY2{kxW0|3HL#JJbQ_f6nH%;CIx5I2|9;jpA4*Ho`x{4&AWU>kcdcjPH!yn^B
z-@cJoji9y2+_P6`mNs(890)#=ikXo~2|1pjv9!Rb^?5<?w-N#QmX5Ml$vrMi+`%<3
zAIHGJ0K*C`bNv>TNz5!>BKo$rFdOvbe(2#Phz31C%C$=J@`YOE^qf|9eB^v+Q3>&F
z24Ky*$eM?e+$VCH7b%9p@EKTR=s8n7Na_`-r^sf5z!8Y2(#HG-L(eCh<8kpn-x(Nu
zpx^u&ch+g-;N;}w;1DY{0W9$z++0pj5z|FsO5`&3m<*Z0Tyu6~YfeDb1p?@D6%{%8
z*MT~fJTOd_&3hmC<)^1^Pwr>tDLKR)=-0bFBMbrGZvckEMl9}yw1V&(#o~YI^3gL)
z<Y<p>bpo|7I1A^8xuU{cT&gk)w;o+urU02nHakS(L!%B+SawC<6A<zP`5h435%Ge5
zR8^@LjC2{5X|hO(j-Wst_G_7={~jX<X1Tbi6ZkK5L?w>uCd<%~Qc&Qcqk>W3Jk|3n
zpWLCBrt18W1t6lC-_UsBmcL?U!*=&9yAG#o>_J%H8A9Jg-lVFsAv%L7(qG)JZTnEJ
zRDdqwnR1^}#^3u6jCTof1T6|+G}ICGl*{yV>^pRnQI6yc`xkTlG|*co@;@o*@Yzf>
zP79cNq!jfLN|L7(_82xPA0PoK9Ha&Q{9AVRV~S>WoGio!7(E8L01!$AS#g!TuR|VS
zcps-lXQk;we_1Aqm%6i5XmPp*2#iz;2yfL>Ye&VU{QzmB!yGsRBrbzm1|}!TBEa~M
z^v1>pXCD|=`)^3CL8=our1n2EYCUS<kDg91fa5dB`wmg4<v8CO0=`gU>iE&4M-^8f
z@W2@sK8!$62{=6=$E(vRo`OUOLyT|-jX0cLU4eoNI^kP(&A;il0gPDnC?A7>MgXsh
zQ=Kb&@iW{mc?^u<)zv2vw16_;Lm-@AS{lRZPV(?!Mtb_->i*GT(fdq58U(?Bhpz6H
zlYrbdt2L;Vz<UMdcTHjQW1GP^2t>cyLcGaW>8p>fP7J;rG&PVM-L@Ayoo%EB)2p`-
z4R!OI%F%8@8uIx=504}=VgG`sp<<n@d|&S1RtSJ^q!_SXr`CXc&o;*aZY3SjCMA23
z;XhxX`E-M<3{0SaBoC;}AE-gzco6sQM@<$z5c&e5?D7L}yg{mmHfGnOfkE-}Z=vDA
zK`{o-yV`gHsjSsHPt+#P&thWkdfX{1DtgMiBn@&vySsKmFr{nVv=(q@klB_kcH|h|
z!D5VzQ5xg}2CYeOP@e<ccA1bVMyA5&>KqwD-XQu!jpa{JqoW7rB-ChCS%`p6ZpZhT
z_x9r!_8LFr7$y}}u3sxe33^4BWhu&K)AdHv;ObQt)zzwDl(4#nM$pKoWz$kkGj!T`
zSoXj~hgn%pPR`kxTUVF-|JKx(=-Zj|K@$Pf!rUY{fV&QuRa9(0iCh3_Z%`h=o&#K5
zhEAnm1n_zKYPNjueb@)XJ+X)bM^}tZp|y=a`*-kcqJW2{(@>f_#!H|cA3ay^axKAB
z7n<YGQvJC^?f{2J8uAHeWxuZ_hdzl1HSd@*4R6*){wY^)BOA2&&kdkS?3P@3SfgXZ
zPC`mX=C}byl=g)^j}K2yz+|)(?cLf`Jr<&BU>OY$zlEX_@LSlyNV=OFC@a942jNu$
ztmk_uLaBRTR=6gc*Bm0p4sekB38oFFR(atQ2IDkJ;6!Z`G6aad<04(-Na9K`2Jsz1
z#0a6|Y{{@i<pB<%Mi(-%?x<x2-2i<ar8x+sdWiMC>>1k~w|=bUr0k2>a*+bfXQn=g
zi&SBoCc${cLSIYE9pnn{^QEDZv8lO&KweRi_0OpA11ITDBcRP7U#Rd!&#cH@V%GYb
z7r7toCNo3_JM5EK+RC6WpC2*k<X-?K=Oz1dA)7BDv*xn=Q9*9P%_VMMit~TaDkJQ}
z75nVSWmcHi>Sbr6&4Fgt(9pkyLm2WE6u8RD2jJ^JG-+eLef#$1gvb#S2KwV4=hR!X
z=~%`YC|;vX9<@GcIq(Ozb&EMW5@=b#vUn23onh>b&WVI);kR$UC-sWdO>gbifOnxI
z1Sp9$G0bPqeW|A*2n<F77<lU`)BnfUdq-pa|NrBX>N2uJl2J%yL<%LV5Q(TH2_YjD
zp|bMK3Yn3}EJBiGg%GlmkR-|8*<@$^Zdb3*cbwnvoZt1=`@CNrT+eZT-0!z_hbHT*
z+4bOK^FNtk6-?SgCw*Uw`F?xDg9LuDo-nmcubG0GxKn~EEbH?b6_cJHC5FWKl9ggv
zUfKs}GUyD`1e+1(!)ga|UdSr#=gl^-bY_mL%B<8ZVNv?sJ43mrp^0tgLCS^|gJbn`
zu0pKf0_PoNb9znrb`FSgktu~Ud~DlFK{mfvpx^)DnVpI(_b2w^jra9Vj!WO>fBmYv
zXlbJ(uR{Bqt6W?ZaI89T{gzg#73WfIDLxqUA*XnO^qfc&x-2UBZ;Xk`ee;td?xxt5
z`zBJ!BIevKYg0!j!w;sli_<}6|IcgID}57+yZtqlp>)i6yWYp)65wb-Y{m04zh{S=
z;48ZHN}Mmpj0%wW$BLh_r~9WPWw|f8Xa1)t)~)^xsU7*(AEB?{c)8nQoPdKdc|Zvs
zH*(pYgcB;QneFyYOck`XO+^g*k0<FYxqi?SUeBm#f0W;GGwOm;x^7-h4uo!B+P2cZ
z$s=PgXEND#l*%u5djeh7Dp_BhlwX-DpSh5%XjwpA?%6%#(Q?l!_<hUq6*-~n$A%zR
ztaW#;a(EAg^AUpXJJWk^vo{GKW}Yn#`G4+u3OLFnljHRL2!A$XSR)YpqW9f5W2|v#
z29A5BlN0<hB3I-61OqMp**pu<dd1nvHk+J&w&;opvQiMv^}@o!$&&Roo|Wz`lGdzB
z{&<DwoR9TiHr-3A7%G0OQs0-H`CdBwn-wuj;YG~*ynd_G1K~B4#F`All6DR{m(w;*
zn-H&|pSa+>t+;78r}a9>2Xzt7S53q1H}CQ?PC3dJDDT3Nb^Q66n;w5F(bMD@tjx|j
z{5ZYkubHmPkqa5n9ZXm?y!uV<43Nrl{lzqx5!t2wFfF;Ie)o-AP_ge2<G-5Q+mZb-
z^s93oK&|PcV+q!At&qqDR!r_bzHu+G)8g)1EM!%E#bj;zW(%qVAG_@WA6(%Ge9}0=
zF=6-G&FR|C6A1$&te}zpwTW?qPT@6XM`GW2jlz`}5U{d7*5Ty1qq*R~)Ioibnk=?c
zSp&*Xb{_4_KH^6lS7Ofks<83J8Ox3B_&Y^Twcd13R<NPHvj4G?jwSWun~&YtIfbbC
zCHbVRe6xS5akWj~tc~Wc{b6#c5UCbx(g#RtPp;NxWfzqU$t{S(%Mhc@qB<F4-<j7o
zxzN(mB4&_(K-!1v%|~BGoqge9hpu~_f!N&II<0FG@D%YxJxN2ZNP!ooNglaGQD#5e
zVrbBJaFw5BZT|k<YolYN&lWHSg4#tJTx|mZYue5t%}6y^#fo9W%0T@9O^)QZ)y<L;
z9(q*FDU5q{cAeJ@{yEIIKTH3^z9cybu7c|>9~*rIJ!U#GGYo2$Cbr(XbM0_6j*oQR
zG!O#B@~VL-dxpfxQ>V(Uw|<1<G3Ll^8onI60rlr+qz6UjCp!EY_lb(eiJMk=RTmKJ
zZlU~`hE!ClypC?5c#)j^$5A%EA*E@UQR!N)vx!VwzQF_}5J13e^4dy}3-kK5aOycu
z5*5-kI`zeHjOvf`2gQ{4JdTKnm$9!RdTd>n1|rbLpfMkI7@7NvLpt@`LsL8RL`1Lv
zgijC}f#iDArcJP<OfFz>i`5)*I06~8TBxIxyH09<=?FhA5bw?DHEWvXPDKsM4X@HX
zWfx`AtyS4)uYbu)xLYxl9gz``auoP2de*F<@1ubGJ3-sm*AtpJ$aQoiI>&S=?$d5J
zRPJhO^4t@@=bamOZC%~fOg+d&fI2zK#q+WG4<TJ+c)0zbgoI_(o3ykUM_JefzuXE`
zDW{qm%V^;N(E{6QbF+6->w^G;=Vvg@?)K7*mrFk*BzP-B4u7R>>qDvgAP&ML=xzOY
z)y%jFvmIN7mk;#ZSsLH{$vo)>o3okOtG+vjqeD(U{WE;>3Fk{W&Cw%O9}bW#<_?|6
za{PGw_weu)w%DxJ>lj?w{0j`;Gz{(7zFj=Jj#c?$`VX{7AdoO;X)dbZ3kB+@PYe$$
zlBOMtrp8D6Lc_v#N7NoYas&oV!_+unY6J*<K*i5Yt^+tS$;{8qg@k%Dr8}hOm@Q&s
z>ndj2pY4jcdKI1%^^_)Dc3D|?VVbWAFjFfDNl9T26k{Y+WQCIO?I=ul8QXJ4lD@4s
z>n~)iGqk?WX;y06ZE;GI7hmY}hhGt*VY@GmTu^j*Naf9{HSc_{gjn8sL!#=ZxOKhj
z1cOw9pPT%|=qMT<WfBwBqxg7v)2e`wfkd~GlJxY`DNQ{ko~ZR@dlQrIQQwQ<+Y-qp
zdHRwtn-^n5ZDkKjetrnQjHD$0qmv%H`)O5yqM$=l;ML2$*6a{GphOM>SXgFKvMY$$
z_J@Hnp94aCZX$ROvtP0cuW`F9H-jcE{Q4eE-W0ZK>fawohVoy@YfCB(-gvt<kMef(
zZBEMD^gbb<eKVRq@yTeei0;wh&#e0CDtC8uwcn`g`TMGR$7(~zO%|(lEJ2;Inilu&
z9aDk;OGNK2ugR0Wq7j#LbasW+fKNT%G>duE<E+XC3ZnG;=;Wl0m6amNyZ=kM?d7|U
zQdGE-8i{KDnW*AqchlI<%m)uPr!}i<vz~5ZRXzrX%iT^V3)yTB<9&;+W!gt>WGC?L
zzmfYbX!z%@GL!5-f2f`I@r4oLi<jfnea{MVX&&@76JMBE-sr;+s`^1e&Zhsc&UTma
z+hYaWi0Ss1xw(fJNxWX>Um}@~spfzrvy_x>bZ~KTxp?uS>_RcFsv0BJj?krzwq+;P
z)XI+b%{t$6b90;83rNPU;Ed~~zZ>uuGp5u~casC7$in{dJ3d?~R?a{}$gC@TF=PXg
zS^>MSf5WN1+zGz4uGpYq?u(53h*IETVwWF<)XQ+qkL14oEbzS```p=MH%3x+-l;HQ
z>Sf(^)aSfTy=Z)(^equXWY28&+`v~cG-K*?`;h&hMzNt*s2MU&u3Wh?o0RRY(A{4X
zrkfVd>nmeFSdnGliIbFWGe|pIX-anm9i7lKLSO3ZgLEdC_{<JR-#*eofcx#5gt!rt
z5+dxF*x8(A<KPP~D^pzA!Mk@a_2m6+{>PiPbPN2&lZS5Z!-r1HfRZp`;<W*!uBl1y
zRvHvUk1!Xzf~nN@_Vz}~2Y$ymDU2lnJBf(YnMvg4<eXs*fApxVZ`we#rL>fylR}Km
z|2d+YC^QIE^!qJ`$d!^D+Ty@4q?3GZU!D034oSb!WSnQ-O74u#%j0>I??>MZb9YDy
z^4%vU3WmsjoxXTl8k&dx{_;!xEQ7E0gx|FRh(X;QM;Mp>O5R5CY3GtC?L7TqKEX42
zUr{!Lg$3OYHv7?Zqkm+oF_{m&N{k8AGAj8d_r<p{9F~^$&C6wHWv$41<kc`VG%(Pk
z=Hg$;T*M+rrL2uu8QrvPe3I5bUW+H*jY`H`H0OM8N|9@Vo{kO&fv&Eu3LW{W!RNTB
zI0v0So5;tKPaJ|_!*B+{G9%Lqd`Z(nOG^tk>k}P!(<Qi82({2fdH97ei`~0dT;3Cj
za|Zs5sl$g|Ka(RKemo_gs516C+*>IaJMe>odF16}W!oX$`Spv_Rb7Zx_uRSCE@3=?
z#+;QqN@0!0)d(X!|M>AE+(<L8(%vYM#Mj?vw~ir%5%qf#<#DJ!7H&H&H8r&u;x1Je
z<j#jLeI2<=yQbA|cY5w*zHE$2koqI%O_ch(`uIHNYPPV91{|Jk++BX+^W(9cm@%<L
zX@3r-lSw+2j)-cjFnXLwHN~oo<2tg=-0O^!lhd6$^N}4m<$D<s&2&wYQ<j);II{fr
zFCx_{O{}b~X{iI4ln~1`euwBwwfScSf{+`)bkNfa>H3u`n~*J$ufnRxh&K>n8@#0G
zx*shqs@mu5?XwklNkr{e47W`VNMV3;Ik<WU7uR(gn-n1w=<R_jhzUY1fBLkh$5un|
ztcAs%(@kK|WMpJq<>c#CH&O)nmG=l=et#~bWnpfPkcrjIXhG&grO7R4=l3d!?(Nrc
zElX!{P7!m>7p$TH{lTi+axJ%KD=hPnQ%EBtZ*XuBX6=F4yPbYBLH8Bi{#@z)NhJS$
zBfEAX`dgC3;QH>=fW3K;)VtDeQSxnR<jhTcX&Lp7PE(sAnDI~kaL<)reS%)Ql=czI
z28Ob+JvS1rA3i#trKX`H)FNbKXc;AEKS)>4dtP7v;f3hn)WX6Ye6UYF5xvY=Bw>`+
zOh#gc9ir+6r-WL@@1(qVVQ{N*+m}BZl60H^dlwr{B4$q00s}m^j=lXsCEwZsJSBKW
zd#liBa0VtLGH!J4T~5x+W@047W<~P!mWG$LW)k#1O+SD7GNPsN_Vz}{vB+BA-~Qu=
ziuQLro>R1s$gedo+hVwZuI>*926CB>MC?FONeK_%<Iqrv4%4M4jvE&NNO^rl)ngHN
zu6C5|gt3SRa&@G}N!#3Y2Y1M-e0o8MF?grZv+djJh`vVVz2Uo(l5~zJt0$ebOcmyR
z^2}E;wZ*6CNr<Uinm5mLtJh=B$DZ!;PxQ=WG8D^u_n_n3r^DMz_MUAp*_37XsaJ<+
z<&u29a?NqSG)A%HUehv@OreJe5n0?9qe{bv7P#N80B_6GG|v0Sj~_QWn~SQFv{|>2
zs2pX9IHl(t+o{tzLt<ja9F9zz9kLJEyAOQ1kB^UM8<YUDeK#bY78!~uS=!n%GB7t}
z`q1q?cn~ReO#^m^q@+@kk~S?g49yJKIoY}hF1Vq@Y6#NGRgS$!7MaHTLTcGx0}uFX
zYL}+P?m!9mR%6wqVudER^?K7$<^$rArwJD_u6(d#;|3OVb*j;{ia7W;oVQaPBY7Bq
zs4zU!yS4lFk)2AGYb#5Q(LJeeLpmJTGiYD@U}rj1>=a<>#vH2t=v>eX3I1@tR4%oo
zoO2JTl)S!D{ZJ3(_uD9Y$Mv@K(%!vAkrr?F^j<$ST57>G&v}NmP_=rZaa!xVb{DbA
z{h4P$STxR7khE)lyW{L!70cj0K>_~whRKPEs<KG+q=)%wF=YdGo;L3thQBrH_liaF
zc~+R>BxLeCk+?OkVW{uyBF{;BpTy>C<p*nq`ujnh1D`}o;^4gu4r;5~gK9~VJfd3#
z2O33rG*ejkDoip(LyXQE8we2(5eS4rbSAdPq<`Acc9qFpfg;0LR0Y^eo%zA_iTnPf
znSrqCK4q=VbfOw}3S(;rK3<66K&eCfiT7RNBKw<^lr5Xa1518-U#N~{I~Vnf$Hm(H
zeu&%U(5)@zVKvlGxwedied?Qb-Bx<4FG_xB|G5($au!j7@=F%ah-<GqwjAD%9>nfL
zp<B`L?4s)jz&p#Und>sotvHEs78>uxeS@?oR8_Ya$^<ihtQksNxzlG?pnY>>+R;EX
zqPz#(O1_o{@5eU(6A4p&U%oI0SC^;Q_Cg2db>X`Yi+?5X-O;%R4wbT|6{nOmdxEcK
zagT?pHc7IS8a*I_32I)oZ{v4nmRuJpoy7qqE?&^p(P8GN&F6sqmELD#iL98)mxzYn
zU2LKXK>FZ_TY)u`ZnLMJu&Ai$2QWH$H#<0&{a+mSp<jg+#>Srf$!;3PNlzYIo`_t6
z@%#Oh(Fr_&86|6f0UR}UcHYcx&C$O>z~6vSqAatTP~TWE8B(2&rU0Md7(|MpnAqgQ
z{cju*JVrxH+vj}U3-=AF4~2I#R1%jOgWVif7H>r*qYv<Z_z)>=AK*gd^FmRMLl*@5
z*~0w1I)7+FLPGr7Yzl-~H&*)EVGJWD&;6Yph7coH{~o!<%K!bH*mFtj>Aggz?61M2
zVetmjo2O8IfUF*I`27{O`JG9vI20L`w4bzkQuCTU(%nGqZM11bW7F-u^xsqdwro#t
zWEN(N`f0e){r<@vhqk-Kd|Gv>3hESpxNBidg_6%@?c>j7b`sU$Xc0ydm6n`p_3Moc
z4E0+!Zzf0{XVRKCd%E*Oua}Ijv`UTxaCCF?)m;5p%{?r1s_ZfQMQON%RGbu`Pt4U%
zd~HT6_2#ZCi=!m!A4ZG{T4WDmrkCit0z*6)AaWQce2Z?-%$o|z;!z@VILnhIJg#ra
zI-680B*?WG*w-O1!lQ+-hva$;tqpBwABrzN`5-D(Veo+$i+Z@^-Me}GN1)=SLEq7V
zG&isIFc3fC)NOE^&-OcZ`z;a+SEqdB%|0_U+U0!uUs?hgt{|vaVSLF3U`)O>Bd^j>
zfv5K7wRLoL5ePcKuWcvRR|yvGJ!+v^vePG~%8{U4Jon=gobN_m$lT@dIo7G=_6=`Z
z7fjK(8Gozg*D|k^KN!s@ku!-#5>NM@qQNNZ-lCl(72mUZ1NNlUv1q5;sa%(({k5o+
zbTvhjMk7m;#uBpp#^rffn!htjyV@E=gcPoqeg0Olro?yX&^DsQori<NcRH!LyPJ27
z@5RX-x`u|x+(T7tzJ}L^Gw?wT2m56(131XCUl{Cq<a+`^X<qhr-LkfRo{7^`-and_
zEy=fnVtR2Vem%gn7V`AK9a5rtZ;c%t7vm9lg3E=>rVlf*lQ;}cWT>pb__A4551myJ
z0c+sQ%#2mk%^NpXk;V4&*RS{R>bAfS1L<T{Ab<j!qG^@rIs^(KN%BplKRd1WM>t+^
zhcT>otqe+CxO}_${q51nda!mE-Wd2Lu99J={TJRmR!@SW7}-NV5D1JJfaC#<?4I5g
z#yGdX!wS4_aVgt~n}w%EBPQGOCcW7Wk<Qs5zrKRoWi82x3}?HUIw@Kz4{~BP#CejZ
zZGi8=lltXP!*tLLZpL(*O-Uf)?#y7})621ZWSFrI^M5dIZQG5!%Ek2C2Jx=DDscah
zQ;MhT_vmOLuwq07!kXYLlz$E56cO+Gr)ID&dex1<p$>SSZD_nSkCAF&zOLz*ujSAE
zKQ#F7?;*|qIZw?&uf(eC_BZ2}k#cDN2d(_*cU3<Pl607;6q_6_UAjc>`E0tOsYInz
z@`ZY={14~jCr{A*d7Zg<u@#sNF0!Si9wd5^Va*`IQd6n<_z5<D6pzR1SEro0%`dKb
zLI04~w)3jT&Ul*ECb_-LtN-0vKL}QAcTV*7_9E?|&&HYb92LzPmjl<!_0*|avY!k0
zTxGZ)KJgPIj4H85ZmSEk9YpSB9}kjc3L)2>>M7UYSVE3NkJw2bVr6=yIc=IK!_&O>
z<kHQ(qN4Llt|(9R^xd#$Z$~y%kF6YV%pEuLH}Dh<Jh?OhcpaoE+NPC}3njlm^#t60
zq{2d?BF}c@$81p^0aRc5rsJGP_C_TmQkwju5%|%47QLc?S|}c*2OT$Uu3giAYg|TE
z<%A<$o`5CmU$b89vLNA?#Em}-_r?;pTb$nI-|Z_@FoGWOPt6`Q!u&N6Ms70u_xmsH
zcC6fX;brp(2<E=PLgH^zyjfO-;I=L7oVwc7)`%I-FcCA*czKXnocq<1>@mxzPu{1u
z&KkeI@*GfGS~H0+qou90Q_h@Ne)5E^q$`*aIc+_Sa$&0y0s@;Fh7dH3aY4MW@|2&R
zurAOKyg!a2d3tdIG)0bXIRm4Tr~5W(Cys{%t0wSlCBJ;Uw>qKhU%6_IKV9Ru*D5a2
zN2VskU?<FW&pGdX2y4W9<gj>}p7c7SuKxMqw*8DsWyp`h7{#O{%Wc#A%I0+Fz`y{o
zeADQ|H$K<an$n{nk;|O|=@_tOjdKAA1H$>Vk+1&_85cMd=M!X%nU$|;k$n5784zJW
zeR1D<InP?i_Rfl`CHeIk`TU@t<Xh97^HrE_6zcl@d-wWX{nEGE&vJ$N&d-Ke;%Ht$
z!IJwH?$Tj4XBU@g2<?NTH3YrB!V3jzRcS5%el3+nxr!2Zle100r*F1z3im2{KCf@l
zTKQFVi&arrNT^i%DjxZ4S-D+_jlL}|_66<Yqfgk`*m}2)03I{_QBbzkOX*6*v2g%0
zjbc13Yc1aVHNK<Gm=kXuA+`e@918p>@VAkC6)Bj9pgtr$|Mu<M)S_>B4_U8)<Aw5|
z7B>?*GWgTS$FB~)^z<f{G_<r-NlUJBMfYWYB6xZRRlaxIhM!0&d0@Xye)j9}p4be7
zf~;rHDEQ#h$Q0TgQ2A$|_W0nN$fWi(-E^9ZH=@GBy$Qnjp;PeP7$dhO0VT`1k!DP-
zG*%jnCj#sT$ybT5Q(zW7;799KnC~Gp{DH<n;;EgzJuNLQpg{sbT1FB}LuHOkN=!gl
zRtabS!`rK`fB80e>+nrQ>Qf~eb9YX!{EB<?Q)&PToUE*@_wSE9On{dTRoCK6B*FA&
z$3y-aL=eVmq9syf#7*3Oa8lBz953XX^mGBWtjx??okdRCoM^9rN<`ES%sG2*y!<c$
z2b}HW9Svr;F?mC;THa%6ZH)+#GVk+<6v6Y#s6^?`Tc3NL>3?6_l=Sqie3JV5z2pMj
zHy)B~<w=wwMIAp$4xu2(PLX_3h61tzn~Y3sefp)m4>%8z9blAcqKr>P0i$fumoFNE
z=JeV+I^S?LzJ5K>dFufwBdcCGjc%QCd)%*k27Nhg;|O0MJ;x*O8yF#S>F|ff-Cib(
zc;xIT9fhT&$l+J974&6E?jb3}p9vSQmDxmvIX~^NZ}$0PN&BTQ7)3v8zs<3FJ8Qv@
zV|${m2`NzJxb}A>HSK)}yWRNM7|0-K*lGv<uC5_g5LcyWkxO;5r{v=<{LYc#VMzWq
zdnzV2cE7mmuzo>Im9n~Ic5<CvosDHg<EpMEI|}(udAFix&lCt@w&hi16~2K1Ky!~z
zd71KAWRUBm2E+*n?DF6RxRREZw(3R>qQ$+lrIVMG{0@ZJD2Q#_Hi3kY2mc^y_4dTd
z`hkOav&1hMN6U(ip=k#H#eJz~&W5}@Y;^lj#620^=VTCH!QsnD<=EJmbm0+X9g<^g
z6A9H|(2(ytCWJ(Q<;Wq{1}w2Ea!qh@;zQz-izf!cFLa^<foIBEU7e<hW0C@mL1T(B
z(kx>7f_FU;3+dzBlbwFZF<aqnNHJYw<yy@iCCZXtlv7*f;_dqwS}z!Gdnu<Fn~_-8
zB`n7G{pssayZMimCtPp$nG#BWOO~GUzGl9dp?^?VbMMG?-KW(*do`D9raMokHN!TE
ze5~MrfDnEwNyxyzUL^r2`|<iBvmgKb&64msQF*bc;gXm2haCn`RGmNlZQkcu<-AuZ
zs+HDR#|&BC$Zab-);~D7kM-fm$ThF8(2H!Nsk9W^oe0i*zX&0MXx|q?uRvyFkAV#0
zX^^V`tSrP}GXUqbW|TmIiN>n_KTm^muFJPBsoekL9rNnPs$yxFSbV{a#(-s+*OAx$
zkO9bUiJqCIlc6WX*(Y|^of0d~ffZVIM@8Q>j`s^^&YaQH%SV^Lz5rUR{s-H2yzAr_
znA|ZyG?PuNul<X*bD2W2#})>;nJ)(s&D{0o{m8y{W*ZM~7G-UIH9?1D{`a>6!<KSX
z{pZ6%Ls8d+=Yp%3YqXA#y34-vJ+_Lmwoz#5ynj;u7#JC0rI^sKdl$ZL@I7mpCQxd=
zBYdP&Bk=g*dAD4-+w+kdWFJWoK^)PKqN3hIw~xm-ti~)?zt}JXGo@Vtrb4>{Svbf{
zI`V(p9zF`*!&ds2Txa_D&O=%%Jq07@3bOJXg3j_*q}6Ss@zzbNoTNoz0?Y)cGoW&0
z_vgC0^Qq^K#MWOw+{Ye*ABQyq^72N!j2iiE;)ojTSw6x=1W_H+XLY_*=d$PKTNk>-
zL$=GWmv^ic8m<RlV!s!&>l0V{c5+=7l>(Z%dw03<fr_(p^GKXV`jWKL%h#{R;EXkb
zq-<*RJ^)Oi-t(?<>s32&)K~3bQbPSl@M|m1cC?^$o6C`S^{u@<fOgx^;2?AmVunRT
z{;OBlKJJd8?BldBG~|B6>}@UYUEX%Edd$KEK5Y*QZjCMUcMYD73;ORlmZ<vbYs%F{
znXa+JL!oQek55LnFu#g?lCS#W+`jVBa%ET)J^Y=#ghWI{#wJKyOUchxqY~GHe(GG=
zd5-H?Y1G-W)gI!xP01}`{0kA0t#Vo5AN7#{K3uUh<U#HIv9W45vmZ)Gf?QqR-(bs}
z`qu07mMQ<J=ZB8_RvHDhgL6l;jNZ&|sWKEz%4=TfIXQ5wR!f17fqadsKzwc8TyE*|
zDbJ>BZMvEV;K(@Uz#MMOfAV}@RFq|Sw950Y^lo1>UcBnACu|-{ux(={f5^TKz>>&s
zyWe_!4UV6g;>mjqeCPD^d>Kc^#wwhmK2FwOr#xMuS@Sc{ItDir@tVd5Vq~2pOl+jJ
z*CVTzYxaC}aQhvxEtFq=c`O^(9EzX?^n8sw`pgrHc^)cXxWGQ)#d)J)>1x%rT&Rlv
zTE2Qu>_6z^*Xe!J;)+FWc*<I0PQc||LFc*721E(DQ;fe&%%FX@Ni&`%sI&S+^A9Te
z;KXAbb-sm@t1MDg%*6-C*WK)nsd;~3B)@;9SFN#nhV&eq-BhaS!-PBIzs{dM%W`iU
z-y5ZfhzNG><KH9$^YXV`$};oQg*_<f=*L16;#<eT+pdER@xhD``l1<Eb9BjW6%MPB
z=Bh`I*Uqb6XN`;c3_iHZDE&rffGLc|X8qeKu}Zt9X}3Rq6dP7p#W@2tgvV{AReEcZ
zH=C$vpF_Ccqur`t!T_k&gL~r7yEqslU)b*uuF_Q(bgoL4>d`B+f%=oPmd6_j$>#Qx
zZc}lS1t~>s-1Em~>?hy(+-s0i`>-Iol)+Hdw%y5==SD8SNn}KLLxcRpR8>yT!{>R;
zF=H#2J*qb1Yv1>;C)O8;z9WTLF=KSn;{`DIzScjmIUiAH$PaKr%tUEvp80bO#~T@B
z@9zoQsAzNxA~zJ72U^R-I_zCl@&~Lu-vF7|M9aqOh2T@y{!fp(tB|@pGgt@6hu0UA
z?7_i3n!cEO*h>~Sy`j#vYDx~#85b*_{Z{5FUUxfiAxw)(Pf1>xZu9X^#O6INRAKdf
z)0hMuZr|?t8-kf$yUBel%;;fz{jIsGi-q*3$2s}lXZR2+Ek5V8F8sc=eDc}E^%wiA
zltSIb<c8W^tyfG7uWt>A$(UTOrV^F?mUZvSL@=+|G4g}X2o7x}H6_j&vBBtH$z`<V
z^4+?4`7W@z`2R3PsE|Z1X7M{ax{l|zO#I5N?VEmWP=F>rUU=_boX)^;$Q*@*1(9-+
z-|k2(-bONYwC#ccXjZ^E?=2bt3!c^1&O{b5u8L5&da>a;dB;T8hkK;whpd0FZrj#V
zPwnA|pZ`SoBC<dsh|9)Ai2>EZ!Xhs>m)tp9-n^+oLfZR{s(`)m?SJ1*2W7u~Yl{&Q
z3oA6C=>6zT(ApsP5rRsrXmCTu9UAG9C092?LqjMOXQP8~St~q`Q^;>aymaez71G}D
z&j4;px@-<`M3(Ry+g(gY<JN)*djo0%erBW`fDBIFOIXl09}J?1J7gi3=Bi1la%A7D
zcU9|U%Uz$J%5SIKIwUQ%X1p(Zv!MwmF}r@MiKR8Hcm87A7JAk+o5)$de}|dLEPSF<
zODn28_r6<7%HfSJOF^?glUK4DxO0xS4UQx-BLpIF2!c2az{Ww@datns>2*RvFXp-D
zLf6KB<)U|6|D6~N44RxrUxEN2!;HI_Q22_^>*~5qm2Dx)+CE`wbL}p5$99P087=Dr
zV*diNnO*$>4Tzb2X}Vg}2i_&6#_Sh>uSydqT=B?06d>|HyEEB&%$!0VI7uk?>?LFB
z)pG|p1EnK@yj3j=Kv<C7K&A2UYx5t;K*n7prRQga_1<=alEKM@lu+ODf$K!OlF;rw
z^0d{TYlqy{by;(Z`uh_&>zCH2ThxLfbNbYm{Qh=W*Z9@Mx*S6P4Jb2;(xf2-8j&4n
zFarZB0b;3}cDeA*_@vuHVN?`ny6(}z0rFkNrOR=iL-oAwx^($&$N5W_mH`}+TPd(V
z!4TX+l|VSnkaxR4JCxUPrw!i|K(?Koos-kj!iR-oa+)B-2R#CTpk>Y*8Lj^`4Gxcp
z0O8<U*^2-~^7^GWZzeDU#LsG(eyloF|Ki09NaL`^U<&}hboC@??m@-ukh=LgHMIc4
zjA>Dyqx_PZDpiAy2bIj7F#?oPpRF{RR0E89@-`kkT-u6x=^?XQx2C-!lF9F{h$w#8
z`1gJvezj0u61F2C9~ry~;vNNZLO4uwn*VIyz8$m^tUOF_ZWXSsFCJ`e=SVm5FcvHF
zurFUovHKi#m|@Y-g?dTpl)q2zaZ9BPR_VO<`nVF}fb)(R?>IFb8V)hHCvu%zrvg@T
zCfjc*e>8QBN)}<GC6gx))=$xiy7H^hZ{4cK&ny~o^p3Ty`*x_omRCu91qO1(PoI`m
zklFPAO{OQW)$XBavi$bs>lR9PawZRg`_R<l!yTOR>=}7%MFWD%LiF2;cIEzsnbJI)
z&+A)LQ|tA2&~(^J^Q$3_p^<V?2oY0d#!y2eh@6qcFCc(p8Z9&`o9_S%qmo>FMsaZm
zwtjTnJRBVx69fN?%rF1qpUYh(guGhl|C_UgbV2EPDeYEtNEn$KEgcBZBtHrY3Lsv|
z;=U2k{70M8#{Is_Q1Htt=Re(U`aHVFgLw6vHjU+u>P3Bu*Xa#eo)6eroJg%QN#g5&
zt2O(%c4wevM1yQk?d8rgH7+`GeWb{jJ*{c5iVgsXm55mbtU1zl{htueV(2<u8DZ!=
z{9uP@(`vlN-}eN@DKA9ns-HU5G&+ZxcP+O{u0gHI;l_=0!wd#F-|`M9>A!#fj;>@%
zq2*>?{E^!a_l};{#;z$u0VAa!4lh}5vHG>0^s3-f&(0iRB>fymK=H&KGO(u=C+lbA
z+j|xm{Iy&%k%}vSOYT>9x3Q^dV#nHPLRy{COFfC)DhjbS%2o4XHZW9^xsiKjXZq|{
zM!Tcb+7h8DSu_uQ`@Q;Oe1o_pwTDKTT^Kg>Ye(}^GP%B~7|+%m`@{dn!LFKo8lqD?
z9>!R@KA%@|Z?VT`x&hB5pltp0C*?iA9o}PP#K4AWx5$rJ{W~|^MNA}k_UYhzQ$4ws
z%gH0jL6^B@DmFHDao_VZFB>hVn8i|>T3cGs`NHu5l<lN(03$@Y)6T^@sgfchrx8%U
zb?ZOU^Pse}BLvaNh^=fTPMxAX-Gr9{?;fxfTidtA#jy!5(cnkd4bC}3N{bIL;u<po
z<>7_Zd$`y4?D1(Bs<s~mpbXw&<zPf@U48wJYq^cI=0OiaQnCj8_t^HlNaP$j<B^xc
zs6*PlM^s?1{l^VOCqj-Dc#>;2MWi8`t{YZX+gk+C1tUmU>#L(E3&8xkx;nBnkgPpH
z{Pg+~>ak}Y7w%CW(dj&jC`7n*;Sjl)Z+ivwLCoEv`~w1AkQqhFm|>aV>jW_ok)1d0
zIg$)mboiM?F29#_ba|nhhFHj#O~aDWbv+xv{6n<?_-IiSUwLV13A5wj=o|m?-1}~C
z&{*f9ul~o^A)%(((M`^I{eL)5Lc;5ewe^0bPor~e_^`5epLNtSzI17avNqU~ea%u<
z{VFaw5OEoVW<u)I|HH1J-S_}hEl>xAhadWN%_)>0LC(R9`^3a>ueZ+}k25@XLDj^j
zv?nXa$*;IlKt!wHo1~6>kM*9r2gnzRqG=RB;&Qco)rBh;_7UMq6g(-8VXpzvc=D@p
z2QQ#Ukp1&%Hy?F5`ejcBxqP$&*q)ZDhcfLA3bpad6Yqn!B*@B|o~h=cnVvs(rU^n1
zlZR1y@&1SOtDrVEj6?rEMeD7^$5L$SDn|f7kuISR%JE!gh_B$zCInAZM}9I8VjU!g
zhPK3;$WZ}w?Mv+8alO*^5h|}d92*0*5#aJph-KemdhX-H@=M;@oT+!Z`sNlD?I@?F
z3eo<zVKhxGrzed6eW~OYFdSOJESg(4_cGZ~tXHxd{p;k4Fmg8j5;iPygjAmN9HC_u
zPKEyTP=K(~On;a3Brf_HOLCeHLl=;)pL-2Msgo0v`I+**ODZ@D7U@b9ct3sqtV9A$
zUvw{aSl*<5UT78W^ar!?j~{yxAvI=+u{8BE?5m9l#?FNfj>o9SVtA#~K7^I7iEWfg
zy)d{>SEKu*veN>!V@Na4zdyBNe<#I<j0^_89LH%6;*xJhelJb(^nmF<#dd;?R*9*4
z5_7{2_z@l^2!55RcW-HGB9{#zn++=a1uBcql<uY*nsYWCmUrCm;yXogw5pEsOJZ~L
zz3|2xAMMBT9l8*sJVonI%f+XuW-v6iDg1<P!e?zmxi?Se_W2z9Af?KzIREEk{2ohL
z3JL;$Qg;^fB=;gm?WaETh)Z=&=BuuF(ebG`>{dd6b!NZ3BMj98t8?_zk-P%AnO%*g
zQ%7W<W=a_KpPls8Q!Q!v!CW%Tr`EMAINlu03;JW(H?r@<;|p}`oJwyLI^Lg{{qoIH
zM#@#5>6p>Y@1ZiILmyTfqSh0%HHebB{OdjG;XJhYS((w4|1Qu0GK-H5Pyl9thT62}
zi>qhjf9&?k$`;Fz{i3m<Ks>sVx%94sgQS#HUj#3eB2HpN0w6ox&Ivov>Nm0<734Rc
zf4XHMm{A?axu#~E?RuX?LGa;CmQl$<*S;-uW<OM?v`E{NM`(F5roD|Uyu#9JTYZzp
z<iOG8!<0-svBmy$0ht4eF)upbTHbx&(D|M2zmvt_4+wFzx=lZR%meYf*`9wZ$9op*
zpIi}0MON!(1kK!GX999F8k3ZgVlBz^M6^lr9&Vokud&2aoO`pDUrjP44;|tIn|Vu{
z(0&93%f;VsamqA`IusZ%r968Y;E->=qe)&fIGFL(>(^$lDpZGgz&7Lb#_FiD*5QwL
z>kkh6_VLWh70+X%@rE!xc^D@P-bgP-Wo`U1CrIMA2e<R9A*Ub08;2vSZYA*A`JKQ)
z3j(mLPUzv^vFwJgLKlA%BIM(}bkopAcZ&L+2vq*KIBtod5`ru=h0hPn2o5U443Kn{
zCJ@1=p3^^b=DvIt^+IswL-j`r_dVpw9jU!z!n>}BT@2EjzWHQBnfcYOR~ma;FRc6C
zoLb*?n3wcpY^?2#=sm?42dCzaf47)|dK+HS3KlJ*2z44A)!)&Kbu38D#PQJ`i`fHL
zvvUcp@bH{-fx!d@H^!lH(OqBuI;HE@p1aS0FmMc)YM~o9ZiJ;r-A4Kze44mcbPS&o
z$4+x_bJN$v`}EL-P$_oj&T6c~cVAK>Kfgv6{gBmAJw49QtU2!bYfF4d_xJZlLiWyL
z@q>u+)LdG18}$9Ra1R(NL_4VIrt=Eqen4cVgz3TV@&D5fa6~7(gxF+oe511VB6G1}
ziDhF#_j>rAAM)zzy|49-Pb`~_1$VdkhT3_QKc6d}{d6F~N-0L^g~eO-&#eRYAqodN
zI|&70ybsY4+#~N?TP+tYhPt0$xf@??!6J~Pwn=7_m33{_O>MtZ$u0r^u3NVbiw#A5
zZ0stTeRJtrhEm%Ny8^)Yq$VWnO4s#g96`p~wQKNXeiR`SsXiB&O3pP(2C|e)FRg^g
z^cdq%&VO2F?8{x+SL6Ep48d{hNbFKbIjmlT{rz4Q578gy<*{!X2(kEUp&a!p@LNgK
zFvLx&WIbqjI2N~^484B+`Uz5_WJi0uFeB-s{i3pKP6(%6B6<RxdTu50b><dVuVUpP
zTrqiHU1Z}XCMK|rSFmQ?62_TZ4kD~Hhd8&!$=^K^d+#1`kKU);Br_*B*L(fejhi>`
z6c{+ZGTImZ*Y!hZry-WKG|fUVb|J&C&b6Lpl!vdv@V)F;+n|Y^)o(T^#g?_bWg00z
z?H}_!NhbAFiVN$17s(DMKro|ziQ8&5cs2}PFbvRb-5M)>3k?cO*;c3tjBuENyQ(IL
z4F5_{zj2SaZ)8eB`j0l+P+2RKKYQmfY-Zu);=(&=m?0(oAVuh*(<$a#aqT$AB_$t+
zhl`g_ExAG*SC%>n3Q_4V!`|Ie=Id}i@>LAj;nkT+qY19|+~`=uS(22boUW_tD#mCx
z+P(d@t984Uue<w(B35FJSXt7jbMiWaaLJx>A9&aZVjaL_=LO`&lI2z_y$b)g+=`-Y
zzz#NX%*zt%r&7<of~*L!Mj0*Ed2MHPbb5!ez);#?V&~muVqwQG^T0I{$D>_6otfJb
z?*8}MiEcr2xZCSye$N98_N$00CDH)^+SpW|%|h3L50+1@k@t%VJs%eEQ!`>e7K(5^
z$8E=ywUHzAStIxVM-zVQ<=C9_KlGk%DF@nFRTUKs6q_i^kn~`fQ61I^LE*me2eUSZ
zqo<jQe26c2go%e99v)NPn<bvosSXww7xyeTl=B2qw|vm-R{LLK0?;YYOMf^drxeO>
zZ(ZaO%S%eq+1?(kzzek^k~mlcaxG&LgrEco4aMxJl)Q@8Qt3I;=rw}w8&fIbUU|C~
zaRzRlq_izCkiH*pyKdB>9P_bwUHcD_F?j~NI(fW6tx0vB4$#@lsPP3{@zcx9`*mNF
zoRYR-r{eRpG}?7p%!t0!V*g#oNY*g63n;BnoL~mE$~scysLkq~_XsK)MB5YtVKLWj
zRAuOfR394X@*ZgRu9L;728eRFd^wQu$i2TgO3wk`5)~;rIy%_Z3`0Vld3ki_cJ(AX
z%n3cK2ZwtRZ(TLUgYlH9ZwA7#o}Mf54Hl4qY?)g8p+Q-MPPvC?>ek9Zd+Qi8_vB&Z
z4usq^ZBK;$OgCNs(6831x)r+AbDzIt9*kVr^o7QodZB?gw6fjBjamAhgsrXoF*ASS
zt3u32!{R5!#^O<98jHM<WyTyc4Cfu|efRc}x2#AhGBV9zT^c_3Pc;l25|=wP<?QGc
zqMy#4ecwr6pVLrT{=k9mh2_Mtjbi)vpZ$k-_t<uO&;GOW^zNI$nS(SNcLm=3siTUF
zWL4ij5OH(PxJM^;)exUQ(-0T$FU#MZ;4*v`_xLfO<iv$wzY__*duORVPHJkt(@)nn
zkM^{|x;76BIS`IPBiQam-wlt->83>t2=8GiWF*0Oi5Mj?(kL~4Cz!8>ufcc}3}K}-
zmwEQ~5-mR=lUGZ6*S=T(T+e%}H@3;YbzAtkyddJ5wTkZkaXUsOrqIuNo4kd`=}rpE
zZ$etjECqzdJ#)fM33+*dv%+f!$Z29v--!QrF;b~Etn?!=0oyYCGl3m^fD%QCKIl#X
zHg0v!?R<^B&pA1dm$Xom&r<L5;%8QyGIK#z-#&;=GS-dWUEh9IMYM<@yuKn4qhzs{
z79BA6x4ym3k4{2|=CRRHu%6-KYLHExNv7?**XPgb$pKt)P_Td0cn2a9kR0SVFdyKV
z;F`(S6Gon%ZW@LvQ;i2cGo3)iAWb7tnH2DwY%Wt9lc)4BwwlrVRYYw_FZi(1M4R<0
zejUB)<MYI<<KLZ9aUUE14N&)Mzb{fc8r1LPbHFOq-th7-k2pGRw_NVK#vppl@s4wE
z&!`YYs36Z_rPdIH-~1U?zX^Rd$?P7RKwuX4_@OOGFi9*Z*e>S0hotmBY8n2BTIg7A
z(Y$Zu-ou6}ZXgXISHqBfVTTHlm<c!{$K3IlH{w96yl4Mi-R?i)A1K_=u`l|dO!ZL2
z$uRFGb&CnWV?BOH^&(%ema>5pw<tgDJqKa7{h(-$9oyN3q!CymL7r!{xS-$KwqLh9
z{SOoNRVF73dWAr&wr?K{KxMU)kX@PZsGJz=G;Ajb33nQAwa@1^r92D44s&AN_}Yt<
zlu?Hxo@EF0Ddc5k1J=FLnnzAhO-Y_U%ew6rp{{hg<=(%GRT^nbeJoPC$dOx&iDCV{
zR#Q%Sj{+(1Y4wisz-ll1j$3&0Qj5@c(Rf4kia!ycE58~<v#(#jZgehH>$?bL19A20
z;;Tj9=07GUau08!pO{o*<~zUHKVk3!&%jdA1~1C%hpo-UE$R-PFux<a>Y4iZ(WB>0
z!wIhws+%T}G4nUxHn_T}u~8h-@%7(=0s=i_+Z^LhzJ*F9jQ^sxHf-(Ki397IS}5js
zAy(`(^6;P>ysQUrM0Qa5<N`6-QOM4B&cGlhVxjIPEw(xx-~iN9-b13IMCZJiq2_p_
z_UGgd;rX@yzO>nPBZ`La9HyaZ)f^7)r=T_+9UkUW%PK59Ds5LvM*MxMS<_4pP!DaF
zJUu%1SMR3&`SSyJ3Xuw)pzN=NxO<wFjbW$ccg5V+5Mos!+M9y>JGB(LKe4Z6r!XpY
z{#hdNxwn+hIhDw<Ka1jYts81R^@O3OO0Sba6ch#E>zcmlfOikM8{Xr-2*n}-6m@V@
z1UJe(f>Xg_zI}T@US58M87%*+SCKk*Se1KtFJh6Beva?baqdm0n*et6Ss%0AgXS@n
zNKGX(apHHz7Lu&qzk7GMUu-OL^&gUQe=$((K10Ub&b|>#TQxGV{rfj?FmP&sKJv}u
zFcSU#tw5rtq7+=`g62{H!hm}i8mBI-U~=Iw+6=5NukA*r9&(Zdx<uVO+uHmIU&Hed
z_4ZKCNk5ghk4bn*;Nj#d)iREY8Ckuia_;w?i!V|q_}$~Hhs4EYl~@&XIF@}#ofUt#
zoJA!Wb$IW)K{oDb^$nF;V+)C=wyVk>yr9L+B(TKna5Ofd1Sh;495i=VU^HY%C@5$h
zutOomse=tFGfR+{1gh*l-GpU4my4YbDKbJ<Lw^47g&?=Y&~EV@Q4s3|SZ^)u3`YhI
z%=IUT=|J^{3N+;*so@Pz?Y&#Ih}PsP4o)Laj-Vps^^_QPK_BR!MXgKDB31wYkoHhN
zcTMK$2@8f%S2IJ!@P*#qgIDM5+0%kepFR;M&!2A&*nZGUi7&4A;8P8aMW$Gl#No-w
z6rqt?u@XS<ShC(Qgf0@`5m(Nqfj?8%#>)_>2Qp1r8_BU^LPCJ=+)BD`=HabPY4S4?
z$6AxbM&u;O#AL1oc_sZmtGC6axs_G%`{N7!^A3WAWBfitCp9N(zUGH|5nTHevVQjc
z=F#BcDh1Cj9a|p~5$?E(>_KM#$}g{PABaju_9By#foNVrf``-c)7aSbMoZ{NiVbZ}
zo$^OEf9icWPx|eapxAjTzXL90+<ok%8J)xWJ)GLz5k^#sUK)2%m}K82or>7ZeidF#
z)nIOfBVqrBjBIJpz=uU?!(#c)|6VbLl`3{w*}3m$Z*@)rn)1uua^Mhr--$8gFrB1s
zNOJ&I;;r8M5K0StuKar23Ig52!pw}yNYBs^!dh23G~sCwxXGPAch1Ycgv`wTnVD&8
zYqN?%H%?#kTlCS%ef#z$h!I_t%7caP-vgh*kMPtQAS(rEWUZb5&RW6wa;jIENqX(-
zCnoRam>Xl`=PizUCp<{_W25;sN2N=IC+o+P`oLo1Kp>vL_=akmOG-})gA3dYA?_N#
z<8?++QSs9+0ki~wFOW+cQQO?sRx7MsHbv#~DH<z$Jnf|jEUo@MnKETZ!5thC7RI8Z
z^QgQBB@@S&__SHTLel>rOVjiJC|-(QsSgxukkAFrxG|O*IfHLUv?4%>=sHLHquA|p
zk{>;6&#6|lf{+fp!fMPS-<>wPz%MH+D^Z3ssisKJAqfr$peu6jg8$<zs{kYE#4c@J
zUAFpD^2u3Q60+t#<?ig?`ie{(PrD($%*Y6oJxS~qkH;OEjeCNhx6&S0{n`2AZIp|I
zq~`Na)dSNja!VnHE_KEB=Ir*8{Cm^XS@(YJ&TQg9@jr1d7Ikd)<deRSc?UreLf0s4
za`l;%ut<|%t~!gz&c39tkNeepZL1Y;xO(|<hq9LQbX5jPA5Ag`>tp@b=^w;z?F{7E
zRkfU@_xgOAZr~5K$N%R+DGh!iZ-pSLewXMoLUeInP%S~%J5N=$CHZF&7!o*cI@9p=
zE?(Te*4j7SU0jy9%?{!QTn&8kGl_<wQ<|DpX}&G#m)P4e<9l;m0g(3v^3pHv%EO9D
zI)S>PAO1Bnl8c|vs_cIEV#IoB_i^i>wWBMKKOTAcmR_ak)!XcXx7pq<4O)6mOm>m(
zv^4bH$8{U~=4PN;FbN3|Kg7EGrt+0cJ)qILX@^|1j+mG{a<u#Cg!~5xs%)jdm>=YJ
zI>GY6CgA_7y*yk2j3Hy4#mfldI!+3I{8$!pNHhhA^TE*sC<G>d@}3~Fp1ienDa~kg
zWEo1l{s2yceW9VD39Q#krnCdMvb+ycAZef8isxk!0MX(;XRc(G26;p&Hnt=YY`d9h
zyqPvG`c^Vk(Tde$txf)CQ{jk!(GPGGV9kTU|N4U}qZQJUk_Pa%0E!{+d9%(aW|O^~
zcy1q=`GjhQ%wt)|FSk|&?l`4R{`3~@reLRcen#}RL<zcyL`qO8qPM_irGp!a+QzFc
zjwWoGA~I(K7-@pt`ov1j4{(qbfC+a!cmZXR;j9z?%q{Z#H!7$mkm%mn-Y)yUYXo2@
zXy&d4S7*Z&(X}?8mw0d81@d0<-Xu`lYkylx+(3x;{rek!-3;#gMM|wL+a$DmcOsUE
z18W?x1DqpEtco&{Ug%go#zXR`Z9FKnKJcQcd_f|11D6%>!<2-jp*pE_^>#cuTyB&Y
z!|KmJdiFn$AM5MP=o7cG>fY-s^v6p3_`LJye?d)&byNW}cPI?>q!V?2p-|C$y394T
z&v|z4vr`P=_Qk=^aeBpa{}-#Ms?=QsEU|m;<~Sz8rvUs$1H{2n9oMO(s^7Lce=Fwx
zd+8H=rMx(@pD;hj_AcwE%dcSUW&JfC*PWX<B)^fJ)qGoVnK3}xkHZ7iD@j<lhiJl+
zp<{?xuC1k&LCu5Pm`SNIWqMEP<ic(*JQ=$ttavy%A7&H5)r2P_?mzX&KWLUI-@tul
z=w;RBUlJ$VR7r=oB%Rdb6iLg82?vGKNo*0_HC<oYmvtN}h=-kHa5|FNH}0Wviu`A9
zvv!{&`7h~@_iv@*XQOX{eFrvQMkRQ@icXmL0>GCo?ELZLfvH;+vf=S9l}_?9o)DHG
zq(u3`I;Ho&sO~EhBXO`Q$08hP%g^aAVSR_zv2R!L-{aSP>7<11@2*aELd|Vau31kj
z`tPG!<u+Bzdp6yR+UspLi-zlmnj4RFBvPH-(@Eh5WiRPDIS2v9v&7ut-@hG^%@J8e
zVIZL=kX_|cQxGb%<xSDw((nt!S?@$ahZjO4_%{C(D|aU$wG(X??m~C8xIOVYYl*^t
zE_1IC&6MM$)aDWKvmcnK+^q#5{=>#Bj3L<Yjt@YQbLY?Z{LXv%Qt0nsqPPtfuB&}B
zJ^(UIRw^iEcJIR@ZY%kQ8Hm6F+k&D9Wm4lX;ef`2&ns*sj9<n*x_985-{`+vLHFy*
z!sIVCw-4-2DwqGL?fAP==*BKRYx8)K{Wp@t5Cr3*rsj7>2>RE>VJ(8je=^yy=+tj=
z1POZIw(IXk;W^QQE}7hPa>7%84XOi2%4?{Iw>s(X)6%5nKRzR5Wl!`wPc|}r{&fDg
z)j!W$XN_<~ExWKDV`j+W3k)c*0r8T!Qay|vx82jyVu-sGB);wnaVHiOnwsub)<#<|
zoyP^K_J6WDriPc1`(^<ODU!Cav%|rumh=r&DS8*5@>_^T?4HLs3^elndrm$F@)1iY
zp)|{MM!)(W>w5TqFW1zzyteh_mSM%FpCVRKE3RVWGj=}=Bl7<Vav6@zhBOR8_>DgT
zH&_hj)h#@(wqS4onpbqNd-+P;FM(Nr+ikH2D-m+>E3>lfKy9GY1xrn(h)Aj!>02IZ
z26tD9xfUd%2i{`fxKB$RT~{#Zyl&L>^z|AH_UY-BsRCvat>|k~1d$;>H$RWg2K*D6
z7|pt!Vxk}nRlC7`&UYx_SN!v*BhxCcEszn#56csA5|b6Me88!2;sTe1(XnqjQc6re
z{qUhfw~dWUdu$0bYv7R22<7OD&@aL_^q+k0G3Q((xi9xoKeJA5GZjcnUF-H5jms^R
zJ`7nBS$f%nf+hzrP{}XVXpua?ZdYToh$$$y102xrtUwe;wvb?tMhWw+vy4nk>*mU#
z{KRY`Pcc*#tjfyQOU<*T0NoJwLz+c0EC#>G_VWx8X1`trZBd&3xDIzQhU%Y-1~oD6
z=e8+XLdxM-Je^k`ceDjvJY(ode&A&EHX+jJoHZyW=bTe`O)KFn){M_Yasz2z5M##w
zAb|oQ0@J%rg!aSed}3FRT|v_^wgWf8?MeH;K-e2$ahEPq6i8`x-)ZODS_d|X)`-P_
zUplgG)X6=0-s{B2<>9lwN0-v};(T5hPKlSBu3&-ZK-T{2_??I+PwwC}ad5y+OSO`{
zzLv~u?l4+IdU5CS<&M)*9jpfWe!RW`0Rfj_0)a{e|K;$B$!?v>r3q&UP{j`kkceG3
zzW*`B8zt=wKW^&l*D;^5YJ~hTkpASkENHv!m@Jp{#Y4lMibg}iB{TI$JsD`IJb-&?
zvqGC}BhZ`7o1o^;h?W$gE62S;U@)+;8JbzbgT|QSSxf0(SzxvC(IaKN$$^IFgF-$_
zkJ<>vaDVUOaeD_Yf&jF}M~)QSzkkoYwZLSnhY88o70FQz6coNeBiB>5?0(#u983S}
zS-f7n`>Vvny3*{2r~DrK34h%IjC=Sg#vdikx=8Vjm1jcfG!km!dvTIrA{0<#9%LxH
z>Rr63K0mh9pSk<oagh$2l|d@Sw^{aj*+ZG^B8mp0uZUFYl6S}oYT1J}=*{cb0ih~5
z5RsP$`VGTG-u7CvEDTVYdR*V^*4I|~)nH^53jbVHr9yVT^TVA@<Bdh2&z{}CD4+-7
zTSrGA0a*qFY3n}!liTs2W%QVBH-(ZVNYI|`Eyz3n=OwF+5M$VWpXj)Wyf(nhvyQT8
z31wse-B?RT>2>l_QMw}%4jFSEyg)NxDXwX^gw}v5`vuA@{*}AJ+iT6Q#Fxp!W{P~q
zGZvbgUTJCEjL*3m(!Lt(o07jEdmxj-a9?=xix+qV@Rj;@9Sy+nhW*aolxI!9nc#?=
zARyd<U*|RU+F#4nVfD6HA)A9BcVj7u-q6jwht=+ff#(l{<44CVKdBUUoUoJN3+GeJ
z+1d%F2+ux=@6pL~7cbTf&8S+-NJ!`)n2bKE9}f6Cg~b@fpFMkqY5adiag4!F*O6CG
z?*o|=rty={&uB~$YKa)Dp%XLzW4-7qM^;JbBNhNvJHbW_x?~i}mh{L&);LmcZIc50
zDal#B(GfMK?fB+%<ZpieM5L5WC$MukGUlSYRvyS;Nbhu8zOhn1yLCM4pp2R>*$xQj
z4s?M$a<0Jl54Mh3Ms*@aQBkpsfX`oMuM}l>f&#z4pYYbSXCY@SRDwGJU!!wjQbO7=
zMuN|8Sg<_@ZbL5ew~7c4uZuk@B__ro_Ra?SJ1Z;dl2-)<PL7TUibB<bK)?aJZ1Sl%
z{|(!f;&<Rb!m_{)c<Fa}L0Ad=j7`i?wkP7IGg15@HPtWK3e!(?G<n$|TFIW?9?^(V
zFcpbc3s1l(3>7CT8<P^KC`>Sm(r<bnFye`!qoU-yu$^x6%0{ZGu#W==ONc+o8(jMS
zGe%ReKZFzG39m2NNd`(Ec5MtF*q);Q!Byr>Bg@(x9e+JP)f71ui<}Q(YoLARj4j+0
zTjG&}C>=mRLDg5nP7x9M;ll^aZYwj9n|@-w*@~OIA|oW}882UwQxu;*6%Y_WOI}`!
zy4Y53M7%c;EiiC{VjlMqbzdiyEG=CSb1F70><F?s%}8}HpnIK54DJc?P`&FBVLALQ
z2_O6<&OmcqHAnaRZR7N!KHLGx!!XFv(b4gGp}5%F*?BbdF3`iBYg0z^VB;bt>#2KG
z#zl0;E!@@#+8tX9yX2>jiOEQGl#_v~(oI)tLIpC4{?J6%k#Du3Vcdyb7jv(jKYqNN
zyw$>s@v%*$`t)XktPOFzQ~dVr6hzf{qRXyQv|hi1jBCK2K8(K#?Dxc3QBm-EK|Kw3
zCZm!gO~<f9?^mf^5w+fVsn)srT%RQ%l)wnUs3dv%;HB2>{}B%SFTPqzYUG7Y@hiSB
zESxSZ!rUJ6*%oJ_<6Vk3_H2|Hq~DX1s>HuOVQ}z)B!I$2&yn>zGzzFYQl!#Zmz7sX
zyV`F%X4WMgCq3PnUbvoBy3i}OO1K`hT@f08tF`c(`~n<^C7<&U{gQoKOMeDn{By+|
zaLw8py21cP^{;O<7`2OBx}iKdAC)Y?T!V$ZyLH#k3bA4{Lk|Ju0QiQH(?B?+jC<;j
zQrcR*%*aTw(ie?Dh!9yfCK_=nr3qX%S@j^r&B0N+zoC5hC9e<-!SbidQ@T^e(!(b|
zDJGxTmwozPR`>8NxFcJ>Y$?xR#una1K_V)76oTXv*kTxz@63~tfS&ZwQw>}yYpPl2
zV}u;-23Ofn_bnQToAL)|8@fFMvwnPeB*vEe<+*;;SBrHMvCCx_oo??^uO*8EBKF)F
z(E0JDcd9R^NzHAggk>}=Jg&@eF4i~2SMqj6UuHC<sgVA1x|+e|N_y`5G*ZC>4#W%R
z&+~a9wIj~A@n}`wG=i$hYe3Mu5|vIzDMSHPqRxX=vy+Yi48kMt#>MA=Vhc9f`+_Zy
zMY0csM<TrrP!nFG-MgP#$L)S&88t<-B;IpNo&OO_%!AiEjbqTeXT+$IdMf6a>^Fdm
z0EG#u>fz{cbpa^J)<8`N>C0TvRc-5HxL3{k)jR$O5p~<{hw-EP8cK3T-Cs*mp9q4z
zV__ct&cKYXn?K8Vk#pB6mTA`8@1jP6rnF`v^-ev>G_MP^F<?S2CJ?5gA|zw7xpIkp
zpl?0Cm3dz$1vW(TdEvF=Dc)Wm?0xozk)ZA6{a#-!hs(C(oX?;Ah+!?+{Ceh5=UJgJ
zqSO^|KF0FKwu22GjKG9GH#BJdo;vV<WJL_E()akn(D5KoB-8oYwO6`n<0B);H$c9<
zqnhi1q8}ZjbG<DwiO;Q$yznr695Wvllx!udY5qW71w#at;>+f1o@EN5W)S@1{95bL
z7GlLAz+Xc~^awKO)_r6Bz^dF8kyV#<Cs)rM$|tdD_T+~y=k)t4gLS0z$+`8lLldj(
zr%QFEa8Y<4{k%X#{K~a2y1<mGKEI;u$I(ER_PH$$pbegaKv%^6Equ#(bg$+9g%?7j
z;<Vz4=?V90tGq}K8%gOK=X(cNMa>|mh)^M}ZW3&{%gVPLrpQXmCQJ&-fgE6Z%o!;u
zj&X+(1ZJAYC)1hoM{K7kw;!WPX4R^HQc;mBj&i#KF1{Bn>%@Hh(YNU%fBP1W-KfbD
zXni8U(E4!Uv%{FM1n=GrIwXDtk~gK=rNfWAg0n6(DH0+g(UbyTB|;B;rr|ryyOG!v
zw9~?g|6QNRrL04*uihx0Uheyt7Gieso!OqC$sw_&4oGbqLI?gIy52Mz%eD{uJ`80j
zb4ewHLMcSX%raysCCL;jQ6f@i2O>ohN~VaSK^YQKW<`=I36UW)nG-U-+u42J&wAbu
zPwV<}UF%vc&ht3-{XcBmZ*v=$8}xc@J=*8YeOOnQhKg5mKK8w{_v2P2e(DYS<BH24
z7qv3hj}~o;JLhoz`<(&Cz4ni4)!#3gsTPI;5QSnwg*IStV|UK(N_XZu%Xd6kCLGrF
zVKQXrN1JR*h|*955X8T#zv$Pi8iK&=fOo?#H9h8R&2H!z!+8`bzT{I&KR2<ZHGlg?
zP4GW%J5_pQLo*!X&^=tfsQm(Gb*WxRM3#vt^kQFr#g5SSS3?tx7pfxeF7i(BHm&i6
zQwE4X@I0X!-`94XynU-{f@LWJN#zUPK+m^6Q`&;#J~_h@=oKJ2^_y@$>)zkX^uy-}
zg}D1|bk<aq^O+Ijws{Pt8d7L9&X_AT1<!6(VDeHAzO=<G<pF6(?pRwsv2x_Wm{G5e
z_qPs)JU|`zSZ~-fom*xq-Y)QClc?bV_FpxbjAn);?<U*&#AOP$FG7n4{L~8A{q>uq
z*5^BPyqQ9hWx~Ty{&MyUw7$qJCHzoBkW(Kv3DnBDPj=(l1W_!?c8d*wZh!yQ<!<by
zTWf^QU*HLQ>6;<hluQYP(?Yon0M@S7Z5KyCpL(eIpsH8S=dWuwtElA-zAfb8)l!YE
zo3)1qsiHfAqkpQyC6D29&s9r4@eF1{TTd^r-5cEk`sL)^2$UI0XJKYWQ8>R&bz$c8
z!P6NYXS;r7b-rs5dy{>3?90mYh(b5|Y`2!clfn3v@x9}P39eZF@WWkaVMy|%=%2q^
zO*XxZIdYkDX}}&#dg98E(vK1>fCC0f{UXS5qE;uBYj4iB_`p=4nrV+;oJDD#&nVFL
zq5a>PdgT&qI$^VhV!_FGCZ0=7rk|BDD2u)TNtuH)4=!A;gk|^o@mjq4Y-weUS^Og$
zOceC3`-_`8o8$CCLP8$0bU?6g5>A}$@g|i{wdVWr-GE)j7bpgfmS+-DPYs$}8@dz-
zVk9oGxq^<cliCtr%ei1yoR!78QVmdS@J`omS?1X>Eb_XW<+u%={#4|ruJY_`H7zP1
zXZD>Sh+%_;w1Lr_u}4mUjhd<V``!PiC7@B{8g3^jj|061clVSmDdS{&>K$8@wpLug
zCeF(fK6mz$J%TYg5FlcXfSG;h2N!7b5v8qHevnP|{YWXcuNXS2?YhTKJ9WQgX@{g_
zl)XfeT_b<$(SoaEra5i5tXnDdv!JQ$HmjUA)JFsX7P@m?v*5!c!Yc3gxyRS8tw`3B
z+9<(nB1*!`mt!^J4x>)aa8z{k&IoOSz`+W!#pH(8YfN>YKYa>m^n1vqgnEO`>80^j
zsjr=#JX`$hYiomIS3%fURZ<pMTSoseIH=Zk)22-{1bqPgq7Y5Z53uKrzTu^@>ykO-
z9oSp(cdL@}!tO}MoLCb%+0nt;mYp6S49N`5D+U6FM=1vcU)^S|%I7AfK9dY}FY|La
zMa$0hSwFLn*+(O`>+EVD<9H_TK*v?OG;QS;WB#4tvKwyK#Y#TO&7Gyxw(y#d&n(24
zJGXMKBVf2+C|@TAw*lXu3;GB5?c4WvaWPEuG)Yoi{xQ^F1$F-X`2%j1Ecw77QV^<e
zY;@9_ZKg=xSHK5QKv=!=u*cIVZbsAwv}p65&CiefpYQW6S%rL;g#Zh~$Vg*Hctm1i
z=Z_yuIYSla8!JBI<Y8{lchGY~yA9+3fHpyDJ;>t;_&I`UF>Dt!1hgP!t{489qt<91
zN#-SacmU}0udQx%xT>ZWaP*ncGb8H?2*IC#uwy3$zPU=1H@EIum4kmj@Kf!Ku<-~o
zQF}dJ`^Nv~Z@iP1nXqqrjCV}z4y*a}iFuoR-`pa~9mrk~dIMJEzwH@bUxLUr-S&N2
z(X%^)N|hD34=h-NAf68=txb^Sg}<hK4B{z_QPh$fdM<(UQDhlzZF1*Cea|IEMn*6U
z6%QQn9%@L2T>YkoTI6HP%;2Ou7}Oi)itm$@+`lEP%>mJ)kQeOe*OEUE#NG#+2)Gpl
z6Ap}<B^t+YpUZD~+0-7d9{gs*U`So4ZZN-yd7zrv9@2Rq)4#Xjy<JwDyCH+6kxRyy
zN`lwVM@!h+pMqTe6ox41A+EqK4#_@JSJIn7XTluq-Sns3PoAtCP={FqG*RIY$W=_=
zy4v5RQh)cu5;#q+N^Ty?A1I^!Uk+pA2S9~L2tech?`ZtxKHKKV9Z4hOH-{+Ane4i!
z=JN{5x}~M11_tpig;!kecim_oHQ+NrD})+VoM+v3daN`z6MRwCDUBMiAUIsScs4-~
zEul0>Js{P)9Y21&!Sb4ACQz>QFI+DjEoD&Q4eOVdNnd<Rb-hP&+C;g4x;|E7B=vN1
zR#t<RI9kyMq`{LT!uYzV2>5Et0@k2Ha?D3K=i1av<POA)N;t+0jAbnqzj-4oBeRJ(
zg7sRq3Dz|qZ2ZlH)?=HAWwPohDje=AhzQ4KJrxubXoBC{OhIM%XhE9ZD2Jje?-s%K
z9ju-)oIFI(C|iWw%jNkdj(m@sWKTo#hHzJl_Woz%*{fTA2<+PT@7Tt7vh#-fAsOtr
zDc-Z&>nz3TgUa#oU_}i+Ok*S4$(n6HXBKXA%TXUM-pF_m&g9?<q)TI43tD4Q5r#O5
zmOHn5*xA_$(g8Fs+>mSQ>3Yz>mKhpOjE#b?bZ_=iNF|TTMY8(*cBA#3tOo)Ow+}vy
z&0W*5Qi3<jm=}IN%Q!t~d<WF{A3ypfcf%z6k2;9<;=~J+GKaLGSp+J062w}&R7;`k
zcJa8pkf^m^zKC#7e5~H&Uc`8fGQ_d;N<41(q4=+SY2wrCn?G}|{cQzmk2UqdL*$`P
zUs84Pq4n$`2&^DXM1c&_6wP+KVZK*v*RGK<e56I7(O$hlOTdi<Q3r;XTw$6(VzbiD
zZ*Nq@;OMJA9dYK|IaKgqM!+5RZ_LQf&W?`jRZh;Mj+$%duOe>z-#%xZ;|_7G7l%p(
z7Ty{r+x;ZtCBu2;7fC64flDTKu~SXAk9$WdB&HtZNDK@3&kgENPd#fHCx}&JuIV1v
zZrB3&2V-f7@9pi&zRK6FF>5;OJ%N29XhJ~4LXh*u04E9BxtkgWPY`vv#*s;dE)m$M
ztjpa$8~o2b9!{Z^pM<yy*#9<?rATB*?{;6Q7O8a!sA}!Z7T6NNciEvEZN+6P9dF$x
z*|blQdc(ooygW4!A)d!9`~zJVy?j|ab`_i!uF5C&TMPz{^t~A1INCl}P&Vv&F-_I$
z)tTGj6=NUk?ro*J6t|YRQ#E4u$UkSnptL`z$dfkZ#mPH|QcsvWlLv>KnS+lZe$L29
zYJ?H`Wghz*^@EwnK=|zrF$4PMrs3b7Mg9>N@xPgyuRpdOvMMYMwYB}cc&%Hgkel?=
zQ064|-`Y=T2$XL4Q^71THhPDKH`cno^!B5cIp*WR%spI-U%n{UE<fQESWbS_=$G6<
zN4axI)MYwCO3a18C)nF#o1HAYYx*-T6Q#W;5?s1v!J+#MPgyIcZ%?_lTEyP_=}@N<
zcd#kGdl!Q`+qwjg`AO@zG@MS!!BKH>P&~gh-mMdv^%`p*C_)H7)0d7HE^Pg0mO5eq
zlwWKA1-%|O!p~?||CGBWndxR_#Rt;VCA+Un3)!7)gkP;~eS%z0Mn;`&4hGCvcwpE^
zs6bdKRXqcRqWFoD*RQ2Mw%6CxK=}gEE3g`uL|z&%`fT=a66!5dXCKS!=&Xw_w2?XW
z@Yoq!W^b`6mD#I{`6V?4_8c*`{o8ST)DI?^ZGG6FX+zrBJanbzztNdIvP~<K_fKGB
z$fl{jmf9|9fms`}bE}bV%=3A>m^TJF|7v)AS`jy{(X39_h$+LQ5up$$sb@yVLpc;N
z2*sO-G8<Q|zzru1hTd$XeiuNu{oWx2sn94~fg<;f7hvWOe{SMtr>Dil1<DUnN*|}^
zQx-I-*81}%@lMuU;9ue-aFw@sNP`_|&utZ2I^P#=v^x6waI<d-8=#!Vy1WC?TBjj|
zpe1f-q%^d@7l)|nbI&F5a^K0f4v}Y;W>mh+e!rb?Kso-jjvixz$^%Ei@aF7+!^f^v
zO7vchczJFpXV2>odhf^tWTOfmJ&;L=?cEwB^$_#FHU}c8OY8bt*P?;ck-BrjarE12
zbxwLNFKuJ+hW{{0?wmLCfs9XOkA<lr0j1Qq5^bo<C?qcmRnlB9mU+XbO<4N19yQo|
zU-E3FYH7C|AdMo1=AxXKOb$uY5ZF38b3YM2Z%5)_^uJ$y29Kif1`Mw%{eu@~L^&^T
zkDO}oXZe(Ze?w%*Go$e3+g#XcfO=Nd3v~)z<BobdOJ~8TfInNzndu0Xp{uX7_qpyf
zFywzy5hr+M=9-xP_>T{9l1@KEdc;@}3)(GnX4aP#Kkm9$&yH15g?=z~XB&;wB_*sM
zzhyV^LgGAvpb0|uw!rd2*1UF=%(;mRGCpiqU-J*%zhGPRxx8fc{VUsg0hV8>v)4`+
zF^&pZk0|`7vEyUq(0{?a1kTJkg1Fs_-Pj^Rrcf3b9v+~-RM3HqCq>xqT@2Q6L_qbC
zq|c+X7;Z?Gp9lYsgSN~zCs_ti6|O7tl{iTSg@oYUd6<j=n{(kYD6XVJdE_S}W(TW0
zwLY^+$6FX1ul-QJgIx!&!=-KRAB01f9!PH`RJxtRrBu6TO{gxd3)trV-8Qu@@8^#G
zM)R|tYVo&L+$^7jU1vLYN6=%4XznO2@Q{p_u;E=?Ay<4`{fNj154N>#W}|9skIzpk
zto%7nzwKu>M`mJvu9VM@@Y_FS0n1$>3#rR5T7=OmJ)y1_yVYp)uVjT8MfOIsfAE)6
z`0ne`o$#`F?e)g*Zq4Zf;Retwt>SN5M#Eh8Yr697-`{)&ncJL^g7$IOlN~DkCq~Ip
zxcF#Qan$CTmbI?y-p+L$3bodc+vF!5^S2I2YF+KFa@Ja}XX5*BWqy#u|A{bTG=oNg
zjw*+*-@&+>jT;C*Mw#IA_L|H~+6_Cet==nX+&d`TL=eWI6BA;W$SY#e&$e7?!w8<s
zYrx`;)E}qCrT1dqXJWdpK7V6Js`pn-zZXyOo4<A7&+*Zlhm~nvxnlnw$$H^X_japI
z?#6$`86NSyHdnfnXYvq&;m###AuJ}g3j4{!xF5JZ)rC+?{;#KI?~j9O0943tZ%=b+
zHUVVg&weT^pZniFirGJVQ>i3G2teZ@J84K)k{o-3uQD*F(pii3-4AAk&(>U;lu6Ao
z$g-Zo^d;JwtCTh9EjV+)F<hl@Ana-Fz`#0!P&*t`+45SIw{}OG08jYp0~yWPPAa0S
zhWP8fLKamQ%*j~m{)95eL=^WN#2VV~_R>dA`~3K=e8oa6c-w>6+^OesPM1y!kjBDU
z&pA(f`>#Hl^CveP^elAUd9Fg=@7j6I|32JxOxbDgU?`v>Qe2~0?*s<Yaur{d<;MQO
zTF8(0OI?N9DsdShPwE#Zq0}y2@*oJ&YjgyBy)oJl(Wkr+WLaa|c#U`8VDbvW18;bL
zlO3n$aRBb$-W2m^l`Ec)8tFrOwAMrq1#>q4xBtdezb9M`7g{&_&O5TZZ(Wof->?$}
z%J?Pb!XKPricPF|p(FMUR;ac+^pWM8PIk*uXx_20Hw)+ZemRbP8`)@G<i>q^_3`hv
z3%fhRyNgKW%iqk$msxtw=4g&Q_tQE1Yt-py$7|A_`y~h4f6w+Qeq*|oxC<Fn1jWTa
zC2qSV;{JU)#m`VN3FmXt-?fCiyg`RK;QzGCSTr*nV9xoezRi77ofTCcz5HY^bICQ{
zS3^_;z$2rx<rfatCGpN}L$Y|ggY^0eT$r&@Q8W{jw1!5(K3y{PD53K7m7>O@Mr=lQ
zmMK;DQWLDA#wGi1?Ts@?9PhnEMI>akoUMAYsW<3)%|T;ct}2hdY<bU*8=NleHF<lh
ztN1N>yV9IP{;u%Hj>6Y}Qv!lh%o%kfE<1jIk4Ey(#tT!gV%_t$921@ju=$pfno#TJ
z!s{Kd-)+V9Lfj3?{f}SXxtJ}ThGLY52jw4fs$^;3HVcLw)P!HJyN8Dy;g=%ZIw^Vw
z7jH~NXKSn7zU7ax$@XuvvysZ+&-;a!>mdzPRj^Cb50;v}G&Ui4LupuiFST5h-m1sj
zJ>}kDcDqz|{Dw{<Zf#vXJz4_z`_o(lbBiRZcE#rrXuPr}eDK&GxvmXj?=R1f^i*e#
z1b!T&C0x(dX6;;g%2^tHr!``x$o$l)>*rMxd0e-PI%|wA53g_gvVN(!aS+0OqnT&!
z@t3L8X#Q#s!Y@$xeVA#5nu<7ch_frZP)S#MfG60*h4I4hhlF_P%i~M?=UT!qk<vfP
ziq9)1o}x9ejy(3i^E^uLMOg)crVvrI$Fguf(QCX7LtA(n6L3(f%Kd`2A!I6K(|QpR
z5nST}I|8wi+9vP%5rGHI%^20|rXB@~+m(Vm#@3CBz^49f?s}|Sv)(@t)Zwn8QFy9K
z$1?(70;zO+t&2a-{L2OAN%bx0r3s7-Og1<dNc>>aZ-UY##oknqy?=19;u`eEVq&m+
zTRdqxd*)1OhPy!7|GPUqN#M1=m8O{Z?32?)x#RoV>wCD9<W9*wy~k`urQgoJ<NIsZ
zk>Enob4x|sx<kv{w-2?~nA0EYzctn(qxi5c@Ahg(9_P1L_nf5$$}4*FeO;CYU(#56
zey}FeKfawvDHLL#e|lJfGPS@J5TwcdiUe(1--O<08cG(S!<vb=xaU=HW21IePPlXz
z%bzSORis0~s7v8mOJ6cdJsR}<5_wsc`=$nU#RPy#m$n8CuVo?NIk5e}sdw$<Zzx~~
z%x}dhE8kXtMJBj^6f!Ty!ic3M?|&HxxUf-2w6Qg{G@svpyH~tlN&A9|C};fxh5K!8
zrMfl8Sa?N$IVh&@eR#<{{qxtgeP41INJFhqk&{X>&k3$?(vKQy$ew|@*A<4w!=LRO
z$?w)o#|l_?ygs`$Xh662B2)XwH^@y1Lc;iH`3eV7s^hkwV{WST=I(WtA9XFSonBqu
zs(fW+Lx=F^>S~eQyVt!N0N<uZ>3if3UH)}P6O~zMv%TgV2^H+(K}Rf*ru)xjVzUo7
zCJCfDBHc;l$`5z#>F?L<*}K=#!C}2ZTx8@cD9W%WzE(!~IKfuQ846As+Xk<5+F<F{
zCIy!|PVx;bc_6R_>fA}rl<+|<4=ugvLKq)RxvQqI<m8ca0rm;=+0uQ1^#AY9j3#I8
zB)NIP#53*C6P@fBso`BYQf(FeH0vh|En}}}|LCfidvkB6V#{FjnKLwm=c})@@fKej
z%a0CfHvMtfqNc#ztK1^Da#_sX;QY!$|1*W9zPaPung5Ee@s{mo%ue|I{`u(h&PR%6
z<1)t{a~KtstFY3F?t%!>KUU#?#q<oxBwATMi=iY8rk7<Cgp#~n;ERVzE<(Xcy+3cL
z4ayP<JN4=i;C(e8KU(?z8XjJNc@oUF7seMbyE&9<n(t`g?Cjid61GL?kC2{_+dhU9
z1%?Pqq+{J;v)AB5Gc8X8a3~*Z!iCEzK_jxGt5v%n<&~4;)kpIOy_3FVN)4YoRT!E2
zwm}Uu39tw%P0sa@{c?wz6#f=ovSJsRi#TUf@OKS^s>htv>VWoOsA^ScTZ!&_TPkAL
zuEEEtY3j*`pwqUP>gnl$0B~QFSV3Q8L&>2>jV32f&=L6Z!qW76J<8W?_7|crKc5S;
zfb)?|vRDMH%xfZ7Q0PtPu>M?vGXo`Kw$d~HI=Zx>3L$C<C7ZBP7+|Of%E%aox&u6a
zXlbHmwED3y#rBw6Cz2o#P9+_W@eYLG6v8i1nERUG4*~kus{blbb)b&np#9O%Fp0I>
z!%m|n?gN>AMbrCi$cc=PO)h`mp4`wT{piM8E503J++XW!K5A@LvDS{H7Wbu7D$X+}
zOe~7p!!6TuzXmhQ4RH>Yvb3AqKP!8rYUl=~_t7<jw@P1co4jUHY2;sPn`xTMSS;I;
z_wQ!$SJ7jaP@H>XzW2?KD{W`*f2dBhdPuX1T5*j?1`T0F%D7NEJ}<y>bs`IgQbOV9
z5~P?M5eOiLLV65)5zGv06Bc~#O;co+&`e;iu)2`6iu0{;*p;HP!veL-bMz&i6EeV&
zO5(Vcg^S48?47Q$&;xsP(|cGoE~{>4`VWKxa`%ml_t*FjZakvjyCJHIoM!C3@Y(1|
z`x{bgdF_)HR{6VvA5D$Dy}Y6}($4qztI%2UQ*w{sY1u9}$=6*%{P)O{+Mc#SEP&5y
zyK^gD)f-~lJS2v`^m<G)okd<gN+29f!8A+E_?{H~2GwpRFlupWkQXpK!5jhVs~v)Z
z2Y@{mAw@(8KPc4j;6wa7)>WeX`u<j%@BY@mc7J;%({XEu#-XZG90|rRGh&o^ZMCt1
zK|2m&_(@{}rpd@@XK9`aapQ#-?(vexlkLxwqmoX%mYcsp32f#w_(5nUhqH|PPQwH`
zbZbo?WJfu~+3kx}t4XtR)8P!gV$es9(g*leALF2P_$=W(3c?7bs@K~oaj+XrD|C35
zKw`r7p&Q>u_<jBL=$sv-VW7Y^*4h-*OTr543)R@q*iBL#&E3Mp!?d)u#1TV7wwOIh
z(b2}{=17Dv2YIF`X+>&0-!H}`YrrV$P))T!^O;@|({^&M>$&^-ikw$dv6oHOandl$
zF!^D~f{|7LAFj>|=g)JBn(%F4tbaB($Yx^0eqlxJ)t?5F!|!WstwVm_6u1A>^CwnW
z{!h-;<(6c1uF~iYOSZ?WdBxY`jY~ZmRgHOYa^!()A8fYMQxkV0Bf+CVe-1x(t?fG3
zuiKbOx?{QP+zujY7Y1X(3eG=&d4cs9dm?6$kJa3zddFu~IDFdLfd7LVkMOG>1pTHD
zR$XJEvCg7+z8fJSj1k%Zl$=4JU7T&U2?GPeu6vBeIZ=WTi^##7=TguOyB_m-cU8l$
z*psSgGabnBJ%4@#2lZ=Ha*ryj>}`M=ZfZm(CGqB!q5%i<4?4c2dLdT&4U7jJ9rqqI
z#NEpD@3Y3=Zrv&PpVsR}^^R{}$|x@Y#Bjc=%X5%DQ{cYyoHXdZ<I*sd?6ap;ElnIz
zt+F<j<nZzjdk;zbR;i2VKC#wQy;#XgHf`A(p6jsVLffJ4tBumT$1Yhh*of*-`-+A}
zCL|PI>L~}W6XOz>0%l&NLaQ=r!L-s+(<_^=yp7<=)7l@umZ0$yvpRLPpSFGf{)Us`
zPU&V&S7rm58Q#2_-8ixF@pAqp5jnZdX`jqQqg5t0@ZeK|p~U_KK2Dl#?mHdH`j4DD
z|6BdYh>Nq1ltS}KNie48{xUj?Y4HCE&Um8vNJLchGc?6T>mcSF`6}|J1Y0GYFSa@O
zy5MHSgN7{0DweSI+L3^Q*ya>K{`BynRbN+I_(S36X=#<VITFh96n7Q!doaqgV&_yR
zky!~fa8x|1HB3QalZ&0+{ojvwCeUQtJmfdXe3+-uzJ6s_t-!HM;g4YIT)wo&EKj3E
z_Hz<K$|w|N%-|6WbLdcr=}Su>ykM${u@@K~RJG{G8!-RK0No6&V!nh{!1qW2R{Z3w
zHq_gt$!T{S&yZU|Kw{<ZjD6DCv#CZQQmxn<>aylIb^aokcc{!iieVydUzT3s)@5E<
z_uzWMm!;9rWBHOrKdB854Rjn?$TpG^nB+V5&N0^%y@>FQ+TJNl9S|$lX!p6k{@g$y
zXX$CKv?Ib`ST@=T>g=;K<MfRPAm}iELqu0g1qZB)$Wyv*4K8y@Nm9yFAZp-Zddt>;
z_=hG}D7rN@*WRzeo&#I^4F{2@$+CXE`+-Ds)tJ1(`qVNC6N_Ek_|VWt<jcGoFG7$K
z#r?#Ot{_4iQLoVEL7%DL1UiCwfy<6#fvGK{nDdcn@%|yAaGD!DIrgN9iG>C7j{|WR
zw*U7m+LA3l3G1)5CGJ2~!@u;mD*Q8{6MFd)1g)DlZ-NqlKX?Lw2OYh4LNnwmgwXgl
zga|*14uuEydcv=P5Yf98c(MA?d)(`-z7y2f8GbJ;uasJjlACrpNP9s}fS(KTEEwJA
z<z0o|@0h`QQPu!HySUqjkC&KJ`L3*USS%@h`trvDwdW6};$NGsJkuHGb>_&rj;@B)
z%&bS3UpX%7OkH;1R37>P4kwiQn$MHd*VEKB$l>|8RII^|<p5_W0H5NY#>5lqkZ-=U
zx;LIoNO;j`ryU7>SeAQYa!Cm<ap%q*2y>x_{pXY;2<dpm3*SU-=ippDo_qvS>Q93+
zxI|zje3)S>rC%Q-ajh*R=!++OWorn0CwSU&O__qfp5O2KOC$ttvd=q3ae}OVSz2ib
zPp|2Rcovn>bvOJqy1Jw;6@g7L97AXJNvy7WxD4MmPKPXW-JMZrWMrhJ)z(lEel4R9
zxL<m?A5%(VX&yU#FsA-7{tNT!%H9KZaXaa$Z+_Q17!zD?*j4uNYytiIyL{h-A{q9q
z3uN|IQ!7VFW@c!JDHn?NKV$96Ank&z8x4F*?dRX5XA#Kc^K&Ju`i5V_WIExWVh*n;
z8EY80ljYbpI;-gv16L>W(o;~*O7DJO7Y`*fo=t?`MxMdo7xVF3K40tQ^r$bt_;@rc
z-nT5Xvr4InUUNQuJ7Y*;jqy?T%{RtUr1WP^=iK71GPRC!7pn%iFvyCqt7e$K-K9*<
z?vfDGlbiGSko9CW{agF*3L*7i6IrK&<2k7zNgf@vgtf0O4qfSZrg45aN+Sg~7s+FZ
z)TKXfq1E&okM!@?I+1}2@%VrI<6{sAQESUV3yUaqiKHSAi)apJo0s%4;W%~X%o(Jl
zmWDIk?nQZveTZW|M2=*<PF;X8;>mD>lrHFv6EBijxDP>fAcm#O%&!mL3|>W@rPY(N
z*R8Q!9Fj2D8HJ2Gj956={mz7I=YS0M-+VVpuK?27=5Re7mgMTCx?p+g8_(7)<*4Z$
z6b`X5SAB1U2Vc|uah3Rbxyc*r{BW-E)jpno<<d`nf=LOvZ9DrA{e{{K=QH4P?|_`2
zY}DdqlJ5sV6XNCQ>N8k&!EFB)2))?t*(MG+Xn1dRpWeO8Li0T}Xw4!ba$bL4)Y?80
zMhd6}C?WmRmoK(>Q0?J-Csx9EDl71EUQqK7|MnN9{nXS{4`TZ2jic7=cuXER`a}Qj
z?~^6z<DQ>}9GDLbO)|?C(ky>d->+Db6zgktZ$y<xK>}NB+8+aoUf;*f?>1EaPfRe)
z`mG^Z!}m{Iq}$PhXW0x89vwK_iGY#lHx<0_r$2fb6BBczSN5PGzCQWM&z6F{+ttO_
z0}`$Ih;7$riEND+GAi7fYv#fF)iqfrFS0%ygQ{rPAkH8<eMmpnGaBwIn8L}Vst%T4
z@r}+<18r^kjrY!cv-V?Lf?W9Z^cv1>r40XyrUWzkFHP31P10|gWIINFZ!wXTF}xGI
zBYoFV%T&EMpNyC4db^?`IAk61Vn!X@{4w>u<=-D}WEXZEYj}W+1`qN(WVBR-ZnS6t
zaPt8jLa}<7imLa9On!_RiRNHuKSJv(2mxrvVxLPva(K-qX>5T}A>kF5(s#RZCEMhM
ziu12$0WG6D==on9b<-0RB$~s0w{j6vf4#rDF87zVJ+X`1Vo$BJXmEbD^PIw^ABALi
zYI%pOg|t$GUj`sBD>;X+=PIA_ZM&C~H{JF~${pJz^b|D2dKMOxeUQ|bw4vvwnR$%-
zY^2Dxk4^G5DM{iw6n?M^cw9Rvc3PZ%F_Ek>QDXNF#nN&-RTJ94crB0M!g8_s{OscL
z=f4edn{H|d?6`FE!_SB0_MJ%yRlO4GdT(92Rc)680|OgolV71It{<#0<T{T%7yY~7
z8rex0*;Yulo`=GpV4R{2nBP^-zmqP%ok=;JKY~8B;qXDL_E6y<Y1YAwpDTpIt4<e>
zh;a$W9cX-a<y1{m1&>9yb5YSVi`CiU=qH!^cZ=;=R!%)rSMZ8tm8ec=!RWYDn_15>
za8&>RTXhy>cS}7!`fMDp0?wO;94jM6PVcqUygF4zuWuE12+)+M%W>C9eLe=u6lg4g
z8Ae`S3I`40=o|e0G+Q+ioBv}&MeSXG+m|mMJ}BALXXlge*gRGfzLPN^-Z)`*f@QB(
zmGspu#SG-Oa*1yrKGciu@!ml5$cq}!Uvyc;9oZ<RM1c9{jng2_)!T_}t0C+%fmoWr
zuBW=6Vmw6wN2@A+W%la3Z9CUuJ}YsdJRuDqUxbo4&LZ&cdZ%dhA%GEh!2Y+nOm@;=
z{T2=HpL9)_zUS9VK2cdb$ZyzMP2M_Dk6@lbv@#e%eCT%(E2L`}#(5u<bet!A({H+<
z81^?m#_3P#-<aW^pY-S{x@K9vVz^wHSC+~Dd05JO=IyhQkXW(W6RYP_#_Sy&mPSvl
zzJ+3>$TBG+!qC+8Z-Y7uW@xgm1BkU)gc=a&C47Gnr76Qo5lV9IiL;wYV{l%iQ;+O}
zWyGy&o7+;Kka9JCzmFeHioF*oJWgB~`@#Y$1A#n8CsOivB)!2zMz0rQkp?ff0`tR)
zvtJt;8b00IO#$x(&32Zt@as@mx$>P24Fo~IbwrD=lJUAV=0`S<3sC4Cejpz-y?^KJ
zZ=}E2pZo*cmY1+9rMq(C*l+jZyFYK&c`i-R?W*vcP;>mGzplj8c-};+6N0F!snQqj
znr2E*C)UmgEO7?)^dAY5d&AeUE67x^n{L^IyXxBCY^}Avx`&Ge&F)Usf4rW|w~Nv+
z)@}FwGW#c4m)~oGHFBDC9{Hm_M#diTQX*lg#Poxb8M7r!6Dhu~Zz+DFl+e9E3h{n0
zqz`xlxtAkSevs}{cDMA$J2$bDulhh|iaj!u#xmvHO@SRxPfy1?z8$6z)zN;U5_6UR
zfl(B?hFUm9?nOuYQo?8~j`6X214>Al5WoTvaQ)9j7s)~hi;4y+`yy(@^kp8{sesdm
zvE~AeLQ8NfTopaiQSO*8vwy#HK?n4MVa;l^Hi8)I#awSj*>%KhE1R44r~LHc7Adaw
zl^fq&e!171ahy5&n|sW2Htbqx`Bl&Q$USdnesXx&-@nd$kVs=I<90Bi(BH0&l!7Bd
zuGfc};7Jq1*_~RSlml+R#go@g)!+}2<+`t@gKze;y>%~5uC7+*rEAEF5Dblo;r`F(
znUwfO|AHLn@{-~cUitJ@&wyJJT<h~YTtd4hy0WaVK_fdn3`D0dM91Rm+aW>%fjcG^
zISHSiB>BcYcwlkrl+(Ug2U5Bn`8_I4zX3>HbJ>VZf}M^IpH&j&5wjLb#QOxWmpSyA
zEyyoAcyP^&kOPQ+Lq|{f*d_LaTDOAPOVtvriZMo2uLqhiz|;jeL!I9AtaZPH;1R+P
znIT67h$k>6Y~<uL6^({nmQB;5&~hTDUTKt^TfD#5!n0H96n~m9;4E1eH>!lZ=g?H|
z^5Q(#GBl$m+wkWd{hApLCRI-*#q8ml>+iI8>BV0UNjT;BcE~D{@4}0keZKRPlj5!g
z*W^o8yTMR@|CJ-GuN3H&|A`fcLg=oz25+U!T2&ijl+v%zTVWVG!M<)Xo_tJJ@;i@y
z@cTE#<r<dyb#bIiBJG%6-1oLiZmmy3TRp%b>2RF;`60B+XYL1Z$^;Q#OF>?Zm#q56
zrJeE|-XUAt?@|OU5;yZ8@@QT@eOfBMzG{l{Dr1tj_x?KR1<m)`FZNHOWT)sD36-5u
zl#q{-rG7CQUh*sccHq(};4!R^V3$5&Y+OF4a3}+K<ox#K|5~oFHs@ni@et28Swf&n
zY1~z0lb-rX8HWAhKa@HR0Bvw4uY0#YC65OoN#GKkcxD8B7(%QR6_??Y+BWAz_;WF!
z8)3hRDBfKSF?@AHD~iP=$JqFaudYS3gSGeA&9zazzy7?P_4GP6#Cm_nTbH`goL7$A
zU7wD;yDg}Ld93;<RG)WHssZ{`(gm{-6Sd2Px_|DJ`~ycTD`Gql`gK0%BQRbx+R+CL
zA**AT78g_jR+St}Z<Sau%*FhscWP+XGf0h1mX;-PH(5bxb0lgP^7^Zt6V5tceK6$X
z#UIjxm`7&f4IuI~98&~A!5HIQ_59U37fTI~W#R{$lkcuxbQ}+tx<(;sVe+NyU`Hik
zj3p)ID+%C33^C<7+hv6*92S+uz-IvH$il>=i02miNt@rHkOrGMkwv+<!a!a}Uw8GL
zqOx+y!-w5~!5uzyDBv<^A{?iAHzEQYVnPbj7oY3I&=8=DnD<ZuaP0E8)AH|yEcUCO
zW57eIA7?w1$D7{WTHM*?@3d`cI_k8H*OIk-)DEG(t|PNLrZx+2HU($Metn9;9~Y<e
z>%_=NGQgTkzX^93WaZlYPcMqxJA^jDr9f&6(~^Xpi><BgxpPmTO>uHU7j;WsktjW%
z3%?i@0ndj=`|a(!cT?19XxnmgkMl=}f6)@u!E|l=EaSg9mau7!L}<Fg?>99UMkHIp
zoNB*k++SmUvUU5FPhgZK?0wxuPSk(?yuR7?`u9#fex4WcmiSm3USl99@xhsm4RiX{
zKkZIb1mMVKCJU+`;l+g8X=x-q6L5Jq?)xza@PV;UYYXVH5bGWMosOaiAGa5`Eu<nd
zjQRDygtSD9cs9m+9Zk;?j%)S$+gLG1GMHtbiu9e`KhsIxEYec^#ZGrhXy?xBW4-t&
zW7DRr0)-W61RPmbCC=HdO`ki>g)tiTUK4YR*o&f5PiDcg4Ggyqi#@w`$vbT<fMW(C
zK%D8oNFZ_tQWSLm2?kG$(=iEWzezi?^dSEaWw>W^=<@5uT#h2M-gjwborS$a<Q3~+
z$9HxmarxffHMbwztqre_o$=v#haQXYJCu6f!Jz{acmTPO1OR$2yq(-}&w{yl(Ugn(
z`1F(;QlG@L;0f(V&l!B=Hy&g;{{l>1bn4X^BW}y(U@aNfc;-N;kGzf#^KuTZI^8`#
z6Tz^)Kx^_dwtbuFNF{Mw9>&lQkv08jLkZ~fSg*SL^Y;#8J|-N3SR!#_7VDfgc##g&
zyLB`*H6cdI!2xzXSZ_Z&0}m@Z`E>Vx7E#H#?_?-hNCdIEShBh<yi`REPE1U}xx(7<
zeF5_nN-48p&2-JsE4p)I|8tV{?Dg#l*T1l_4RQzY0ok7mJsIE&JhGRpI!2M5`$n5~
z`QrOPc70|O`3pV_i>vI}CwmN%9)DR|+t71ju9j)EDAfI#jM!gKot10OQ=MhuGd%(d
zlamJnsomG6?K_}dyq6VfQtyRn?5D9j#|XeC8$bvwgm5X<k(Vd7MMgwqyHD8!sUc+%
zOd`yKC_sa{gzDF#supekmece=+9^VC1bx^a&!Up$F(-=pkDHF5pkmPlq-SIRfeSqm
zOhgDH?<n+;yj3q?BlwNp8$ewzuSEm`P>KlpCW>H@`1ol{PR2fJwo_cB++M=3q9?0;
zEb9cd&8M={LX5IU^U3!r^2~2y=WDr28QTu$u^nz7<{hroIT)2J>i&2pXjMvC`{+tT
z@~^%fEholVXL^Ube#$a0o(1bmQ&{jWrlW8HX%wV&trktOak~?K+Jd_fT4h_-oPHu)
zN-FX=pi!>$Ra03r0^|#XXKKCGu(6>e>@=tjm9C?uzN_{hVNnGIABgk*_v|d@<x7~o
zErU^6w!UmN2Iw4Y2ljNV;PZsbETsATa`!bB3Wsr<Lgmhv;q*Jc1xJF`K}Ar!h(jsC
zso<2q%<rToz~SFlS+kv0g4*k3_-|6c=6pnwbV~YK;-O#tAt;o){me>z`Kfl_TIpz=
zV!O>UD)Z}nq4JBlV&RDuv-`h7Gv}04OQT<6^n?+Pc{@9~mJYy$1Mnl1W$o!=T1!Ll
zl^fGjiWJr+#;9sn3ct&1EEl>^$oVbcuzonl>^oHNiGN$O1HTc1vZ$4{+x~r7OBF22
zg@|o|=p7b5B`|2>p^rE6#_lJ45I?==K<qhWkp@@5N#$iz$W6_*&N6e8@c;+`@FbLU
zWi}0?n5@gSxZ!*>+QaAcY92j@{72=D)`vs0_i?grc$H|sgV*iLHp1`qdzXSYXMV0H
zCFQ>7wY+TkCr$npn0tA}FwtZAy@euYg5~Ma?oAJ_7fGK;YC3l}%Qp3mFANbn@5DKd
zvvY93W<f<@k4us6(SMqNvineffdu@4e5{+~FB{36tA4H){;n{s=p4`1t~e2fh0vK*
zx4`6psh14Mu(Bw0&o)j++erHKhRJa5G2I|9q@d7llN~tH+~3zn5IFz!o47cWK9$+)
zzF_u-Y8}kz%;LZ2@rh#TbpX8so|nmO5U2t;?a!z}_(7Onzh;euj~(j`uFYIreroGs
zVL8B)SyIx4_v+=hnLM5S{QC}m+AW!SZ#nvrfc`_j{4lch=;xIO^fl*Br>K2v+Vft(
zu_6CepBu-y<HuHS*BdzgVY}UsY~(z}bUf*C$ThR91eQ0;(bq1PWR7|U@hLj!X*6wM
z(bYX?K`XA(SX^&<V(zc>UNLX)$8}Q2HY6X3iH=Td8HH_AIONdb!?@P=-mkF|S5Z~S
zHaW24(>z{nl3HS8wa&@ts8^!Dj^vE@QjJzSzc{(SHxUJ1rR@o>Uso3~D0tp{-s#;~
z9$*3+uZFKnMV0>b*)fzis#VFA7kzz|raMmyF<<ey(2}u?Ee!CQR0NKGOfnHbjznx_
zmp=O6ok;8g&cBJ0YVZfs)of_}kiU-H;XQl8Zaxna2~AA1*K9ezFWElMKI^7cDUoOU
zZ8zqb<=pIVOm7I8yKW-<uI<>EoqsnKcqWg2jaK!>_lZp`q)Ey47ju>`%ol6UYSkT{
zuyAtnywcTCF;>s^n2xw}Yfomkw@IC^&W&1v7<!iZ<8E<UuT$o)BuTpqvMpp~Wh^Re
zaP1qVd;n+wAR71iAXb|gXQtYl{>!(1sKxZ}I#=4XNm2234*Qze85ilQc20t5eS77C
znOQq(@F^5|VSPn3#Q2+DW}*}&CEM?Ah&GG~mB4S^H->KRr)DWY3MTv>l&G|CU%x`g
z?CMIQphfqDKb(8b-JA4fvfHK#Sg9o(I~?;r*iaGBDszV3i;MeYo5K*$FzgD#Th22y
z2fO3apnUFkS?KPz^76vYLQF)&FkS1bvVhJjNLL-CPjgA5QQ-TL2|@I~<RWl3bRsbb
z7v5ld_N>T;5I!MmB-D_!s!;yo{Y{j+teM`-PV3+M=7{j!e;B2#_re%K!@a7%GYTT$
zMa*W82(m*ETl&qWk&7emtH+Db!&^3g2}vJxT%K>Or_6Sk84~uqkJ|A|nK4UrZ*;=Y
zN^JWWCq3;{?wQs{?Y_U|acMPd=d4@M8D8D<>78;hdE(Y4Pp_tUy8G@7UagJWN34{W
znY}Mww(neWtR4t0K6Yq8FVTl%eSEFuzu20wi3xC&vbuF~gl}NHcPOZDlv2Pt8wek@
zJo(fF(=W-t$He_Clx94bch?=3y+g@gB>d=j+^rtpVRtBy#|{(`B`|`|U*GEJK1(*A
z066tO&Cp_<!URP2*tDGIcZheidYZeHzu5f%PjCel!Fp4}c`m{x8NtiY))7aJA79gM
z_uzmBdu87yoE~{)`+G|PDT7qlJ?r&rBlvXIUw@<(PEG7Zt($y=K3F(J9)3&FoTU91
zjIG+&#HC>butsbXdC0|OI$+H@Y9IHZQYc6Ab8@2W+iLx+P2gs*N)pFO{CmhyQ~bak
zI3DnnlnO+Ip^V3*$3Jl)hErMSHPiUg)0N5y^CR?|bYnkwu?c-3hzRZ1@Z`}iR-I05
zckQltw3XY^XXs8$lCMwyZCsjDZ;ff~;&`@fp?*X@YoPo2C-&42<@%8kh66WFo3jab
zTe}GT=t<kJ`b3c7xLbzo-oD$dY0MIi^Z{9>FENdL>iom#y=_j}gifs9^*rA+e$LP!
zgj9@&3)GQy?ZNe2Q}Y*{c+AgmDRJ~Koec`tKN9h<YXb4NY_C*jrl+ljHxmzYbFaXk
zt*%b0b6v+>ikWh0Ydw;A3!Hl^Z;Cs;WjkgC&b{~F;U|RO?#x2=>LtFq%MVL1eA`@t
z)~fuP{DZy7&;@7}W!M28!0`}^8ew~M!<ApH5<^AEJEGgK|7&C<pap;i>_e4!L3aCx
zVUa+B92``;*c-opA8{cItX`P$M_{^-mdd1>gnH=S?h|TBT?m6vxcYfw4qXa7H?ew^
zHrenqA3h8#8}1K%0|Vw4y<rZwn$P2jJjXqsz1P=3*yu&UVq~RL-GR>5Rn^ub-jRHx
zk&%1WR4XgGJVx7vp8&jGip#j5gG)@G=I^D&Q{<mS-@7R{8w{^}njYPiFEshZrts5y
z=Z6<{@60h15AC|d?Ra>s-C72|{JP-bS^MI?(0A#7>%lZhZ-XxvH*IV6b@d4!e)myQ
zV*RO;(~GHVSPtLWu_bKhU(-5_4q+s~mJss-9Q#PQ!mCcXxu8wqIMeAWu#--Sl#^S)
zafI;g9Xrm#={NO?!T*LrR=)zlr>OXV!4+Nn&QS2bTpu55feaBWH#8jd5)ycR6J)94
z4KUk5heW53UULma3J%)>Z3es@xE;n^<+0X&N;Bb70M+f;v)H-8G>^x=xvLAr6kOoe
z99|^4+TAli{tMl&iEV<nf|*-ZC9)5v+QV?g=Mw%BtN3Iy?$GD7oG5(v{5Q)>np3}-
zkGKCY<sIHIb8nxaW5mxVaw0V*vYUu3EQ_H{m~X$P<Sm+PAZEQfH)U2gq{OtXoX8Vg
zExVUmQk6ZE?`PL4dFl0Nu5*I>i@`85befD3H2U!4zdP;q{FWnoLg#pfy?fr}DZgEg
z&bqws@y7I;#2THG^3Jvh(z`8kK><B|ZcF8ipZ|^PF^Qt6nekQt2PEpL#aCw*<2qHE
z;>wM@hXR8qP;7#GIM5ud*5rzE;LOA_<Y_RY5q=<2G%$LQAt`oo12YTQ)vJzz<9GM&
z1`MON8n5xUHo^Z7xhsIn(3=xbund)|lUpmY9-86A!&eCEo%jRMEvK=myL?G6H0g3A
z?aAYk<Im%qTeLY);X}cTs(Z<|<cFiO=Wjdw>%4~i?>}i5uZ&*j9!plo^Rs`3>=ZH1
z*jDu_N~fmu9PNG{L(RzPq|NiL)_Ws&9;7H}Y;3H1I(96u822ruob8^;^A`)dZ5*ss
zUtXXS@L9AifWQCv$k&ZrDxT?0H%!#k4=XM7AsPuofK=oe(fnnnfA2wmHx4*SKr<)Z
zJw4^Qw{UTBRr}vRl**>mE=B5N#aDh1nEBRUeB4zO%~FIhxY`0px`}Mim1LI8*@#79
zPdFZF;d$OI&AjOnr%k;N-HC~zoRDT_x~zd8Y!KvQa`tS$$vWe$e>+!y_GNlMmUI<9
zZtkkyTvC5HcY#m-w^t>ZVa?*w{jZ&g;hyKN=5DSZd`kNIW-zHHc-d?`c@j%=#1={+
z$E0?DOM)E7%lFjO60wlQ;v4bh_6^1@saE?L;O-DcvD0;SUxWG~qD5UjncHZ${+5MS
zDd0x6wqcr=K^cdaEg~)G-WLY_*%Q0&P40zU47Oe%!9+Q0wDTp6PurL;Eny`=8e!bD
zi0CzpHo}!8*8d4v#7$c6If21Xq0fhGksg#3FUC8kl16{sOxpf$(}T|b>5cu=t4XU%
zLdwdk*;PD?$ee{Wt=N6)=VnRUGiT41J+%-HSNvmg>x2dd&r!-<ot<^UXV&)2QCyg&
zrd-+CA4Ptyy+&0*io>YbG)x}=_f4#%kazGN;U&0^9vGavtHx*e=`q}{$o<531aAa<
zt5KcU031p^l)4WP1gTYJ&(oFJ^LH^(17{Kh=3^VMRmZ^~fm_y4!MDA`r+qYIEdOf|
zbM~e2{(dq&!Flj~tHSm-g=x-7qrbye&B=d<86=!w;QF5`^1RX={x4>{nQcITjz?ml
zbm@;>TF{klE&YtWVq!S?#c#iCK7V}IaXYu)aC>1)z%}qe98_4qR<Z4m)q{7G{&>II
z=H{d`+B!NXi@giVtQtmV#o7E)o}v_b>ZFWizRGez_7=&s4^LRZe|?6IJE&(4-x$m@
zXnYUOKXBVpyfD<N`J{!5BFFlEWY#e4^?ClWFlFkD&jaIIy9(6CZ-i~76dsg*gF^%E
zi~09Sa57amNLReR3P0=7Ahz1;)YFJynDxZ0(Dcf0WKox&g-H~T0VW8Jj+U@WKIA23
zrKD)h8bHLxX-(O|>Qg<n;8l>ljp`;yjfw!U9BXDA7@kF)qqB*!&NdJ{iHq5KS85c@
z-&IdO@;Om)5yiNa{@mOgfaDmxG}uc$EV-~)flZUJyk4GKhIjLo8ReJ|pT;QK*lT&$
zryj%$f8653{_kfe>1p{h-jDb$F0Z_4h4PA3Z^avTTN9NMJ%Xjg#Z_1X%dDUXK5d%B
z@EtQE2(&>GVqsyyzHiO*Hq}OV?kWl{EpDT7S{}Ex^E>1_z>j$*2a+Q6ANTI1HrfGR
z$`BwPs|TrauehyV=wXT>7<N%RWNVWcb#=>H9zDt~TQ-*O?ec3Nuk1ncpbcSWJl`7f
z^NF*A!=ln4gXGZ@c}%|hi!Wn$)_Y<7zaJfElabBFY<0f6b5`f2DC6&-v|LH?i=Q9q
z<?-GuFIO942f3T>d0SZvmUj;N@0!nt^`{VicohM{`UAdBvifH-ag8c#uO#KCGQ?*G
z|60#zMn*<yaSYA3sLZiU`q83wvR}4t@^#j!jusig%XsKZ%k6qd#&nx9Z@yFf-wXNi
zeVc#q$`=g3&_rNJB-xz$obA{)EZ!GWP;ArZ)?%|}3qk&06K`yTE#Rah_~`kwND-P1
zQP?0^PK0|p9@}sQOt>M*D)wN^KdA|17N`alsBi^&r;(i|VVj$N9@~^_TI>4Z!oL(=
zS|Z^~;o^Xsu4r9G!@{S>ZTXo`#f)CW8tv{4(}<DuGH0^4J-e*8^Hb>m{d2wySRQJ8
zL`hbIV91Ir--D%=@$usq5rBz*rrxqrvnlFnJh+)19Q47aOw?0Yo9Fw*y`KL+moa1I
z(|#AZj(T~QK5bv$ViQpyui$cqr~r{L6ahldtf#fuU+1AiPmP|#G8&QsYcv*`g6s_8
zRlsNBdCp3+*cm0UO!hsJsGOLSth{AEC8DpYN_Egse5n%(%L7@<F;i1nl7qHgZ_x*G
z8$bR-do1YQ4zurT-Q<f-v)Zt_95cSR<oi=k?Y~0LTN>Ql?(cq3Hc0?J4JNCDEBE#-
zg>Qh+@&$y^Z74`Ww#zBu5Wth*CUrj$>#%|R;E;NhA!_W3yoD(9f%)vXADD_^MPJj8
z><_^Gkf4hT+dr!L{1t4SK+Pt)5l9EM0bn6qqZ^S{2MGv-i26-MMMbLnrcnqW^GuJr
z{6AmU1Eg{U@Ik%=mP>BJx;~{zf%<5$YmfeU5>znfweKxS0h$mx8mdMQwl3}@B$Su0
zKR?_l_d4=n;iLMsenFR|jePD3$2^au`|p!>EY4(`dq@h}NE)*$Y}`R1y$WoSr0wVK
zOR^ZE11wCLM1hHiy$=HVRxsk(%_A+qYFO4+vHRX9L&eW<QE-MHjC@>XHMvMGXaANy
z_w^UL!$-DQePGqwqx#Q4i&z499<2fZ?irt$5Yln!h&RW`1Dju1Z=hityZS$LuC49P
zySG^cUrFEv7w@a%q8zO99bc7n59e+$*Jz?;@-;fVedBW(BcFu5v0fPt@qDkM?svR8
zGm-c93Tds-w}!uOr_<9V?FX;CkN*1Kw@G{L!&AQ^(BH8@q$r$sfwmESf0Oyfw&vzs
zGzpLe5~Y-tZcfs&Xmri?I_Q$aD|$eU)8O~#Gmxv4($CE|tYtECqq~?xoy5Yi!|K(U
z!l$%Ja+`D?MCyO6H^Y3oWt7~u|6EV`wX|o;FG~~~^42uC@=Ld9ZSCkTQT(!Q!~M*p
zxs2{5M_+@c-wh3{lu2djSj<rNyT!Q#-^bajG;z$2ECQ%r92mWxbazQYhbwK90KX+G
zt#z&`0K_(J>KX~Vp`&rA6^N>)ZokZ*hGrhEl3FU)`>!@?JW6ag$eDW(!@$<eEqh7j
znGuh&ulM5>GrS)dsx}R~B7>qe-|3X-ef5OZ_sa@*M~3-}mt=lqGAQ5bec^HMMYHk;
z_V?MR)={OU(@ab=`FQ_2H}?0(|1CuQ<xTR3fA9~E;lB7JH_4T2<@6Q%4y7Ew-S_Bo
zF88OChniYaLqdBmit%&xl^(XgL!;cR%DnV-4Rkv82VaQ%h9Z#g1O7r|DDmYHVQI&X
zc|b4L7-;gP1Rpr6?>)tEqoK|(<>Oz|=S5*NGpB5ZC7Lq?()P%Y-W<C!x+i12Vf%I&
z51$WS3-;+kUC%G89RiV|T<iWuB~R2{e5|;CaF_>fPupgLYx4CuDB&W`l`~v&TJL{@
zeR@*$XHGc3$Aq$q*`DH-Q;(Vc>SQrG`x|yG|Ba{Hggv|D<CzH)XYar66aPDiXA)xg
zC$aS__DiW{^W)ej8F5L1-}Gq)v(4a>)3Rl~e^zIh#(h|*_PUkdQq$vs|Meg2--<$9
z5YumhME@TpAKkXIRQozgCNSZrG!mah{VV6VjD_6d?9B5M9oXeUX=VCy@18x1`O!eo
z<BY@QJL{P|B#=~p*ucO5x)lQhp5Ygm`g70%t%g71BRlX`7J11K4*_k0*fa5)z<+;#
zy^}Vc^q3@Pm3$pU4l5)JVZXKYBLyAyxh?eHuX2&oVEs(vXlnW$yPih}c3<VS(mUMK
z8<=)eOu)7-Sh4zyVESD4Ue&2DBH@j84(HF?<k4@3@y;<Huo$R$5pn!}=qs!w0!+>(
z0ZDbf7jKUYX1^elZKTO~h#*hrn$oa%7D0q}A*T%p0JLj>t7`Deo#5pEPmyC@`hM4|
zX{2vb5hJ6c6Buzl{lQMF%8GISIpE_p_&u=KKV8Ue6Q`0Z)g(VAYam$MPFL>3%ytan
z@s}~e7%D=8juiyj9#9Oh$Zc)IZ75Cd-)}D^FtkNewY|HpWP+t+a0i>M)7v4w>CTFj
zMmx&3pvVoP&?Ca&mqTBS5lEkloYC06_*!T#O~>3<eE;X3i@3R;X%@+=IH4BCoXABA
z5EKh=O8O-}HUQ+r*N={mEBP+X!zS9*)P#Lh-bJ!KIwoewRi5TB%#MKdYHnrS#A9@<
zLbm?5`_n(hg^v$p4Djt_$!Y8O`c<O|vR-`Uwa6^tMp{5-W+s?76LT^$GN_Dh@BgCP
z*q47U@u||HD3cv)%8bS1LKi5Ursae^xv#IRtWbR9#C2=9anQY~o;*R}G>}C_@7^It
z3YH}`7P901>72H7<Cg(qHm|uUr6oP>&w%aU=pIc^yHF@HZts4;fbCDmL$3hy|Ea5O
zXOpW=WSAGYO>}ZyR55oy^5EQQ+5ByClgBm;%_H@x*ZuhMdq%sM;{2>*{$-(p(+l;i
zt1AWp8YRUw=l+tI9WhrVHf-lUy=S+#_rb=fK{E5r<$$K)z>6A}xRaDgNX5?&L9~o9
z8UMXQDDG98Tv3r~@ZUTe8eP>J)@AdbD)`2_2%KaC0|S&ga%odYO7K{(QheS?t)xI%
zvEk8Cz;p>eyjDjh9&AT&<jyE)%&(P^Dy$lJ*}f~yM2u-==Dw?Vd^Oa4dUU`kKEiYA
z%^sG(G<*Ek{&Go}#)9Zc+=-9h`pxSc3Q|B2^7C6J=72wiKSMt{?7s)VCMB!}Az^V5
zeIj7tslnylnf<9{nLhu*Z@C*Y?5!EpRZ|dZf0V?i!75^@fowQ}ZD02tOm%+!f;#z<
z*nbrEvU0QU+Q<1sP7NJJpSzWl7ssWR1$=T^qMUROR+>+oYLv1#c~azl4O}lkAC6t6
z@Q4xnGk~IP*i{}08kc0XBl~6+P<!s)t($xV8|U+pWmZGO!w}*kC9A48H7)Hq**wyq
z8=M=RW)Il&y+i958Tk|T?ZohxBG;aDFEs4hLa3;!`WvV3pijXAmz8y=g5lY5qV&u@
zh@fG0Ea`%l1(yvdF&g}AlKv@ALG<)zGa7mXwDo^Jlf8mi%TvON@y*;&I6X^MIdWP>
zEva1V%UWKtf<`8srPV=&wYT6Rp+fB$kQ50GFDMp-il^t_ryebsOjs-1dlnS~5jwb-
z`Dh$6Wdk9mqhCso%TF2(iOe*+(Rxe>?7`cL!9V7E<#CT2?W}IP^R3o*deBuGS}d-9
zmcH0%Y~Nh-X_s-%o^}_it*5ifHqfLg@@*%(342-|hVUif@JJPR<wJZ2<|Am?oAlJ!
z&kM6Ci(B=E?9g>Np7Zv#=~%|=qPp?V6XEMBUcEED7;)y_%)*8J=-t|ia?%%z20JdB
zL}L}hP&1@id{*<3gn+<lAdi!~#G=n)0rK{3aqXSMSB03EnZGqRA1rcuoQou8cXzVY
zO8kA6^mb@6YF<(BE<A?|)p5{bBzvRD7YTzdLf%^VQ<dL-<|AE7%P4SO_mFXGy;{wL
z;W&K&-$!mDil$bTy_vXqe*Nt!+a3sQ&>z1(Ys29fl47*wn}go3i<eKN@2|GFkzlE*
zrBxmJ_`y@O&oCDdDx{$izFpIoR~N$eJea5n(DmRG?d$5M%ZXhGkQ!htoDuf@Wgtv}
z=Ab56UxA!7%tdg3W|^72{DuY?CcDT;c8cJ`dzlqjmnf%bqg~j_^Yhq3E-x#KqYXbj
zE*te43759nAT?_cJxpKz!Gr}L(CO3Q(d~&iRUv)>W^Fvgl&~bZ>$x|}ne<_uPq+^d
z(R44XNH-$5%=ex^9gBzj=raVimXwh0?lDWO7hAKoSB7jCR8YZ3fqyR3Y?!=!M`OQA
z%9d-ar#U-Otho*!%{a@>;n-RJ-0Br|Vt$r<!Mcd1P^d)31ldU)_WMy+b`wB|;cS91
zS?IS@iC5pwQ>BRpd(4I1Rftd2D|1@0ilVfyr$1ddQG8{nCXqp5dlj}xNL@qD9ab8E
zf54?85?hHFfM>?(SWUBW?(J-8p(CKM!b&~)3G`@)MZlfi<}grf&K!K&%F6HcDfj@$
zO@}S<b$|c9<*lGhVy<Y-?c2A{=YB6NEJRf9)C>I$ugv`G2c!1+PL=NIiMXKdTwo4P
zY5%8t36iJMo26xBAZY>IbUJ;XWxfu}f1ijBV$G7dfFTpXfG2`0=TZq}-dQbrpnyTf
zsrprS@eiFspF^nw8}I9Nkj$|j;NDL2``+XFgc9>+DHfV1F5FE@cyrFFibFJu=swd<
z7eMam>55z1X4>hoxh-XPbjdExT~y_3yzPWKat=Z6gO&qyE1a;<Fj!M9iPqqUzy8ea
zy0u9(Xs-BU{e+fU)z&txrk}AnA6<gr(V~ouy{<B{BmtT*PGSJop5(d|9Kd(4W^)eH
z!=?LZkYKYyo_(*QEO0*+YE$^ktM{cKC187Z6b~P+f7;k)aZ&Eq3QAkHUby0+X-97P
zP;;2$rSV6x+OHexUKSR7415@Flfh<26L>IBoqG#S%0dIZ;kll^HA0m;6D~!qRFf$*
zbeV4gPsH95ctCa@^h?ZdWAJdK;=4*a>wYUNj0(!Ij~@-ZJSMIyZVTs_H4v9npy?+3
zf<{I&&6YDvxK<ASe{8*XIG6qVKmM``NwN|dp^T7_gph2aP+1XCsO(v|lu|-gkuoBY
z21-I^DI+2wdqq}+?5yA8)&2e)-+z9uf9~$1BjdWB<2=vDqWvZKY-Nv_m>_?w%qGO_
z?P0y06NJ#=3V2`~aI%2v*VZ}ebF9b1`1tMnsR&ykZF7Y~k*<Tm_XNhyvk~XpJ~0rr
z<i#t+R(yfyrGvd$lizIF-wnO{BYx)}EY0b#<lWc(z5TS`qv>}s6Wu(Mqi?syD;*oF
z+yB2UFL%CQ+n|6Z&Vr6E*l^~v=wj`u-5zvO?tMFoai)jV)=0%jhWq!|vXu<&t*Dn$
zSw6!-vR$#EuRP7imq=9GO5rK{;wIfX>@paAGT}JH5Z)Irz9=gz%gl7}@sXP~=40Zk
zwH#<^*kr!0^Ql8^=Mk2HRzBufJF2U&Bp4D18#@!>-g2&^M~b`!>!^zv+>cxsW~=J#
z=t_FJ*(9)x>q{t?d5=`Me%`(a#Xq$B8G)Ikh?6ZHJ&9l6fBXQu&BDwKBip#&Kx?0U
zd8mAbb~S;YZZlIz>o3nDH1$mvWp6r!9s7}Rqe{A;hl?vT!Nn-vtbZXutLzUYbr4Gj
z1L>};?&9i0K9ON@DVDywmI-5k9CE~JhiA0U?dM;ft3Px{<?hY5asw}DUQ25PYo7Ji
zmk&v)g#Um;bJhM}s`gP2Si;>24~K%FZ*hxs{>6Pg-A&)V3E!sov=_Q`DfM3<4>QAt
zZYS>qT{f?x#Nu15Q!KF@x8A=gu_Ld=gA-rHu1vp*NjAU8^45dq+PYWjM8l?9r}{5R
zIj-L@wW?|w$H;UTWIcl>8PY$Gn7;({oMf>9YO2LA&H>V%Ltl80l{bh)e<fX)+pi()
zTK~Z)-epEq^oDZjp{WillQ`-5(>_mrUi;;IEp1+3K$=>@U7_85D;LD`ZEadv;JN*w
z+2OP<O!KY!JQr!qQ@N*jaGhJ(i}{R9ebp7kYUUHRLIY3qU=oKtC2*2B&icey)D8pS
z-LRX3F<e^mt?nZuFV*?C<{#X}c7bJ=fIv|5)dQU20KAfPaZUMdHD#A56lZpoGIidL
zU;D7WLcX_G)noW?2%n*Fniw||zW3rpux7PByI|aR45XetuQ>3|&B*_8V*%jk2kB%6
z5-?5#M#C7CPy`>Z>~BjpkMCd<2%*bR#wPO53~dqa8)mI>2wdE+VR9<Q#(kXbfI+^B
z&v426M1^ZN=QTYcRiIff-n6u*zWn86E;oMuC@n>vhN&qRz<J|G8!fD*aoM~{UR+Lj
zE=8q$>{zJ!RHFQO2g*-BYptHu6fvt_J8aY0M@;+JEIJR23N}}hDPNJbKXHA6)-lWF
zC=-xTPuV)`<Gp1=kacfYO`hxqGJ?{d4Gb2p;xMlfhH>TkG%kSWE$KcB$I>Un1;`ij
zs9o8Y>1#caUv~uruR3V9M*WpvYkqN$Ws4uQ4q@xLoo_E;28A(jd#OL6pnVSkK);J;
z;2f3$YVKc9TaHj9%aeb9ZTrCMG8%5XpUCy6fBSh&{M5_=%knoo=ZlI=9^RN+dg^iT
zNlo90?^8OphXW&~ccc^)6uf*HZhaqj2yn2E_BD2gJZ8+T+Wcdb`}Br?u{GP!We0B4
zJ6pbyly=f0HgyBZgos59lw|$)EsU=N^nM)Uw{uKLbYWm*q$EXDnPVJorWV|mbB2N^
zLWn&S4caH0kPfbGca_w?SE}$w0gb|V83poKb0%QLN+*o@^(luz!P#xROt<gS|JKLe
zg#|V5lutv~Ygb#hHf=s=%5vZ+Q+Dg~)?c}~MDE$_XrSR7q*CZOZ|-_m3XVcjz+VSu
zrI57hG?8TN510q3hg)s7LxM((r>0=Epp1$trvTmV3m=X?sl@^aWILZ`55YOq(BOo~
z97WSP^z&!<aPfW)1}jbbO^Up4PuEgXkr>m(L%k_@1x<@LG1Qf@Z?o`je@!TdE?TWe
zfG@+R3!p?am;!;0$m(UH1?fsX-7_qgh0(AbF7_G$xn%BTFn3LLH3ugr`aky2T~gud
zwwtT5GGyV6j?H6zAGcb#21`Y@lA@yC2e0d@*0xJEtnu~vF#fv6XLK~V^?dqh^}u79
zl1@jEK&@AZuA)ZY!_H5m)O$~u?YvhVbJ}DQky@s*j^BPAvwt8VER2X>dIkL<)q<)^
zhWDu{%82zfXxvLPgo;z$j19M4*Sh%c9tcx=Pj`S>neyr*jE_@NuG+N&O~7dYB4o}N
zP84qdc!aVQ{MU~*E0Dro7yW5_E@*sMiI{xg*d6R1+f{$sCeKXUjhtOsr09d>EaZkZ
zp(h!K#-Qu%%60jE&|Y`P)J%9S`;~s3+*Tn8KJ#EL+us-y#Qe4XGL1dtHdqa3dDT+|
zsE0NOPszRG_%4r-&6nZf77~ywBCcMzLK0OcHYgm|t##e5syredKMq<6$`F2r2lqnC
zZVU>ro&WaG&;HV@alx84d0L9>fDn%pi(1V_R?RJ0r_{646n3hq!Ph}~ZE=7cG7KFV
zN`p8tQ#VkfV5;r<g{)ImFG<*81(3tdTT@m0L^;wq6R!L&r2Bb7HT~|t0#V?&%LgJ{
zP|eHB3sByL3wtUgp@IZx&;Q<xRoz#S=W%1TouiE8WW>JSSV8NZa^=N_@5`)IY7QOw
z{J8n}pB#{?0O<!?lODmN-z$msU*aE1Li-0wK2vA#44D&M0EQvBM2f*3F0SaJ3oZ1?
zIgRMTP{~nV4J7V{bHJg}F%MIy-7z4@&o3+}B-X=@c4&AQX#@|_5op_JpC&AdVX#bv
zRazdZ!%+{QVUHW{Kh0w?G&cSySqJ88HP1~XoS^)GPpj%qGNriNRx0YAXKGJwu8{K9
z--C?*6^E{>vgKt95IKNFttI%GsH?jOQtgEcSK!A%R`HS@miJ|K!)tyZ0u*d-(D!8G
z6*BMuaZ6RGl~}ocVpIOx7;!r^)OE<?vbHgm;In<qc^iegv{C^v=?F%{$pQj%+($6k
z{T@iMer9UUdZ;^RKqtIFtSINXd!+S!3?sg7LNM=-Pedp>w%x^7@D7e1{Q~C{qtWY;
zKy8{vkwpa*5PZRrOsruf%LE5+Aj$bHy1@7s)~pnd-oE`lJQE%k_En}E4RB|tZpVyI
zF}O7m;o&!)j3ldF7&x5Kgs@_S1hq}ftMExGD(*0KFw8WTuq-yo)N{K^sHQIw|Bh^0
z%3hy7sIvgSdNis-oyGsW<?mF*mi4uTvN9z3)XV>R0s6S6C4wY*aZ$moK#GGw2@hmA
z!M*+sm(oT&AF=@)DF-}$ayXFG_~G4Gugj<~bCOqCxS$Tk^q{<!mk0h+jiLXxGrfAO
z&ziV$vm`m3di?;YH}jb)h3Cmn-CWOz+Q*AGmhAAx&*dUCjTC4iR$o^q1`287ccr^{
z8Zf(KRoL15{d;s)g1!))TTNXZ1hi_s5S8NW-Vk&+YZ2Mc=C5pSygvi98?+gjqd-!j
zRsQzP@P_I>Sy|Z>^cuI-_=!9-oSBJF^QI%g!gcDFkExLXIwmU%P5_{LxJZV60`x5;
zB=p*;D@{6t>N@!#(^Bm6Xga#>w8wGWXx7#{52hMsk`L{m_ft?ftw4hWv4l&*XXL(k
zB1s)ayYT!XlkdnYC%t>DQ2TCl-*m<67bXN5y5aSo#Zx~m6LN0nG-d`~<82fqwEg4&
zWK1bH$bmfo=UzY?Fo0=ovzNJ(u}>d!AQ=q=Xh4}kaR$u?&Jti9Qb*85uC2^8P>{DI
zVmLjdCfo{LJ{;NuAQNH;kK78qa;VV;p6@w&=X<r013|6_-nI-{?O#3lMaCp{7tQ>2
z)&AzC@~BMA=S*Ind}4TTbd)O60eL-9e>cIAr{eQD=v&kUnXS1$x8Fh+sU4-Q-O+SO
zpR%hf!13ClsTdv6D}7ZHGogFyk!^wMw%(4GbS`%Z1U-LXLlk|zYZJXmK0oNX#1%sJ
zrwR5Y>StI;Yz!o`JJHz1oM)2?jV~!np|##XzZkpL+T*;mojc4hOj5IO@ZAY#2V1+Q
zxF~rFHbP%XnJ0IL=R5ugE{DIP6#j1h?CBu^Ny%Ju&7P(m{H}Ljv}As5Z3Q11KtF!a
zkel#h0&pXZq%0$&nIEBhlYEEbW3FflK4C)a@7<&%ebwya<pQ))qtqO|LRISxSF_GP
zPC9MqvudYn#q{TKs}8x|L)TSf@$|SA@sw^ar+VAIbB9K6|9E#lvUAg)fD`wVADr&T
zb_3)E|A;QQl34$|im+bNqj&Ef?10NKG_*J&-;Jg}T${S7toG9FMw*t+o{#)dBV^;T
z!1cAs_1(WLvBqL<eCb|~+uLyaSr1L|8{J=PJ||Yaj3@Co=rIoc-YY?9EpoEIQSCQy
zbKl*~QE0IS|E?ad&P%onPQqv-HU>)9%LT10ebe0w^D(lMFzT)&b4u!IoImD}(^ran
z65i4H>r0oyUms#S%p^#6S<h<4q4o1l-6Mk{H`4mjbG=>)FOANc-ykn~TB*$z51*XQ
zFckFKOcN0|MzB(@rEwiO-h3&)djsh}kIw)SPc#_cS!9u802+w3T#d)RBPlnl-;<?2
zBo}Oi4OQ0PE<L=N_6C-jN`FTL;%Yr{4r6SB%_iNTn2QyRohDXrZ2)HQ+CROj>(0+R
z9Z{vt@ZX|XD^=9VyIMTaDpsqOBwj1OPK>b;_im>vwk%zwUMO$ya@n%-V#ELbFm9Kn
zuM{3|*{&%{DSJyG&8@7Ls)hTN7nc6e7}@IaF%{qjAQp;xDFQVfBMu>r&qG6_5?tB6
z#Z#)Jf5vnlkB#D<2GnL`+rP5&1ch?aG2RGP=bLqYLu1Y(EB&+dgpXDET7CaZp}GEh
zTBJZq$4j5q+f7!5BOa_g`B_Lozu3-5zi7JozyG$^?nv^R_w|qCc0R-`-gug0{rLU$
z#cMVJ*S?7z|96nWWDHtsI8mpkpM%*p#^&(qW<S{3(sJS?`<`q``bF)XE%H;CuE*GH
zgaus>(J{M_o%Ool3crE2+Jon@&e_%P-R}LjKDOQd<ggqoak;>`ul>FCUY1=_TS=b<
ze+cP%noCKizfXeB${A2kG02opMqVUtFl3v;P{wdvg~>LF<?%_2EXPsLJ-vpt>7t?x
z|JJ<iIsSneUaLpT3oFHCTt_D(?syUfB?19ux2kGcLh6rw&bM`ZYkQ71qiy<5oiM%}
zehv|4+RMsve`eB4<5H7OXBFAa{1~S8o0j>%<#eRlfMr$FdW~QmqEd;p%aAwbiLZGS
zsL3?wVdK8AzWTK#DE3HH2L8#X(Bzk`tM-3is{XhC;vZd39wzi21fn%@E}pIg9t79~
zpJGdAjQ7>mKjDTXOYp9ePSG<_(`9r|HFkz8PkoK8Hga^R@qFG>e?mMl;Q7C=c9_Zb
zT(l$m<qu@Z8dPrKr#6;?VI3lrRDdJ~!?LYXGw~H6D#LV5F*lNEyf76?PAcB(_>w5N
zoe-L3R(hIW{?e22b0YU|1iTG09XY@M|NgL7Zb@y3EvAoy0Bt{Dsn#mG6>krfG6uMO
zOdwu>e~hC7-3mB@#o%Sbs#<&13)|bP3(o$VdzI~01plx`{75#M`<1(GO^8ji#E5+U
zkdL5Kg2Y=@KZpo!a64SMu-uXzKtlpH0iP019v&x0$IIAmYfBA;tc(l?L-6?|-`xTN
zR3xXupX?=;%BEg8<a8U&AFj=Q`uue4wCLx>_uF}2TKOnH^ZwsA2oc;zvEFQC;9BqX
z!M))S*rRIvjbw)<eBI!CH=xEMyt`bhu+{mPLoik4y)ynvTXOs1NYmPsB;_^Vh4|xp
zO13>c)pOa&-E=;EG@t&zAN=q+8oJ3wvXFsH)c)OomY0D9?P@wCBVpb(+s`Ms){5to
z%E>CtyHR)+mnjcVUOAfY_tot};Iw)Y*1=+|=aswBABfQd|E*2x!{;`I&@JM!y_fnF
zXV5${LIpG-%b@5Bldiwz-;WgEn*AC~wx#>V#{H}po>lmF@*B`CecKaP=GFdf+oEFQ
zELZTqBkmes?yZm1xIdN4fa$=Kg$xJ6d5kJhE~YK!Tuk&d&*8{+^@sgI%KHh;HVBsf
zBM(fNIplDMD?W+3x4Jk^<*BZG|Nkwn?@=!7f61p?Vc`<EVl&IDq{!st-u8B1l*C+4
z1v@KGY-HR)3jClN7NH*5Ue(sthKL{vSnR(r&zEp~2pu@H&Euv~%qwaR2X7u%oKC)?
zkZgOEpb7kRZFraQR#F0O*H7Ed=Wh8HuU6Pzx#&DgoM?N6>l-K&K6-6(Otc6mq#{9Q
zfuwOowt2p$VAvyYFkk=;f1PxPlw-o|p+IC{*R8E_;@=90Y}b{t8^k*g+M=Y@HHx=x
ziBKkJul*`E9|+r~wcbLUvO-|RW)d*z`?p`xUQ1v%7G+Crz?NjZiEmCzeLUa*d0t{I
z`><wHT*K_SfrWQr3N^`;I=fToDfnnDx6CFOEqy(BZ;N4bvBbUcX;8hls2@jWj7O)`
zHV!T>1>Zkuh~+&r0um=VP#NeUtn5Kg6TLU)VS0=B_r{|6Yg);kuq1N25H6t#n^zW?
zq#>j6t9X7FsL_S8nCIc@D_i>Fj4yLcxOjEGQ|P@dbqGX8Q;6o0WS}BpUqWMP@GjZ+
zcM@!=(PLbc%@sqs6x4PVoG_|Sp8VD9fjJx~hFwS3?F%Y@Zzq0qXr=53q$F)gc<1`%
z4)esprgQ6Ic4y8!di?l^FsoC?^}+=sxap8HoNVpl8;0@a3N0j9>ES*FE2)QH7%C!7
zU=<iUI6R!}bY<(m#~!`Ym%;#wY(ESGC4izKOB#)d@>Dq$<}&h$j3lT@&CJaYp0AD}
z43xGTi0!$i#}4X`S_xHP(<p2t=x_X(eIRoMT{>TuC*tu$`VM9*q@+wtG%PzaO09hj
za(fnCY+~#c5aQKt%=u7~@@gMuF9Lh@^ToBKP(1S-y39%vz74(B9&z!_TsjZZ0ban5
znj_f3$q7X<XgL`*Cgh)$I2<_`f4zMS@;Y)m1ujTYNlCY3-hW!Cpp?x}R4f9r2D53`
z+?AbLE3m51Q9fQSv^n3m4H~$lU6htj3=0oAl{NHAyC~|FP^tVqK&iC1$W~PRXf%M8
zrc0F3f`}i$^CN-VdGdt`e47M1ArZjvF3*hUZf^egdK?^Egjd)XbC3%ozUmie+`5j&
zo%U)+^m0BDb1*Kz{3KD{lb(cW06z6{17W-?eH%wwRnrlgt!exOQW<iC0B#umTJW9f
zg4p$oK@qG||G)^!K)o7Xj~#k9?Q|@;sYL>vpX=YEb_+kVorR!QN>;w&&6g_%!cuaB
za(A#E>D7#f`96P~>(oJqNx~4;JnK0K<9Z+^cyDNn;T9QpX`iTAEy7z(U%i$#PtyDM
z+6~Wtqozx4=!IYfIwzs$N2s-L00Vr@^%j#haq&woF36h%;xyPi@Z$bQ1CcISp9e%b
zH{(qUiD7+HSZnc-NJUqO!~4E@tVvO%Wf=X`e;Xw8_4A&5vA@fW!4xEfp$YnLs2Ul0
zirs+wCqtTp;r1t+Ok*<>llyRU|ISP{wgbH$Qqn>TnWJzpjn{(|ky)Xrp#35uI(Gh&
zL+hAnjQmB6%(EyeB#HGI<#np94vbo&sH-o(Jb4A=eGKtb`0%noHq&{65$BeP>lJV`
z^Y%tpy0jVqmosqF&CEVa)<I=SaRv%;Y(2CuFc}e;aCoqWrjz0k4B+0px%;c?*Bsdj
zO>cKZ6%`5Qr_$2!hq@84Xzc}AaC8G9#KE9%Kiq}y7HRU3oJM5etO%k~wH*2_F$>6O
zSfdBfT5YX;jZ8M=xS(Z4J%?mO6bfmBoq<sZ$8Bvrpr|<RR(S3Gqkj=7h>M95P`x#*
zO?wHL5te?qhJeT)=7w8Bv$p)?+DdBU{W2bPTPJB5LWUa8j88$~Apk1+^OsAM0w|P^
ztT;<o<+H!rtYm!rzMcl@O!)zJ`yWAJKXX}+={SmPoyuD}<!t))i`+$;MYD4EZLh^T
zzkO5HxE&<yi-87OPz2u`+wA<x1cW{C9lSpR#@@jg^!}KIMKL5LPtO5w$LIQ7b}K=~
zLfH!Al|>mU#(T?p&*i!j(fXKr@hgmC6msd(u24-9+~@Gy;x0rO-HwJ5IjBLw=}^7C
zqxD|xPY~O!3S6u))iqYwvO|t3Bhh_BOpTh-1XvzEHnv~^!hBC9VgB*s$59IX@J6fg
z`YSo^b>+wrDpS3kL}V4Tb!AB70{_i=g2kUXnrnJ7`M(7{v_ZHp>fo)aft^u3I5Grz
z^HlhNh6Bb1acpeYz7^-F-|fIpfjP9vT43m)*pt8iTl7@=?m=_i>WIW0DV=@THPshx
zo+4hK&u99>T}Q`GNs;5*#v_uaQe*BI=uOmT`~IkVZAA+F^o`{^9SH~n%bdhL9r8b0
z#-Ir^Km81M505vH^wj!|&Mbik>p<rE!)AdqJh)*PbEcjikZw9#TjQdlVD-H1+ze7;
zYkxwkN0xJm-uuVujTs4paXTnjA7}c3{}y&T{TKrgI)Er6fMR-l>*^|Y9ni>%!IuDw
zgi-~320+S&nI44}APx7s>`?NPd(!}j6CeD>ckVzWsq*^k&H1LUUsdiv6@v7V+o|BC
z_x15rNJ7I4D_p4v`Bz3_?8AeDPmPh)C6kosPWRfMiUf-6k7BjwEz$cEwFnBRz_CIB
z(tsM7W+;;11GhuGM-}!6Fb6>x1qJxaIWCNLN2f`A#uZiTf-Vz<TOoMIW<Uag>m46O
zgWdm*P50oB)QTd*jHKS>o&`Z>?U5+C!lj{yERGR;+T~BO1&k~akeQeWU!1|Y%T<+?
zs)^-MMA$H)dqZUI<*u<R{Rh>Trb>poon<e+WKg0a$-cd@<Bil&!?kN3&;C?OEA6R~
z58WXccl}bJ(7wU{Kw%3(X%gH{4qcgtv?8~?j?U+Cqn2okxK*j5@iX`xPj=+(;b7oo
z2-c#2f<=b&3*21}_P#jz{F0N??k!Y}^8`z&Yknv9cWof$y1ICJj~a&ICT!vM5e}Pw
z?wriFJQ}c*;tWD%{>yWE)9`gG(k*5UV(qk*Ra0_GOOwq1;Hni96$KR<ng^hAAO@xE
zQA<cfTl7z=CwlMm^mu|U3_wYJGfR+A)aybjF_M?!;pIg<YZ(J^@VZjupxn_WR&^Bd
zVh0Yy?SSDPR?k6?kx2MEdyW}HpSd9SX<e|~Thk9=n-EJOKJCT_IZeeRAr^8`9C`4F
z4doh`Q#<!UU$3`&S67tWVTuSnDzz_|Q0b3=(-RH;xl(W-Z^JsfhO6to10h>JIT86F
zKIMu%Z?rGS^b+l9_VK$wxAe;BYUc*TM)(I<9qG-5Ns0l5t_TG$X?%R^x4KH%;rAz4
zr=E<r9)O?{K3ODqQ7ycZWFSwX#3Upn7+G|2qDf0P{dBuAw&!*1;p4_8u~H#whGu36
z9F&V^)xh#Ju_ceZmOc_)PTq=?NvTh(^?`U)monsc9<}{8HI)MvDA~jF*x5zDHH@c}
z*H_#n9>~jrN1v9AlO?BQ733jwSh&}n@~@mZ!{<??^eQ(!eYZ#GVPl114%^dug3O-?
zP?ZIAkEdL4q>i=2fVP513*UhKA%Zd5wpsB<i4m5?KE0*>Vzu*gb1luyDnATVM{XgI
zC2L;v<Gvr$it*&*`dZyC$?H7v7b2r4Ns)1JRhbsd(K@LkaLuZm(TW%*HXRXUR;S<a
zARQbPlh~uWy5^;R2M^v)J+Pmr*Gx5FoVEY$d%|a=hrMZ%LZ!xvo<xyT)XoE~_%jjW
zVBEKvp!bhzY(Ll6Zy-Hy(P7nnx&zfCm+VE3NsXw_w~be9Oy2@b(X0IE<L#IyZ&(o~
zG*@-xiQ%V_>nvwK=01px)#&jwIZKWl<{T*OZfoNp(NbT_$=AlRyy)z_u)K;2-Fow_
zL<_AYAFq1b@vD7GMgG;p`P7u{szi#un9>oEZTg3I$Su*_TG2~YlGJz{WjnX!sQost
zkZva-s*NGabSVNn4&SZ|@c3@}V{<Z0?%#F~iM}5w<nB#dqo%&&0xh8OYc)r@I}jxf
zrc`z-rq{Xcik2@kpRVQmoH0dAG*AyC);zx@H;K8^|E;u;zpn_R>sK~CK5MnkD(ve?
zI&=0;PI)>V>nSm!P=50Ek<ughSvIi^ss27Ye0$+?iBo>(sK<jZLvTmVK_^s*n;86O
zECJXE;eWwA0cq!Azd+bi#(A{^tZGN)nc72YfAV{Wb(A5e+o(KdI=aU8xb(I^KE-IA
zU%h%oR9lBP)yLOR6b<D2Qz?j_RG#5(&t#^}`g%^$#KJ`EPLjT%AT!|XYPMvK95LOo
z-D=;jMsVIV(2zw%RV?fe)8zclns++|*95QMk=rMBw^07azcnMMX+x>xwice}7#Y8z
zon#=Yc!LMV^#U1iq^)9^I=<>W{k5lu(4Te0_}9MA|8`Gv&U{p9p4w`g$hXOr`!sYk
zsUA<MNjvQYkI~{R++}|J%EYsL>83rTz(tOlj;qF^yC{>Y2GnCoe5NDN2x^_q;>d@`
z9wxpR_CD>Lc-wwaFf~bSZWw!icv$naSBSB_Hq-fCl>D8Q?vquM^Fpg1iMB@ZdpeIl
z7dBb$+j+3-@K6T#=}yhmY$p>cGIF#gZ*$q?)GTrPgZBfZH1i`%;{BOO#f#HjQAQn6
z5t_}`{JeP$!vZa9l|e5yVcDrAUtf~_?*XK#iVd?}tCI(JI)1a}Z_!rAypK-~0@9<j
z5eQM<a#DChP=7O#X*6}z@cCDxX&<AVW>PQH{4WS%DoPMA_DJJZaNv-gstLK$qvTf-
zbcu$vJ^LDEX>|VegfF!sYtw&&(6Er=E(d-bk33Y8!qrEIDSVOl<nM_IIF>L}AmSSo
zARHF3a8N%zSzT9W0Vi*KaPmZ1GE{`!xx+(%tKFJ)`uJ*R%#ZM}+?cjaOfe(dcAZx5
zagu#!Sv6ALE?!k>G`T}5xUi_m`+~U;%T~5}@jX@B`$)al3yxF=ecrnNQIM@c*v3B<
zQlkbREI3`M<LMT+dqsR%yW`;TBOPl&I-;hXikFre#?xi~5H*0&QoQ!}SK$EWXGE;d
z6HIeEP_yd^+lJTq7A!8ow%xE(g!w^w;Y7Q&;5!9QQiA?@SXq&(fBvbaZ)h4Y5A8*9
z<JHjDyXEvDNiQbSc-=l4f{yp2p-x6xEG5Z6%PMc|)7N>_FJkOhN-Se)?nXr&rlk1r
z!y7{ub8~a;fzO|HAa8z_NK~>wd$?Cf=;qxux1w$oX~5}iZ9fjrfSLQ!sw}|J;&;u?
zm>OJ}2w=G^`zB!V`YO1E99+BwCsG8%5{`qfF<0|eL$i1}$8tZ%Op{tftlxQ8A>b^j
z!m)NJEr2Qtt!N%;@;G%0p-+@C(_O%b;WTkiC(mnw=<w+f`YTwMyj3%@Nh{DULhiJJ
z`1caosyH~7GS#t#WPPpdXuH8Dy?K(u#O%ESeZa=4>wM8iF7JH$`D(<p$xN6w(-c%2
z+bmotL_$x)nESG$t!>(?SJPh(^!ma7!jf2>duo~2uC?Mm?Ms_~Q(BB`ewOLv(MyUL
zsHX-KpEm}*uYUitYfk3q%|u*ixaI|&-s19iH2gLFZu5lV>%+~uat6>o57*<4z+R2w
zI?jqHH8|aha5~Fyf>nnze<TF|HYINMPQY|e6j%3J_JNYl_OpZP@+Y4B5T^?45<6|X
z4sW*XR-{o%TvP%o^=5lbN!0)`jrkCmT4<jX6;sYJ9Ta4uEhr|X*`Pg1*LduJ0(?7t
zMf7z|#_g?a*Zsow1>g7Mn+~HfRX|BE$fjy?C5!)2>#v_`=LS2kJ2O8kvXIDV+AgAL
z8oSGs4K>(cQ+3JyVC(Akrww%s{n{n{?xEr-x@*q8eFvpRIhVW%LoR@;QR)f!m1#Dy
zfBc<NQj+L5j`?SlRol}or0u3)AogsmzT~iTAZa9qN&ZJ#NyG3!fY#XvpM)~z>e=$u
zXrB_xu!#F~)TZ-KJ7)^Nl=k*_47;=i_0pAc%`ZpXG5x%&S7JFjTypVKeE3J@CA|+P
zq*+{M53dXLjYJS`w}k><T&}j-qG~aecpy!*&tHi*qX!1o`Wc%J71`CY$J9WRn2$jk
z#zw`jUhVU$%xs4{CHLM9T>Vqat4%NZxfwUc7dy&4c*^L71t%}h5xg&656$TxhcpNI
zZ(AzA)UO5*ZS7D#|2Qj@*Y#Nbn|8_cbk_yJ&9-KQ8E0!?m&CJO2|Xy@24CR9Yg|qX
zPCyPq;{NE2yLRo@|FmtT>T)`;>0<5I&@1nFZQ7h!E*xxU+_0j^^m9`_EypfWwe9Qa
zw=k2U;BOYW4FzM$m+D_=B|-g;F9=oZzG>Fe+zb*idc!u+o!7^089mQdE^-{aQQ;p(
z3~Y8)J$twg-yEflH4UxynA#tCk8=(FhgFw_+(T9PzBV<nn6~%W%S4+)V94PdYt8XT
z{#*U_4<>wzcY7j7PKcG4lu#+?l;yG~#KfGH;WUj!4GP9!R3^I|nW@T|4EkM7O~{H$
z>Ye>t+d%DLU{KqAjc!8_+z}z2!1WNl9vGrT7bU>+H}d1bA1V2Kd(U;;YS>s{#cUjy
z5ye-_v+$SkUDwU7p1le6l8ro-^^LFmzT}s2h}F7jOGKIKjMoUMAJ5e<&t;e;C@YpL
zFO^f5yD-&hKKfyHOfTnFBD2}c*yU~P_duS{BFnPgA)ppR^3%EPL@^?3XW-cY<U*TY
zSBCkXG0jTo-cqgRWuTu64|`X3i#f~_Yr{kj2FMY>dyZ^5)ZEpz0^n_JHbeR7^d0G|
zIpg+P!j{=_w=iyy{ZbY!sL4&j%u?;Q)2_CQZ&56qk>sGI=D$#=`{Y|SoqWCtlsu{p
zK+B_d5S|#NtCee}8jB)k=6KTal1w$dXU3vg=biL<9q5Lw(BTe>Y3sn>5}+T;c29%(
zav<bwqrDaXz)%Cx;gW8jBbep7Jlxr5YtGipyffzLi<VK3DD^qy>17&a$kAk-d|siU
z_QcTi<;moRxTGW=$AkYdKenb4!nboR$VL+zin<YnhrYo6)SwV82~6+;qrtOb@{bap
z^($3Q2g{|r0%0C&1n+P2X4j%w3O)`7xRkj{Ux9tq&&pw)(i`*HjrQoxp^380ZfI!O
zaw72L^Lt?hYW$g5Swklu$tl3@Ycyv0F8@Tr&Xu$Fa`PYM33@RSO}JCTnFl`FfAjAI
z!-zZ94y3}*uP=gtf~MZV!J(~HoQW3m0#r<>fQ~mKt?F^sw7UNpVBE-3gL}g0XrsMM
zih)$^Yuw04DmfVPOM6zEmU}uOVaM&%yE?n#f9lDRtLf@DaN(-1V2b}k6>oS=tn&xq
zF_G84(7q$t?8Wy2HXkbP0_njEpfZ96?{q3YJfX5?xcP(Xu(I{|$B&z-c<lYzKCf?<
z*>B|BBIokg@64ILXkCeeaHb*I=XIb9B-_4#0CxqG<`3WVq6W22l&{YuuTuwxO1v9^
z7jbw8lmW_%eL-MMLw5FiBf;w@d4O6`ipttFcBBQHmZRxAggxKHq`Kt6CpRL;6W~j?
zc~)a`5h5TjRp5<&on@*f{_yT!XywL|e1BBwQ$C()s|r65QTUGE;0Yb;r=`5u>^L0^
z-_@RZlU+yXZ(Ivyzhhxg^Z7HphU_?o%CAyk-67$XNOwRIXj9XGTOr83d`vh8(rm$(
z6&4l6qzN;b?xMJloRK3Ng5Imh)?JW!Fepl^$4+_mOk=7#R9%=&O!kx@$`wZ5KixnJ
zYM=0WeTJg<XpDv}2i+QcULoKWV>e<2H>U-Yf~HwGdP1L2O(;BX>FDh>Aii^AN_!q`
zCBj3j|3WF9gfISJTu!Be^aW6P^b^i-q@yY*Dng!Wbj00s2Pl%H;<YY=r-Pd-_1OXk
z7!>fai%TB`_Ksm4Ts&a%i(3P_;~$++`xI2|Z4}z+@sSIJFKKD;{+9gKc5tCgvcLi$
zhUUj}+r7;$qV(1^bed^+DLdbp%uXHeI!kcIT^7n=NVp6ZA)ju*Wv<Yko}Q;wu*P)n
zzp>4!r#N1h#uLU*2&%D#=>-f)FI!t%17E-_*=w?Awbp9w4<sIm%mYK)GwFSzgU3Ga
z$qwtjSQ|@z^%4$Yd`4Gm2wVPFk3ZhnZ<+hzh1fr)w5Eneb<dpOZ?IHCgpP{dlR!sa
zIf4C%5g6()I0@ozm`P`Dg#nwk{Gsc5yQSmx&;tz5sQ-Ocs4{!~uj0Oyg$RQ#o-?V}
z1BXO^Wu2a5Uu<5UTds+k=I17<Ja2JBWAJ`I+N6vo0=y<<e4#@BYJ{;N#Fu!9B$vgh
zehgUPbavj(4m10$*#TBoydibNI0C@1&+u<@fXKm>2DaW`cigu9bd^E#>G}Ji+w2`3
zDal<k2r494=x1eS9`yYqSRsk_Xmz#M=`sr~HBuh#FuY9m+~K~9x7)vt*#6TDN)T9j
zB*<CkcT!y~2l>;MZ3AZN@FB>dz2gIBIt=5K+y*g++s?^JUz>LvXY1+H+?NIay;gV;
z*fUohl>R{*E&s8pwAQiHE^iX*n5%*@Z`nhezCXV^{Pgax6FSC|)nT9ZGyE+HIsM`5
z*VVlyvoRv3al~!T1ek)8flW+8wpsdI-Eq*RuN=0MDJP%z`VMpdIKBrld6?+#fqK-r
zcMZ`qtZZyYOI>HejA;}@PQ|+@A>+9JVW1^-JUxo%Hfi{ECf~cKE?NUxliJCL51E3k
z{|)VS2S`6ifBgF`E@<qZXC;|vKhF{i>}ULVxtA6yd$xh4VfJP&`kgZE60^pKNOh+B
zTIDgTOKE0TY=ba%178S7H%#N%`LK>X<++y*K|;KJ#}1r<7L4l{`{Uqv&4+Lhj$pX^
zNX5T+^eAxfNirRv85o%Zp+&!WM)H1AQdZMw^&e>mrgs!##GXCRF^|j6hJ#%QLO4x=
zp#=HU4+{zjZA(n3O-<W(&@$SZNdAp6h)jvlQqru3p>e)bVvQ{)+FdR#z*^uzecW#^
z^B^H1fg|GAb=!02vdr_do<Fa(mvM9GhrJYZFsZ=-g0q6<2LvGI`1De}{t4=B<B@qE
z!rSbHg818p7vJ~E*K8gu<Zm>4V)NLnP$Tnc)Gi|__QUD%R})Dgg=a`4izAYkDWx1U
z${*R2$w`+$cbmv-Ef`&ZVe4iR-k7Vs;ze!kng{zYkhdQUX?k3%xSMRdx^|tQNPhT`
zOfSQ5qLZ$-t&Nrxq9M+&l^{tloR#Dt@u8Rg&`nOxwqS*)L6X`aC8_rH%`Ln<*`}Z?
zWzs1>>1jq|yzkxXcgt%5BhPl~o*fjS03YSY`Cp#O{0P7(h6AsQYA2a?<=mr~vWlzg
zxB(_Q&CS=H2MvW;b>k!wr@oV1-qUOg^VSi)7rec4U^?{y!TFfwm5E|ck&!Sfc9!+V
z8cRB`H{Y5VojfUPNl63z!knv>I8$F=Z6C_vm82mox~rF@JoE|l=K8-bDHh-ru)N($
ze*?P|AJ_T#uQ4PjR$jb&*WsAw1`d-zAqI5kHZg3fvdB7pvUA(XPM(K{92%37ck?^V
z&k>?_nS*cGwyQ^Se3dwpDz3Bf&>4k$w#k(hkxKQmC+{;B*XRm*19&C77+%{XvG{&V
zKsmRd`v4TxH};qIUE73beN>e(iWTHJ=e8VcfRwN@{0yHzfHE9;;g;D1)Tb1EuyoU(
z_M&BGW8Sibga^%1c|2zI$$R1qK%H+ghc=7s%V>W1;3=<yXrh2Zk7IGq1TUcI=&`KR
zOXo`EEpGBz!^iTqkPXOHQ58DZQe$X)l46dL4)QmadtcaCX;5mfajN)}5XIr>KpMH<
zZiMx#iPs&aZ`w6t*d}}(Qr(R|=brdHC#rr>Bb!e;>8e4wTzm9)FM8rZ@|}QZe|I0d
ztJe_@F#itVRKR^Vcm%JWi+Yyx=n;veMJXbv4l)Qi5*02|Hj*6ZmuEq7C(D*zya&up
zHok_N1<GS`%=V|wom$4g%LcsVr<Yc=Ho7sao8P~Jag8|L##78a&03W+<Y_bG4ZHZg
z9RyEL%NlL1hBlpLlK=Y?bJ@D;_iHcPu+r1ioCoWHb;!8&*PP&p{Jr83_b@v*({Gdk
z@qQY^w<__Q{<wL)YBKMjQoh+uzeevWJ*~G|1^Ss5kMCxm5RI*Aay3`Drbh)b`Y3sC
zz>P-C^tIruTn)jxeT(j$NWPQg--?)mQ$NU%n%FL-5d0u?3$Aivr{BH#huOW%&Ypx-
zf#9`mBv1oKb0{>J1OrAMoo{ZpUSQbAbT{MYxzF3u=>|pSN9QBX-rY|Be_Vpn79aoN
z>O0j&te?!wrmnh+@}>#b`EzeW$pR!69WvTH>-)vUgkqG|pjQz9D%0@T+MoAthYh^0
z3nSuL4Aj==@}%F<ZdRhBWqueVaX?!wzw?}~JhujZ%e9A7i#j5gJGR=sE5u!j5iJY~
ze&r0LsW593?DRF;D(?MJHw58n3ke)4XG_bT)rDSVc=EthhehZ^E%(!wF}S()omVE7
zO~)2gBYm~<M{|QtuTS2cW|3c<$mU5F60NZ)OC1zC;TtEEpW@Tju~K>Jlv&rIs0=~X
zFD-l4%#sc{iIJG^J<*MW%_@k65X7o&W}dlRaz8T6B~Nys=XcLm-o%lLD_1oR%~aT&
zRip!&vfEfsCDI{y=FQ~mj_5n3jr7Eg=d6G7be1&o=nqZMx^en~N`~74rOa4$u_R7d
zDC`8k_zHhqxZ~L;bOXZ%k_&1MOZPAPiNy+DCD1Ve$CQ?qs_=o2VP$C<08&a1+&YiD
z9}}Qu=89`robP{`D8Q?kL@%{wRB52UU0I|#!4)*T@E0;ENd(uG^_9j(VheRiREFEV
zT8U>}>9tw9-$s1HRQODaye8mC4^RU4$C4jPIf*!obHGJ(k(>YBZEj^1QFfCQSwFmP
zc4{{FPS*7-=3~ugH$}fFluHa-5QK#)u1aflZ`-7={qsx)%M#Bo`6T?<$`UQYn)K2z
z@|kms6d`ik3j?c_{w)M45C-`gNuR%b*_C(Z!yR)m_Hu69n2&?BJ|sT#t{RC26~4ZO
z<+DOxRG9>2HL7YRmiOW`wmHDXp)mC!>qM{ls*`J_>9mLL#zVYsx;nbnb((YM-tlJd
zT`qe+xX15{HwXd3!hG>&a`2;sq@@^5$}pjT+aCOdXkkKfVNu(%KJMf)mxj?w+)C*S
zTkXGLJ*0-MbJ{AA2OC5pyyIFln5%!qbT=jKWPM>I+Da}h?&S%0L>&x&MoHbc_Oekz
za@WHy{rij^uA=j4nbG6Yckyb#4iW<+P^v|d6#2Q-!t*?%AqBm6<Wc&V-F-w^Q>F^j
zN!N~kmoArz%~08q-QA75wjLhha4K7xql(je+EeDB_|AWQh&Q=JS$FFmD{fL);k&i9
zmJ=0kK?PGS9Yo0>CJrCE!fTPXege#FH&kE21Q=B7hhHis_1xWa5&ib!MZ&OaGi$jO
zdu;o)tLxkI^uFGuxqXdT7n3L!=RNg%t8Hb<!qP(f1a4!4O0kk2!FB)YrIi!(SekB+
z&GKXpB_7BoxTAcpr*DuMsOug-)c0_t_pGto$-`@ZSuZ}g1+r+o^?kB?KNRqPE_!;-
z-9hZ#?=D|l(!o5vh3-P6<`(O~VL=u3@0*k8!4Gyt8BORsF+{DtIsb^hfS?&PZBpBv
z7LGn>ows7$@20fj;O9@u3K=1q>P!1}z1gnhQ#8F{@1=hfCCj3*mlk&1q{v%VU1>6-
z)Isluh9Cu3S*!c<<>m9|p?@3pI+(;lrAR|cIVNXLe~-p=zO{tPfRSP8zOW|~ty#)c
z;UAh-!q!ed!#d*zW+evPKDOs<a@*~ZbRVuEJ$wMMpZ*`eprZU`aPrM_>T9`oZ@V|+
z4*AI(KmI1vEA$i6W^7-#ZygG{E+r&{EA#8B<%d&rACE2QGU?fIcwkm{;leNUCao{c
zxA}U(C=>LYhMJnuv9TW?UVS;|4yPIdxop4YL|bgHQ`9-{eoqCdB)DJgkY7WY9nl)d
z`st|d!$vqV+6I5M>xzGRc_S%k!rpy-Pi?649mP*M?;G@$axw<?v8Rf)ofxukvs=Aj
zvvUoyS3jjPV1u=evS0>y6+z9qlReZR-ywrVN&S$ap@+NE{J?6^WIR#D(AzC+E^^Nz
z?#Zh9;Urbb_&waT*8~p}x9kl%nzERtT1D>tI&phW)b8}VFy%6;C?>%T9Zw|PXf66v
zMp%xy{BC}>0KNotS1HE|Y(Cw7y18Cyc^sVsFWiLFA3AlE`WGD&_9*H;b=vb@!|*aT
z`mHg4Lo~+H4DL0%h&aC$R|}f~@P@Tf9!T-ZwXIr2Z}U3DCDI2Mp8xR&UmpU0uw*LW
z1>1o{6#w9%=f8O)$pD`N979F;uI)d>ljS{ryiq=RSINq<W6Wz|+3$|Zu|3RVA02kh
zT=ArWz->br5%Hz8$)@u{ZnDQN({=62;p7w|(%E=Vt1#PhR=1rvi9)DuZ)EEY8Xzd;
z6;qZAY$bvdT8tc6%0bB6Ktkgdx^*-MdAcCN&F~`l6h)DRkx^4u7hF4PYir5Ui_HB_
zJ(14>*IUch)SHX+GtA2ZN@{G8k{Chi!{b2l50X0y@h4Aq!2s|l5ocA3*a0$?E57OL
zVedHC^KPx27d6GRoboPGc-x*CClA>#Eih=v#y)Y#&RREj_+BZvSv`6eI$i!K8CluC
zeoEP`MPLZwCW=nmJF|m_X1|m}I0MTg{_0WN>R}<L`O_!O<AiixMf+$C&$NoNp-Zwz
zl2B>9!xs`mq+cK7Rb94v&z014h`!^=F*gyx6Q*($)hFk{W&CK5GW7emZ+~DW;=eLo
zg^IYowzlZU?L9nxQ-?iH>9J?hFM>Zup#oDM_&KPi{4?78gJh3FB!1K@ulteCn9jp}
zo@}PXg?sJY*t2GFjmX0A#sH}|Jy%kRk(Wmk0-eYGrz-^VH^vM52F|q1wPyJl?P}TO
zFCfU)>^O&d9%OU)W=e7(rsl;{@<*=QYbULg!S?AUtB#9TF3tPPYQ|K=TU-_tN%l@x
zeWf8{_1qXIw0efWk;1jQ0727Z%Pb?yR+9N71^p}zYk5|uz!xZIhVuwSV&H(<eK9sJ
zZedt?-IsGbJ{ePBvau=&NgSJwPm*#Rn$pQ0CBSLb%8gj(ih5UnLszrbJsIG=I@j9z
z{d>BPCUtzLoL@zzh10?qMLDlWN$*OG02LA5ajxZBr-a2UYqF5ePYtcxT2H>=xyl?k
z5D^weQqj~TZdMJLzOg@U|NHrX#KM)K#_p3vJ(KwxDNXs0XPWp~2mk%Xa5UIO$MF_e
zJA~xoYMK_V(U;v5Av2=eb2<}GM%%eYR;|MoSnpG-XU8ewZxtT~Ao4jvg5*q8Ts+bC
z*JEoL&Mn;X)Fjw#OR!683^7oO+_hx%y-?mv8B2Ig-J#^7CiR8#l`u7*c=VD=X=wXH
zgEV=X9q_jbgxKjfIc<~_DRz_^-X_Sk0biCIH`+Lcr1yVtZ?$!*ki6O~k)fHG=<MhD
z^VX`#py^Jl*S<9~wYQ1;pT#3<V8%qApu?=y)03hc;+fi?W~F|<V3<7YbSJ#MCzppN
zKh|yjyK<Cq_~CAM`DEgh^;$Ol%o9I7%RoLYN{-dpEawjc<^${#DG7&{3pwd%^m2&f
zp274LHd}2!N7){3Tsr2VMB4OblaDIv#7vwqO|jnYP0txl85&ZO0MVvwgu5<v@Av~1
zoLz3NI`^^?l`4|GUWS?Nkst(0R-nf?zlZ(FH)W~9W8^{Au9zBJU6rrLL1G0d`-L5w
z@(7p9#wzjWLkt^;?#}fj-&w1?6Y4fA8}IOGiRsDsfBoHd-hKRyNepAewqwoF9_<==
zV;5J?d|jAO+8bldL8Ew0Zx1VNlR^IGY6mo<Rv2r_aE2`3ii{)TpvjEjRzQlldTRDr
z5ACC){rz`sg0|W!MT%HeKc;4Qn8q7mHEh&);nPpc%Rm2|dUdj^F74@4Eo}b%`x|8@
zyo!RV?lh;>a8k1vQ#!l8tG}pp*}grxSf*?9RlaI($)?%3lM0!Cy*~WZ*ME~S?em=N
z&z$I|gQ7o0Z&=g~5KDfgi7%@tFcktM_8@)kvOhd92RS3aI{3)fGNt?H`6@=;kN{?1
z8Jq?ga%5o{8X%*rb`;ViJ`gO??o=me@*8hEpLB2bXVh4tCY@}H(o#vwWXk2_8q6+X
zAx58ONQ%rEU$q?GHT%&n_NL#Oc$K+>Y+=no;u@bjy{2SAlTArRQTEzY)VTl6EXBt8
zuCG(m)GbAUsj;<9lM<qf9@^n{6VFb2=9PVCOMGc~+gFhU$=x|qb`tzTFw>(bCsPZL
z+mw>2C?fnPD9A!CDr~SSFm95s{QmW$1;*kuf#7(y+RIRuV~Rnnwh|OXInpqUin*`V
zZG_DF!HX<ghYqgRaTD`^ZTnk1)g$jl)!3F9Io_G3A(c}tzG_%?&%W?MICI0z8TArt
z&blK^wi|vhY@K*^Xk&wd+mGqB_Z5u81qv>y??dJnww(yiGHogFt2w*r_JL<7m@OB@
zT<z^?0n{AE?BD5kD_1F^16K!&w}U<@$Y{7ZrMJL!0JZx9^j9Eh9fHd)nob4M%J_L@
za`vnTY>5-~^T7s!&@#Jq6FJeNQ}BKmpL?SEQtSAymBgNoyEe+7wqM9LUD{G8=WO^w
zHT?ni+1Q31S>4gI$yW|>Q7MrM`1%^=b0R|RvN(3#B}{$NocpSL52}jghVji%ne#0^
z2^UPw%@=&ya_~B9o@!L7<r9l+!>r@NKWc7`PWvFUkA%V=b7II8nGg{WXqlazm8IVt
z*w}cy#1gC>2>v8Fpg1$kB=?X&c5g54M;^j|Lb#u9jWPiI?a1a(+Q143#Rd8eK~~5n
z!e~aFtVCN{S~@QNij1FY2}X0oVjU)U?M`88L=fgX+nQ+Qwdn|knK|VsI55JFZx*->
zIM;$cJS1{71lf9Zj=Ugmruv*}S)cpV{k-Ij_^AGW68eLvA3}X<Wk7&=v}P{Go#u<6
zjbhdu-Jn?(o~mx<Lib`{K^lR8zq-#a1^xXc&dhi-Xk|*2^)?8t>LpK~Jh{=#1by<r
ziG?mWaJ6#*KQ_DPm_A;HJ1)jRDaFAI$=tZ<ifOl4lkp%ygI&DvzXd;JBT3Bc&C`Mc
zAJMvpcg^8-w7%NAe)JWKEpexxe#4eFZdBbRvdcS*1Pa5Y{cb<gg*a>i(or1><&@fk
z1`1ot0g|I}wzseTH3xr#sEeA@<*zRSly@C+XW9iA7{qsw#kij(h_lzCxq%<>9dIsh
zR#;1NK;^0ZV30$S0bD&vS=kLF%wQQ$Ooo_&Mng!kK#vUfA2J$&5kTkgvRBnmw|6h1
zrPh%U!^p@Od+*+9OqoDO{;yTOcGA+f`&s_F*OZ%8dDKXr1LYFjaU5r|ljtiVGY5=n
z^Ov=H*>7(Y={c>Gf8Q?jeqS_Si2>6E@ge*brRDAzysuzu+~<5uuntIFsRtze=fg!=
z%?U{+TDe4hGD%J6=iQ9^^Wf4)+)NeDGvhAWrG6YkBTSL6s+8Nyz(Dk){Y6Q?l0v?{
zTLz9rw%-)s=$bj2IKMk%&;7ghID@4M!+bg#3`M^!g@nu>yY-bVL=Cg;K`kdIiGxpD
ze+_J#g3sijXeME7W^9bCZc;hm3r0qa!W$bK(NQB1`CA)E2(q%@WWEf~0J1=s14Q!j
zi2(bdZ_3Wzi_W}qTY~j{fQo;P<b&7x;ll@xSQ05j<d!2HNZbt8%+!jSW*y$owqsIc
z4J(nDHsxz|d)Vy{tc;AWuFHAk)ie_9>nGwzH)Wg8dBvul8)G<<evPk|acLz*vy}PP
z@1g0*sds*?o*t=Tv#UMMtp>h|Ng(~1y%ej(Y3=mv3y8G^a|jZ}#st`I!WH^S)$oLa
z%mW7(6}X`TGXu{Yle(jAatK8#LoVy=<wNv7ws-m&rIsh=vEm4&hXc9~zcdJrI#`h{
zA4FxJ-Q#Kz^?Y-pSZ(Llp?`F<)!l&TU60R#k>R>_5|Vm~K)v1OMP4tDElxx{)X(V5
zJs(l~6Jq*=i?W>IgA0yu%S<VZoSB*Np6Fm>*nl)bA1jwwyN3@RXep&*-&Xs?O2T_T
z6dRlX9SZ5u)bTs@(jl6%b*nkqytdNy_~!DOTh6b8F^@ez%!krvb-$i%;HQY)nCNdp
z^<Hbi>_7x@(|Bvc&!NaZLlQs<!M$(bYQQ>WZRri}uZ2NMWMUSHX_&ofR8T<R2$_V^
z17_<P$!gxJ$68IZBO}5j;d?ruFcM%t|EC>DuYRjt_(`TMq)xc&aC*&X*ZJJZH*NE6
z<4U{ZPfsWY>co4vsb3McNbt3%m5-=`2`{pp1Il_Y%3jFppj)<v1-&UOP%JEVJnNxA
zJBcwk1m5|}o$@mX?t~hF{+9)R6x7ww7FN$`Qr5l({B29HWdkosOA}Pr#m#k&HfCm-
z=l`Fn^Z$rCKJH@Fhi%G-=}yJ2T(qlcsH?w|6=ksCdC1OnPEB^;__gqsvO~{*{2t9`
zif1*ZR7y^b-V{X>ST|X(bV8JETFg~mk(Iv>UQJa4)42bEn^UO-1Oy;NT9O0qWfr32
zk~o>Qm<~VnmJl<%D0=^{xn59W_2><s4sHH-X$x;=NU>?P^(RFdj*GNrvOdeSnBK|N
z2M?Aur8qzkZjSJq;h_LxZFBuRBmy}CHm^8HCb96i-{PK9H5HyKb_HI3HYV^db|4%a
z#>i)U9n<dq{L=jV1<aU{leLkIB7HT&76Tu~83#KCuYZu}bykXpTKhrW?Ctljf{wQo
zsYVlDZWF9_FX989N7h_3IKJeN`c-T3x-c`LIf>+7C^7-Pi}Znw65#Oi<gaNTkjTHB
zldgSj{VMw-Jeg*Q^ZI|^9Ro)T=o(2SyEQ_n&nP7Myy^GJ1%t%f=S;zNuxL$k_Td=H
z5jbI>KoX0*g{z#a<+J;VXAJ!w*B#*O9al@!>siL~#!P%E4{3c_u<uAbB_#?{_Jo&0
zqCRMd5UqQ`<_WnBP%I!40KPoR;JspL5}C%8rH!biv<4MP_w{yXx9YM}+6*i#AL3~k
z8L>D1Z&hTj{l-*9zXV<aiTc|vd!ZBE9g~@z?F{f0a~EF_ESby3X|FMj-F6z`?shL&
z*qgG^GyYeW>2!pL>x;zl>KA>Ad5)b^T5hdL*=u*|hW;lHY9eGNT;yPe=>6H|6$m6G
zkkn=`mw<IHbF}l~+hZv(YSaEOT)lZ}J-OsSg!6#>WIzMMyeEl?^vWejODC<sexJJG
zYq7?>9SR}2v%CGdGaQ;8ra45*I(G=>n?6WCy85o9{hgr0w_o2q{d(Ul34R%>^tx;E
z%3e2hK0K}9DzRstn}ZZ|DKLH^;_T5%5-nvP|84o+vTD;U?<rgYhJBNIfsBA0jl>}J
z<1n)h9sr+joqb1(N5F&o_XBGvNhBor`SjC}RJ^^t+iz{(c$i7&3A4O<-2O4xUw?gS
z0K9YiTZW2ag#JGI*uDB`o0YAp7qlq$GH9KNqHdmuDvV}o8w$fe*@T9M0y>MUt+AB?
z?+1S8k_;Y&yxX^fM$hKt3TX&tTauDkQ|amHU=AxKSGmFu46ryjb0Y(v!dXfg?M{Q3
zTY95lz?Nf3RdFAufI(Gt=(MeWDmT@mQ#*-cx?%RW&QibHpPINOPT84&n%e71`&s)3
zQ2b-9w^%T}`wHKlMQy`=tsAPYBVSXdr>BiY)uRRLp5Ekjg_nS19=l!H5z4D1nPLQp
z0e*wU9`O8MJLn*OxybT;nR{xs$>7KwSva|-YxUQ#QEY7Q+K;_T%k2&1G$vy16yXWY
z3fZSdfZ$!KmY=+((p9VYef_BOIOoN!5ftMizm=6=Er{$m;#ypk<hP+%q<xBE3Od0}
zDs!q~G~vvRR1ba_E@8lUUz9goV`+@jiItWb9GeTZtPC5LD3d}vx3%lwbg7Noo6+Rp
zwCUo|po^Tmf&#Gf3T2<f#N_WEZC+te$3{=kl=WJ1vN2Fox=2_WEq3?2ghfQeIayg)
zBqSxBa~3)X%5F{$1R#HxmoIQiS4d*1AG=y^mBI)cOR!O%IMJv-iaq7M$u~ApOVB|5
zY|bz9bN*5D%Pr4tzTtd6s6S8rle7AzugQMO0HE{d&TS}7kz^2{#Vi}CIEnhi@XS&H
z+aX5iNu{J9CqObWLxwa13**fc-4!%{i$h*Jz}N(Mlk+LDbHd5NK}4KZloS4ScWZxA
zAATXrjAE{h@x`g~btvJfNyR_KoqfkLGd}acce8Zy)7~DG^l(iW#k+79?Av;4j*nHV
zS7ElvIk_~u+?F7g`^pwfdi^srubb>$Vs$B__uF<@`td}G!G3K)$77Qu62pq!h1{oV
zZ(WUA>}5Xhb-hW60erulM3fe&o!TI2pqFr~x+@;o@Y8L$A^GODE?cS5IamRxMW;25
zra5qK_$jy9%6~Ychy1ZD>Q;i&C@X!+Cjb4S&5lG~{g*EkBuI`k1Bd?%pVH0QyNf*(
zERGO0j6!iR<om$yKO;SzL}Cs}%ghYn1MvdvfBvX~%X5zRw%r6x0IH4q!H*j4K_@}N
z&*hS@mwo|exo41_xx{W0TC;DQMs&YUWyA5zJx22$C8_?|s!FLlHDtyHsI6Iv?8X3_
z3$R;`*V{lkBNeZ+Dph_d*Sp}s$@iz2QGaJ-WKfX6v<I4eMsmtsy0CbG-(|;vc)gro
zYDi?cS)r8NFu8C(w;kOb@OD<aYFNaSr$CW|ETUk$_L`cSvsHJdmCbcDRigzxQnDq<
zFRWkyFRY@W)VqZEb^6|xE5i!COX#myc>sERr`qZ4+t9bA<#qZ7yWtK^fwu?fgzhgJ
zs0dMzNK}1BxkYRCXZUZnO3o>#83wc&rU$A%ZoC`&et5=ranH8#7Ql7dABrza+@R+%
zGA23acRD<%lpODSACw^a7NG6J3DB4Da03E*9P>Wt5r4)Ml&6QAID${`s+<&Ng-r*d
zn$9&!GvE%OAjv0YJ@|0y_m*bIyqk>NqMxf1<KwU9bMf#Dy`DyK<@xh7+x;h2hcb<I
zGMZ4nx;@^{8GgX23)C=3W23ZG@>g1(lHbCnN2c?JIE+FgDyv`lD4uv~^v1i!F2`V*
zxD{;Q#X~-eW*fhMe`}U>tI-~>5L1C`*@DRHLwLXrULGEay#J4_?||m=fBV1fl2FOY
zDw1R+N%mgZWbdr(oqeMaLUu?(vUjpql8_a$S3>qmHvjAM{r!H=|9SrZr_(v#b2_KK
zAD{buU+?R>UMtD?!M7vhjwltDUvE6C^@%&BjXM;C(@28Efd+I`UL}cC=uW`D0Gc;h
zCUOdj{fhe)k+@(#`|<2@p{z19pwfX1tn~Po_@WulV3%GUCL+1E(}06AP&EB%Y%H3*
zYH)ULAzOfp+<)nPBzfpwcKt-w%oE32X)@%c0`q-=IY&Ms%dFQrJVya8A90Xh@NG92
z4O&-!)n<(kTq*@0BP(}~*mh&)`<-R_UmnkOD}y4kOs()@6)r@%QaCJ}C}=Mjsh<dT
zmO%$EwKTE3!+h~9g2JtFuTd{vl;QXKdhuN&+LjV{OEPrtaJ<v0HW^H2FaA-7Z7RW1
zvN9;S)k!=w#{B;L0ta2a@Mcg}yn3H@;gWBn!hv_hms5u)haDSDf2Nz)SM1U8oT=cY
zRC66Z&sbTe%U)`Z>lJrnS53H#I1&@@hCch7<O76xKX5%lji8VIV{CsHt6@qD<21Zs
z2sxsY^$pQbKR)xC)L-x@9|0D*58&P%Vtb$gL?#1SpwN%cntf_I!D)(95TM_rl4Se)
z%g`HNPl9Aco%PN7YLN<6XMA&;>1|5<kVcS5Rvhdkd%hYcW}ljtab+WadM!;vhXfa!
zs0FO!<-Kp>ARlxKbqgcgW}{_s?Q@H_)eE)R$hDJUwBFmimvirJ(8C?Ar>In)H02`@
z&1X#+aVLE<<{F5z1mTXnLv!IdjmW)pD8nxWI}PQSWFGypdLbyo9J8aF>h|>=*SC|m
z!6oB;$$yuC^trK^w-RH;Lp^gNbZVUwUB1;0r8uzJ7*t9Y<vw%YxW4`Q!xPtaw2Li9
z+Dr9@sXfCn8jOTR$lN~O5@#U;6((#+Fb3I*inM|rCI9tBVV|Uj&|xI`4TrFpU3o-r
z1GAp@3&=D6J7S52fOD47!YUZkm|B(`UHh8dxqiUj*dT=F?r9!w=}5d=!IBS{R_{s>
zaj=g$$7-&+**OgtSsHHmtOr~(45ZU#&7I^;EyQYo`BZ$Zmd^@<`^|&qMAv8`TQ^!h
z%Q5;ICDC`D^RwT9l)ArtcPHT1VRo~Key&DBJO>moW~`pd<b|9u_af(Q_Rst`;XvWy
zNgbj(+;4KUKCF8ge8{ICuu5LL*w-Bro}<-m369iU;dJCBaieOElo4O%=FEdly3<5J
za-XY~Yh3;U0EDm%@Fs%-Xi;90U#G%i`}pzZz7J8$Ya#ECSy@t`L}Q9kV4mIHpRQHT
z?vF1DzuX5rU`M`d(LkjFUS&X9Ko-h)RgOEcvwC8!j4^k=0)EivT_*KuG8lBD5x&R+
zp9UFl^#dNP#bY01TicbW)4eDdfHhc;A>X{Zd+aJ4Q(}fvY|}Y`+x1Gft{*OW>;`i0
z<X!bKc!1wp{fwo!GgR-~_Z!5Iryg77ypJN&!Y9F*g=%l{l;UCOCzOj@N)^f%N1_B)
z)GS{kLN{+=S*|lY8Npeb_#ZNX?OpAJpytX0ev!d|R?`6A72s&=OGZj(mK-(w_`7~Y
z<Vnmj-MN-m;(PbUlr~*O-KYj3n?S65i|}f8MyXZn=M}<Ntv@rL9Xw0$(9SM>xcaH&
z{LGR%Ow6jNa+umTi#z(VTONY(cs>xr2P<#xq&mN2el)Sm0<gJ6P9uCozuxO%@epu4
zu0(tRBGvxBr}jJbUqs5JpPQOQ@8%o(+^Yr0EXcTBmi<wOhv~!4FJIn(;6m@M_8TIQ
z>8Px%oW|#rk)6E&tr`q|aEid0#n~CX=y!yLoPc4DAHWOCbX=HMd*>N*9)E$*O$X(U
z=I>WxQoIQK%lJ1t-JXW%Y;qkD%i{{}+OUvW!xIe6Ji_)skhsr-P2;>%MVJ@&Ro8rb
zyIimd_z;k7s5lHHi$FSh3ayF`(~v5TQU9jYz^KnT)$>H>?3rSFpQCGcW0<zxIKn^3
zUf+f2!(U~Gm$!T)zC<YUR;xGpK<w#=t>N~fovrP&TfqUM{+-WnK5%dVo-cTP$}`7u
z`5F{sR{PCFo+(}P&eeoL*okEQgM;R!+T`TqpmpL;wV;${j)n9RkjDLfV08X<iUD`h
zukn>IX+Te1`B4K{(ONSC^Yr;@6TTU)@0v?mxi*{8-X{uu4!}(zPO0XhEAWVqI~$KW
zqx}A4E;mxZceH|-x-9c+Ra-meQ@%1WWFh;x$w#))+mwX9kk5*d&jP_U(35$1n!JXt
zRkBv=(}lN!`Vu05^XaV%Q`442nJ!XANT>Cn5Gn<Ii#A&@Hh5Euu}%8BnnSF0c<bI5
zEuXg{e}AYL!Ru<x_WllBmc@@s&lp&~n_war5zSGy3OzjQ7A39}!?aRf^@uB_3zL-;
zPjC)5@E+JSNL|qqn6Tn0`uJ=Jl3e=x)jy~fYSKd>Z`W;jn+b6^yv<K!Gx|`LA!+-H
zS$NrhMwZ{cE`-6)foZh~*7IXvKwdUs4`4z2!ZuuH<>lL7qZ!kHb$;tG9CJI=<8=Dd
z=$tN&{2py{+~}M(iT93X7yN9@1sQQFET5};FQ4@orYQ<+zi+vN+v2+8=9h=zC~fZ>
zlM9KFI=-x!?<q#{$u`=}V$4UrpJDp2JVmY2neX-qn=AUA^rOxG+go>I3iLkcUH?ho
zQ0ZfnIMKSexyy~0kgAkrFk)NQ#1kfErzW>u(}@WaGwmGaQYO~^Nt{0poIrS=sboMa
zttu~kewcqu3J00&b+=cn;`V?b#Qy2dk50kZZ{9Cvp;_0PY{c5IF3G!8zaCm<S5$<d
z9A<qe2<0QTT*k0~OHj&&4N`BfN?-FscfUMLJfhV-P$*7+!U{GSyzxU5y!EOk)}Yeu
z9)=g4y=VW`AP=jeE)R#?Ur%Gn^#x5tbaCs?bg!0Ki_$$YcCyu_Zt4gU(y_2=je~qt
zX?Htab|ZpBqET)0nd+%G;mCyLo#&zqYeKxO8ZdfZt5t%AktJLd8_~4O-sR37y0|^t
zF`X^~J4|zd`^!6CL@iLALSX5ZK{yGY4V)vxTfelm!4QXD_BS8d+;n>7(X(?6bF2EA
zm<5PJmkT!7d^EvtT=1WwCK6i<SO`=EFv;Ul1X(j&jT{T~)GH3=<5%FPpnFA!CMeWh
z46iPC;~l;eCe?QKr=Pf6oC>|h&7nC%|6%-Iw!@LuIft4$*Xqi4j~HrYwNn<d6Lr1q
z*|{1_`r7BeSnal$DMiikDrq~G;-F{EEA2x{Vf}mXcC$nNM82(wrN{5fB$}b$ccE_<
z@=kca2rFV;HgZ_;$F{SCIrFFXj!%L#KhY@!>k?Y1ncq2QszWY4ZBtR4@2|qLvb+pT
zAzkhyLGQn|E-rV(NXew&M)43NrnuY0itlmdL19-Vueab>#G(fEN1rPZ(E<9Y(e9br
zJkA0_4SJE+Rs}*N(>KOq`x_&RRO^2AEKCVB{xTjp&91i2yny-W&en&KB0hUuM>DMy
zsL`|YQdR5^oguV5ITE(W&r?H2@ju17y?HQfE%1iknKS>}Rq*k=v|)ZI1&;;v7a(+d
z-~SjhYS-?2>*?wOYoPO3*Jj}chdjDEdKa@gxcBFI(M3q_AP+YoM_jw}PsZ<p*pRD}
zA+~*ZwQ=!;fP(q0KNtvI@j%<kVB*Yzl3>}sSMBkWVkW12N&2i5rDVp{iY1nG{`A4O
zvWJn+OqvhhFZ|G0FQzAIO5<d+3Hn5mbmgA6G3KqHuQ<DRj)J3!o-gg8{W+*-40)1c
z2HE};vH@ONcG-^R8^-Kd;Cx5Yak<F_EJt95VP|851<miv*{BMO2J2vvLG9j*tE^&b
z+6(DNT3SOOsN+om)lCI+(<4fYQf<JqXXwjR73c4RbWT1iIVB}r@>-C%Sgr;n&)>aw
z&&A8HWnDW#H^pI9mJh)Ai*9C1WUH(J#<GR{>DQI4?h?ew)8zi8?V73`Kcx*DaP3GN
zrjX&0Bml~mBm?Q##e(-S)i>UFV4rMoo^EkEerdAHjloo}yR=bN%iyljo1~Mj^7Z9N
z&qqsJTO#VHC+_ZRBjp5$sHz5U3VTE|<eP{GX!vTSv(JKCNU!fI@C>0(`V?*1{rajL
zFlIq(lc{&NhP#A_@|0ON#9Fv?AGd_g=ol?yyFwg-g+Lp_@EptvzzhKbQeYG2-E|A5
zd5nF*zYS(==thLj`ly3E3OYk83657i3y(@E#YpwauQ^thyzFio-|N3!wwCOkD(mrr
zfo_15k=iXyt&}P`)rJvv>h99?d$gj@x}WIRQ5Q~dZ$BYApa+Eq`r_(UUL2l%$#L}o
z@&L~GDcP5dqykTag7ESO32TF{@?Ie#D0INTGnL=fp_GR+&zQo?EP3BYcxq*Z4Ka3f
zEOAvkv%DX!e?WRA@J5ONMO0c?Ok<d2;&Q}?TX=ycWRDvw_uq(UhXB>*h=p>?%#QqQ
zSr#1gu}JUk1^fU}G%m;lQmr-_WnPMoU*oA2+m>kZ%p0A>yI8W|i^w>Q6oP@dM7VI}
z*hF7n^AuD{W7=kXulb2tHp?>oclB$3Jj7Y}r0XWP>_jfE?h5vVwCtpQ#opL<{1wG^
z<#7%iM$#+dcGxFvA}3DVNj@+fx8u<O&UPPZI-EK^K_lp?;g!^d@4tVM&Bi!*n5du)
zY4YD57<KX}rC!-LV7O!Z(-<5WAqboZ)n3TD7JcSVh5$IrP8de9Cw2l4<k;^NzKs?x
z@L*&k&sNI?>z8wwjdZJNb$?4;xRY6#N-2!0HH@p)>C;%^$8UrAimno~T*fBKbw@uh
z>pxIIkS1)ELkXw5(c>GNm@0RWYnSmifW(J^E(#(5cVOUl?4&<xrbjSams%P~M`Ix?
z8(9Gn(e(6mDqXh_4t_zK|93)7KP<JAhf%jzaM+`ezIlp!8@1s>5uJ0?^z@dL_y{R+
z$htf5Ncn&#2qA`YF|W}S%&*z?n9eyV_k-uw8*|Q<U-xoZJQH3htL7eXWRRXA&VLB)
z!tFrPOcjBQXFM&RjZb_GltU^f{hq<wUz$6>Je^+Hjp=3}D*ElsbJ(SiAiDtML)CuA
zUL1)j;H*`VFP*vVo#G=LWuiS1@%g!X@9u5VRIKR7qG-{t?Xy@ZZMdckc1ZGs=oE?B
zmk8Q*FYp^q?2LbS2yRck9iZ;-vvQZVuU26Z{6~xphB&=HRUvs<gB3QL-@gaeb>w}H
z?MiK!$o3UZ&0W6+;}1)^aFJ@1)pM+PwdeC!pNElIZ$9R@-q_F&?TKk@j$#u%zw7T`
zXnQQW`vlzuM)%g(1t6_I^bCR^`Q$aYH9z9#4`O(T&0u2zz4P7c-=V*XkG~9SZB?=r
z)2u;jP`OQ!o&YEhyMwk(Kyc%VJt5ZE_kibUe<p-}zEeE;Jql_JYy`~Afi2>7`l;ef
zNDFvZLrVv;xpW3JULF$DfYsP;#;RNVY;6tl3J8NtO--*+N0hMvCJiDs07gGJ0;&nx
zU&G>i@$$_EyeWmUy^#J0xk@l;0vN<I<9>HIHiiiiWh2kmVEtrmXKl}6zzSd=#4e?)
zJkMZBeh&&MfZIlFkNnoNo{6V}p@udaKB|26j45o4K~yCGkvNHlUT5Q~dK>qChO>17
zTnZ}Nv>waKvDwW%C`oo7r~YKZ2LW~%e3p_+0*EL_;_l8)p~w|91p=BuW)1|hRJyMQ
z!6Gt^(lKzQ+~;;;k1kLqYRB8d%G!cwl75|zVc_Bi)4&|+EYY_Y(K8o=#C@|F2?XFM
z3FX?k@62(c``}U0XEgxz&qvt)@RykZZN!B$pWPwt#R0NyOkRORYe^z})8lAst`jT_
z#$vz;KlEA4%(QMl76JkisEm}%)Dc@*%(kX&$KMjG#yK*bTpsnZb;1G@V%oOviC+f^
zIS{de(U6b_L9Q?(fnm(a$3~}g6$UwH`1X?&K7vJOl_S>24*|8;V2yrJr!X^R!ahWE
z;iWzY1z^!L;Hp7ML-XsNRFrgPB8Nd>VIITAAjF8k_6C+R=BG^=;|_b-o`P?HQAn6N
z6Al{V+1c4F%y$XwHtwW*pEcBT(Sf8pmU|?~PrBx`Oc@86QA@SO3T2iMeSdn@{PYVG
zZ*{MWN4rbb|6%>LFdKT%y{bOGb6FHlu#8%Y|N8oRBcS=aji<h~6U5ntr+*zz@sYjW
z5aACuGhzDGd#=y1zOwQ<HB~XY9}vtz9+H3WRlUCdd#{TAXR%u2BX-u3t8T4{oH#`5
z(#DMAE6(Yb%894dF84xPTcvU);h4ffeDOe?3=R?uLwqzJ$5yk7zd-5hEQ6#j>x!_}
zpU|$0i1cuM`~E#<Xt~RbTN6N@3W)7&oNBD9`dYSfbmRlZA*x%R{Hd@~VmK3vwgW&R
z7X!P0cq7Qe4WQV8=jO&&TV8Ki2PVJA!xmFlj|F3tGq~+$&LD;a^?X4AxYT_A<__i~
zs$>e_TNjAY{XxLuOj#=kS~V1fG}y??$5t<fLMC_noHK_Q*qQ<y;gnBF8C>z<12`*u
zCqPP{p8fS(3zc)`A*YXsb{qQ8>3GDIPTfaJV*Ok!o^bF3Bk!}yWjD8TdzJUVi5-q7
zZSMWXK3iUND>m(yUl8_*(dmxSJ!ch|c8L2XCjnei&mA~AsvESP%hpb8wXf{wi7*)k
ziQxTJy9!-9-Z6OU7ejIDBGq!AB~6k^fhl$aB^r!v2(CiT_ZwajRN~cvL)U6!JvCjF
z3~Rh5pMU4(e0f<JHUjH?NHGjjqJIg>@;jE@agkJ?%Gg9QU}qQv-oOK72y}KJ*UT<2
zzqIy!(UmU}y2D@-*%&r?7C;6Ne~`q1FL+KC_?mcoA3~ltWD^0SB1jzNQUZv!Y=jM&
z2F}E+(P!%lSP7650+KNW!Z#kJunb6wnVx1y?6v2jEHY|bbFaOVL^SaCvf=IL6ZT+3
z0ju1fjlzgOw585V25jUYh!(XoIzS3M&q)@(hP^eD+}<`Yl#)5OfNHa~#eGx_;Qb%f
z+)z@*v?%mOcXTMM9w!zz#fc6dU&BU7D=KKGYPD*FzRwzY&Oug}dHD#$?5zFKCS9J}
zrH(+$R}}6j6yA8#Ywh07>YIYOE>lhG@xs<Ud%StgXyDxt@SpX7E)uv}#(EwcfAt^$
z2rtV4tQu6^DixTQ*Vjd!tkHl6*{xfSD4i`AB}*7=6#<UZZ7?s_3CqCbwc5G6!^*jr
z=lpe3@MCP7efwoYfN1{*ZVMM%C&)uVTyo8gm5gx5OGrw!Wa7yME|Y-9R$E>PtYQ_G
zFt1>Uq2lB9E5su3soslr43%`=m~QURR=z`Y2P_=$1T}0ZgD=-JgR%+I&QddS1~vJ?
zTs*v^5AI@<?_$t7N*k23fy7Ji&%w9u?t|mv+J$_LJc$rur#IW?^vLJ=nN5!$3Q02*
z4TCbTx}nr&-rjPJd;LyrC)E9HBuQ>O9O4V-Zx?8$dM#c@L_-ulg2}#Vh)Mi&kv_A1
zl(s0j`xDmQR?!c%A_}u66KM9rJAyiLe4dH`+975LY~-QF2@vI@fvfiNi&_BaR>;5X
z-kI&S2D^8Dx~T03BakR>LM5p2&M{Ge8QgmUsTQMU!7Jjz8cb8ReprH$vlz=qYwu3!
zm8+n`lxkmi?e}AON2(Sq89u%)1kU)au19DHXL!rZbLmW?mT5j>&QE0pP{t3xVIgll
z9_b|H3rzc;KK~wG{P#(G*6(XjziZsW&n+y$>p$(Yh~qyj2w0Cf7QlAlK7@Hp;tM@5
zovau{qTg9ZJ_0P2bawcyJnSV+#IH-}(v}7tJ=EJ?r1UIJkUTi8$lrNOICmbsO8VTc
zgShmJH1vtGP@BN@Yt9+2G4RT9$oN8l-*f$9yLEj7<SPr(g#@@FqF}WaIZM>?ti`0k
ze`pb|4h4?HY_*$8O%hjGNzYhswTSo~@l72xkoo=Cv^L{T(qN5OV3tUKIkarBT0&^h
z|6tB(j^8ChTHU=?Gv5v6qnUPpJ#Nc8w)z)x(Qe-c63jOX7x<ruk;bp$cj6D*tm1T>
zQuOoqLUxx53od&33@$L2IlP-QpoUR3?P|d71U~&SH;wmb>FVb1#s+-~^7yF{CwMu+
zK$&Uz(yBPoShmLNUhB0R8f|CA*$c{hu3?udz2iISpDYSN9VTm9sd!4_`CHHd(*xUc
z*7q?m2Pcel;82TyvN5^aK+&yI&Hiu;#BwF?VY&19ha;n$qDtJo2mP)^SU87c0^<GZ
z;M@Z*DQx}Bl><WOMr{8EQ5fbpnkSfV@53_pnYADpp_8g9kTVdtb-}i<kE7nO@viSj
zH2tzZ)5QQ#6M9zce4Q;DGs&tEPv>*}-MhEv!5V)}?qZSTG#}JCAUg1TyHzi1uI|q|
zlfa~wI{@YMZL@H&30o&QVGyXoN<USP<CL73Fv!CW7HmCJm4GOdkxVS?c$4riHk&t+
z#QFMad;e`H_>D>LmTca#e=d<gS?8VyQ}&VWUSbVGLft=qY+@r%6B#}YpgN2SG(j{U
zu5;6P=Zzd#^2P~*JL$)}6<s-emHTaU-bCFC);no3*mMEInoYx_Wbp^T<F827o`^B6
zy>(H_#yBe3QqoZ8J#<SP6BBvb3_URvT}ZvtA^&1=9}BpJaxvc)&~3`#R@2)h`zPBk
zp{;>LrdQ*UCNS2)rJ1e*S!HQG1j|S3XU4^mUF@Zu1L*bsQ4T$AHfD|upgP~n3dPh&
z<yw-pfg;N}eQ5{^fcV1zb3FLl*P@%)Ufj)PUJB2-659@P{9l_MnwK^J(VOE1dpC*=
zjdVv;_zmCX>pd}>4>oZP60InVhyAjC`c-UU1sy%>dpm-E>n>X-H<9{*A%N;RxPhlj
zi_{U(%7s<ZkIYUYlxvlkj_C1^+wLxD4)BACo~`%V5d}LylIky_L-Rw;1dwXyOS{FP
z@1`;yUp3~OsM>rV9sM2Wm$N4uIfVPborV@}!Rp(|%ZvI)P0L=PEK{J|dPyY|=RPa4
zHS824gam<x8CH%65-$%%v*uisd$&(T8A?U~!3RsrGAHM5!Kf)>q|GT7cn(R^STl}5
zb=yBafer$A0g&r3C{Onzw14q65G>U6S>vIr?e6JOCwtfr^@9k7_Z^c!z~)8i2Nr_k
zlk|8JB4?zEud_6~SY-bdHRqzJkRE#TD{{u9iloJeJe2cHtJFCzkq6Rv$hfurs*Uwh
z_j`kL=bX}PU})dhUV+pZ1OfM8;I(Bjf}|K06;S2_H*sb7Xn+HV`HTv1y~5;@*a_7h
zRb*E0D;t2A3841?oxIujRdO<T2kraNwLu<7QJ)ns4LW%TVA6uFn@R3F4|}vMw76Xs
z`CzA5H5RaVwGSKssGgX+go-z_TwINjBgua#P82Sk`NYc~GoSHRob@RB9eu8?SyCPs
z+i7Ss(|FIf-vurAvKoju2c!x%-)o88*=n3h3ZuBKxhpC}ac)*2j=^^0wt87=a#`QO
zF&GF0O7p@Bd*&OC;o|Aq?3F$^@Zk`IrY!{e=g!eiuWma&Cb@(zC;!K1rrkM&TuWe7
zdTsDppCcsZ%Re*-K$K&c?lIb7;qTA%K)Ro?F~s!@Y@!GQFHWK6xhc(j^FD7MVWg=*
z;pETw&r1k`C`p?pN;Z0q;vqiQB!pg}Pt6})(zC)23cwb+`qfYMz3FY+D1G-hm9*ZA
zpB_-?Bpf$(MNqNc78dSK{odP)=!t{bqZ(G4Qua)3ZNbNdma-KPgc@h!66gGat4%_p
z1=RWIqC_j#H=yfUIyZ9>3AV1fgCU66)cLgom^@*~IV!zRL!8)Ifu5dpJd#*Jz}^2*
zPMZK6ahqJlrWR7Fw7X-HjBtm1ZT0xM`+lL!<~r|bqPEwIZe5*VcYdGigs}`~5ae|M
zMMX3Q=-t;c^aQ`7%N6Qi_6)W<C=GE7u2zF1JL}V_zzqQe(RBfNV%dAw*Uu(30C@%z
zyf9ojc-4vBtt+U>UxI0CRzIX#a6w{%E1uG8TeUL&UFF*;U5C5tA*ZzmbjW85W8Rd+
zgoHh8;fKcEt+Sc8ow+FwTFfpY8<p)J&iLQxNIm(hkDvEMB=S_A3EectCNHrFz8~GA
zlO%3*_4HE2mbL?@)7W!jFS^gBx4}pyTGo~~1@MbSSIS^$(nDG<5aCzbN7K+n5uB8~
z_Xl>%vn2zLL=Z7GCE-h{Lc&{;!l%G!wV`7mscNu7P%Zp@=}c(E!L50Dr;H8rUP1(l
zqr+`09P}*4(gQXp`GXeaI*n8Gp_2CFRRKRzU<r?1nkshl&za9cGLa@z0z!$o;){KV
zj~*+^no5TA^5$dwAH%~reO9KW!+U#GvP`UmjuBB&EvAd0ojcJ{X=*bq%{U`iWq{xF
zc)sw$24N6jn(x(eZFxI~Hi%jjl$BeWZQW5C;Z!AM1qC32I(1x4>;$da`MgaqKWU1<
z6$8qNBcDPH6Il-pOduyJit0}g#JiT(_hf+cw>EB$hV9b}{y&+*6_{I2F@k(Z{i!xu
z_C`oku*f!GJdJBRv`!GFW#qhZ@qREKQg<T}AE~<SgGb@YHFutRD!B(6LF=lfd(859
zf_H9RW~BZpcHT+yW*Q#~9*)6eQgOpyp^q=r)Ef}%wdEC{i4;#KCBipJCm<l(u&(gx
z;&HPQ1iU#C9A+9l0PTZO5}27m;#t_j@$wMK)7&i9eH+YVFjMcGGxX7t*Rb@i0#jhm
z4<Z(T7pG5yX4KdFdU++&C%{q;TIVZ)xPsVDJ)4|#Q>Hbek-f9neUGE}#A<v5{s5Jr
z+_L98kMHJBDax-f0g_4GPwj?(5`m?Hl;YR(_&5uOk+Xb8&p*-7=jb}o6z2x$PdH92
z_WTFvwMtfEx(EviGO4pcpZ+lruSX4wn~jH=Zn*`BLR$OJ%__ieI_H?=yaef9aTWJ3
z8XE~Aq7NQ~V@CRUm~eo*1Urdq(IBmao6%;+jejCpioljzmCr4Pt`Uov4j(B|BDi!G
z8=<5j*iAzau1NFV@7P>yut;n0p?VW!g0XUcfRpVeh&+!7!b7L6uX(bHRm@=HsqYQx
zsf&=+H!IJa2%bmyZs-qEiJj5LLT=MU@%&DkFh69EcrW%RWkJ7q2(FKhDuL!Bz(Hzg
z3T}fUpZp$hTq96a;l+*(5jEbyICz2$lARrU4fq}xaM2O9U|RgicNq4S0yAuJWSs2n
ziBXORt5c1QEa4@!wE|>>g(HtIVpf*V1!V1F`!G82<+`bC;hkJJL#zPNf>?ArVV$vp
zy65>E4o@Mrc}6cpZ+dv_fwT6a>*%UG#b5z&R>0<_@;4bY<9$|dwO&eR&Io*QGT%V_
z!HBr<b~Lb>;Zwn7(r&hOu!Ali!8S>;cu3-c_s-HoV2%k2)=@ozh@L6{LD<M|bqs?q
z`VpO<bY^L0<|9A9H|8T?&2Y`zCy)Us_QEVFT2FO!zn-pJYO|g-%U|-D?~wBH<M;W=
zK6JyuiFOAF(j7}h{7{w-+PS{9dVEUkx`|XJ$exT0rXc<zu>YZdM7Z03zrnwfOW}$-
zRLV<XVH|;2y_qjXa)4L$`F4aZC#x`m-OtjIF=+ij%hH&<9*9|o#~8b;>l-6YT(*h;
zwtL`(*AmXcS<jlC;b(BJX3`G_Pd+`bXn3J7`x0Z$Hg;(tZ2i!K{xrjd<Cp$23~Ou;
zlF)Cs6?x0rI<iFGn%54~Mz$;${~JJ&Q-7Hd+<-8uAM13jr_BhS&*K-g3C&LnJJLe_
z{QXNE@degsX542l0ZGkQPw&b%Vt5xg67!+vy~%u|lzoMYDk&`uTtkx-V8F=H$OEb+
zSQlq7*4ETO=ke(iDh~ZXc0W-IfDA$6Kxu&ffHnKzPI><y7v-m{HRGH&a;gKKQa<|+
zeIRA_$i|5^Cl0<+Nm?_id9`bv{PK}szHV<y2AiA+CobbQ>SzhvAQO?cd~zG_j{3&W
zVopEWYJ-sm7E)}(-Q_4<h11g7{*o}O9y9J-4U{!WAz1wCZ)2l`!O06i$Xs;&&O=KW
z#0S<5B2$1<ew-0E&!bV+QGb&W=>AFt2Vu#d6uVzc4?q^Q!<Sit=mM%Lt9GP!PN;Mq
z3dI_d%fW91X&lx15&@>Mpn-v3oA>6=QPsS>c(y}f6Tl4v-A2NZN(Xw8d3_Xwqe2fy
z-wA=AzIi1O5(X29<T)|HIYsWr&}A2_xCe-qjB)^}E_bU~C4UfFEA4HrjW14RVj{<B
zQ*#5gY^h#*--`DLW{I{pl8j0zdHW~lxEl>CU|)d|f!sYu&>@UG--)#xNs(L9;Bf~4
zQaWbSC#yA44_gc+YPrkig1$M4-a8MDzZxed;fsD1OC?BJy$8_6(&TmCO~$sv+QOOX
zVmeEnN-Hagwy%y%@lea?B`836J}?yu+58}CEVdLh_hkux%R+ckGP(p|;4`Dz|3t-?
zo`*&OI#NYy0@G}mZ`)?i!co*_OX(_OJLio)eOxgL+l2GEEPkLMM-WhQfa9Fw18}kj
zw;|*$BwSZi#G{V9DXDMXz}m5w<}mU>B%#CvQ{DX<t8m2s4K{5Po^Zi%O97d(`{2<-
zM3j}HVeu%6YJGchc!vih<Fe1JFZE5C`yDpS@8-U1m7aUf&UkthCB|&9Pl@U>)r?ed
z=6MRSe_C?esno_xQU4pIeLmT|?Pq9t`+(I8n(#LDAnP&066DIIut{;tsCgHWiw}0c
znmcHI6-dlL{5k2yfkJ^GOmf``0#E+MOU=_HYB>{#VWkF%XZHgEFT?aavET87zP|t}
z19Z8)5AOR-s~ZAn<ZRXiVTan3N1juZk=?hW@aL_jlvy(4Pxav@PG$gCXJ`Bj^Q70W
z^$iX2^NYvq@5#mpy37+Hkm)cRsq6>aAFmuci;s7&=RT*E;G%>!pT+sgKYXU~g>7~Y
zjx|_Z+i2W+W<CNi@RwPqy<3VD4@SJPt~x}HeIY}Yez~+Qz796$Mtro>LyPX>kQX{5
zE#%!@=yAnJV`r~J&d=`o*;_N^@H;!Z(e>SIPvjF{vFF9K*ClM(zWo$@l$PtGAkW&^
z=j|VJ6G13D*L0HY+yw|r-@bJa5jCjbO~US+lvRUW_=xtAR-aX!b)Vtq4Dqw3Bku2w
zr_iDP^g%Wb4kawu;Jx)Sig4r;6cRd*fKe3quLAH}?3^xQ3OpE#%CXcgi_E?p&HNzj
z2GCq5OO16=k`N?X82!K}!E6X$ddVON5LF5^ZQ^}}FqiqQk&IsXYtefk&pL)Ph90${
z<H-Wo*7(5W+1;HySml>9THR+A*KW4&104sz0k!Um3JS)Qn!3Mh<zT~+Gs#Z=02+(=
zcpeAKZ>ATTxfwkVDu1tV7%Ja9i>M`F4O54^lFLzXF53#9v~((e+c(T}G%XU1=sqcK
z?Oog@K9;;XQgh&ri}+W27H(m^({-s|K-kH!kY}DXs72jXY83A4bcSVjQ$@W$#P!)<
z%YmBc%p)E7<)>x4M_CQqrWzC9iI@xH%7Mj&a0rcv@3JVBrL?*WfD58B&{;@k^*wTS
z4;uN$<X*n*zhG<AnbF2YlsCt^y>Rc5APrw6*c2)-<Jj@Z&uDsF*yRJK+NdacaJf?=
z!4Nvpu#sz6=-PO7+VP5|b2ZJ`J@04G(hP#ke`Rs;rZsL?Spm$BZ{ECr?B(pNR;Z~(
zOgb2*!s4?x6EZ_`w5q@J2$=eK>P3ZS73uc)L@fYo&vkS>xfz4rAveB<7nm0Q?)BzF
zDs6;|;ZuokhiOU<wgkF<G?pk)ta%4t2HhKf(o?@k5|wu8T*t8DRo=4}?s;2Son5>+
zljcc(PV|Q2Zib;n$l1N~8BJwFF_;%JBHGe)Qxdqxi3<D~S8dm5WL13$_uKxuI@SWr
z)X@Ql|1s?2+?Q=E3N5**5Py9RPCW_30jhL)U>j3MfJR_>=hK&$*P?UeQOgcEMQOl(
z2jh0aAY*sYKEnY5y5<X^lHkwQhYJub!y4a}w^ZPdey?FO0t97X>oG|DDm4{vkm?(Q
zJHpoaNlA+@62_+Z?w7^`2G^qX8+^b#1Eh}u0Ra)T5+flI=O%3PWt`681Ai1020(zM
zeN|Oie}$rs86`W_W%*nDkb-a`KkD(oXiQMBk&<7H{0{ME^u4sklbI>1y{IMSkzCm<
zH7id8#Z~)bE}|>9=xu*-1s`24G?b7&wEQCV=ZY=yu&mjoHg^&g<B=qd0~#uk%h!QG
z-{)0m_RVxtRoqwnfnY)ZFi3ZslKd+n=p2|}IyzV7(2Insp^|0E6Z|3K>1Xhta@`6$
zb%Oj#ply<UbNgK1Wl`QXE0?849D;LFV{yFyJ~HGHz-1uiGarGCHyyw0^K}DATp;Qv
zN05@@U-Q+KD}o;<zngDEYC*{>Q!lTY(o$76j##f?XzQMY?f9L&O|<>7vVN7KaZl<l
z0g6csU9O}3Sp+Rg6ZI!}O$lAV&R-g&F&Jw#QQ@@4{A@|KPT->rYM!SIu4X*BFk<^-
zb`y9}8Shjy87N!viqBlXhxaMjqWQGn`sV0G1qCDhBlkj_D^E@qUamf3KfgQV^^>^8
zjJq&tvd(Xq4DW}UfJI~5%5I#95UP_~Dwx3r1UfZ1T7tCKdtPp8*JspAu?AFcGE!$K
z>JG{qff)kiJ<d5p>y9A6;GN%k37k1YLkr6Jqy23VrL#in8$|k*e)u4%O7`s;a5zCP
zWXz>;DrUz+4G>Pcia*^JP^$;is;FjXpjH7z2qKA+6yBgh-Vikk*)hrQVJ2rFZQ@d3
zjeie)Ki5~0W#XDjTM2+O_*1`d*>=sP*Sa!^r#k|#c&OM2fGpV3!qhv}6uvIVst^Dd
zfkla7`yYsklGvI2@q-s65JdRkY9@s=5eRQ!m1@84St!=8cAoumK1*#M0N;jXgLQP&
z|0?50AwN(P`p@!No@=Dm%)k6YI=E#9p210xBOiny!KlNX#70I4YfFwqP&%b3yuZ5u
z3u}xB$De|rWNeiS{EBx0nh_5%gI%nSwl<ukMOSKsAzt3(j$?dYRo^H$3;_NG?n6LH
z-}iB^6{3q;2QwAaq8tlRdLsy7L<IQwMEETO194-k-3!}%k+6r3knSHDX(!o>BX&FH
z#BSkgHtq!nIv-)b*Mu!4(9Q!5F?Qr=piA%G&x*6<7tU|G`#O#Nz2jnJG3@oZ{r8G`
zBl=FQOVBh^+f=L~wYH;7C`|FA2+=v_w^mUY9+{gAI?8@+<|%cT!y*qOwP|9(e8S?t
zr+lob!@;Ts=*Zyb0=_P#%lvxTh^<%s34pGMKe!meHWJuKtTNE$ypU$J9RtF;Yt2vP
zD0VuMG?*1_d6#z&A16b~Xt3c=%&G|ZS8SgM#=g{>U-{|E4@S8VWyOd_36`nLV5i~k
znU9W-6P0J>y@P3~pa5);hjA~=+k+zQ#P#QYN7&O|i>nSS_Gi!ZKuS56(dB*~j5*~+
zjNXD*(4jQ>wgjtk0D^#&2;?nfgsNmx{U3m$#zF?5PQ_43vL3wXz*a&KVj4b`AE@Fi
z{NttC3fH1(guGlI8D4-HxZDuy^xK^t1OYoZ-~BD~3f=RF$58F$PHfuIuTAht=}(!&
z#dv|t0<%~ILU)X+o#%k3y?Y?@VtC%HTqma;Xdz&ZVUY%4a8_Qt`3TsQLuAE?G>hqf
zR!<wj*))g8=m}JjPed+fq~L?zr-_8k0|;H?ggXDF9h&&pa%x2y2Mt04$5!d6sX?e2
z-=;nlE*kR=PDt&<5eRi4DF-fUAggKZpFXYk#xI^Aiu(|!cRp78sIfL&pz@Bb$&X#}
zcr6yw+a9bEZSOK7{s@c4IVZemL|vS{#F_f~+^B6H;U3UGG`*dA|8hiXSWNL}Q<EB{
zzDX8xTE^#xX9rEPOts!HpArA_g?aTZ1c;gZsL^ICQsV`Z-T$3}hGw>|i;*J8KlIt@
z&o%F(K?TS%o$qkkkUKlm<aE@VE&&wpMb|H;x0BzunmS3K5>E|_>FMc1Z0NJ{aZ*+%
zoh!@Z6FBrFjEts0*xJH%>Cz<#Bg!wsXZqI!FOAIrJML&%Krj7l_@6d|zM3*PR65hg
z{L0KjcHk+VP}5ZA%L{}q<vM?pic7@?>NwaHZq^<(J;3y#IdY1X^UmRb8!ax#@u#A}
zt#@sdihy?4>?LBhumJQ$`h(t(x4>5t-1^EWvfr)@u}B1XK#(*#93Ar83^(#wwK4LZ
zI-DGWg-^G@wU75-S8$4S>t6!7ok~^5+!mz6dC-Psz?tJtA_{47ul3zsGrg^o^iR@E
zBskRq4;TxXJet3^-581<gGOu7m=b*IAX)7;W^c=p7;f?kfB<kyJO~2vE;t@RX&te3
zuCBAn5WI6IRz7Pw50X6pzlEBH;nSR1DgVy>BX$s}F1hn9&-Js$*9}+h+Fb&doUDem
zm&~Di3%=ed@<$8DzdTqe47brgWn`42m#c^SJVG+>38Z8x%vav9{N(EQp?MG8f4yUX
z>##mPtU9q;f+(owV!#vW2I}yTrB&(2<l!<rpPx$$Wn6XdG&0iZT-Y8=`S9;KjwpDL
zTKy7k=_rIkS0%GTxHSBlaqwK$o6_w%Q!VF%ht+s>*7HvvDx+1~puuXeN-lMm4cPv|
zzSK39kk7LhSyef{x>CYqU9CSkGIAd>ofMei@#H6Y{Bo||Ug&ALG4(&CVWakyILKQT
zNLb&9bL6XoB+8H$A1r#}#*?7V)Y_{WA{N5t*`k0<!TZjm2*hS6o*<dKL3$?UpZN~z
zHaMABS(RwBxs4}xvV)TYKrlIM7kKk<k%2fCpt9FvgjNdW-Q82)Hfm~Xu@IoSj+5tV
z9E2+S{JfO5_Zib6ffCUx^@*gdOBHdAO)%(S%+2wTxBuGGkw{cjRHOnXo{<B*$9#VR
z1U07o^4Q{}hpiLxlo13<Vp9{cyErvIKK}i?Gic9Y?<XLjx71`-IaZ;&3T6l$Uphdq
z2df3(o|s;c&fNSGaRqj+DP6GFv+uIWG{h}3z0^_yk{tK}GyAl!nsU8WGps2-J?{V8
zU#fPaU~1#J(zdNucJ=EZWkGb7V?tZV>`TR){tNLY+8agfR4kln4z_vfO`X3yyt~#b
zXwSa*kaqo2_M_Wc?yKg-Wt6QMv$#$o>HP9JBzbWuj}5BW9r;rcP)r#l2i2Z;;!ouw
z@{k$Vvvzw>{dA3Wmkh>P`^qbX9{Gbc7G|IRA6f+ZKY0<Mb6HdKDorg`F25q^GE|xX
z^p3F0R6l+d-p!hd?L;oSJMn}b#p?bKOChbNjbh{O1ob>QENYCow9V*o@aptA8t;wa
zFs_OHIBdlc@!iuiK2N02JL`Y{4#dvsf^KGvUa4tl7q5kjs1q#%XAgKNvXtk7#On=}
zfl$&n#R6*BYtaY-!zX#+1eoCiB@=eUAPvm!*NI#0DT5@Kb?c-e%3v@YChuAW_n+xx
z$YIE1m|ph20OC%#+`(2sxA39vJnsXh;+1vZn>5FfZM(}>Y+YRA9MuS_oxlI8^@w=_
z&AXqW<7IF@cF2=IgZuQt8GU6nU;_ryr=*;k`Z?cwSWcVj_zSlx<nkv4*7JG|5np7{
z!5wBP(LQ3vRqxBYy}!80!<(rybhj7Q$x1HzrLM78>vzv?Dkv!xEoQ03>dDi#N#t?b
z1_MJ1N)I^ZwwnR@3!E!cx?mGIj><QG{*0>+s*qlu<4@-}5d;o4tg>(E)f9M|Bci-1
zz8J?-gn)$ZSEt|^Ej-_&J*Rm2hbBzsrymZQoa!1IQDDS=3<5&4W6)FeTrwIpAX-Ep
zgO&4tCt7d?0hM3W0gvO|>n@9#ccT{NUESSqV$tOHGdv7wJrW|L5$79+_$?yK)EV{?
z+PDg=UIK20VMnl>d@K2(*Rm(5*bh_;&gq+BfxA-s+t0m2b_n+%G8xkC)>EuYwcn{N
ze*P@UkvOY?t56tF;rKlD{6ey(Q4;@uhzKqZe|ZlP>dydNY;F();2Zc)qch?jeOS7M
z9|O;Tl}?s-b#YOL(90$73ou3dUYe`qv6}k+Hj5C?n|~G*T}cI4B{<Bn`X$o!X=<a|
z`~G*85(}1HCCwY=8jd}~0|c9dCww<7wG3Ot&f5`iNn}o+%%*Tx*?qHRJ+_9ytySjH
zyUIxtny0d^3Yo$9Ti%z^En4!&z%OCzOdy{Z-P#vK82-T{5RfVsPogKp(0oqIbG&ZR
z5UwilK8G;S5chO-&3T>#*?8p+EG4M*Z&r}=a<Z|3bk)&Fc_W<jVkjHLM4mrCXh2?>
zt2K!Y=}bT`>|cbd2o{r{wt*l+8H`map(E9FUqTS2TSzQx2|fpK>&f~wN0I_uL||&#
zS!N5Ue7Gbi=o<X5=pjbSufp9g+dW&SO)#e=mg)0;$KzB64~5MOsKvX7Itg>kNhOn{
zEnL7)-6g)lFl_`zWEf3$3;hX~es8Oj4|MGiaO?wzshaz~wnyDe(Us?RXDU(aa#g1-
zAVY%CrQM8$Z1lxl20o=AaaCT53vEtWYE+?;@~5vml61{2ncTeB?f!-%h{_q<!}zdc
zV0NfNHlpGVG1A)sadMhbnx!^E*^I2yncIo8awiE!R0>9;6Ng#VXE9>$AcD=EdhVp(
zBqyW=E%K`zm`>-x7qqAKQsWCrEpE4(<6PW8;JiOe;7cE*QDNPreLw0@{^0`{Q?-_@
zz=mODv8;0re4&5&!T!Ad)>20G($AXmV_^Sc2v|r9SN>pL|6<;zTJ9iw*1@a|yRCce
z#6QS&skUu}F2n^Wco5@a=V~m0OB0Fk9z*18eLe7h>g%CJu(wxYAroKU?J%-?XjuJ&
zqAoIeK=H|w%Bl4b_PH0*v)1?RxjX=h(@dd*{JG&M(v#8NUYJ-~xc*`4x$&pMci1_X
zs}Xi-W8(sFL<aA<<^=1ueN+7lnoHZtv2#J6K7R&|BuIPUaNvr?9iRSivMEPLi0G$5
zUxT-Aa#E+6U&y^n8ZeB7QGj=^A9eRDiwMEM!^gK*8s)nJ-7X(8{kty{9^}#Zg7A~y
zy;n2)z(WheJ0y?WZW3J)tg5C)2OF_V2B~cMKHEJ@xB(}~;@GZ3)B&eaoHYkfV?)g*
zt)dIsr?1^^89h5dds*G7Gz^+QE6^^mGU4a{?F}Y5i$$fJ^+00P$bKrNt@MT^Gi%$-
zZKUj(^bP^~Z1Krf%(6H?j?1iao9vVNyQ}=s8(8A!4?pbQBolb}!3A(Vm$OgwD>U#5
zG?!uZch6JhiUi=NA_H1St)97HAp^x6L|$U8W$zQ^3sL)<gY-t~ZnyPK5rl;-EIs||
z+Bxtu#N#yq&LnY2Ciq_zzudRy&R0E-xAAIjzEa_Jmr;krwA2Ieeo!ZWopXj~4C1(3
z-juIzK=1#mr;(biQ+yXU=Zi!-X{F!ivmO}IkOXTy>uA={^0M|yj%eYd`HykF+3^2E
zfzt)3B@l>Wm`w7{M9%6K{;=f*`j}sbFKs2zrGONa)^q41%rXrMP3|OIgzRA}BdI{j
zjvwJ`J||2jHiE#r?=yGHDBz07L0=RhZXghtVQ7OH43c*!15RT__G8^9iA#@XMwiHv
z7~}A$6AbIUE`Pk^&&Gzb)g3J50~)mH@!Etkkb#Z-0hs>7-hZT^wF_VAocwjyInF*e
zCFYM~;U(0^itB~zpADC1`P|pZwH$a)9wxs;Cw?3!q|aW;+iLHD%=KX3%Ev>U%05W>
zIIUaf-O{*r!R2#x=$A?76Zy8*eEQ_nQ_utCdkvilau`}ZLDO5@eT~tLuf089wZKgq
z$sV6s0yAYW7j;>xhl}}xBo(AyI)jJ*@!>8#aH~Lan8KRROKLBaA$*Cg`vDM7$>r>S
zXJI4$hOAOIislorQ5FNfqOHBV`Qd;~)==8(>*xg7tO3v<RSYI-1o~0)ab}9Plkt_W
z-lHJoi<x!hf3V$tb+Eu%?A2K09hD#uVTq?VJ*;-Yrb9$y%!Jc`^jsp+nv4g$SP!r$
zaCiThd|%u5QE&k-9?WCl>;klH*lBQO(!HmIs~0Gg_}BqE<J4_VucDfE#sg-a(6p(>
zaQ=K_iZiYb4sO-pyXgyBgc)?w7(SP>(9Qt3qK&xnpU)CNrk_F@?%34H-<>Jq(yr*v
zIf$Q3>4JX{O8NcdW?ws(J8wWh?y?FQI9ghX&BoKH+=j@(J=nEJD`nqeI&!)H_ln>0
z;P3%Ox;^oeqs5?B2elDnMa$5jj%XpYR8B%1L0%0$iCCfzLD$e<>|#A*$8QKe`MK~i
zniHx<d#Dy21vPIOCCuvt9h^>yNjFUDud)4mqZYoX;7g%Oh}&X4c$OX7KsvR2$tR&K
z6D(W%M*x#mDZhdDOY1jI;OySSaQkEEM^yUG;s#z*7Ldi`jMtq|%M)z)v1B^@Su$NE
zE{_4NG6O~Y@`GDrb+;UZz?`DE{x7v_xq$nM_9B*ZwIF#oh@LWKqBK1fGyC{pGj20+
zo9K~_Cey`ce&Ri0QyqXaKN-9Y)72O@Cr94)FNV#to2&0HQEn$s^L6i7J(Gh>;Qwx_
zAZlVXRR~)>Sh84@_gSGD`GewEV4`28dJM2@ZQyy2aj9#4mcrXfJwEeyjw(K2{sT|9
zM`-so<g@<%?Ey2)LFALVN5{LmhK45t*}@e7Y28TkDYHr(TRp+B<Vs3l`wpvEP@he1
zzuAq`{B6#iwC=m^O~IiLYGr7MVQ+=*)=ouGqouZNaN2-JNOT80b#&iTJv%4kSgqXq
zj<#(R)+^sRcRD5=cCa2{b1ByUkpuFWh2`bDDkB;nPo#3LgsSX(^qo3s&N^-W@iX$T
z)w))^THNZ#P#(ftzVB~S{tfIq2;AJEtbJxu=I|m*du>ZzT#NS!KlxK>SGREYcu^z*
z{S6o^R+hM9RH%C-P^K2YU+>nhDK3r*-(6#>S^_q#=JrB9qQ4N|Zf4#RbGc3f`ug0h
zJ>C59na%v;8L8Vx$-|gOC!bXY_%Nu{lN7ePQuYb_!|+A3Ft!#D4`9Ka1m~5WuIO{B
z5k5&Jl)bxmB&!}P<W8}pl|e?g4BKAaLdw~niC;}*CDE$|K^JfxKosO7Oivi}_JXl*
zB|K$)a-m?8jX$+ngLv4bs<ywp9!7un_?;fY%J+Ckj3e>zV5pIpRwoQn$`6;c8g*I7
zpv?tlS+zdUd@&B>7`FEIp&|S!JS*qNkIy2aq@N4Xe?+->ojdDc92HN_fhc2nXMWo=
zu@gkUnw@)K6z*te$dTxB;fnV2Gc9{r*-d0+5kOV}4@od3hOh}6t<u6|*l6EU*MHnZ
zUb4ga$?JD%!(rv$ldj_Ild_K4kTDUzuYaQzX)l^Ou)EbsEzuIZed3)Wp4f^Jte_KH
zc7C+ce+ixsNgayCi+dD@xkqZSx&g-`ry0BzRUH;maTf%QvW~5zz0diCvJMXw&u6LG
zY*cs5+T?MNpTQl+mQ;MIb1A|6-eSj-FvzmS(Bk7J>^Zx50+z{O7+Bs(C{p|{<`GzF
zoD2*=FCOllQ@>A#x=ERD4Avc18B2gEUrYt24{S=?{Bdu75WY#sRDlFLE2|tU3JTTP
zQJ#S(Hm}&%H$A|=NZ}4Z1n~XdsUf>cI+jp-bLJPsz0WRH>`7+!Irz+80b)8_P9AW|
z*)N8I!$RZfVPl%0=hm{(>A{nRZQ8YS>+~lxC_TX8z#IjNgEC;~QEl4;>dJ~rO2XUU
z?w|+RXiWv$r3BFVhsw$pViu6EuPrUX8dA!p(kJ^TWGdSi|FgCSI$cWFmL=pa!~HL@
zT2p%UHz4gms5IjSr6jaor?q#;!(rsW22<`4P0cqBo8cJuKnirPlb8JGo%{FC>RlK4
zJN)QqmC?K5ZOkh_zwM}V3Rw;+L@o4d`?Uo8r#?9N1_Br{FU)z)<j$4-*T`4ZS?!tg
z^8(mRWecszrfBE4C-v?BxOY}oYdkDU%3{F{%zQ=|_7^LDYx2UX>&0F&V=4ws?C3HD
z%Rs=|CWa8?8X9fo=m@a(WI%xZ+3_19Z05y8t_}f~534gulc7JMI($fnX-^X0)R(SV
zZTw7eofE)|oP|OOwZxbbUz%YLY?EO`)n?-r7UsAZx+nJ*QqrTMB%(*g$G?f3Biz$w
z(<BQZKzKkuZm=;VDAf+_FashdDW+jf#|xIF(;@>ijw_=T;VI%OWOy!gc-UAXS2G%J
zYgKvkrvl>4BuAcQ;7FA?-X=Ry$(zyz>ZN#jV8!h%M|w7qK4;ZQ^v`~L09#?1NSd3Y
zeoF(n4!+BJm<rb>FzGbi1ovRzEKn#o>o&jzL;p}c5~E;MgBUFit8A12c`2U-zD+X5
zU6V&gg0G7~HLReZ0CAIGGPLcu(`Q=h153A~uQI~U2Y(i)jwhxJ`l)eMg<xj;7a_&u
zRc>xBJZf0j2Z;xmXv2rb#eQ8M)@3nUT<2`3`oGYRn(KfD25RT?_-IRToS+HikBJG)
zqBr?Hr+!uMzyNNe@T1aKEi?YhwSdCv?eD;QjcsOP(rjJvLkru>Xe>l`hosJDdAt8j
zT7qKsI#<J@V!aXXrNbd11<8j#!{d8xe_Q#Lju<;X7TIqs{O%iz{+r|_RjcnmRqQA9
z-A}f#F57yvA8Wc6#m%_G(00gi`KG@+2^TRkx0>tBMOo*y!}=n=eQv9{xw(*yanuWN
zbOmO38)2)EpP#R}Oa43Slg>zDS*syyg(gJ{+ZwBPn+EWchnsFCDmIOKf!Vvd`lSNb
zTUp>j2ec_3MWZ?o_$sX`biu8n^^X1(%uB?9kB>gZl3x=~he^U{_Y;H2>yT+aLq(kX
zxf(U#K?EvLup@&>4oEccVLSvIK#a1p0WRg?qlJUF=LfC+SCbB$T6}E8m7n%hZD(TC
ziU%4}@pMG%v$41^wv)2z!T{>*yk!I1RP=NgMK9rE-TA)%l4&1s8^|TzHahs8_;-Ek
zG``qp4kUDjJuIM?5;^V{fj(fjZb4{jzxzqp-<k<~l`_eYSr5Q~7)Ze(NTmswkyr@C
zQ0`SrfYt|S6tS_f@Q!$ljZJ=hBX%v!%Fga#{viAzP{UKYJ_V@>^F^+Ypyk=HV`v28
z5GaY+IXQ2@1u#^+{RGNny&(!m0%-#K_Z_d}jg5_9q6)y>v$HL&t;0+xLt2{Y^8h7=
z#P%)+&)6zPPEZ_pZEJjF&+#{`UI<O6Z|h&9nnJxzhhnHWTe#)rGIh!uJOA(w?D&T7
zON1}d8yI>=pxg66xZb6%G>CL%kZF`}pnIH_k=NnKAr(9;V=pthF2=}@h+ZYoG&~Sl
zsrPi%+jU#7?Rn<;Miya`eVHLEKVYnrV5XPwOS_bINY3FPIakgl_Ajr}Qi1DrPdlU9
zwa(R-AMNe4zg2dXt7mr&yjL9ME1KX>DDb3X(V<}oLvdtH87!ua`DPDiZ-|jHV^1?g
zANVFDQn+=m(;RWpzFv1&Kb|OYyzq-KR^m<22HqJX+)n0=hAZNmBVdS@Ih<=)JXP=g
zB=y&Zn*?ZXz~;|pjXj6^K1S80Q9W9FV945~Ndc~sklsDAe8p+@OGI_`{a_Q_y=t8=
zcQVnT4MXGp_loH%IZJ_o?PjAuF`Vm6bg6~~C{2wRI|M+nUb)3h^oTo2#jh4blE+76
zozqxN))l-~!vj-O!0v6CaZFbc4X%>n=jUgAfT#D$g`I|&fN=6H9fA2{*poXgbf;#C
z1^%F;jxWNv0z$QoD6lc+T7IF4dxVnsZ9K)a1~L?JQuT%$1~q?v#VVf^pYAlCOp~0v
z4Qdm|?EAypW-%@vc?nev{X<}appy_j1ZB9lLws}h^8HAN6v|xLfP@IJR;kFNKWh<M
zrC$WY7s%d9N}#^+Mc(2s7`6Qf5J27>CYzEt<Pqe-1LFgwaZc%i>JuSsYgi1ba!s<<
zAIZl9pO$wr2qGk}v1q=<FjmKc&tF-S-<&k?ynrx}hl3|e))5W$z?#9NGW%>CS79^h
zXA)jJpOx?N1q=@@WVBjE)u-skJ)%BH=uPi4hLSDy`$cqp)V!T>tz8KJc$Rr~)?${U
zUdfX_8`W`^qdb=RioZ-8bM-Ipv$9yZwywXo0V{)VZ8x;u1<VA(^gd2c2G(gvsuBf>
z#}C%!uV;O?bpu2KnEPK78@3Ka0F-Dq(=8XVTKyJ(!CNPxb6~;%`g<ba+LMrIP1r!R
z!RlCuV0?HIg;#w8*q~|BjM-N*Vj1nD7t^^ttk_UGv6x@r0ub0dy|Q8pUD%xSKLM}a
zf^Xz)77`Zp!vg}a5<7!fFZIPFz=&*J0bv*q^(rd=_I|ZFTjEO#Z_%RB9Hz$ruOE@f
zlgW>EW5p*+#hQ`(Sw;t0KwDsxiS#)fn-a#bs*YypPtyAreb%bc1>ILamx~)Q*VdEz
zWdngHYm%yexaquCN-^dnP%bde+Q%TpZnCe0&0GD^7L3Y15Db+wb#r{_vX1AbxZYYS
zeXtw8lqwV*_@zh-_FmS<ns$?Q)dj)T1$0*s*_eXyGsuo5Fc=}>mO5#Y^XL=jz{ugB
zepGpAG)pbFgIUuiulHAYi7R-PvV;YRo0ytnd-E5Oxc_z&86G%owt1ves$^81^84nJ
zObP+-<8=$1Y8Fl!IcAd9*pN%@`GsNb*X7oDf79KgB6;<~o5*X>mu)osIUeOc@h@^a
zQMs?SyJPp;$PR#f;WHgTu$J~yE%|+=!xZWVCIIB)dSQmF0A}$2AFkdzoXh?X|Gx-X
zAt59bWsih}>^-u#kdPHwS#c>LBr7{RJ7lk{Bv~OVdu7Ys^Y?Ure&6Fa{Ba-0egDy&
z>;1l7<9wdy<4jyF@XX;gcr-dPBJ|)v=hiVq1)+UKtJ*ds>8j?k*!Wv=MUAI`cFUHr
zwX-XLPjb6?@8G4&&GmLm{1N~=$go}qpkCq&Xj>%w?}v9k#YO+F-AH+US!tyyLj{i2
z%U?V8uaE{4ZeA_X5p<;;`_>MbZo`vRmG~>`FrL|iZQ-2Y?hW@W)`5>P;v`bgV2z?@
z_QyZ$&-454t5x%G*@S+BgT2}{uTGg=DC9n$^H+?V*wpXZl>LCi14tgC2f_9G!A%gB
z5_wr!c>;6Wv(Aa*&S_$VwYos_cCbaf7HMb<MLPyDs{alA#xP6RI_~S!?2k1l`c|Tg
zS?w=2q1Tr`Z`Wnsd>ISlL8mNYwBh0HWnmYQrz29WmgCklLkqAX)PX0$StP-eEz*mv
z$*6VKpytbOtRlUP(Zb%<L8})we5}qxSF!nN?VDTH>@?UWD$QS?rMCNBb>d>ma9+!J
zZBC1xD8?)<Z3+aJ_wV8HD!aP4D66P2>`_4Ux=2^`0J1G`<aWCRJ_yJVZ6E#a>A3?-
z9qu<sRet-1iBAjnE67;=HJ6o@U0e?2zu8h)Y(*A%sdv%2$gpZ?0}L8qCDFPHs?vY3
zE2k<NM>%PgSMvwcL!)+keSN5=VB&cvA>=t1og)~VXB>DDZ0TTMwKs+&Z=fYOL`Jf=
zR08aemeJ;lOhbKfV0ML1IYS*zd~m@6Wh<c*JlVgE7NB)+|3ls{v!cDK+r{jIdpr%i
z!3_~rdio~vfee=U9hA~#2K3n(oRujw1|c#{4u(Vb(oQ>{z|JEuOIt!@`LF?<xi%rj
z!Ul&$36343AF?268FKEms(1s|2SQ#O73fjLHB>;2Ol_jF%X_zE#~|yYusdo00a)Y`
zIw=Toz#I%298Tl9uQr36yf_?#TA7mRjDKVBJ2B|cUHw@hwsRf_1U?U58j(pyDP;66
z4v&l|MjpeLxd+<}XtY|AG+;0ZmAZ+5DV3P4eiLpN)FE?@_dmSLQf2<2knx|2s&%pp
z{U_X%W_sErSn$H~Qupw7e>B(QOn5mnpBx_#^F!ndq_Yp32_TULji_na0erL`*}uL-
zrElN2Q)3Bzw(iXoete{B%5LnPdQFoB?KC|M6L;Mht%^j93a^si!th}Z`uQw1W~37X
zRMlL%e(M)jyh{;%XJy{AdgBN4zRn&mFYj>nzARq*Z1NU(w<(F&H9Pl;$E^PO-}a^&
zyeJn3gaG{44O=gn$yl$6kw`(|X8SCwC(KecU|}acn|e;E_u>X~(9+^!5sa%QhT1|h
z43^gN?;7Rx!{QCWfpk@7sE>8FlHiyIYa*@6;wz9Zu*8*=l{bV#VfK{WzXYvN#gXZv
za9c;9%>4J&Pb)kleQ~u$_JWBNST_<Mrg!+=%#LvzimCjCaUG2`C*+X%2BYzy)*;(+
zoO%t)4#D*M7%0q@kOjEWu~MC$k5ZG*4wB(7e+uupy+Q+eb~FL&%=u{E+MMFXxW{sS
zq4!Q6xKRjTUxo>N&kqn%y?{Y@Rw8;-B6^rEI$Uf{fB<a<`oQ&c{V0~=+o&>`@r}{z
z);HRD@Lfo;9M1k46<CES-NMD(Ijea-^_smMmiv(kdoIt5!LQ3Z6u{yIal7AIUm(BI
zX#vL!m>1pY`m!K+CA5W2=b|b=J|P<w(NfHQ^s=s4`+=Ml6Dc0dyWO|u)I&lHofO(1
zgCr+F^IqyE+}A7_<s+I~@1dTj(|O`*nqGHP^||_3eGJ{(>%6S-lS&MdOAMIs{cPOr
zl4dX(n!6>fqjYk7B389C3RS{$uO;|JeR3Kao#OQj>NNT}w;#=!N+xcrmbbfHT2r<p
zTdIGdY`lPb;ZQ!p%%+n*H{xq{&cM}ae$q&lOCK-uvLN(k>i3G2sa((tV#lwCySdm0
zc?SRo(qvHrQ*No7jm7NxCt1u2dHt~O1~jR${TLY^hi+P68S=zQ4N3wKo0l5uz^$=w
zbo*7<K5<Z-)hW!`__^T;+d!<4j}pA4I{mjaK{(-wgS2@>D`VFdVH14-_ooI$S=g~*
zPzRo_E#a7vZVWoiGPjuHZj5VaX6G+Wds$WoDG#aUL(Otor@4y>axcW_{NYM6JVc5G
zxq72RU!&6H!Y!iddsj<DV1;1Fs;Kqc<59Zu=r!JJw=R&_02xuGdw*c^$;pDO=rAy9
z&&e)_3}>**V;~|MfhuJELWnRJ?TU0)W@C&`S6%Cm&S#xIXtSw2BWV<$rlq3Bzh^p~
z0Tvt)UGqJPxjmMvK_}NT^;FA+rPUS;y2r*I+`k`4%GW;dD`+nzW8&)jnO`68`>r~`
zs1zU|P28<!zt{G*gxB_tfJ$!7We;;x5n~8eFp`c`#Oy`FaCu?As;1^zB5G=yW=R1Y
zo9^pOFn#_R3n-BDwUY`?N(?)IXttKqU?x*Ff@)<+(GgJ91P&~^jNe_SBJ@90^j?U5
zP5)MFHG!j@N8$D-tLogQg@-<gQ3i@_%)Hdf4!^YS@k{HIg<l?<lG-%&CpRGd$1A?l
zw@eiK?nv3cR>s$^e^5&MiZ}`2GK>oCmr&Vqq6lO#YxzAr{e3re%@RZmw{*NhBfMIV
zE0OrOe%oiuUyar$s=V1YG<|dBFc~&2Vz2f-DR8aL`!%j$|G}c=s#UIZ!H})pmu#?V
zr7y_&_F5iysPfFe6zOt+?rh1zr~O+xiSjnzXB?o(&3f!~;Fcej;SmNI1zTM>iiKOB
zfulXPMQhn@;bLu_t0|kpmpCmNd@mW&ZWAv}2Of0d?Spp;X@C=OycgwPngLChS<nfV
z<~AXjTwyy1mxT$?Sdj+flYV1B{eH;Sj>|p2yaGy`%B6%J7c*xS6cBzw>=XU{oF66w
zx3WWRIAdzOPpU`$p+?)<I8mKoRDq*Q+87jd?0VD!vPnLxcR~24?YBken=@&!Rt~7)
zwLS?-3t|$#Zdt1TY$)xqtLq-{_w=GU$p3)RZ|B8t>oTB*u|BM+sBncsNb7{a%wP3&
zsaDi{N9;Qy+0_neJZWyc?i!`WawQUZp!e<Q_ia7{U~r1@V49b7tDiVK^0Ul798XW3
z4DtE<!M|wCxm>ESJuL1*`w8phojpv}s3mX9$CY*@1l+r!!FhW(r<_uFUwNXv#85eT
z&E77c@v|tLpe(UVhoVWJv#n+1e7y~xjG-gO<gWA`wNDjl6cq9i5sc&-V!<iv;WFvs
z;rA|uwm40F^8<g>1JJ%=;`-o~Jg)~Hq=G4ZIN|E*D(BHRs1gQrA+4x4jd?l4$jiP*
zI0&o?i|X{OsTYx~-at;WpH>nX>JamYl4h`qFCVfsm*L9}XGCl#>S}F!9zTE1O&XXo
zsk_L)?{ey*(oZ&VIL9DeoaTHpN-$dKx=Lm!W8_Co>C#?gqLpzQrJu*0X3*C#msJg-
zKjd4gm1z<$aY)v8*<F7eyJvF==cLX{S(K`SW1ko4yYyWC&12`Q_CeX-Xd~Ek6bwyg
zGyNM&Uw@aJ7Gx}xyn!cyV|j_0&}0Q$gHBa<apIZ9DM9gU!R6q|v-`39ugj0U>+ylL
zQ8)+yj`W_lp0g&as1rOy@TMj{?E3YK|A>1bx;BS-4|W(ujp@tK;Ozz}4dCtrIP|Qt
zvZ$oQ34rAQ`!fGmquN!KCie$sAS&LQkxui!QVWJHpRh(%JI)44r2eU|Kfo3vv`B=;
zRY>%nc%!`3a4!3|u#nBiLqn9FTav420`VIYlsko}!SWn`V_gQO0KEO@0`Ig(gF>Yv
z{$5cLR$F!mruZ)1E&0c`JhycmMn_k<jM&G1uN>Sl|B~i*GfF0Ww<M<{jqe|3gG0Q#
z{$yYAaY{q-Cn|&<qkP%rqLTlG+cP2nQ=KsUD}C!d^2J#+m+z9JnXUqdf1lVTp~xbX
zJ8r4G<23m~tB==3lI6~4NXc%P8*b9;gLL{2o>b!d!eN90bQptA#P~{=-s*X8-oZ__
zIpwCZIK6^%b~JrKKH~dOF(qNzeT;f|1~7@#fj*pbsM<AAg@{k*cPd0xZ>O65<>Upr
z{ogTl;GyLLpeCTB!Wda_zvZEz)u+>)yk%@PKAkm|X#$rk7;3-~UAGC65CApPA%(;I
zb^RNGa2K@lk;Di;=FWhNCWM&Y&%g~*yQLKGej`wk^b1xzY9R+1Ro?7aWe9lzDl&2>
zhVZIHfCT>2F_ifaH4AxG_-<}UV~LAJtCcxFWT;0EU4DTfAiW2$sG3qfD~Ej4%CPU}
z#PwD=&A)D>myb{pJ#n_W#Q~MdMAs$I`mMw@_Di8#!e`LvaKWe9ltuZ-{_hlw4L`we
zPL7n@4yoH}J*v9Hrog0;(-hd3Dz&M^I`;YUn8CYEt5+9%TzyMxnszE%_hVa43Zj$I
z(9zSWx~uEn_2|o&P{@*%<dIu;m%F;TrB9r_Aq4eZAYVA5@2<=xbJ}BOe0*HfFi^*>
zDBMwt9sZbdA2<6tU%+GS_kO|jG4tQ}r8Wi0NK?MAs#7X8@iI*&yU)eZovsH!1^wA^
z*V%F1UTtAxdfh>%eOk&1bp8x&V_ic{V6F|v-?NQS^!OgSb#G$-$2}lYFp^%D4o}O5
z|2UH&&69jII`2L2>J8FnI&j+e+=O8i4g<GYN3Y`<I^)2=$J{noVZDd4IapjAXzuaw
z6hpOMZj)++d#bA2Y?RK@N9Aa9aS~M)KGJmrQmx@aGI!8Ao<20W)eNM2Ffg_kN15|T
zcMD>UF5&9|qn>@2#BS=i9gb;0TVmyFmAwL)IY|3kmz|^d0<EF^0R9*bbn&-jj-M<<
z7xVR)S6TI8-tjz`Cz9*f#9}wZZBv&gDn{S-xHnz<fi-_CcobWn%@e42JSYpw4XVuv
zkwKrFyzq(zX9<Fvy(Ur5+<I+&x58wL$>#@4TWpQ^4Gf*LRXFOMggaCox_&u8eKNgf
zv;3+e_ZLXtU-#V|YrjWMXi?ky(dTUPqUgRK*UQJSK%D!AVFQDb-dEDZ?pvMr)RqMj
zI(NZT@wrTov#mvBw<v(ld!GsHrYJDozI~IBl?@nN_BwZ&4h`uEb}Txdm%5+Amgfqf
zNKhyZ*rU%;#xN<Suao_5!n(MyAf~tNU0+{P(rcO%VP-HaSRm3&B<XS9a0^GWG@rtb
zWLXf_-@Si-vtOA6OQ9avJru#W;8h*={GR6Dg(VKRqT(Tx-}}^j3Ot2!4|Vuz0c{0u
z4t7^CQAdwuX|xagdj@8m<z;=|Wc%YO*2`uGj{Ya^?C+yR&zvoG`7)zQ`)JFSQihzn
zZC{R>mEIvl-n|_K-wAWj)bq;UvslpW-JRf=954}sEkjUE^m9MPBWV<K-y7GL?pYrf
zwXYs0%1Ha8ZUnMSpj{ks+8&T~@Bgyn3%^nEWj|a0Z~W(vEhekyOEkXU^tpz7-Uac)
zv<Q-dD+lw*XK-6E9&S$kFf=s81zugQ+HKM$u5+ZA`=ZrSC^&l06(XIn{)vgRn);)f
z;G{+vFPtYrr+K+!kKS+?l%q8F9kHs0{^iZx%IvX(a}sEhHLmtESuRzEfbW8|)QkX(
z8II?n@^tYQP}<%@_aW4V(NR)D5Fmr~`c4wFy6~6ie;=8H%$I6&wqovxiW*N)R$1eD
z&13?<2<JZbhoMP^b5Ib-f;QQ&u%Nq_Yp%ayuIS8z{S@sA?AS14Ws)ass2^~T;1f$b
zQS&{;D{0MVFZ{|}O$qFZLDF@$a8a&yo)!nK2M1W<*~12&<YFPgI)|zMh&8XQyS!Yq
z3HwSE{`iYlS97C@c2I-EPn-myPhrSeSL4xYR2)erwM0Slu{+k_I_4$Fm|0*8Bf>{3
zCFid?nuMOZN1NAgXK;Ce@Og9&6N9%HE@ptei}?HfzaNmq&kLK=ge+dupZ7wXwiO<?
z;djvaHh!iTOtgb`D0ozHN}1M7{C)TZ?+rZf*wpeGT*GtIiv|)-`)qE!Xy}riT*PR}
zQ<aQWRz-g%dG!qSR8DfOEL2VZ&BV8ZcClNCa?j_*+A+JQdO%`Iv8D%m4t}<J!+(`w
zz(7cv@5zB>TjdL7wf*{(fV@cLLplY<&>vTlF%Ta*SyOvnI$OHFY_;}^3%AoH)e2uh
z(oO2%D)YazV(~=gAYeNn07_{0I0#AuU%_=;EY;czOApu39S^pDZ_(Y7<ksSI1>ry3
z0i5MoqlQbZt`zIPgn~^h@jq2gS=xEEj<X8Fv@zy!Y0QZa=QrF`%j%hfz_1QMmW0l<
zM=-1z2m-%*a%`-wt}YmbkpIM70|<=Geh(skRxwies0NyJm~b)WBfvy&#smC8lw33j
z^a7SSpLO9-1x+|=czSv+fBmZEAdK4>N#0l#jk$Pwuc4g$;?$kIfY^Y8C;2p#BE$tP
z&-HeeilmYiF)Q$!P2y9n0P_f*g;Agm28bR{ozrOekD&Z;fXbF9&d6giqs?fQlF&l=
zyvqv&`Dk<a9W%&j6wp6&0-Pkcnr+v_NycApDZihAVXrkqK=-0EY^RA2FZzndXG*Iv
z$``Yb#&^wQh^6;vhQ!|{o&A>&hO5z0p!o#0%tUFwh{_}GOhN;BZ9LYV3&4u5N+ry*
zD|y2eniW95;^)8J?#NC7z$M9=?jT9EhCz1$wP?v9b`+6v8~2`eRv^Lf!Pr-d?8=tS
z_vkbEW7AjPZoO*zd4vX&W8!TZCBEONo$JszoQ^th+eC=LRR?1(xf=}s#f6gM;ue>e
z!GBtKs0$pArluw>S6BpLD@q-{gpf$ziBmGFa)fG3$Z#hKx&TCy%(dARLb>G*=**9{
zkA6_{3*PJBu3_eS(dJJrEZ_*DXuv@h&bSC=AkB6D5sWwNU&Av^6msSC6icE{-K%~6
zFT<Z^aalu7Y?#=i%TbBI*UD#K;BCl&)z~X`i4|6Xgr9^>?05>;j2B2Jrl;^n$%%Ze
z{I#)!k(G)1iTSNsI0+e_ETKAZ*2#Sj>-jz~=?@lF&}!Ytz($h+8b*1$q2k%T9E}LZ
z)NV@_Fd+hp9AGfS#CHibmTqvhZ?3O%lk%I}ks(kq1O9Kwf`)C{+SsrWkiq-~CZu32
z8{WGNL0xJpDsjvj-iLKCj)8NIUNo0+-7(bSMp?|DP(?sqQfwfCvcL$paN^A_Z1LvL
z=Gk)WJ?zW$<s&{YvQURa<GTenr)lTG;RyJ*_--FWRn$H-o94v((*N9Cl-o9B<RMqF
z{Jj`(fQkIQMU}JnV^JdyUd5J9nWrH?u!5&w)U>p;U_j<g6MuhNNUNImyTd3zaMbuv
zZBNgdP744y#wi`1cqh%PNl0b*t1yy^B&+0T!gdDaEQ@I$3ULS$j{E~98eU{LuEa1}
z!9axs7pgS=BO_bI1KbYIBw^+UhdawK#LY;Kh7WajMGv9JO-oO2vgN;mbvd*Jj&`@?
zsbQ9o*B{m8;OxwcCUim?t#HtWCr;q~))7?nAztRHHPHZZ1aDw+*F6!DhrPUxsrKE@
zUq~M@ZJ&}Fp2ggnt-MJp!ajK?Pd07QHcRMzGp4t6xEsl4L)I(BS78_ZXg7<RLvw=)
zEK2kVaWwoWV8I39Y(DwCN%$NO07B}<(aF!FF$;zgfNB6qv{YLTYQR5pz-I=QwYK}J
zOyoHKXKo*JcN!xP*E7UF)Y;`acg);0hA=9+-7RTuBZ;dQt;nzhPUXx+xC3KAd=Ma!
zYs<@$HPCe_9{~?B;*%MVUa42Oxx2G78A9nlROR&aj%Ws)8)i1n%NA6b--U$0-+?Ex
z8QS2AbkF7gm#ahpGf?#c-K0P*jA@6ZT|NIHiB$XV+P5wQ+#c&EDB|`)ex5~>OC2_x
zHyJO-H}^}1Nlj_r(5&l`6x09BJxlI-9cx@J%P7aKbx*U|gu7qQzved##E4-hQ4}S{
z@vV9~RB8p<hs2F;p8uJrK*R;l6{tyVhtck!ENZ`(hH4L3#vfZ}sOv5I0i6Zh>d@!I
z1009u;H!$R?zKh<o&GnI;Y?aFWD5*h3?KN|Kunsu4~2hxe11P78LL2)1x4WYWWs|p
zQmcXOal4B2AViJKTu(-8RT={%oXA#29uvs}%ojC}KFqdP9WUCh+7_e)vcIJ7x~a^9
zjHWEOzy5fuRV`@f*7L6Qo!{z-_-M0H`c&fW3x>kP2%S}lYuQlDDLxs9<AG8V5uYHL
zQ2pXch>Y4pPpi|H{0Y)r4`CC~i}Tmd2cq5b#!QR3UQ$vLCkJ4z^rB%zl$+~{K&O{<
z;Cskc$Th(0TioYbF%mL;oN@mc1~m&;`UeMfv>%y5=*@1_ZMF@K`Y&&%_Q>n<YkIgd
zwQ2i7XM#S1Wb<{lSE<*^Q3?xU@kzI3CcU)N7L$01?8U<04MqDU`mY&%!Y<soh<OO@
zCH9Piy$u=mL0cK~i@d>JLfuQv0gt(3Z?<32{s!|>u=Ly!5YWy0_{XB;ALyijNu?oT
zVB!aa`XIXO>EXfpWKA02sHE$56*bmAP!Ica>N&U}EqJItEWH2PsH*jTa2YPW(iRq$
z!txP72UO*z1d>Of)a!Y7P|(ZtVXM1@BN_M1lVIK$pCfi0Y+&Kxmo=B)$N#xY=&QaL
zZur6cK5b}wh5E_aQ{j`|p}Wj%-g(T&@<}(beW)T1->Z9tn<@ma(2?ZE{$&suSlu6~
z6A4woLn0V!Jz^frVLV31@m^iYclB?{hnE$vU*W9Hpk@4GvX)-%RYtC{-;OSa3X&`-
z>6xOy*vcWISf}hb`Na?_{QD<2nDjUL*uBjuElkdTQZ`yZ2!GV4J~H*Z&swtzMBn1K
zY(UdF|8YD53ix^3!}eB%3_<3eM=h=1vTVBAr(HByYg8dB9zy(vs|HW%a$cT3K%B4+
z#{NFbI&sCNLDrEojr>p@_V;AUzg0h-qzvu9>EU)<P)7bd>1No0_?jjwNVGsqqM!&e
zI?QIn@QBagy1Y3K>60>Tb+CPaJIR(;6`jq<b-$6OuBh|Kgtb6Rt1OX0UOlU~*%U2m
zqrIdhl*{=*zD^t$pYqX_M`i7*CMwrDSkt&MHMz$1X=e3LKNon@9*w>7N!ojzHYdUM
ze?JOBoRaBM3lC_4`cJ;79g=^Tb+oy13^!W2n>%Xag1BOt>a|!%1NPzjCFeYL46)cx
zto@!@>KnMp7kpk|-#z;&iCU&J$}=c>zV4iMtODpZdwZh^i8~E&8T#ByJ5XA@NqAZH
zbNVHsXN!Xlqn`|4vbHKMR(%?Ge$F6YcNrkkzL<XXW`Cp=HP{&S_y~>O){ZR|J!lt<
zniV6>ZEo!mT^r?8boDG?BNms~D>1oVJ+jqcGC&U#M($WG2{ddry|u4L$~tN0ciMtU
ztVr1sRbGbYtYn>Q;(*fHvllfJYoWJDZz7l01I64Jo$sn0vr5zbB{-=N^i9lK%id06
z`uPue7GV}XODSIYA*e<92Pi7uzDWu3AK&?>FcP6U@W$#C8bvW$d3)31x&0M2j)u|_
zx&C{YZKYi5sKxb)!xTRsg^F)l1m%O@igTnfRB{@+<1gRD=_QR8kM%B-=CjF}t}DLT
z*ZPDjIBlZDedUrwn3#>v0FsZR?{#FKS*doRl$x2mgYH$shDSb|7RADWy-5SXYf2?W
z6O2$;e?NX?^fG^OiXP3Uy=-2qTKkbM)0Dj9{J_Gn3)OR0jvMb<|8{NT`?Za~LvGV-
zVkE^%7oH=7ts^=sY-E=ZxiTk2=cmy2zFxe$KUX(Bs`yjL<Ez2S3ycS&aW-u8MH6I%
zO)%Af;=n*Df%mXt%Qe7gu{rV55@^sj!@1*>Jom;8I8s3=g~SQrm#Zvue+s8RQGXN0
zjYs)^`p4gw!J%&`yDIl`&__?24Cjj|GPR{lXSsAS-)7_S&Se7nzhWk@3N`04(9b)<
zFA#mLw2n2+nGDie2gTRFd?)|ly?R1&4Wpp+YarrvQMW>T<RSg)Dk~38Npz!VGaPEo
ztA1FZ8%sZT3_CqP&b0Q`1?z~+TAw)@6FZ?5zL9KFa1k4%Y+hrvSFCdgTDppjP%<t4
zo>Y#dZ(^JclR_#bdT7KEv4q#|y7?ZOc|-QR*;l`T(<n$n<PMLcv)QMCCXSdEZvrrR
zm~g)>UI>w4h^R3?*g|~boBDfNn>+lp-rJ`Cb?>|UrI^qtdKUN-6?3SHcv$xdO8D;U
zsFFz}5_Re0r6F?@?soZ>D<pK-Nz%b`m&939mUH4gm;V;cs|&KIw2gK?%3Eux`x@5q
z?D8*7%gx_sSJ=V|KJkk0m2{sFgoR#6#<_9v9j&HCN=zdKg>ThFBPQ+n6)a0*G0T7A
zA%5AKuie-OFDDT<hUBxwXys+CCR8VW{O;|NF3UsN_oBjgPH1Hgh+b*@+#F5Z%bsVU
z#@n~iJeWPimlnMdh|tG%81K2k?7)|r&zh)<`<(;%gIxniL|mH1?C9!B-?haO*Q6#%
z!*im6E{V3{WehQrH}@u9M$QNn4%WQO^bLUv795_j#AB0^Ru@^gDt3<06XpX5ZsyN@
zrvBc2At}q`P?Y+*N?yuYlbE)BD`?pC_z!H4!6nEq=Ia>mvpW_R=l9)>z<tGLlHs9h
zzqEkH<@@zao94KxdSd2PBfaoyU)}o_a}9oN6$^q<&%*Zu8f(vG<z{zur|;HXn*bAx
z_7MN7;@O`kp;}|B&rty$k0xCn7vU&x4vJ*?d3vvs!u^U5@%AivmWyj_eG@t#5{ibX
zEGon&>{cA(m4&vepV~j;%1g5!eUi-B`ZU?A+*YYSq#g3wu>8Y$2n@4S--iC}79YDK
zmNqtb<CLHe&&@6NF=axh_UNtu4&^Nlh!6gvP9($vP*6nwtFeNqLJ{y<;I;$<_H$2?
zRze4#bpLH^#g+qH5-g?qayJ?yt2ZHZIAza3+3QfCrx1JWQ}EPePE?%r?)Pl!4>6BV
z7g#%=xu<Hn2CWPA3#8-d6O?+$=kIV+b{rxEZm?Xd9}n!ni(=8fk$aqcFv&HNkE+D3
zHK&-j?>kn3770_L=Qr;gTdrcWn3^w~+@8UR+;E6=DA6tvDJBDJz{<*J_I74WV=5C_
z!b_LH-}G+&05HZihMXls(8Qq^y>oPO&N+ZIbz<}y8<sc;K;@oxB|$z7?kQl=k!xZi
z4TLQ7-|Tzb++WDJ9-x|JJ`rS4+}y<7{C>M#;;1RbrLhl1gmj)@GXA0S`Q?h~)J4Ae
zWPqQPSfapaV$=Wl&)_$gFtgRZseaZD9nR2N^Ed3@W#~ol4=C5E=vh=L#pW^E=_#uR
zf3Sr%etz=Br^SSu?c0#^Tnj-D1<~<I>te5saYNVD%9)(LYa8g_<)VG(IyJf-o^7(S
zkrYu&ep=oH+U~`lF0Nf>B|~KR*!ZcDjCygRq37!ulo-4`aD)<r^Pd_?I}hO`<pk|^
z`@@IB!^8ZLXM@5O`3Q^ueu8Q&<KZ-CV=j+sq-<g!mCvUz5m2UgcR8KBgX##%g)as!
zqa-yri&o5x;?q7ouYVKHd4C~#g@1AU&Lt=GplFbpg)4HwreIx)l>Ry4T<9yedE%=v
z{v9+}DzATCO3G^N_HxY(Xf*lXhbwXXgn3kddM!@Vf<0k*u=LOOBTk;UF1m-Gh|WDh
z@@P|MR~vWFwn!v~i`PhQe}QO^5k4wT0u`^;*7(TE3!$dIul;31wh&%Qh>Zozs<f+X
zt<gkpZ^>cNo;qCnH>m-;#T+V@p$?P!OV{{IN8i@H1?oK|CA`p_Wcc7lDXuedr*G)H
zm|16VW;@D8=rlo3v1+Sd=Y8@=KI3yYY1!+)()R?c?`u3>y|K^x&I={xDcSj9HfQp8
zr2F9E^D%=EaS+~cP5pwFx%W7Ldg{&vhuw^cdQ)w>Xmc2wT8Z1+*a~=g-%PMq#ygg_
z-<ua2Q~cPl#GS}R^N^AFRs^`8q8a|l#CvpDg5)S{!eiTOm81Ra*i5hPnW-zATd<xa
z?&~l)Mrf8vEF-+v@jwwOyWD9Jz^|z*Flk?sXTCa>|F@_Ro4VtP%KVH%)HJQ_Nu?mJ
zUgcp$SSX)h-OTmwjs2L$bqq>WY*Hlo>6GJfRQuJd*N*`aky7&)fS%aew;L&M;C{0y
zti?eL%54{ymhhjC7MU&mr#HKV<PVLnG~2e_2t512d{%G0>Z~8+@#p-jCt}UJ!oM(>
zw|5%v|MXSY=ytdI!l2~fGz4C5N6K&Sy}EFb;+cahfcZYqVIT%*h24&uFq>elXV{>T
zr!vVK;F$s~bHlDH<MChW)J+4@M;&7{^koY_9SXm!xUYPtM#ZDw5Z2N0r;cNo?2=d}
zn$YSNvA~g@`oDuhbTW1l*QGxs$hBJzSG(k~HdOHYhdI2D1G>LWTx7UX<~cklE2|bt
zb?Ktg`6-78t&RSXc9Qfd>`gbz+Yqt3?@g~XeXK70DfcxuNi>V?>gwWRRRAQzk`hiY
zuIbc^hOh6Xt~yryU~`PPc~Oj;%IP4dn^QXK(H`ck_sX6g?a`wf9^?`C;)P}Wb95Rj
zr^Xj_0vbnECf8_ieD%Cb8L<(vhOnQa?CYLb;~s3-cTpvhvu5^joR)tt5;}uN^+?&+
zeI#j6#yH_<r9_oar*sG+kffx{%YC43`C7n||ABSi8Xxs8RY=gzqrSQy;kRPFu3)x^
zleW3a(|j)wU!h(QN}aENRqf$fR|Yje*<9xtI}xyu#jm;B%@<ug>%vUFaEIVHaby<f
zTIxg^YO?oK=m%bSQIcow#v`^C>YXR`DzOjx+SL36*bU#52Sj@U)ismzzq6*fOvEQc
z-c8i0>bh`NeTlO=Vj6ADe0j*O!qwUNa|hP)<+=Teh}I{qeH&(n5`w<M61YCb{nQxr
zBf2YElHtS)&NA<@A0B_m^IBGw9Gki0dbJ2eW+npue!toD+Z%=t-FvS{p{@?+e5ea1
zCfHDzf@%FGdM33#j(JSgwLK}k&S=~V51P1&`!9vQF#j(9`F+OP7h4z}FD;fhyzx-r
z_4u(k$(xf~HD>fO@=J`JG*8F~_Ka+Aec!OZ6_KrSB*(9FUkpc%o&VV-X;j3vJ6+^o
z4WWlW{rF3#8|@ApLcbqtU(v&yAik=zhRe8f??iB6z`xh@zDI4+OeNd}aj^~$tqUa2
zeYenmArM~oA;Epc*4@qUr-Vobxg-7DTbB1^xrTBBl(X(D3!y<Ig8haMN#~3{H7Xiw
zscX;&FE`%A^3{|4@B4~IV|8t%s^Tv07Rx1Kam)0yST^5@4;*_^?e%J2CF6HX<l?>Q
zK+{jPgqUXO?1SN}yyCDJ5XUQD?EC@d2WyV~hfi{O_9z%%D!+?yqdcRvJ>p2w&MX;G
z=p7#yxqtsYDD{dBOp)?^1;8j7C&sMPrWIn%#5MLAT@HaI0)6Gl`l035GJ?pF4%2>f
zU!i3Fq%W#X@v6%Qv*H09yu<m{p8%PARot;^6oxn;SO+wjk1@+fl#cwfsF@k6nxr&V
zQ~FvBX&g8QfFfte$#mf@;fzksnV%^U`94q2X{6Fe%zcKkRrcl;ib~@t0$R2Ga`mmX
zcM%x-gCt5cNL@i6jdrf<we9TL5!#q1I~XbV0}#1M(lCMx_V51kE8!2nk=oD{mn@O#
z9F<{3Yi47=lbH&g5+K`yuFUBL#Z;UAa*pd{NUZ-}Z;bFRhUnE3e8hHzYP5UsXm=rM
zaWSDr>NT3+h8+{Ji*XZ8bMs`^9)ISa#tcjsXF;&LSjFS`OlahY@((NNUTIL`nwJL3
z&ZFv=ar1VAv7~7Wt@9{BL$2ysNm{kvMla@Gf_u5<OJ;k&0nw4ELx-~;rk@@~&w{Ti
zFmPlj0I>0}cS4>L_MoVz(dGvrGJ(92aV`8Wd^@%CP#4IJ>oygNq?T|D+_<Rn+H)el
z<2qkLCwLB%y9V3RWYt`O$*@e==^YC>0RZ5tCOAt8CIV@=(d0{Bo(<UfGUoN1pIgxi
zB}aLsDi%rX#BsCaIoJ}*xxUPI{DoKUMD{iz*p0MRzM8Tycy{>qrSuv_KKF3j7Po;1
zM(A-?_OG`~N{xc{lw<jFL)Ntl!8aGCTTk!ucpb&zFY#vge!P{9@49%qLX|b8zd~aO
zTiF`tm*eI?EHAZ43ckg!UYS5}tgokg_P0RnW%Y3p^X#<Nf^w~LtjmQ*vkdhf0dOs~
zfOESOdIywhVd2rf8wvc+z>lYwdhnSX1vwi;Qa_32+rOnD1Og^B!NeCeiC^yq9M@L}
zJ<R$&@zI|5Oj|i3*`<O>wv8@L4}2!9QFYui6MvI&kjs-K=`^=kE~gliC6Nf=Or{%h
zN<;)KY+@~3N?XU^OlJI<%%X{3=8@2JD*8FPW&7~S*?}OkZPS<Gz6n!^I5aK}1Mz<h
zS}R~pp=Vo_89G|3opaQKpttq{9&LyWteB<x9LN`Tz?3Aknx*w*cg-6p{>vOBC<><_
zCUXpn$C(-aDh#wTXs^!2q?MlQiPa~1q3jh4iF6pLyCr2Y=Yk$<TBJ$_8ABKpfVX#h
zyR_dL25u23B0*7{^KB?^-t83=UYmq#wt28&%BIBahbn=Z&=oIunyGcU>iWblL3^_Y
zPKX#eZ!R`*)L<LWTh?-uWO6O4agt6!ulTAF<Gk0;-tW9a5Ef93?fm&#Km3e){qa6K
zYf0-Xjp(Ts@#-0M?JvIqt)tr(`)+V`Qk%zjjXcFiqY)Cv%)N+%=_^u^`Pgc-LaAKe
zAI+OZFumf@xQ-L$r>2Je6H7w_TO`?4iCDzkrN&-V<NXZLb-}<sm;axT4SZmP{*ox<
zuy}c40f1ITt9Wl{SD(}#c`sN){W2lmBQut-QGI%z@e<+rO1<<N8-=Ah=RG@E&Y?g7
z+gCi=ynPq^IM4_V>4Lm|eJ0D}s5%f$?B{*yCzg^pbnMfgJzk)*TUu)^zKVg+lLo#u
z?KIU$e5e!Gbw8rh`{eL1ocSI7v3&~7HIT3yZ$;G>3)Jrwy4Z8K-DB)OOK7JwUu-g5
zW$wgv{WW*N5KZd09ImkS>M`w6y=On2XC3P-_z$|g!JoZ4Tt<LLi7fp4$3f73RXi>a
zpzdHDAaoL<I8Re;FQ`SEGlb((tmnP^F8KoQ);QJKFvU6m3e1P5v#!3{oG;VQH!3VN
zERp2hcpQC0;2M0tknV}y6eDNc-iCRfw%2THNvP)aXgr6%`I9RySKEAljx-snU?9#<
zh$TkV-tk={U)?Wn;$flaXjI2AwoJf5a#Q#o4Jp2@c$!A^f~lN_T#Nnb+fj=3Pv2j>
zPuuv>GiGMw%b)ix>GMd;Kg!4tRUf@%t}k(V6{p2#Xi&`Tb4q-5zx9~(#X_y_mrnVh
zub)hF(62<3FM2A6@nt-#CJbc!r&yq6G8r?S3Tap)?s~+DH2iu*EJ#x!bj9vx6&TIg
zgG`yp*vZ!cR5I`Wn*PPgA2p@-&uhkv3imqV?k+kX!Z5<F!T<`4jUw@0Gao(8e>An;
zS+Ur7aV<7<1a9h}h7s($w5oepQ+gIX!bEXOhGFZgst0TX#74}o0bnI4z1om7xLLit
z!m<5OCVEzr6>tw=f$!-wgvm^1etzeoGyi^ZH`WHqzYCE7bnS=c+Zl}9ygeq4%)-4L
zJJr0o;(I#y7BKsG`qU<|9Hf8ivNX9P(1VN+#O;3NI>THhd6y7Yfd&gH#$+<tQksYC
zyYN0&^4tk_vk}I{MB%+;{Qeh-yK3o4)QUdr8c5e_Ox-aCKi1P|XMvc5V+`I)ReDZI
ziwY9p41I%|#t5EZ*pYaaJo_gobD_wY)q@&oI3bFuJi$;$Hw{?Ng~k7V%Whzst(wBf
z1rKOJ!~Oj(_vT#oE=%8m+9XUEs;W{Q0uce$_F2b%*V5ePmXYggB?iX$Z&1=eTII|)
z2AQre{M`qvPvsJ3uFuVp?*GcTTkmkX=|{-c#)GWnQ*tw}BVi10&rEW+1q45<dVS=|
zMsIDn`|c*cSnX3ILym22rEuU;hh)1#wHX4oUt4~i)qjhNRc;%mixd%z$!^nu`cG#n
z`Nj!7I&W?ShvxXcjCn+D5I~Qc)q6f-fD;7kWG@SUlmdp2_=_I};+q5b&EK$p(y{IG
zGr3eTnO!zTW-Ta@DiI=f`rxq8adcDyj}<{AevLfUoDwQ1%4q$^L3rEj_sgQWL50lr
zdhL&oF_7%9+TqSA!n+al_fvI<95iKy(w+ZRSDoZL(Z2<a#r1)Kx1NdK&i)O+FkRBL
zd~o-X&CPyjK|$!-Hz7_rq@rSe`=J1hOpX}Lq&GDezJcwXbxdB{QY032k(s~3eHf}t
zmr?HPvW(w><J;JW4eG&Mo&Wp+^l@zd#hxPKkRW*jtd;YP^~%^E4(#jFUYSh?aYR{+
zl5gjE(iSwC!yF6RAmjrU;qD14H8_x8hkGbf#C$K6c(gboHyHe}+KdXB!{tG32&74b
zoQL5;0cyh?G;*qKvA0OKgAv!<2}T~|OnI;AahV^a=0Z`l3{3)bz`@60#$&noDZHhb
zT}Y|(1xB93{?ZNK=Xosgu>pRH^k%P26A_IZP3wIj_V6Bjy-BCABJ7ka*{?0;JY^7)
zQbZC#7O5t$T7UeA1pW$c`<(r2{fAtm`srhBsYFySOC|IByFG<*k#r@2xDQblA?6u#
z9E1fQ13&-ty}`D4_oIlqDmLPyE;io?XV<Iqx|0p#R-+riTGz)|BZzPjpIm|%#k~Yb
zXTWrgO!{x;l(Qg=M0zY4^Y}hVxhR2$cYsf!a1a<DrTfMBTaBOZDxH`NVHMdZ(@2D&
zlV10JC4;F2^ZeGZ$3W8pn|yBPh<rrf@(w5{oH2?ypGoVCZZcv#Y@V(yxi9a{kR<T`
z_$JFVJe5C2;3AJBhClVM@KF=Rhg|H;R1fz2uez62W8hlHWsZ2vzUnw_)+oTf@cn@r
z3S8(`)s4>A_#3K+@>%k6R}WlS?B8u}`OJLp<WI@Hn)D65Y}+{hQ(~m56A}C=`*<3D
zO+pIC0CiLR%ls*LqlDj_=75)C%C|LdFIp-X>7=xA82ib>ym&kyuFd9BUBK_LrU+KH
zg^1Ujc*q;2-&Q3&bDA<^ZyMM|V)<-J9{NsdXU7)5pA}9-@iMa_s&qMxXKkc&8fxhf
zvFO=jO$wwGXD7{jC4=1yyP~Rn^8ZXeOIY>ucT)psscC5e-_UACYW^jXF;M&An4POt
z{-tVNpn$=BJmc4Wb3F>IS9UpxvKsKbMLo{An7=2;eOmLtA=LUU4gy)w`7NlIB#jw;
zsG@Qa6XR121h3$XW3$H*G#$<$s~g>K$qNJUtX0B)@J4`@5u*X9C@{5iIBMflFZlEa
zi<34q#;V=)ihyq;J?^(hImcZ=zqNKV(@y*LzoWnNN7*Bg5odkL8yCKA(<aQSxQeK{
zP<lLLTidzzj<268nmnVFd6-HD1q?<U?*EKgX*x;txPW+u&OH76W>Y*I-{!fFy}P6r
zZ(e*?Uaj*D|M6<T`hhkP%uw@{EQ%O4uS{jIo(#Nj$IBDLA_>y<ta}z;g2lUa4G~lS
z8)l--d3&<tSD<atl?177J%*@3_4^!=MPy2iFAWX?;^c=rnl8F&pp_;>8=2UFt?h5V
zUrYOXBYt^XFKuLdQcT=Y|3c}oOfK=K+2=Ruf=URmIk~Qo76#vT?)mt;0?d5s!DT<{
zwX>b+s88c3yi>J5SEYAa2tA*K(lD64!&n{yHW8mJd4p_^6_c$p800yu`G|DK4@?H=
zK49XYmkQ3VFjCl{X9~Y3y@>zhomHE1$C6)!mB+#X+MR|<o;p>H<h!cbjK;+Y>947i
zmBIfLYC}Mz12hB#qo>`(NeYZ=@ezvPjO=lp(r`x>&?3jew4!Bs2zvpfJ1{W!T%bGX
zLVK}!)i#^L!-!v5cUvNJ?oGGC2&l>MbPK*!5~$M+SLBa*cWr&v{Po+BQK>44g5-fx
z`w<nN3CVu3-c7doIQ?ot*j@h6&T@(={7;V=56fF51!G@4RSbC}Kzm;S0FyAJL}2d%
zOo;iTy>C;`ZSLw_e2cAHiM4V0=f3z~1i{7Eq~Ip@@e<0`G2;3|XQ|eZaM8!EtzVH7
zjhS#gPR15d-1scb9+tn8VUJ<n5oMo1Q7SJVb!&_I!^uMj>v+EY`$6}lN|G)kDNY_C
zs91Pu%v{A)!p8`L+2<-))N36vP};lnLgs>Slrni5!G?+34Mugp%H?n67cQ<6I!8Y1
zCh8_aAk+;0Gv7vF4>FT+8dg4^9GFBi>$J(Fb2M|Ik0Z~^T=%{L*H}gvBQ#r#kJtUY
z3vWT7)HgGpe`5w#heaQG1?W1e-)M`U$(l@I?oVLV`Dgj&&{ZcHhOpg*lPdQ4vGwM2
z8rMxT$$ZzTXc5RkRXfc1x6krW+k2EijGQAGs+Y=uels1XlNRo{*a7;;D1`1neTio_
z;*B(}npEe&(yROUc;YS7_*<sXTN(%2Gtv*N`^8HZQ-*M}R4^L2hCbK6xHjy^Gon_F
zrKM&CPB?3)v3MprB#bfegP1VwnYh5F&=rkSp_wPNYGPc=drbrrZl2qP_uQV8lq@s=
zV^wqDw&v}T38pu{&~GbUuj)&bQ`sf#FdPZEQNopx)McxbYsI%MFMC~$y+|)IJgSC;
zmw3_s^ih0^kw)?aC@biIR_6o6<Tr)7s)BQ>310TR=Y1OzB-YY1ADdSJuM}tVI%Fx3
zD)Cy`vPbR#3WE)m#z`B#62v1SPFu{3u~zY%t&Ib^;xm5eFKEe`ngS69iiL}7vuQr}
zSgIPS&rS~AuGiMj=Z9M^E1B$|ng3%5YW98){zJJCuc4sPMfT|P5^pZE$WS!bqH^cO
z5~CY(&sVTUUdR~*>{I@w@Ok>`llWsSFD6}_B4^lLM9z1Y;BL3&N2N9<CiG<ZP~r)A
znOaYDj}!|aqhakQlanS^t<D;c8Vwe@gEMuirL%fdyMtOaxxSW@k})o7vU&-G3x1GG
zdYEBw<sVkV?ff~1$PJ&?$_cYCV;()MiV~d`TOA$m=y8FE_29v*S2dtbJ?_M83hL$$
zfK^J4ar2^Wqe_&qA{tkpKYwO^!$UaETMwN>&PTg06mpED99?E1H}SX+?7uwES*53<
z;FPe&#}O$mb`2SKm9!|Z_@2Xx&+zWdglhB88V^tE+p_uVy&t{Cm7JxV?;w`00wH7f
zu-%hhzoA1?mLGL{<xOoZ_8xR#TDrtvx|sViR%5VDv;!>s;Kch6$~tCVV_S|U1lu)#
zPtL@Ff-1{78*~1I^d4qMn?iQ7__#O#KPz^T7QS0=FE)&}#a0&4D5dFeFoZAb$~qJ7
zw*>pTR)>#+%g?Gp4A1u&dpYWk3nsLc=M!h{{stJv<;W7`yIP%(1(pF>3`G*4*nTxy
z1EUrKPv!UTJqxtvbwzh-MXZ7~FLk(Gp>1kK+6S58P$rG44C!GyzJ-a${4UH_?=@WT
zf$7ozYk0Jn=y5X~5G@w~yjlC9;1wvr9}|<A%nnNYR_JnhF*bK+x5r*Su6Qm><Qz8R
z{bPOtFRar71ofOhVT@L0W>_pCN6bWAO;CU>X;zt9<R<_#H$pGc51lU8J|A>5)Bc(l
z>%c_W_zTwF?CV}p|4L%~zV$OBmR@ut(PvC7#1HRKMD;U&fY~qi!%^Ns1=d~VhD)lf
zQ3=HH!Q!K4qz%M~7>oUg^8cJ?KSVoZ((lBy!&(XMqYg!h{S)tP*`icQ4b(RwQ66xb
z09^0d0?A>h%tmL?{K}&{NXiQJ2;llv$!-NrLvO6Ty}cD9R61zMOeT0_{(s$Xntcm+
zke^D|{s8t8^HBBqT=nw2oCBCp6VBPaf}LI?bU=R%3UWeQKxDZqA_CPc&|=CO%5@#>
zwbfat5Zs2E2qEeaaAD#Yq;n8_WlxsR_}_3LGRIllR;{##EJ&rSLNt>{<U5rz`vrGZ
zX0hHcl{axspFdv*=b>;{vY-n-`E>+Z%>dpjzP1R)OZ8!f6QP6$@@z?e3GC#L^9<)m
z@+q?{YuVRG_7iV%AsrE#6{xf44<@;#M6jqN3cD8<7OE5uZV3a{fQgAoFFM1Dx-63b
z3!F%`d{-9vq$bVMm##NHH97X=-Zrd_mcHqlHDOQ?G&|BMyvtYj@b%^&D&;B>uFfY}
zd}2BuSi9)_3`r?DbOY?{OQ}9X$MZ~OYpXm08dC6_37BgDoPxqf14I`}LWKzDl|Oev
zTY$6xozvj@#_`Qo8Ur6b;k|45_xLJ@;<yP*z%N)a;VqQ-kWT2uXHad-cmmVjjy}L>
zu;h&{DCNf%C&BT(K?^v$9N4MBTnS(rrP~0uf;KqT5p~`jB@7cw%Y{EhPvCipcn!}w
z#-W`SW%^LEFtH}vstvajXJHRIZ1LFS<iX_~!X~oz!Ms)L7b45C@aLr@@2Dpv<N{*>
zqGI?+lGpA~f1CjWnX(gL#{0^_b4x4oQa6I}H9cHJDhvrurk?M^ts7ZCr*-k!OTeJ%
zm5X=i+eY^AXoEGo;ZhbFA}C;V*G$uk7B-GL{BPxh>K@j+96Fe4(Ol+#Kuvzq>fF<u
zkL}TyTl)!rlN~vx?Z0(>Rjx}y%Kv9R1{U5Sr{%<@XysrXe1Hl<<o^IjHxZ1T9UUpG
zI9P2F%y|NqK_3}Uetp-}&AFrS+b(>tb+hIGb5tSWs?UCq`dyR}F9Y+Grul0dbMh2H
zoYEY0UuQhs@~Guc_y^xB<oW5bdqQV9>8;8=A_*{Y@p+L65o?HE$j{3ZF>Z2SK@hRu
zVtL}2@^kV<&s$h8i2Gz-vK?tGC%62^bI<5D?s>nLg#}J)y9_5alo-T$@$Nr<AwYAD
z-?q~GM6gW<Wwog^;iYz69n~}BvwZPf<E}W|JzgIDtN4!?DYbaRqt+V;+=@lm8y>Zb
z&iW*8+=SngX@&!k;ctb`r<|ww$n8igkZA$V1o;n8Z_V!?(1tz*sE7GeF%(W2fMxlg
zi+Ke;$-N6;kj}W}Xyf;vh*gfNGBf^{WWPY%w-rnW#?;_ph&APXJRfxGUx4FaCM`$B
zc!rgSr6ulI12_-$ZzTSxc|Bos-2+=sB<9!o6p$$bq<^J6j?SaLr3<&4;9lv|x}eSq
z&OuIpOoT*)xXo{~u^1~yUQYRS-4*_?bRUArzj)Rf!2krMUQ5l@j8cGhG=AGZ-6r*W
z`p1&o!UmJbWU%!Z!%Z}dYdbLQVYCL;9i^VLTvg?WKGoE(y4wi~op055vRI{~J5$~n
z|BvBemsx8BecSBT>Xjd=#*VO5d}A+Gx!06;$C1Z%1*yli7#*W`0<WJWD}LK2p0)iF
zBp73kn)o)<;q<V5nL0_)U%ed)7wr7!5QxGtkR!TuDV0m-c5iopyNOA|tYa$g4$KSW
zBmNKn;^+F;51Bakgsvb90L9W}hX)I8gzIS#XdfP~uC9(|Rwc)8hIipfkuD=?t3^q+
z>Br1PI5~ssk_TgA<t{dW+R<|Ri8|O<o(mY-W~2`O&&D?cd=l`uL3xGhd4Dtnc@_#R
z=1W$?8LgmgEL@hhYUGMlJUA=m7;gqm1F7Y(LFYPjfU7pZtQbt}`KvT{D3yXgFgZWw
zRy@xiDv0K5*GPqwT7!-Q@a_u@(HsVC#l4}|h0uYNnm>>|((~}dLLfB&KZuxGAC$DQ
zY3b=5nZjvxpiVbK{kpK*mu(XH2nC^SiVC)<w%Mg_fYbsu=I{yE&rLtxV5y@Xvt09@
z@jUGt&8L;E3w0%#5AI{kOk6B_<=^s-;!qT`y2dQvczb=MutmQ#(ev5zToyZ<DSDH+
z-4RMEK;>>z(A5c%+5AIrC(K|?)2p!Ck=7GtP~4}rGPX;v2E@(+#Mktux9_y8KqUG^
z0SvWmA#L+pG%ZlIm!Y|Aq8;KUT%gKadM4`WN=qXjf8GDYd~DAvg%k11b8uB5?j9?F
z`N?$03ekN<wGY083TA)Q-=HiT#0i%P^~-Go&b~+%Q$@dIk#O9Pcoz6Up4C!qV8uGR
z=p)syZRb70Lz#O8)`nWy5?ZO8npy)<dT{_3Zp~>G4AWw!3GuEMz1^MN(N;)`r&RZH
zVlN30ZK+J)pdB?d^bttbSP1OwRNm}>$P{o0Cv3b0nzxR@y)K7}OfL23wZugYtbEAo
zt`EAGjJrd4cOBQ|x6|_2OA!^p4k#v=s|{W{+#Mi4-WJ*(aQDvFyzMoNMt7Qfa)_^}
z<=Dk5ujzJ*e$fly<5-fuAa)<`8p_N91bu)1GL*=v9($=IJI8tcOM=Jj#O~tM6dv|E
zQ^1qd@vpJR62H(VIzf^SYDtFs_H^?_F&gUWu84`^m(wHeBZHH!rt~`KiT%TO(hn2P
zr5z6lv_5*|QZttJbbEdR@D~sdPgVUnzt!3AgqYadhf1vhvbDZPNMb?iF%qiER9iL<
zUKBWb33nN=Q`AqKL0<`2RKTM7W4CIooJ*t=C8~7+nVC5v+P+1vSLBmgB=6C)VE1h0
zZqCX3fA%G_bp-AEzdS$;UdN!Ach9`-)0ZE;H(#WEO1F+^Lh9Z}unYCp-Dh;I)XAUw
zQ*Dqz)0d5oTO{@;Pahxf@BFDJPmA&4kbGpQxuGRlCLeXre?d^yQv5|On;L@n1azZx
z@*(4R<%C>klon^ZEPL-{7k1KAna@E)5Gw`9S#h!BY+F<bt(q%yP&}AWJoh+}&xM81
zZTutS1$%ozetvi-NVy|*iZO0Uh5gnXBZ4>I9AaN^dn34*zhlQtCMYcYiP_rQd)$l%
zG7d_&x2Bs3vU*vLjq^u#o3qgFRM$Mi13mt|De=Qdi5nv}-JE8_^)~#S&R2qO(@9iW
z^>|)nbYxkPu)wzMm9-r;_A?qJ3R9bpJro@tIV)@mmkPODdrW7)+Ogz<atUlJzJFq_
zbKgIAt`4t0PCf6~o$60Hai$TKBOh~VO1E!ulkf`;=3g6c+I*Q*gIsG-ojdqB=FdBl
z8+=G4j6v-K5Afg1FX>4jP4#;(e{dN}HII&tKK%%RPG$XknCca&j{06ChN?)V+n#H+
zd(=;-dU%2E0vq^Cpu=BVyF%%I9-F{N;5rT}L}H6c$&j4NM|hNVb>o!T^)C;K_46?s
ze*S!$tzevisskraSU|w?^Y!!To8k0oi|FZ{uC2j>M^06S|IJ-yw+^7aAUeeJP48y&
zRb!~03`#37teW0B5Ja)wsNN-<yu9=&V4&vP-vjV+scTu<*{uRO*_?6?Jn3Ksy_Jw?
z$zZ-H*L_;DG9_DO7{7Suluk9LZcw9poEhDawm%Y|Zto?Ue57#8<=sOw*UbYo@nJ>Y
z)b2;llc@PTwBPFNY2=JY)z^(%cw2vRsGrTS7V9|B3Kg2PjmH!hi#v6D<g-l0UfA2Q
zmvB}Xe>|ci*%fs_YHW&xb_arQiQ&;ebWDu>SoN)<kK!W9#`^j<F}R@s^glQSYG<oF
z4_(yDb%2@LcTRD@_J-}egASKV{<FtjK0fl$7C6E~5`1!!zKVk-N&)EZ2$a}JC0bqy
z{0?TYYo0^Dd6j#e1!(G=7k~fxquZVy7Uv^N<UdoHk#TGDGE64!c*thoqikSsep8^o
zwkdq}xAjWW{EkPhYInaS%KJVf1P~Ou>~mWOqjvW`K4+LEa`U<B9Jp<D7Cid6OT0!W
z|EAx9oq^I!3n!&UrvyvSOx?kC6E<~4za2mQ+go}4;kl*FhswVRFRUwXKin{}h`KHF
z*p$oy{t=ZK|LtqWev7dgs%9U+(?2V`>tx8byzIUb#mgI`!a>A^>ih%y9fj>?P~DiP
zF6S$?uVGu{+H7KXJhm-3$Qc^CQ*1qRNhG`gEejL7!NI}c=4k+*Nl)1+LLUPFraz@k
zL2q-@6x@BF3^`ox&11^Vx5?kvA7(|aL_EwI!Cr^$;f@b56R++4Q_9f<y)jZK5&{a5
zL5C8{l`vtYa&L;9mbxh}$fZs;yLSXVIsbvpVL6+rEGy&7*B+s9nRl#GW&Wg{=i8X_
zFv~_~1KbsW<cCHABsL+zM&~NY4QEF99tuk6{9kna1yoe+`aX^iptOL9f=IVWi%JZj
zNQ;z+v_S|X0@A%vN<q3sK#&vx=`N9yE|Es*ZV)7X_ux67^*#T!{<GFO>pka%VP@}V
zKhJ$%_Z8WWEeC?AQlPC(+U^5Vrp+VQyS1^AKWYgXIGjTYWfxlt$3=PEB3eVQ{gkjq
zG9$)s#KH4ECLEVTd&XVg<IXDHO@8+VlxOa3QRgAqGNjJ(@Zl5rHrF3Nra818=Ui@k
z+tbF?#wPUiJoY+A%;d5-q{f>6b^P;Xyk3%dw8QA(F?GnHgE;SxnCzh^sGI%gg^Hi2
zaw*|Q1@zBmP_4T25p>?Vj58BjO-9I!^cjQdhBhz^H8#6Bzb%ix4m~Qi>$3;e6Z6i_
zo?G3cwO0Nb$X}Xr2U7k?hOqEm=r9`P@z^{evm3Bd!5F_>JSS+}tA3ONe_MuS-cBsM
z1>%drxk$GO#4PZ$=+5x_>q@CMz;|)_0RFsClR`@JC!~L2f4{tG^6cH)FlN`vvVZX+
z@3y#J!YreNi_FFu=oZ`0Z&OJ3z;V*m%~m{io}-hTjLbjbW~*^NM4Y+quH5I2gkW;e
z(24w#`;Ph|9WB{H?-AB|-0<shzIj=ZjAi6Tal-+4{uMW%Q^hkQzbT)cpgCQJ**iKj
z&E<yB|D-(d{dy>y!}jghg^kl$0s|2oJ6ZbWVd70K)<nksX)a6qKX`k~<C|7ujvY{{
zI1;5w9tY(=X&Q>FP74RW{7v=jARaeu*AnDGZI=!A2&;;lkLzF(duQlt&GrvH`s;j=
zN>fCC*Vevd*ac~2#V3MHcC)j^%Sl5FDi@?jI39eBx5X5FA7tHsmxjYcLz<JDOHF**
z_pyRSTRx8>rGjz|NwbDz(4>}@mV&4|Xy!i|PzTDjRDJ~~ylxB<&>$ozm{(Z|Mo_@$
z=Q@~i_5g!y+pQxwz?{%zC}2+0Yc;car_`btnAe=p@($K;&A@91Ui3&Yi1Aiyjy)<K
zEIYVvs)LP8g^2}R_CTL?@7tB@jhE;fzi4rXOD-o)k;OerQJhk1J`U@58Gcg!Tu?Pp
zNquXEs6Sn`^3(gz*R6ChDl>UkYS*OdM|8-`IF}dlp4BX1_y#<RUM_vp<rsX*G;>;P
zr(Y_on9Hac<@q_S`x`Pm{!^X8X*R{zPAkZHI?-$VlD8oVh}tL!L(s(;>2DQ;V%Z9M
z^?Wn6vigmM;4ElNT{4z{&|G$Vhvqv_i?8ki!vk9O7USmM5Wsiae^Nmo#ShQ}$n2p$
z)yNE?Er)+h&O9|$3XDk;7c>PS9N^&>agxt8`ym)dGYtWtRp59wO<ts@U&c%kCYi!!
zx%GXR`@ss`j8rGJWSen)Bm=eSrRj}Dj!AK?O~-2sie)2@rTQ6qxnGzMS1%jpz7bC*
zIzGRx;@a=UEEG)N_gTOF@^-|J6m-q@&&xD>>`v=7culD0(w+&iy<$nU?2G3Gdt-4Q
zsF545oTlSPB7-Rwasdv2ZAkY!<jy-h|E0L055epOvogS(luYD&V~2txSt$1HWT~Cl
zC{69K1y=aeI}gDP6t&@dn-EVlDz-@my2Ipdu2@+ZV#Kv}6P+WS-nfAHf%vlS5!9aT
zf%e1h8qnfW8_%FobaZqOKzlxLhAQ9(ligYDBX-3ciV-fyKZ+}BlOvBm4_sSzNc|l(
z=r;W1=#JLrXG3|@FTKJB%SC9tPCH&Ll!6=%y_a-K8Jl=tYb||N!PsC{OKG`p$))K)
z#SMqepUw%Y>9*K-EZJ=0j!RY4jYsr^hk{*rdiM8;P*m&%+{?)GlB(yl_~Q}{>+u&x
zQ8znz&v12>{Y5i)tBFIbou*e-KL#vUp}DmzyZFW`tABxj(XH#(s8hmI9jzp_5Hob7
z5>fxWH;MrmwXT_(lvPvKB5W7xt=OVl!H+FEIvVE@7{Z7uaRxJp$xMpH=a`_zov(Qa
z#inq5pIrZjJQ0jCdGO$YxGI|rAw3b4ZH9`+dpj$P9QP!a%tBhNV%|b=y{gS?MMIh|
zsg*2_A9$vakBifFDAq4NamyLa8Hm@M6;t_2LNni)-8RGGk96J4{kEjA_A5VcpT8IS
z0vUY#cc9?jQ;Dggg7(@utG;S&XT_t;@z|uHh+AuJ2sJ)p{g#$!Hs<>&m(CkmMCDIN
zJ$tOTObL@b&VH(sWeL5>)FpI=?~q6&)|?y#xY<ds>{8XSiB{|&!Es}w6h)(r!QPCk
z=SeY~cDO#jvxbkNM0Z$2rOUR<OI7DXRnI{u4*TtuNF|PN$m)U$Jo(xVVN%W`R4~$Y
zWxlw9N5Yf()z0d>y90OSO#VLE#`XtpT`Y=xZtm}V=2TP^)jQ$qu!rA+@93IgT<Eee
zi?4jV#`NwPVe6(p9|!)oXA&J>M2>f^ELOfhjk53f$WvH){yN#$335Ko@lw;3XJ59{
zC6-Hnt|l>g2eD{+0keA$BwFuoARuf8lk6vXBbCnge!G_I=k)jA&;J3jGYPXWCGR#n
zg`%9>Kqt1Fw%qAD#C3WD8Y^I}seB>v7TqzZT88WesEkl;9UUG!gHKc9M&Vt>w%TtI
z+WX?5rE^n;qgXi<68%0@sqdkpj#~!Fk{sKiE6~+Y!4xdsOgfFowY%%Lc<8n@eJXOR
zTG8U>pD#SlX&1BV!mH)?h)RpgzaTe3tvwT}@BDK;-DvLeeVZeg-h!VxN+@<||8IEP
zc`omlmv6@7M4l}q`(lV2IQd!LdE>oR?a^pXdg7eKgqg%kYG%7vkS+N0=T9(6&eH~O
zRf{qlH~PimZwO*gq4D}rpxj5fk6?2*Z}-l)z-Po`Qpjn89=PDn?arOe!#%>hef895
zDztgN4lt^T<-0I~@mVi8;_mD?B&q!z8v2THfUfam?MB<_ChD>;%;iM4j*F?@a{ajZ
z^Dtna<G@lq>~Jy4T*-X<XA#q1_JYIck?i!-F87g7YGnu6%g#qNm#KYA_kESuE6P=h
zn%%hVZzL{6=H=&0TN*=MbyrJ5yH3(|N`;z&-D~}!5B2e-KKAGYHM2)c5BO)^;d*yS
zX`Bg4n^)uNL!W=&$C%lVfvzF>x(%u}e+lHYVDk^AMMfVlruQmx%!W6Yy|Ae^yLdp9
zii~5;WPR%&N14EIS6)s|CRdK-Y`$3Sle{gr*Ps}Oo(a6Dh<(7`POISdNgccOX}`MR
zfS;0p&Fna5(5cs=+~j&zM@#Y!N4n0z`#vMiiyMvAX6@?xnR^KheKfWA(8m@1JroZJ
zP}Az?+bng4if5heQ#CGYL~Cavn%AjsdyLMF2&_N*vhEl5#HkFQ<A~Pmp;s9ZVLTuW
z;Ip{`Xb4>sJssU^L+E|Mv?IC|W>`>5hJD>QWe$nJ{U{&_3a`lG{sDmKdu{zd>qN_f
zqeu&CY<1Vl(h`fza2e-m(>8(k!jFLgEPbokLWna|Iw=Nu_s(Twil{XoZ1SO3S*|ZV
z>v9OsWmne1>M6J5mz@P>!dxBq(yHjoh>iWL`@J8S)uD+oiv#K|_at^UXL3a>j_stF
z4>oVaEZ7(0qptby%P**epPQoEX06!%*5p|3l<?5Nfc+r^XcaCtZs`gJ*R@%FZDd@J
z^irwmC`WK;j_-HvXx9-l_c4<B)vxZZE-x3k**iGwfawxxz1w;NHH3-&CnJ<KR3S+X
z<@rc}3-f-(POOCzq?M36Njv;HgiR{b3EZ1q%$PXO%Bn-n0coYA^&>UN`08%B5sbU6
zMU?S-UQ-qRmrEcZDmr)@d<W$KI-w(bDijMA>LMaP|9Ijtvz^6XBdq$ALsNc*&30x=
zcY^rt`15PQ206Yq#QtY9X1-mz<R)NEPZ#|#Z?C@QQ9;MNeXAW2s`}njk5nyu6vG6f
zF>m-?{E+_W5?;<r66*1=<nP#UilFeevPX87F?-oAQ--O3q#q6)BRku7UykA74b;6j
zonbVJX7%1T0$yXNf&gjnL0H3+GJxEn%Yvv!D3awq2Ihh-BeW0!4(i;IuoB|h!$vQ8
z89%;=M*Upg86$4K18!iTxq?znB#6YXK1G=bNSn}uRgD$5OQt^iY`S~6iZEskGWYR9
zzaX8CeA{(L<y1H?_0vWCR=a{?$BV?VGgPzJ<@NArxx7$wiKKdWtAXOdalB3FNcO><
z)R}uG=@TQL)DEKSz0!qhd+YOgV*QCwF!XCup3xVGAXUB^S?k!Ruc+gXcMrmvXm73J
z^q<yOc-eJDIaLEfU}5Qgp3VY6IcQ|#;t(K>YH*EN3e#ery)ws_z4L51Zni=3JYIg`
z@1^)5NfiVuuCC(X<kV_{6-5I!X<_l%w#ISZ8{^RM<Sh2P;3;wWzEOJ!R5*=v&D??A
zMJoxhyTW%LfZ=ys%C6Yos&@LOq3@_poZ8!EziWDE%Vq+dP{f`DK5voz^^q-d3xf!f
z-_8kCVjA3$fma3oUZEo^sdx%iDhz}mEW2vVA-ZdS--U|`7gYfkc6wIy&ZQ9HghByd
z)zB{v9c5x$_I}U;C^k<eqOb{#E8yv{a^F80ZW`bg+B^^ugDEQgHVHOA5Ci+qMTIC>
zb7>7(!*(|yrUH|Sf&zEG7WmijLxd>!voG?RF1(suE1CZoED1px$^Mp4;f8<G{A4+t
zlC+7klI(tV8lXC6J6flNV-Fjtc3adJ&hTTr>?GMOvWL89x~@zeTi8@)%^jj2(fB@&
zkxu$jP%ydeg3VriX$GpEp&=IJAs9ZyH_&uv1<vk)Q5H6yc?;Ton8%q522w~9R&fY;
z$`iW#%<f^0HH~QG_@a;3{G^u9C1o0!2TM<-@K6x^y7+^aSzP$W4M*v$cYkeO-=ldd
z4!--xpU6X`*O9o-cG=Hw)x??hO2yK|_a@eNxQ<g2o}%46KzAZ=kHBBE#n>q!xZ#QL
zUEmde6!d+lr<XObh?>~_8{8W}7;kGqWOK>7l&GKX%Xt9Lp5tuNXmi|TF0+EHTnDj+
z?R|g-%k)*)C}E&vG6-q40S!HtDnin5Y1UycqX0ppIt;n!$Y7{sEF>x*(DfKP6P4<@
zJBmmgi4|=8j*iqQE`MDx-sh+5kWyecs3DjSw)L~CQYn7H9l1LAxn{=Z;Oi9)hp~CP
z;)8jZXK|_~&@o_NeEj?*fv+IC?eYxNeEdL<{$wPZ?vs6U>E7#cf}G6afjXm*%-X*O
zYa^%1yuZH%Z*3>1(&;<!!~`=+A!Fm?v(6qSCez@#)I@&+KJ%78wo%$Tlh`T$sRNQy
zNhkh-RY|4O`YShyYu98yQ5?4842;uR<J|OpY|FE#b$OdpNAHjD#lXfAFh&I%eMl<`
z419&{*7_rXnxg;Fi8+$;%MpQEd-14eR^3=rh@tPD8nSmrcEMpti}B{>CP?rnMS8kX
z%-$3(cTcXEFW?1P{r#A0p$b4mb6CdE_%avNL2!;rIv&bLdVQs0HH=@y%)(vt+vT_S
zQuO&C05HU^3j~3%)RWIxDEZuV?mnWrLgf#${|d)d+}e(gp!#1Bf0L<6jOxQ|?_K_2
zk8w&s0J3a9Kfp)n!l-pbp79`fEdTwh+2bp12fSSjtiGUYhlyfhCs>?hYw_=tNpw0T
z%!;hYxW>Up#n^SP(4LiUf-TQSXA8!T)7RmPq-b2~T}`*f8;!pGekRZZZ2Z?IZWZaC
zhU9AWjT^gzPZNoV>cPI?5~**;>KbyAU`@ZV4I}2x?Q#`ly{6stYFs<i94YtnyTn#N
zkpNwrVeTqC6sxPyeB=&5+FC#648E6~5w$xmxGaO$7#y)r@(gpse@}s==Fy{@W;_QI
zbzV+rIGn}+rEJV66TXWatT#<9I*z-Stbck$Gm|cLco{{CkdpH7@Yval{Qm!gJHlCK
z(YrxfS-$;~hW7f=O^q#;FD5_#;KdyBhA>VJsB`6Z<pkStM-GjR!T61|e$Gw|zNnyJ
zcYdQ2JmUYr&84_&0B)|lf{-}i>BDy;t%n9Jy?5{rQvdu%G)Xn&`YHDEze{W3>|fl5
zm}$$BJIJ~D=o^`v;9x&tC-y)~3);HYeCYoQ@o5+tt(<-J67s6u++e|so-CLP(XH%l
z$yU_0j8Bga6w_Hc?~Ba=@Xn?W%Lt-zlhxq@v5FfIrSIQ|`$bMp?ws)O;X{4|Sl=*z
z#KR_?W0A>F%Ksvh42+D?`RIh%J$TYW&Tr#Vy0V2UR*ZT;emA8(=n0(A9WV6ZVR{Qk
zxANpbV08IfSeK2?{x6xKqLD>wd^gGQmLe2d)Rpm8QETkdG}SjrpW2El#)vd?6o7Ty
z_XF6sFI_?)srk}Zsg_Q-g$6Yq@t@}@K$DxUng|~AjgD#NVsvB=%k9kJ#Cc(qFgpOb
z;x}&8>@iK-h|rA<4?o*obUPra_5yDuFqRc+yTMM-90x`bFdBnp&pMt|(i%5e3d^53
zm{l~0wL<2jdiTW@>qdZ&d+O!o2$Kd7Iw&=gtORKT0icHJEnG3mBrg3OY*b!ec&IPB
z*<Y1RgC@aIGV$iUR}Yl9g`!)5`vCp4ZgyHf1~S9t<mKU=@2~g>+gIGSizLJBa(Hji
zMNpk+HE=wYIrQ}P`^>)em+bEDiW7jtda*{Po$JWv?Tj-FZ)BPgYS@RMPO5MW{&oov
zb27i%nVPVxK@ki9P{1G7+wZcoc`2>t?8@uvZf&rpnY&n7<!bXlMmsn|cVw$Iq5-TX
znSjT@2z7kFz|+?~l`%)1FB)JxOpzE3>4KdYxPw<egv_M(FU^F(TGud}!L|Rkg5;7(
z)1fOOutu-8(`PS!As}>|tY&=t0EjTD=0lGJPI=ub7;F<z7NO&iT}^=Gpq*hWlA3Oi
zS7XDQxcoQ%9%^MpCTCb&2h-K27b8PMV5_mc=k`^cpPxTm{=HHvM6>B;Yvs>yg|`Sx
z@>AFyy|`03F$_QsIRlz0`!=|Lwwof5ng~JEV2rJixwNt}w7kR0Sg_;ZoSU5;sl*<(
zP@>IKqzef~Ay*Fo&T<@cmKUj~@Aw(csp|5l*PlW<R}@=DJpsp1$~Ul4LV&N+tP(8L
z9-_}#TdozFQ$RypSm2HgAx8<{g<5=NdAYr_)A5?={QQR~7nMAu7~G_ND)s<BRBsiI
zRc?Hte{$#OO}oRl#0Rzalk~(68;(&o7g$lKvPWR43l9G4zahMqO4gvt<vt<SXjp&-
zVhQ+RU=PSN{Lk5j(<v!6hPgG)o2Eu_<-qgml>GZP4Jcf)(quyMrR(CSh$ilH-_O>1
zMvBb*T0|wFTNZY#1ErHf`Y{Ij`W7QTZZ)G9|A1YXf5Yz|KeU)AL~NWJ_ClR;+^VAK
zMe*|&C(4H!_`sjEb^7q|5EcjuSo6FeAHGY2JYTRt^(KLHQqto<gf>u%`#j}v{XolO
z>_Z3hn6In}H<<)PL`n+_HM6w_kbwaS267z)p8DoqF4(L=QT^-J#M~Urv@u3uI$gh-
zcn@D$flG4Oc`>W%y)=Qx-j&A?;iZu~0E(|rB*V3$@=j{l9BGAq4(t?5%gf;9B%O3K
zVSL-V5$czdK3oAD6vOm#;Iq!43}gC6Enm48IAk`<u;d{SX@kNP4Xy5i$M?H57?p!R
z<)1GVZxgoc)b3PGe;B%aAwV?&&8*4}Pa57!Pcu9SRnF>Hpc9gm99wbU+1-VQ%rL}h
zj!dQI`-91PT>s>>^z=Kaly5Hh`uWKY1pgo8>YOk>IvO7vYrb1d*R+n#?KT6W9z-TD
zPdW*-k-~SQmDqp9!Pp>e9`=Dg6Q~J|ckBgd*n0n7^kuec$hVM8^z+?vE%7_IhL#c+
zzcQ4%c`+tFz6v=$3Vk|JDN0KJY}^6oU;+EW__-4h5P&t(kVN>q`1tcxRD`FRbDc9a
zS#WHHY1sqBVVEI%rIGgn(h9}s#)b_Db7WYIzjib>;_p>egJ-Q-DkKAFvV;OO88{9p
z4dXb@#)|3@)@8(2l44>?y<4Hn3@+7`m;Eu4^#JI=*baMx6;R`QZ=todV1aSo=vSy?
z*uX{+QYE2?Pmt|=p^wc~2DSv6vohWxQz7gx0uL#Om?<oUq5fw;JKZ#$zr!g=-=|_F
znh{0gtD~#?MCStPy~^y=lyPf2<ogttl)z1@Kpm~Eg^b*QfRKB7M2PhHQxQtjU^Lln
zHZnNKTtK9&bpTU7Fi1=64AA(^0#M0v$++b%W>U%Y>$FYiq-lYU5VcojjTVvxB@1}P
zp*x=nU4-Z~LzZ>lkj|wY>*Qe8eF#=qvc~7R5CE}?kxKEMD0Ri#?5D5M8WXF<-dv6B
z(?4XhtJpu(`&+y!Ismvhh!6sOh%ObO_u9L8L_4u)uSwHQrZ3giDl=vwXPS4Q!G+`+
ztJTdt?<avWvy3I*5Rz)V4f&YaJ!=u3eL~7(!)-TbF!v;TsvV=*zxOuq?OU|Rf%bQi
zv(kh_*h0K$V%r4*98Mx3Uc50Ei|0r(f3P{Lm&OA7yW$E+`#-HW>heW{7#rE@mjBB8
zfr_&C7Vq|ZFHGa+0SL<#93JEjJcOJYgPHNgZA-+;$}vV(wP#iCW8UV7Zg;sm#F1!a
zL4tLQF$Qh-kel!|fYnmSQRQvL=Qf!rV8K%mJ!+0s&+7lW<Z||7cTDH3BAI@<QYy4g
zkg5PPfZ6QOQA^HN3eYYZp{xo9!Ly4(ARLv@mKfbS{S_T*NzQ4ucQF`^XN;!rKu{07
z4qdI+PH{tQ*uykbFPD?mEA!IP98RAkRpW;x!@Np7c&N@zSW}@2ki7svEDQ=88X9KC
z^P_66wa0~yKLem@BPkMyW;CR+%Dxtj(2?ej30*ji>Q=fal1E}VvF;?Dq*i6fx!4BZ
zYt-^aT>EmuGX>`L4a+?w`j(w<FxtqVzeS#qSq59mVCojP!&2w-v_HexZr7GdRxweE
z{CWOkwx%4dEX9dF1vtv$JAem%Nb%oG3sZV&$!M5jYHL$SkeL}{<9_dgXa0?V1mjWh
z_p?!pJ1xYIg<Pfg%MWw72?-kt6|HfX?_fyu5MtS{icGK*O8uGeBAfi>Ty28WAeW-?
zfFqpQy-dtPF?n5+n_V;g9Ig`<e^1p=+zPAzMd5R^2d<*1>t4)LrV^aEW8#=60R|!$
zY!*ZG2!@5<9tkz4b_CDk`QU3Sl9%c(4}18Rw3}XHV6fclnVOn{AdS!#V+dTn9)~>l
z7-D(hK~d&AEITm8&g@Ce_s3xvR=a;P?eWmo`YS8N+yQ<*K1t}>HM!{-7{VStj#UQE
z9$d-J*~K-^40ZD8k_dKD`2;~YD8-BjnZ1L&0s7219Vlm1GJG$yudeRp7j(?5^W!78
z2PB`wQ&_;`tCb0O9S-Mf?;wf<ikyN3?D-`nB~?{-gZ?&NvZ5RO{d001Le@)2ClpN+
zEP8WZ8ElrjC(Od+O(RpYOn(J4TD?V#TcjIEMUIDpiR|uQ@rb?)35101-3nqV{vGyz
zp-h8ft=U~lRvn)8T&Sm9ARZ*RspZvcx&If$M$n2*#QzV9ouSW1aE4^URh+I{h{Wan
zRf=O=2L$&({q&iuSJWguTM+WT^!`8<6JMoZ+uB?Gy-*46!4-wjQ=Ey3i4UTls}G*r
zZb-7FNeE>_4x@CE+0xvPTzJU+>R*7!*;pu6fej!W?nvTT3+W?VeWcp0EU(nyPL;u!
zM#q<ks}PKrlF05&A?=C1hV;1uaZ0Js8aGX*km|wA$x(N-7_o}=2aKeG_Ky4yLrPH(
zpJE0ZwZQzr&uXR+t(Nbsd+RD?Ezp_4r<0ti`5qo5aR}26pLfO>mUM1bDy7oW&^Q6B
zuuMOy+VUadavDeIVYHiskdSnaw*z}o%OnFOWrzV}&GJP9Z~<A*5kg<0dgg^QpPLHP
z0xLCM<*Cv8u8p$b)@hcjN)Y7H@&{-E8{R(3zo(rg<JyT^Y8zwH{p#(z_DH|*#I;U?
zd$jFu;R0N7Dg2PAPdtpZ`>XX!R<D*cwHo5&5vH2azts{eKM(IEnL;>}KA)pD6Kh+Z
zpKrSNa`CK4j-ZR5q`{gzELNya;bm+A!(C|RDc@j<_Q9JHIyuLf;I8pjy{YD_TDb#2
zHXuy0+^$w4X12>CiVs(~YWv4U!?9#@Ht8-ynr+C)bKK<^q7q2!g>;~rpUB`~3zowb
z+isdWATB8h64~vj{%jdg*+O)XB3b+7I#?zvi9)-?-bw#1{6M}VarKMkL%!(nqEgC%
z<=cOWai>UrxqF8VwJRt-M=`M%;Q$jgOr9Kco=#6Ge>&3892b#&WVb&9$&*0;9<eN$
zTfmgT>CYypEEbPgzL!#!Q!#xFv=X6EKss-9T%7-9O~2)UAA@Ug^rmkvo~z;UZo0vl
zY{z1A&jgQSu8Z2~u4q`NoCA?urYppIw@O@1n>Nw0gtt8~(4Xnc4q7aJcTJ|;>Pm;!
z<OI{X)aj{`4m@IL#HMybO$If&`;X4-yEzBYA<m61h6lk=$X;!$W%bs*#~qpSQU_e*
z`Z2Jd%S3_)qIQoP9eBS+w*np(l}p^3u39H(t47zv%gY-|4i1FS%&@GNPN_F{bg1xL
zfOhI!wQ8?$S?)4Lt)pSd_;Ur;yjPxT8WpLHVM9Ev0GEWX-rzIh(CLMarv|ARKbY5!
zI-L6$45}vqnx(G$0q-&KDg!n{r(^NP-wMksr}w7l>rAFR>lSkAj+h)H8|Z*?2htZv
z>tA=&jy0*RCR5cv2qS+qfp6AeGO@3^h&wE7Y-m^YaWl}UeUtNnDfy<!Dex+#BP-wm
zHj%gxO9Cmy8yN);?F^NtNXanp26TeD^<x50)d%j|)Hz!3u8i80E!m0XKCIz$QRFgU
z?hCWXtci9k=y*!{7cP<-xd0v<69LL090hYeIi{l>iB^7XIZ!9?hNIidK8M@3(e(jn
z+wwdIG|u|^OP>fXm8EmS&sv3fR|3P6gXj*ZGZpTF2Q<<rI{QQU2|R?&WU0l?DO+xm
zP|vw<kU|99^&hpi6{BnyTy~aX%|g(fQm#91BS_iqTo(&2*5rP}F>XV*3uLa10<)vc
zP?O~{{UA0xSsb+Uo24tnu!HG~$9L8SQ1ds}M>k%r?Vp%<^R49gLW0*+y`jTW-2{0#
zUtD@0;#ISJHG|KNOidizPT3S0h<#+LMr_LH=84k#F+S!Zz$%^E{SBRbU`+uE@IzKK
z9_7~bt5wf~Hwx*!?Ih1Zx3ZUPaM~&P-xXTwqcvvn?vj0V(aBp)&v@f%Ykr>g+mvsY
z{aLQ?26sNm18;qp87r{KtSVLyMk5~;Uc85Jkr|FmHhJ|vaF(txC|CWC14E6&Lw9B(
zRa~dbi&Q~nmJgX=6?G{G9Bjg^4`z_SjJ=(8w&`gz-5eQtKi3bA-;muVK)MMGEvG7o
zFM<VZ{tlvdQnEn?u++6lPP<<|4o==~ZX;&GLZo0@2bHAxJA~r_k7%k&`aP!+fI0qe
zpuiXuf=W>43js$1f3qt3952>Rtp0EEV)Jv{fkn3UC2rEHfyruv?w^ki9y~fE56a1t
zSF?%v^X!9Ia>wLZNamhBza&rYl?(teB;3;YmLbR0<ExxfhhU0A=GVe5C@Ojx-PgN_
zY=f|wj*L2>0Y3g8H_5F2?*hsIXmEs6$wK5#_fpd2!!>%akphdvpvjeyva7&V%hT?|
zzyJx1mW>xc)gk_7oMS!+Z3I@{?+M`OpoB>}1qMoPa#B+0hrld54E;hZ-aQtK2mzil
zGV*#H1H__$29^jNghe7ywIDCVFjq=Mgh@1`Q7vk+6!<J;q8U)t!%uB&gQ+vQL|_YE
z=;){cvnPLaD_g9CF={;DL0DEDDr`u4<WC?EyFzGp-x4O;;0$+z+6a4cfHCR*mN~3S
zxsS=$o#5Sid*wg-JCx}I)D0+9!;^I|K1Blg!PmgAKCiIw6bcmdz$1fSWa5V-C7?mN
zXBEVKFA{vZ&S1yu6=r0zR(^{gHZMZrXJGap)Q7$XRgKSKFI~=0HD#}Hy)nV%I~4Oi
zV(-)A57_gSCOB7v-}o`$4&AM+e&|$T)hVUQw3DU6Lj?pA7<hr$76D5Qk%M-T^ZsCj
z-2MI|tRqa(I|Vi{#yxp4k4Lu{8yOh^yi8Tgc>HR$cG?E=MXTKQ!uk>@A~T@x#gYt>
zioYJtHx{iMVblW362Nqr`_5yLf%rde)m2pz39~H*<|WgMVrpDeK#_;_%Mu%Vlj+fQ
zBo0=o+{Zn)u_t*U&3nyqGED<|WS_KoBwe;%f9r%xCEZeD>0C7i7e8P6R(fw}tzK3?
zajT^srpz*N)>-IHiNUXZBNp}|AMKJ@=_IBzgn$6pKIMN||Em<7Zf9noiJ|)Bb}6tC
zpx`%|&5$P$-C9&tWx#Wx#faJusTV|}l{XgZ9$5pfW0)JTPYuQG%--fH9enb;o3p{T
zrDQ|#F9FbkNt=HurYXP@bGj=*2}O>IB8vITOL$Brl=Zy$L#-r<mlVoAJxh-pMyxYc
z00+aA;NWrR>N!&Umd@zHN2gHW7Gf;|1d(^Mea*)CP{TW+8IU(Ju-ViLKw^hwNpMgQ
zv8}315K`v|(>ch4zH#FQ0IOyCGa@R$yx4ZB8RvgdtU1na6-xHB4a6r}d`#XHVw6}2
zO@)rkPbrm!<eFkZvSfAHz#==ztt*kl=FBP~Wgn)rO0(FftLB*r8uFLmTyL>4iEDRw
zPp3PH@a%VYtLJlVHE)iTS|0!5-C|?O^3jPp><;$mv%!0u@HO-@v_xPI1TW{QTf_7)
ziYt0%Ephc~Z$8`4Z{K7ps$HNdsRqKG-(x%b(%o^Sre5zXZzsx})p4qNHnx%M96~?m
z-dyQfAFU>Fs`wPKL<aApi+IDNb3vFd$egvv&|#M>^yBc79d}!;Dt?9!)mMchcpZSj
zpaz)|;qzfysJtO*0A9jU%wZmNC6LQDbOtEBRltza7)#-=xl7;iMed`y<LjuWq$pU0
zigkd^{@V%`ypy@Wy`Mt@duK0pH-Y=eTe(S~0&s+9A_I#M@$KX!LP1El?lvC$Q{#as
zP#>`Dw>0+!yK*oonp8eVVcOmFM6hODbwgdg!)~t9^4L2qCBaVdd0)>9YJWzn-LW}|
z54$%f;;)JbW2Y{8*d0jh>(bIvXxZgh<sU-pEbQmag<`?BR%V>7H5)j46qicGWgg5U
z9~`W3N2fjjOs4RX3zP<>rA`OQD28Cawq(w`k+9Fre(mn^7?so^zJvRxdf+S+SRfgN
zm{?FLO@5lJj_8x)oevURlr<7hJq?c?tOt6aPkDF9p&2+0-a#bHkIJqHnP!7+b#eM`
zGtD^zw3L)8Z9088G^vYJ2%G{qwUg^~qNl5-%MT&cy{t?pQ*+7&tfZw>mggK)3RA9Q
zT7>Y0OoDypvhN)-oPjUs-9d%FRt4u@x(>_V^Rm5#8S__MeC&o5Dhak-;=SFH)D$#l
zMdMY%HB|Aveo@J>(BJLlH>_lO`1jBwQmKD=O@aJ+l!>gkE)NoAETnk!1>n`j2R`E;
zpX715CHAqxn52Di2hiO%)2q%R8*9IA1EzoN8v8*D#_il6u{%(H#z_$>rsLgAQZv(U
z7Pxk;sHh0=JUuNn$1?rd1^FFrOwKK{QsZW+)%<obnCea|T#sC+s$p6<bHUWr9!9C^
zj5u4Ec{gNJQC*Y#@Ck*Om4N4}Pu-(hR5iWt<L0TS@HulmjOK*zXWfr7fdphl#^y%?
zfV;y+uD}Kqbs+1`F96D_zye?`*h4BVSvwE|k~Y~CtoY0fW(#@}-1#gwIl@C<S>wVi
zi-VpkQb`@(@Xxc4T3MHN0hOYsp?QtSun6BBz+5|F6~Pv7m<u4mw2kP6fERR6fkcOo
zchO-pE(8zGgKnn}N#^fijl1<f5-1&A0j-~%*xeO7SWI<~Xz1vcTHU+{E;hY@9yzPu
zA9DjUmZo*%gW=-J=+c9Q;#=z@7wl}lPuo-(5}<mz)<64smBYLi<`1XG$}9c7<5Y`u
z-NjOEDa3A5;U{O*z`!+6`xFYW1vEcxJaRHjCX=Y$e{C`dq|c%y=QQ&4jvN0a6ZFpK
zJgo5`?mWx(+xZAs!EAZZEY<uSUEO(M=u$)Yd<x(xQGIvMAem2H;G{KksjH3T=O0#7
z3v%4_<4kRv)hjyI(0g`HPF!|rB0fja8LQCGDW!@@9gpAY`PI(plKsBolA3qd;_W0g
zU@Bdu4Fn)Llz;$0ZkUKe(kdBQLYDXhj6Nx=Kpg{nt1V5{$AI?hc=b}`6^yu@^i_F3
zh@w5{KRUv$Fo*lq72ZrInd>_Oe&>fr1q=$DpiO}|NbD3;TbrqUibBi;&+26(ejGxq
zoY!kwY)*4%65?28%m7cAbS?^ewq=BCau)&eQ4M(ub^ABwNOkz?BFA6;F%WUWjod0R
z47*%Qb!wGY;7nrsPe5Xs?TE1mR#Gv1Xe<zw9Rffwf=0XjV3p1lFC%s<sdK#a{c=W%
zXVR@SR3y=sp~YSh{GNdu_IXx&2E7Ax9F`a$=>m@aS>C(rMj=_3SceP6$y0AIn;}Of
zO&EVgHBpj1egW9oU%!4W<NoA_{nOrtx1FQ_uIB|X3k>n`JoWhy$l>Y7An)Vfp8P-u
z4ffDx1Bl>u62;y=<k?>cn6_y0I{B9rSX@Y&WzX#hN*1sX%@fQWa4nv^h5f_Qud(A=
z?`lBA0KZhr9l)ubw=*oan}<($^2=QhPQG^T8v1zcSG`8&DHQAp?Kj3#xE4MWAU^-|
zGic-p<|%ld8kzd~`kGu{x;d~Po<dB1ZHJ248-;q3_Y*?dQ2*%BfR_JYb;({{U4_&3
z%k&*8S!j*Hf39E<a+DVb$5y?_B~hNh>V){?0)_xcRDzHpW>i~y`<XLm;<v5gFDD@+
zSg8mFE(&nosesJP%jvz&#a_4Q*U-?z-R*^G3q;+0kGukM#LomLpVf|5z_aEL=w)GO
zk9y?hR%t7n19Wx}zM*P^J+?mfUex<YE_D!MG6=FVczINDc5Xtp251{?$}mOWHzDr^
za%SN!f)tbvQ3<i^`1F`>u@B-7T&i*)A=%Hb(}id5l40%xs>A{w(CtjFi?rsaYRvzt
z_k+k{EXm@UH1^Ua-LqZ6rLnUCtjAq4vF?$`@Z$(Loy(NfBl<BC-FU#5;(>1SmGLVu
zH|U&&aRHn%-@cvl{bukBC*i_;yb`+(Vzg!Bd^6in_+lZo32uQ#=KFW=#1_?=1S>|z
zHg*BJlyq8yejeHp*kNI&*UNYbqnM;t`FEvaDxiRY1iP%eCyw<Wf^!QnTb9Hjpni}H
zhCvn<7LY@XjoadMjIb7|tgLL?M5ACj8{L}PkHJcN;h!RvkW@@jd1a-tdLO_?EoUab
zDhRf%`g}xU(66<g1&$B2m`D}s9v3hlzgmCGNSZVNV?vL~z#EcBe_$L5;F=COfqMXa
z%0*_Vgrup@;jiUJLn<i*{=Tf$A)hif5xNLF0+m0=47z6lpP_v7IVVSmbn{@(7Bai@
zw2=->hP)Vg*=M9Y@lSmFGFs2dBU5ws=sx#hdgj(RjrHN=n{JZuK>6HL8s9<P7NNu*
zG`X+Q9yIG5G&yCnNAw&KtV~Fx#V%5E0Z+sLPBKX8$kqZ*>!@W3Y+;LM<EersAwRHf
zvr?G@f)`yjW!%H;VJr*`EjpwLN@*bF|5Enl?jD@UKMK!|KfX@4TzQ6k4QUMTobRcL
zPt3cbcX0o#)<(tX3hZp>E^^gkDSh0Lu*}=qNFVKa4&dqRcWk<4YXPeGb$4-lABOcu
zN0N7Y;WXB)1`d(jNA)eTl>Nr9Utjgk>+9$M3zet!yRlF_q5>Jq*kQ;aVfiQnN;_(F
zY)mCpL-_7h(vQG|ad8)*0rN&<FHLTHXv<MpVe-6PJ7ai+P&cN?s^-w4b@NMpetuEW
zX%wV&?E==di4U|6DCtpaFtB`9q`T2>Cq^I80N6Mn$^(nQ;!5v@1b5KS`X{>_A$@dB
zhmiF0@_LO+4>{~mC+zPUFhgrdy%P;rO+Tw2m>hty*kJrPc7LQ_@PW8cn=(=q_<_+!
z4OWa=#<fRb!Cbc1eD^l1#EseWWG<9MdB-pK9gMg5w7IE}>4qUy{!XrSKDv0~gjo_F
zc-X1@8|NevJNbDL-q0F6$Z7{U5JYqt@Tx+S3Fj9aL<iwAG2Lcc``e44sheFc7+iu#
z%@%>TW&#j^Zf5qHXB3iF04UPRVyAoqECn&GGJOso1?WC1?)}<!QAE|$)Qr>B+R{ZU
zu%T*yS%=KU|C{B97~YhW7K(+-3Y{KeMwVV7afhXf9`-j1=|EUQfhR3f71SA^)4R#U
z#D=6m>;VXwV4^jZp#8BM%;#aIIb-)D5ki7D(NdzKnFEVpM94)2a!eR!Loeg(9J@7$
zkMeyiwj~I+6V5j1LyfZlEr21>x9X9UkZNTP@qCGVq;r;)eWhm}K8>ZDqD#=+|4YM2
z?i{WkeD;Gcy8kxqm#ogys80Vaq_E6(itv*Ad+~#+wJ}7{u5Q5WgY6Sr+wfD3fi+71
zL_t%30jd0-QMz;z1xmPS4FZp#Z)bzS+nsE!02!9Q$lt#J?06jfiFX`3d-zIWBtdiN
zgk%MkVq7~%bSt3SfU5zMV|27=7>NGS$r-SAtVN*mdUx*!Kr3$D%$_i%24=Y?cf;g5
z2>iRV;7akd;(7w$Fped8y)0BozeGpJhVX$!_3$uI9pIp#BEELx)I74o^JdFI4ZEFQ
z`-(c+76OV)3umG=T}fr78?d_G+>!gI^Lkcf^(i`(qYXs$K#75ln0FVJG*IIuNNwPM
z85!OF!};dAQC>_0DI|k4Ulh$97>)^jW^QJdv1eE|OQ$YE2U+tvo4c;!u9S+ZueaCA
zj#hqHBJr~o^gU99^2bq6YqQ1vCK%2pxuL{}QfQ|kbBv9WdEA^w_9C$ljc*dy0w`6X
zF>vdh-8}78$6s@h4@r<9=PP5B&(zExv;fWCxg6l3Y)`d>!_omfk&6n3a^%Cn3&j)>
zqp;)S)@e!yNRm!;+fb}B1Xvw`q-_-BT8q%-3@k#EFKqw%)Ee|fo2*j7O$B9TiXNG=
zi9wZK0Nbb#I!ApPShO0e9Q^sx3WU%Fosu8T#UhoiHT8@pu!oM3ojksBP7nXObkEy?
zk{RZTfO0@Yt%hYJ<(qFkJ(O^A|F6Ug`LD+7JvED9if%R;zL!{hP~nRNsBYuq&-h3K
zo(jednn~2B10j4a&D^UF(h`B9R4?$$ItEyC|A-Ne9cwAQ<{Lf{Qw}h+ehr{67Zs)Z
zSVCulUC+5EqaHb<O55Y6`4Xg=YPrWAEPE9%$D}kFA75TVNFF(3UuE_9U|+_i3DB=X
zga{hVb1KXGe<#=MK9r}ZWZOGB{1djJ+{D(OCFgz8C`&lW2e|xg#Nw0d04WEqIFCH%
z)zdV-AUF@)bToU%VP!MLNDPSpWo1Mt2(n*q`5=IdO|FOBzYb$?<Uez8T0CIoa8Z%@
zd8L`F<8_R6Xnaigw;jsS)Ydq_>>XfQq6sX!!Fe9t3G@scY>@=>Hq%i8+?SwgPff)h
zrPb9d>rOH)2WG;U>u~_!jyZk&LF0Ols$YmI-CQ@~7vNwi2M{*LoagR`^MLz2vf*Q`
zH<CSO1AAaIrihP^@A^v5=3uyri}dH=&h6vWt5s*-EDh_Sj+G*cJ=pR0-l=730=qcy
zIR|XH;OyU38AgYKS$&H!Y(B<{&aA*aq;n%i<o={(a>2#f^mIxf_p49dlT4OkHNN0}
zqQnpMW9=N}@)r~GL|_;RX36lSQL5kFZq^?-L%W<q!+;W2p{AVQ;kmZ3_W@f0V!Uw<
zp2>5B19@b;$~C&s2U*N65woBJ9oE||tPbm)^Z)9w=Efps_kh3triX{}+EXXom#DG;
zUKnEIISJ{VN9k(WNNRM7vLqn0U;-&yvE{at5*|ab_{&y;x2Nh%W|q6FUQ41fhogf7
z7^__Lh2j=?tS0vB9+6)qeQfYD@Nx&PD4U;!mJSL|0U8-q4YPOn;6*dK!tdE_);0@D
zUc;@!D5-B`QkUNh!<ZgSr*a2C4paZ@OVcDIEt?jUO<xrcMw*~j(T=*C-urb6N9Zq<
z42)$QgxEv@*v+SHB9&x+dLRfR3c^JE*cCvL0UY0e24J!jTV)Ox^u>PjVu8IY0|?u>
ztT5msJjD{KAQlgs+_@g1c7+6`B(=h~0(dWLfxVlp)oLu1d<Nu9vkDdTVJzp|@41K}
z1`0Ua4yuE;K^fCuTqRw56qT`C(@t<Gg}CveT~$u>`;RLa_T!-;+E`9yC@-OuD=Fcu
z%KNG~Fy5#!d}vZKoO=8!WLB*o6WP~n;Wz87zWT=Ufu6z$l)3*iI9-I>r<1J(o6hDQ
zEn|i#)+VhE$%i~XnC3|)@?QvGdVGYw1%qB_{{i^x5<sf=R?(TMfIupwOI%(>4z696
zp!>gb61dcN*(r&7G=TS4^o&Gx#dq7qy_Kk5D7Nj901O&uQtUc+oq6j$=DUN|ZkfHa
z93(B%qqLRb?RA$(l$du>c}>4phNxVvj;IbxOcBJnd6KUNuCm!$a8EeGoj0c5d|RAK
zS+p|BT)xWIRwqgsYtBEn6Y9UUOi$JX9xG5mvV@9=h`{ko*90)L4ZD7|n+}&H?6;uf
z1J(ye_>?%?A&?+t7WamkRaGD~j-vKPXCmi+#<N!?5&z`oEwJJMo&ZR?yZBOORYx_f
z2xMW+tFe+!JcM}^1`<>u8?H$AxxM{1^kuv^SU1(hmF1xjJ3KRys(daozv5el2m^%h
z0yuq%L}oXH6(cJDVYYzN;K6KJR_C<&rt&Ny8g=@DkF?7^JzMS+Fl18}N`WEefGodT
zTYF+>XJ>C;B*;5H`0uVh1!xCAh=2y0&6)NBhSnP}y8}a?+FBF}I0#VB1!I>j00vug
z-EbXzr1ULIfWZdLBgjmHlswd`Lq)2i#+pkNY~P^&)ye{K%?6oPR?+(7hrrq3l?DKA
zYAU679<g}e!;hAn%e^Yfh3zGPuY9}@)V7nIg6I@(dzNx46H)Ap*_T<FcEPLYujxzy
zo#L##?^+C{QR8|+8716y%N>R#t4y*biHrG3_w4I;@NiKOwpva3V8hD7B4qtxnT@RQ
z_9F}o^puD;hs8za_lLDHe=R~UWBC_rf)57mvh?}zJfTG4!;e>y|HK^97|zq0rg52B
z0z?*Y9nAUV)v5NF46c1*!a;X#f+O7Gb}`H?`E+BI9VQzCXZ$?lpLK69DW9Bx258u^
z0eGH~Jlwjc(Y^va|Mp9B+xuXO{>xH3hTY@U$0c8QI50)iHqHpVU}(Mh+c|0WeTew7
zRnXkQ`zi*>frrJz9^!@l!xG#4qUEkMY%kcQGA~b9i^v${zXAL3reH=!=g;9eHG!$c
zofmXO?q&4@6)P@1zJTX%w>%?QC}Kq-td7!9&Fb+*Lq6yB;(#nak^Q-5Do1H}5P72P
z#Hwm)5LJ#n08AS9ZpjPLkOB!tjq4<;UOjgJS|{^s{7j=_fX237MniG1|0_j2{A9zS
zRt;{?>Vg<M8ikT1Xv%Hyg<_9JEgP?OZl=^~SZyz;5N0|yychwHb(Xiq%>u=APw&^T
zC=2jezGpecv5HAO&GqZNmZ6T+Sgo39_$o3V+m~$4Ez>IKv~=e4=!}gBfq{m4JTebA
zjU$}MuO3DQVsG|`DaRSNeTJ8dwRuoj-zga6{-NyL1p6J<PBmKb&Umhko6PYlH&n^b
z9v!{rgC`PLWg06<e%Hg<^sy5jKX4;XZef~w0;=HTFdS_E1;p}iJ?^S+5?gKo%P1*;
z;Obg#MP6Q>nXpo?9Tydb?B!>zXCic_Iu1aSH#U~p!a!M8@?<J#?>!9z!`R4(F^#E}
zRp;WiSbVluixb+z+8TudtJpsihwW$M6AK20XZNc@lgO{?6u0Uv)QHbHIt)sxeOL0(
zP(uwN15^-x0~{MtluG&0^{64q+W>ll9noW&nE?K3rbseimE7$GUH#|JV2Ifsep#g%
zoa)dFzyb$>5HtxF0%W1v<c<VFziIVPi|gHwz?mghRyF56nSt~Wm=-}5M%NU2Te!xn
zb$V!U@Rta-!GY>b?qlfV7z9KSz&q&`&h&nmtpyw^u(;it-i?p{@5MM~+$)ED4u&?n
ztK+jGL8l4>4?K=xtdgfq>LZOEPsO|OMe99$NPq&R{%z{Q@YX_TudY1E;o;?7N7ob<
z3X6(fB*mAF?SYyUtNYIdTAud<Kve7Ma9WVF*YGXqnzD0p9AH(H6%1sLNe;X<+#L@-
zjffYrAKnD%C&Jm);9?-H#PGM7jZGGWKT1kxH~zEqqyASSbbfmi3jOy936e#15Kjen
z5oQw`=$eX(EBo^RJo@;Y?oI_Y0Ua5EP6dE1u1JPT^1ve0lP!BNVVSx32Z(x3PENpw
z)#mZPkN_my|KHi%oCAvY|E><6)2S2DR=Ca&7Ajddkbo2*NNNkMG7Jk=IKV;!=(~4>
zfi_a{?w8_<mQGY)!xL!6fn!{-;dZrT7y(H`QF;0I7bD-lf1jC28(xO72bFAO>ogFo
zQf#5Pn3Ntont84Aducl@tFym!5}Dg%!xYuj4F4W7+lgxV({weOiu2wj<(Rs$oV1%&
zhS8K&iFTJQc#7N^pizVvzUP*fkfxH<1LQ!uix)+UMk*fJ*Z}P??5Xp#4H7#k0y^e_
z#Y2ypFHZ8`mL0-T)ma<%fslW5Qf{_BRfPq4qCo-yWZ0|==#L=ZscjDo(tz{`HWs(f
z1v0$<w_Hf@;ewz9BqZXo6GU^5mSu;|wzkioKU>pbr@r~hz(a!pX~1_@dPTF&qq%NY
z1Qqk{9R!d{H2&1PL{AU<7U*U`69ND5;-@#@C3f>+b6xJ8XSWYeF2FV5HI>$llaSq=
zx2rj{xl;cn<&OT<i?!oBWC!F%x3=F%yoI4P_<(C<f>UbLB?v^lylUH%`X#T=p~eZF
zz<4+(FR==(BHR*q1tTLHRl8%)cEbQyXSiK5jOW>6>;eE&5Z#-bf9{Pid#<|KK=B=X
zP_bZBHWb+u7aM!_{NCciJ@7wps9#+8SDb>@6&LldFovz@*&YUHtLA_X%a*gi5ZkXF
z@_v7mwYg$hpf)x(j*ca|*`HuWZgtp*u)`q<hbQG5P=mu@N;kW4vr;1ybW=d7rXVcT
z?;qKvQ%5=tJA*dMJVuufPFHL*$|WDGvJ@F)3RfFPw2qhP9rxZkcC&guTmBf!<}d>e
zB5%HIuF?`7a`MHmi%AMCL+ThPWd0#`-W}AW6a3YyZnNTu#&vjL;AOG<$uKuDK0b0o
z5@4O{mvhGIe*UD%^cc^$?JJ+&yB!q(OmV2n;AODfe?MsMaCcm4<4kH|)lBf<2|}Ru
zdzSJCwQxQlE>a>}yI!}i*qLW(DE85WORqf0%B9`njJBtajfr{P@B~_Z_kBCCn#e<V
z4xqh<6zL6)Ltb0!{ncQ;$|rdWY2jOF<OFJ`#^TaRe9DXVz54|^Nzj0TRzif1??M1<
zlIweJz=3}p1J#}~$7O#w!BX0s2sjOhFWq_maV*4g4XEv|u9r|?x=6k~3%fhB-LGHE
ztkQFlg5u&3+NW5B_?|SH^c&#oa9v6YTnti9t#Ru9djsL(Mie$cHAjpZTV>B(dkLhp
zxk%>lmJ8I>0JuOegO!{>`ByOtq+nnl0skoChUmP;Pi@R4`tYVn(RM>eZ2RKkqgG0L
zZu<c=UeAERdT4&Z{Pv-AuU!$MyPWh_Sj-mPncjP%%aE=h*ixDh;e%p6K-NDQ%@C}S
ziC#RLl!N6SlHp#DTc+CgO0I4-EdcB(ua?;5{seYB@(3S$wZy=^u+6drsN5F!fq+by
zM7BmGc!HqAe{B)hlETOTwr>B(424R?;c~NhO@Uzlt(qzpp^uMHq74gXs=%L~8W>N$
zIg7u@R(j!P;Il&Pb@QfJz{lQ*qpSg@gL?{+r>c*ZshzkQlBoO+#RaVK4y?RLkFh@n
zKg)JzOdNEapIrzSYsgcb(w~c5c;fis3kl4DBYb;lFP^-ebgqzopg#q&z*&s@?0Dd8
zcf|*am*AI0XZzLY-w(`5V+RX;s&NSJv?ld%ui($48=Ti^PyU2A9|d*-<8E7edax*G
zSIi05Zt0ZVzhOmJ!z(Oz(CSIzPT?i;yv}_z<$r#M0sA|Lby`d{y_3~tXC{7nkRkXd
zpBYYl7z3ywRIGRJPO@bUr2MG5ee!MQihQtdCKdBF&mnjEuQ1tv-iqtwt<-blr&s=c
zYXc|jlxF|7xyX+dZllD;m)I-e-F5okuM6)KjL`r4OC;nce`${3rpdyKhe$R_UsQF_
ze?A4t$)`wEn<yQb%;~t|PEYjjQcz<Bxk=UQ9^_O*IlCGcA5SzAz5VYe#||*m#<>b9
z4rEcE0t~n=zC93Xj^Mn`@t==MeDYC^6TNCX702GaN=hVX{X-N)@bA+gB)4Duz`IFk
zAY48CL8xx01}A8_FjoKcN94T4)%0sT|K7v!#mSfJbUv?q>-7O?Pv^_oZ@C&a)`mF8
zo!f>*ClBf|0=94Cnz5_o2Q58mgF}UPcdvC-nVcoo#2K`=>k1HmYb$?C#=3UVhX1My
zeIE(SN8DFj*2w8d5>_Gpi`F=|C~kd-NWE!JlFB~1%WEE&GT6BqJlj#W$Ig6rxX^rV
z_Ruf)&wI{Ux6bgiY1f1^C|_o_HIx9;sXDfKqA-!@j5}|NLu1Mbe^DTp&&r?GA^t^F
zNe_`&U^A9`3zr;)_nMz=L(qNPkPQ3(p|rKXVgCVv|I`wEAYiDCw$sa&%IX^^8r_!r
z$mn;Db&Zifa^xJ>^b3~)`?%RcNE{IshDRNGwd-+sxVSRVX|wtJ`xBFpY$_=vfOBpz
zov1Y7#cLQ%Y3Wxb^*Hd(o{!HEp*^iKuF{NWem>H(4d{4M!=)dSIKB~+G4MON0X>F%
zXy^q>ikeziclTqx43H6t(KDx%m+Nwbr>3QiT{K6-+n{RIJWnyy5-eF_40S)*DUW7e
z2%pdXzHn@5cEE9!ms3rQa2>>_Ak+e!k5bwRZD?j@1{uA8K!cWT-VTZqAOUA<JuJ0;
z3^E>QAG|0iPu`avAtyrDs_k0dUyeI}UqUPED$P@U-hEfoh1*HYc(kgJo*q{fOe4qw
zVsW6<7r$yT{`C1XRCeI*1+6G>LYSHHF7?#@+T639X-nd-q$k0?FR3G3-A4+4KV?qy
zTN<xMj6FFV-R^ltyIk7<;stOmS1Q;AVYi8~G4j)7)tf>}ss>2Jc}7i?(x0^!fhCI1
zmv9R-{g9hwMUl>leM^*x&nbpSa~Q$csL0C~cgJg)X#=Lo*-d^JR;D{b?X=}0!ORBm
zm6+I&Y<o;h%-Wi@^lO;3VhGk^Uy0NB3eu2{1!C_B>O-Oc7uBOVmy%D#y%xy)as{#Q
zP;Q)_2lcPL{z)3&%$yt!@3p;y!oos^#q^XEutkzqfFoO*r&t`5r~T>ECt6xs7>t4R
z`i%_WZeSzEZ7&)45CEab@4$5f*j8HM=1rKeg5=-sGcwo3?E9EQ<)l{tdoV4nAA^Ij
zp`kEx_$h`dLSXn+SyeS^CQR(}6&(G*dmlV?fJ1OjI!!f^FI0h$8B|>`H=Vkk)dd!-
zkbC_MtAvNkILC7E;>8QNnRNWiCs!TiC;(5y?TFaN4mYmtE>}GizD_QKtJ-TIt?=YY
z@=d0>JOKTnmRBJMdjZ2rr_i+Dnh4CP;AL-b4}}ScCkzgOvG?i~u8X}AIKLDPftIGP
zuP-HqE&Typm}-aRcSeAg?!herMYSrRQGn<J9dUH>t~zNwSko5e<a88=m~H^S7!FQY
zAI-n87rnM1q!y1<y^P*qnZtH`%vEW$+r0SHh6EaUUXh(0SRJxv9tMVhOejT}TbdxM
z5)JM)Lp!d#2Y|&}MotcpFjyks01dTbc#Ec%mK<ST`>(in@AwQW6$pXB28BM{3OX|R
zRE_>0Ki~?6g;6G{(VRPH<ai!T6ScHBC<wvuO@ZwRbq{Dn%E9}U86xk#H6o4pr%#&~
z4ejsm=j7yo2=<0mQf1|P(^XLZfKPG|e-{P@pG7CZGTE^5+1XR2vQJ@`(IWzQawzP3
z(;Ul7lu7U93<O914Ax!cbf7Jpboyg>`N|atLj_}nk<u6H>gt6892{~<YG4POl$Esz
zkbt0|pdE#HJtDx)4s>_;c0d*ahNZ3)tYJyXBk%}<Dfvu$9Q>|6-@pnte7$oV%3t(M
zgA?xZTmU&x*V!ZQkh83E`jAJJn%dB=m+|RSGH0xiiU0B*kXC^Okll{~x4O_;JovPp
zPnlm5&?M86HYQvtbE@bTi6Yg+T`k=}&=M}-dz@}X70__iwCzK#y%>GF2mJovQamNG
z5AYN~`M^^zaR2^&FlzS>0bkXe96o6U8ylgxs(V7Q*bDOIi&}bbe_!9vcJ-$;99OQi
z^7Z-p;=2@YZP{IS<@8>2sszevqR5L^^bb~F^<zwEIGC82n3?NV4uI?nR31Q#fbcx&
zjJ>KTm(y8Yuu`NAYpN4)Ah5Y&P;vCao!sajMo{|q8q8TlM;M)k`3^olrSuTQu@Sk~
z&2G{Ugs^&lYiQW}`*-+<BRZ0IsELn)kcEY1iHykS>lcBN3uF^EWd)|~?_fs#`YbD5
z@Q_L$0rt-!YfM|%m8pJobm;7AcRaQdaV*toE<i&^od-Cjj*gCl5AYqcv$KoY?e2&I
zS(Wn5dqY|7NC<6%SI=>(!64ut^D0Y#__2f*mXwI^IEhP0Rh7%IyeR+rUX%N|<x1<<
zmoO8C0#EoIOyEKFCi(u~P9(2@OyKxtRN%$0rp*LdBQ<P~Jqf&$enx0#Yl%BpZaP(B
z33gxMidURUFdc#w0p1AKFL-?)12HPGGyU@Ja=yDuvpKGnxHQPiX0}2CY|<fx+Qh^J
zEF(toI}dN*V!!3q8OlF6Kqun9as`;DfFwFP-T(-X@YGA><x5~kz&P<X=3^$Sd{9jL
zOhg0~IbU^yfU>*0yR9vlPC}?7aLpT@lutTJOXDU=O?lv&YsEs3Z8N2y`{dubXIXIW
zPwmCaq|H<T{zdMwN)k7Oy>FnTbeE{M;4A<6<K#H@JLW;bxp`L^`Jvs{9{*owUjmim
z+J+m2CY7QRl}Mo^G|)&%rD>~(=1GG_n#Z@2u%i%VXiy>*4NCJw2+=6bv*tlGkKei8
z_Wz%C)>-F2>-^_id#`2j)%U&c`#kq^-Pe8H*L`EVj!jyYDh-i#*kQIb97AFf6N5D$
z3^*S-e*8Ep#lq~@f`i2*B<fUG$P1u~75`b6Bu!T?iiV7rFJ5GP+X}gPt<C80a7t<_
z3?~pVlhhoTpQG@N0L*laZTohaCma`TZQ~hkHnZ2&*5V=d_4V!DyZ6dN*2U<?9&`5l
z_oioz2pOh<zNW#z>w<gssO(_jWW>{1pY(Ka?18#~`@qpi6@Km12Ds3Cxwm)RP*<28
z6j0$gwlHpsU;WMBFTOEO_wdecExu1_ZE3+M(&MyuPCc`=wKX#{%VB0%kf2?~A5WC!
zBFoT0UNX~Sw{`2*RK4_Z1My<xQcMaiB8rM&M~tW+#w|VbyMot4DmRC6>eMOd4ETa8
z`;dBWShsE+4Grh5lh`9QHE>bE7OaXsdf09FyFW%~7+~rWN<aSo{!m_0%8Z_+rHvzO
zWQS_=vtGS+4Q<GnRlJ%{qwR;25qie$Ocmcs+RO%-H<|G^5<g68ZG5KFl<ssteV@zi
z+`+q@N$}voJN_?GoaThf`N%?Jx{1fAQj*q7L;jzmQWM$o(M+*P<=?qWQy9x_#JFa%
zWfSO%PChl`{Fqr96Pu!pa2bp#yN@FxdgnHm$GDBIuJwhoC{|b$!gbp_;`8@`i`Y@P
z5oyGW7dwJ!)~(yZvHIG)xoBl!MBBD=V4HM9Tbp{8$Bh$rBO=&XCMX&znrdo72M&~C
zphmMj*`oQ_=#E@sYHH|nFJC!%D4In?s%G|-LkuX6?~d`sN=iy76(b`uh@ZJ$Jy8Ad
z@U;@opK6mwqL<IxvQ4JHnVGU6s<{O3``-I$Y3r(`s;>SrB?Zg~u7LFh|K7br?&90H
zxGrA2xV`d9P!KI2=+XH2w~pJ6DqL^2m%G#hMn!keX4h@`(5|58G^^S0%$dHtVIK{R
zjocBx`onLl3!p$;KQaws2<LRpHJf;8!FcYuop(MBanhhvUZqOnPXxWnsGK`!9FCGc
ztY)O!ns><p-`&cpe;|%1f!Yp}U!eQgvSrIyTRzJBcfWi&wXjoKx>W?x6_s$Tp1SO_
zh!{+VXKc~SUf>pL#wijdZCl4*I2yyE;L~nP!C6r3HWDfMZgd@D7z^!so5T>izXJjY
zUS;bO3~l<_{z1Bb%3rvirp&~-pvdMl+LW37s~|D)XIs8YnBNe}RT&wgNRU~2=E`|D
zZ8{z?X4jb3{K~09^eY;k2PN!B!!_*fg~yBz2nl7i3TeJN+gaol!LoBReS381CJqi|
z561(0_hxddIAxv#WNsgoNm8^L+(WRoUPj5q&`_PtbE<sxq^g)#UrWsY>4R7+aHf6w
zmA<N~s+n2*iJe~HWqw{h%EvDy<=v2SMuz7X6dCb_iavko>2dI#*t5BSvw0i$aw*$=
z?%a$STR;%8OjQ*X=Y!wma1`+K2me5^k%uwBjB{4+)MS^^2AjQq2>axpJ$VOR6crs^
z%oJ$5U5*epZ%DZ{zVLUH8(_*e^lvoss6F2a&jB=7lj=01@oqSkfXAfF55x^A0wreO
zyvbN`xTd`yreO1v1L=nO+UL(Vqj}iacx5EL6oyl53Q$=Vev!RauBxS#G(Gv}4`jJ#
zaW)$m5C*!s_Z~l1lOF_8nxyvQC~S3~p1id)*hcKpVe2Yr9kVhsK^Ml#`s5@h-x&Mo
z;g_=tZK3e;7&n9+H0qc5Jupy7)=Ap8Z{MKG#scjCYa)Z(b*2v{Z}Clm;lNBybdbS%
zyb}|;J5%~F9ZunOR+fV@QK@|5L}F4B$CfP!kFrxWiaa+TYA(AK-^@`Lc&Zj7?bq|9
zaRlmAH8lpKb3!|I921EW6cA{+H6zK%2w(hN!8f-_hWYI*yXfHXt+H}gu=-;O4Ola=
zv%fB91KRBUdnCsR0UIPYNJxHG1hZ#}iFrr$;=&Z6hYKT;-T-VUE;=EA;vv~HWAFVh
z7<Vn1XPt%?5FUc<-}@A=F&$()e5vO>%-0~Khwuo&6Sp6Q#2=Tk4#uG8`5H5?08?sB
zz0x9R%kp>cpi4E@)3cU%3|B`Le%u7i;naLwxxDy#%l-+9KcTy@*R7%|q+b#9?4MKA
z7<#V;5p)~pE!1^WUT_#ZnbvW2&HihT@V<S+gM+Z6Kq}=PUKXri22s+96T$U@dx3!j
z1(TjUd4k?5KuF-<85^hqjaslv?mT>`YW@8mh@Ew9&KHd5Q*Ppq(FU!;1r#gKH&cjL
zn_wq$AmXzPC4Z!r9S1QagZJ;lc&Q(%!<=gXt%f3{p1*hU@ng20NECoy9`80XW&|*l
z+$nQvey#LO`<4Eizr$XQ8Hh{m@&nv-??X*WXy3jBWmy$8q`@`zEi53NqAC{V=WElH
zlo5^a7;+>N0PqR2s)%_KDWNwicuwn+uP-YHzDb~HAgJ-^#D)WTsPWin@@(4FeZ;p#
z6Pa!AC+iS1q;MCbzSnZXksCTJm&=!<`wP%DL%K!_^VWPsdIYT?#<z`9>i!xT9v*#^
zT0Q&WkpKh!J9Jn*qE^>Vls2zTg(xLH1^&A1vG2$cPn@vWu{fr|@258&W(510sBNoD
zzBYdgT(|c4*|U)r*WeLSI@{rc1?l2)xbS8XdU6k%2lp~f2D4mM*u=wQeEM{4OUuQ@
z%>sM(ChI-U)d9}dP9}FLk}MF5{CCmspO<$;A}*<X)acU};!DDB9AX7K<o)<h@P5SP
zftD71XpYPmy;pB?$DcqN@~I)|VV0RV=I$;rrt}LpTCDLPC2_aA?HS64ii-Nc-#BsL
zdmR=Y9)A70v&)w3=g$RNu2gh8o1@$4%9W$r!%jU*c}42}0rwT+kqkU%J`I$1fw*u1
zPy2pk<ez!p0{UX3x^sGZXgq_YOd+0@o*uj=ZY$)fVw7lcyT}oKooCv$9Dc_7rASiz
zxBru*#44vE>G^?SBRM(f#6DCJGmP=E=g*_T29gqZ(lS*EQa3-)RaL!#=DW}j$XEp;
zKn5e3Y9S2CIk&Pt60L@Qg$I;<nFLRE@%DDSoLlpu+nEH=A(}@{A&=D`9~gjNFkb!r
zeQ*}$_UU}uxiXV0`-6^=5$=yULW#ZWf))4{-v?Iqp)l1rwDnzi`OrCUWuy(R>*UCr
zNyW@HOy}$8w~T+cFgIu4um8Q8xTyaslBMT}WkH3;tL^z&=UqWv;E^jTQu8GOBOo~}
z@1|M5J|sAp2B?4Cx^~CA``e#{F1gr?N?(Ir7s6Lf>`$b757kdrm$?gXyLGantZB_x
zG=us1mWkB<a+S1)rF6vSk2kj}jxf*!AzXx*ac$d%{bykCm>BsQ?Vu_zFE8oQkfYd0
zL0KDVmZ=3Kg_|~Q;y7UZ9dgHS-@+Jr@szM@GR+P%*0;B}w|=w)<t*Xp&J*kqKej9-
z!NC_iey^jY<%>vAP3|I>uT|E=oY-9Itt^Nbu6f@u^MzaadQCGgbWn{>o#@?N*w3gn
zl}p@0%fT>R$@OImOrJ249}0=TVaA!&eA6Tre;3o-Yjwtt^UVg)?~Fq<XpySf*lq09
zw<W1nd^&||`INN3(-3E)@hdOb(X^d6m*V3)F3c_|E?&ECj5A~iCL)7_FJ8QO^4S-S
z`4FPD$IVAZvKMT1Q&+!PS}N~1e(X2lSp>2KG!|epU`5O1=am%D=sEZIN+{*}DZDND
zkS6D&10QYIC3&Acur0~Vypxxy#J(5>9mmm+^}9%xKp(~;2(_c;JAzf7X_M2ZhqhOL
z6}@TERP3N0S!-h{l=aQL@m+Rw+6yO-_S?5v7Di8E9ara8&k}w66$M$V%?V+o<)c<r
zpg3ydI-We)VN&XcZ8(xE8hz9+n_eza&uFb^B8Eos96rYR<>tw!@RLRnk4~m-O^EB-
zQjcF9I1JV?u;=mUjarFgo*TqlGf!BpWhZ&~&{1(Y|E=jDGHl$C(b2@xvrU#IDJlEJ
z#Ty9sXWoy-COu`6c157H0X9QJKnDRkTrp}9LenFfZEn9U-!nZPpLs>W%I<J$G|H78
z?x{xk#*NX3g{$7yWnTKL+OJ8!Kx;ahAe3hBpE{$XWAR2GadCNRY0A4;|CN>Ck>HRJ
zZ)x?_T_70ZxHesPU+H}R<9qjrd+>63O=+Oz!wa2vGZiffyG7ikUHI>)c)$6abz-NP
z&RP^!F^;aTE<rmPMAk!fAoA*}Dl3lhgr@<ZA?u$N(ejyRJ`iEjs`sBcZUOSCH_~nN
z^c|J=JQwQK3Kzus&?4a~!(6y0w?4JW{2E;JtQy7JEk!vCh{8hK!rBY#P%7+w2#KzR
znbWq2-7K_CtnRvr4t(?0E;zI+MqaBu$+r=a{AD*$+g<2q{LXZ6+hHqpBppCr?O|0j
zxBuF+9m#@Lx{NXrkEsE!hvs&o2ep1ESRsDIkHyqxczpc%;~P*+GRDs;LE%@?o8M`%
zM)4Z_*AgtYA-acG)~uZ?h<{wx*w|=5&IYSf=J9$z{@3y2Z_AnG)oGhE%y37qNH^TQ
zdv}njI!rpPUa8T$Vc<uc$@rQzYd|*PV8?V!%pU?+6I<D<^Sp-M^(O95emm<p>x~ny
z&cXvhKX6$uC@wCp*PzMJ2tyW>i6I($cri|~B8kt8)-cCPHcH=LEg%o9UvXxTjt>VC
zV>w*ATc^wZomdw_yWeKmZu?Wy;AQTu&p>SUbUwc{4KQUUKTo4DcdEN&ecT80_{CN)
z>Dy~uF{I%Kf;a;78~rP9y9Nf}g%6WBRN&j&v<QD2bIuTf2}6<Dsuh}Fy5MP(>Oo`l
z@;)g+zfq0JD1_`uL<FM&f=&*x{5H3Sh9_<Z;|4{^K)CRM9df@4ZsD0Wex}~H6&&ri
zi@~?&=H_691eX&eiOD(l8R+S0`Eq<hW=)VzE?YzGGd0Fjj_q}WjS!5L^V44)y7{zr
z@W^<}UT=kpscM~tW~XDk_gH2HPQxb-GP+F)cs0=&C+kMuyO(kJEApZDWSY)IM+b8-
z%h}A_`dniFbm_r%r?irG4n~@S-N#`XmP@^w{0&~kg9T8PcKiddTj<zcFf}!WUg!F|
z1me0!$|g)SkM^ACdW%o0)p%@chW7&UmzGkpZR=$?eO+88zde?uBln2N`qy>&&0-CZ
z&xi%0Z5X#jB(ls*@|(Nh-M4Xuu&Tn0owugs5Q_0}ciFI9l3p~umPZB_w_bRX0O6cF
z_rSpl!i~06Q$nWo!E$%=;wnuNU}nR)@i)liQyLm)l64<TIz5+7LwjI*ftx_Y5RB}m
z?ddC+c3Dd*m2$hkG0RzuDQO&Hu`@q>;>Wkobzv?m8`g}eCFxKbHoi<wjz6$-*RCN3
zDZz)8ef>Xwo;Y!$X*zZG;?SC&p^u}?)g2tG!GK^?EZ-9-7+kb4oC@Xx#%ulyLrFN3
z=cnqm0Lqn=l@A!^)nswpt?Uz16xzG@(m|P48?l`1Y!(IbxTOgcNdmZy2}@4HM@INI
z-5D0HSU0mZB%Z;J`SqtyDpKD{a&jU&p6Fm9PrVo`=$yK-9$AF1TKMCp<{=lt!aLgf
z*FW@eNPPGGGG!$thnDQ2lEra0`m4E)?PS`}|E)E{>`KM#Z7@c3bacTiF`Bc;Y^27n
zHI7Vkh!{ReeUDmtRv%c{Y{k5nFTXl|mVKkIq1+Gl!P{TqI?kGJkt7V1x?K6?i{Vrm
zvtjyy@QQ6mFY^<{@8O1CSy{=Qe#XGS2SX1Qq#z%NQPt$NM6*a?OdVXhE?qjP;o2Tv
zlzZ!_vy6|)>WxFoUHj1q&32b+*-pP%BO>K$aIi4O{nTGNdM7gSM{Vu##%<NBojx0x
zm>`X^=KN?Xilhq4l}8!CY<Oik{d6XdV73X4d03AD;^6|Y#DZB?=j>?cozPNKv#2n2
zITBq{Qv+Tc$p#h;(rtJ}-8zY5qVfCp_ubCh>lGgDhqwe9%VgbQ*Gtp>^f0MCbqZ4t
zBawFdFBfPh-@JbPiS&pocsEJOEVZ`1o9SCG(Bi-$(1wl3l<zVCngb?3?McN62@0tV
zfP-1h_M)7qHK2`aFoJ=@yY&HeMb_-P)vGUO`4b%YuJ<>&ZQ$4eBo|_)Pv(;b+#~$_
z{E5$=HQxs^vb3@?F7<D}_=app(bM~P3IaQRhmmkDahMhnTW_6&$dtp)8YEKfn|ga-
zoLUaV4uN1~h4oA>)u&WcXkpU0Z{MNS;i)qsd%*2HdUVla2uT>gG<M6alP-N<wzMh&
zSc*iMps>JlH!coZMC1kDj8E~@j>-7{>#4Qv2;@B`CJ)*Kpcc@G!2X30sTyq=(1vhs
z-hA)&ZRf&Qy_@NkAj1I*f(lej8xIT1j;P9ryLY!Yx-rwzJ~(Lc5kBMC$(TCbZ2w$O
zOG_&}G}J}C3wN}>t`0RCBt58pJ#KBBK8|pQ&4;81{Wj3`0FH2mjOXu<s_gCQi6%5@
ztG|3nm`$s-8H`+WIUmlEcjP<(n~{gIhh8<>o1t8M@oQl3?|(;#51;2QPPlm&cR(J-
zfA=`&*72!f*Ph~Y!aQH6IAxSl8@L$POEpsYGL(zD+MNWjbbqpOjs+YXhL;2dEiEmf
zL^h?5INV+!&d>jz5F&{9IP~%$*6kP-X5hI-_`b^s;+d2F3&(iHKsHH<AldWiraQR$
zC#nGEW2k1XzPhMURr%+Sb5tn({PU8dqFXH*;-~(@TSuvxW09LQQG%)o6i?<xeK|_}
zN|;QJbzZHmYV7WvG#bri=H1D-VS9*K5%<47foh6Mad0)7Q&IU+v3maN(|A#t{O6Bl
z8+HV)a_Ur66Khnu{@>n8{1@-VTYa52P<gyWgl;tU*}rR-#q@APN7y>kb8~%BCSg3X
z`sdH@B?~<t;A)C>P{h5tEG9iI&9Sq@4-7Lr1`i!NBrfi@vIMK&w$@8)+A;~v=p&in
z*KoYRfK7@A_Mhlt`&?N$GCI2b=;fcl?YzX&C7y7}xl3yqY4|5ZcR;#|rXUT02pn#l
z?85Aoi@!>V@ivfxy8Fh)$I%J@=f04hvG>rk^I2A2(ozA^=W!TeY=UX#IkfhG6GT>%
zk|Kq(xz&j%g(Ke3*dW!oI@lCYg__K7a&nMf7bGMEv9NC5tPAo3ufP79SQtrHdsg>$
zBb(_lQ(9TZ>li1Ds*IO3eL!~)qOlz4AT(u!Y}^KK|13<k19=#0_5jz0U+7JKjmT5a
zTEHiLFh8GR`nJSxQ4WJph+?1Fm43-mlt)Jz)4}O<FtK!LqV<dX$DW>^)-z52H#X8m
zw3mkSh)UPnNsL-s0#E~$>m|(+U<OhAb}3f~xnG^nLMncPw3&#$#^Y`C=jUfQ+Cs(E
zx@K)#COtt<4|%D+eoN3g25b&mK43Y3fQ`(|P7V$vW0^}$ljpKs-rpBfz^Y{?zrMY)
z!ATEg*+f*GcHSWeq_}da0nntJN%fTJKvN8m<Du)n*D`MApXf6CpVSxGi;SIDmf`&8
zr|PYfht1y66eb5$%M+9(6Uss(+FE^tL<6Lajf}X-GBRqIm3?5CSQReX*%h;~m&Lq5
zeo<0lyn_YFD9lfj+y)P0iP*Q+{z0CIGla4HXJ6k&?Uz*4bS_<Q%X93Tf!5Ey(!U_a
zRlNGSD@$GqDw>*MMqVo9>$>F5wVWb5iJd!X`4D;;%QLoKt+h#<bczv4Uq|Ja<@uW&
z6H?z;N?HDenho~|S`8FF<z4C>{Zesjo1U)jiSV^^Qn*G{lwZNT`E&(t0Hb%%nZwk%
z#<(OSBLlkxCOk;m|9BRmPS^(L-A}w!>%V_b)+wc|EF$Ma6_=2k{;q>^1xTPm_c6GO
zSs+iA_0Gm^=QJK38KL6~^Ep<-DrLF8tPvH&@Z|LL&U_bZb@i_aN}=Qp6;<%zbL=R}
zcj}aLuN5xX{NeS5N~^CAx*M>hE!PHiwR?0fPn0u*JcXWKHOCjf{y@oG&)CWNwPjS8
zo`YIJMMFPy{Cp?Sddnw3rB%M&bAl1(C$qPF`tIui%|T1SdD5nCMV>M#4=xzhQ<z?L
zqK3-rJ8`_&n1Zs21-nRDfD{yAxFlLW)Jl;Ih`HpUwetA!z+6RYPWQp}9~I(j(hT`o
z?+`2bG!t%Ew1;QLiFSjtXJ>v!XgyC%^ps`~=)OAhXE#6pORU;myA=0q{`mgAFf6TG
zvTd!h=2O#4{<sGq0O=%Y`oYZ@<xB9;xNrKU{>vmAqG#;&4D;HcC=?PFp2Azp8T^8N
zw6{CxX)bnTGRw4Wy<K)?1NvgoiY2gnci(40B5ulEn5ta!*fA=;pKmwt{`~Ek(^oqG
z&54$;=@Y$a*C_goSzcRyzKgY%<BXffz%HC|PMBD1?JVNpDsw&&<%k*O?tR((@a|_g
zxW<#KAD{63Kj%=|+IJ-rsPoEE8-kFaf|wo#PeaFtqc~^8f;u1xrzNB$YzhME4@^C5
zdfwl_<pIdBm-ta#Q5A^TL>67gzMR>&g^yoM?EC5`vU*NX;q2bB<$~Ue#u2z=0q-rl
z4Gjzo+^V4(_r)UvwG8;_-iJ>KzCAc-^AmEM<9jxvO^X13ka=#D3I1aW*H!r6<uaR!
z0-ce5BItI*eSWh)$xa*HnFl_s>r%xtqh6)0nYF<$Vj-w$X^~!zdI@Ve1$aEW0!j2d
z=n<Na(}|s()DqM_MOjQfYLQ@j8pbQHNvo=<Ib&=?<#C80D0z8#DMU&Nyfo^z3JDk7
z+4nJK4dd$@Y=Y=xpFoozC?q4De9*`l88w}q1lB;uM&`0zq8oUq%5raQKYB=3wz#Nh
z30>w-48CA8D;Y{LT54^e>D!=nWE>5{-qPBnK!#Gx??p|QtQ)PsI=p1Qf160u;#_Yy
z;Uj%4pnEq^pt$&%XWE#c`;h4t8H|6tf_wl>C%lGdu#k8HcyLW|x#l1KmR}yyLg^5!
z6jZ`st)Ua#!2;+jvvcRW)>@ot&($AZ#?Mo4VDg0pCT@wXv$M0EUFrw(?%$pa93o#o
zeVQQGW);>mQGoG^2orc%K*Z=zf>91glB>X*(qdHAB-H4d1c@oXGS`c-_1Fr@PoM4*
z5I{_~#W5Nt)Ve0XT6S?X3(;j3u{kv*1+WZ1fYGaLS~4j742;!Kg**ifrSVw8Z@NJb
zQ^Sx}(9_Z755)TP6V-8Axj`H9sGT`Oe!DaN3slrlfBo;F#D9k*{smA_fBWCT%#(1e
z6mpm!Q0Aw;%Yny4&EH=?>e7!mFooFJ!L-q_(mWHEwM^HtP|b9zIHYUeUU0v>dkGIe
zR@TA5V4K4*W{bYQdZlv(N0vJ7zJ*0^mm&0KrpFLOsRrMm^hr`lF8##UW}L*RpkZb$
zI<u6NKHw-qbd&RgcU1lJn4>OM6Vw0<vc!t7zyC^?374-f7*z7Shu~@zU_`F@z1nHr
z5@qcDxit{C(37gO74U2B0ZM4ynH4;+%Ls<!+ac?!Xecp2qe_a2k<IzU(rN5J6kO9G
zjgW$4(GEc+gjSO*t6PR5PPVpNzm5d`0*;B2l9Kw7W;8N(Z2jLb`9wGZY4wK`0Ha>i
z!35Cv#DqkmQ}6KWZD*=_f1_c0eig51x6-472`LJyh2mVd!e#z^LY->7U_``8iJ_~d
znaHw6UjdxmMpYvOyq10<{c`Q{hIXw&t3!xbnT*3)S!l6K{}v;S);Ui}PN3SAG8T~Y
z9yqf08|K}%`bC+mkjK}viz;SQ2)c5m5y9JS<g7j!n>(&=F58dd*Un%WtWoMqm!S;6
zdA(^9-)8y^Jxk-E#I0N8Jx0MUTz+M+T?Kb<fe!dNpzz1b^Yt|~$7s*vasg~c6Z#nr
z6p4EMu~Z;MT8|z*TEDmCII;-<zyRZS=9!3^%!9jS5>ivuFoA<2gK}3={Xxchiq7<l
zoSd1mKx)<-mK{Z2X?EVj<3)qa#2oef05w6T-kp%PQI+jSz9I4)rVW_FFw8TOj)yx`
z_Y4MhX)j(l%XlLcqa$83P7YR;@S3%<w3NZ!@AO|pY1i##TWxJ7ehfZ!6(2vswHkAo
zDtLTGfdq5hFl_i^w==Fsk=EA1;TeNp>oXywH$rD$s%Q~fXitaUTvpca!{1u~L>zO<
z@NnOCJD<G8r9?}Pqd#@(Z{2+u{05YOiX`c@9kaeObp*wvOGlz15MouB!4q^JZ3+ko
zK+MTwGfsc>T~kMg%^`DnaaNjV8LM-s7IbS443aM&iAE7`3p=}JtjtkJE{HvwA=0B`
z6Y!nyi^|rtwjRLJj(zlKAM1^TgaiPIOLG%ug%Mc;wKiha@)R`QzX0h(OZ(~TSKsmc
z0SH=R*0)uM-)$-JD{Qupm2`?dCtPBryPPkLtU8ar*bbB+0(2hyDnK8?L<6}a?uYwi
zzedvi_IYC>gU4?QOexqF`RSbdco-q<m?YRn!o&r;$pgf^8c}5-#>F9g&bhGjGipEq
zu#kU6^nZN_t~cqviQn{@oE97w+(KlqwZDqltf*BS`SE*lGV4|UJA4)`A5=t6#nX}X
zp3gNNgPAa!O*1x%svcld2-{8^RY+GN^N%YUXxO!#hrZ_%`)i+@esE_A8<2<j2AR}+
zW<SNxs_+R1hnh(fSq4X;mz9l;o0&u*If6+}od_9Zi9rYj4p0;;V#y3sf1CLJzS_$d
z-WGH6Y5Y}b_fhBemG8(*5S+QIKVv~)%jgl!-+o<vg=NF0AmP|H=+B0@!6;4!$F@IJ
z(!Xk^y3kC%e6$)6V9-ljTU!n`y?Dq=y}XojxsFDsPSNl0%rK)i$d8ZblaYyL*~uPi
zf{1~m&>u6@L46vlt24e@Lab<wU4rQge=tI7VJP{J*(Jz%iMzYi00%<>+%JE=yL)A2
zC3^{5MfRj-#6`GBMXjl<sxmJ2o*r*6H1VJBd+}nAcdXZB|MWnCSRmN~46Q4`KS*E7
z%4{j9W1kFF*~ewz^A$cJqd@q(hTBoP)OP?}wb($S!)xGbtXVSV9Vnh+bs!{@bz2%5
zE_lQOB4f^FeZBnhR6R7ASf$IArArLOFBIYjwIVv0xSt=V#WbRj-qYc2M(M0`Ad)2b
zh~eZUa)LPiqs)a<P_aw6ii_qk$j|L$7_KH(4)a;Uv~WBd8GBhx@ai_V=kP$;o~Vkr
zeft<Kp{k~aJR*Dluff45Mvq_|BWmy|Ojgc)^nt?#I61N9;3#g1l-B6_X62%<&$T~y
zq)OJMpK+}6+ow+%Bl8}XtwJkpF4ZVT&HQC#)DDyWp?K8!XXkN)(1nb^4gphIui118
zd1KfV5C$v^n%%uWKqTbATuk8JXD_|G$`Cg2P|mNwXenH7VUKL0?!M+O**~XYlrmdN
ziDB6p@frNZzps3Y(Ak>s70ma@>4PQ({yM=dXdZ_zZz;#-<?Zm#9t#&in09)*9grnA
z<7jMsU?DQq=vHzUHeW<J8nY!nek32eJ0a8rDPJ952_T|<Hz;8`KCA)v#m^8Oo0BP@
z+@kPUzMBIMR#GT=8zWvSB{&s6VO1g<phU1-#@T|^HsiT(AKYsJd&HWbCp>)$4^uU1
zdql@Ee`8R^jxE_Y#=PBL<!TI{hXyR$Bx)MqtDeJxs0cHH1@TxslpFONVTW72WI7Q4
z%sD6+B%@_QYH3O)x|F?Evo}<rLme7q+&p9<f^IAElji%`xiZ$-6u&{&+`K0#&i@qc
z5R6wwv#M>(I!idP&IE$dSc|zD#krfnFf(pi-B|DXQ{h%oEx3u6rh#Z^fL<xPv=-u~
zP$N}KOMbrhHe%yVndRd(0KlblJq#^<<6K5R&`T^k+vXCc93^~Sq0qo7cOmL2<S!55
zntDGvdg(oz)@?%b;9#{IbSlxH6s$yx9ki<-NjPx%*}pYczEu54yi!EcO3+g}E!Cr1
zcGl9;vS~}*9b#QDOMhHwozKM6#KpbASFh&3=uIvhsqHSmcI)IWzMtS#+-0NyHQ+yQ
z?qcG%XpH^45Y!=!2Yf(*M2GA%CKS!ws|*`Aw)dB%g@4~vE{c;GP-7nVfd@lhu#=k4
zc=p^79Hj47ixhq+YW5C+3@85iD{pIv)K9h)2YvlV^<#NR0>6c!zxc6Ww5VB7{blb`
zkb|z-3UTQpRT)98Lz$|wLcq9b!IjIM@-OWd#ymf5k~Zpe^JDhEH+A@O0VhwTjz{BD
zRKOKfP~ogs8b#B~(8<%MH!XtzI)G`x+@{rG&k^A!jxNQfBG&WK6IX*u@;rYhSXg8B
zr&sq?j#2t>Zt4ChT?npPa~!16rUX47<lmv80)m3PGM>&jlTSVSgNPp`Vze`umL>>z
zMsB9bf@B!hQd9ffjJERAY-CJAe1;UL$A2wRsFQgk7ZmW{e-BT(fpl!U8QkX|!Su*@
zcXl?RmaojSv}ed=Pa|Cs!?J24ckLn+jV=dc<5=oo#bYux@3V=EtEHx9m1pzi@Fqj^
zA%_;R1#-b%p4_kL5kOFuXDe=Q&ALz-3F$U`5FkZNiuJw-00eTtDBI?a#d33&H57Kg
zI!h^-0QAf@7Isl^U3pDhs6@;o=KDA#uUxvrRolD<Ce(3B<`oUCRfh)4-O%WOuRH7z
z!E`*0O>b7HkaD_C&R5F0eIz>U;w2cpH;Yd!%oIN<p3<y1fIB(|ij0cNajU$)4EO0G
zSoxw4nr8$&IShP0q)og>#c82F4z2Of!6P&4Bp`;zMBeE}lhVm5X$&yuEo6JZ9e8X4
zhRFVyeFv5ppd`+3rkR>q<`^yL2jb6XL5PJ6mG9oWcfwF)|4Cl4n%c(ky!Pa#H$Mgg
zoWo5TlS{;H(=)hBjR)QeJX3WFmzTM6g~=%&(Z|QoyU~5<eTV-6#4`*?a3sP#@U1z6
zNcfs}wR(W^JDlJThVNcMTdo~yXx5E97Jh&xWQg-b?*xc?dYZMZPwjv0M5=?=7l}Z1
zO--oje`CuaOR`W<v9kk)Ss9jMeKg&kcd6XA!XBzoTid<)-^A>b{!C8BM4}NRkfrC>
zFJ4}PSDvNOEd_DuQ7;DK<YQ!s{MspgW{IKOr@DqtUP4&ZY>i`CRm3tB$`~W8z?V99
z4N39NiIDetT))KE3mO?dOAv_LBpr*4ULU7z#`i=<HgL|^z(9?5D=CrrW6D@wlU$S|
zfcWs>!Q9uw)kk74&Jd%`Q|rcbT?5nhC!I}gZqIU9o+T(A*4Bc2mUW4$Y1QQmcoN&$
zdNCuTz^6G;HDZv)mb}Fs63NomSzxz@B8L<MH$1+lnV6oMI_Z4f?~wnJr-F)(PDH(v
z2I}d~&Z2SECT}io*+0}e8&rIrBa{|*obl3XDBMB-?%FkqJc=nn0SpVG)S?8FXdn&A
z_WW&bo;AQ8I`agApu*#G-o^evsIDG%p&#BwhC_75a(QX)lKZj)a7^J~Xms@ULd6{G
z9l=W@r7J>X-M#}c9yWqUAs9tBdPDGSsG5_8ZtLK{0`AQyKBc2m56I_09C}F0N0~OE
zh#&jsG3cH<$Jri0o;Nww0-PaUz+sqK+qH$qzHSvI{T&(#;by@5mTlx&-QHqao1m1W
z8S`4!5+D(ZmWBohXV^syRTvhrE%rbEYtJC<CS0&d@ENe@-e!>QBJ_FnKeUEiFcaqw
zK$R8#5`<*Fp)rohzy5OWez)$8{<pr%x%SNf;((A=i!<PlCt$b(h8Q5+7vJjep+lfz
z(v1ppkSF5lz+J21BFuTb;K|@Kp0B*`zR3C4Kfz6-M%Ucj>{{^>oFKHrz~mI1ePXc?
z9L_8et|3sh$vJl&gXkEvBo-2$7Lu-A;W8wXidrB^LxurGwQ)<k*IX#>ASiyV%^%Vr
ztOstz3?4UpCPZt<>j8$hjCJ@<yxm}HrE=x>s9OvIhm<T`RAr2WLrc8kH7tJeVE9E@
z`Gq<nG*e9kQeUJ~d4`U>GZMVCxXZ``RiD*CM~48L$sp#9DUQvk0^`7ki`Br_hgo1(
zaDGUOh@`!IX^)u#tGGb2!_(sDr#spH03i@w@?;JrAT6Cxl@;Mc{lcA0o4lt6TbGSB
zR8@~+Fi*KC0(|i8Xd-%h>#Q7kusU^L&PUgJHg}4wQ-_Q`MGJvy>@1M-XPCd_2<C1J
zT!OJe<~0%~i={Kr(H(W}3Xt|FzEH`&5J{%e0ue`JhXI$0VRPXFX|tML+WD-H#ntf#
zG`_fJ5P+gNg6N@PVqkD-slv!_Ilt5A&l8`2B+1T@&P+z4T(?5bcD%k2+Ro?xJ0B6G
zpkh*#QISn=HQ}_cT%?@VE?pjTqT>U?j3t3Rw}BV-y#1TjH}d78aD?#Ko~<DN1q;0C
z%y`3W(3h`Yv+aDb_TsMaO;J%5OI-MJ4<QYQ*li|TC_*;hII)XR4kPmxz$KrtmN!T_
zH<1`v(!MFf`~!XKEPzNa;5aC<u+UIcfAO4w9om4?K$@45$KHXW35-=kt$1`#C;XQC
z1_np~?a}V02afy1PPD^)$5OgHSQGCuP=(@RbL>()>HvuJY^<y_Yl-7HsSQP}nTwHF
zqTB~=*c=Hb^#7qBfGqKKOWns%wV3pFTLc@pUqr+>)BFQkIB3=$S5f(dh2f4_?>@7R
zcT_C(ny)k##~?82pf3umBE5K~1?e8tYWo@K>5(He{?!^<rK3?``gX0t^?KU@<AZj@
zkC&*gA}b2x65l5(iVQVuo8&l-$YOjfAoAG0*<qjKv@jjOdV!jJDia0RfbbsCT$5lr
z{An6g2)MUsmV{cHrp>dOC}Ow%_z`*M&Ra)EVgO^!u+Bk!Lh0waH&D8%6}{C+tBP>3
z2-To|N&UCqHImz-@<H_(?0f!|EDy3|P65QA<HPK<8Fyi7N5mtD-Q)s+A}5SixvP+5
zKoFE5$&!bSn74G~nkhh5^q?W{ziG6dOgNNDvWbfGIdpsok_JXb%eXKv@}2#Uosf2K
z+$KSIO0(cHR!AsECF;A3<1hNirJhIPb||gpT5M$@i|fY5<PblOZQDl1&f$wLFp#HD
z-fjsAce2!EUETlN6ID2wkjg&fIxhGh&$%P)!ZC16sM9$#zrN7Hgr2D~>I`~W{5)Q$
zOfs()DRKS(sY>}@6g2<;-+Jg@-{${RQVpPN*eObwz8;}<A>XB{q<JDu!8GW90KV$x
ARR910

diff --git a/02-tipos-de-estudio_files/figure-html/grafica-pcr-1.png b/02-tipos-de-estudio_files/figure-html/grafica-pcr-1.png
index 765554918c0c00b5eccdaa8ba4875b7af969cafd..13adf30782dd8fec3aceeb24f76ff461abdf5aa0 100644
GIT binary patch
literal 36149
zcmd3OXHZmI5atC`l#EIcC4+!u$vFrJDj+!xS;=w883Y9ZK{7}b$w-bvmK+2mXAqeo
zO3r!D_;zdmZPixokG1MO;W6CXr%#{m{=WXs`}kVvIX(^r4g^8?3i2|l5QOy@f-uc)
zV}k$k*b}P}{B_4cUdIW7D1*`eU__#J-$D=#q#z@u?w;~@&cjH3Y)*V<Wj^WE<Bd0l
zG`QyxG=Jn&W*u&3NzYjnj#XFdZVVgRYJB&REiAGnF*LL-RJWR2wI0*TstTo}xsCax
z{Mqf7M`B5@o^CX*9CxKA;ps3XCH?G2o?f^#B6yD5C6^adwlAp^Ze!4aKcNgl(^xd<
zFMPsbGT;w!SGO$q>%Qav!*4q+2#pfq2v|V@W=LdY0t3r*+Z_qoppiA+q*vvSlFQ~7
z9iYz)EaKwg)21J9M|Lgky8kqDs>Hra9GY1>X9vlB8?p%RHw@%Rcp05lTU%>E$Urx1
zSCh#MnVclxb)-#R=zsGg2yR2cF}RXmaX&z)YgA3ykcKfn_Irp!Ys3E#`Zh#f7}bY@
zPi-D)e}Yj(K0m+n`0?XTE2NdZy#P6083p1ILGZ}hjnDD%q%qBJm>=&dKOrV2R%z#k
zgTIJnAJLQj|L|=J_cT^U0FmOS^{<yZ?nq0XtjYrm!r+jQkh7L<z60iT+oOhK0#Q3i
z`#54ZRQAI@XXsG|_-UR%IP6(?-3sgAb>kgc@9QGZ@kT6JZkV7%!4lZ}%=yxIYx%O8
z`UTHteR_N0Iv;ojhLya;wR5S&?v6#FS8h*^_;7cyi#nr`(qGh9brxUXLCIzH;=_GN
zQbWwXaZ%rj0X*o^7)c@g$o1_EDfId6QYPbP2B@=;^yMi%6mll~cArUwk{UehFeJ*l
z`*qVKM#4RRQJ<XHm<=&&E^;!%!Vb>C&co80=Z--j0!Q!371KVOLg<TdrSJ&sG>BS-
zi@p_xCO4^+oWVoZ?$S^EqM>*wBl`ei;nAbwDIU@Klwgpr=**%g@MAHg^#r|cEwJvu
zx|Nrkl4;@<8(P^7#Cu3f$jDyH)tf&Niykc9hzpL7zkby?9bkMGZpXQEj8d((2`V5M
zE(O2OEC_x-pzUjXDnyf=^@1L^zEOsQf{`rpLp9m2E3rUfQLyNBT>1&33@X-uOz3l`
z;xDV}I@IRSO|TbKS^r$;09*$hcGGT3QVEnL^r3exRvlU&qR%Wg<C2pDOZ@@dhB>;F
zJsluwCv89R0B=uc1=F9XcZDsJiH>fVSdL@P2_d6#6JlZ4fAMnrQnxK>#fmj@y?ArS
z{g$aa`ncYZ<7iatQ?gK|dC~Fhyg~cL;O8!7Oq%KGQ7kWdaY0}v4gr0_{KtD;l~@Q1
zb*_nr1ESAGUc35GeWgY3uOByjEmhZ7RpfK%SsDtHMmZ^>T?PVSmN`c8<KHrVks}V2
z)Z?-I+4@fD-+2ZMSqR#aLU{#s=V7-a!hy5ib#`alyVhISVbZ)^q2u7_ChMsf7$u11
zbd$3(LT@N_IIJ#5Gw#`KNcF1ww;?0hr|72@P{>|PW)r_^AbK-0fE8czm?sF3Q8zE$
zlNfq<%&mjhwP1Q6xI)F{IP~xFjZFvKe?yxPY23Ls9k9oTK9?#7|9<UIwm9X&xAApv
zf5M&|M8r3iS@-0E9{t!WBjhk&RrM|Cr&*85+O{?Rh`AMZZ(2#K$P@JbamsOA-W?~M
zoWbZ8^1yOOcz<!TRhip=D;SK<gqL02*@n`|e=xIV17tW@n6!HPgR85xyH-d)hJ<8(
zjC$EDY_)>T+jmnk&Gn0G#@4y8s3~}ACWbL`nA7hiJBx_9(Y8@yFFLJe;fBM`7%A2(
zu~=BWjJ9b$F@GK&#>jiiRJ%Y2)^Q)KBO$K@3sQ(4=D|K8jb>0zB7mgxu#Z02zCn9X
zIUMGhU67ei>>nFX@XfKso?ba=*{AvPU#?bZ>&{lo%*QCek(^<7Y(a7_8S`Vkeo4@_
zNig^ueHhZo5zf8!1-IpOu)e1rKP5-*#y~Bbs*zL#9_ZB1OB4<hW~6vn8AxP~fVyLH
ztgzPud}y0UKz!ii(I+TbzK20s;Q`|V=irBu(NG@t0|a<(W?<CITwyvbPq`3#M#fxu
zdR&Z<<;Ql8iXbdWz#~zH+A`fyxB=E-!Gc4;3HNVeZHseRJR;~~EomsXo#8JW6L$SO
znZ#qxp<g#@#C2nTnzQM|WOKR-g+j5jvwL`W++$K|xH{Y5;d!QC@DQ(G5i$x*xEmww
z$J3U@(tGBI^_q~1>t{W~fdGh;Oy40dWI;)ZVVOxsnQ0fK%@*%-ezZALV}HCgcf8=|
z_u|Ei?GpV)&qF=?#+xk`yk>9T#>K@^O88#HD2wZW-^x;mxn+8Qwd@)ODhCTao*G#8
zJ31C2*-728d)W??#!9#QX6}Oh`h3t-dNQh}Oex}Ovp$;dce!DBeRbS?y{7E<gpF;s
z-fg=;qtMfDaj`9U!S~!Y@Q1^MzFUz_#WJVgX&+epQ@1Tx?#L;40!!#q!6j<)q0!s4
z5MIN!H&ZmymgFn5RZrAT0~7H7O~r|*jo&{4y^j7WC@J~=4!|xiFNea(x&LT2pT+$!
znk+SbrmwHhS#r74M);m+!qD%sgUs*|85uSW3(H!r#HFQ)iHU}WhKI)~jSnH#u!=}}
zJXP^=ER*6m!RWK|w!<_K0QVX6pBESD)nF!ydAa_kFtM}SLiWVGCz`3Y-9S|_DW~1w
zGVKg+i7~4{WMn}8^YjUt-8bdejKg5VCq46;#KD9?8)D#4Fw55!ql7&VnEiDs%v^*e
ztjE7MdL3ICZuUeozD!XvS5Z;<tw1ZIp+VwbSy|~o$e$wfvci={r<Dv59i0{RKJy~W
z!tQPrpQ@5w?R*l5u(Pa<ZVersJCb^;cv-3Wd3md-D)Uqq&0HRW;H473OKb4Q(Q!AH
zMXgnM*rCFco=lb9*jvlv&ztNE2I%mYFBUpY4ABg`Ag!#ev3?!)v?Gykpv}!qsir}D
zzw;T>u86VGQRYzi*;?*`$4Y86!%Ydi<o?G&x3A8O+(oH2CC{luxcPZ4SerIG1&5|7
zij!a5>~MsjiGaqmq>8q~>bc>o$f}x}zVc$S670_F6)BCe@tsS5XIZayhU#V!2(&IM
zgd=!~1qIR}+Lw3g=Ik~zii#Aqn%~^LcQ5d-u}s_NV)}%mgeSCLot_7_gWRi>%hP>a
zHf6(~ta#orJvyqcqB7*N{GAN^_rFoBJN@-(yjB&kXk>UKnUr6$PhT&0XEI~%GhxS{
zW8;s*iLckr{n$I7XlMjpI|<(JIC8^oS=v3hzPuM>!uR@YZ8IT!%Ek<tnYOx0jjeCF
zujg9dIO1XJTR7MR^JUfi?v9Gn8vm~2wBIj~F6j7YqWS7D0IQCJHbJ)t3KEhcZ~YSb
zMu&y!rsSSGa_`=vw>dhI#ub|*o<4p0B{|uaUoUrmI_bbyyg%XdXNP{iiP5n!5L#Pv
zZVRcR9`Mhc;W9;y{bnOqj%H=r!h@mpDO9=l{NE^SY#&cDx%67DF9t%<g{-I%`r7I|
z=Fl-KMFj<JIbROjojbJP`55GQA4W$<EeM(D{<_wWtc{)qKE_&xBP#hWgMw~UOg-9M
zlnENM;v>gXP%xfsk`A7po`!Ld`2#oV_di*)ii<gro1hM!E;;Y)>{MDw{_+JnD7RO(
zw%$k+^SXQYu9C#%fs&!{X649ukzOvzGB(Ryzx2zFH@5=6ju@&eh9$yvJtytimB0KM
z-cp5I5cw+ilz%T=_^|lL&?26J<#DsriMWu-dHu6j*EbjyLN~XSL_)?4pWcsgPVK59
zB{cLo)X3cj`Qa9q=fP^4fkn+sy8EAxUDSRJ*7)(7$(bujAN>i47rkLa=V@yXR5ke_
z6@5nCB4a;4w4dWasEn@~uMDIH@HJ=_x(ud?J&v)QZ}1Rn<Ps4P;o}=ybg)=t`xQYc
z^2-_)CTwgwM37bL8nUy$bwMBGiG0m4pAL>y9XJnP_-&`#Ki&flNb^{Ju!(pG@=^Oe
z<=VNe{Xq$dQaR(fIv4ASVs;jmp)0lMtUiu<yq1bV@p}6O@y)D0l;>vE1i$?}2-1><
zy<VwnyLZd!-Y-6OtEY(iX1dZuJPL0=zXvj<7RU&g4%>~#^KJ`1$pZHCRn}S{4Om*{
z_n|1bP1=u}jvHJ03_RBu6Z%|66a_v#6i^azx;&iNUTF4PX!1elZIHo0>ekm!EhyOC
z-F2JyFgGwT;B89IIGH%lx-DB=c8{j-wFYRkZU<adGNbE$0H_4LgM)?$ILf-Du^Ox+
zIyyQ$JY2|W8RmC+sB2wp(B9D@cDDNM`fQ?^-)>f0M`vny_^H{?L{K>l+~!;^t$O)R
z=qM3J{CziN*L#*S$DUXfX3t)P%)SQ>#9S4FV{~_SpY$mE4xjVcH=UYI#M}xrUTl4j
zOU4oYvzD%NfnB(=JNvd{>6~43cG>cc)6pb0%L|4SQ36RZcy|lBDY)+eoI{Z&V~q>R
zN5?+0DVb0bl@@g@7xn5}8FBR%GCn7JKjV5)_I>!?&Q6Y~-8<I|(3bDHgG`%e4_c~k
zwgt?L3?`C_GMfdShx1s+327F4wx+6U&X<2z%lew;%@5scu0Dr7Z+nB85g#w9(Z=h5
zjv`|EgaLO^he-^Y)y;5W#Gz47GsPE3zU94TyuEfqVGOlE729ORLjFy*Kq(F?EQp|9
z*MRhcF?u^oKraZYA?*!_#e?QOj8kKx$v_XA#vuM$$ac7kC)daG?7%HYQRyi;WMfx0
z?~$cmuSz8_@71MR%nk}CfKv$rLBC=Phc*%0tB#u|YmVQVlTdR+YwO>`!S98}FhHE@
zV;6=#^pItENlI8$7^o_H|6bY<$@=2bQ2V04XV0_dVawYGmJ$?>FP6$)4v%Pta|}mp
zi>glNFvF#2|14qTb-nn?{52vOY?Wt6cQ)shSWz|->DyR3uz98SmJbSKD0(E2vvE^`
zlTA>B@1|tZ$mvWB$>+BoquP*z&=^j<6DYm1q`=rp4|=lbGcrV4NuHhBeMkWLM)B3F
zRnGvn#i?D>%JgG3R9qEDjp+DHzTqPg;T;ULlZb%5*`~&25{d{>>4w)DMU)F#>c2+b
zR;Wdn4EEnLQj2Bkd%|EFkNY|@fxw{_U4S3Saio@=nj_AL5#xNkaU%NAKt-04^13Z+
zmSN-*UirKJ&FUI@`*Jh`HzdV;?_+#YMVH&ZXvb8ZTY>8N&Tp)c&t>|DE&bVKv6r`W
zE1ymh^qmUTG3|*cy$wyz$l4x4axcNVq)d>*2@!jz8P*8!^TrQykixCDKi|W}ORoLi
z(^mzquok}ev)yek^NoxH{zEtSJi6>!utz_aX$;G%zYY}_K*a;SzjV`%w!V)ddDNmY
zVy7uL_8WFzrU;5>ETa3P*Z&Ng`jyVY0qI(+K@(=?H&3IfBY#d17)1Y}QGiD8ea`?0
zCzM8xL(EYyV^fyW83`{~KmVi%-F@TX>#6&>q>f*Nr^5)^v2%L)Y+fG9Y*MPM!8htg
zdkasFgOb?xax1W|Y-+lI8Co{aVx)L?Y5Dj8-a>B==5N&wJyRwOf1=F4GJe>v{^y#B
zhn7>#aeR$;+j7h{fA@wT_|9#JAf~2&JS5%L4FA_iM}bC~Y4g%f>^guGezUnCPiVp9
z!2^5DQVrQ8A!2`B*K)&cN$|KIrpV!_2%X~gj4$x9+F36fTd4DKVi&`W^Pq2X@ecY*
zq!31fSUGtL3LLjk2C&Am8ZW}TUK=#qHDw6djR$@T61la!3wvUVCrUraq0Iir6c2Ch
zh=WIuGJ&<NF{cyS#apX|X^12y#0&e+q^5pRB?5M^)8-4$<4LN=;{Wymdy|KT1;bm(
zC>*<*bj5!5=^v-0ebClpjD?@QH!@xf*QF3cqrH0=4TK<^4*_C%evw-CE(vAj>~&OC
zg+-ohG>)@~Z4_l6m~jk$j!A_2&*0m@;N7@E`jGCb#wq|le^%kW(NN*x>^I@2fNZ*d
zTJw<0+@#h;Lo@JEN@CY*HmwEJ{!CHP^betY7OW=I6bAjYD1sK+zK2iuW%QCQEU9+t
zfe9+gATDH*QK4baU<<hKXIhC7M8BRLN=$h`H~P?sg3ce<SgBiVgoC(IS}}HuW{5`l
zB~HON8Rg>cR})u67b0FaAQc_r-hWT40kEcR^TQ1O4UKM>(Jc!HFX-nBCH@zx%r*TH
zVIuU54<tRL{{$e_t3v<fkqhwlOv6alEvrM<9bq2E$)W&HM})oY?ziuv18BNO7Lz`~
z?jZN@B7xstJeee-n@^%`rs3+_ci50NVFJ&=F&bq^0-w`*`@9bHfbHTguuZJ>NbxCd
z&VhyJ%VGH|K0t8DTe-NbZYzM@3_x2b(HBSbmrW~EDlImV=qz=KVeK6+u0Ma^;(p~T
z75xWiUg#484niYQQ#a!w38FW(7@gq4vlyOPRPFnc<$6qz3cNu*MO*lu5?DC7IUm?W
zH(_T;I}z^i{fQOFi{q`~6UWE@An^%+lJ9q9Iat0%R<(*X`2^$yk4Lfo8qI1Kjk(jJ
z^is}R-?nx>9yps9l#oU1yF)8%6BGQ~cafHh$*5?5++kIW+P5BsEGm^#t%n0(5rAR1
zT+KtRZ+z@1nVE*<B}05xM24QMV%YTlB<0gE{?`J6r1I%^nznQPABTEM$*)4<{T6sj
zmz)Q(#mZ0NviRpRM8Noa!1x=-?SF{t=h$!f=Sx*?uP5C3c!wVEsXr(Cnx-5FaBnGi
z67G;4Hl*-wYPRp~`E)~%jg__*3ExUC1J;iG@c4MOIOk*V!QWt^AKD0@<r@dnR@TiW
z3YW3BWr{tfbf$kiCQfrT1eRw3B#v9;-84u*d_L3A$dD8C4Ir%j;YNHFf$W?+de)N*
zpdGT91h2$SxEG3fFh?tm_Ce1gXj-^fxM{UCAaCOy3v(ef*>J)IUz6@i8*dzddjtuZ
z)L@Z6upj2nB;dT0$a)9jP?$GRyLWU9vKmD-2(%xIGn-wrdPw1Ahgx+EBh?!@wY*L~
zn;HTY@6tf(_6Ct5CHQDMVAL8ZJi4W5c8~sbceaoOr<Ta%_mJCj?|@z8z^-7BX6A)u
z@T8GnxJ`mOmTr}bi4V~y)`9(N1@N2f-R#ITHR7=>8~g#@hl&|n$mUj?mZBnY9WKB<
z4j`Wq{uqN(9t^?{$6nIichZP-EzJ)lMAzc?*z^hFlYybI=YI*HJLNDUYHCDTN&-@U
zc6N63E}|Z67(OcqkzXp}3;PEeg_-L0Fm#$j$@2$G@X*s6XkgelRs^9JBMR0_8?twL
zw_YIrQzWv0&cA)OVWj92w9c?rYR?h_jFtY;osA1{o~Pwq$Myu@PR4^Hb<@?X?<oky
z!|B~+QfrxQ&=K*z0zM9GjxSun_lr0qrKU$WV1s2|=NWdEX8$&U9DLx}wHbo0zVx<H
z1d9?WW=pPOe^e71<~g*W-!bpTKJmLBj**m<q-Mhoy@1&jnW39rv;~CIYzZt>@JNZM
zZ?4#oQ;>*^Mb_UGp(c2-vYsxzj|d2>M8Nvs>Yg@??jOWG(<n^gfWH0QCEg_Z8ddHz
zuo(jTSGRCT=o1K-ac1o@YQ*!p9x0JPT#xq(L!qp{vxrYF)7<|~;wm{JQ(4f*=e1V^
zEW0TOX6iL}EDXy!2$znm|20JLn^jM*FcYs;5ma)MM8NNoW9sVa2Hkdg=&CXHxX7qV
zH+%UpyPOAJXZ7+{I??RPjIl6bCL-9nl~tVm@}Z)OJ(7%7+FM)2Ma60lnr}C&P*6oq
zNMWa54+ds32ktMCdlwYoaLfBm-b%k_T6GYFai95zWHNJKtf=i?ij-DWX#fx!9TW3O
z&bL(c_=M@3z1%nMzfpZfMGTm?0;Mz?X5`y!w$`@S6>5&Zl-;=#cyKJl>ZN3Q`S7#@
z60Z^A=$GG^D11*u71?FP)h#AL#N~X4wsPSlnxSL8L@ToETY=SgUpL)H7&LuwJikcp
zi@DQlf^~pDi}$G46m70-@k+9?vPzqC$E?PD4H+02rQ)kDND^XGKF?mhQAKR0JF5+k
zM8|mABv8nuT@Wu!o+4Zu=C<UwjE%%p?su#?KW1P6UA(N3v&!Qj2jNucjn3Bgb<)?&
z#zuXWSnbQL6s6pwL8_RZ*hE2czpuB_kL=R`0<W&FwrZBXckiAlpQh^Hp&<!^51fYV
zTAYfA2%(he9~MsS|Ds~Hr)V6n2Dio4y6`26U;<<lED#x3O;ow7h{f?c!B~=-Fj)By
zF2TU)a;G=9i!_4tHQcJFZ>=<ek8*#EVm%W+s*Kq7jW6t3vfO2c<nYJulR-iv*e$c2
zx|<RLRSp%hAlw2#D-jU!zGEv{=FxG{6H-DGq_lKqY%Fh0&rGT{Pd&O3eeVyR7BrHa
zmqlfTh&b1zdQSL4jYEP(<oF;y=S>IJ?-A0%yIWoSAdo%~f`W`r)+}&Y(|Fv^<xb(!
zDPyBh%eso|%=E4OEnr+(P}EE8Gf$IUnMyh>O#<k8`p@Hk-8ue!daZ*y3v61-sQs&M
zKC91Uq)u>*l$FAag~^|2z`_e^pYX3^wuIw%nQ8U%GVl2NTr`~mb#6n+y1J7opjWpr
z0zqFgs8bxH#N@C3?T5*utQ^?<?i(|y_$7_QS9(1Day90LUgt&x{vdCdfy3-iieSi}
z(af_A9sYd;F;mTgmXV7*EBqC-GNEb3ByMTuhAs;MfXT(MHRc`&+Tu4CTXVL=mJQSs
zDFKHPBSmYwC}W}fpMf;tu%TRbwxiM5jjR$^-_N)RUR75Df2S`S{dXbn;h*sT`K(ZW
z+TX4iVs@qO!xPL^T%Hv$GYug<4T!f^%xiiw6|5w00j#8dEogkKF?jm4oOuxb)*5?%
zx`C2Te0dtp*<N$QWp%pjBz*eevSsnOEvJ0*IA{2Z{~F25&TbOhVsY1V7AHBEcvd0V
z7q2iK#(Tag+`{r#{-P&3YZC?D0BG6F{)9Zw0#t!R9n>w^)Wd5Lh?Ax9EBVopJp&_T
zp%JTJRGk+d1pK;cf(+$PsKJ&|_Icfx$^i*Ij!}reWv`BLVI4+xy49R!9R{}cvK_ql
zYR<PVxgGR))Aw&m8hbnE)I#z?LeoEFH-UkXU%q_fBL@zp4UJ;$#q3?FPeMI8_Nk(w
zcn_nvW}65<oBmrX2<d&-WWSB>OFpA*l;|7N&nwtr7l{a8W=jx&E4Yl2<>*1R>Sc%y
zilSX~rH}=V9|x79pDm%8Ma%d;;DgGIcrl5p`1VIeDr2bOw&)d0fTw*wZi0CxqTaE+
zOwxgZs!+r2H#TRnKmF&o7GVFt{25&1<fxJ(#00sQ;B-f2a$1y|8>6ME+Vy%8Byx1u
zJG8vV$>w^m;E7L}nT#`iLP^TYq9O#Z7`{JrH8Od>13lWnXAg&|kRtDe(PxfgYtXMO
zTViKcpOuF@Z=Oi7t$@xPhk{<KQpi(J`#T(gk@euyw3)~d9qNQmOTYi7WxkottsP_2
ziq=#=@XmK_-6U_L3%l_BQVOPjyV6e=hW<Lbh+=>;bA)@01P7yWGJ2M#>`(&|U>Br}
zk)h=Vq@=7V9JY5^>Z<l3B6Gm&Lm?kL-9;15$4M9dYbYFmHx#{0RN{%rCe~Qvnh1!8
z*pb8T6Qmj%qZ9g3YoB1(W=OQ11i(&9MDM7g8b_*Vab6VWlT{|B$Kg-Auud|^G&fBO
zZ-WhO0`ZtsT{>w?t)AKsKRuHfRmZA*%7f*PbCE2j0&>GU;9L#qmZe*D>8Jwunq@pI
zlc5Mv)fiNK5a)DR_?6oVJ-(n2239G+5gCU%t9J2)y=&tHdoEmkj4P;9D&6mx4#r;?
zrL0MYM*o`q_f{^rY}|}U<k)6wreFL^!YJGv_nzomlA{w+DtGbsWY<U`cB;2d6@1g^
zInS_&T(L#8cU0l|o<{<%e^yL4sw6|}zKS9C?}DFrn-_I*{}Rz*Kosj2hU+EYx%W>t
zja$M6zkyhHQWuMY`=DTBaZvGIoKumt(LV#k(FXK2mrf!G3!Z5t&r+?Lh2moP&{YPL
zhW*^lWn{t-9JP}$sd?`725e1TS2u|wC2p9$6{0}ju83KmFp|{@*+4uUke5G7Kzibt
z+0^wBog)l@H`Q8NO;&(~se>b9YzardOkCnzmkaO=g~;?|LChb56I39aru{diupq_L
z6Q**{Jo@R=ubY*ta16jI2(fYywB(pshduwew*x}RIUK#Hnwovhk_uSru|YT;i{B0C
z_E$a`KASV=mj9&%kRK*_um;Uad;e;$LKg@sv>}|hVPhQj6eEMJbP&sg-axjdS094Z
z%%cx6KF)*21z@tuS^yKPmOiv6OD)XR;Sd^JU~1JHaIy6COkuz{aWjo+<v#!p5HmuC
z<{0RHJu^o*tM`Ek%?%li_pV0V6a&gHZ@%7kwXd1Z<bMj{z{eCB`Xph*>(@S8LX#EZ
zX$3JEwxqhonQ}S47|09g+|>)b{x-g_6LpJ-vyb$fW9joTT%CCM9u&f#Y;kB;w7NY|
zRh@>0UU^>N$d0P9s$-1S>O*_49P1`GnEX>DR7G}_oj)ndCnhFBM#_5Uj=>dOS|IJ<
zo1#Cz+x{m%3Q$Fk`Iy39z`d^rG`1uVEj86AoJTT@fXD*pLj6?lYZ^t7mA48TlhSjL
z!z^V2Tr)Ip#XFr=*V?`L7b12bL}IeU;+gwGL>S>N-~=V(lU^b(qULS=FwGEvX~}u;
z2@<!nveX*Q!;c6Fkq0zaW{+Oq*xSR>&Y}e!4h9B!%QZ+wpWS<I|5LQHr5ptmT;Mx8
z`>1~rI3G-;wbT*%P*&|gRAKeE!-gxm1V3994Oy3#X_F6bw!4kD{)M#*s0w1z?#J4H
zhhqyn$q;bo2M8qbd@$))=+o;S)~6%*(Kmh@;(xM{Rt7}C9eS-%`B3a>0wMim_$ldM
zc>OE?wE*7o*CIr5lqlGgqoa8?0~WJeT~m3LtT{Tk0o6q!R=h>6CDYQHDQ=F)BA)Ng
z){MnYJFW?Jk1V1CL8t^i!WV7-q;P{xt0=8+1*69#Dq1bW7954E{VV>14ebF?=bfL(
z6|b&+C4AzoNcXBdmMS-Q&KDDm_n?*~0QjzAz^NihUc&fWT77mL0x&ThaIhMo&uffP
z#4J?4<XkfonXdl&J7pe<32V?WDT+Yy=DfYt$5!-Y{oqAz40FK$X$H8|71_K$AOa>{
zmUeanKYy}4d7@kXrUy)<=*5<eoc|s<t+$#KW=?95NETnYp6|xs;#A#~Rq?%R-oKk}
z_&I6U`z@dUF~yhSfx(F3RrJYwz8zKA)a2U{O41cUiDgu#{ckF`&@@-?w%~Vt$-=_Y
z*nBjiEOFUJeU-tC<DZg}0;U7&>+4{?2LoiVz`XmCQFxP9bndw>$8|KUm;>Ece6^1+
z{2QxN95%w<b1P6$Mcl^Darv#RsReo~vX0+(<feJ9PnMag<tm})gJ)nqc&Y4nre<us
zIN#`{Rirz1j;p7uo926FDXTIdvfVINsMUP(Ck#x@5U6-P=sTDfZj>}r#V#Jpt!rzG
zGU|>YM5m>yIjBKku;fQ~U`mdvpL+em3Q0f#5)`Ur>CLs>RCx%L7EuGM!f;AacLy*a
z=;`Sp6y1^3)FkyMrkA@|*J^%Bj`xt?Ca<s%4ELygPrmk{zKDA40g663Dhi{;h>Ojj
z;bgtAgjBlWc-|}UK<uD(?2;|v+EG?)5RUN~=Q)Od<_&(Co29>wXaAgQ&SVgVf+Lgb
zOo^zro5L|W`a2b19fCk0VwjXkrEO-aixMW{V`HJ(-|ujRHGKM$c)G5~R`&+sE}kYT
zD~gB<B;EFOrl5!D&c(Z*ct!Y^=SBS0Ujd9BMz2B4^L`6(gmCJ_fQ8}a$F8NG-rgt7
z%&X4g92^AxcYi59(G{kf$bY-%Vz=FDGX6Ocg*jS&FJQ&#mE%fQ4wUEW<ywYLSj13j
z!l0I>Cg!jB`&gFdPo6yS^w``(_0@|t`<%N%U=E1yORkiw7Q!_+kyl`~6DR3GQ%^~7
zolxM@WhpI)Ik2Im8Nq-aGqjf}Ej0{B*JXF${7$Vq8@cyWg~IU0>9uHHLBU44%5709
zLjTFDt;#^{ODYWw*}qpeF0XA|5wNTD9IC`uk9EcmVYg(Xis`lV^d9;bD?fhB#DryG
zzwPF7v^7_MeLCoed9p0f{NSR{hF@=zo(%fN39n+M>KIo!QlGM$Acz*TOIDmfz5b{p
z^kLenlD9@fGekY?`og<nO$utM|I{xrC?7Vyb1R^rK=}H>d*=UUkbNZewqB-JgG{#$
zq=-~V!`?@Qd1h`%Jg}>IC@C#1oxA_><3|WgXm{*U&ccSir>J0R-^oo^PQ7ZIvx7C`
zM=KDbVTmkvOMh<ld_pfrWg>vwU-xwuB{{3|l#OF0i(Rx3Rf|+`1+S~bAUJXF$kv0;
zM)TDY*!5hOJ_$rn@I8671iIiCFPxkz=Ipke4bIIK(|r&Z)izV*R$~QcXPy;P-Wx?V
z2Ch?YFDX96@3i(UI-CexC{$*B%wo7spwQ3|VSc5cgj2j^T`6zcxfnKx(LSM>G$6>j
z^D^w|FoEVcbP@>w=J3cjo~96DX$!yGZ*{iLl;2JMcCDYYb7Q#L*$C)b8oKH?dz$p;
zK%ZC0$hBTS9(BTT{;}g$?nr(S6M&PeR+=o`{Byjb?<o*R+Y8sf0|_|{o4f~7g!<yy
z>yM_bd3bo*o<4;hJb2LG->=WqS;ey^<^5T&XnZBz&lh|KS{_LC0;AIvB=W$GJnH0?
zOa0uHZPAjNYL>s=sK<T4wLjB@!zZ^N8kk(ju=Dq#-sY%XbNa9fthmMTLG!lN;=HcT
zymDkxZM5(^w6k_@078SX41)n|{-ODK;18n0g;cN(IxTPS29;x6OqnmQuN@saADBZ#
z(!iOhqZ#`_F!`#q82Z9zU8kgoFR80b;s3vfok^Qu624uxQOvGaeV>Hna?7o`e!J=X
z*5?|!+?mJ3m=-%057`I;4Cz#L>wwk-l>8RQUge7eWR{8ABY%KLmLh!F&|{8{tk?7t
z1F0gBkA%uiIzsc+axoq1=fJ3~5ePP%oSf|I)^Z5NPiSIY@nmIXp{BF-LZubA?Rk8o
zgVp(ErRkX&kb4$CeE9H^1A)xXP5Z)YIqV-y%+jR8!ID8@71p(+dei?eZTTeaCn{V?
zFT1jl1k#)wh^Xaye2nm_1)bL!@m~I3>ineF<b4V<>ha&HeXwCxmt?wndgvGiGfNv}
zchvE$(;%=DduISh2T1)>Mci&__JS=F^Eyg=mE*C$e7HA#xG|xnuKqJk>>NlY^2M$B
z8o-1-$bS((PCU<dZ`E-+KzACUI87H;Gr2`wg~lqpv+pa`9i}TSBgM}w6cl>E$kq3#
zBHhwf-rC9v<n%S3K~bka?H?mW_vD+Wo0OCJZNYbnii(4x`{De0%g(}J@O!6b#^>sE
zh3RH&N{L>L9Z<dm*kAp?riQc|+{>l|OMPar7%dy$ggbs_xb8Wl9mXD0rds+0%RG?J
z<?MgO70KU)H92M_VinoHjGjwKh06c<^>?bmd@$`28KZ0m!r^ej(6G_d#mnm)Xi?A(
zNB`?@#o*OZO|wrMzQ7GN3%~!X=`4YNCNq%dtRiTwe0{IduP;^{yh~^u>LK}5y+JQ?
z#D|})CYsf;(NA=XtWFgKA91<yipTj|S3LF8Ghp$bu-=F#Ba?GX0oD$Y<eOW+ClYb}
zt6gI^cXqfTQdKb1<a3_tHm?JNd0$^1AM32erKdN;QN1rY8+mDIzh-4Iu(R9l{^{_$
zSfcjYCQ{MT$_@7TKChq~6GBUcpivl}yaQzvtnUt3hfSEhwZJYJJE+DM0{!YwT+qj6
zGJ{?4Jm0KAi*jt!6z)+|`}Hs;u<4X%ArKyi>tjGY<D#cGb5B`R#nAA&6_@kiDh-@<
ziPIis&*tTRfd#ya!G?wgFv@moS?Dl&tzqFf;{0MF<(46`_YsaVn-#KbX{^K*^K~Ru
z6q6$U#qdE^`<lW>ujy&R%9&=1s{R*4UNg9$tG|6`aHXzoQM=jq5{Quwx9Yd~d3jF;
znHU&gAaGwMaOjsBNm*GHxbCHSZB5m=Y(#<;mo#5_jumJ~fFi;4Dv>DO7>`zL({i%}
zMe(!(<}W&MzN8lUX#H&LEn=#fO$_7rHlNGCNVNgk40JG`(8~sWh?Y^?FcOI@EUvR!
z8_w2(Im#ME)i0z7IT?Ug12Gm3t?SmT5YT<?*atdURRW+*5Vz+6smkTVCb3hj%eZ7)
zRJ~xy4=~h^Pu`bYyF=pMSiRL#_AMM^5C=^KP=A#xerQ+w!!IR1zS?22HTz`(nD1vV
z`#)W``}BZEPK255QG~|W)%boYDd~m?-Na`d?He?a>3O`PQn#y+rHcns0ftXP1oRW_
zPV@4!J5g6|rX#IC!!7Wm0cRB2x`RlIi*q>~)8f>xJNOgENzjB9Ur;b8e$M6(&K%DH
zH=YniRAKPmDeZ8=OM2zRESvND{(|06t{ev>fI@kXj$#_QxUi!+T6|?-EBXKf2?f?<
z_lQZ^tYa4??Yz(GT@HJv2_;23idia1PT|Ev)%p)Trd`kCzZL+23K3>*F%?Ho@7c$)
z(RG!1Ym4tr^}(-`C_tRm22C^k$qeK8)47KAOYO9Qzc22A`6Ck&<>ea)i~EqVHAx&Q
zwJFi!b4lL-CKyx}{w0-?k!e%Si?T-b#<6M@|J^SyC@275p%W(UJcsKE|B)2zIeWrO
z63xhWv>GDYOv)*llzG7f&X?=Qdrnn>&6;C)^NHwuNI%-`atwqvKF4!D2IG^H?%<p2
zy@Y1GUDvX<rBoCSO#3_ayGYivo_kEz*g)f{-|T5;)r)Mu0f7*8tDLSg0BPj@4BePd
zQ_Qnq0*t)ErOfCA<hq%0S<|<}i`P$(!}9bvS?ajZA<}VZe*6#<Ey*2n7~hUO(yw>b
z2LY6-WM*!Dymwbt1+CUWuB;p%yK)!O!(G_e_}v?PUh@S1W+P)zNaNC(Lf{uXiFnjn
z`5?9FYHl_K`=xU2d-reESCLC{Ox=O^j{#?Kkg*Wex3oJtG2tdWb@-~r_oR*b=-9RF
z#h&vBgGmTKYA=RKQN`5`D}!p5$>k#qkyFdcGjd(hR5rEbk=wNRA!Xh~7;H=U+VS)A
z?kM^MPQ$J1!RORI_a%QPZP}ah4uC2NR2!F#`ak#L2C>KCe6_)QY-AuDZv<k5W>N_o
z5oK=N6j(u_H3|M>WB=v;?*}4qlKc0+YnQcNr|TkdM4w6R9-^dcqO9flQ0qW5(qTF}
zOrP2RA@nIjw)5GUM$!X-*Iew5pDVJHC{1*`tJT;)1D^JhzO=X=iuAo7?}#T5U1IaH
zy`%eH7=_31{i=i8m72sA1M3H4f(muwOdHLt1@~zUit_7AEZ*>T%FEpE8#R4ChIO7P
zh9i&g_`oCVhlA$?q8T(I;%rOgc6FQL+v%uY@-I~|2&tg9Q*s?R{v`7y5_4y+A?3O_
z4|C^*BKi!dOnxlC*V}nGM0FF{GeV9KUQ{z38Y=5DX<^d}H18gSD<Do)b%i1IScRjb
zVzQq~3Ta5|Usbt9$+#foOjNJU)HI89vngjmg1K|&PJsD4XVAwTZJZ^J=^GaN@11@X
zMk{0mWv7~MN|Q&#(Uc@t(*PNTg!cslC<p5~kt|dP7<Y;BaX&Z>#xT92Kqli@hplY-
z;q55~hhF(q`dyzdOc0Q=X=liCTqn_R!#zCed`yNu(bnd=FzvPX^d7N8;aw4bX7}q1
z988M^;FsTuuYl}bMkb2(6FG}=l6bcO$zfHXcNc9CZrBiUX_3owR&`tDjd0qU+{1jN
z<2_?8h>MC<%B_g5lQ9tuss*jwynplq-p|5)b-^aI8D%La_rcYuVwc`Y62<|w8u^#p
zsekT%TuD_xNq6yQ36-g`UiK||E}{VEe&m*C&+y+@>^6-$6B!+xAlFZhH!C0i{P|PK
z9Egxw1rm+h^~9-Pi!7`jS@1|dt0c6d@o0T>IH>|<A{}a?HuR=+2xHTr9akT$37|B#
z5d7V0;(WWvTbxG!rcJ3}JCNvlT*z#zSSr+zg}>xdR(^5Ap>)o-bWYp?t!odBzb#h`
zx+-+C&rUE9cKGeTTAX8ka#vba6q1odS{+n<>Ly>u4fE$MC%W?yYTs2onK4eWa*TKD
zJV-tJ@Z;eFa9(1AnBYbF^+)k~c0>HvLpDHWI-I320dwkSubF-Nwl}tqv$-KVo007}
z^rl^nrWQvyMg{DdcVLHFcR8;o8{|F<G*fd1<$`^B1Sba1MvU+Q3&d>IN`%+~I8!!^
zwDDm0<?~3Yu>tL(akLQgO4kdb_TKqHq6H&9YcvULKJ@+#7{!`&a&1r?K2erXTAW`y
zyLi4RukQN8Z(S&12WX;>0?>jk_a~=&uCqXoWAZaHBt=dEa@7Pl&;4l1obZw0PMF~Z
zW{Y{c&(YQScCdRr`Qk<(MePrhn38zbTco5I-KL#@77m7U+^MCbr}U?5KmU}n3O0?L
z%nCWYrzs`%<3}L2)_3i%sc3it!Cb~P+(3;0O5rr1+XG&B)Dp(oME+;fB&a@leE(A#
zxvut&&oNAQW1u{y$Yb>Zm$znhuK(m-+?*gi|8Bv6(Cvg=<k1l(7+L{I+^wB{d*2bi
zDWH*sVq={p7P<ePxSey~;$Z6>xEBV|D5SlkgLd3U%Yj4T3ZA-XKCd)CDN>7wSj7gH
z4-RCNcUu5Ea%m34)Qir-7(bH*2K`j^oY>ge@j>R7Iw2#)GDiqoAxunw=BTRsW{BzY
z`bL@#qkr0`PkoIu<NWSuy}Uk!m}ljvtd-R|Xd<_3mm&ZKB0sBsd$*?faualtk?qOY
zIDGKz>}(7N!ADm->E3&H{6RnDxYT*7JU%{-$#s(<+_+y$oD;LR`T{p>javnr{asm(
z!;`YKVgwt#qR2@`nFm)V=>F;wVq;^MP0O@x_-Bgs>p`1?_;`=WZl;?3Z;hJ9;c~Kl
z)8)}j+tom75Cx($f)e*PP#^B6d#t3oLC&7yj)2L5Lc^xaA3+yYiUFv7eD3Pf+sso4
zE+z=bs-!jQ8_j&{1mHfi|0Y+^4hReWQ(ggq8Jp@V>xl>|v1zNSdC!faEvogvL-fvo
zMi6ped`~uBV#pbE(r0*8Zau*Pg^{xF8CC$;LT2EO(E^0V(hd4Hw=M1tMHN%b+Cd+0
z2}wo7v17c^dBS!JN>EeRaQ5o?8;j)qM%=PYT38XR4X$~#>6n=Od3O`Hl4&0eDp3iB
za$Yi6_t$gJ_1)&;;^Tqm`2DWD0Ro<}t-&A{asAHUH#Rw03wkHe(2T4(zlc7lNWWym
zdlrE)C-Ct#(X+x(gZJ|u{b>}L6bx$@mFm!DXx&^d`|wA=wDOn_u(7cPZ6n0`(Q5G2
zZEtgcQUJ#V4uNaUp@ECbzaHUd5|5gy>fZkTN`Lb4px@O<Gs{3t(;@q}q1zb<gidr8
z*q&}5+^=#RVIAeJo6{*NkARjLS5j4VIQ{A%_vFMKDk>@h&>>B8?GPL?ljWaDw1k3>
zPtvsyWI2Khs#$PM@?7T&yU7qr2igNJj^J*E<1?OPCK|wL=NI6#G?&(T9<GB;J&)5e
z5b{28aCJQbAPOA_4|vRMW^0G8qeDYO!Plav$e3UPN+1^vDkuOQyl$QI8ps)XoA1EU
zI{_&Q!{}*J)Ai-{g7b)C)I&e=@lpW8>%ES*GzvB6K*U6f9Ze<}`sNq$TggQ{LIA&l
z5D>pQS^B!b%75F>ki)RW%DGFOT3sUF;r^Qk{;NW8{m=0Ql-{lAp`zGilaH6I>`i|(
z#1-_tII*pn1Mm$k`Ufa-cz76q>y6_2;_B+7oMb!rX%0V?`)@$M6riQ2IXTO}zCPpQ
z;{!Qv0YbBH<ujX*@4+|fOBAOc#^=Pu696+r&;M3Los+iiW27e}<i2lL3JnYE>+5TS
zhb0b|m9Bn*;iqXArXGhzCcyV)G_oyhh>A2>7^^lIM`B{i|NL2>+p-<k3@-R$niKm8
z2t&Stw|)a{Eg+K!jKKYb3bTI99mDHGLym-vaXot<fC8)ffZ+x}3mb?{xHaE+4lWIh
zrb*5=`H1+RZ`E`1jh$m0%x(L+0Tpmn75Y8~U}dR)+6PhF5B*|<^L4?3Z%ZJ$XXv$J
zQ&*2L@1Q4OEnf?O1T0VhOb>>$JHp7Yk5IjReSksIdHeRQuI@B&M~()E&&6&Rw1rqn
zYas(+J(7NPzvTn|!%7}q)<UpFLOViTeAeUWdh>aow3V;#W(K$^kd@VL7&#sT@H`vL
z?NRGRRNrb53Q(fqWjY*^PuMy^2QV7bg<=^2hu%PUedZ7!;;+$O$=12$;tf<lJPHa5
zju>(M<!q4@d{%(WArw3KtSPePEDU`7Np1-wbKyo;z;oTUXuJXk?JqFZXF19IYB$mk
z4aY#rnSW1au)4r${`vhPmhZ-R(E`BFk(3){+Dxz0lWYW6b^FGCzg*zf1nKTb<5yVB
z9AHWlt#=2G?d-}3Ize+T-(}!)I96EFRF{BTP*|v;(cja<O^%n^6q}w-nDPDltMfs7
zpAEf-0(RcWsOLbDj6NKI5XMRb?f`g{+wbZkB)rM|D+e1DK+FJP;5@aDN+{f~9tS|`
z0rycm9SsWk!{%0oW{M19fR`COy2Ao0t;!y2d#xlR(+GAV<}ISB$v`Dm$#B@UzNfEm
zKV9hhVt=B<&`?kB9OT=1mkE8qrkn#v#R1AQ8Wz3AraWVtppO*I+{|+aTK<dQ*b>Yf
z9Pv6=v$-4(ui72hK`T$ClSL(l7yN8=95kZ(oZRxmKnr(mX;aP8t5sb+#nKZf@e0fp
zpZr-`dLHpeM^n>v=eN`*hFmkCY%qUSKXOLEK!C`7E32&)6ch77-xY~xJ6)o_1|6k9
zF(AP~Y&HO19NTH~t8a_-ASRlh2E>E8U?&e-NO1+s&JVdR`kVqQtCjAN7@j1%gykv+
z7{LvioGP30_$%vuM>m6`^|+i3J0SN0Y9_a8)AmBM7<LCx(j3q)Ap8dYJ~=sAUN)<k
z`mRsYx2K_{UuD)$VcTlK^`4iWF-Q=7sf8LW-<V=>=Qn4%fIGdtWK4**9%xR^k7~Bo
zzRE%irS1=}?2OSer9%|BH&NB|Mz_jp%-(nZVG(Y^n3X;tJIM^aRba61o9|;%Q^f%B
z0Yvv;p_5jJzO^9Vq7xl7AaQ-sB2YPH18%7hQ##2<6@u#`=i(LR<*mGFi$6Bs#n*U;
zl`3*i<7m*I)?BU$!GSu|B$^RQuP#--0p4wVusSp-b`10Oz6c}~PyF&l*tj(ah~Fm+
zyx0W<j)4otDHI@gz&kN2=mKs?Q*%7%&b`j?2Rz_&fE0r{{J05voM6{aKaAv+m5<#;
zLH?-Q>t&@ESpH0x(DUYnlF}_M=i|h1f*M@^H#oCKLp7!3PKtC87?P>Nalnm*j~v+%
zUvB86C8gZRm_|LR68U!!d2h^tf(tGcft)Z^>$n6`B>K9=jD52YxS<s=qP4#=xPWk4
z8P0wg2J&BTZ;H~&EW!y)N)7=hcH`(P+Gg;RNJ^2MqB&W4dFq|VF>8Rz)B|n?;2E&Z
zIUStqF`9jhn7)(K_re^bV`otw6O%L~q+wWIi@PD~FD%~}*GzcR^Bp=B;<sp8ZzSH>
z|4r~2++{0i#wP;u|J`2Jk^|p<Fmkv$_Pee*wl1#U0&l1<HmsrYf1S?Zy`G<IHjv`+
zyCnd0u{Z6u6vgghQV~MNe?wrphQ(l1eZR_BHeSLkAN5vDpQmdg)tVJ8ZDn2#std2O
z9LYJ~ZoUSs=t8M++riq%3Mj+C>kSQ}3hUyc08i4jix7C&C5)%_O7vyl9$Hr+ilL<Q
zM!u7_l~Pepxv}i*)t{nnfAdn)?i>I}Zl=+}lbg7tCy^wwQw@O1>`(SWq(v2edlm$y
z2JZJsBwsz}=sEsn!dU^M|Dy<$Y+aNCjkcSbk<k=kmrnt}nZ!BdNjucL9Dl34(V1AB
zIsqn2ss*yzdNMAIqErV&9D|CAVq1A?K_?AWhtUU>QdA589w39x9xP1v@@4q0hrslK
z&S_4E15%ZWnry}nr7K{)j4?<}u~zVCtR`qN|0nAVWtmMe|8;s)8Sh;IQ>1*z)(J;O
zUy>Yvc_n+=2J6aM_#E0v)iO`u9^TyVgo*Xd5MNB(13Mtq00H>Nv_6Cx0}~^~8z@~z
zPs7e0+%V*3u!-6L7g+idIGcA`aVsh+@NWf<8^6*RL-i#*uPmGVwUmrnugi7OwRv7l
zV}|&{uoe!ke)Urg8rS=LM&G`@@3Ht17AbLc0(xg^ufJu=;-^_4L26Vn?a&0d|1In4
zjHGV%KHV#wWhQlINw+6|Tm}0#b}Pd68H~wTIpf#62uDv<nLh+A@>~ftpZ0_EmPZ*$
zE#bR|>;a^D@pU2SyFi-+8W3%3>%!{hV=g9GQ&C_2D7<3_0DViDuNIa8nIT^a(ma2q
zFR`vnn_Yik)&E~w0NYCgg8`a|h!EzSkTcIi_cf!1DuiX}wxBcczaPk=wzsSSsXPcW
z2B=TBTNbd#jJRUs<LBXh?0|^(0T9T(ezTfIEhosmi*|7Hh^pN~VD2w-dgL0|FSyiK
zPa{sKs^RCaXUH!x>zh*754?WeBYrPARckO+D?a@#gTQ!l=lN~pGVOW4Yae!dm*}`S
zbZ4&gg;zmiOyXjZ5D-+PCuRtl*_{HEw4_i%p5W97A0S~=mffNA>FJ`ht>YFT)}??;
zUxqK#xv=`r*wx#P=BbG3n?6lKlR5KV+l^?}HKZmCqZO5m+S@Ff97<qaCL;};v?Ol&
zkoTnJLmm(|F`_<krFhs?KF6?Z@3tV6asv;X;J*}K3g_``wrLHz0})V*Kl-zoiB|uG
zl)A8eo|&o4mj&YA#>U2Zt=IN&;hBrRh-WEBem@r-YA+r>!r&tR)WcF*6}sNbUr6#{
ztL7l-ZG;~sRGQzk2vdXAOvoevC1pW@t*z}4<*Oj@zayZP0mGQkKdF*ns8e<(R!~6E
zCL!fVO;<n1(I;%hWyzn6FLq(}kk=4N<{ka{?i@Eno?v9~Xp?(YF)ym`%~Q}wRaI6V
z_)hi2FcH6A;;XEM6?4gP#I!Aj!5%|)HT6B47)J37Q%jDAXVbJ@o33tWUtfCA<1q$j
zFsFq+<+GQmDf#mD$dzqo+)9M8Y%P*VdWHTG*Ct{Kh#fEdzLCi8N*Gg`s%ykCuxR?d
zyGqxcFnl^hLQPG*No!>4)HfJO6?w#5%Ee>qsA*RD{3|D*i*}c~{6JQv_Pf|6^xX+G
z-~j3Qp}zqs^&voyDb{*accz<5s&oJd0Jlh?HAly_wKd}g6=#;v;nC69=;*s-ZoRy5
z=A-1<NFx6?))mHkqjrWO6Lfc9M9gR$;?~o%i|IdaR`dxX!pFyl@NPYGzjj}J?&Sq$
za*^2*F#$lMS*?>aWux@!RrrwuT_yrCWyF=0CF563e;15#Qh|&RT-`JMuJuFp=j}t~
zb<iA$gRTWVk^k_}PP@`VrDg(Pg<XJ7u^l={w1NP{|Gh2Fne+-0m7ctRvn}IUkQ@i|
z!Q+9wrzvxne2;a6=^0N)NUZ~BoU-P`X}OJwumPDy-;|u-l*Z`t*<N=E&MDz)^4VTk
zJ?M<vuZz;{0u%x_aDNuOdaDUE$J<R_$nJ<xA=-BImwoIt;LhB;ckj?`bNfukJGJ^j
z0T0n>oaaB?LVleHRBmNsi}?%1$AUft60fL6%lPqIU@zB?_noj$bAo3!!(n1~s@A!{
zbzMN+D*h|DKq;_|i-Qvy+K`^_wEVLN4D=e@9ZgNYF(?#jl|+omhF%G|M0PBW&`LYK
zAUsP;Ct00b_#(FYF@*VX(4tf%TFU?~4TGz4xjk<%Uu!O>Hll|<R#sL}ip#hT7&SG5
zhU!_f7TyjgTkAtagR)ZggKrm~`!b^BNWpkNPa{rD--!TCk$1%J{QM5a1)}>udC-$I
zv}yFuBS5zTu3?5#iFtwqWS-*KnJF9Qd?=$3n_5@88VOr#1L{+M*z2sV1uV(fbV(J*
znh9jmE7Q9_vt5J7iTGS5kt%?z(pbDmq!#l$023LiPJwVmAkQAUt#j*_pbfwFK@3jg
z%fZZsE~*1)Kv9vUO5{FmwT>QVe@K5WCg8<ReyLz17F$v8N}k=ov_gCsWJ^$geM12=
z<A?-NiG>#2sNk(GbX`0n6ubZDQuDhu@gEU9mwDY{LHnA)MNM%+$rrDppO(u!OU)|^
zXv3&L%=9gvq}k3}p{~Zkcn^8FZF)8IoEm@Ew@nOw2P5(!By5KGpnoa~<(F#z8&&+t
z5_OfXnMjF`O|;>fy#F@2N`tty1uhBlT`#yY1}<vXiP=}_irBJ7B_>YJ&jaQeoczH=
zZc}I}jL%LW$RC*QFK75I2=5z}z`bxHM5+9x6%(Z5iYX?;5y+4l^QF8ge`;OjsIqGC
zl@i4)H{iV9t6f_m*6a3HB(8O4G@N^5SwLIc`t@1xUso?ZJw037O>AoKjKdGi?CkpG
zZ$6F(`U>9T$aSST1qJnkA=>(1=NXy0<9b~U;IbbIrl(t5EiWVUOfL_$4?wnVwM|mn
zVlB|$fR^n1a@)@j<W{qa2a*7tU!0=R!u82gq)6((w(oKM`SCVkS7Pwfg_L-|_`>{e
z@z&(*Y`yfjMqCh4SL7jlO28Za0&OXvX96#X+5i%m#<s;o5w|T6p>t~cMTS><2?;Rx
z$2mK{y&k4#4RfDYKn;*6v(QlAG&+eL#Pa7>>ZQITIjJkFW~!_OJdzpbU~7C+jChC(
z+@Q6{4+BvM2A&IGY%?}CHbiA4FZ~GyMx3<%pe?en4iiIw#uekcVY7Jq48BM1mL>iI
zTY^e-G=7@}IzUBrMTh`R=d2j|BO}fs)mczkI3<vW3xsAd(5CZJgS1yM1=x>65c!HJ
zf{q|N4pGgufPq?NcT#%#!ueYvAt4A%)Pd=*nS^x*(^dQ8?05-kR(6q@D$@bePDoG|
zCvwqxQwl0H1bmzhP)2fpHF)UIx95OB0Q0}Gx&5mKVZb=PP*l9RbGnj#>^lV{N*Hh6
zyaD-)cJLaO+jBz_2q|y+zZE2bTQYz(cs5if^bJ&KR|fW@;Z2+Be;gB_wAdlAFV}m0
zoL+V1V5~{hHD(VIJ<ot3RttdOnt;F=;RDnwutn>VjL^ENJ3DT0=Dr3oW?TMYLwal)
zt<EqM(hK+dsHf%RQvpjR2c6l6GB6k<B_+ia1bniq8JI#wGh7Z{UnswP+2_`@|38|0
z%eW|`_FZ%=3_utqB^|m$6lstFq@_W+1qtcK1O^b0kPZa_2|+rf1nEYQ1}Q<hJI+16
z?|<)e_UHZK@P!{bv!1oqv)0|$bti#1z`DezO905n78f{m2`~u=cq|VU@Q*wH1ZZ{Y
z9p{usbTOq_CTeOw0VP|Km8*mm<bapGckkZ$awvFu@NWyPE2a^AY&9ZZ0@NVHu!fGw
z&NdfWJ&v^Gg*rjRvkzaYf8AAQiUdjL46H&6#a|a3EAjw>tYzTAz{#nUV{(4|Z)OTO
z`CD<8$-{s)2COg$Z-~?iLq!?r4Z3SHyI1cza_E(>L#*lCv-RKKVYZg1#|ID==z&CY
zF?ikq)Sge^Um#>wO7f7D<FPY1zi*6NBuBDJ^f<Ej92$-9xBJyyJ2V(1nNLqYPx$Zs
zHkH~>8mOq8!-ia5&THlA4fphXwH+;OYLe!7H?*<~#auPn*^8vL`+ft~0IasMwe<mK
zf6Q(VGRxjjsj%RE;bcX0^2;PtWoYK?K=@u&!{y{~N1sNXj|>k!vjyQYio<IEDnJa}
zM`*$^Fn5>E>oqO8u(RK5U_A^)oIVBFICU8$OLvFkUi0#yP={wL;~w5>8WUpcYx<FV
zvXL1a<L#QvIpU3%9DYbhQ4uG-R6$Zn#r3K@qx3P8KK?7b^|_lnkFE>#N{Ds8T<Crf
zDi8Cjt?$9hXp&0tzY%=4yJKF@pO2MBmCCA(jM1;mpG=3fSI;vt+1hWm3d|?p<btR%
zfXkgB`FOJzEcD=@AVC@t=Y^h9+tCc=l!=1*K&Ir+Sy>fOPlJuDuOg*3K(7jU@2^8L
zj@oCt7s7T{I%n5UYym%+JU<A|7+AC5-I!ZgxGr0gF$xJC#Xrph(;8?dLPUQoHcHB6
zhDuSntZGw6ga@~_-L8I#nQgmfeWv2fpW}1!WIog)#2N>IC~yJ`!8tjG?am!1Ncw>}
z2D?0<WjOHHISH4PnAoLD7rhOx=A)y!%&t7hwxbR~Yo_WR_24W=`hudfuiC&tf(;k-
z{{3I)N|iL}+ZHY^Ri@!gsTMjaX9>L9tJ?G^6CKz6D;|>un@3S2i=i4XDq9AZq!)LH
z4c78BwM=Pa1?iis&mr)rq|;OgL-U0?3YRfRx-P+`T(BVWO^Y=HU|g5sG*_xTgX%*#
z1B+F)B_;1MtZZy)D)SHVjx%G1ZIWSwUx+Q!I(SE1p+XA0vQ-q4e7bm`G!%Ymv!)tN
zPs2X*0L_UK9E{g<=x4j>Gd$)kb?cDi90I2ZF3KDThp?~DVhI229Sz8er+@`W-F|d(
z3NW7LX;U|ZAjwApCB5}#7J{9Y#<#4G_-8kt&5(DOl<siYNl&5>2<JZY4GI+^H8nMI
z|BVB0xhp)ax^G(23qlL@=&zvr+413ffAE=-Lg6)RayL|j%5VyyND+uPYL6a$U~W4-
zIr_HjcApaaV2J&jsIonn?Ry*(+PHi7i>;)q_l@>KI%y>Gz}rz#<Fo34r~XUOAdQ@E
zR2K8+5ifcyx|0>rSeGj1<oV&xlrE9IG01iak-SI!8)j;2CA?jbFQ6NWl!QEUD?d^W
zk!B1n6R{y#%I%Q|L_Jo=by|2qtlzHo1ggW7&>+mPst$a6NYVN3<QO()n_%?fRlRLZ
zE69>UdkR29+DPcmMzq&U<)xJFao9bW#m*#Ga|r~7S;=F_KxILovcA54dwUxS9tzp9
zVYE<60bWgF$=N>5X7$gM%4nsEwD(UJq&XRdVtC>)4@K3H76pyNnz-P}rwq+_4ghT&
zrUJ^L%pkcDOv;taYc2=TdI+7~9n>p=aQ4C0eDhaL$-PL?d;OtO2DH)XyIGGEDVewx
zY?g@S@fwKUYPg<?N6^R67&R*|7W8wg+;21+8(tJ#*r48amm%k#qd?Rdsi~+SJ*E=u
zpgt?X?wX79{`wRcmDCtHe6nIkE4JTEC~2aBvn<mJ|4ngDVj<V6%G+yP_-f0qh;k}l
z`g0->5AXlm8($Pkg+BE~YPtDcNBo^q>C$JyJb1{~bOz&~q)F-uQEBWJp<lP8-_~Jd
zdDu@<A!w&X8IQ~E+7ufZmrIFKjlcEY|L{>`R{<l1b~NVRZ~}u=Z=NqO<(>Gqq<<8D
z(f@%K)9<0UbX$l;i^;~=YjS#G`jI&vyH_bg9A^3Z@7-w4f7hCukg%dSZC^R|cKz@~
zoB0%B#b?w0sw#fH$!SVm-N|Vg#Rc7qFqK!>QICrMzIur6T)Bh(bVbr<Ul>{7(J-13
z^qHM0pGmy{*Oq0mDL~Qne~ru6&+|9R-K2T?9ZkCw`bVM7)I}6ZHbQ>y4eQ<E*r@zL
zWUm{?mynkqEJ`jQ5DQFb&K2iet1n77rfkCWNZ<;hOqi53`9Fz#YamL9G+{u=s@xmt
z?YRVWfEU8$C8?f1&Y=p{UYSQP4{gBJp<f%7{QizZo3sqh*b_8%2j9FT@WL<2w;I1E
z`)2H3r$pPklxrm+%!QZyHYZFrJ?A!0Y){^fC#IMX>ojq(ommyi2i4=x1s<J4JjmHP
z+N+Sf9ug8EXW!y7ihu{!z!la0S!ucaNv&TNH9MX35v9f3on$L}+CMQybj5ecMx@>3
zS7|5$5ja&7)Uc;P8S29E`@S=~wanE-?r7Zb93j_wGnXCN3N!wa$==~{hI<K>8@{HI
z(s!}zr|}+o+A^7LiY)4OYosS1R=MOj?AtiNsufj~*1WB~yvaAJKSzw-_BH)!D2+de
zeU8JII-}SGi(nUy1?M}Jw3YP%)&5G9qR+iq)!)u7zDW|+;2c#(Z>~~(L=8T9?W-5@
zv44fdp@P7(Dh<!mcIgU&u$EJ~NT2$0IBun6bE<1-x9Fsm0cSKbIn1G6NjUwBeStfZ
zpTWa|8KJRn|2YXS;3R}<a)NxKlnBJ+p6{NH{-#9c#4F@Yk7VIcswfmzFKAa&8whcu
zAgq9B%sE;|tA!uK^SVTFU+YYihtx8%C^qw`KDH&s*>n+JLLFug5BC&A+(012t3U>Y
zPTR}ly#2h@5@e`}X$g9%H~uv;gRtSzP>odd$BL|bs0gAt5kBX02=Orx4v#W5IlYdS
z7h^A%2r&-VhmuL5gNk(WgvqjpxcWv9NhWQc^nbR1Yr%vIFEA<Nfo;?IVB><w(IXPK
zGqUUsn;ea#KAXtu3IS#-^*ac}i%9H256yaX3Ay#bzgfv{O?Ql{&|bwXu3gm@JsUZ<
zckAB)bifan7m~{qBB5Fd$KCp{yYQx_QIhy$P^bNx?s2UtMh}4?93-sZ`xH7ieSi9X
z$LkR`*G+X7$qC{@`;7+)cjM+UPtwL;)<}WWvc%}!Qq`>x?0hY^pp6%j#8+}s(@6F4
zki>;ON2zBna?Xt#LybU8lc8c<`-kV_Zqktol1Lh}N0$nyBsA=B7>KYdj%&N+N~Fb~
zLm;-PP%)ZBe*NS^pwu55A-5d;9IEUc+*$-wxW4MBT}sA?76^qN@G!9=5I<PZoL?@U
zx2-l!4aI~i-}rvp5%OorBTI>&&o^`^E7PTZNd52;7oJUmijl6&wHoIkW^Hox`>vg9
zb9x1(O6gj$_~^6Apv9*gihN3;48=y+>BDQ`@@!Oh&={19^<2Gn-a6{9ZlaYe^+)V=
z@A{=$Ci9<pUV?ds=I^G_p=1cd*-*`cu+y70QvM5-Z;cms<z@cu^9J)dRLtO#K19){
z8-uZQIgsMC1hX!pf1W&6^kn-dFoJC^^rrQGHl`Q+BVj)BxWWu|Lx(X?{)M3g9uLbs
zM*b8GkyoUu8gceeK5qaI8oli}&clM{g)4{o=zdmZ|Kth8(oWYQng);sh0LGtqoUwB
zY+!;J%FD~AP|weViS2q;t7My@L*3Y^X+%>xnvS}~sGqI)1gkCxsNmuq#zmW<vUd;o
zN2qQM;K7+U**Fb914bZqb#;Zr`x}t4^E#M|NrChUWc;V6r@=P9S1$u4zt~&@WZ2Y~
zhQG~F1*hh|Q;WD##wPt|-zWIv5lP@eQ!bv%%rKqHe}{FTs*sbHhm5){Z(KB&VRCe|
zG<tY=7>ItvnjN4VA}>Fm;Q1j_igzKDGlR)|3fZ^bPc_Cam8QKcmj-kyaxr>HBbzJ*
zC2Ay#Vu<k}Nxp)z2I-kcZ9&%co7hX07`W6@V%|=IiWl{%9JW~<iVF|t@jZ5zQUQR%
zt<&wcwziRx$4N>tiHRND3eHO-6zOWsCe3(A5@&X4f1&g=d!AXJVChg@ZuTV1ap7jY
z%D=nn9k>*Byt(O!0`S0Mq2YKMHNGj7qCOpUxo%s~YZZYB9VO-<6QUt^-7Ma%X{2e`
z<~Yr$w%>@ObK!03xquKC4qPI?2EJn8!xhzKg>kT+(iHD1P2;q#nuUF{l_ZQGnThe6
zp_f&~nW1uBBv6c=J9qFmyircTL)}N)$~x-0ma3g^Sb7FMxzZhqvXb)ao>eq_d<Hf}
z*Xz{wKFw%lFt^oJ)$P-@^Wyi=Jf~DOq{3gE`~X<+qk$}n3%iVW#I>VCJTkBS_);FK
z#~BlhspNkIY_llGZ{s8C4O7py#fbktevC$;91qTGtd#ZZ2;>2bBPEQo`_R+Wip)gf
zW8NV*vZD5=$r5ka;vmOttCWNkd9w0oAWosbD3@N=^_$2@_YFyS9x@!u<%+YiT@gWl
zAT%;~nI-cl`4Velok_u!SA<KC#S?>;LcfU9ilc#8tNj$}dQqDP-Xe)N^J91UMWu)m
zS0=ym$b$a!yV+kOWjvY8?Gr3k#yEOE6eDWu^8N_tAOIhMpnq3xH^-aMc|9QQ$n8;y
zk9^UT&MVsO+V{curQ?##<Pn08eRGBRwQC7jf^Oj9cqG$&y|>%XC%_c;MYN;%zKctV
z;{=9U{wkXP(=c^@jus}_?fzMO!d>oeU)0V364}6gOsg*j29-|q<)ukW8GE8;hh^H6
zKR5Ak$o;2q{@V5~PBJ0!XA?TaF*$AGRe?*&xCxzqp52GY<+o+6JqRH@Q44|K*{RsW
z7Y}Vmct=&MCw7HM1b*4vBWQE;E@|$3!gu|*WhvjuTcaIg6z}q6yK?-E>>(=8lQvU)
z2`P{C)`HZpw6|~r*n;|oh906Rn|0Wu&pkhO2ULRyb+?g^i)f1S%FeI9&ek~xzh|D^
zzaJWWk|f#N+t-)s>I~)W-w#VJ<D`_Iel2_a1r4#1e)C$v8hWTlz;O|@qEh>~$SwOr
z)<ii@no?#_9l0Xbu)677dGI$2-OnnU!^43UumR<#EdVKCN~IF2RB9}CnpZ;I5^yxb
zTXgR8ZnjKub4KS$jQpWtc-#9Ulw1Zq>(HL=RUdWiV;5RkP*{17o6P>oB`{-d0+HeN
zPi3Z5U}SWoW?6F<1R1~)G!5RJiB^fIP))82D8dfWNiDrzt406qlQOU)lWh)Vk0OGX
zLU#(~Kry_)nHF?ZjOWYx`nAWdEMKoem1=2m>i`?vyK=EhV4gCqcSyyt!5<%zY`!`=
zI?Jr16nF3P;%&SVc@f$e`9mpD0p|kNM3s{-PiR<cd-`<2bM_g2JXXj5nLeRA0~l7Z
zHaiwP05J;4Ct~ZDLDnrVFGIG^g>(0Kt2^JI`ti!Mw`z;poGSGGD!LV9sb<(&SJf#B
zyvIC@o}(%CV{zBXw$+PxiE;flMsp8;vo#$aaN2kz;|s}FAe$7h?4o@A{gI}oMakp*
z^64Ppe-oavYfw2BMVX+fkXg8=u>#QzD?failUy?@Z~K%nbX?SDmpmJo=qZ_gRziZc
zwXSfSTm1rT$B-<)YWOFFf8vUv5da7-Xr<MtMU~zqaS=p;M()XRcZ?nLW@J*%%2K_x
zIJbGh=KIhihWyTlXKHg(w^Gh<1DxcY+I6BY2LDAMxMP2-8Q%OZuIu%dxsqX+sr}6;
z@zF~mQ|uhJq+(43qfdP7N-IOKUI2yz&B?*M;-<%&+Vv7X7LstU+LrJn*`#s8Wu&yr
zuB)r~sxb28TS3-PdT(kXXw~uCj#jCr=EGfg(CZjICV8TYzm_+Ut%ch=^TyZW@7u{L
zksw*n&m|<mQ~Yrq6~ic)kxa_7b6oe^q6-zxfU+9ppIxcEW@%YAZ$=>yiupzJlzGqn
zZZ-`9Q5U$LmG7qn>W50Lo8Tg2V^elJyJvOZul)Kb1>*gLqC1#}s!;(s*2P(1hKQ4b
z1GCWI=2xJ1!7s%n8CG{de(Aii+(cYLnyC)Tb7P_4R*IN{ESxxIiP=8%e1YCM+4179
zu~9jOu$wNeQt2g9hr3JM3{K9|J8=z|7N+*Fk60_i<XpHs?sDvdjn$(cqicV<gee;4
zFB7^=P`XGSdaLdG=b3v52saS+-jP|nHA<=n^x!>suu?QtWOUc;Ea>{TbGO8oTh6<x
zpF^^A{aMG?Ps2aV@lQvF%SUMb>bF)GgE_fEuDO9K_(Jfv-<FI1#RoGpROKQEroMS2
z?~nH4={O>YPO{;)PzEUbd>J0~DpBoeyGuBva%nMMOzvrLF~gYIcH~>xnDk-HXBI*|
z4KMYyL-;r$5*4!>&mH~rg9E8Sw%x-Ndz5{7T-j=Cw+26Tezr!8K6bV^sF+{ieB!|4
z13GEGho+#>L8A;lB?1bqcOpB*N*-etFq3D~y{5MY2;iNE%X?Wjrv{o1P%IdUuRHll
z<4K~wbBWO@mzNNo^-T2aJk$vJly~<!Jl+RZr6xqeL$Fw9Qs^|wl(u)VcArn771#Ec
z6(b$;y+@UOKfkq_2#kv6RIFgop?dJ9VCzDt*cqziqx{IG6zr~8DZldxF|Pgv#0yO@
zUHv>Br3JKhSTHFwpf)|z^Ti0%91T1d$zH$2_}aMG+BiPJ4o5Z(A02pei6kvBMI_S}
zJvf*`V>nny5o&zO#-%%c`jq%qB^(-DI5hR2^i&$n=DyA_-auB2_m%0^M^%YBX-6~P
z=*`^yv%>LZL>*S`&kGZDac0lB;%)PkSKC3XT73Ln&6vW+%-F$#WvYr5e0{QBZ~=P*
zTGCm-=AZz-L&$^($<le<YLEbqVPcZilmN=}+sD5zn$uKH5D25s;8qX`lH9H96jX+>
zj@cD>aJ$D7lVs5mslt!_{PNBrXe+VHAj11hQEtlPIkXWOTkMQdwax}I0&a%2md;-V
z?8s{8?V;YU@+KfZwr`?hGU9vM($D!%b%|<?N7Y>8*F{G&8zh7)p$&8lbbQdS;S|-q
z^J_`Z0<V!p*uPr_>9+Xy6F*^im1WuV*8r9SLl1$)yX*NQX<bGwQ@ikHP9O}#0$~Jn
zzr)XNT2zGj--P^nBFY%Ea$%157|QRz@5qE&s|mrTzrxNo>pgBbTV*-9m`bM9`!XNN
z70Ct9?SNxSml^wkl0`=a81cgOaCtRrwwM-PMiQ}u1(UwM;er{=Lx&Q%Z?11xJfsUf
zq6Z$J?~0y{RC<OiUPd4!@kICUrcV@LUh0kfwm3$QzpFjq^==LI&y&fvHe;wP0jK)o
zRq%Zzq*N|H8HH(l4^3zpQVo14s+ZO8pI~x|(rqJ(49DzNh}iS<{kw8xau8xLCBPZL
zAxfbAQO+uPvEi9zhdO`ySA2biOn=VYGuY8ya4VC*2<&Z3C*Lh5|AC99KPgWVUgc&0
z3zfpQ=Votd^e^S*>;IW0TqIYBh!h(uKK;TiO($e^4OqZvea`e_BPbijk_LbE?3~lq
zXaU9XuSY;_Ko2%*Z~+e{{$wubs$J7CT>Tw3)%uW-+W#Iw3z;qcE)ot_>?Zsho<v)r
zQJWLLfKHlcSOER0pC+<>*N-L3a#(1qaKIL8)z;O)!Rp%3<OB~t$F%f{l{{CG)L2AP
z9P2u|_+x>UOb**3Xqy+eAzC36vY`DIX*c4`OOFiKdz{MWmvEu}heK_9CnL&l;iII+
zzbo$#w{wN>qM#rNvZ9y)GiLQq#uv{8ubapGF=N=GL{@7|mNyfxGRlA`YXO~*S#T2C
z$oF=xkJ9azz>h*Sz3rr&uVHM2X0l;Kq!gOB_w~Qqa^WI&3b1x65bsK{3h3F5hm$@x
zYu4ACiWDhj(D{S>%T8W=pFzOmd1CZN{*tx7Z%cd6hEtw$tPXePG|>!%6rV)-U8YD3
zy_Clld?>W|PwYsV^FlkL!+t?C0<K|#SRy7WL)v37YzlYseS+!ieU#^p_KEL5C!lt!
z$WUl;-D8`ziI9KE+l6j_xOpm5ii+`CZKvJSg?bEV#sx7nXGl52FkL&yC!R(nVL$R{
z33OZdV~_6Db-%SPp@|+pBs{xE8N6mIJd<T_N+Yt>G2H&n-x*<^whNUd)gPChbIX7l
z+$I30H8d<qhD}Ss;b+F3tiSLU)ms_9)sxl;iOE#EIGhyVbYNx4$;+1X@R>Vkqs4<E
z6P?xLU>L0lf6888f%KwFD5tE->9-r$1wY$_1K;;TGq3!+qR9K!78ELd^>gvbD0X7)
zDokn5(`wG;aGpoh$wFx2Aw2h1lVc-g`zEPjhMgP*|2r;F9`|0>-r*0x5TzcmUVo}0
z0+$(bmwc);b=G1IZ>YoqsYIeNQ6!ene#%VYo{947R{x!v3{f=X7b^J-x*AUqmYrp9
zKPM+M8$@tyE@rRmvXRp%H{Q$55b3c>a8R+hD{`5@N<{z$cK$cp(8l`>W^>OhA?`~r
z`@QU!UkbO0EC8~V=u^28k|=JijPw&C5I8qs-zn|qZuGVhbM51D8F#dz>9we%N_~kl
zKD{(Q<3kyapU!-jW3ds1+qwX9g>J{!nZ1do`em(mO(ps9Z~gQFQH5ctxP~SMa?IzU
zIp3)T9*%6dPOq=Pdw_tM0%fM<#+@*_<=L!dmY*2R@#*i%m<ENPjdwbNROtFwj@WYa
zjG4AUQ7)XPIeYwQq}lOfAlvl~ExqAi)$ZlI`>g@Sat6s33Fak)y~tVsbxQw}2CvVf
zVn`^z8;Pr6nDu`Q9h5hag&cE{-e@2U-~KD(cyP+?M?)j>%9IRsX!-i&k(VV8D5QLl
zNj-mErHUOJ=^*&J7nyPc$s)I25)&2uF4CSjPLbc@Ivpw{m0j!=_P7y3q=cjYX!MVS
zRI|z0G*{)@P8y}f+0XgaL>7+*U(*F-t_ve=ZLCiCEu_I<Z~2gt<MvYg`B509XiE5u
zUPM%z!(tlCN)g`0*Ao85hq{eePZN3R^>=NpPh5Fbp{E=7J;odle9xK)Vz)S#3)HVk
zKJV(TIQe}W^;pwqHr2R+T{zQLs)Sbd2L2D4=icC-g7GJ;ZI}iw>i$6Mieb>J?;ma`
zN0zP&i!o338H;l}>@?p>xkW*7c{X!agGfgu;#so^<xKk?#e{)!&;z`g<l0BKc(R*K
z3LnI1QcnEYCG4#yYM*rEE&0L-fuZ0=)Cc|x55#8|^_xkj&-IUtWcwb0dB1FgXB<H<
zkfc;J+_81_R|M<XyvRX&C;CTh<Z${~V{FHGJ9ALs&XaMExq#!K2k12=_NpJzsV-7!
z9l)gREt)fMYbglrH@W6kp@%l(CfoM($<4V*tdto!b07MRVE?auZu95_9R&df2Z#QS
zjzadw6|_+okM6$>e4^J)HAvs=@S4{#3*7M>w}yw$NW1YVv)>VDv2rNQeQ`O)=T|G^
ze<Y>S^w-6WK;Gwb9#|_rzkhfARl6g_L$VV>veV5w<?Dl7EMFk=C(q3fH;(;&ve+Mj
z_Ktjc`)QBw(1k?$&>2!aWlyZH<MqkD22bOeRs0H+Um&TrF6Rvw(c~Z?PC^DDK7F22
z^{IKS?9uN86B>(4p6pFL9d{a>Y|ksB6R0|$3%fu3#(Ewho+>eWe&L+Zj=JljGb&Y_
z_N%Y$X92oR^i-pRX@AZ_L3Vu161N&wVBv-1ez2@3-Xah7!$f6E_K_w$ApXWT;5vyB
zMyX<atMiYYWQtrhARhnYWy>YPSGNOG+(HChMfW!X+6t1tkp?a((``PFOO{59>YREw
zSOvC!(PseTMuOJy+?yXHq4OCj62Jg^ank+$1H5;Z7z0W%3+cjt`2eYJa@Re49{;)t
z!9E%ksrdb41~IK~q;rzQebHPGi0^z|_KIvKWrUkOGM`eE^XYo90Yqh(U>A(Z@s5dC
z+_-%A?HebEvW?<ty`9^B=p{WwAB`vDy1gH|`Lvo2TpmU==L>_cHyMjE73ff#N3ZN_
zsWWDuRol%($y&V<M+@odldg^Yyy2G)o&YWKw&S_&&=jFCJXvg9pYEhwM4@#<)}KH*
z?gOXu$(JJY)z#``!kb{^)V+Jva;ZnCDY3)Qv7(Zr`YdDG8eeT+y~wIz_NsI!U1~RF
z=T?yX2`^*DWdzwG#H3O6ieiyiIZG5H?|JrR-OuuI#AF<wQg<PGJB9J6zWV4ZB_X1Y
zFaY)#ViogTv=a0vkLEq^;hOcvZsmf!lt;|mc*ts%$#Qo?a$ZtI-6MbaC`(gxo1uGP
z<(J9wmm8k-F)0ZaULSn~K9Gh|e@k07iekHhKy2W`D<&w9D;k=j3&c2-Eb{yGKRtNL
z+jl15$_J?|ww$mrLx=lq#PE*q`@y<M+1t+?Z(lu+-0*UBNN1={c5rh2V<;2$sEp>m
z`py##MfHEfUWH-pW4DdR{9>|4r)AdIIJyj!D}=b_C#FCn6mGU@l*puMga3n(hgas)
z)S1XI7e))wtJnFpH>_70yQnDzysiHEkY%e}hPqlbbAoVmy|G7E$bUo6yCQY)W_Z=}
z{OtkD@Wk!eSe0Pj0A#pA@#pP2ik%RZFYtzc5W^de{7A3sO<Fu&O?KSvCFim>Gfkd4
z;CD&ux;!?RtOwO5Uii1QY+!j3IN`Xz=`S@YMr9Bok+Q14-_N(stnLtn8vhEV@$!fd
z!JH3roJ;lr4b_{+1jApz4o_PC`vxOjBp4Y@b%!P4c#Aj4EKc6k2ivk>qw^jN>1S3C
z7SAKdj^RBkC8Ms?*Hg&$Bg3o8w1bQ?I8l8C%es+Godr69;B|^|ah@P-LAt-Df)iiM
zO&(WYl{2vNo9QfkZ#i_9U8KF{j@9uc7Tc^PQigQ0!Al74a_qSzg}c1UB*uVJjj97E
z7F|hrqPfwq8&lZYlcjh=)gUi2$QARzY|qu5Ejzk)KqHwHuJ!uthks-k?z$-6IxFfY
zLS18S8}Qn#)^lUe-htiFhMgw6!p_EC^->mW|3+g1;XOW&>hQ!&%i%J{P_;GVD55$4
z>dVPZW7{xcB#N{S!9#yF*v>-10vwf`3tV{_;<OQ4&g@6eZ#L_Tu!msGQYTlr&f2FI
z7c_|x2tGK?+(xnee8#<<w*1`G!i4S~_?PJ9^v^fsQi<(m<l^nrwi8fS=2WHbU4`wY
zbpWA(N6*M_xIduyV+(r=Dcq;XH;+&P;ij?hO5OgL=bZUeW$#v2A6-ByY=a$j^;|fv
z(Tf7prmKS!55cP3eCczVsm0pa8J^feD00tpR+28@4g}!k^iR57SFEnf_2#l85Kk_`
zw|*DQFkKxyi@z9-iHgszYYUjWkd93L=Iy?^Sbsj*7_)Hm+srbJn}#QyH4~WB?$`@Q
z3<<?4e)$pUpJk;NlTWnPI|rkFML+DSW?`Edmz}}dwT+y6<YLRm*#fCL*frU~>qFxv
zh&|9gG_*MdLU*a~nxm$(L9xq|&*iNad^W$Nq-S{Z*g9gZBh%A5524^Q6({-;VjRSa
zg{jk{+B2bFE@VO8qAB#USnWmTAL!zOlbp(Ikgt_L=tN1nREhR$p1F;RHsA5|i8XmU
zQJ3V2(@?lMB}v~-FWWXHXVp&XXmz#Xao`NO6R+R?g_awQIRWX<=bV4-A4fc(;Tzuj
zR!#az+a;G|GV1(Gac(jNIk~>y6~#=l5tD07C<(%g5*Qbk)cU2JLiU`We@*nmKd&F1
zEb4wX$hW*(fkGUHsXUWcXTP7Mv_my%QH0;NhbtQa-5wfE$fHs$gHrkK-MfM8-qOsu
zo2nLnD?Nnfp-?T8j_A~j&!$_Oabr_92BMMti3J(X0=;bQ{A*m#|31-e7-D{X>gQnU
zYa&biA_f^fy0YsooRyW;>zf+8e&7w=X3$W>#HVeV-Z1t1ewWR?=1P<bL{X==4lpg&
zC7Z%DON$8f#c;NJgcUbNWYKx0H&+z<cE4;Aj7DW7$EcY_aCEfv0`0!d)PC2-Qrt*d
zEnCid8!xf|T{deb!pA##t5f+FIuDH%$o7O4CKtllJ|;QO3Z7mmxx0B<z4@K@pyzvX
z42}sIWP1|sC9w&HfZiw`ovU$OJJtWfx#J#BJ0^x&`;)5>9wOJwhtf)woIQ`>$LJNj
zY<b^vNK>KYo@?sjxMQ5Y_)gE|p@6DnU*JuLt0*n!Tl`DHjCVVk;A&+1mS>M}mFbrr
zc#e;I2!H@2zwFsT`F{+C!|UjRTyH17Ja^(XH=Ptz{deCtdM-$5rmHqqwqV7@%h%9#
zJ&jSOSpxcsvEK1<(EMXI^CJE7|FJwN9^iPoe(=bzYUe4$HJV4O-IEgpf>T5Sw*3`d
zn@`>cl<}o-_YU$!Zj8J&#^voRL>HuVhqb(40KqqnkJ7TCb!6b<Kxs8Q-Q|}z^kua@
z2|91fW_IU~*=ht^xz*3W1t*!U$?UMI$su}|%69_(agip~JZv+Yd<oAA)!xCk1wn$j
zziuI>!0267Y;PIZecNU6H}dS^XsK%l&vz1~-}4+s>?GK{m*M@#Wu<w(DWg9Ra&3~_
zpSu3C6^LVUu&4W<y@|bc_Z~N~b!zZ4*8Z3E3){>4vSXjrqD);D$+*%B3{FplK&H#m
zM0YsNYjemu7IK!RFX-%k7@UtbM_Gq|Ql{!u9;C)+ZgVb@U6wD0f{?)LY9G)4pg2xH
zPmjlVM|Z#T#%T<0s}Ts!m1|Yyg5?_*tnG6|>4uN{V^!|nQS$OJnOY_at}GH@9yfGw
zA9E9{^3A<%p#cT~xO9jc*BYMv>?V9(n@DxnAM;S<lOhd{fk%z$CsJe}`&RpQv7#EI
z&-2A7kH1;gbk-%^&j*ryVpb>A=C&F${y@Bd0@TN}F+0iBV41_b!H2wV=Cui=`)<nY
zhBJ$%*}ZBWhq1CuC4u(7dmk(%(|r0$X$C2?8VHpTh!7A+^o-GSTs3cN7X_z1P!Vix
zCppRLj}ymnlP&J~++B$RQOhE%jPCldz2sv~y26^Zomzp55ElYHe)Vm_J^k_xlD?%t
zII|S{^Ec&8efCW7dZ8sj#Lvu5#iA^)d-ntw?T<FXv*SYpm%$dy#;%tFYJ1$bR^GNM
zpykv|f72+p1h9It$1w!0Dc!1dJ=RXByQz{x&%u_TSiI6a7&;&y3kY4>HtbiaBa_&U
zcirAXM2bl3>F>DAPxfVRSgW|{e7dcqv<T$Y&b;boDlEEBc_vTSV;Tk)tS=%k)Nqu(
zzmTP196R)4WD3JDli}5raIQ~VHyW=MJo9I+y7%$f&nQ_n1!)&$zrk<EzdwfSmMM?d
z%&cy<?NKD$dM(k(N`T-dGTJ>Ds$1T=s(3U!Pe**^;^`W($2@`U*_=n-g7a?w&*5Wa
z|Id`T+#X&0kT7+P=kqozzRpq1T@MQq!)-(xY+y^1w?`x#t`v1M33iAQ1YY@%P<nif
z<xh{OIbtlIvfcCyTkBhbsxK%tB9CshFuxP0^CM;5Yj|Msv98vclH|{qc6Ss%<FN~k
z+r%65Ky|U`onc`Gg=Iswf9L%b`pthX4M!IpsY`Ba`H61CC+h1?l#%ZnLMb~*-kQsV
z+*Vn>@7ObQ($RaYQREbycJN}5DZu@>K-RyAqqT^=b^bBO!Ql^GRh)3P-rJ56P@ENn
zkIs8TauzP1UR>HD^!(jkG`zDG8j~l&%>@T^u_~rRW5RHW3bY4sxPcidx3$JFyzHJ(
zt&>;T;c<v+_qkl|)@R`ybAA0$I-2sS&Z4M)H~b;`ze4Z^gCiAFyPRpY<0s8Lq166g
z%mV?$mQF48lrrox=bq2|Tx(cX=6#+Jv8%ZF^@|mGdTPxPh0L{X!(@wZj@ly!EWhq|
z8({PTuZQE_K>kPk6#od;8)sGw0l)U3PBh)7c6!lx+1p+5)#_8-!}+u(@91b9%H@&6
z_l#F+6_oYgg^S)1hzn_7@cSplus%Q|4od3*#l`TY`%)^4p=|+dQ6sm~3nU!w<><e=
z7EZtTX4b5af~-23OnSq!)i`D;G-B_~?cEZA1~FluKSwxQ8Grt8It|bNb>8}^HlLo^
zg(+J8pTQZ#XE!~op*9jwx}ZoYp^_d**XFxZx*hb2yF*xEx9@#MX<e}^8=cwsl#$^o
z@_|`S@qMvZz;Q!qS9RoDjscI=OJ`ja6?q^3-t=;c`MVqtST9u?wUry~lpBq7;2xv?
z@5>9?YZ>V2S+Bc50QCTb;f*H1&B&N48_78C)Gbn7^L~4N?1|*8+!IMbN0)c>Anp4s
z6Qa^esE)|e0+W0#W};FD19jbUPZRz`udZ?ZeZ}2q#;(Bowr|PF#Uxru;pRn(?#%V7
z&l+lE|2=<1YvD^!&M*!jQK%us))Yg19W---%3L7Yvj&zyz$nUWcZJ-8Rr<HMB&N1_
z1;0)ZRm~=Fo9O(M4)sRr=JU0t+|~00zH_@cbnZDkUu}ZwbUj4<=52*jQ0w=Sm5BvQ
zFR^tqY*qQqM0j-dhyAhsj|vN927W#(tG|Wj?Fx)Pp_iNB#C@o0wX2vT(TRq7)X3yy
z>JX>9FF^4UR#Zk3EJxOiv9C?ai&UrcZB5T-{WTo&_j#(wc28IMLT^PkYmiiQr(+Mj
zth8!GU<Z2w{WWRHh|WZtd$%}*E$2objB=~A5J^cMmH7Yn)HLoGE=F_cVx9mfIcf4P
zwAKPFnxURvmCMpEEDHM8{CRQ10@G{5JbppdvZtrIS?)28#idTWg%?kpmaPB$8Xye+
z!A&gNbV0WDy=)VmV1r6h7w6*gtDcS}%2&UCH%7OJN|^f>I3*^{e3sAN+i(3T-rsXV
z#MW)O1a5C^=N38o041gw!Su|#RB73O3}Ix{=rYU7#l@5gKb%KU(+d#wfwbQHqzI+!
zktSHnK<LZ_vSI>yS(GBFe0KqVFuTM2^2K9+rdv6KVHy;PL^K{gvkaRa#3?Dlc8vqK
zqRazm?_EvR@4vf|s!z0B)x~-3x`Fc*lOkHatp{veLzjA6uW}}D6^56_3A~b!;Vy5U
zTP<E=M*h_)t^2^mO?D0MaQC^tUQ$cr-m(eeoi?O~2Y&s!p#m)WF(-@3@$vD@Nw|oU
zAF&-BZ(8|`H)RvqD)nT)bg^ef>U{}nR;*Fl`m-fW9ufPAUXGVnR=V#mnZ!GybOFL=
zrImUdN-@=~6GKDR_BO73LIPc!7{ymZ%t6g)pCqN{>XuIky~+*y*xI2eb>phaTeMWy
z-oa+|s$Ljd_ma|9`A;7M-5&d$qmT&tAQY0vQ^bMtDk^Po^pXlv>T>htOCR4GSFz8}
z`O&K2Hs5SUw%=&xsS+1fh8#Rj_Z<L(-2Iwp5dARxz4*IV=X|)ITsdr=FFO=|)or=$
z74rSV?;XbW8}f1yxeE`C@%+1pU2m{KNe~Af-KE!d`P~edM2R&cZJmArsb01q1K82j
z0MaT-d&^kb_EKkEb=p8v7xt%f74*?}9CG7kI)p8abX?Z;+5&Cg+3D=Ie*EQU&6Zfd
zCEO6}0nr5it{G;c^!7J&OvkBypZP^9>02i!NS^+=Pkogll^xoVwa=kaS51)$j5i@6
zwo2Vaj&Xucpmx;jNPEYlgeZ_oLo%T^7nO6NhMD~BnhDi&0*Jj)h7M^m%9-hC5;0ih
zD_MSUzL0LYt*W5eGS7sHZn5Tj93tJ;VJ$?2SLh-$6wg|k5%eY3D1Ouc6K0;X(`5bn
zqg;x5oSE<wZY_Fp0>-#S=R?l070V&s>W>Qcv`;@>7^&BM|B&(8Ma+30VKs;yeMh6s
z{u)x9N57=?cy;0xb=ucnC~@sZxa_q}rkavPhF02?udJWzxzTt?6#};+-u|DX@-NKw
z#&3zf#`{@cykawTu*w$ZYb2i>o-!CS*n20k^-X~}p`6~C`Ni|+P`ruzilLbPsXS3b
z%-xwoA3M_TM%<qK<QHG__#OqbW;Ne^e;FyIg0pAOU0|p$O=51Q()$<v!(G_MROG3>
zS$x8lZhXME@wbBwd;=;U;x~?DBlHUO?D*a@#kEH&%t`<JPOl~vk)>(*4rj*Ae`vEX
zH-<Q{h5E!bx24s0?Zt6P^A?pAp%au=byyTNOs%_DmI~+Cs!y%`_}g89;BH~Yc^#F&
zHTNfg<xbEOqTWkMyZh#I54qjx$#OR?-=fdrA#;A{SVVuBXn2^XyC(G=DgiS^%pe;s
z<*yKgYAN(1LNQ34{JMQfv5@`ZF}*AiUfXx0VEQ|%FX#g+y*Z3^0zRwVPrg6F5acVd
zy`C6zGm(ACa)`wLD(kE&5o$AHS_4BQMXQQYSZVEfh}T{4_5n%sZ4@2LD|QNV*(kGM
zBEpk%YIO<AdrThH(zVUATUUK+fA7V;l87+RXZO<4v>?l*b=S^&Kd&oLqeCddK0mcj
zl1#wMj`YyxtmA!^>ueX!i*v!mYvjsT1vVx^l^3^3nfCAh>fHR2N@k?Z)ZGs6a6`iX
zwuv|UQs61asI23o!HMVblz3mqSryQXP8P*E=LiSG+bxRVVE+1*hg+BBrK+l$>TE|E
zKFLHQpl~C+hg%k%8tff`vH6nC_nTYxBDUIxz-y-=7o5m;UnnB4X83rT2ys6I_g9ca
zbMx~zdx6fRav2cqOvn(Z1eb&2b;y3?Lh)5*imxEjcCw~=>xo<D&kC<SI~5|!&UoT<
zC&<Gia5VK~@<6o`laP$8>;e^B_Fm0_cMUO{=Ed_b0jtvdgwHi<3;^VSB{u{*g&Y3{
zM(b9E3047c3(8h@u{;Mrpm=a_aG8L5`{)D;?10H;QEwRwE;>2^Vi+F@2?@h)2zVec
zK$H!aI5#&JitZc%IRhnGxw#$IB?b<S#DI@>r!Sk@o`-5s#EWx3j3!EB>*$eJ@gvdn
zl`Twx5A|Z{&(lue)f0GOKLEZKc1S`Ou<YSS1-#w=YkyuEGJS1K=5K7GAG~mDjO&q)
z%j^BS*H^`QL$&yc3}!y?l^aswX}(QmF35xQ4x)}WAXEx|zxWJ(GW^DXFZh4$Ptjk*
zWR6t^Ya=C3fxB|iSr7nKA})*lP@#{Fpm<84fkx;I%O&8xfTdhiQVMO_d3nVLxF-Ob
zxea6mz$E*Prcs0Zto|6sLE}NH_-voLq-4Vd0^vt0DiUat;B=@S`}D~SdOZT{GHlo;
z)Me#^8$2e&h4{CbH?XJ}<d-!|mb2b`5IaemC-;yD$`d@N1DT*Ykx?np9U9I8PZKCa
z(d{0|#>PhA*;%G?8$&LQNUZhq?=@3<v2$_dx8F`uA`QMJ!37{BV3`Dh-Uq~W8szhy
zC-~FZZ_toaOah)5;K@!<$N*e@&SWwG?Xe67x^@-jKsNy04i%J@!(6C7k}f8A85krR
z<cGQFe;@^BsMe+s3N}_&Z&Aa~yydJjI`p*?^%_?LFEq@T6#ch2({p|z2#6^FgUl9G
z2&T=D20m(f?+AKQ=4!G{4BvbJydbu=Uz9TCm6R;VGMfKq9KHAN#G!B@>3<bt)_(r{
zxx8Fk<l07VUY?Rl6m_$qB<15OcPP&c9#{d?8x+X7a=p&sO_tiE_@Qz2DmIl2@E<gQ
z<$syiR%T{oFygqps(Ez81&|J)o;nz|OaVM0>>%7+M_`9clMV@TS@~`DJ6plIob~og
zC<t$!b404_d42e=>LcsFdicP{2~h6952NdmtgNJz_wQ!{tVlq|0woc}(zOREQ~!Ge
zCc-O(;oprG3dLAh0x|Vr1b*%xYARr>v6L8K`U!V`)60Y5ssS0aPdo$_k9cLWXbE+w
zxE%w^j%(Mhndr+bQ0To6yspJSPY8MU*iMUphN%tr%4MxuSyP-3`)&a4#)4Bo5>TBy
zl1|$I%m*+K8G3ItV&P~&Z!q+W@bH?Jd1?c8Tin>Kn=OivDgfid+E5S@!!FANuxY^I
zJKlVv2Mk$gXMoIUt6w|3^grK%&D6|!j{%FTa=C1D1sxS1f9tj2cX8;DNt+5=E2Uzh
zPl}!AYvMdPj3ngd0nv=DCv<fluKnNjgcp%c-^Q_r3kZQKYhUy8@e!){`ucv(dQEEw
zyY>Y3y#V%o{(n~yMlSt-stE5cBz-q>hvuPJ%8&*EF=v5dk1Y!nv}ArORsF&6nFb>t
z935>)o8GbN`&3{5;M;hO<D8sq1g2$v>kE)?5MD_H`3h9azlh&FBlEAAjgZGI0M%5&
z$i)X}gZw>UssiGYv|$?qpl4NJyTWuq%lrXsT;L_xxJ4)))B|P{5WNQ{dz16?^EATl
zJ!^ZGmX=(Pi<y6Eyq-7#GR|#xVf)&XUEt`n1i}CuNgto;RA!5!1NgSO%q5BX?;3GE
z&omyKH|8#WWD0!2`9K5;#nF>vV>U&@SeO`)2t7`^c;!Yt;OR~QwiNUJ{iX9SkB^VB
ztxJKQjwG1j*UqbF{c|SC9)IGfwfw(hrOX*t%0MpkKcevM-2Ziv2;q_P-+M>c+DU-k
z0-H>ebxC~ZA0W-@O5k7HXr=)o8Ij~d-xE*9iz-CT(CF1|b&M(5y?$-aO<)7)s;;@9
z8n}hKflqo9$}DSvdmaQq>RI18g#ShMr}%_kKw7U|XeJ9nAxs9u&>7(J_YRlx*9xCQ
zF};?dk4rv*C{W;G*=FC~=~@`t0{FJRbFWmmIw&)2oxPU6iw3<P1d#ARenTD&h$~%y
z)vv8k0SyH3SV;WV)>e{|rH;-pXh?l1Al23N0+Qr~ox?}H^S}xv0z6$pYXze#AQ%7(
z=-)m!Qv3W2PKA!DstIpgW#WDNi7H&q@(=A@(a}Ikn+=%FqKu5Dsb^zAP4+WW84Gr=
zt2Tjp8rgF%h7Mqz0-<!Mgie4$=DP?i?i~SY)Qay`V9OIe5>iq((Mr2<!5vnEA`8VZ
z0s}QQCqSrWn`2oBLQeqX{IL?dM0^b%2$levExR$VaxvSt3XMQKB9i*9&G{mB!(A9i
z8$-bJOaqd>ka4f8mn5`-J`Jd)0$b)Bu^n)IJddHg6$sI?N1^(D^=Adn>nV_a=4AdD
zU(hn5;6q_6H_!d(fX2XvYHE_`vT;{k87Fl&ToU$JUh`{6tay4U{y|xp2WXUlX8I<!
zy>DOh%LKqigWN%@Yij=4IyiJ*Z4IaY3k?yN+?<V4a1lmz|3+0)7LkG~@F^hLxn{Tt
zem(`Z(VfXK&;%K#FfcIqVW65?pHs`Qy97#@e{jDYIU>Ir)4tO}7uFHWLkB&NZSE<@
z@y=m63V}VkGKT2lOXFBzufkv^ytg{x{4MSrV)20yY1!bz12`{~Mr^{2H?H#!C?4Pt
zc0M~j@$muM0N-d|SlmWkLjySZZuA0SAIKq+nsp%Q8+gjT6h678mBPT!Uv{*9uude!
zg7t-=p~AX1dJRBw%jL_3?Lh1$w7U#8!C;<#j^-E7g^vMPrZej9!)CG50<+eLM`7fP
z3s}|&V778iZc2raMscw-ps9f+yakJ>k96mREDhw*Fd;x~_h3Ah+xWqD@*ME<6%8&i
zVB;-2J7d5IR$N+oqZgD4ip9{-P|R!B28&{Z<2?YxjbAjE06d-ocmL0yKcKYl;B8?a
z->w7rVJt&UY*JD+%mpYbILi$D1rH(0q7RlHnm%Y6pMMNGz2o^CgBaTcUb_90Q=!#z
zQ$Pq>zife5Mq6nj;!9(MIRtvdRD;)`C;zmHJctDd(e-B7VlQY4ojo)8O;~<khy?Bt
zv}OlQWy(#4Szr=7faVQ>pG1M%ngKw)*k<cpM_?aZ!^I73G2(_Z;<~J*+vT1twpO)8
zd0i$(2G<Ed&!H6jp%^ehU_yNdJ#fj*i-s{zfRg|mta`9}dk_$}wsCeQI$dudVA{Jv
z6`rGoJ)iS8g-}}V?CqTTmBqkQ8qo$}D@^J`G<5UO+DiixK~<V9Soq3NRugTs`8NgM
z-lblkc$Dxd&P?{qJ_-bQX{_fqn9-r@P#n(pVrmOfa9cK_@_V656&`SLfr{V5HuRwH
z0`M?!qP(I-tM0?*KGGtooCy`Gq)Eck%k!9G7C_CeIz9n*#G6b?cspN3s5NMQ0QDru
zgK^~pVE#-CeY~^;c0Zc92`-uR<HzxpDh395UFRJ7D!WCM(a`1_mh0!|XASh;o{2sa
zYkQzq)&Vk}o1I+=TvAw~24J_`eH?J%@^wNLB_$<0JG-z;j0&;OfTaqm*Ex@5LT(ZN
zG4>LEtf?uD205ViP4pW#&~Y@qJpjQxP;eG5Mv4$u$KFS2P0i@aUJ-KVOy)MACXfP4
zbPmMSn&UQ5AgmNrXKLNC=qNw<li(%1Eq1mgh9$=XMPvZ)9?sgH;Gsc?1i%ZZ(waT{
zeUg>|E5tqLC7dlf2q5PNNq0?tV5sS|_TkN8Se@C=|7ZdNtbq_f-5TIxJAMP^9!S8%
zDmd#SjqW9!ZD)#QjZ)DS)I+n9;8|mHo$ENSllDLybI&zU;zzFZhwCnWiT{61-q`r#
z$mKH7F)r5+xdknvVg+5S`=9=@0|v}RX?h21Fl5ktuxx|N@1*ZHYLQ*KbP27lqeGr8
z4g`<Lth8n*9k~c6s^fE#C+Pc^m#ME9Zh{4+#g8@LcC1U<TUs9R!|}}b8?d6iPA&ZT
zjke~)2zuEFXg>J7vA?ARKn<%a+EAP5-hXugaJ-{uuvRi=@P~Mcod2r2NF@P~inF8e
zCs*z9Ca4EVv|5G&3oC11t|!%$Y!cyJVV97W%F0SmfB8qfK;#=0MU*^`wTe!)Y=EQ<
zJ^So_GB7)>%D@dL&-^Ki2KE){Gps!Q0NL&zZyQZiBqeDH%Yk4Q+fBsP>Y^&q9B?80
zJ8P$V&;TG*f+Mj1;>`)*R#O&Q;X_q!G*JMB)K*u(vF{lG35NRm>bVu(+#C*o+cKCH
zV*UJdE2S2Md{+PfsIu>ogI~LgRlH|#Z1IT33a${!*X}oRQ3HSP+1}Y{%IHeXO6e$|
zG8~4@V}K`{8_#K>=OgfvYf(j`(O}G+0o#0WE=BhiG^bI@yc6NYO78Cg1R$1OiNpQ<
z&fhPQvTJ`eg4KF>cwD|tsa<0IzOriu+O=TaELvjS0+U7%iCc$<hrj^9VNj)QyEg(>
zHqZxni_#!Pj@KJbX!vjt-0xC_2Uro?*v1M}Kb>7H`EH%<<Ue;2W9xV&Hsxoyo>`=(
zB;bdE6!^2CW8nXyXJX1cl7GZhxTdNdf`;xDMqvcRr=L@qm4!FZVISKpimHK>*2jkr
z@#2YOGfI3JSR}rtrcOiqib~k;t<6n1$xsdkWGS#hYG}965I`^qY+O89v4Seusmxd_
zg?_FZn*D%Jh3&2aL}Bn_QB?8y^8yeW4T0_e=kRi=9X7fR^7BjCgdp$)ru9DIyM>F3
zi|tha+GQ2@@dd;W%CtA-)liTF5rb=wK}sPIsgSvVU$oc?q5mKKxq5cNa4bbswu_kv
PJ1to$C3M~cqgVeA8iXwi

literal 36040
zcmc$G1y__&*Y-^#t<oVS(xEgcASIy)NHdg3NIP^RNSCyd(kdm*(4d5jfPgeej&ygw
zXLz6I{RQ7zA8RS%%$)n2eRf{g-h`{GJs~EbC4eA^SXoJ46N0erK@g@TJ|_5|dtO*g
z;M)x+C4FZIVhBb5!H7leSU?aHq%8kf%Oib#`n4$?3L$lFtC(hCn7r=cZr3$V#lcMX
z8Xp^9@?F7Q-l{4KTUDfH@+F}K=e|%P%H^$-*l$cuPMnE!xom#UP)bZ@Y+2b`Os|+_
z+_qw<N=Bsi{HT?68apcP_A52}nMly6rgn8#^GWwVI%4gvAYuVuD(o5?7-TnZ-n5Wb
z<O4tFRGBYc|G(t_!ylU_u&DDf-v-=OhMwL_V8XW{Wl!>v$}%n+U8`Ju!c9#S^4$zl
z%+(;TTqUJwnJF7usaU@t2$s!O#}8;T{k`IX{s8zq6FzRx<Bhj@nqu@MxIs|GmM)&g
z!%L^4?`sEKEia=K&65@QY9}*NQc^5hM17^QX14H5E5|E%;I8iObkUtW-YuAKW!&9G
zp~#MSi#X)kK}606i#YK7`e_Gf{HQ31!&t}H11*i$w^^ycH;sw=aPaMBKEwa#A6JI&
zK*Z0Om^Q{aCF|eCKX@}wP91GS8V0^@wRR_AES`REYblOrIP7k5{#8Hj?CMN~PmvTS
z7=PD$>iu2j8O#i|k~g<*-W-sjMpUV}i3j=n8MujitDQ8?9nBE8T$R3ed^?QW@_chu
zUW^D*OxWZWbfto^EE@iy6#(Zy=rG_bA-u9JiLZTEL2xme9Lb)JeSf#~r$fq=9)B%k
z^dc{W;qNh-Jlk`XC@J~h59#wgOUQ`X?wW*JO$A`S8R~j@hEDbCPD+!tN?99;si|ZH
zvgA0bZ`Jj%H4;}zS*a5<VrEwhqeL&Hsfld-@#N*@u?66-)A#4@S>URQ+Gc_U(#eZ;
zruCOr`y{BKz_w!;n|l@B)$xb*Z^)9~x#%_4SZ1KU)o;u+hT6T|IPvOn3XiVB%$LgF
z-S0ScG{Jcckr`YSOl>IdLE&Lx)=EGes%~P71-Ovx0Cjfb%dq}iUC~s}({Df+Emf}0
zry)<kCgi%h^Nj|V8kmvLj}`WN`|=knR-AZqy<41RGT`Oamy3gZC%-XJGU*S6v)KuT
zt7TnzAw;E|gpa@hD=RQ6?ECQ7DK3iw5Po2M)~y^#Pqjq_H^2A^4ItYA0dT~me(u^@
z46;sgLWF0yl;K0O0!?}Tjq`W5kzoC==8J<IUoOLLC57DYz>)c8(2@EEdKGHq@LFwS
z3}^s{Mih*eO>~$WZ+<Lwt|?H=GZoI{ix3qyM;r7U*c}I=L|c4$yg}k|dmd-c+kMp1
z7n<AP*~C<=DT+Vsd3)ZfYl<}Lk`>vSiD11a*<2fc7Y7#p)7qV{p=n45EiU5YjF>gQ
z<A_+e2gwL<2#*xny#hN+6jS8mjn5!CZiIde?!Gd9La$5Z*=8Nd4<4+E!<s^QKn`c4
zLSS1Y6TYWLfEBemy7LuOp724kuU|hXV~r^ZF(#F<B=rLC*9J0*n;QN?kFU_7-1Po0
zla?07`)R)Mze(q<;FH!K;Ho~e!zn1)PBhDFjNun^cTFO|3n9^QstNTHEO^1@@jh(K
zw1~VAR^?`L^y%op4t{22K&C=BJa253jz#J*WT0N6{hKB!Ysw)uRz+8|Rxy|CN}i1n
z#`{#-cz<E=8tolLzTUWodKNkZDz=8N_)rc9mm#jJdu3o|SeO#8S2dm!a9&#{OIb6G
zo|Qu@=>lf$zdj14@SBtlt%U4O#Y&x8Z%kAZ6BCn!<~BC^^dxeonE0K1xc?>W!E4qh
zga~XyGZka?&!5}pf}q8M5Ur+rV6$xUYjH(QO}-rw6nLa}uuZEC+_(NVc^_>uNV-jV
zE@e<gT5K<s8aJJyQcQ3^jjlyTMy{=`kupf^_epLQ>y^J;{qu$O{;63UgBwbQ<Eh5t
z*W!pq_zTv8tPk;{cYz+fv-UOHiF<NK++%w-ibn8qzfk%D#qWReo$*rL$f(J7v{buD
zhxqabfAjHtlCJT<ii_xs&t9sqZKl`im!&L!zq$5czm_Yo!VO!`G^;{ExGBx%W<=L5
zu1+pIO=oiz<M-e;Zso6ZSA(5rNV@-9TAFbkHR|Z-fVjB1x0;TOGt_GRE>3fw#FFez
zfyG}WqmcY9K0EUrxIq{)#>Ri2`|T!7O-=p#_b&u@V<a9~s<cRZ88Xs!i<^rHA6ooZ
zkl4tC7M03Sob6D-Gw9#1kGESw^mKGY>?efulSLi>KJ0dmr5DGX%}Y>=kp7sEAn3NP
z2c^cq%7YbbZCI#?6!RmSrp9Arn1suef&Pa7cHtepyg0-7l`32`dZJ%x2`*0cWK|}a
z*YMxbnXQe@{PUg)uqC2a%gTJOZ=9@>3_Wk(=L9Q2vI@adZ^$cnYO8;w=Au1o*nrET
zjg1ZFCmznu7kxq^Bsw339RD^NTlq-MWL@sFry#Wr4JCa~_p?dml2O^xVj2}ZM}EF+
zaGI!!pYO<x3^<B@ls?CJmqzERJ<RFZ+m?SIof>iK*-{~^<M(3rEQ$H$j2dO5Ng&hb
zN7#dW&z|tbc42+m$_PI>If3dM8bXgHNTtuW+k)^64Gq7~ZU1D?@)>$2%C>(M(bw0v
ze?hstDp8}Gd{_UW1ElHa_vFcwtA6ElW`1crX!eoIZxc~YZGK6z*sk10_wlYdr_Ybx
z{n_(sGA`5I(8vp(+;D4{wuxmqIQMIrkW-4>+}zB*dO3QpX!xrZ>FtC36@yY!0sK}w
zkt2z$w6sWhzZH%ax1<L0hWdKktUE7Kes8?Ji}8MM_r-C=T)ZYqW>jkI5!$^c-b>9y
zRp{}!H8gF_%=BbE?%F_7RKDyWfQ-wI4n|F`_Oh<dktUbBZ}IMY0Jlt*RXV!%LJCPi
z5}GvnXD!x~P~LHo|KYgi?=@9rPQGB{Saqd!YgH8Tke!?$JclD9Rc$XkqRB)egAYkb
z(mwZ%319!vPdWas@~wd4a}8{PAI85w$|hT&&UylmxRHL+>Nc+>;1m=T;~R&~YnVMB
z+p#5sQ+xlZ@;B`-59F3L9ZnRLHB0PwTRa=i2zo<L68a!iS6!X#id)g<H4!HC)I4HO
zQ)6K(DKA;((@5fNF3zO1H0%_l<%8`0A7%vAwY6!z|C*f>XnIj_&dRTpk!;f+UEjI7
zy3#~<Hs*<{k0y?tY8e}6{(PnBE%s&6NladY{**DDrfq4Dvt_vQI^lVC^(cEvcWiuo
ze63OQN~pW}5nl@092ou2-^}p)qPR^SR|%K|y-HCPC)-W9NLnQcbv)h;*l6lq5Oric
zYa3}9QPb&DwdYlCe()b_QhmLH)|Y*W7V|C=2y6%I(ZJe))c$WdQmJFp8Jx!!r8%EI
zF?E<fqj>J3!Y}>BfUoRs#?i_qJ8c?PJkcj{vby?j>mOHdK#Mx)yfG)PV5K`>$ZM>|
z-q0t%acE@<;aoUi4Xd&qka^C-$;rvVv3=sBA{JTex~4-!JNk$132VxG*vWZp%UH|L
zIZRXX;Dp^hsJr!fVgH0h`OkC|+AqHWhRmMs)Sk%sqQ1WVj`#Zg{JO!dBWbYBHq-6>
zq35w>Ql1;tqa&q84URKXsH{MyboalW&>jEto#_U5Zg#A<t&#lAb#E3r)6>%l`5T>n
zw4`N+Q<SpZL|jh4zH(NjgAzC{&$tDpG6nY}E!d#*9FM(yeYmYN7p@A>)*HXQn78d~
zr`^srY8|KQ-kcq72n!42k<jB~V<sgf>6-Yuo`28uJsg)|azZ8Ps-=r*h&!L4Qntn`
zt&TRQ!CtH!97;Q0ihJ#UzN>sMVWl9=fs3DCccDRCYCOS?H|6fa)rRzE`m3(GRY5x)
z==syk?U?C-S<*%mtzAXi3k6OZZggDxSYU;Kw6wH}VTj!+=YHX|Ua*tOaXZx%zKe@3
zfBz{uHzM5c7e_N!L5%07yNf?_I3(y}b{4v_z!siMoh-gC%n*4H#E3hdvEDv^b;o`N
zwSp@kJbFH~%7w~PC_GcJ&s#ZnROw<-J>vkr?RD}9cpy0i#k9vlbddDF{1oF=Z_UCX
zUj2$!KjYa}Xmq_es$pJ_;P8+m%$A~9A+IJ!ShQ5B+^=B@i<a~CJ<E*C>74jS(a}R)
zvCqAVfhdKv46;zP_KCYEW@TJ{vo7|o2#lu!OFos0ADTFc1D&GjF_{S4hj{IkV-wL~
z1!N*@7}9}{p>76ClOA?DR(ICFdfLo~D{p=5&E9e#p^@XDIieZKWRMP3f`m%xIrHeZ
zRJgd@T*{;*^g&_g>+WDPDYdS#pN+0$faPPKuFT*`L68HzresY?_K064+Zh~idoD&~
z`g(`j&_8VupUomgm@!<c&iq3VI(+^B=}Fksh6YK9d&B(9#6(gM_gUIJCY1Z;b*-wt
z;knflc3n!4QYsR$ri64)Pxrhv?R&P5*l3&eYQT~yZ~Io~)aqNO8cDDuE&&N0V=*GU
zwy4yABwn|JqYZQiPrYLum7b84t00;kjK9Cr#O!rb1<CIa?Gkrv^<>lu_c~sIOz@VY
z5S&d4d2m(j&V`3m?IhY2$gWyH&y8!Komsw4ZfKc7KFET$s97FecRjC32Rzf^HA;Ov
zm^L<3Vq&k8B-gnyWSG8Ls%~8u#0CzB&a%VV4WzP6O-e-@)~C&Gw4NvgcQ#+t&Ew}=
z1oH4ZIO32cqYjMhC?12myG4x<l0Q^r(akAQzZ#~|XKD0!IxrhczRN4H#ODxO%v);Z
zj`jjK+rZJLzr*x=a*-r5P#<p<!%^GI*W(`<N_y_o<p9l!CQ{@h4&O)J!e0CkdRfe0
z--OpF1+g3Msz!$_NEvf*avz=D^O}O(yz33I+RR8&&<S5uoT_T#u4g3;mb>TjS}$J%
z-03e_K^=sjpAS*G)?mDU8%<)B^BsL_^5E7&c#y{YP+Phlvu^R7wc@vupE_@=ATiQM
z1`5vi9$mYueIQq&EUI{HnQqlj?^WZ=+cvRHZ*{yV!IATfP}jtrtt!p!k)}asu!<0C
ziVnvMn9NIGb?F0-iJ-^O9wCqA8>OuRbs=0Y-=wkziCL`e@W};&(A#(T&<-{*PF2v$
zpr8HQH|nAv5=fCeZ3T!-A-fK{^~=B}j8P0G*TRA}Yb>zx4@(5LGiEKfL)8<O(bqN1
zC0vSp8ejuOMjJiS(oPTH;EnkpYqvc1iNku)_d9F9dFVzf6Mmr{kDPz9x(<nH^CL=x
zUY`Ppt;SI>1Lm_FU-i(3$m;$z+g&{9SIy$m>MalMj`#<VAS}kG8>$hTj!z%62AazO
z)FyR+p$J85b8=o_%{)rBr5&YlD!**u8B<0V@Q8eZu0aC*cnq-5?t$Qr?0$HM&9`QQ
z_pM)9pzJ)Yl^TgM!HSx?4wTw51rXKsf;Wxy=uaJ_({kc${Q~0%af3jyXR2l!1b<1$
zy$qf@7!IU!tI>Rbfn@8@1?%(A``RC)!i}GC@L~z@dkwLm3vUeraE%IfW^ckOVUtFi
z-+0WaU%;Ro6b$8O60HV(Lk>+dB!}!sp#2It`Vd0|u1ayXwpFJ#rELDxx-VF$0ub3y
zz<VA?_saB0urni2Oym?&tp2t{NVzE98f=u8fOl||AfR7v3T_>LS6`uv1QjY!2yA;|
zw3@y%mQD<HbVgu=wf;=T7Qv6WsftWWqg<*I=C-P-|6ghIp{4EJaDQ>)B*t|4eLtS5
zjd7_$Azj}A9R3cIJK$p|DTg$+;W&FBc)mS&{^NKud`OOW@pNoMXQHFOk5P>|MONRC
z2miR>0}({x;mgqW`<qEL5F{?JBVs?sS2O_vRm7CiZr+JO`ri$-wSI6dFX2N>Sd-!d
z==;6{j!ssT!RQr=k22MHPzf987M`TL;WT+&dJP7}snVU77r<`(g5PLWK2iP!S8y~)
zANJG|H@!aXZAHFk%j2eR@ChR`ojgoxsamm}W3;7<Cp@75>S&nB>xacuz)_$zfS5H!
z<oG3FWLGS#;o-DO*6~N>r*AL<#C5-sL*1|KKgOjHe)_xRNQ|A}^$Z9Sl=k5L7jZ$%
zzu+5TBYpmCKf7`8Q4|mvLEn9OOQhc8=V1FJVEar@Y!UCkyL?x}?TGi28uM6ns{;LW
z1U}I_*@KFdg8|5mx#pYLJN?q<-j)b>LSwPIPB#57H&E(Nz5dTTgdWH6@sL;~+olxK
z1%Mk6t(v66yV>yijiPF<#w(mhK%we^!~6c;3PF4Ewp1Oz^@b-xb})qk$(@)%8C~*>
zeG!FWy^EK6{x;9#1CW0VTK>)1ho}h(-bKyL`%iUZ^v7RoNKJ^a8$|V2D*#862t3HU
z%-xfPUigNZ{bIy)6*WnJ{P!N3zg0@$Z15=9gutSH>Gbp$P>c|G#j1*%ila5V2WYnJ
z^P`gm#nLezU55zP(ft^a1P;ux&E34kB(znAfNM+aX2TCk=|{F&O;1Av2d63t_t|tQ
zxw(?C<Wz^kJrR%r=388y-yodjlLJwMV62!6znYbyy}+>IATl+`6xY_JR6#OFK#R!*
zh-4qsqDTt2b{UYmyy@Z~e!kCTT~UO0ZZPb{iz-m?0wdKTlG$AUU>`#ifP#T{eF1el
zM*KS&NX+j=Y14Gvq0tNH7a~X;8TAoL3(p$q-P5T5dz+lYqbpA#__O<kLU0A{<w92m
z;YgMMd%oKAc0z!yMSDfTRwIQ`fdfLvtkRKuJO$s92~|kFpn%eRPfNc$0l<1bjfyTO
zpgjl%h-73LD2dJXd8%nC=$=Ngs+W6*hwOfV0s`^c;c;<b<~2mZ`Mo>8IrCA;v!GtM
zC)<~SEj$6Kt~|#5&QAFo3LOH}7~Q<l&b*YU|D>mX%oYa&#@gpURl-c%yxE}Eyu^_{
ztN;!I3l#J!*2-<4J+Kme9?l=&*pCP!!K%JAyrrp$gA<;?dKwg-JAEWY#(1U;NQPv6
zIkfW9f<w{GXO5V7u-0;`i4<16=Z3FX<1~>O8M*|2&r)!Fj}Aw7!EMW!S_8|3gv3#V
zkpb#7A5rm!ssDhSq+KfT(5h(zR8!_T@929<O7;SCv+{?x5PAPPJqHL*N%gPYItzHz
z8p8a|?`GIplN4mBQLHimisbHIieeE6Zn)NHJygbfoiVH21kp9D7fw87UnI#>jg1L)
zra5blGqS{d()u0x9d+23N`E^bBh$fMEIqRb7<M#pVzjfQ@+7!?06g8}bSG6)lG|ux
z(dlCo7S4QG4&ftr-AZ{u;LgWnO*x(4)T4JYd*i?F*0IC6A<4);!Y=FJ@Q}^mam&Z+
zYx(j?*Bhewl$$H3W*GXluTQNKncqlkG;RE)sc9y3!){BHq$khm42o5b|B#S|2jSC8
z32ksYCnnHMN|yCL6I=9rcqg_iVQ}De@U(-jH~yEU1FxpP>vs6_a0INB=e)9w(e_`8
zQN^ro=&+42zzkypg(BkCj^WPEFD_H1viRvqxcN&HSPnETi-@scMRV4LaDm~MKaoh)
z6Ay?<luD5m6JvT{q%4N>!JV(9O1nB`q60|q(A<F!7A-F4e*PTfn5H6W3ipftg@9UE
ztx%A5$mw~AS74hd6B%Us;N$*j@hR`p-U;?&0b(zA=gu3!vMOP}0!<hsGSoJh-@921
z^XCfPFU;nYV<J;$(N$AZ;~-Q2_xqO)H}0{A*exPXyrE*873x?PJ!4J=acsw0Ug{4?
zSuNu$HiRLmOn0?p{_S`W;mV!jeu+@$$dfNEKF1$jI{-E8&x&GsKJ+$M?vmk$0UUcP
zrGM!Ki|*mYDpN#?wttxwSAMc;AsvS!hS=gB23f#Fx{#kAev)|-7UGN9Tj9YB+k~Ny
z&D`iD>=wB7Pl*;p=e2c2^oJ)HS|6gsS75^u_uv%d99?Pj`g#O%AhfnqBmS`CVN$yU
zb&SZgaw|itgKiBuvc(bdd|%3~lGl9ktk4-!Jn$7F%Ca2z5LRub^m61>h7eyyQ3M2#
z<1iKofhcdc)>5f+^Q@Q5O?m~4Oz2@0%x3L%kMJz7xe7DYAR_`m3Hc~!u~cn*LA=)h
zlCivYm4nm>t%71MOkj)*n;%gsTkS0KBb2Cukl+y|@lnNVwPXv8Q?-9<t@`~f>u7eI
zf!>e%6#EB-lPqzzK^bxAz|RSu73msx=QySV_}gcvl0-MOhkM@s858(vS1Lq<VdsuO
zv<L$;d;p$CX{bkz>u}{OHj!8)aZw3-iP8gX4x;={D@zfDj8gK9RS`bgss-R*o>I5B
zzig6*KK-y3h;7voyqBO3!<5;y4juJYbg_SfcA2WcJzMZA@>O^AEaB-1t>cFzDl|>R
zP1MR$8)88fJi^2L75YgDYn5*#t-z``flGTJZh#d2-98_;jAb*}^7S)WE(!*hyGVrX
zjVXzzblrJ*oo1W8T-2Q0p`={Pt==1E6-81MeJaCqD(J;!;z4fhHlxE60D-|<PT^k!
zklOkTUp|#d6U`R4cTdb4!VH09qUBTMOZqvW&u12(T+GTk#>~?G>Xr)f)BdNe!}J%{
zFVtFwI|u+soAcjC(VfS={&q7Bg~_DIy^IQB)xNjz2PPQsD08!5*%a-#K`v>t$G8;J
z5whI=a=10RL6Z+Y#KI){2TKoTjm6}Qj^@^NK*l4L-^65cG%XQCf$CB@aYZ`f?^j8&
z!b*6X$A4Z&);ZHf%{2dS!3s3i1lW*PYE+H@Jm?m8(Wg<?(!jp}7yP?kfn*gW0J~cU
zaryzX7;x+ZGQatip|4L_MLTdHuO@Zl8!~R<7Tzfeg_8NLn4j3l##RoamX*LxD}W>G
znK#Hjyp5PAKq!{F88hoV0cvGFN(suTYsEwBeF3^gp(gTe(PQK-E$gEh*XNV@ELD?U
z#9`3a(T&c&02E%$iQ2=>9rUfD61GCvbaLYD#>qkd^mikPyr8X#3}a>5xi%01EhK=n
zIo)|kTpC`!s1m(U8iixXzPSN+#!w<Ycz-v8DczzHb6a^(aq}FQT_O{D;cS~v-6&mN
z_+v5_6s#>^Y7l#NDn}N@&9j6~R6)>Bedk2o!!ix|s?%vl!2;ADsfoXjp~rM`C=P6{
zv|YqQuQUMUO<^QTPISTk_1`iUm*O{4o2SpkN9W`q8|SAl;JuLBn<YjdNWfQovH4v9
z3mbt`(Cmxn5Nsvt3950fro_-|^`+mQJ_$WS2LZNsajMT`IjF@~9g~sl${B5Bif-;5
z<{x@xU(-N{T%*G}jC2MgLs-|ffbbRhgl`PEe)n@f3+Mf-{Z*qopUid@l1Wa|xjA(b
z`o9pr7DWbXwO09EeK3q5+btc=#2|&=nlseP{SC4v#&-9vAEu=M*4b&csF^?(#>aC=
z_4QAakp}Zfj4sKd9A08jN3afRmiO8j5Yqs$;LZ%yD-%}2*z#qK??7MX@Y~IxNc%W=
zX>C2dO*9}7I2M5Ua*5>ycfb!=sIVWa_aFive<{akJ;zkdRKqh3`hGKB|2Mc{p}SGA
zuMxfQ1NW|%sxIT@Q)B%xqpXU>B`-)8zvDsF3UflX7?2f5<1|X=jVW+!G-&aF?BSx<
zyxQ8->}CL2G)%K9eT~TQEqpsB<2pjhf$INjW^jj}4I=e|R!<N6$Ev}yOIXn->F<<Z
zKKX~983M0C4D#*N`BV78?XL!)-lWXn(=A+`d?gU3Jq*37RdtXiFs)&0ktxI4&iqdE
z3azsPWZ+QB_`xhsd@T{O6;pTcWj6pkD!@0$1xk!RgzaB&aG<HV?I+jzLyHNnro#qt
z`nP<H7GDJlAV^59-Xkb?2!xXv8b0JT=&K%RyEeuJ^F>W<r&eNrAx+6LGMla^hqY2w
zCgRTZN!Q*Lq-!>}0<qk(9HZUBID|h-BiR;F3gA5x_Zrfne+E338ra_kHWT%FfjnJ+
z-~UQm=hLh}=&QaEwh*IlC=}7oi9-n)W&h4}aHs)^vkN`?J$cF8pTng--<O>R?i^C9
zHb@LQs=Iy}9Xf6WGDBq8%ic?0M>0qDCO)9PrhTSY!ed3j3KfxvzgFntMgnj;!U~&u
zLexp~`yTcK-P$>E`h14jrrg*}qN43fINQURk-I~C&ENA^e*>@Z095&3WtoY-6}sBG
z^p-!xLhVI#%ctB`4kg9>wsOF}xP&3Lyw}0i58nIP1Z*PW7yN(@@zR~g#ypvD2L&6b
z?kJ;_??Fg8!E->|u@(SHTio4u-yu_K0r>d-DZK{wmPT>9>LwlH`XqH*FUbDe3n0W%
z;*%aLE9=xsa??zIlL$9EpA#@<%3t0)VlWLe>TFgo@?76Buu}F3T-jwoq^l71_{I|;
zM{oSVPwVHmNmQSZV0}omy<`3xYitwl>~9OK`Z3T6MSO;qgGR+vct_a4MM+Y%0>%du
zQ;gZ9yysJQgr6wN%jekxiPH=^NKaVzP_V|tx-CUG)Oe_X7$!K1#<y@gyYn;0+FxeD
zT!lVo6H_kSpwJdb!%zw(;i|sc%Z;-oL|5zst-xpAUb7fB;Z1?7zx*A$7#8!K?8n3o
zL`I|(eE^&y3$eyEo`SAzQ?viTlU~~a3(<3bDxRWUmX?}AXxcJ5ddH9p4NsYagDRo}
zY<MRnB?i8)h+(?bPdp|pLnDXb5THruJ+Fp?Jh$T8=W<C(j4*TCL=4qX5}D+ix{v?2
zK6?w!Zuwr4(wy6+!=MO-`DrLzcmZmi<gFEA<%9b!UrSWu3hY|Y>BPs7315aC-+Vtc
z4JCVS85w$rC3c@*RF@rdY#v_Dzy%yo3;KMt-;#@Xsj&q*KT!q8-T9*)ws4OFPenY%
zm`ts&AMZB}5ySTBA9|XbH{*1`2`oPacLdyRkTtWe06}IXCWr&`y*F*nKW@BhBV0{6
zR0%!>uz0`p)%m~Pv<$VScQ<ZZ_N60yPT(18qKQj)BwUsmua3L=G9)+L&OY2X2Xqr4
zssJrg<r3@3;l(mqP8j(l=&^;TT4fxeVuHH3j)NGBBt6_JR9(n2lg*$H1<$gr{46N}
zEP{B@YthEh(WLxk2cV2F^h!nmM2==0yJ8psN?h0hPmeu3`s=;r-g0InMvtp*U@!qF
zhrz)L*R>HOU&H$3%;lj{l%G15aQJU{Xc%?5gKXU#KD6>1!)V}x=YW1GX$IdlT5ynk
z5Zv+Pj_2ZRTacjTuj3)WhNZNeqe%Ua<y>7409db6Vjz5cn}MV&H&%+4B=ouDM%8eU
zuJqn7LEDjH#J^Nqs93-9-_UbuhPXLWn*g292AgCzSw}`dfX&8&R^jl(faCDV+lkku
z*q^il)=_W{_syw~Sy}$0M(z|;R9K&jiyOgdTqUMY57uJuJd)^k2JB|943hwm-lRiA
z$>-1b0cwSY^)4%;&4<GJ$yX=+{xQFg1j0<O+<9tj-D<PM+Uv%iMdBr5S>zzOm9tLH
zufvB@r4R!GTmiOeU%L2?ZJ33H1>j;5EIjxA^#(n3pK11m`otEZjI~V&DS5(twT-Zn
z`WcDeB?i4D=I0?u#x#3+145MG{B?kwbGHWEuHk||q2LJrM@xF5qIQShzLU+KpPw%`
z?>dxl6rhGUaE!U=J|!2i*3XJ5ey@Z<Ki;@POvu4Gxv)Ms#&O+b(ce$NHwmc>F|Iy)
z_6$m7>+tCL*xA|nIMW-n|LlAFA(MHbdrv<I?3-a@Vu%kP08*^O5^-*J<1(|OZe+PV
zop}39xKT-0J7|Lu6aUWfE!W$Yw9%MMSd1s<+3HF|tj}RQxp+a3pDiBziL(%=C$ZW&
z%imPeRExi>pImv}zf&Ic^lF3%H;BW=i!VqkZ>1Zx;{okA3Er2MW(0jW?m7H*=3Ox@
zR~K7Q^W{Mq>4iRD3O6c_Fo28z^S!HAZP#u8<_@}#Y!C1-`BK~_Vc`lL0E^mgyO`!n
z!JxJhq9?*`**=-9Zz!R;^mb;-G@mJ6Cj(qec@tShWIw3NDxC5Sku5awb4O=_kKp<x
zfSjTYD_~98rTRK<x9KV+DoS;u4?4NH6t*2{Jl{pe-!*#ELBXEs@sglR{Ath3jIV?0
zder4xW1*_Q<~DH=s*CCLZA-V)=$MtsI`k3$g5G{2THiiZ>+gTz9RMOeZO%%!MY-O(
zTf^vNgNV|{83_lI(ho;7{{F*XwKq04#Kpx&jlDMjouBes;FqOyj}-zCs<#l>xnFb(
zs#N1_wf`-B6uunQ)FwXkXT}af@P)#R#alkjZn?1sJU?&SQ}Mz>MN#5oCsuPM{7kg)
zZ$I^s%22cVz1>9nos1wjQF<&&ae-{|o)tmM3_pHgc|V^S4ey~J4H0gW@lcjZSCT~w
zBS*E+#djkutyi^^(+=)sdw}F@Z(nE*99r?&{mDM#f8`4iQt>seel1c_RlUs+^59Fd
zLAI$nekg>mXm0%X{%zSF9^Fzko;&aYE!EEsvL07d&{KQf`Wp$9l&WHb(5!h@*{EpJ
z%3WnmmXvY3zWvyD#-Mqt&9uAXsHmaA@C0A<BWMFt-k8hN<RHJLa3A@XKdYZAvfDRB
zIIJS1K>DGb2wec+tc{ldeGGwK1<UQi+Un}*H(d<i3ove?e?1ode0eb>)MrN4&decb
z74p}=qy|s!x7*qYb-87F1wI#}_`9-9H@IJ>GWkV(RYyiJ4jSzG`Fo$-`q?w$06>D}
zjh=wk3g~lD5#1tUdV2b3+TG;Ildf^Atcz~eSZgjvjQ3arBsaShxpl?XQ|?+faOy_s
zy&!xHF@ZY0^=<?^Y6(+}mWaVpP78!6F~Cg%i^uwAKUs6E;#oL^a9&c&9jLP&$a$RE
z;4obeOi$eT-}xfD^nhtb1etv^8|iPaX1&R02$o+6**Bc%e0lQ_5@V~UuJ8Dm;jRVX
z7~)vq2X3&Z3t03dmX?-w-SK|<C2!~Kg7ne8x&ZaVi~XU6A{wAi)z!A6okxlO3(=ww
zk5RoAZELr`WY_Ze#?l9|v71r*g*3TMAIA903n-yi`lZsCvx$L9ixa=lSvmQWwHKU#
z(VX0<ooNQd=*DF2=$}9IVopqC_*yzT#=Zx`JEY!+<5qw>1zjQHZ|3EpIj6epYz!<>
zdEhjU7Z|U^Fc%oRWg0H_<x7da3{Zl?hgbu3D)-@mz0@^&?=SQSiSsEZ=OwBqX(`jY
zblmQ5tz%ZNCED3;`Cpy2Ny^hd^3uBSEN)Y1;3dT3<FV}K(RD3>gJcI+pe_AH&M3xl
z_r{LxIWH5aRUgv`Bx!1Cxq#D>Tz*I~_Ts$ELZMJ)&8IdJcu7@ym5vKCgpoAcBM?~+
zceW@KdEy!=x0U6`-YYl6Mf`sG<n6S55CCaW@foPp`{tckqz|2ze`jkES(2~l>j}{h
z34Inc&@oA5jw6)F(9_3}IrLcUBY~XZtoE+gxf*$b-Qq;0RsXd6%p<?^<6uH6K&d7G
z)#Rkn8A-@R?KP#!S3R*pJg@*6jk-}TeK+wF=vQyA#^njdwLr8g;%pDs$H8UzU0ocN
zwOpP?L`10aAzt;S3j4aRWP6m~yV=yUB$*pm#39$Sgn@4k>VB|Bx?SG@A!fi4gPL$C
zI_+WmsaOj1%@<J2+&aY|A;1kV6{vIDm;h}xjm5?5J=|qJKuYZc&r(;YK_SvL0$0e@
zD-a5Ym|`ATCUfhOKmR3+WxbELdVAa}9h5?m`Li({;kViUI`=<n0P6rHMjz6w9UJ0H
zSwk#kN^Ujoef|2?q3K`*JebWt+wPVwGU)_eUUlEcdQZESWE7?LGq&pK<l^iEn$&tf
zaqlLYvj=2!i{ngliD50e@dy|mPr%5?$P-3im(3}}i-M;o!)1OHGbJusk$sjoE)Gx+
z@B<afUYKcpXT?{+dE_I7Sl|daNmg2%N3R;k!t09w%|CnS9Hcvrl}^;5@$z)d+1VMx
zXQzV-Sn^Dsa>Bp<tgGXN*w{nav&?%pFZNNi<^uSv(s_6B(TLBKI_YfD$w`m&j1`&@
zxEZitEt_QFxi(UQZt|#iT>~vO3oTU^$mBGK`Y$y#hoH@hzB<WRwu-{$y#1wm^7Nd*
zgF@gF#6$o<+vx=j-!rfNiESD>xj<Xucps#QjpY7Sq2K4+XMB%<<#>XIfJ9z*a=`kp
zSN0p%*nOOKFlq8SD9rL(-SN2NwVZu;cnE%ws5b8=0!WPh%!_+7G(<n42VdW-GHZ2k
zEw8;HfOx(cj8vU`uQ<;hs;>K8@NX6!<C%mRH>9#0r|Vt$Th4bue+q5OOKXudbRLO9
zrkDVc;Nal&{4{iIas{c0^yN;hG;kfD+v{T8N;>VKCtM{YPT9)_;l-l+L}1TBR|~&`
zP2l6~k?J#I6`YGNBZ0#98_-|p>tv3N0daX)+tRe#!+mw+f3@V+V9vg`x5r<<oP&Or
zQyY$_r>AdLwP{-*Fm(rTNfkVflQs6_ilv#&us4K}4B14J8-<}vVHDbf*;w123ZV=q
zCx!Z~H_|BffrnO-g8Qx+&O|Uw+SRbBCi9pypMwsE-k(2d&cheFVs%Nr78NDX%PA7<
z4-p98mX!YiE>p|}a0ewg+U70sDOSU4^x4~ZYPw3vN&h~B1#@)>q$WVKT+3l~8Lvq*
zy4x&-xhzaMejn%t)(1&<hh6jThvmT?bq0LiF3Cj#3zi5d8(s`&su95=!{OoJ@Mx+;
zGLZPMAO5OdERsOy&Pi-%1D7L)j=Tm{N`<O*;fcdGO>-f1@r_=%goi_Sa|9_751f)!
ziz5_DDN&}`)w2hEd=TA+8k%bcZpZ%Cn-P60Vpd0uyt76QmdzGpJG1+HHsG<vm|j?r
zk!>aXpO9f*x-Bab;HlPsR7Kz=Fa#WU&n{m*XXm}1ByJOrx!}9}DZ+`(FnHOtpDYaw
ziK9W@(}RIOeW%U{jX62X-)wCl3B6nN5WQnpHadg@0+nnC{;DHgTVHPk|Mt7w%VPIL
z6f<)J*PvKbYKqIxgNcS1WmRv3MwmYCZgdT^Azs4(XzT&&gwoBuCZEfbB`5~u&ks_D
zn!9;M%Ll#+8obB~N9{q_7KI8>@3Wq(OowVNJi2uf?GPaUyqh=ttAOQSobUB-xmM)m
zy?S5x?z>sTGzo|ZECWc3LV#O|h+S;FCN_<d`fiZT^-`2Khm96thXXp1w<OX<wRWc2
zS#k!r=ej8ff+r>?y))Lm3rwWX=SXii!Q2QAD<??<0Eri1OvZB>)yVp|_;VnWOmeU_
zP+c4xMGbtVT}ii4qHm(0ER8m}@uU>^FYiqR`(gjtRtuxd`NvEC+|ho}=LkAk!j1X3
ze-1?Z{x(_U`m-h|ociIZtVTKL^JCb+KQ<FWt^N>CK7U4{YsgerJ1NOKT6{<bO5#6p
zPK&gXZtD>g+%K=RSai>6B$~%EN8Sh<#uIr9V1MEeBt($QS|nZf3swt7g2FT1tkMBq
z2ibj-M02y&e^d^KE}lrAYN&MHua~z%%u`V|$THuM32_6hFWg`AGP0oeX|(EsBmj6m
zzz~8wx;BqBED-hISy_6#2HP$Mtr~{!;JmAVPp^C0#L61RU%2>#T<*n#m;p8eBpY?b
ztJ<s&pm>4r7l5V)${;4tk?Rb^fsMFy^7t-M{RHF9#EmFeQ<QoFegyjw56|IL`B>7=
zhKnEg6y@E+pXnD@T$~AaF{*F_y&tZlS*TpCmsy)dIwhJU%@-SvpsDRxP|gMPUftV2
zUjO6V^|IgpYRo~Jrs*5U7=c+|aIK+@!#f^1GANyfcgeuf{Z}_q+dB#@6S!Ga7@klV
zv#nrJ$hzNi6}7D(r~sCpxw<?`3Ig3)df~8gLAf8@I6qSQ8DAjt>0j-)vjjLPf7$G2
z#>Rv{X#hC;?9;S^(NH?5xK`5P6@bKZZZd1>gbzKZpNs6-uxIt@AZia*p!kt0kADhq
zHoK1xkvL2+%>w!#Ba~Engo(YBGv&^e04Ar+zjfvkTm6uii%XMa%E3M2i4g66?Y9pM
z)i5Y;u-iUd(>d2;{wbi`V5Zdo=36ETk$Tf$rN67q0Qyqn2*9Tq1udVJ2?8JZh3?nq
z0h1FSw?A-<_bR!SuxHWb30C8Fj_%@6qrE%YnGSIL2v!i@-$w!XlzYe)(U~YyC^rR8
zS4qHnX2{)!#*M0!7Ye;WTH3JHY+F0I)So5ox~lcqRJ8zKoV*0JTE?$iFbaQMn9!Ao
zZ_iuNq`G;CkMIOY+b-e0gcZkm>35p<uX4K?alzq?GzF2>AgJ4xA{58ZpOk}E+%GQ*
zjlEX#M`6yr`zu2rVEBXv4bf>_OqY+*^u(jPTs`yQOtqk{S;aAn8zyMIP2xZ6@>S#k
z+6m?{!9m_=p8|vbtn8O+@W1jyMGdW-_qjLUE~>`c?xb^sfpGBCgj;n#vA;_=9N$v;
z<ROH3ICz7lIo<Y+i~Y_uOgpfQPNzUq9&7AF%vNim7zRll@soB!ej#CD43G1jF6p!N
zO3BSStX4cmpY1aL3+|sti*T7&W;eTdctS9!6_|BXc5vL_`ilh}V>QxHA7>ROo}=5i
zHZ?#_<Gun@(A^ywD>nxXcnzSj6txJFBcj=yNWOaDvJ2EQPC=7nhX3UWl<E2JBk-vV
zi*M89U$v1_LGra1ar0cug-a-%k5Nz!Vye6y^oQqYu#9&ffHDE2Zl{xOZFMzK(67aR
zp)>lrhV8mjV>kW)0{wWZyKsFLS_GN>t((afRw4?eQxDrTAsGjIMKOBGlBX`guRNQs
zF<_zkyL5;c(6wwjQT4i9(6H9w;OqjT<TW6d{8B7_2c+WY152UyV$}68-{aJBD*{B7
zBzRubAI*{sT(cyrC8ecm%ekLEp-~9YDJMBqRaFT-F61s~pIvGJwe^a}j>k_$zO-K4
z)}_6UzQ(;iNvH>jWLHb4mK+}$!3Bap^t=2_KDmuouhd#$r3xx3=K31ynT)k{*|b6W
z`5839FBH(&_|Oo9JJ*Kk$y!JM`Ed4vrynZ-){$tU7N1vCTEBX_R`&da`zYRSF;&0}
zGXV6DGP(T&s1;s-@ew<ygD7O|?d{`JkhjWpJ_C@bcqNFIRY&nuUg?b}Z!#@0sH)J&
zs_)9dS3#g<gEdsf9@*SI4z&%_o1n@W86FnT=(8<r5}{73=@tU{FHk+YRsan(IZR9F
zCuf0Pbx^`!)h))WzKk6Qp8ziro1X5kC>!&sg1!g|h9BMjSH<I|_Z2^fO*PX>BXMTW
z0g*CmlD-V?mbtsPx9NC3GD&g|B<OoPU5rJKa?Dgj;L`vs5=rY_*hzQp168~@fRz{m
zZrr}dQ3=b)$N-7TM@je6nJp^pInY-Bnjj<{2NQ#SG7)(*;?D=e0}0oVL>mC$!kq)A
z#>o6H8*>0-*&^AXMNJ9lpn{KrRyw-6tX?yj^I*Fdo9->%?Kjx<c#>bJiRaIFm4S}=
zzkX*Mf>wQ1dOV;xIg*n1^lave4_UE0?C7&V=hbv4VA_ESyX|Fp-j<<~2tYbUz|oVF
zlQ97r0`2bZUT%j<e|UNioy7x<DAZhW?P75Ml`by0f?*ZZ9U}eaw!wJi<nyTibU+xo
z9aN>6sH>}g8pQyyW?wnp285cNr>vlD8nA1t!$r@<QHfwE;1YOf4o=Q-kT2hU{_xG9
zGQ0iQH|E1cQ0*qDNvw#rob-Zs8SCqFDh~n|*5Grp8_MXX=h-^n5m|=#M+>IDcECw4
zL4y3Td8hEx+Q0T-LihEta+~3=An#ro1-jeC=wDnc!Ue*1GLL=*aA3d>2%Okjgo7gS
zdHw3Idf=yC76(d|rXx8BfQN-1&TT8=xnDAa>m0AJP&GGa4FDJdB<8aG8_Xjt9i91~
zAVEiX9~K^oi<_IV+qeZ#NsysiFip_RmNsnKI5=GX6Kxrvm^cMP9<?<!mjDaIbpjI%
zpdTKRJ{^b!BN3orM<Bpu?JvYiXEr-B3o~wmB2-0bJ-?==CObQO>qLW>J_Izq6F)j~
zl&Ek_CM4rB7v&W(Re4SqN7zMvncIZbdgbcUF`)=3!Goge>VuNnzo+B<ez>NcQ8bm7
zy_nmKS38VvWo9RUzHc?3C5DdM)u3$|424VsAW*RXVeZu1ItwHvR6FD6{rK_Q-EUjI
zTi&34F#sa~a&q565u~_F_w%^P;=ywvFT&2h4HwWi3sqjK`BV8HGgbJr=hot*YiRU2
zM=F`mh;}BWsIl=pg1gk(#-?K24orgNXd?CxCmpfs4n_<YvN-VD!HC9fXl=Aiw~wQw
zI1}Vi$7l4T0G$QF<Yu3t(l8HI01)f?WG%l=tjK%dW%#vqf2qgG^2v;JY`o4WA&d-!
z^pj(wVZ--^K-U%M9?Np~KJxax{Y%^?8VryKaq#o^lhxvtgHfZ-Xv1%&ZTx;ag!yh$
zTSuZoLdyr}kag&_8P1*=RyVRHeX;T!sF<VF;h6c=p7fP7$WF7|CSF@*dRp4Gq~A;q
zBt3qXCgQs!IfEX1QNQCe^)&ec^kR0|f201@-W{AypKTpAHNaO}%RC!8Kz`N}Y);_r
zS&z~9cih!nJJ?GOkmJ3cbHP;2pnM?S4M<L=RE`I*m}~uCRXJdZ(CXS+lE2>q-OSiG
zv+reo2SvnmBE`m_u3H_3EPGQA(7^?GjMp?5t?@>B55Xq%tulmvG%Coz82=pKyE&({
zSYi2MF`|UGr{d2bz5ivPwR`E++I8{FR?yf8Bkurpq0^K|Gl4~I8nEQCkHAIvf?5yg
zEEq}{>=m7P_-6hG7&=-8g_7@DNv)#G@1Onfi&fngO*gH^{Xtd#?I6ZKBO?(kYu|BM
zyKE5iJxfXYe9R4?;s{I99xP=5Q%Jafq9QX)ta$;sb*OILG(EtO40k>M2y7_(!N;fC
z+S&&P2SD$%9ZaKuSEYh1kZm$d>Lw;&gwxj6mXMGzh1U>7z6x~S28!}ppc4ST@8>aY
z^tc<EVF$YAaP$D&gML{n%8DFcfr@hBURAxU8AREn#*V)l#r5c2k2UG3ITEoS=;RQ$
z;Xg6q8^2#O?YHxRMOg8Py!^*iu%6E*V(E_WzJ`T&+J_`=T}*(i3qoS_F#xkjeR{89
zIy2$+7dlFX9T46PDlMZQm<h$lry9Mt@VvUbfc$_Q@~j``%6vM)jSLeB?=B#@8KJir
zpXzl)fZ(XFIi^K3TpW)9H^CpRxbRiG=#IxMMm}N6&`RTc1UHay?!<(F1%G;cd^~Xd
z#}}7qHb%`V@AhJ|@0s6T+6*Ly#^g+XZ@2+;yU_KQvH}>=$@R$sGf<qg{TyWLkC3R8
zyJdHoxmi<&?=7OR-W#Sm*_E9Tc9L^R`;@u5M4uXR0UYK&7{Pe{aOd66SM_I`4TJ;)
z_nv3Mw6yl;Bl(k)k{%sRIjgYVfY2Z&6Mm6y86O{C=0nfj<DCWe=b4SpOMQZXatNVw
zP`{Hqv_e2gD3|>km?&T4cKd#`_?K=AAzDzfW4zkDlfe2sCb=J|L`8+1{L@!C+|-G&
zHJw@f_o@F;=s|zZF0$$$eT@Pws1HV~M~xdmoaZjBdpZ9jytlU(j2G>K2^}yo5_7y#
z1a#aAl=;XM{<4yilCrX+vkL$s!>VnD=^Ptx&V<f`@si?_l1CofpDNr^?kW3rf`ME>
zBiR99oLjdPE#Jgrg|3@nJa@V*5fzTl*F0n`D4M4TWo&GICov3>&ga|@PRrnffRHj0
zvkLo;!7QHOE&Bi`5OW?~o$m!nTaFa#U!ApFNgkU6da42}&XcW5Dp%ni@o2Z36$H<7
z0KD)a7UPe5n_kO20*$G0+=NF}F|3_j%XDXn&I(2UOxD)v2LN_+GrI1T^a22GrqO^G
zKz?8?eiy*!)B#xIX97DIFeb#B@~nN?5plQd!_MhGs%JO2<0~kznpj7Foi~=>GY^vf
zHa=%Vh^@NdCIgb?o(DhuI?+f>`GS*JBfwds=|o9(Jrw!GsVB|vYa)|CGhj!)Kq}cb
z0A1c&{IGa~QRnn@;KeiETiaWurCZ5KU&Yjbo2~+>NVy-knJ*<J0|1GN_z0~KO!zMq
z@?DOZ8jZ*QvXS2z3S7&{X)jDQfEq{I&=Bu9&v?H_3Nk7FO%x0GoTNEnWQZNQMMOLL
zJ03=NG;tHl=)A0L49Dr+6_X63$ND0#blLU5!>Apq9?ZX-GyszXf;Hd9ti1cy$wtYl
zn@3Z2UFsETFrgTAkA&~ygXj@gTF#YJ9nk!CYR8F7s$i9^DH|aYP8NlW>X6K3{>6g-
zr779quCl~GI?l4FbVsqlBf5t(mS^sT46U{s8HG00+V6OtOjI9E%L)(ute85x=<(64
z+Hw%1@2hZpmTZA0l+wT$m;QF{=b;tcdhaL>i!aODbKlVc7?klXV0saB7b?dmnxN2}
z<*=>^04?8X8Cfri)5nalJM8|NzJrpke}F`}NK7^wqg1`JMjcXc{cJk7K||rPC=R6f
zqFIgo-FHp7rXh!|gV~||xGy846Os<K`ige%zb7PduYfFn@rag=25<!{f7|3J-Y(C)
z)-O0Jiy?YdZLmORcdJA7*`n8W8=h{l-oVinR|U`2>1d1J>|4Ce`n6K9OKG-IT2GxK
z;o!1+Kh@{cD!N`ie*CzW`Zdzn$huPDh@%?@hV1M_(#FL*b^qH75Obl}j!LpFrh4d5
z#DR~~b$(QZk+7Q9ynlLleBMXz{49p@iqOT~otqI(8L5zBwAP|lIAkjDq1Un+FJ46|
z3$S1qAg2{pbR_O0ai}#U@0gl`p<j9{Bct5u>+kw1o#*UA=1Wpw_dzQb=@d};o*j@k
z6g*2iTrDc=Y_=)agYk`mvdcCFnZLf3@kX;jmKGn!ysW6PPOx4_`<0MImrXH8L{m##
zx$qpoLF-2>%dMZ5b(shiEDwa-gVxeZ8ExS9_iHpifw}&LM-ex%v$Kx4zlXnkwJO9t
z!~dA90^BX{J<FNcochH>x0T;-<C?!V+2)mf5Rbo@`wLiU;P9|a!5A%ibm^m2X1KX@
z773szdWA=ejT#O?lHLL&T<|<AbCE*O_?!Dl<Djo`_h2*VAo^+`{+@2_{>D3FlSeI|
z)x%*|n-y2zDtz+rp2#Ogx6_RTix$7|CjY~VLwDI-VOMQ%bx`Kd=~<KlxK_X{yaq$H
z@psV)IjH8wWjNc(6y%q(HnIhvNaI4}-yGHZ6G6tSl^di13N?YS4180glt&-xE5w1t
zZz#DXqo?v9kWkhGgD2jn&K)=g!?xQslMZKKSS_IuCW4Q96Eu{A8e)7LPgsSc?44Su
zlb*x+G&`nD3<qC@fyvf7Jfs~=h!jo|DEhqvw)~i*>`(8-*NNN!b^ZK9G9mWySWk-~
z^ne-Z<u~A(ut4#`AKwV0d=DCh0+}@j*RJDDkuC*>Oc_-V7l`!ZW+_H$ZG)0p875lC
zURG2A{_d{mUvoe_`GmrJ=6;d?(UH7!pJsK}Ct*-A<_;|Hc+it{&5}yO7xCw98WpQe
zp*IZ|W4XTS_T$DvTrnjiqQVf6a_|>HFW-YHmNfXbV&TQ8Yc=a5F_s)VFiUeheko@Y
zi>ca4&da;$Tlt40#g2CrC>hjvaX1+yZ3l=m2v^^Sp10lX-S;p<y>(EYH2@ThdztQ3
z-!MJZwzT&?V1Q5#v76xlQcM(%&(PNg*#;;jj%!;kHVkJ@N7zzegInezc4OP1W}s-r
zp@O|%L_ILxcT<q30=|;u=E!;$>Z=AxUgxRb5DF1oQ<!GR2JUNNcW^$Ap6;f`;fb;5
z%IlH1(jm}jwk7W2@s2;0GydX33xEkO*IF)%va`)VIZ-<WNAlKAX=)F@xX+G6qwQWm
z53z&VKt_%((m<U#^s~)870&EnRxU8fstd!pi1TA~a?9RW`KJS|)vu!qX=MXz-eW(7
zaeR~gNC$q_Q5ldN{z4+X4~BI?E(PMXqyO0?SaIZ!{5?TR9zAspjdGWjLEz&alj)Qh
z8|mqN*bfkd*`<+~+a+^Xe6vO#1}iKd$a=Ct<8$wgF4j$(06lOn);vD!{Q{6?IZspU
z7BQUwCb%`-a0JE<HAO}~m&IzW`T&0g0yYF1PHJBd6?8;WvHO=qjTVxAWtBi`zaM_a
zLOwsHC$1}f+sK^gG=cX7ia9<vu{@Tt1sBnj_s_h!s%pQy>rMkG5KR0p4gdoI2R#p9
zNsz(-UgzT@??Z4k?cNj=08#%Ywu+{+w1r#mLBP|F72-V#s3f$XG;)9dE_TmiwA3Gf
z3_ba9%ly*PO*E7+M`)R?UhE@PGudzb1=YR-z<JOeDdm%g(K_kY<$B#WSfNBfNy&);
z(1f~XSh;@bdMpf$Tg|k}#Mx$71F9DKyIP?-bqrbzE0kU`?zJ3u02BEO;M#bf;9r#M
zMp9eruW!X&5Jvu9dAF;Am<wBX`+3OMHge_`0m~3;X{~76T*eP^^Lhx-i7yYH0&ym)
z^1!G4uP&609TT7MiMw^ZtV%B8rfx95e+DKc!-lKpLRca2OF4X5sR9J;M~L0<0uJ2p
z(Bz**^be_jk&$Ox$%`%$Smgp35OB{Mo0~Bg&H(;l&u|$O=uEK>Y_dh9YubI!(hxx6
z;FpJdCVukcw$-6;>qn6_ehg@BBgU{(CuXV}&_9Em<p!m>7bk#-<%nPjSQ!EU#QqQ{
z%@Lqc+~XlA6C4|NLI&yPTi;@Vo{tw^zhi9%bUweKyNt@YcN6-t3Wn@Ir`!6yWkdHj
znYzET7y;0*MzT_k&*fgfbPLGLzy$j<(djA>WYKhO0{ip6pFc6wH8tUV;!DMbwVQ2t
zjQeL7GiQH~0BDjBao?&!Z&9K~`JD!NVtU-cyFvJerx2jX_ESDb<v7_q#(<u1oR?1u
z4laPiK_&K;^gp2NGp9byXeJdT*;v9er=a5Y1FW-m)8Co%f1=k$H4v=8EP))RATaww
z4#x;!3ey&bW)FHco}Br1v!rQ@t4GuWz%&sM`nOLxI5<9ljE`7!>i6Ja&LNZl@F=gn
zUQrAf$iaFh=4C)KEDHg$5BPN&euJF?9v+nkFyPnMmi4r?@Q15=-fGQUNx`hfSkAsR
zn{Qc}03+&g0Jv@qN73-ATSKkm%;k^U&!2vLU;(_BO{~Lld^AjDvH0_6>T|PvFo-xY
zF#+6g=Nf00D5jHYJho|~OITP0_tL5~95k~6JRK5|Ko5=yt%9_i&`1M>{tdS>qlTK&
z(!LcJG^yWa9j?}?ECPD)$)V%z@1%JCW1!m$J%-EXiHM?_-0m`MwYKe{plEolamO?n
z35Ea-c(135FaR_`1+q;v(rMbC`3Ovg@5sbfO>VG?mA(T^kY#@+@oWc`$rLz+Oze1z
z{}mX9CiC95zQ;Iux1HwWBJHzkkRR&Z)gyqEo<)a=%xs}txY1<P=PvM@E{ckZFnqw}
z^n3&y;*G%VnX8K#lF)^QnF}!6?0GyFY98xoYuj-0FAd<;;7b~$V%%ko@TI<tRiZ~+
z<K6joU&n@rt6%-RWx-XH0)A_O6X0F|j0Tm0W-kE2?&QR&BFqQ=r@6NZtFl}Bz9%7}
zf;7?!C@7`0NGVJZlvKI|L>i<!R6rU51!?I9X=#*Zigb5LgLK1t&U>wOJm0r=-koRT
zW#c|>JSNw?u5pcVj<f#1K};VKex)d&K=Fx{>B<2{k!=CNKb>+2Z{^@VbN%~7c-l|~
zgRFJUkC%!$Ul8a9FL>$hYelW))>FCwl<4mMi>@BV@GOEojk`P-D!U))3Ex#Q=?bJ6
z;bLfbJK9-H0>$>uVi$v$XUz?~lY>pbdf4T?I*)*gJVeit^l$tgmb*{hYL2xHL6aX0
z9P)QZI1#<!5>-n=hQjIbC{n`u)i*vy8UhWS-@ty!LL8i4R7BhWtt+!#h1!+eEG)(_
zOETka18rMY_jEmv79fCf8n<Z$cgvVR4tgsZuZg)syrJVTpaPu*iM*x}m#4(H!>=@W
zoiBPpuF5#cLp34|TJQF1Ig~Q{(4PwvJ9pk5>ae%vjk|)wB(=k?2GyNM9hDSh>d+GZ
z@bQeoc;ZOZDKRY_sGQp_rX~|`rIe;F=Dg~RUps+Xytq=v`=j^`?aPe5vmEbQD`bn*
zxk~+&&DlylOVd?y&-TCV8LNmISR$43-l?TKlX>ZqlOxGKu|4q-{~0dc=CLlqGbMH#
z|7N=t<||`?_MNSvSBoOmo8Mw4ST+ybbe}6JZK1<`wfbXnZ*YL*VtKRACeKa|tijHK
zCfza&vUzFHr2v7dGwebk(s`<@Fdk0R4R4{83@D36FjqV!7nZKcrx_`%5L&{<auWZ*
z4u^ocfB!oX<HLt1utiKklfw1_D&#l|YoaIsc}cu>6q@zkl2@iBH{QTTtcwppx=m$q
zcASy82f^*OzqapC30L5S{=n^R+i27tI837!<@+Hla<ToQ))>wp?(eDkn_UahF2><f
zY8d0#Nw@2IoF@qWA+<*B4Y|$PS%bJve)q#SEqS5A0Z8Oq=0_?tWl@*$uvP+*h;L&^
zO+Q_YWHq&w&d#4b&Fi^c)>^++^t=`oCoRJLO}@Snq>UPUH-li7>FDefS<NfifZS;L
zfTpG2HHOKVaM&^t&d1A{V8f60!vvrV!pgxBBAMita)7HAFTY4-9N)95Fs5)l9A&`g
zDO4V1{uRk7P92r2swzIT_x_}LsVITwnT^n#QG32wQa|4t=m_s=K&MpE=3B&K{HD**
znP)2A66}Cg7Mlkr?|*zH329Hqu+gg@*3VlY8I1a?JLIacI*-9$<UeQD9ql;Ma5tcY
zgcPIy()%&>1Sm{Z_0IeefLMl)PX%vg=V+HGC8DCC&FHR<wFYWHsAE7in$2_?$$5vR
zYq<L$+Rx1H(W-5MLms+xC=!{JRJeK!<(P-Ux@u|%;J-tCSmbUkC^53%J-S{|SX3_0
zUangt*c~W|AbMZ?vh2t81+$PuI_)-eJj-GN>3jtAlZF$9U4Uy|J^k(qg#M%<^3Z<%
z+Kf$vODM#app=aQLp*{xl`(oKmmz<txZR0_RMWbAm}tHek~~?s6kNKH$3f#0&Q2G_
zAZc;wYt9Nfa|P*UVNua@bTzhAsHM#)rNKr(y-B>b)qv(c`_OPV;U^dk#-;}{@(6-Q
zd$}#QOOD~DJiee7Tp|0Y!#RXb2uDS!Dk^?VOsq5MVU&qL=a2s^EFg0YgSBr9Ilga>
zQYsKq!TNXu)!|)3t3W+i6~8Jjlf`d|suV@fQ{uf}2XzOv&@98L{_W*YAN&u5bdYqc
zh8`EljoS>Us3D8gu&hPd0yQ`H?SLDRVepmE27{<^`v$hX4%-A}PN#kJmEmjCjaA2b
zi&yU#IY~;sDvyZxc!9OK_D>H$l-A>QI&hXDmpDO_u);^Xz#TS8UFkD$`Nr2(dbw8b
z8ttANvviCcSl+rS*#b@?)`IQJ_ZXKm_Xz5Ua@U~fZNR`39T1wNZc*Xr=^Y9+PTL&>
z%)T?7i!LeRpWk=Fv<CS3d1FTez+2L7f?=GC?h+_f^?wq9yTnR~xSzKmMWAWg_G6x^
zozrmcy(T9*3tbl&j*>i|ldi<sUkJpxQ*q07Io(a7tTM~t(iv`<!|Pr6Vph!SmA~xr
zIHLzJzt-{D5;Ub~=Syl<J$ZxKO-LumzvWs}6ndBp9fWcra(r*UW}ktRv2yF5fp1PB
zSienyYdimr=%@Pc^qkRTtYpp3C-RtISA<ZYo>gLKh8ry}(+<2OZfO2suL(oA2v59&
zPTykb*3GBLU(ZZ6e(%vQqUVpb!92IZ2S#FOsUoPMVJ~M{D3j#0)U(Or=ZVy5(OY@Q
zSC(N9>XNMJ`?fKxZ?x~cC|Kb3U%TmFg4MmhDY05FgKJ8Gns35J;MUYx$4+GIhrx#h
zt}b!8mC)&V)D`7FX=(`7{>W^qp18j*@b{#w_`R~cRdhb*Q<?MPWc-FPzA{6)t&IJT
z{#n;@{+`r3xIOdiQ7-J7S6h?7*^jPm59K0J>S83$Pn$U!t(Zb|AFMy}Z*wq}-QIk9
z4f9aJVV&y(A%BS*Ox+%jTeur+_%0eGF=3=Ze@xYwe`bkJ6}x+SE3jjJlnUnMFRgWZ
zToyBHPgtyp;jTHQ*r`k`DcL`Cb(hQSIh`K!&+nOhf*qIqe@N(8G|u(N?%%vFsYdmJ
z<U4zF*=(DV%FtAPcC9GS#Hl9<adVw}wTG&vY^mGZJMzY*)kJpJE1%h~(FakOxQbXG
z_DMygjT4U+JjOm9{c&bFbkFe(QnMqFNgAU@Nv9E(e}DRBX1co=4kM*QO8|0*mW`yx
ziX)oc$F3pFHu-GaV5=f|R68P`mg^~EvY=qEe3A?r+p-j1&U1&??+JxEs<GehDE2pq
zHt^%MEiFpk#HXs9H8Oo&d__Dm$FogK8)iY$&o(2*)Z>~I@dyO33HA4nxu4r;nd1Wx
zrSDH>2wYTlKAUE@wAA0teYOI<{Yuv8Yu_(2DZ_kYt|>nB14TZD*TdbNZmMb1wn|Z-
z6lL%k+j}V2>@+{x9Yj;jqEehY*Uc*>Q6l+coNBM;&2EyH{^BsJuE!r}NoS?3)JCkl
z`*F>=9vTu>pgrVQ$|@cXqQL;8+p*Fh<yXa5-|G)>9wWK~A4_o})Qr=pXt)>_{xKR<
zr*#5PZy<DVg`yG{n+#}X-aL`8LoDq%9EUe1$DZBb7~qrvu2PO;ZjBYl?~ZY#I3w*@
zMZDH<g1Di%f{8qskH0kKr=leZ`BCquSBQ|D<D?bpNLNsW*qgzP*t#ztxfU#}1oFPF
zDN0@+rOFr?9~8>!QlkxVPJ&3p6#y0*p-9dvp<1F{N(F|J`$XYiu*DxL<_HM*nmM|d
z*#;sp{#wcD_=>&My?{TE2lk3a=)`)fI_w!mZAu0EJ|A0xu@G{7b@&e|Od`{A`}z>r
z4jwAVbkkZ`#zLkZH*8&MC&4}7QvWs$`KdZckL$TCke505XDy0?%K-m$3>|^tOZo~+
z`RR)xW!eP!2c`0_=Y<k^t=_?1kJD$2$i3$d4k4Gm%Q(pbM&~&Nkk@i@R?xM^IO{r>
z-6f^C&T`12gLey^gdQOwAuPMXoAnoUL#rW*EN*F*=O}|q-EZ=qV6#w!F)f5Wj;Hf3
zU(`BuPmYz7vk$dv!eX^>^co?uy&M7NN~+0qUbFa>jEqb>G&!K@Kv2@2fpj>nkiCh%
z{tT4y`WlVjR8Mq?o~->&n29osOSUxVfgO)7%TT8v^pys+2qEGfhUHM`J%YswLVFG|
z+Oi){;l}9$mp4yRQc|Fsl9`zqb}|uDQ&U87NeSGmr$*FJzLuKCNGD?=n&AX(Z{UHt
zo$U?DEn5p#)#pF)ArSZY=ur7MLLte-B5SO|&vZZa5mig4m#(Phfc>mpCj|Jf1A*qI
z--v^Nn9ort!SdSc*RMmVc%+8*Ab;&AR#RG9T2nLry5!dF+m~-LjtyDSX^C3DNLnNh
zx+!zX@MQ@`cijZ}ABv%63e>^#=NP;>ciDPifdvQs1Buf`MBbKL`TKp#zephX(iMt4
ziviJ6q_lsPHtoX7JA?4n1VY3QZ_V!<&l<H#ctS!lJ04VkChR^(29uhL%L9a=Y>~)Z
zcyVzt>+W)zLydtF)=N@G`M~9h8=7eWks$<@s6Y2QIb|I~zO`Zu_Yd!?Fj1nveJybL
zX4IcQuMr8jStKMRi0>0teO1Xc71F|&Y@bJl=hwt$@bL{?QbEX|$#1l0GF_8@c8=sz
zKDOEv<7m<1n^^VkZ<yYF=)<g*<tLDF`(jGs1BS58ba$`5g#0p60vIPPeywxM#+%U@
zrf<c+W@qbN8Z18m+$av_S&lmNpT%bEl^o=*&!td`3Z-;fRiJ7J6CV46biiG8bbX&N
zthKcj>#uLYjqd6i-OX(20>Q9We{Io*%`IMYYHwx@?)8d$&+lEgIQjk1^!UrxU?)xT
z>hVyVd*5$@fi=y{xvs~D4rc=^qbI$}@x*nQpElK|Nbs<aHRic@_q0i5#SWjBjz9r`
zO!irq8P%K?y+Z4y{+2`4-Du+kyq@FALbfOz6&@1WO;b9lH|{8hAlVsJpU|4@Vs{M#
zOz+onZuxaT(tbmcB;o59`UP1Tne5!$Rbd}1;fVNn3wZ$&Qc{I88QxMK9AjBohj*#K
zvNf1B>i_|a9s?z%amk4MUO+@d=?@7dN5!06xjdn}+tR*fhdeqpN_yBYjXdV)TpH!~
z9I9s>t3!T7`1tsE?&}R_WZXt7K~6ShB7Aaw=P0<hb!%%2=xL)9%W^Hs*oJBKQ-4p+
zTD<bz7E_sW$aIO3$<4bAA5aUJiqD_-&1^y44w0Rm{l`94O|%&6hqscoZh#YS@-4$#
zvJok)%Jo{0xKOr*r>WUfA`ibG8zE+Mx!7B3DXY18M9OYVp68F1)Q;3P&&5T2d9%Vb
z@o3B~jaByI<jH5|?4*T=^7f|5<LAM2kNwXNJj*&G_v;UWvTDDa7k5FL<eP(etLqmi
zjbVL^j;8b#O(-=)GUn~&2ayXFT~8F=`W1z#^RQxdZ!b9(f4Ws8e}GXv063Ed%Fm<=
z?mW*x3_~z_!iNHyBy<>l)67Sc7<a22e~zjoRScF%6*PGY)D58W_+5I*5Z%0cK?;q$
z@dB)kx53iGIwo36VhaLyMImUp@Du=Glw@R33Fg$P>c8%?f8>s~3^#}bp~gi^D6NHp
zo{>7p<3Y8uXVt<fxJvtRgF(J!CpIj{DEC1ZB24e{Egqk`UG`3cn1m?IuNo?PLZh7t
zXra+B%038I6PzYvK7A^MW&jAvIHC3n&E3AfzMva}_}Y5P0^pWVu_?Vjb3HX!iL)8=
zCbgzC<hklK1sU)enVwTWp!S{|PqP{!%P<-npWSXGf4v**@4qAoCbXlY!xaL1&bmW`
z>Iq$K?KN9;8siM&v;F+b#3l*qWZ?1%h1w?I9gSZJA=rz&*7zdm%cN$2r=4M6dB*EV
zA*dnu^LR{;=6?G02{KK7);rKx;cNX38fzz4dt~n4rpQ-CXp4G1P7A-p==M!lH$#nn
z;BtjbrrcU_4C2qVj#JBaPed1Ofx3|G_sR+A47kiPC|!Asui|S%UOI4jdU*Q?FIsk!
z#Yd6zi>U>Un7hC)@gdl2Ufo30IG6Tjn%;ImJ9)T=pREHOquDUCtT`S>HnCB-NeNk|
zJV7D8Mn{`gUzXsbu_)1vn8j42!>iMnD2k^6AA@X*ZR)BeIpfRwgSIvXTi7xY!e<7C
zJcs9F`Z!fq^!>&Dp3O6<ukB&9vW&r@A~wlq8{{q;E_YA7I2XB<bYCMaz47|e-K#F9
zE9BXCvMDY#xRp69h<atc1~Y$qPOa|F=;uV~Rn};vzq361emwIX9xfjK_nQSs|8z}N
z8SSKSLh*q;%DaMSC3Mj{y`$INTDgl?Rw3e!N)%J-_2Fs$YJqSy@;x4)OpfiO`J`_l
ztoI`pI<U#u4)fu8Q?-Q@8D3y~C*zzKdi0nzTGz@VXR|&;d}zndScp_!;SJ(2W282_
zsJ&(bOp7lBUYF(gHZ-wv%kF38R1w09B{Tiye8(CaQCp|0bD@;=TCC%l0ROp*j~y^_
z$w=W#aNq-D@54suonKTgDeD$E2hLyYcrexDh`H)no2?ah?!_zz5~0_{^3Y@5N!UtW
zQA=@>-nq;kW$2*zpu<5qEq*)-b5}I!cydgqyuDb{_;v-^lBQN8rVf0C$LXK|4Vz@V
z&_c0>A{k;Oqic`pL$!t7+}Ehe=Ebl##2_O;GSu&Ns*VYDe9qf_OQYR#J3QiUk^?=$
zMDy?Z5Pa51Gv}!iA}_xn`zrRYHF|in#t^fbS7(+*k_Kgp$9M+35k#0wZS*=)I#?00
z$f%`fV&W|wqoi|xZ7@K`=oUO?d3ivyq>N+*i$~l9XO~HSaMk`ECtE5#TX*v+x_@C5
zT(uUMRYAgg7U78);$q4Lo_j`%IL0iuLeV?n_;6{-h@~Ui0UOny=cZv&D-1V+XKKkf
zJ|9~uefbn2)?q=l<Qj^FerkXTyv{Y6UlHm3-j_Z;gV;S5a%f~ppgA>B^xI;oD^B6g
zDiMdgHWD=IBkIg4PVJUhIt(6&!ivVpalx^p0%nBt>A91cS?2~VJT9B(YFxhMCscFW
z08QHXOz)Bn!rrOmcv(7u=2w?ljF$VKd+7Lz)eQZhj^RP1LA4T<^KKj!h+})YnnXgk
z#kBNWceuUF2KdN5DbOu@OW=NMS>C|AA&-~)GTeXiX7Pt7h6-4SFzJbJb&ZqG8ddyE
zbmHPeNe`9vJrCfaAx%*i`;th!_~*@n$uw>IfTCp2V`9v6LO04+M$|M43<c99z`6*$
zEXDc!lKn(}&h$cHt=v`myX3)_z6uCr$MHK3KOBj<*B*m-S=f4_Z>d9e47LS}9?};p
z?;aU~ajA4oaQ}g9@vHLGSv1(KC?~W(t3CD4o`AX_CVV}&;Y~^e<)-PxSlSFzbGJTX
zH`Z)Dj*f2scjq&O7q1Z3sso>ngnkP{w1A|BtBwZR+I(Uo`DDk%%P*uB^yK4$OAHm_
zUDbKmcH?a)So8kjs28nwC0S3SH$8eyNs$p(yj6?UxVTooIn7M7&r$becCZQSOyI<U
zKWcl}Y<-tvA&^}rqTVgEXyoWzEOmB0I)fQ2>niqnyA~%2ad;(|kY*ZosaW1UcGk-X
zhgG$|_Uu8fi`^E6hV99=Aq$ND{%w-M+8H@>tZF{^((l|ysSreebM-uW^f!ZF4K-oo
zLPKKLO3vM#YV^>6pPy}p8LUmr;&EVMC~;;hu~_Q_i3$GmnhT;<$Sb$;!o|$u)L-o#
zwT0V$2+I3`*bQXTT&_dE@Jx!PF$@b!Gu`JJbiU`X3cMkoa5vO8M&kT8uc5@rY)NRf
z&*A#F<d*$S*=t!sAD|A<b`^bh56Yv0F1Cr86=JUjUGlCnG+#vS4jH9|IQ-+Hi|Hv*
z{d2vBMlR3iMNfQW*0#0}0}w=bSy}>4!*{N1iR*?IdFZvDzgFR^gd7Sl0)qLR+e8Q=
zuBSF67Yj^JW4CqSakh7UyBTMuCq{n4`}%9~0`*oRV;|lM4zuT0T*nvnXEOv8ObQ(+
z%Isyt?zV)h3uUsIjA!E3(8b>$+Y=k1kzwdsT2MGjSX9m~m)e^Q{2C7`gwMx%QY-W0
z4dMUv0(><u=9y_7d5<{cT`!+-s|*A`fe|t$!hG=V-+V${?VZp5$c$JS?8L(s5;(M*
z{B||>QbSeM@xlkTf6NB~5G?+f<`Vq~K-%FMr@XjFfe@y*TrB@hpv>N3c+?oOuD8K4
zvbh&CA_5&9ieN|jAm$A84M#Dr|Dh0Y9GM;sE`{(LVS<;l6SV)g1#NkO^zz4)W?{A|
z#WbyF&WDHCyWBcVcVbVDJc*W%v9p)E_WQhI3_1;yg7F@Gvyf{_Vt*Q$0ivDUxN7<(
z>Q`l)G<-jY>ni3OV59DHMb8*qrf8Q?Id<DmD9P>q63F9P_DT6<xgNjY?T^DTLi{xy
z9oMO@jeTuK8Q^b%0283#u!^*UOGiGDi`?GESPCJ~?D?(O85K$^I!=%K(zxVg!{elL
z6^tkWOpQO;RywH9Io~*=v*)upoGMl8hlz4Ro=53j7m+2#Y%u)LA(oZTlWhm@9PsQh
zi|X^UtkIQeTco~?7~*5KPYd($2iEeK7bhM_AeI_VQ$!r{Nq@jw%nXGm9QY9x(ELI*
zNgyWWiLv}8gpOMMA^Wq&6Wb4o%g+#R+7>U8W=ZV+c}zelx(IfVBjFjtI8v^OxEm4A
z`rc0sjtdoYnqL|lDEs4pj@0?kDFf6ZBy>5qk|rz|IdrHs5aQNmFPtn0`fQ)ooLzHy
z(6X=KNGX6_UDpPGvE=i&dP=y3m2&8}bdd16wxPt>Uzx)wC&Ko`*d=WAEp*oPK0%uf
z9AJM$epl&CHJ#2g*nN`P8t~}_V)U)@y7+f)qGBS{mF@fp#J9ZDHos%3$ST+~&wxHD
zGn$`ZJRe9RV@9*2ET%(Qx9%)`I-4wkFln%?4#E<$u7eL|`k?2{;BPyF`)2c-&)yH<
zh_j7n3^|nWo=@saNR6%|9C@oRJWE$kc)-x@5uq<D+m*R(C8qSM0yTBSiO9-<jbUc2
zMW&ZD)sJTET9n$Svby&$%q(x=(zFYK#M=)jAxV>^XXQOcSGV=+)leCgDI%6{z2ZYy
zY>yZuP%^=iDbu}DD)@3kSvkG03xP=DIvi=8DEE3Lshh%xb*LcGSCwJKNDX@vD<lcC
z%@U$y!Kjtqq5RF;k&=_-fgQ%;K4%fPcTmY;ROR_xcc!n@{{Z~BV<s557sq3al+u+c
z&#rbvwIt#o=oP+IKc&QAr;X<mog{pjG<ET2Wafzz`s26IA1^CdkX>X|yQ=tFcaXcg
z%rjxKUv7j0=`Zc~`U;}q!s(lPQY+07AqlaPS7Vz|gseP!3rRTNcS~gS@sEcS<w;S~
z-&;Q8HKdvb1s)))AbHn+0n;<hk^^qSz{j?I0qvMjCM|zJRa;(om2gLpQ#HUWDpXOG
zu7T2efBhRBr32xK{BIipsh(oq&Ke}1Nx0pbYWi-qx;LA$#k`*3?*@>lZV}#JU@w{>
zHHp2L;xgZ{mRJ&atTOOIqfU(H%&Q|5p<e2?`M^;tf<10{;z^H#$n~rH9SkmOc;&h`
z)(%d}&K;WHXP8zM91I=gTD#l(7?q4L2N+WI^Cj=d-=4^ZHo!fF^|hbYXH}w2W+R&S
zFvZ8j+%8aBEro+Do?J6K&0;$~CT7^v)l0n%G0!E+cs=jXeNnM}V^~7d5OSyB)_dHT
z!GX3OhgyXi4VIkxL<9GF18oC?6yN7mWKcw)qJC%An@NKazx-_2{SP(*T31^x7L}Ei
zk4L-$pp6`;-4j+}j?%7@Y~%FK`VSDS6ste0B~=}xJE>`CJVY71naaPnYdIeB${{x8
zt_v`Y7`?V;N-i%SG%Ev+vWZigK;?v0#TY@uvOQFeOq?<W28M>7MnFM76&;`MxyHrD
zB6gF*2vZ`v3+vVlRjX69<J&_xdJp^3d2VBg-y?F?cd6K0k{+AepyS8(CLGs(q0R6@
zZiyMz0Twu1S@WYu*I&nYls$Qh`&CWtQlb`4y_oyS)cn<$d9Wbei`5=ox;*IB=mBYX
zD)0}A9!s(Z29^A(Cbw}!jGjj{lw5cVXfJRC5p{*pQ?bY(R8?LS2>x{Af~j=?hW#GT
zNSX0UOp=0-$INt|$$@Xx4Q_6Z9Z&Q=saiaA5q2?PZvL~BVbR~qS+bzW<7A^&D>n9D
zcW4TC_)V5^ivD0iauLr>yY%F-PNd(W)}IS(RT{+ElS;2;iP3_e_tw4m1y^%+Um)n^
zs*B&K`PO&ojRWzR7<gIzkE!o>xQ=y!r-B&YXn=mMx;^VG9j=d*4JE14p|C#>Z|-?m
zw+e*Z)CA-3HasXx<i2y3Q4d%83=@X<(13#2_JxN`mW)&;d;NXS;3Cn@+?I?H-)iBr
z+!Z)ZVUJOYGi46;xP{gV6c6WnW{mD@ba?D%xx-U?9^^agHr9#!w{@cs?&?}%=B=C+
z9sLm2=l}<GMKa=3Otgk)jiMka#?r=XR8W^jprhl6qh`v^)LhFjGkyaUBBHIP+6}-v
z?x$j+R2qT|dz2n~l-~k|cTP*bPAZwzYVpFYI|Y1m(qp&q8;+0tu9R;G_UhQcu^7}Q
zAozPs@>>#>e<gj36d~TXwpo|HIX-EGLMcZKC5jH-K_SZd`I1VEq-BO@bzuYQ2CbNH
zx;nevqWPRWE+j|K%Vfv&dW{;d=*Te0(^`+e1UJ-eBZ>}b6&53Mi!nvL1qPT-=JM~V
zyMhK}2yDzBi_xnNWjg%vYtPiZT1+3<@FR>wFMPGG)EljUONxMwBoRi`q=sGpqY=bz
zMoYt$t{d0aiuE7Wc$_u$ZLixv)6W@CIk~MLY%L9NafvE}tzUH+3)C$VxbMvMt_^Ur
z{N7hGUlE$$(W^R(%R!j;krETwqpzF(Fkv>9CfHSHWpTC4+R?}C!SL&|mp}ih(144$
z_LGq-H;sgt01{G?EaLMEY&@JGfxT9@A2A7A^1?U4v4@#qa;jvAqyYi;!}8%$d6>ah
zfpaZ%f7_QQy@nXE`zK?1CAzVccGdsl;IKzk2BEH^bkzNp`=9Nq=tFqW62!~4ogwJj
ziz9hXoanW_eN44pQYPu^V+Wtsyb3dZ2h+n<*Yx44`iu{dF?_AVgy+U`Stq(Q6t9dX
zMFp>T=d2E`A3G*}pu7h+3$Cvui-f3MAn!ge;;-xb!h&}NA53eQ#Wu&&BoTaXr>{9u
zYz+>9A43E=M;YZm+HAChh|xNcl}rX-Q&Zwuy*a6C<r)|k=4N$AV<Iz{u>I^wqc1AM
zEcltN@P7IG&6;!u!U`eLJ~Iswt5UfLM2&6f%G)#C&LyZ1HUl<0C6Xn9Unx<T9FD<h
z$XT2|@HG|Y>p7awib~%-h^BR3HZhui2o|E~9Sp!H){+7eQ*b{!0vqaeuhwU+W;~*D
zBA~k}j%#&ywDHRI0Dl}pxaM=i5{*B<KNw(qwVX;hKS=o7$v~a>kVS;tTd%KhywV&F
z3ZY71W=t`GQ=)`a7~TB0-wz^n7-~{Iw=}-AL^O2VPvx|~h4ChLbm3v0OLf&?ur3Vj
z3urIg!#w(RXZsQl6bav(>#7KNQfJu2GuPUTekqA0!T7GSe&SnwuIo4)4>cmCQke34
zgbTfFno+&0u5UVXe>hEEE-66WL2V|AI--j;ibK{@C7aUK`@A_hrnLZxibQ`0$<0p>
zhT5lMUZ18FxQYE;6NAaD*hyD>n$*`a8UNYz^zd<WPe5G7c}3Kfl>q-^7I@@ICYg?S
z^98EtM|g<Qjg*E|uO3IMYXjqV1B@?kU+~+0S1L45YJF%o_ItH{#G`Fdps))P2M`>l
z=-%?--r+_O2sB#ZLST3)QIGy~C`ez{aY9sZHAp5?;mI3B_uJ@D)zI_pRk2U>yCyYU
zM8rxL(3RbP0={Hk{b3oe@YJEwL`_fqZsBqGw&FSqvmk%MgBh;I4-Z~ZiI?P@7wk5C
z;i~tn%^6Ljhvc;nJSNiTiG0y`)sv%13kc3?x3ew+q|4~EB+4rAw&LP%BA?&x?0iOv
z@IKR>`s|AoV`b0zi~DN@pq9wXV6K|*ZYBv#xSk|Pxwsyly;WiBR}j5w*F+TOu<%8R
z=5~VUNX$#&;*6Z*-A*m_%tZgJLU`AK?BOC!RZG$6c>U|@kFrACDHpcWn^SW3p7%|E
zk}`c@xy$XHdfs3(<|Jx=W8AU?LiiT+t?I^4aBsG5m)eCS^*e0hP8~ZPR^OM!6r+hR
zv)raKzV!gZq@nQO^JKYOA@x2hMBN&&3Nk)28|`0%S3)J0o#feXxq|tT#`wf})-(vm
zm`aS3O>8?gMdTUSUk_p?%5WDj*mo|Bql)r9_ig@m4oqZlLeh%qgIJW<xS*O|URF)E
zNmvWyt@?*xTE)nUdiibFj<l3l%|uSGUaastU>qKO%rlk><CP!!O#ZoIbA+Z{6ei&&
zJU4fo`=tHCHd~0L!(8T%AzjI9If;|JtcFKR;ZkS*EU2(?HQO5z9<U;G!RV4*+ODL$
zt9IJ(o$F)OQBXsNG4WE};o#{bs;ApK!Run{Vuz9=Gj01&7IM7?ufo1<QTwa53js5}
zce$P0^m6@jXXga{UeL{vdOH68;q#+Jk3!n~KnpI&;6}r}BSVDu*ijBU!P~p&EuJ)N
z=`%TezO^P^s{|LL+na%W{rtk*jA>E{YGp<+kUh)^o_-5aN#)~M)m0rAO{p-!t-_8^
z+Gf-;(lP&<AIJYpIWX6VOY~<QM8B;VUiY-U%nK%EWsHV7{5)snv)$CcAKD(*@H+Mf
zzoJBhWps`ft>&-9wjIE!4D$5w47&lp9!2@qfAR~%M#x>M+uH<N{540Put=ttX0BCx
zS{8G9pp=1(f<vTf000j~LE%BHxA8tS7Ec>*e<}%P7~m4SCcO83vj6&@kIovJZll9b
zpGRO|JDgEApA?L0*!NGJeTMk>6;w@AzT8B?QvIx<R;JyeUd8qIu<`iEnSQ8$bh>~7
z&CXojv$@uR&jYFTr^m@yMsoVnZIB0WhhZo}ik=9gI>nIg!id^}!Hapr72_nIY(;D0
z+3a_x3zv=vHE<MZuAaN<Er0A1n5Z+JIzGmk{1^_!I~et5TGLl=9cH-D2~IeG&bp35
zeJf3uvfz3XhenL=)B{6R&A=@~55+ULk(r(HwYKrU3)h?|M3fwzrAy+RUNbWWGA8ec
z{pbO;6ayYMA}uZLiE{nrI?z-RerU9+kgjHev}aKdH0r;p)P@u3c5^(vz~rqA>6M?F
z6m=}>-bivw_IL7mU?Sl$*NBTcf38EiWN%hYwg3TWspO&)X+mTIJ|_}GQA^9bOp`A3
zP1ndG6t(`Z4r)C(sQcF|Q{IQ?S_jC_V<NJ&_;IJ43Fw_rT@*pQp6`TT*R;`bMrbF8
z3e6tWt7@FRR_SCxcYp}I`4v(^-ZaMZ);^A7f*w~Kcja9x#oGt3?zFz9S@lo)GW`v&
zC$3%vtkoVzur{ywJzFF$6O^Hv+b)S)TPRL>N=y`2&INM1)jW|Q50GKJi4njV{)JUp
z89Jw)$UicFylpaSyhpIQN;5db=iW2Cq*`6?<!DT(kr*CtuOoH5R!XL&cW4KBNv-RZ
zCf+aS0|})H>TXqqqrNgc$n!kab7)n?KAqq#8%bNcX|Qzf4vX46A?0t$xf$;QYfH29
znz}7iOACiwCrHMo@e|LfR8i|A8bN-qZb>gQB!9SGc}xMQup9pw9N)HqpNoBUXHsKo
zd2~Ww)#!X%`__pWZ>h}B%QV`i>$@<T-_yL9JdM>|+1n^6up){uQ#y@J%xXi-GPiTe
zbmpisT;14nt+%o`&nl?%7zkiS)|oF&Ur<FP+&~c@b0`qQ;QVYuVqr6>+24aPyH8dr
z^w)LI&$Tw2m|)M<BVSi1pFt#rj`M?nBw2}idb~a(mtF_MorU>FQEtWVlbNRGx#rPy
zqT8NwI%A7Szk%&BeF?7!zP7u~<J?>@G``_MICxS8kStx%G7zwJkQv%rkzH@_^zaH$
ztx_ly9Cy&*ij$*$Fd7-j7ADdep4bz~GCA0PdeZy8P)(&mWC9$qLOZ`-44;T8pL(?f
zYkL6ak$@;zyXQySK*cUVBHj+=(y<R5M4RD0EjzoV^9v~(@hOrmgpMnOo`t)l%A>9`
zM9yh*?BbE7&KMK)zJOJ<<LP5zZjvm*|F0W7cAPg;7}{f67@ZUuO=PsRwpsjq2!0v$
z1I-SGTT;Pf_%n4w7g}0QUdwL@_F5+{+4^_7S}FH7VYj|<(~cU5uqt|`nJ*JK^0e;1
z-sG|v9J2OOoNaP_WczcmlK?f1&?e<_jPk~t@Hm!*`GDQ4_A0Tf??<bo38U)Op3`I;
zO2#u26lv_N@8jc#Zr;8fzbPBIdnxR{pTDc^QbU%gby3sMK-LtQ*_3B$G~Jr(jGA`*
zCy^Id=^1g1p1v=Z)D)Txt!+N3ty3ZYZ2dybYhZYMU5C!-=I^Jnr}|Q(e9NYHqWn({
zx&OUrcHg!w?76BHXjek}q@SN3F_YA@c)_il+=2)<M~CBVS{gX`&*Q0BnD%;sdVgHp
z?(75%n5ImTXDQ>aw?AjR*^tb)@iR&B<%u`N<7?YrpU4)I;V*k<F!b!o>J%yuaC=Sr
zi*-K;9ARGWDE-7>Oi*?2LADc@>#Kh+EC0PVwDPl6?7S!~4O<){Xb%q&Zzv<8m2nfE
zOYt2=;Vtm`ug13xBtqeCldESxdI@*m+z5AiCt=<kkJw*<RbK8yh;PSGZKc{K^fE1F
z5hQ*DjE%;>#M~c#Ez%w>NdNDtFPSG@;I>)o1pG;m$5pQuRQLAw0-}JydHU^p+-vel
zqGzEGH~h!{{guDk9B6fha_{an8jXhAo|*XS>SPbSXj~NuQ0u=jn3+*X9bYYEu5MQ?
z=7}#j_1WIYb;amT`_caQNasKN|9O4)=_-M|g>D7>`0*Qb?Usjqf#1ihs^$7hp~)mO
zGgEq2ofN$wgSB;bTz8LMYb)_P<*LRA!n6*Yr}r8*`>-&aMewk9#S!!Z?z$7}8{cH}
zl$3PasrR4#_U8y_r%MzZd!4X??g|MjZ6zog>1CIF$Nuou$lU<%44U-6tM7h%4aMQZ
z`+74gr~8u6({Q7kd2hW%u0M&Iw@@QtQd={ePr+c!s27H2Jp}qFr*$zsZXTZ2V5;Zp
z>d;Tp0)k9Nl2#I>u%;1Yd~GJ`8?RhpVIPbOteziduI;+U5@+w;yHV1PSM%L>8JHed
zE4B;O)WWIg@HpqITj5g~$Q*tY$84m|uEGDbsyX6yQJMxjC(^N_SSLa|knu`*S{uKN
zZHH}fN}2|l^i}p<=9Kfh{~~N+JcE|r#yM!>xXaGYyDv4X&bZ<c!5r~i()O|x#q~-l
zG0o)l$@F#3SaoG#o1rM&pmNU*SvMA@lezv~j(|Hq3e1$_S(7s8c8>`6oyj%du`dp@
zH(J;nZSD7L?JJ@GyCorS>8qhQM58j0Zzd|&=Y9OL`?kzh#H(=0h&<+4-O^zHlw1F5
z2l0kx6v8l?uuq>qBbU?}x!<<6ed6I>u;UcJmqo{}vcE2}Iel)2db(#_;c~08JErvE
zkIJr%yl*3B?B$v$o66RLY@3`FMJx(Be6xm>vS%7N%stFn)a2>%MXFl)(iS!<gMxZt
ziyLt-_Gl`OL&ky}%l7DEqZSVcpTB?qHS_Y!yHLk6juoGIUlW`a-+t^FiWGbk?|D*_
zcP3<v*Z0Xe9;a!2Vh_57GH5T?EY6dyN_nMD7R_G$f=}MxSu5V8I>;!#wr4S#xjosw
zB=C3Us@0tQeC%N5GzTJ+`A#urt_t3FvP<5KJ|)KQ-D5*t%U1Q+WK<0{qvCFwP`j_6
zQ(4+9>6r_&;xxlKgjQBK-7W)<A3w%K7`j+6hpvCEtE(GQtI2N3yxHB|jl}P$`#raH
zD1W?C{-uB4laHT2aPqoG09|axk-Mm&kv{LO%$Z-j@ejzo|GFYA_m=k@NU3SM`<Y$F
z|MmxJa=DAn>|CzbyN2VtiLJUZq`@woZ!@5(71FF}UBruuaz$>*6icfL7teqGB-E^I
zT|`^ftfD@5*Md%ML*S{hzP7u*^?7+Kwtek51M7kG$y7a^%n)hVtsFzIbt}N7C;CN5
za<G~rru$XW`xcEOsL6=+ka<pd_cB+zF~pa2$TmFKjoS{#$Dex*pc_QW$0J>po?RGS
z*TYIsd)@-wGk6#2o_AXviLT->IfXyEcm4dL!6PEWG?pTR+J#{-s`i*vXUK5%Oqndr
zm{Fu8?se?#Q&Gg<6W;;}6=Xx0xcH&R&&Btx94;b%zx%evP46)u*BoJpxm(!_?Kx;^
zpZq-5l^L}TVT<CZw%ivf|32X;ZBvul>-tko@pjguxSU%#7}ED%_LZiIi;FvzNof_b
zR~uDM8swL&vsv<2kV!B!#}YXNE|4XqJyO1bA+BTLg(}UyhU2=>Cm2=uBsD8&Ir7J{
zJ-|*}geH6|*0(qD=$=Y?cr!0cT{;zX+w?>cT=K)DEI-T3>C{!IY}H!W892lW0TslM
zPY&ic49B<F4H!*W^0@;yeROX0Y`*72P@Ui4MLO4dbH2k!c0=CC%6-(nh-E)>u$+VQ
zRVE6>Ud}|ZN*uJ8)vRe4lpFq8N`p|`EXr3e`sLMMOI6*x2v{1HaSeU;K^ZvA8h#rV
zCtPot%dRSxb&{7S@%HnXxGMVJmU)-4jOh@8dE<tpoq36rhLZ2MB~9@s6gqaLq9rNj
zzrXdQJu1(NKy*iqxfhc?k0~~f@^mm_YtHy0N&tJy%F0S`(1$S|dKvi*$d6jnjXeUu
zaTm=2QIO;=2Rlhm#jdMP0q@J<Y_X>%TjEk(n#E7Bm8Qu1^!7QUuZRQX_CqVM2zqo#
zggzTWqp-iL8uZY_$I~JJsk8p~u-tMCF%MnXU>u-qJV_3Rh=QgR!EZe|?pfvrxC|f<
zfdl~v5)8@z^Y{P1rkU@`zZQ2y4yWG(l>>C)=Jm&~;J{Gp+C5bpn=KjIr%;+8GcBqk
zIz&sxYn-=$HC%tZk{cf%?{TsZnAx?pHPD+>SeFBcsD4Tejt4zGJtY-^L+kumdktJa
zT@hDR9W)FRaRiFt*!n)m45CqdAXxxTVkryd8tk|54U)A_zFBF#`1V9{>L(+y$J0V6
zfzh#(JnDDa%jI%IKihB<x=&t`%m!i~_#ys>zyH7UYo=s17S<=C$9Tl+x{^N3GH07=
ztF4=AUS|WrdN~(oW@noM+`B{2|0I&U>RNh&J|4Nh%f<lmhW@OxXx15&EJM+(`2U?>
z<pUgXacB#Ns-cMoS`ztrd4sRtM*|AfbR0gjqk1+d9cF+nqfut25?u}8L$9m!!o&0D
zimad<orzQstpri!qxU!tlc-3}K_8cT=Uyc-Xgz@<K@OUec<a_J1i&5~p^XXpL4kNV
z0gewG0yA2h8VbZFl66kSu$qhW6WziJl#cMSf#R2gAQrkkX!FnbpO#Ow`WyBpV0Yls
zJAeK5?Jr1}E9XKVrc69sX*=BYnh%%uBcvPEkM4XOBo&s$R55n8jpEuIujF}wm1uuI
zee6Hi{B!5--O(aL>Bo<Mf+E`R^gsv{Fy|mFUoRgU%!iWo#`x~;uE{}osG<8YoD9qh
zT!>MR<WNzVupo^pOiOE8-0?MQjO29Ooj=1=5Ly}f?mj6@!*7AO=dwM&-$xY_6JyvC
z+~eQ@9$3#RwZo6i1E9fv*XH!xZOs683<NWAol#%Du=*?&6DU*-_eB14Z`M$E?E>WX
ze>nuoYgt)Y-43;=g<&#)UxSq42d3c($bC#qOqRy|SajBw+E1DzLyrZZy!k0Q8tci^
zr>oEo`b0xRgX~x`nTnJaPBsOS3Ih;7kVvGarY2~*KmY_X8T$1}rLezgt6K2JF5rzN
z{?pLO6(RfN31Zr@I<=eSa_8ah6xu@yp&>`gJbNY#_@88n58y}L^Ey9;qkfUq@B2j(
z?LDr*KukeNgcO4F;s3N>;?<{p`ep2X%)-ok3i^;FVVB|*RuFBW37bH)*aDOUm<xU}
zF|kpPG^_v5jnHTIX{f0!ZaE7HTb-)K-Fj!x2{N$X!aKjf3Y44P`Os(s`eA6o^8_(2
zG%K&bt0bMESFUjG-Qy5m&hnZg5BdYm!WiBC{rxjGbK&#z^Q^l8k5MSpAXe%y0&!JV
zM)J{%I9^lbMI&*5ui<-xZ9%3!w>S9D6%6OO!~VPTQE7=Z;rp*&&k&&4sVb1J=ISR%
zv_HdwM`wUXkBp4OdI@SsI{K+{OB#TD)u6u;dXmAk4UU$XKlEx+Q&GWaYj4L_(egLI
zRj3m6fqCu*g>(==UiR*c6R6ND9qaD?2It4-4xoiF(uL;z;S*L~plAj(%s!ApoQBJQ
z(S-mOY0na6!gUyVJotKaF^UH$y6@h-1Ic9i)99R>%YZ2RA0ulgNZYe<gO17PaPpuI
zciIcEqann^(8uls=n!BsB?6rVmRQQ@qZQ)+T+_KZC3^!5G{~6?^+38^Sl7|sJ_69;
zsm-q6zZ1p0JWH#k=*|F&wd*X5X3PU#DPBHv!m2PYuis%}*?tlb-7Sna13iS7kBJiP
z`|I%wq<A&t<T`<ObqI6`sP|>2rNui8Y5>nWU+I>=K{pT+fTwvuMoEc_`4--_QmT}W
zT_+#BO0Yt7t>54NdafWcuuu!HsX6xVCM7(3;eQOG`Tx3M1RKkWCe)yN&B6ZAJg`ke
zE!YYrkOJ`yG?D<?2kLB<@H`kt;+p{f?@5*ZHv`kNzP|A-jE7YgmM>9Y3hbDjU+&KV
zR3kq-JK(fXUXu65-||7po(dpwK^rAN#vCgJzz}tLo$u7R><9^;19n&&^s5u1M_Ed#
zi6CMFW>dMyHi+ENutAY!v{VW``9bWN$a|YH<=!U~CvTIlXns(a=RS{IA0n8VZv_A@
zsP=YK6BAZW&U099Ab^53U|qGG@%{UE#hyj5BZ7<ZcB5j-0J{P<4m5M}2?)Rm1}C5k
zn5?uj2JiO+)*|Rt?f(2CyvEPZPcQ8J2CNAX`azqH*747`M@k1@?{TZY4?RU`q)NN)
zfEz4k$9f_o<FcF{Ewlz8J~&g?L68;!{jvJ``X{xp{)=rWxb*63P%K<*Y!3l)6vAJB
zJWx1F@3olV6pV}BSQP2(swZG+2K1}}yWoMZ9&CU1IzyFFL9AxN^j)wP3UhKAXPja7
z3U!favPU6<C`3TCjfjnlL;pRgE80^4z7h2pEXs$;hH%^i$OM>tfZm0QA{7FEK?t3c
zk0obQg>VPsJ75&X{8EL0G_eXuf5dc%z$Ad*SawN(97oS@aornGt7q7hR)qAzf-Ee!
z;x;x#U5jvCh0vTF_=HGhZS9FBI2PzxV+O}1+aMgP&j4I;d1@Olj)1O~la&S4W?w*^
ztgS7#cSc4AoY77yGPq!iLG<`~Ns>s<`1mR4Cmi5I&}J>4X$PXUf8bnjXTLc(D1wM)
zK}vDy8*s4z)zs45oUc)m!Qw^$!ZuAf6dKUg(1{PyM))59AC>}XiqwOHk&*Pxk+70I
zt+4YqIG&*a;BC?2;ly=YI!+S<mh7t$Q=0<1Mr|K&CPa6A$Xq=}??Cf^{@j%f6J;y6
zns99<R|Abj)_OP)0YPEt0rZAz!haj!MkW@N0I$>ptT_o0<E&b+=Pkg|8m|jRDxrUR
zij7d`z-~F8rK}eW7=o$<Eo7gU4(B5kf<u7HAhn_5d2uoYJrf9E5{va1Msw*+_V;7I
zJU=}WG1b%1c;Iooi*7mpH&_!yfxuTlvk8xnkAayc4RxBS(~L$Xw89fk!^niU4NK|I
z;-b^{UGDx#fhl)8G#?ziQPAD<!r}HeoBRl90Jrv_e##5TeOGDtf0=aepzkctq;QIb
z`s?|2bp7?kX*8K{gX*Vtkva$~(F*T3jRY;n2ZV(r=Yh?TLITo1NKMDADzF3rA|*nv
zpm1CQ5=BH{3L2oj&cOB0+1!Yi6=ML99HSSI`b~zcTuKQ7-&{yo>r3sAmv7)4bOg8p
zu#a|o4gs(_HX;J^B@p|5y!`f|5GnZz`iWAOOyXEij{)DXK{d*#R$*0tw!e>Cjx6}g
zECa7w$<WYHLFl=%vO$?9#yGsygoE>g`U}3q9k4Q%Zr%QmBA6XmJ$u$C0U`a{xf~Rk
zH9K#^gtz^R4tyzfmuAd~iDSG`ivcMTDk_E&5sc2%uZ@gq;rs?PM?ij79ek9;M5`io
z*WN?G+VgL)+ryaIPd<dV63!%zc?i}-XiW&PqOBn`lq;sy<L1HvsD&y(A`3Ct0zr*B
zcTj$}4BcD@hB<|agQIhJxZ0^u32QdZp+NLn!-8!H-NW*)UmLEwuP8?kES`B@fHE1l
z93FqCe}F&cHB_Jj`-DM-$1reNyy1M6YgF86C>%g#O_UfbfFb$N`T=~HKk(IB+S*rN
z!kD)78{^{R<1<Nzju#sR746qA!eEj?f={w1{PSlc3wwE-gypJpQ1H?~BCZ#F3Zb=H
zP`*~tZY~eP()8SpOtP<Bg+o;uk(4^{4;jF(S%xXI?6_A;{2topM^(cdvUfa+Va-uz
zctpE}41xNXz~P0Jl~r^&+bIkxl8l8SAY~4Qho2vO-Vw_=I1@$>q~_pF546h>Ct_#z
z9f8dM_pp^#0J#*nWD1F@eWfq&4WEaDQNFj{xJD4U9{gFWB|TQei-R%6AL{ghq3_;d
z8WL{6PIKR^`$Lba652l}-qgqiN?GIZl1w}#`LNK(uvk%k-2c|3cKzt+2m#h!f)B`&
z7%%hanHU3!VH}hUpRIuS3+<QBXcedz{T%+nXRg{*J#d$Yr*9ar53A)xN;n3v&|Uq<
zE@z{C_zcs`$@@&U5ToVp2vQ)zS`Ep*GVq!8XV+7JZV$@+fw5qEshs*<0B6uR?syE-
z>kvTI7;r9%es7vghC?BTFx4fcMxG5~4bZdyiL-W`)KXFT93P*@+5t>^uNAsDeoJa^
zIFBBfgBd%MnTf7Ba5zE(98aNe22;ot(&S40%(re)CEo;I0Vw}Lf<5){Dki+k{^tr4
zuB-ViVd(Q5$XxNS<KM0UzZ6X?#CQX+3ZOQ$t~){42&%Noq9P@{p2o-fGw3teTfUB>
zrQGPv=&S(Z9%>1{RH=|y0qe<sJ=qL-kYvv;9R5D7;C0*w>;tWeF%QtSi$Nryrfq}t
z^sFN?03;ro3{u37-++D+2sz_M1dvLg-=U$VcD0pDCz=JY%R^Wu=4NJa4w1`*!w|p}
zEy0b3kQDekF*ZricT#EG<WF4NTZE?b!p@(UmnRH3HcEE&m%(lT1Jys;UJy;X)*x|B
z#C|2H1xDn#&(nMor32Lh?Ge?CT%He&5SE>$?kMS79g^xE(wvt%f%YAX=<)71$#XR7
z6PSs=>)$E0)kzpKkpS?~Yh^q8=jMQe2Y?)O{Bl6e!Dl`gzf^*5mPXHCYlzk$a05KT
z;X<^YD7gKHhlemd!N{V$x8x7>LT{!UEGs4`K8I5IDm&=Z?~j_h^_~MW5TdJAAd{jG
zAZluo%<Ow$+M5O#Qd2a}H=1*Oq7Hoq2?$!mPDZ+<LZ}x&$WJK!=+O_@NMMJVJn|cq
zlvHBbC5ncIcxWiJo?%7*k5UuVR^bOpiw3~_zw_&v*nXZhYSugJ83KM}kn)lRk6*s~
EKYdM`?EnA(

diff --git a/14-intro-bayesiana.md b/14-intro-bayesiana.md
index decf046..9fbfffe 100644
--- a/14-intro-bayesiana.md
+++ b/14-intro-bayesiana.md
@@ -332,7 +332,7 @@ quantile(sim_inicial$theta, c(0.025, 0.975)) |> round(2)
 
 ```
 ##  2.5% 97.5% 
-##  0.14  0.85
+##  0.15  0.85
 ```
 
 Es difícil justificar en abstracto por qué escogeriamos una inicial con esta
@@ -388,8 +388,8 @@ sims |> group_by(dist) |>
 ## # A tibble: 2 × 2
 ##   dist      theta_hat
 ##   <chr>         <dbl>
-## 1 inicial       0.503
-## 2 posterior     0.61
+## 1 inicial        0.5 
+## 2 posterior      0.61
 ```
 Nota que el estimador de máxima verosimilitud es $\hat{p} = 19/30 = 0.63$, que
 es ligeramente diferente de la media posterior. ¿Por qué?
@@ -410,7 +410,7 @@ sims |> group_by(dist) |>
 ## # Groups:   dist [2]
 ##   dist      `0.025` `0.975`
 ##   <chr>       <dbl>   <dbl>
-## 1 inicial      0.14    0.85
+## 1 inicial      0.15    0.85
 ## 2 posterior    0.45    0.76
 ```
 El segundo renglón nos da un intervalo posterior para $\theta$ de *credibilidad*
diff --git a/14-intro-bayesiana_files/figure-html/unnamed-chunk-6-1.png b/14-intro-bayesiana_files/figure-html/unnamed-chunk-6-1.png
index 3d01d1bf75a5695ed3327fa7994d888614436b0e..8412d84e1a266f58ca0bd9bb8a8f2aab3eb9770c 100644
GIT binary patch
literal 9580
zcmeHNc|4SFyEm4Q?4h!SN=5MtsTdTJN+OY6NwQ=HF&Sp5NFsiv$ev`6A^SE)D)lQn
zLzWp5W1o>3!x%H?vGw+zKi<#ze9rrx^PbP={4w)9_x0S%b$_q#b>Gi@T~EZt3nqeF
zcWve1;1D!DV`Ry}!G+`C;56Xl0we*g><>TzfAPGPF;E17K%AVMTwGk-+}u1oJX^MG
z;pOG!<KyG!=ij<@tAK#Owr$%41qFqKgoK5KMMOkIMMcHL#KgtLw{PDrAt50tDJdl-
zwPVMQUAuNkOH0eh$jHjd%FD~|-o0BvK|xVbQAtT@&z?Pd_wL=dZ=bTV^8Wq%4;(mf
z@Zdoe6%|!g)kB949X@<mO-)T*UH!<BBN`eSnwpwgT3Sbs9@W;?K6dQb@#DvJbaZre
zb-`e;o}Qk*zP^Eh!O4>+4Gj&AjEqj7K7HoQ8B<f!bLY;TKY!lL%<RI23+CqL7cXA4
zu&}VSw7hibl9iR!<;$0?t*vcrY_43nVry$_XJ=<`Z|~sX;OOY+<mBY+?Cj#=a`o!f
zYuB#1y1HJ!e%;N@?Z%B8?(Xh4Z{GCq@bL8X^z!oZ_V&JY>z0p?4+H{%LZQCCzW)CH
zw{PDL2nYxa41~d8ckbK?3JQY5;laVdAt50M1OkafqEIL_8h!Wf-Fx@$-M@c7G&D3U
zEbPI92jSu24<9~^h=_Rf=uu>3<m1PWqobo^Vq*UM^Uo(wo;-c}G&VN&*|TTQpFfX_
zi;ItsfAQi)LPA1fVq#KKQgU)~N=iy<YHC_qT6%i=>({R{Gc&WYva++Yb8>QWb93|Z
z^78ZZ-@JKKP*6}<Sct)3ii(Pgi;LgBeOppeQd(MCR#sMCUS3gAQCV48RaI48U0qXC
zQ(IeGS67F{V(aVcaX4H<L&Ljw?;0B$-@kwV;lqcIA3ruVHGTT@>GS8$fBp5>moHzw
ze*N0q+}zUA(%RbE*4EbE-rmvC(b?JA)z#JA-HpfNdwP0$dwaiq`_|Xj*WceiFfcGU
zI7lE6h(zMh(9rkq-$zD9e*E|`IyyQwHbx?m#>dAeCMG5)C#R;Srl+T8W@ct*XUSwT
zg+iH|oBR3m=luLUl}e@2XbTGqbUJ-;adByBiNRnjFE2Bh%$1cD7K_DZv)}Fz1Ov?G
zhn}(Z=irbHU;Tk%M(5l(ICgWG8lAMdlRPuXzVCDyF*GOshI@P7m2%|C!N9VYjs=c2
zQ$Db%4(rxGMU=+!8;W_9h<kivMa*oEKC-}`IV*MQO7Z)9ArZync2NTT#vln(Mf`pv
zQ9js>r=Kg_pPnCM@h<Da_Qh72w~&Iy2K;A}{YhA7{T~TcRSrU=5W@8%OGgg!T(mt%
z<>c6If-?vcRlj(Zdz;k`5SPU1e^`_Z#wk7iI4mZQDz*6D7^Ahz<ZKA@NxGQ4-oQ<n
zLvwTCU%*`-p%F3i2mM{6HFq_=VcqTJ(u>;@d@h8U{`!5myxG&0y?RKtL4>gU*{An`
zTWAp=J^%Cwn|KuZ@_jh@oLCK8Uj!P&cf|<D$ysg){7H%y39kO3(MqRqiky-nP^0yy
z(R`ptzAFYeQ3V%KkO1E?MQ&9^7ZAuK6m86{x_juMNYO(y54ViyKew@!>QK;vE~X|7
z=-WFS&EG2t>fim(Q6TXx`rOUbe}2FJDpcxD{CFL-gUiQ!{HFSWxnbT!_WXglxnM>l
z^U@`8XvO7i&w_J4W5wi&&Xei;KHqJDmC_uMI0V1q5iL+S3zPhD*Cp8^X}wU%IfYWR
zaQbiNEk(I_UQ>P5-^UNwH;}JVW)Od2oe!hq<$(YPiexztn^csE1FV=4fgZ0eNLh#N
zOZ%-E!1yjY+z!z`UdjL+`W1%!Zrkes1nCt5#b^9?x|9)yNE`l3>Gp)dEmNH~R6C=y
zN@K6aeylja_R6FG1_H`!U0wQzf#oFX3zC3-2YOvQ%Ii06OK6DV&{1yh4}i9mJu(A1
z{hUteD8i?x?IEv4K|K#D5XwV^*b=5;HF#)mwZ*t?iYWBfoqibStY_mr9s&4Iko>Hz
z)wqSD$htW78P5_Ll^HLEwBqG<kR~jopp`QZ4K>mFg4L3iBLeDGnFdOng@Taf3%J-0
zNy^xm`6G9f)s(Gp`5^3*>Ad=ufz-Bk7t8OB!kTQ7#!`(TT|Bu00W32Jf?VKRWTatP
zxw5IyZEQvX1Og%!lE8tQU}!kj#hr!4a>t4S1~?^eU|fpG>$S(VvY^zAcD2beZU>`r
z<UIidhlGVSt_fTaJTLWN*I^+o{m?+Tf&21PF?j-$f+GJQ3k{k>2@7>g3REOSSjV>h
zL%N+8k`GZdFWb16x7NOOVM*us;vVZ&1oR@b4lJRCeF#VTW`q!K!zR0aK@fsd5}?I?
zlb)g#gvomt;L3(?5&7>}e&iy&t)c~yYH07WJ4$_RTAAR#f1H{@zJ=5Y$;QtI2UD5K
z0{ZP6JfgfV<w?$pK?Sk~@*(rd(i?_(xG^HIn66vue@dW!eJo`Uut@*^6~Ms-!w@@v
zdVELTeriHk+lxtbI6$qvpV6IMC;H=`EmFo}xPz5ZFrdPx+SvQeF(J(4mlmtc>W|&S
zt3&)lw;NP5Y<Tn2SDnw{ZNPpA+?cK|RR31j2_3v1$yG~k!{k=EJK1x1J$V7Y%gty*
z;Xub1urze1zO@;qD@kbSN~}2NgRiyA+{sxwm|*w(VGSu%aBoPX)#{{b%y9{~dYH)k
z;T~sN=+rBR)h-8>helE>8Ohd-kov&v&R;_D({hfCJnKd_E6yOTLvg+N$Z2HJ_ZiQw
zd>ihiZ|We!8w3QWoECHNOG(U2cjQouG@(0WrE;Wi>&9?qk0B0BodURd+Ddl=ewocY
ztTp27U}M9v-t+>QMcG*ZZ>=0po%^7d*<~|`x8~b;WKY8ew%-@8kE7^i%K2WoV!^#F
zA&Aa@?6Nyu@Ae*)2bn<=9H|@KP;8>{E{fgFzfnhV<61Y7#ZRt<b~*~F_!;5;OaPF&
zq&4?5Fd(||k&Nj2<U?@IHX5?lKOUkLG|0$0F67VYeVHwCF#I-}HDs#f^lGzu0tn@t
zGVM>myFcEj#-dH`%d}3TXeZSjYL#2nc?Hg^3)f0?cHC4cGoEoJO87siw62<$EJYSJ
z;JbhMXcLk1)|w`m0NmN!_|xoT!$vI0n~|%-=z+`+4uzAjFY4QR+jL+b_O+s-Bi6<*
z;F_J9k~Zz+x^<rYV{3}=v8q#3VRo1ScN&@eV53IroBPrM_PBiNqdhClmI&7d<;0CC
zukR{{hTN&O_?SAi=E<O{HYFbJZS~2*6_z0SAdF;gasgX=wc#EQcnPgNspf`JDeEBy
znc^s4mZ15fl(9&R7jk*9<hdov=kU#wdPA4tUJLLTVoW;!4&-rppS_%{XO`zWZc`qA
zBfTWL*8<znUby{2^<}jZI|HfpfwqG5Uv=f?cT%l0@%r<`XYfzA&Fyjiw8Ci`f1M`E
z+1XM*J$(Dywvyqz8S(=yUS0c?dGGnr5Zc|b3RL>odBq`HIBcQXkS+!tR|r$3dRHb|
z;YfK?NkpM$s{*1?)ycYGF_79Ww&-FW6A`?{fvRkp*2z^3#y)F$AZPgrm@9_Z82V7y
zyxQz~KjzN_F647~(b7!D*)ps;#c=YgDR)RFA+^q@yJwfYt%yGF1Lu}-`>-^wdJ9Cs
z(oDdlU;zyK3fZ)(wHI~V#;M92Q}NsE*twnv8D`Vucf3w~F8#%<Z5wL}=iK7dGY+vk
z^iqIxy*o7KGKSUi>DJjnVn}fR`Ledv?)s_U@g}7EQY~`~qRBeuA{0W;!GJu=di`e$
zD2k@mNK;nu`lhBadON{MJ6WipOWS^BxkkZdaWtgzGT$)-753bdwLdwlCK00&RT0Ij
zYeMXY2VG6a-dvJF@S}D`pk2p8DxU*2eMd&UP=t^{+P)RRdD$6I_DY3n!eGtZ%kFxt
z%DSSD(XJ=gG~w^v0%}wvH5Qz~)iW0=E<l}ZfJ4wKbHqUG`T8-nGJg^H83lj6frH%4
z>GLe>vr>Q=j{pr$;m$p)rlhB_ypv-xB@8#`733Kg(|yNat*9bG`IP(D6y#>$uQD<d
zIplVHmT!$$Je+zV4;X1qaU;>_;rRJs>)k`8HYpXvCcgnub*A6yn6>1Xu4UAuE4XF7
zi3BI%XBqJlX0emtWgRnDTzYL&KK3QDnjif7Pk^ryoO&6Le^UyoyH@G&>JeoP&aJ((
zwr=b968v|1v`~Qy2M&S*o7Iw~M(?ewqBPW+qQ{sZUDX=dMR$`}x0{M6rSC>9WXgjl
zB+Js~&e|%riWNY94a1cNn0>uWa#n<O_B5sF?Y&LKSYEV5Qb#+wvyqe+Q>FH1|0Y-1
z7ZE3xTc(2{7uhYomxVU#BLfz?`I|TiuaNx#wjVZu2#Ko(uT{Q`@pZKJ-E=y31mny?
z$~XF9YjWUkV${i8-p*olJ^k!(ev%pg=7_?{8Az{h^95)VxjQK*+;L4EkA4wsrQP;y
zuC@vd+oPg2lBTP=sb>Vbb1**b#6tKI`^O)@kjNK>Ch9F$*dQo%@~p31N`+p2r`9PT
z29^1Hw0(JN`#!BLi>oHB<nm!J?_aV%5XrZ0tN~CO_y*O#4daRMCz$Achuke+M`8dJ
zv&8V2l$B;>KEX2U+4+e<G2vkn#+yuQy>KaBo;{k;n%}v^wGwwq_`vm5k4EbkvO0_B
zU@bQ1lJP%6*R82ARFN}fUh9AVM3bAT&nD8TxmdsNDK%ZWhtK)_B5=KBn(kc7=%nE9
zLg$>4KsnU16y}mV)c)BoZ`W1{n5CN#nGHc_5re-OD;atAKmcn}hut$dp9G=)jVNLu
z`5Ex&A|xZk7}D7I?vdge9|5N|t7r`}EFtWKB(pV|_R8uGT|+&k>VRLC)R`WL5nM-#
zC?_ngxbw_YL<;%lR842vrCxf5tQ?Xus4?_P-;|Lo*ETaZNahFoQvM3M!C<7ww+|1S
zRz%1ohT#5qIAl29CeDjVcF$6jKV`B;@~<L5NC8;4Gz@V^S0QQLw(cOh8r<IZS8@Aa
zuORAxl;Z<dD5XlK?y9n<p-Vx#U8K3AhJMZ)tSa3wO1g`_C8B=sODor^^4#%hyqx7H
zaeRmMg;yu>@72NiNVK#xoK)DWsXN!P4?MPXEk;vmRp;;t5vX4nUZBjW3i7TjiEl^e
zC8S)I;jhhpB-BfV+A5v!<W$V|waGSW@-6NJV`#4Ab0u&mJzw7lpY1``Qft7d3vv>2
z>YvpWb|p1oWPvl~4~xOgMt7bQ66**W!>;4b6Qjm*-GXb-p1+TdAHVf&%DT>***5~8
zsf%x3d0*`;8T2x04Z*CFn;ZFUX5Wnh+H{O3c&ed1NmveA3=l6xuI(G)^)fiUi8p<e
zlnyaJUh?7>O)ug~sI$`n;j^;!fi<c0-gYWZmT@m-DQWrXBLT1;-ytgZ7;wg{yoihZ
zh~mb2GW$yOWmaxdBjJZYu6F$3yT|o7v6hflr;150a|SkBnVFM(T=&PTCwsUq2#<R^
zcdVqgjofehIp3y#rA76|!k*6!EwF8*`GeajyH?IDz|)*i9rPJ|34(P&NG~D`Adb4A
zUPPG%>UgkYMtu;oqoLXUbj@HoM)l_z{!KJ0CO#N8$2SO&qSzHts$J$cAa8mqB!gau
zZo)Bfq|WD<tX(d|r>?}!e`>x3HWg-`=`m86x)i(>Y-zAYz)poBDapymxYqXg>!@8G
z(b$Tm8A^&>D|x1UK24Qsv@u*bx|3xe_jPvgj(Tvg=8z_WSu#&@LL`v}*7$wZj$7C5
zf1c^Bpl~@JD+}a<4gcG=8h-@*Z>QqA|7jSl`=9&64gV9_@IQqO|1;e1zyIqUSQs1g
zYQ95<r^uK-1^bwP;ztzbM7-iU@nTOnaO6DF4WjQAU6cJhUHL~hNM6pb{mfHPHt&xx
zJ)170cF^E&XHya`=hZ8&UBj(+0guef8Appj-P2nov@C8L@K=_!ksc-PKk<@#z;@i<
z*lfG(Mb24OYSpWKkA(JvV3G_63vpxKXJ{RPbp_7iQZ`1|JJfP2O~v!o|LN}9?094t
ze{z_%2PwPk<}@aypzrc+Pl0i6QM`DYz`x_6!GG()BhO|{Ag<q?=)t+=V4hYd5a+k@
z%E~@C^zi|qO68~^tZpm{Z6Q^VkopQ8t5WKeL;DNiSXiax(q#-(+YcOwjGwY@<YJ4z
zXu-Ins&6(^ySSNAsXYweg-lzp8mx8<1j;yUrvcUiLAsp(a<Z9hmc;}a;Bq4I=grN4
zbsRrUOXZhWODg=E@Eo;#Iw0Q(KWBI2{b#6#%f#|2$g!bIURg;X`H82Y;b_vN<F_es
zxJQ#$0`|3Py0#3(!|xK2B0^EE(+M6-57u!~Iqvcj8NywAlWEeQf5Pd@w%LGU<+V3~
zZw5f%FUgCnHaEB8R)}~xP)2_agrN;0V$_x{#)4DG$EFpARA>u;2H5fuEn!`hZ%JS{
z5WYidnl8jIF0OFx)eG5gDipW4*w6q-)lDA&Ne<jyAjzSX_TDZ-{<9Ogb4tf)9=)Ds
z@HslE=8o~AYX2rGpZHK9yQ>!3ayPn1cRe{!nePT4gq|q1Bo4sczl^0QJI+5?@1W4>
zN{*VHo{96*mCqT+8HCuvAj9=M2W<WL;!Azl#M$zz{=X=%v6^p*{#1;`9%-#^cLuUM
z3$*7uDR;pD8aWv9#fTUpiQf8VRCaPoXJDo9-NLh-6d+tWA)$^2-rsZ70d_^9?`~Cc
zdHL=i7r$?<EKY~bJ~C_ksnsd?ME%fAm9&1~u(Kn*b9%-M{&il#x+mf4%+jc}CsAtL
zS7(Gk&sqx|xEv>2@FY8=(@quG%*GZb_&f~t@Q7*YsD?$)E_qI+Bz>3N)7j_$r2vmL
z=Uc4Gq1U;EweWiD0Ih1zV$O6BRlBv%_MF!?dRJ6h=TEI>-Ckq?*~`}jv5X)k4KeS>
zVx8q<arOlWNYqj~i|o<0oSMa}Jz0*vMFMA`)7t0l*!e*o)~#R0yt1SfuiEv28J*vF
z^`&WjV-@FaE4B4i|KT(Be$?8V*d(*`z83G?BSOYz6Kz=vgOWwG+NSN4(<_<N0ht7E
z?2T##v8+sl6Weux5rsWwCQMh|`nJeay#QP>@MvcFMGY0!nL`H;7P&OsOAW!{=jAT_
z@}wg_Hm2U=$h^J45U4w52E@vFZF~2;z^B^#d{{tq720~)sfm(<Z=BR~nF@+Iuo<(^
zaMl<UiMViXIS+f(Ouom9T<1Njj9>A~v%#l$%v6n9`w?wuy8ZBoJ4B6JAp6$f^Ptx5
znzud5RLl8QTa6c$W*_cqmXFZZMwM$UXV|sEp?~$@AFs9bC8FN+xM6@{vkt9~x)Ih$
z#H^aCvNRpz@KRgL8PYQ^r;{jGY3N+!KWWpGu+dmQpaE0T^N<ei+yLv<9DCToZ4Ii9
z4i@<Xe73?4qPyJKUVo1f`7Qf(#x=WOpDKv~dJ~nF0Wqp2CzPWLRxO3;>M!XjC`i*t
zTmve|<9?6VO4wkPbXqjor1(OA;-LyZz_e-Y=06}&&|fpfw8vb>&?ji?)^+Hc-OGq=
zEgh`1Z;hUTZNJ@z_pZ9!Pak_5EGhqmxMLXW-p5&<N`JBX#+h7K*iO`31((0GL&zx)
zqUXn)0bZ_1?%`O!JOH3II(rk7SWXZbgwhzIRrMx2z~#%TX>P%Nr@Z>MIUOURml7u)
zoOuzqFccaSa$0gyxkMX3WuFbdHMj$x>I@jR&SRpNH_hU20RM`YR1m(s4I-KL_>td#
zC|WR|Op}q0&HCGmnH$nH<JvZ_R<#N{VB!ckehw48eI5&hQXHU!jIE1dVT@O!{?xtS
z<gqU@$lwXP*_6uU+J#|(kL-(IowbQIm!_8Lz;IS<x()nDBbF3zpNoLNuQ(%^%dP}^
z+#Fh<CD>{+*y2l_vXj$~is<=Cen0xUtUgD1%dfP9%6w#|<c&W1fM0$o*-Tr8eE@py
z{c>umOwmkjlY}~l%6};fy_Tbft|X_kOvd-(q&N#>oKYxhOjm%w;vyo411NV-PEVKF
z<C@5ZJR(qp2u0HfM>~n@6an?{g3JxM9i-y(zM5r#^r`Xn!F@V$x;hI8Ui?aUPs&{3
zIaFM|nwdQIM|~|TYF=*<_s2v!-(&RDwQIPZf*lC=G1qUfmBr+Cn0v@!81YGv=T~Q;
za7P^@ZU=c~y-=3+WD1IYAC;Wh)o(|x4~ess8eAwF#aK+vi<4`daw|}CtmQ&fjiq*|
zoh#M~1AGFbzWObIM)mF?#xc<%IGcr+mO4lanIH9|*$bBv5VOKv6E*t==#l|G3-kmb
z>}6W$6}5eL(NiNvxa=t8;62`F4E<&w^UJQN$e)d89sR#!)3wty7|+2SdnbjnLyBL>
z0Dy>t*rav=eap#U4@Sk=75H>?MXvri!L%?I2goFHtuuM~I*OGf<?^Ii6S;DCjq5d>
zQma0ES;iRGuDGfjrB}*PT5$2c=e<k&2I76{rw6lRoc0Y}5rra>snW+!b0Nv^(cSIh
zbi}a4*%U@<q~7vOj3~5}uTn1*t^bDLUP~Sx6RsuRL<D`IT5?89f_lP5DDGWTF{l~F
z@UW}$to`c?C=cq-CjMh~0Gg+8!ux5s@#W76;;5YyE0ZZ1@g}T?=nP{Vt=KN*d0jb=
z?oTILHMxuce$yT0oLou9ba~X+V;-9y3?Ou2Bx6tEXmi@09})e|#|{ip+07R*4osN6
zk1mW#_m!z9BQps>bn|ioGkCIl@)^W<v8?Wi7&OH@m6HABh$|?YP0cEHHmuf#Ig@+4
z&Dn<Y{k9Epk1z05p2lO{#*$`_a3>3eFo>*$1hVa0Fv=OuUV@}9;R)HDuJN@CRF!*g
rb^x!s|HTh0{_$rQzdh|`^GZZN3Ai!J)&xE-;V?D6U{rX@<=%e)YR(A8

literal 9788
zcmeHtc{tQv|Gyc#vL)pvsU&rah!8Qlm2y{9ipqK$nQClhZ;TnG+?FWHlC4ElcEw;Y
zgR*3qN{n@86k{1<Y%{}*nfZ>vz5M=oe$RD1-|f1d?;m{Tobx{K_4R(Ab3W%2eb(lL
zw3M=xkdTnH^~q!Bg@lB$LP9_b31NOuAddHu-{?7e`oeL36951Jfk0tlVG$7#QBhGb
zF)?v*aR~_tNl8g5DX9$`Hf-FuQCeDBMn*<fR#r|<PF`Mq)22-d3JRMyZ&p-P+_GiM
z)~#EWl$4Z}l~q(!wr$&{s;a7{rnY0pj-5Mq?%K6WU0r?m?%f(18hiHa(bUx3yLa!t
zefxg?`DZOHt^ND=Yin!k=;$0ca6ngA_u#>U`uh3?1_p-?9WpdDG%_+WHa0$d_^_#|
z>5(Hx%*@Qp&CNj|kcEZC(W6H#EiI27J7#5Nb@Jp%YisLMr%s(defrFqGd4Chwzjrs
z&z?PZ?%etF=k4t5E?l^9@!~~$dwT~5hf9|(UA}zz%9SgQj*d=FPR`EGE-o&vuC8uw
zZomEZo4dO^7!3CC@VI*Qs;8$X1OmBs?b`M0*Kgdo0fj=nyu7@<y?uOqe0_cW{QPd-
zyy@@n9}o}_7#J876cijB91;=|8X9{0_H7sp78Vu;hr`3e!x0Ds5{Zn6h`4j-&fU9r
zBO@d4-Me@H{{5(^sNaA8Jvur%CMG5}Huk}T2l4Uo2?+@gA3l8a=uu)~Vp3Al<HwJm
zJb99woSc%9lA4;DmX`MP>C^P|^o)#*%*@QJtgP(p?B~y)=j7zPdi5$dH#aXY@Ad1~
zZ{ECl`}S>qettngL1AHGQBhHGadAmWNoi?mSy|b;ckkZ6e_vi+UQtnjLZK=vE32xi
zs;jHfXmm|YO>J#$U0odpgTZ34_4V})4GoQrjUPUI`1tW-Q&UrOb8|~eOKWTEr%#{S
z+S=ON+dDcsIy*aYINayYpTB(h^7ZT2uCA_c-@bKsclY%4;PLq0-rl~xzW)CHfq{X+
z!NH-SAp(IwBoav^()aJ*fBg6{GBPqcI!Y#!$HvCS$HyloCMG8*DHO`o)D)FUot~bi
z(P%R>Gqba^b8~a^^YaS}3v@c2!C)|%%*DmUrKKeni^XQMmzS4091fSu<?(oBO44S0
zWJ~&;yzDO|q#CvI2S^y5y(%QML&*Br(F-9N)FEE>&_y_DR-r&-^V>@m;YWvp%AYwG
zI-@B*!IZD|xQDv&sl_t!+Dc1%qSGU<eJVQ_H+TGewcLHhQ~O&|lwylMR+cz#KKn?d
z)M{sp#bW#EgG&t_ExOq}HyWqi!*jahFvOfmU_v_CNjS=qr4Ko&Ep!_$6w|Id5^D{3
zoE%*X5TdHPJ^LV`v{w%(_G=t6V$(lvc5R<oY-zbxN=Z<xT7LBC4j$w*GC484^5gu-
zBv@Ttv>4<vGWSwFv<JU^rp>ynxR<^`^^2u-Jw<kN)oqj1%{AQTyByTcB;LUzs#->w
z8lgR!Za{`QX@(_oL<=ZWY>N#SDG&lg;WgbrKzk&zk)lwd2GFv=4$~!j6Of8KRw$=V
zV5Q{zLg4&~ZlV&#HRb@ed<jrwn@cv5BXJ4v-@?XG#_JqKs+YMND6@kUMp4iOpwQ+3
zT+BbGG1p1N$!T6>n+0}h#Q{FSaA20=f05O(#w8eqM(cyUUx4q5Fetv)_k1h@J$}!!
zZ9juOY@~m7e0WN0afYeUcw7gRYK%u*FB0BPz(-#J2&-!AQdpM!?#G+aT(_^2AUz|%
z*kHTG^Qd<yQX_F-ZW!gU<7N@a1P=(lZ~MHFGQaP=TJ|shd%~q9U8(v#%JN@_3$nMh
zbrWd&3M7TyH8oEyhX{c7duGP|*&jPsP`D%LR@sMys1=C*I}#g@bF;>j%?a+BW?hU#
z8If&rriU+oZ??o9&G!al@$xgJ62>6Dt4K&oPQ^Y_H)BjaZPqX)VAmhsVz4^j%Jz|w
zFzQK1=skHqi>VQhXf;^?%-f~R!`cCied*lRf91xOzpy@~EX;l4+|4)-_)F&*0LaT|
zF!UTw(~w)X1b-Q2s+YLiQUGP~7^iT=`4LuWQj$ja5%DAs%;F1aO2X!<<+I<N5CzYW
zs2=XCkl_=B&VIgKY_Uxv^62bwcUKE*1NbZJ{iJm2i>UY<ao^Ihsa>R?57U~}y<5+2
zIxf2WO*UzY9vfu&gDs;v^&w*SPgB<-F5}g7ZY}j=nn#>LL7=Of&A;?xFjJI8G2#13
z!mF7EI)4^>!pw)=@ACDziR*_LRm-35**-qtE*c6XfVck2{Z$5d-*xerr15J#lBFZ+
zOaMEw_B~ZYRYECzSV4pItO?ni(tqL|&8y6)qBchu(0Cy1g~&Do#hHpida)_dC&fHD
zR~;3oO&C~-$VSXe^d`U07qCN&&VCdabp>5n+@=%$o~r1)&M+6*wJ?m+nXX#iGd#H3
z$%nQ;+W$TOhy5TgE*s#ENZc6o9$P@6Y28|?wVtMfo6lMeE2!Ee)~XWS{5YIHD>%XQ
zTXa7!CL(@IY9OcHJq1@AKBe^O<kZZ-!PL!Fth0&6cwbgryzTR)&DAvxnxFDMOUdaj
zH)#IKua*DF&&T?t-n-6<!;U*vLG;f(%SL?Ckp4T|S`yJ$ft?Y^ou{UgC5^>__xD#h
z?jdcHQ2Ng$_z1v&8@n3N*S7*JlLxi6WO`FWd9zJ`Vd?UF0@;e<%gwpuCqZqo4X_)k
zYedFFf!(GQ>Fny|0dN`mhN7SwDpTp%tcY>-U8JX&?W%l}EU)X)I#f1?gl?}|j=SCH
zi@BBZE%%BQ_Mt{_8%x-xYgXK$<}G0cgW35t<-NeLDiJ^e&y7CN5$RAIKj+&hvhngQ
zRDl_(l;)BF7c=7tqg>{T`YFw$jAS)|%zT-U5tT8XO)Vj5Yj|KKcNT<cYiDn*9~wO_
z5ECO@$7#&_Mll<$F_qa4Ur5z+RUB{(U*&5|x?NlB97Jo0JHdVsYT*gi-YIZ<2Oyge
zOgwd_&AZ>=dzf#jK()&^vCI=(x<l7`>$aT0yzj_jv`>>b&Lud#J-+|wnp8u5HJSmW
zW$Lx6K$KH565~T|TW=QVlTO$c>ya1s<#p)$26B>^%p>?kHxGYT+@yAcD|eSbjZ6T7
zxQ*gpgt%4iz0tVpL;tIuqCO%yd}vhz2o~ew0B*0oj(FESi&LFModFb==O--~X-$<=
zeR`t>(F+tiG8>T0t6s>ws8+^bA+B%jMoh!QkK#9k2B(UzMRT!XLnH4SS>sf4z#*|g
zU(Dm$qSB)5l7ZDo#RzwTt-)9*l`%z^>BBvjTtVQ{qEsb*y2S-G@Z<pv_C`G_$-@_-
z75QLKl+>j|jE4=j)5)%<twy()_7FZ)h1-4W01$7rX&>+;8n(#$B@AWOn!7s~)cNjs
zlhtb}h3S3w<xMW}{I_i-dgP^OqK&a#JaTKVSsz8AA^{1O-M8$Hv&xH+3hDcX!=@an
zqmd$Prfd`I1vw5EEl`G?9M%kOYitd#`WSzRzm@Z4dDnG#%WT{jhkpT0R41LHc1{JN
zYroQ`gJzHK=V#XF4@&GM3iX3p@6D8<Z-P~V545l_QNHyVwM8hR7tDG!<#a5u!ofqV
zuA5BRL!vc}I>mXqVLqsNh3TMGbWb8tS%{0&0a}(%+>(FrQ9FR|z*A0#mKL?CbYVsr
z23LX6)z?sk_>@ZMx6SiyuM^{#k2&T7rz?VcT(g{S;2gWX50HY@<x4#Ia5pB2#~r@u
z{FHF5Jy(;LuE(q*bZzr939+O#)Nw7T)Sjrd#H{(8F?)eJ>NOwHk3#%bt-<V87L*D5
zZ8K@)$$nJSSUT&#i$O1fc6Dj;I$B+?P8Y)2&-CaAY=<B=ibI!9tpnQEz%?HZdlG8N
zxm%n{d^G3)T^~W*n8Z9=O6XXk%jz7Y{dL6QsRIJCyazm81V=Q7y}J+uDO@dG9|1I2
zp`NDlJUgnZUN*~v%()Z-)m|UVljRoam1p1U3J9hKmQOWZ`9+X7Plau=DVUH(9JCls
z^uX+_%C}$la{1pYVxahPn=b<TbfoTz<=Ry47{ZK_9ev1XU2EHpY>sSsy2^U8yMQ}u
zX?dXZT{Y!f*1dr{xFIsQOGn_g9+jEJXsZR=T#Aq&F(}fDNIlR)f=&7=M_=sAyj0%4
z+DyY$r3lC_bN?Q|F!lXvt|^Mw@DkeDh@We+8SDH}a4NH1i~F7y6POf<>H)6B$1e+$
z>F2XfWPAOuD$)OL=_$Dx`RBFo6IRn%yoMbEGUbRQ)rEaLu-mBn;3`AanK7HJ#ku*^
zwBUQ=6<3a}xrA^wmg((GN7^1DZ|Zwc9=Tr7M~=*k6AO7`B!e>rhleHmGRyC*58=s&
zZ1WTVc`qSP6QX?zG}fI5GRW=>*JXc!wque8F6OWMxVV!W)X<7Vr4hFb+Q+W<=B1^-
zQ)>@ju$ot{_b&f^owp`gy-(H?R6SVZ=;x@tTI4<g5cR3Bu2v+5`6(5F{%O4=k$xhh
z?9J_DLzc7fY5Mp7>O0oUt~fnkLo&SB$ddqS{X@yaqyXGUO&hK!7ZiFjU|m2rAaBAB
z0qYYdvWIbl`Z|?)J?raI;NLxcQLLu(%|mJODi(hh_TkR&U9#IZAir8(eik-|fBk@e
z=6aDwE=UN=`3q&%0zIkS)R5P9lGVWD7JG74%T%C}>!v_7(_$02V=r;(NlVfk?HIG1
znI!a8zTtz*c@qeRSC>}#x=!DbME9aBE=k$nN|PN@svec-@%o*jry~LZ$Zy(yezA3E
zE@I^2pnJ3OuNf30&N<vPaxA_0G{P_aaA^`UO3Wm2w>4x#dv!P<c7<Hh#s5pvgcD{$
z-zpf(`xJJIbvL#{#YD1FKTJw`Wo2Eo8^`G8QB7DJ*P6~S3okouJ2}6p=(~HX`p#{e
zu?Kf9b3X(l{F3!tx2}cP^TbW;L|&X8q!>*{6q=$LnK}Yw$!9`z4aCE8Xpt{5adV1r
zb%6)eO>l14WQ@-Q2VtJu(`t-#xM?FRJAmIi^_8lHi~4vkHQl)*={wzyjs3Kd4uDz+
zMnp1%0V^}*{A5Jp#*f@Riaji+`(l2h7+~!G2>vBMoFgMo7&7SvgoB3jU$&S|q%o2M
z)K{Dzw@Ez1$(?#0il54!oiYJ7&9>?*X~={WI%no!F8@}XIqOx<Svu)iQ>bAkq4XtG
z>}AbgHxpg5{L`x`wBrtP1qYv0{4kvJjOr$z%-e?j<pr6qJS_!*9hD9#9LlOC=>6~?
zckR1K-G4|0XcdrEzt<OA8+3wEJyeu&n^7|4H+I}CyLYjTY@{~PaxHO+>UFQL2V^kd
z=$;_!KEJUkw)ORlfnjetf1!`5E=0nnrl;ugX!prJdnmOb;R>2W$TgDPt+5prCg(UM
z>GgTxNp1zu#f|AvY};G+24XaDTvpXumf()-JB~Fj`jCYc+XOg}&iZMw><z@0quc6&
zxn^Wgw}j=|Y}9OZJHgK<Ze1T?6Q5l=M3<q_*wEGMrhBu9muP`a>Cfz=1i?o2K#aw>
z>?fi#$M>&zb1qySS_|I+t8Mc!AvZI6?<Haq@BZnVTa0Q#`26_GKMuE<8t`?m(n{JW
zU{IJ3v;Jj5(;vs&cR8|xlz41J3V1bSoV?uyTwIJO@rH6Jov49V39lwvyq#m273DeN
zt8~UBNJ{^_ec-@`GU9`8-n0U^X;6N7EW3K@`-y=oU!nxc{QBp%qK~NwwCRchp)1Gc
z(j)y>L2*$02t=UQ-IWu`N{+6aNLE5D!9S6#L~h5*iG&XaK7pK!F=OwuxFe0n*3wmB
zEnQEq<+HB*YCiv$6CB)>dJ!1(JG}U{M8stcx2RrbOpM&WSy%U_nc+t^HK{9afIjbK
z)V|my;u*Nt+9lw~_H+tTeA97vqSntZD&@DQQIMvMd7t;HM_lfn+ZVu(_to9H`65$t
zY?pKLz<0l5WlCv^iaY;pyP6T$(gTt+@=9=^nEOBc|0(Bsjil1vq}<s(ZqH6XZAwQ*
z9K5dV3l#fT26PNak2>IH`KY)KRsmQf?(H>Y+Ad)mozkE;<xmE0uE2l2&jsqBIpJ&H
z3x$OC90polG<Xmkv09_E1P8u)2v9=l;uiyg==vGY@h(n8PWSDL>oprTt{3L{t9X6D
zDodd7bh&96o*Hy@=CWsvI=CDO2W??v0oG@JIBy7PKg0k6n7u=N6!oCve1aXV8jK0R
zqtOjlkL@6d2H7%Lk<<<5Pz^Vs!zV>*ye1(7ZFg`78`&nMM%H2`SKNB&QT(@(d-nx#
z>T_DUncZ9*>N7p5ewPE$ab{Rn0~YH<@Lzlh1i0P^_ADNtKH;k$$-8~7nH3f5#65d{
zh9!KwC`)FKXffh-OZSPS91`6>?9m&z=%1BW%3NXn{shjVro6+`8R;udSi3mE$)Ib@
zlN-l^)HeP}SZm+^06Z~#h{0ls4^&K>kkgdbu2rV!Q|DoXw?43=4AX>aYG(5h1<SSD
z8IG9mxN|Qabn{3O!06BJ|Hez3h!k<L`?1c50ESUH?PJ6pyxPyV=mFQBk?66~UT*xy
z2I-KH9XM9Mu$Y_$)*@P7_pH0f#>YwkVGXN)*v!>hhu>0m72#PFG}zXaUFy%9e4Fh_
zB=TQ+D}aOF-SWcRY<_FY>+R4kwF`O!T7JrgWkJIpLKZHHuJ0<CX8yHDW%1M7o!q=#
zQB>9q&Rxm{+(Rf^+SDG-itp%<SiDY+az*tUie*6g^J;pEY`zi;O7N_#6}jMGb~UxB
zK45CPs3<#!uEOH)MUM1u7j}voC;@lp(Rv44a@6{HSvNq-24>%U(}A6bSWkiBUv=K{
zcy03=s$Kj*tMaEyvOCQuCQvFjh>ZS6Ct+>%p_Tf<1*`?(bl8MKteD+sRCYL|JIc!r
zyJY2RhvhI%^-+}3-JEv1n3)rrUBhXfqFcklpT-Ai$Cj0bISu&smQHrbdQBU$hc2*(
z^U_oHU;f&y<X86vzj@{g+6-0;ZW@!6#yR<ZZY)RVcOs6!IQN{haxrE3FTD1v;TbAH
zeCeF@fm$~>=Qwj1l58NB;k$51bPyWs!3X}YOntn{f&_IspYh26-s@92|02uQYkAa(
zwtwp%lN|@>7X*`^0y~ZPhEJEG-{DV8Nv+PEnlAPCrb50KmLu9OFP?*L9W))|8(z-2
zubOWTr;)?Gj6930WkFq=SG5MBHDT}uMwKAq!NF0OTk%8ZVYPiax5rQ3p4kHHEUF?7
z6|7m#^qICBTQ-4p!xz(Xk@FAga*-)#IGFaU(~u&g@!^3Xo*9u5me0-13s3cstxJVl
zUr|}p8=ekpaWR`!9g51nE@ey{he|D;N9)6eTzFfp@GDmP(R>zfKrtnG!w(Xm%lr0+
zRQAPR%&znG&n`_cs5>_{3|X_+#aFGxg;th6C~$KX+N_q5WNTP^Gr{@w%vOa+8KVJy
zjF?_Nw)nYR&^G5kCr+bWJ6-StzSV2!j?#IL#{?{}`L_;noeGSvRRl+wc&gF#CHe%+
zkjbn-d^i0h3woEcgb^P534R8!vRgwu`aDxLBtEnjVi+W>Nx$x!`BOL~srF_OGx;u<
z{a^*;SsjI`%k6gTKeV=S_<xGc<wo&I;|iWNJ1Ir;)dV$Y0({-%mLvXq?W5qyk>2nh
zE5<{Ez5>vsao01imS-rrK%cApWLf^RH0cOR6Lirc6N~}{b$+8oEm8VUtY3N<Fl;in
z<q<1aQs!yqLO5*ci<(3DVxS!Wb&&NIXd}K_$Yvcrsn$lO;gmsV$5N>v2t+I+R~8X;
z6E|w4$YQ<<!UnPgX<Ibp>c8L`K?zU`zwL#svT}963h`g%{G<A^?66yd|4oeGTY$fn
z9tYe$GU5XPm?P&GxXfZJV4nC?AQS3jH7u~*rJWgBwVhjLzA&GQNK~3gEE*rfb|1wy
z?*taf0L^9O{VqCPJBU9c5=M08PFvH);Wy@*Wqg)pi%mzJw{i<+;n#2TM0un1ybO9h
z%!Q#H(<@Q=v!_^C_hhsz8T6R*IRTft=tme+CNRJ5(w{ic{C!&0MZ`(9EBhCo%6un=
zT@TXHC418oT#Y+Y=Y3fHZ}Cd=OKCK8e<}(wJB=<1N9*A~^y_;b#dhndmW1*mSizl#
z$Z7p;AIYfcH;BarwbZ0wTnN#zJ*`j8-DKImOA(A|i9qIqO-KrRxOI16aZ8|bA6tVY
z#O*mwlan5&uF=;V{JnVoB&<WNbIJ3OEWn{T0Lu&gUY`)R<nokTsE5f4YaVFNeidK*
z3;#Qp?p>tGRGGye3bNbVfOYLP6XT|xtpq)N(kx_qP1S&0$(0p*0Jhu*Pz3<ubGu(d
z9P^dwf!oYhg#^h<TEL=saQM#VV^~_|PbG&Ba_SD$G^8pJ{Rt2ZM-ONpi_C9Sn2G9O
zXVB0Xh62rK5|MZH_D(?lacuXn<f1;AeuCG@DlJXPBcj?fmX|;LNKg<+%tF%i$XU?|
zNp=t}B!`(EY;$BgPhG$eoOxIq<}DSnTuTll<|B-6bkw+>#?Xq}tznbJ>e&Y5f-0i+
zt(~OFMHPKoQ6_G?f**@VPIJ)Vp}p~0hi56q@pUvGFd>v}Tj9!sk-N!}zHgQtYmZ_T
z%L1EdR-aCbWpKOF_wZD{ew3zB(!*F2h}UrRj{XYyU-%@$+VjMKbahg|co&a>vW45y
zI-8iUg0nf~C0{D+2z==dJ}b=nu`1Z`pIRvYhUNd?s><L2te%KODlap|`5(&&Ss%AK
KR`iS8o&Ny7_GVrH

diff --git a/14-intro-bayesiana_files/figure-html/unnamed-chunk-8-1.png b/14-intro-bayesiana_files/figure-html/unnamed-chunk-8-1.png
index bf2b9caf16de4c1b9e08b04195aaf1c3644eb93f..4eba332ee2430dd35b64807e0cca985328d2fbb6 100644
GIT binary patch
literal 19227
zcmdtKbzD{J+BLiYQ3OR%0VNd_1L+hgkx){lTLA%S=>~(46j8cU8l)R(L`r(WA_Szn
z`@1IYea_k6d!FY!&-tBy-go`>24TgV^PX3XYm9O8?vb?EX?!w#6bg0v;RE3(C=^Z=
z3WY6*hYi1Ju?UHQZzs(jC|jUVlpe_cSpE~s&rzt$sE5M$<ZL76Fb<9vy7v#4+cdru
zYo2uXK2}}l`4m@JF5Q*5_B9Q@K{>V?TE*0p;R3#=Xr-^pcpS0FV`2ROlDgM<95=kg
z$7He-&SK$WpToYQD8yJAUcPq_Vom+x+|Q^vBWpH0KI*-Dx^3U@<m%~dmg}{B4=K~g
zf-^!b`QPa}jr;?B9-AG$#kiu7A8%Y1#DQ=4c<u1j75m?N%J4I}>)~Tf%|+`{ky_o{
z(3bjb`!O@_+a%uI1`)mCuN=M^af%5Gzb&~^^0HWsF<4?MC0k4;pR^*%%hNMk>2|CM
z;aReJyFk|C9yO2Ekw=CbTg~T;3@dJmSQEFX2<ha<4&SCN-Ajn=S14uCVinIWE{3og
z$aHsiU&>EDwXtyG#+CoTb^Lo*|Ce`VyO&_0T!U`&?5KCTS(jprI0uXhf)wUHQRRtZ
z33}Ay{ioOdFGv4R+|0jxAaXhQ{5w<e!J5u?+ba|5C6+JN+6PQ4%S|vPy1gnnS|XbZ
z7$+C>{6@ZE@2_9K6y)XCr#to9W9|iBmnl%X?c1}Gdl5p5Cv4+FjD@-RoV7#lnMfY1
zZ_x}NgMxIQKPSF$!SjKZmX`C;LqlU|u-Ia3Y)sx83t|+<fEuf$pul>6XMKBnd-7Hc
z?o^#O>E={Ne8>HOfPjiaCgm>)F)>f8T(SK5Hk1qu3?4u3=#&vCHyy5!Pg7`#xFef?
zbo3arXq<B6fr9>AJ(*C}=bcGX3>P}FE&ZwZ;=-6iB~?OkP7uyvO1T&rO@_+sQc_ZA
zeq^W?x)G47+zy%_C=CAi@dh#EiD1Kp1*{!7yoiVh9RK9WlNC1xJv*mwGcn1Ci+g!_
z89bA{YqM}6AojioXM}q|P*6{$i%V8-czAf4+$mTFO4*_lQ<vCu)AzG1S_Bvw88__9
zrP^*AjW3PX(6H(>z87uIzlM%KAU!%*7ymE8Ldo#TfR7*ZUXIoK3mg;`6>)QLaM03f
z|44XP?y$K*On>XvIU-{P(fjBR=dVluah5s}s7N?wc(`G{%K3BW@ClHI&BEXvV|3oh
z^K|-kuu{Sa)6CXtY7w4YJw3OGSH+Z-Lwje9>!{u}hO+6Y6l(RQrXddkEBZAgSLNm9
zV@-tUBWH~p#+HlJGD#F}GccT%_I{nVztL3;zpzjjN>wK?khw-jCn_v_%yd`K)#rT7
z*w3F-&Re6f9x53<VPRqGv%Qs@Lk?M$2RkC9>RpAVnnWpCS)>6|Q&Xa84YjqNo;9QG
z2W!7$hK}|athJb?{dD5=^mK(CX--Z~XR=Jz=g)mP+70ff$$zzkU#D4KUjFpylaxnK
zUI*py``qZdy=VC<@?tR<lfnY8I;}gs4+G}|2Ie_KC3P)|+KM-$;R0-Vg=F0LaMt8<
zF|p;3^UA~&3?1w}xCCW5nr$281{|d?O>>~`8_1YuO@4qI9T3oAxedp(nd^Hw-bggx
z_VMG#Z{NPn%*=?^<y~uPY7(#83kV6Ba(Sl7fM#-f`SPXc&U$KGT(xd)U46Yf)IA8P
zj*90Uvd>z3W#K9~!x_3-Uig0grt5s8aA00Hw|Ih2TwEMqC4NoMHOfHY`SZkXgKj!3
zvNw+%@E>VKEY?aTCMLdnU}smnp{~4p`z8~Usa$$ooRBb&)l9<vv>potm!;%9zhG}q
z4}|dc{^qbLkFwC+S9E-Q>buu&z*pgox&F@2N*Wqbo?G?V`?rrn!S_qS$c%dW`3+@1
ztAX2IVmWO-ofypNsK2{0PxB!(G-dyeQD3%HIOjY%zj@SbYhh7cUOwMuQ6s#C(|r6v
zq82Jfr=w$02y+bO%8%a1%WKjh6L2o`yscg$M}C!rs1#EHh2jkV)@6SF1HZcq?IeU0
z+{|ojxp8q?P)<w=RfQ=C#jLHZot*YPiD(@I`Tura{(|0(8HGbCii(>HrHhP8+3*qk
zJUk@{3AMB9^$iVeVH|0rFW7T+Tf+DpHj<;AOB^;A*4EZ&#r(J<Tj8j1))}M6e}5if
zBOVr7D@QV%(|Dxnm1=>}9m}al5Ym;-`;0{+uaSb*VZ~fxJzrwf|3%<n*}v3gksn?n
zF8&QN(`6+vTEEv;fdN@E!p5!{)rl(?9fNb&`lLcmIlg4Kzx93F{??T%SMq`ff7lS_
z2?Xb;cG0UhkiB7Wv=`SrUf~n@)gJ%RkA=yjCT0BYKi7KyQmy}4`L80IfwO+C1$Rj$
zS0T$1ql%3>Uh{uT8#+3BL5D9>Q#P?C*r?037hl2Epe~$^f%bsHJ*E#|1+lQm;M*fX
z*Z<X1uQ<9+GqV;BWR!UI&UPO_O;E|P=U%lh7e&+Br<S*uCp(g#I`LIs^0WQ$=Z25H
zvlAf(xVimCtf8;-US2j!%HCmS4*BwhHA<uEX%wG2*$myW-nrE$_c6x=&z(O{OG6`s
z!5@jZ_=+&@?b~DB@=z-P#(37&*Q;dIPe<nJwLz5>77=lS;*c{iPv~(UK+)uEN=V2>
zb$<W&)Kr||2O|`O9{omyqSX&1CA*x$5140I<ki)UbClxZ<L_h6V)CS9j_TA;Tr%N|
zEu+q%Pr|hZ`DD}0g-ts6KRld1>knJJjly>*+5Ff~L6YnjB1xB#YdLHVSJW#@;$8P3
zOAv=z_W0~g!G}tY=U@#{^SJ-kTK;=?;Q!5cx%NyA7UwqUsANBVwV?-HwMmCHR5DE=
z1K+sjtWnP+E@Yh#A1f)@{;WNj)(yQa*u8{-`?@F+($#hJS!L#R#ENPz`|GhLcq-LM
zrn(Y7dGbU$lII!?AdXlQg4f<i0g1s0Yyu2LGNTZ~!0d*g_}?j@-xS24TJKLioG20e
z`SWKVAD`&My~$z6RjE!bzz&lo_zh6%zkK;3>Pu#IuszvOUr%_E`6&R3XqN-K8rIEW
zm&zs``<awv;e@<Axq+@P&zoNtYU<~2u)B@L`}5mxPXq@uX%4~1C;9)E!0#Ru%j3C_
zw;5_ViX{~cSk<qZCw5v)opWuMwT3^$i}`1&6|W!#zR)BxntXe2wav6_x!%hQfA|A}
z4ra3|ohKu0pt^UPjp;L?WnEjFnjleH_~leo!ajYHfAr`w&AWH+02ji;&d$zavenn$
z|I8EUginp&8V(J!%IRpJH`@%pY?uK2`m^>O^6S+$fy&v;(m|8WO_R~C6xqT-a|wWR
zMn-dM)16smE8jvwLXZ=vBh}=6`Lh;Y?$W~1J7`vOn&yjEofk2^!fmDOk(A^paVaT=
zn>SDHEX3xc#auHpGSn-bot&Kyit%s%D^$w%pbqZptBJ`;<Vul7Tl~B^sKRO2pn39p
zD4WHv@1z>GU!#4EFeU47Ry1E;bS421G9<Vz8DchIT)4hCJUne#vD$i9#Z~m~xdn@u
zLgN8WhmE<LvPnkhylJ)P&}Or37KZ>Y+O7UdOiJ>4P-uc7VN}!4Q7SMUUf9J>Ha%J>
zZxV(TTV9%#U*V=ZKXN@oT7JQas9Sc392n`eMe)&S<@RTmZ=l_qp}9hvolB8T-dMn>
z_A10PymfHpw_nqL{P?ksj*gJf>n0t8g72-7cSUrj+G9^oDkdZ*BIh}~4r9Ubc!1%|
z1#SGhlER|?R89~=T=tDRtb8z992g1Tec3#1X&`%#3_z9rm@n6_w&*M|?T(~sKEbvS
z7VUYLNW+8W(HajCK*(2ogA66|#H^U0+Mk-6>(522Kh7-r>7`5HbCIPwdVbJ;HUkhA
zjrp+Crn+_ia~XmCwaI9g1v73-qg!Z=q7(Wq7l?`9d2(Xnt{6AHJAeK4V?gE?`v9Wm
z9Yz>>iP)EDDn?EBW(_Avh3~9Q)6$e$&xZ-P9Gr34s;RB5WznuL+goV{@L<nz(Ky8^
zLb&fsSQr`63m$$-Qqt=*`T6-@Z~XF4O}&-XJL_Ucv>e%%rY`a11;*@`btxsGNAm=}
z)qTtO7O(U6x8^u_lb|tR4?hj@A-ChD%IYPj^X-VZV?p=6+lfzJRW(gQ#_c2k#ARD-
zNAqg&fi*cJqx5dFpPuI~T}s)XR5M+H<!51GIXLzP;$srx>F{7ToXhlJclgkGrt79h
zO%6;w@Za<2&ok6ZH^K}vJ94M-_yJh&?AQ~%^Yb&w;|0J@=7^qF{odxDa$KL9_S!{X
zfXHsrw6V!(0yuaZZt`Bg^ay!}Tkq`Y==w&NVO|`)GVO1(>trh-_y#ihJDsv(FpL1g
z*2vma%jJ>~7vG~+@b>X3n+Q@hDV`u9+TIv271$mRv>CFSG<Dfkg)z;z5MvEk49ePc
zkL9swq9il@_&TY|6NXHN!%AN&juHg&(rO+pFI=(p(_5E7Ke%^Gqrs2zuJxP}0AbUI
zfZlY!hvZnzD$EPn0zQT(+T)w_?}}c0CH?SWN_u*FW~ND=n(Bv$2nrvqg9qZ`E0x(v
z1qBM^<9<*owzCv@cr1QN0yDwsWqQiW$_j1lG?l^(T^RDl)6=tx>!MQu-IDLIRsgEE
zNz^wUFo~P`OHB{SK8s6D{brDFE}6m_UZdq|I^G!IveOxDZf<@se7NgU<9^z3Z7R2o
zf3{6Dvt%X}CNrdfz>P6kd%yUl9p;r+g(SvJc+`6;(#BgD>39rV=VrKPMcrP9m#WJy
zol0nnE-mKHpbqOLQ~$ifi>8%6g<@!zHFOqp=t?FBtt46e$)#zIC6?!`p{EMqnNvNo
zE%tlM4O~WjBi5zK(vc;Q$L@OP`FH1XAplnOXz#n9x&!qLf?;ERfXi&8%Kc3F_OEcX
zOPIzKZ91KPY)bZ)JL%<l_P(Ubgz<@qcMn8GMVY;zDQP!<{~p0(mFhV+T<Ox6{fteo
zb#ILNun3?sjA5;zY}Z*?zqlM843`G%E|kg$&sp!b$KI!gv9EB*es<C2a5q&xt=^YB
zI5LufM!Vjpc&VDeX(C^fgPxuqk`%MIw+Cdw@fV9{@o0Sn&yaH(!F&!$XU<JgVWFFr
znhF)iFt=aSkK&S#77A6hq~W++_w(mZ2GxR(pFY9Y?}fa#ZW*+MaV(atwjS*7st#43
z<{D(yYmLy0cIp{5V=v#`gqFHIDR5Xkn)MLvbR9@B>-y0ji4xRy<3yN^6Fe-{cxzR!
zaM;+`V4x0S(#+f+A2s9tlAAj_J1gdMF;=}?ZS&~ZadB~+rbB!1QFenxFQG$4wdl?c
zmG72s4wV}AW~!Q&cenZK=;<xO?=njl$arndV6anD8-#x1xZ1*nBq+3Ff%E6jlao{|
z*ONy(?@UMWIly9D!+WNEZ=NNT95qY4D1d>bfht3<nAx@9)!*I}cwMh8%IVklP-sy9
zIOp(f4qBYf3Xe6-dHOjsvq!oW&?5Vf7wp8u#GK(R4>ScFNr{t^t0S_K&z(DG#Hl($
zu4cJbTE%sayA)%>8E(~2TYh_HqF3r#tH@>){&|Q^g*5rnKv$NF`N^njkV_;1nd+sC
zss(c;vzgw0e#l((<jIfPcjo~wY>V0{4#UErrd-jTc$Q*Sc$N-dfBzr;>{yAik06e_
z&U+#Rr^guoGt_=|%2zJW+skF~T7^s$Z}GDl_pYw4(uETGDl!N?*R%NSf^9(K?BTUf
zOiV!J@gDB2o<zt>{a&;_e#6%{Cqxq9VVsHppx457$LxWS>eX~{ig$~=Yh&_^xNtyS
zS@|Z7da1R11|hr#kNz7LdcyouP!Oj|pXI|`7%us{XW$9f>^O=TWbSi0a2$@7;amUc
z$*%Upkwanj{gwMBWy_*}e^r~n;a+<B?crz@SCgV+@LuL@!K~_KZ_PcSU~BjhH7^u{
zWe;GXBo8?)eNybx?fhbX6fkwZ2a=_@54eQ&R3hXl2cBb7aH2zSvvF~84=x-Gy+wH0
zx4D8imiO*`^><#Q^{9!JFofT^GL!1NI<GWc&tG$FW+>d7(-WBO&rR+h6t&?(P6DkF
zgSF$$RB<){x|+#g(PRG&<dX;#TYJv-m{77YzRi!v#quR_C?BfvIOE;e=DficiAGNF
zZN95$jG=)6GNB|SCCzkEq4a6J{hUc>=-^`$;YnjoRfH?ro7Fu0k-`UW5mX{z7aVrp
zou~N%bzo-3ATkU&hqqaRSs^((O&{*PIX(~NqWbO0-Oa_`L@DGgefVp)D4<A(07nKi
zU!m!+^Ku>O8EXE$*944HKN7tW2;r!ppy0}Zkm*-0=(|>ox|2H|+LU6DG=U)>Eo65z
z@}lzvaq_IsU4fUi-#bPY16U<Z;x7E;YW-@!1i7X^DDBsR$!2g&$rML;2~Ik)=6JD!
zVlfUkf5f-I5Bftt>%67tI$z@wye?Omy=6b~Nl}un6L4;7o<h+x{1fn+n9J=qDS$fQ
z>Aep@c;XDzVD_{7_wK#wH*)?Jss}KE<f#?%QD3_iH6tS<`5d>Fo@RBIc==puM>%y$
zapkLxyt^J10vG}*hMI>vy)Rz8fUK1{y3P?bV6r%YX0hM{vV)YExO>g7gCVQi%gcvh
zN|u|3mNwdHqaP}*SL6LNd~%wa)N@E`@T5b9=ZFX^Gr{l#Hs!%h7#tjwU@X2$M^~?#
z`&?J|U`DYrH9eiTS0xOdDP64CMOL{hr4xdWlu2XGx>QwTg8Fc)L5d4wIqF6#Dkio$
z*T32T^at>2Z;DG}J4h^IU}-+ylocKaSPD1N^bK<8k-+F^0sFNnr`?VHkrM(iJX949
zJD1un{{(W8{Zy?lplEvFVblO~LfHZa66rc=n)@^G;V*0YFpW58c+K|Ldk@#MTrxg?
zp6X6heExi5`0!v8K45S3j0;|?iJK&Oq_Ay=2<eN%X@KbiU3jp>EooU5VNp@eOCw*)
z%E|zc8Vy(Q(a`(^&KFZ`p_e<cy1MGJ-K^==IMba5?S%jzpOA{z7NTr&e4KZ^OSUC*
z1!jSYwI3249UU&a1H)e4-kXcVnhFX9j@#x$O|7kjLz!;E<hWBXp?EzoGBSd4GLR&S
z++GNa606yrjrp|wJLcn!kYrwsP-!qm+uN&?4!lvg1SFE7taVyBo)3mAoHU)*<%#->
zUTWv2K{u_dtCNc4S%<NvH%KX?$AilLGYl+lZf;Q1yPcASZnOvD?T?xllQ3!ArXf``
zyGJeuk>~1u{0yA_(ulhRD=VvT?Swf8!`!lMfeFTdGkg(RiF>ohX>#Pu{Ts&C?Zczo
zxuL%Tzs#;;JJXc{7}PlY)uPB29=l-ltg2`I>^cbvNyU1PDxBOWPc?`6d@Kbdz~zBp
zH_WRw;-6%HJJ#TTH?nmB$nY>bL)K6#`>EI<`W5?fn6flsC=H+$gB1-43<QBk1v8gT
ztnV7dYxgDk8ss!Dq%|rHlFv1DYs>bKfEw~n7)Z5Lm1bzQxw*OC-ro5Ffq{Y0Q9<W{
zUUsQl%B{E{1P72am-F5hrqm`mAtCSHQ8wkncq1bW6T!zG`;C4*YNxf1l9G~Qi%9?$
z9@nm2gH(i^6qlAJlxDksvz=&4Hd)%qG!!O~!X-9{QdG4VG*ku>iD%EA0k%_9Q>*cS
zTvbp}*@0<?lakm8f?jx>3E~B23^GPj>o6+^sIbx!5}k>Z@P51#oZ-e>OQT5X`Syuk
zuO)0kwE~m{h*lfmL;LUG!dcL;T28gQ6F?DjgaPv@SJ6BRGt3jNhs&kI`_jTN$pA{V
z&+4V8L7)zx1bP}9jFn<;+8RWiaGnI_1>^>&?)Qtv>!2TS>RX3*FbI}n7C=rJayi`Z
zRku-(=65QAwq(%t8Q|68daowecsrC}*T?Es-SW!H%7h+cK&=A0{r3I)TxW7y<-vLu
z^hLer&pArJ#ff+e>$FDP2@MVH)o?I|BG}+_DKH@5W&iWwca2uP>V?+x=z>AFm><^#
zO<{EW2@Pa2%0a_<dm<tt0+JYpB`4@PSrt1oFp8cyb#ZCPVR8BiQb^HNd)uoZ>d;Ar
zL53#7`k$ua-Lk3NRW~s)k&~0loiHt#P6GD#Y1ik+Fo7WP10YKi;tqZ}5~>$<iN|tk
zd91;gJ-8;pH|>Ph^p8Y;{+-VohXBBxAt-0Mn+@_wE?hv66S_NBWE^8N+`VA;hD1(L
zabVt>=%J*fkEbV|m4&73<Hyf(`?a*S5n>D=L|1qBNj~BYN4`(*-#^XmC*#ipF5-&8
zM?$E3{YEBvVzK$3740vl6}HRHs~pko#+GCXrMg{FJfweVn6ILQZDCG9sy5As4<A+r
zh7UF>b{7n{I3Fu27Mf3dgE0fh6iB*L<ecfMg(g@%NTquFC`OiB!*&!mxcPZ*|G`dg
zWh)TY9l>*;bcj*e>8$>0;SA3J5OMeJ-Krz_(GF0@1wxx<r8D|qEj2BZM==!GpeLnV
z_R`YQuH8sT$3nU>-oAYy0G{|Vp#^PxlNez=(ZJ;8MNeKfsKXX`cGTlC06#$K5*Y{b
zN3Xdq51Y?^4KJGqUVAwY&<J|Ib=Y>ymxOf<U}=s)_bsAJeD;=qKga<vHGvTkb7tIh
zk|9eF6b8M_8$iFn3`~t+A&(j&BBJH?`)6>KkW@LV!ZY{)k=3|{9dkDh0gw?em1Qet
zY0#L@_2umEZYnA$7@T2HFIfu^qdJ+Lxsmwg%air#bz5HaE#2IH9ah|A^(!OGw!*bW
zJ@Y-wS0>l(^CQ0?%CL77FO@$~Xnlr@Y7Jd%b=Q!4Se>2xE5V9T$7*OF?OiaP0@Wss
zqg_>g>NR;vj51Fd&vogDBH&E!Sp1^kwao^WmJ4|lK6~{YQe<w9(3ZR!!loOg{)cFd
zu%@SbN6nOO2Qk}mb-6W9tl;XuS^wB?#Hl0$k~L8dNSPJ$d9!{%5Yq+TOt-#LrD{&L
zj%k~=43ngz#bo6f+kbnY8Ta(Tj-%v|B{Y2)_+S{whbn*HgOHNP@_{gPjsfZ6?Cn3j
z0DynsHm*cDEU^5T%sEtfDKHK=YToxI4-fjOX%Q^&@zQ&FB+W3A_GTxOVXTYPhuyeO
znNCoSU{am9JpF(yOWVhP|H^d2I6Dga<5Z9#14%QIiPDj&y~iP5guEYK&<}d)Oosp>
zv@1UXd^41Db)h%FtYTcW=T2@?G+!{FkfD+&B>}HNq3=>AnNdKjK&KM0U5fF9GF3Tx
zAKDiTWIL<Thh_HbGf<mAOJZ9(3R?uB-=!ebM1r)3@#!QR44IH8+6{hyTJ#_uR$+SZ
zu{7a82p|!$N`mL}<v;E#-BmfOAE91so){E0*%EF8L<=IqjK){)&zmBMu=c!+XCqR{
zKG7>4|0Bj;9iIZE<k?vCax4mJvTl~VUXBuwTAwj5o<lzU?qzQk0D2!D?9;Hm_>mCg
z{leyP2R?H7^NKN8&_7g!ft^=nB%!8eqPZ>;b?Waa0TTjo0DQ?;lcawSyy_9A6V%5g
zi^(eAP$5z-w_k@+)h3?=fW?<<=xhMA!hbk3Xz7hwIWQ+1K*YoF`>YciS|Wvj^BzE;
z__#P=GKv5ENPEgD1gX)}8-m&lu!<TM+!YH4n*AWaj>g7DSUr%P7n_+XyTtin;t5tW
ze^_-I$#kKI9(Re59tV|Xb+V1=_U-R&$ir0b@|`<&Uti+_$LTU+byO@8@4J!&03|pG
zFrBG<?TNH>E8sMTxjvTQ=58Qtpu(PprB<xmkvBL0!btr4O){<|M8;mZoX!0c=651n
zL69itL<Xm~Co$M-helln#v7S$;b%by(kXdv_(ywyUHK4lqh(xn0yW6h%4~Cf)8888
zU6f<B@CtoC{{}XznJ%accNdCiyNqAZ9!WQE0|3wQ+`a1nT{hOF@eUEff%`N~n8ODG
z|MG~dRujh2hYueDp#l&&L_?Lp@vaoh3DA40xB=*BvxWkGqjs3f0UD{_sDO&kUNVYz
z8$hjCXEmV7%EjWzs8={AKrYmThIWQ`sS2`#?9<JgH$wqD#@u`Jb8HL(n+M>WFFBWe
zxyF14B&~d@r7*OFCLJ$-|IG|rh)t;2<6iWcp5ETG&|m=&10^wF)(osZaQjG2gbr)b
z#<yA9&>*n-o$Y8QO0tNlEKHe1t1utUy*|2d!Z0{G!c4ECf`*{rG8yECF%9aG`*VdX
z4TKcU0*)@+v|@U&coi7y?H>|M>gG+?q8(S>Wqk*{$yhx80h;r|$k#WR6022PUSya8
z^u-je0X<b<e_96d-?RXn`S$T9dIjiKBXD9g2Qv+qJJ<v>s)G@rH(i+x5DVz=Tn4F~
z(-d$L=zK$vt6J|amg(9wL?`W3=2!i`5xUJm46yQdVV;3yyG2h=%BVKAZ2PM<5~vQw
zj4GfUT@F^G?^5`Qc$2_oQQk36C@b4TxIU+yHCQE>Cl!n!#39kjpv=I%0ywLrrnUzF
z%&W25jbL_WMtGdGa;pyTl?Y5OzhHuL0<<oYkeRt~U0}NjmI)YfS$TOYAeBKh;|CIP
zeSIBD+&&zJoYSiR3jmAf#l^({2LU?hcRDTSrZ}5OksbwD{6mQOrlx63X`yFrT^-+Y
z?fJ^R2_}(on8(`jdErOt<ke(YxG{eLDq`YmG!SbDk%ufa<RCELhAKHs6<b2Wx@}RB
z@88ET6oY%gv~<28EwdywJ9|kjA1o;2z=hZhJK2DaZE2b3*=JOMi~`a8m{GGm2tre(
zEf~3=;Fbe#+?LNF5I!^6V|fpeE1AQlJCk>5vUH)I(tH3=4J~jSX3uj0qS|?MfzzWN
z)Fc5=34rkqT<S?zmW6&rzX4YcFtl6~6p~b@*`5rVIFSwaypvSaE_*L)+<@o<Sfx*}
z`%_PLe=v-(2$^ABJ5G_YfA07IOp`TKf50pt4gH=#6w=nK37CxJ=Ag@yKuz=0cj4dY
z(*}BWU}d5i2t{O4gMp#DSIt5SXsYIX!$5Wo_?g47hh0rWU7dkO%%94*=n*-c0I5gK
zC>*svOykH<CfbRJL<9(30uTu%(AIft7?c3mw`t|D-|EaaR3y6NyjQd_2}00L)AeXx
zxV;eVO)$LycwnG`Yk`rBicMECS01JkFn@SAlAD&j%)|y4S*f*$XP#r~FvRAbQL2Dy
zCxI+dj(m~;JWdPEmzcto4TdF_@D|{PO$tjNUt-wh{60)8>eC098y5a>CAgAzrI8ky
zV`tK`fUYY!Be0j<p9*XT^iWB<z?#ves5eY}-k$w~PblOKW;RWvn1Is4Bgl-%$WUEQ
zamG=#&8w);2rneH`(@PG;jOSlvAwZNz|k<c8xe?$C1`V0g@BfHmjHMV36P;+2pAl`
z0yy)h7H|e=35B2iHmEZEMAHL>S)SltfEmm%?dtXGGGXk{#2zB+7ilwukm!{VkqC~@
z&EK3K*aWP;1*vqLa4))W3MM6(iO0vssrN=sI4?ZV(Mb@HGS-IE1$m<d)UYHe?hQNW
z$s*%$u|1+LS;fD|h?K$66;B+@ph`y*FY0?P04TOK;D_1RI!y|9@j(xQp63mw9B##V
zft?P2VV&ilb-qCY2U}zCa{En;&8LmJ=%fM`Y<Aa;7K`x#l7mn`svHUdgFDW0_=H^c
zH@Z%raJ>w*qZE{`{OzwGj{vpS1(78J{ZE2ETfc)QyrsaPJN0n4^01RJQ4Mk(rXGkd
zMJUmG+ifn4G(eCobSC@Pj1~cV3WRVT@fk{<wLy!vca7j0A?r~B%OHp)`!0}1P}?DP
zkQot}T;bYz>q-z0n97!fBqhJY_?4E&bikhP`$vIW8E-;FPkVcNuz~?F$RE%`E=^J0
z>XX7{TH3M1U=1%HpS^igm+<%R&A#H2LI>Mh_IHt#kWkIlm4YQ!Fany7F7O46D&X-!
zCU+P*nZQ#5{nd|*W9HQfs`hUHE62xmf}1%&zXBuIy9dCZhH)4IOo9o)4TcqfrO?v_
z*4ppu=5Due4et$k+p<Ic4xB^u_$qzf)(Au)gce}W8wIZ%280PN7Uews$2F{w6>Tjo
z<sfPT!gPR3g5eJd?eERahTM6)Advv7kpglRxP}IQY8!3}0pM5^GgQ{Uo}dO3!j@Nt
zQcfH^2?V6a8S0)OABal`Qa0W1MdKeNplQg;$|BS{T#__=3~}k0jO@UNrrCA(1TFS7
zz*OP7A45X=maO`R<mKd`{Pp$qJ>nAP0<{^<AUS=rJa6Y}^>qdYUq0JGs=@{a8U^oe
z9Zn9%&sKWtZcPv0{^674q%6F2%iBIt;*B<#X}(_EO@um?r;-77s#p_G6Cbqf<dG31
zy<E|rzK0A>57HO}4z}T^Y?cjwSX!F<u-awQPJF;Z=`;Od7GcId9rNa=xHan3bW*rU
z38)YG18TGnjw)3B^2nPr?j<Md8!jLi=T9$575t=6@XQ?R1xj$_pL$L#5U3JX2s7m8
za96Y1e`cr&!8sjc2ZaK0uAoLeQKv6PJR$VU9%v!)rl+B?oyS7OlxzPc7+rxO$1`jh
z!Wo=150#|zulrH$HHbrjdj0l@I{)*up1!|HVOqJr0W?&YlRg9<5GI@ZFA4{GGu6TE
zrz$Vsoh%a#)GOTW{vYFdj|7fvl<_PNT=}hLJ_^){OGxB?`GSv+FMJhLLxH0QSjm8m
z1s7c^Xl|%>tV_29nKa6Zpr-p$LGZ8byEJ{c2BCp-{FLonPQw#fS#K{d1b;)>%uvZU
z07c@@cj^%4k;FKxO%Zj!ZPNMije^_kDP#lOKS1XaQc}bLV9R1qD{9xu0p6}WAwf6=
zWMLE!9mqaV-B(stAibb3k$BAy7Rw5|6H=nh$V63CQ~=IHxeE>tcK~sCNAr}v_8{q5
z;KQH2C7SyAQ!A@?0QyXjHzmKLk}!0?ANs`w!VX=BN4i9a4>NW~%P`O*XqX!ss9k^v
ziU22KKmZHW)}eBTT`y2dKo6#m|E&jq^xCc&HM1?(NOXExib1;oh*>;5KC*7F&KQg_
zIRoPD`PWhA#psxGAn-2Rsu@2C*5FMFGT~jgdz??hze(F5kR<<PmZxXK*o#=t<jA%7
zjn)x|Zh+E3n`us9Jv1pCc(CYP5tW*ibw)!5M$FuPJZ(qjTVuzjfT$fDR7>j)d;k8C
z67nd59u0%6R0#7udIecoS{f;)c%naKW-uKBi<Uta;{+Nq;@AWwAFvTPrVr5R$wtz8
z19Icmr31aeT!sC=WLRjQv-<YVehRufv&5CrmtHECrNEKA4lYmfgBJ<uDWjMyVyJ*8
z+wt`3PqxE7gI2f%Iv%ttDXEs7_1R&E#d4rK_v5@7XOKRnzo-cus;`3oyLUuHpi%tI
zg(!;G4pR)mv37QLI;}DSDFacd@%S-t*nWWHzlU?l2m|m2em2}$yXW{Pr~9qWB&!}l
z#h<t-2<LlqK7YPJiz!UZ%xu-l=`(Wny-+|Poh-YX*{#d>+#padtD<7ah_f)OSED&&
z-M)+}OHBB?Zth|j@9~kB4$!ol(hsSW(R<v!zP_IxQOFv<v8n7~`C*81*LjPI{IA`d
zqLswoTwM4hrKJ^a{qj$$=^JjIw>HG$0p^t^E-^H6-WwdHDj6BJLe)WwT>YH0H-qyT
z{1kxS0U`$p0+oZv-m!K4PPZ013+A7?pKW=mSe)7_3jK;(95fyDmMCQrJ?>>ELNk!u
zuh_kmj_7MqPr<}PM+5K`Ej*r(aCN>-wC)ykt^~li=xApW=-z@2T?;st#>Vr&%Y%bu
zKPee*W02Pad;5}&Qe*h)2{wr2$2nzXyhO{s&vuo6Ta8Q4w8IU8xycqZ4WRGRM9&xx
zSEviKva#h00B=l(I2#OXX`VvafIKkf9B2%n{e&pKG1>;3SZEzqUuppj;UGr!)YIp@
zJuxPlz~`T!c05Z>?b1MLqfc;xkRl{BR2gn8Fwpb2)CoOk@7^t~sBo_JBtk3S{?o-A
zB~8o|jBp`<>4Cpf&whH((b3V;((>WMcfGS<+KJ{c4B|{yW;5(b2ZOc?2t6RfNz2Fx
zn2&?Gdafs9An!#dnMr%wZ=vxvm{#zu$gyGF5iJI26c`MGf`fazyVn;myWi-uBG~k#
z<8L-YO#_gUnvnrMojZ0bI_BmpFd9)g%(2YV-At`}{dL-sY=zYt+Rsx`bI3;1v8WY_
z?Sj(=9hub~6&eoFA*y6}OZMgHljlTBeIgrR+<g6#m{8VSHTo9ct(%X@C}bg==4Xlf
zSdXOC)7uMr2oZDf*)bIh;8zj*$*}5epsJzKcBLKFx-e@2WPC;GOZ-pGi%%nF0m5PT
z&ZusLIBa$nUPGNNz-t!_H=i&%ZFL@^LolQoncV-)tsEoLFw9ft?NE*UpjYp8M+vsG
zU?U3T|N7&|Q&1T?yL;M1%c3I1-yMl?!J#(#h8jNVe_+ityL1=zZ1kao&gXC#Ry6<G
zawSKBefU7Hm*yEVJ8QhYPV>{(#}eWzT^~fvA_g2+i>JqsHT@qP=dThp*OPl<$(Y)E
zpNcDnG8|+PPNV9NP7Zxnu(Ycr+8=CAICvKUhMg6cQK1p3r6n@)RuxVsNoz7%(W`Sa
z5i?Ugj+#UO!{6b<KeeJNj^f3k(JEQJX?fyKPURca!17!2ylMa34j~408_-g$$pg0U
zhfb}ZiZF-8{$9-!tf6})T7sp}e4I*q>Uqx*eQdGkJKXkf&|V=+POhcbShgRGdQ^Cy
zD`P5UM}e}8=BIwmCJl%|=Ec>km@F$YXX*SBf(k`BxsP>PCx6txg310(<FsY!aee<u
z0qQGo*n02hscoVIp8E0{YxCY&&8KMPE}^wgx3X;ylScN=u8VV5V5>IJYXtU=6XBHa
zbvdMVGj;87QbgfRLS0=hpDtF*94S{678O1IqS=)Zul*jF;}c56t*r7fCZqvdTU(x<
zF{c{c5U*>CuD7r6X{-BTqp($?P6dvRZuf4{WefLZBmRbXY=YjGZKMaqlu_FK28v)(
zZA&94BtXkhi;~3|ipYqBs#q*}Jy0_@pH(n&`}R#dt(%OLv{#qoxT=(tl)8E(GUfvl
z^oZ*O-sv&NC4~7i{-;}=q-oyXD>5$t><9Jk{(l!h*NJG#%os0&AY^egO#LY>|8=+2
z$W<XI)IUlBKV@jwPY<0#&0nDXw>DK31;F02;jB-EnmGd#VY8N|r(D~$e#bWwoP2yQ
z-kn179*^<JGWy@Tpw++^+Rm82^MQ!~L+GT69TWfTZ=du!aa6sOX3jdd?x&Y0-ZKYS
zDa)__49EWKy<sK^*eJoXp#uHCzVxf)#94jk#*pI(lK%Ir<#+~k7KQY>VkVeyKK}z&
zw}U(b8#M@cME`mPp;AB2TQF~hv)6x%!tj=2DnQTz6qTBq%I~xTtcN~KL?qtVy5_M_
z$5O3=_0^`Vr-nWR<=c<EQB}tJD(Y$?3Pb_84nWcOL^vq0(ZOS4q9yV!7wkFc%0OE1
z4VDunmn4F2-@I`H(YwK(de8$s4HzBZZu_7IfyYS*gY52ldl9Q7=U?t8dRy`eUW<1v
z$&9_gVp5t2p=@gU$>2;tEg6P=F(xJpz%pMf>B0K;G0CBXhCw&5-)IzM4!?Yp6?V2{
ztixrN{=UDOBB2emKV!khC_R20;He+TMfM{og8OClp%n0`3S2d_>zt(}4g<p~<t+sl
z8Y>E!j0=)8N=;Cxt%vKFIkjln64(oM56v1wfNiM_<%s6TngKZ=L7FG3s<Vel-<q3M
z_kTSBv@mBLE{-GAsI1WQIfcqUxuSw!ZvG^dN4#&Qkvw&`XeDWJxk+IIhSMtlTJXTN
zdNjikQU*i(zMHWOgFeQzpdh4p!yvdh(Z*tJ@;1HfEo5@PoG*-!^T0*EE_Avqcx<XZ
z?>#9cB}U&4RKC1tDm<i<^6^3gsKfmDqyhqew-PLrO2Z}C-q217D-MMQ7!Y9ifuPFK
zOPm#i9fzO&KuR?+e$a&IJ15e?@(2e0gu5Mp%nAn>*5vwDDG?vM5ofD1_3?$9ZzyFM
zgPV0&J7iOR*2rA15XF#7{QrD}bB9UbqLMSr%FHYnn5Q{MPHvc>nFV+aj8$M3_)H!b
z4rb7KRT4hpUt}8}TXYG2Ym*@6#7Z%+9oLkptF675f{>{h3iN!q7kysVYp;Fy`0*-&
z0DXg?QhPoC?n$Vf4hf~xa3f&r6U`Tx&PAJ7S5}0t($dl%Glqduu<x=U7uZIaiTZ&?
zt69bUi&3s531Z>$8eY5a>3`gCgI3o1N025OMh>;cRbk|6weEKyB;Y0>VmFBYfZY8l
z?)<dmDYcsAgk58gxDZTmr5?|c?T=kDKsPY*@a({Z0Gufp938<*{webrVkfw-b*#h=
zco;<a{j3Tn;+p**+;Nf~Yl1rPIC9k^qt2;f&H4damY>qw8V<(?IL0Vt;YZ))jb>Yi
zV|lL;lA;7Tb-~iF|GGv0FE>AcnZGI<+>St?TUc1Y`h!xyrr#a|q&3(@DSJ;R;BOsU
zVFy*h5=J<1P=nq(2l|}7{Z7Kez#Z_NWPks2FQua9@4b{(eI9<X-Fy=+q-K)SaaJ;8
z2x5km6M6+EX2g9^%|d((@j`$$ROYz7!l+)dRg51kJ2e8G_wj#cjqiPq5lnTJxhyjV
zuE+v#;X?}0fPqoN&Tbnh)CwqNa5DR#1FR9>wVIK8WE1-aFClqv4%_dh`mx0qc`Api
zuj*J5OTkSG(8a}$KN*E>`RHGo-DQKVa8j`OGn~o5$d_w5{iJXJA@^jW`T4G0tDCnr
zh3#rk>1=oB3t^zxdPBV@NT1@=_a_Iz30CJz3Hz*3f*MQC)>sW68CCY#D}V-VVLSwp
zl9WNk5}-EFZxX82EQ#@GG#a!GsNeJ%b(0sWDD?L!{}l^yb?N8?U-A0hxpPN4ITx&~
zu&Ye`;X@)K?>bhHGl4XMI5O$~Qm6Q2H+vCe54rOo0ymcGpT(*e*`GdIpcQt``AC>t
zz3W!chFKt(@-0>2RD7V2EkW33WI&S)1s^|MBJp+v_FEL?Dd04JZxr2ocZ-1`1CFYZ
zg{ahjI~$OV&jmb=7v%{1&gHmeHXKg_`Pkgt9Qfe_xKJ%kT^fiU$ZYrH!XeuCUm%{e
zXU|3vQ{TK{Dg@8uKkic%*W}&$eOO@lNCK|hcV}10WayweZT3e{#5YgCh6X;+8p#>_
zJw4A78EfTvULYlPDAzb_YrdC@J$+N4)Ky!Ybcf#;JlMO<ng_^!Rm05U`oj0x;!MEC
zDeY>Cvwc9W?oRvHM{;skY0yjxr(Z}&M{pBumAsO}iWZEj2dflU>E4k`5!$o9i$NvF
z^=sFR2aA?Lf`JXO(m*yeHiodjcmm83&d?h_n)gvFxdhxvma&VSMdlOYpd>#3_R*c-
zPLhyc+f6|{IMfj@QrX^6SNHCLv9a;U?C~eKj+QQDP%k-625-<Hri23F-E3e<;gPSB
z0o!Q6dIb--2N6)--kv;<J?KXG_d|_!7D2@Yz7x?oApTAVTS&T5M9aFKjd>Fbe7`X4
zz=C!9&3yd6IBLpha78R596jh8RQ{8QoB-5w{sh{?3H-k`X9Re`Ih)q4A}gB<%IoHk
zR&m!H_!RSzzYE}i_<Aa*;=J=_(Q(gTt$mDEuVP6gpp32ul?7OMkWJ$|%>7c{k6GZK
z#kOPyp%E+<J>A_LDw$5BCU&Wz(MMOkW={rU!76Ni28KXXL_`YAL!d`jcp$qjV5?D5
za`LWogO9IoB)^lrh)5l{>5(2{I>gIoT{rT#n{3bL3z8ZHAeh(F!lHDmScuIXrLn3X
zzWo{#$pD%U1}L_xkS>1&tAH=ae{H-VeYWw+g2AeW`d|qR&qg6pBG9n3Vb_L~9+p8J
zIAG5`*uCo~X#cH^aU|r2n}*C-Pll?<IAo0R&e;aJ#^pP>c(m<b-~S&C5nGe%{jHJn
zBb|sG3%(9;ekUd;3v7+NfvqyJhw{mZi7(_KSTogj%U>_QHtjk}i-{M!PCJ$w8XDlt
z?<C;PbJb35S^DD;QBjOEu%qQg9xv?P7AA{43+_IoP(d94kpebb+bS#fLk<PsR{mjR
z5BqyCWp=B(i<OJ(_T+r92`QDsRmR>WuOxYI5h(yuE9{8qppj(*n*nq!#$;~zrI=Ms
zUbuX0>gh$$u07|_`DvpM&%r`LC&=$N0;TPg<DVrNN(7_{3n;T5)Xqw!8y2$v6?KDy
z`G*D}6r8*a+YHVJAl+d4K85V4>+hUk27^C{3LrU6GzS}Ub(wKf^!N9}zy<1;r?2n&
zC&$D6O)k@+xl+T(!`U58T$QsAWSo}1o}ht(C){?&)AJv9@>&x8WhZaNH1YBFbcM~2
zhe(im3;NxZf~5kX2+Vfib_+9!)=r&=JIUVMwdTQmY+<}95Rq<Ss|7ru`GLq7A*}~w
z5=hy_?#WbO{O+H&l!lG@Aa<qB?kJyPFjSW*#w<Jxq<e9*SQ+FCC!nzMJ^(>v2a=P>
z)xT~9K0Mr5U`%LYavj^Uvt$=-Nk2xwkui8m4^KR%+f7;Mq)L~Yx^U9D=U2<sl*FwP
zLx^KVV@ZC<D{!e8T=l&8nCnubxJV?^TaqMYgV@mRJ*Ik1cQkq#muufSv;NNHok_c>
zZ6g6cy*7){YPY4OB{zZEYg#?MAID6wdb`R(nv<i?#uXHBGBLI3vxC@eQGZaJL{bf1
z!o%Gi6u)GFp8<<EpNj5zn%x7J?-c=oLqP3N-Ex&~gD8vvRumQ_eX}wm*m#ol7+uQ(
z`^Nfu|HHO|K@0|Jy^xR)pha-@w70+i`jnZ>7pfO1Tv5qylt^^<Pb_Y!R&j7J{jw#?
zSpD|(>sJ@UDvpU$TT$2{M_XmDW*<kWk@vqTiL1Z)kCVT-C+s7<^S^qE@gZ0ly*@-n
zmiOJoMk!gddza&&p32zR*vQEEi^iWkb4HVJ3>y_e&DM^JP=U=!!osz%uPvyl2!)yy
zLnLq1|8WkkfzWy6FI)pT5U*o+IO~R2+zoJ>cB)p`F4J&w_6!WDCU=HPCV@E6+S+<)
z7fgqf;MM?RbZ>@ga_{W@`}aYhQdLpWX~@XP-~;F)C)WjpHIOD|W@b=&K_I&hbKU$F
z7OK^&)C^_zEHzbEM@OyH+W1QXsCMwk%gf8mWTc^~E_ewa2NHpe&8E&f;O4_yU=tFg
z4m1HjpDpm}z!tC{Kb{O!!>%@Xz{r8P+vMbmN=w`3vr#B%smJnGIh$ChgBkENfIAb4
zkCT&A!GwMJ0f=v}UcG{WLx(jI_ITaCJvZBnAO%4FLU?b!wxiKJ{S{eROgC;!4HTLR
zdA@!7c4|srMy73NX9xDXeEoXwiZJ*YK<hZT0}ji~{Ll`4*rO=?^l3~gu$&oSN9`E}
ziN)$*ZyT4F7Ue3EjO}{9)N%XE*=o=N#kXNbO{dlZa+QjTsvrZrm+$jaOG_QTFf%Yn
z;l+07!-Qu{4mO74xVR^;Sn{YVD)tIAtZ#0T3c*f9D}DXO0!K45=F67_@rbCY1%k(L
zQQQJOf;eW+;A~(7jylz;0N5b6Lf_Q1uwrUv2Iiu3LJ=V2g6p%u7Y7@AVW^z`?Zr!%
ziftC7_D7)$a<H(B&dxgQti34LFjbux9VG%623+-2NHY-@gppwkR#7(OKivMCZsc4h
zc@>WiHRC`QUd_p2xl#*m;3S)Kl$6HI7XkQbR5}~GaAzp&LMf{3mR6p_LMf#Jd!ZK6
zp&y|!^^0@e#Srj;5FRx(wWj9qrQMT+geHcDyGx^Qx~5W7op;t!z@!X~52nb}RB;0f
zl&d!la(VE#N&MI$U4=ZOqNqe3&#{wVyl9qCSYKaXU0n@Eh5Uj7KOZ0NFDFq?Z^9p<
zF?&Wh#GF}J*c+e=+kD`!nUMLGpQ(p?tp~3B?Nu?n9N0*joz48_>)8tz@<8E%z<@t7
z!p(XXRh0<J^lc|FHrwxR80+Ya!=ABUKV*RWhd+Gcb{77|05n~Y?o=|YzZ9gW8)e?b
zN0}k};3cBhMXe|)DN|Ehi@a)h;?r+@TU+le0QdOix9!uGFwGAQ4apibG&D>adO#ld
z4D?|s?Nry)xScI8F9)?ZTw`E(xKg);%vaQ6eMWwFUnyD(**rQutt{US0ceNTM7he+
z!qs8*T>7c(?e7mJ_tLX8ckXOMiz$%T*4Bp2wly_1UU65iUF)ajVrG7X*D*A7Fwsmp
zumar-Qq0}my?F5!s%jf9lAB{zW~ajCPy=90U*8tww6bN~4RL4Zintp%I5=>Pu)7x^
z8uUxp-5VPl3o4zEC-gMf@YP4n#mT9F$HK!yMoGCp)j^xvucoM|_~_A7)=<zga{(q^
z5tfjUu(Dc*dhJNThpM{Cu<+(AR#nAKZ${7=(<ge8&%j@wks7Y$Ie+`@=d`q3hOCqn
z5DY>OSf+mc3JMDPHBtm4EZfHWsHg*IU`-$1oH)Z+)r3NE<LbiM=Yz>)ZqCTX#U(>w
z7H;vx1jlS9%+IT^{|t6#!In!qTidDio$TyvaC3yW{CfQw`$R*z%b_#0N>E$&_xAEs
zj0FV+VRsl*9AJB(DL^BHX+NcNnl9QM52a*(Vf|1H1G7GA>pC10zW@7$8z$g?qul>j
c?>fYyPI&(OTc8vLoD}Mzh_rB~(9_rdA6QdB&;S4c

literal 19561
zcmd_S2{@K*+b(?5q$D&@RB|hgD1<UrloFcEQ-)+F^Gr{rJ42Epvm`20Wu7Y`Nl4}?
zndi(iYrpio@Atmzd%pMk*0<KbZU45`wQY|c?%}$w^E}Su*pGeR&+Fa=xifU@Sk{q9
zB)YTGl9x!N74JwSY6%)@{H4t5NhJPRYay*-MI!CEOZ-RW+c|QLMA}6<D|zCIeds{5
zlk?`z=B1_baLcAWYyN24b<<+=`*r)HbXLVjDsn|d)<jX?QBBx;n~8Rnq+hmy*Y=d3
zYb9?h8eTjkJ<PIs)$<XiL;EEosj290O2~D*E(v#Y`!dy{H6;5k_ib{$-l(J4G{+ZB
zzS^V0!r8;QlflO4Z%Rnunn<^cxA(6m{>OAJ6$}2kAmL8@x|t!83jeHHX@H*+RR67)
zZXT_nd$-g!V(%;#LHSr%7*ysl>)39Y`9(juOyA4q)Og2)W3FZrrOrgat8im~S5L{H
zZOD1f`$}r~4E0S3b#?W4)~LM^<=rFpk{^vyRTDgBQ%3DYT=T`n#KdaoD*aoB?XsJ7
zKDMvoFOtczxsj<pF)=}Y`0%05Yto?gaZ`bi_tZg+gSN-R$UiNF)-!$T>+AELx>MXf
z^1Y_UlhSM^*fPSa=|1V0d-Uj0^20}u^j=GJS!U*FCcUS&{$;<1lapn@p2}z2fhb}u
zDo_6(8=3$9c#=q=<$vh5I$z%wSh|NivgP_xxu<AyR5d5bpta}b|EZb#^ZKs(*sR#>
zJYIQ*CwKgt=az%V?S{6GRy3OD=ejP;a@MM5T3$;xYyM*?At^~K-AqGMQ&T}fVYnrC
z4gF3L?GGz4sc#jZ6k_NqT2h{q!a_pSw6t0~PQ}WFt-DV9{P}ZQX6{3W4sG^(^ytwa
z%g6lue0}nD;q4o$b?_;JyA7w^F?r^>6N)EJyw6-`K;>BSa8Gl#9lgUoM8Ndax92jn
zw6tqBY!|ZX`r_^F{qyHfDKDnM)_k|>R|-|J@=9D75qH}cslGQ`(o7rURgyVngHsHv
zV#YRCQCm+9HE&`$5_DcQ=++wgp~m!bv-I*HzP%hArzEp%2amfgI(rqqb(x#wyZ&|G
zmK9hAbKe_QjK9$>_Vn^<Z*Ld2?qTr$z}{qL=iuOAWMp*m<Oc_VRO4E1#%;%~4GOfk
z1_cGJSiPQRB{g0@w>%7Dsgop8(}onArGE8xUzu~~5asO$X-{oGW+^FYX=!;fz*r{b
zOaTvZ)srtxy>E(%o6es$GU_YxVHIDTZbxY2E(Zn%M%s&~`l=5q$7f_HOQu}hMrSA^
z7b^7h%6bgjp#KeOYs`hFW_Y#HV8zo@7m1@XQ6oF%d=RhgfO3Mesc8a*f2b&}aghAe
z`uJndJI~&}<z=ARH|W(mT)X%pF;T4c!DE5X!&BMn+`h$|`aOE~+dCT@8$Z@$O^BGh
z6#wD{1<QA{NPei?bqx&-6V^?6R%K$}A530azrVhQUhCKVtGGC;O!YUW-g)aCTvz!j
zxwzyu4&Jvg-f*|T=;%qwn-{mWRli(iklnoaiNkmOj^p2rQfq2z@+NB3hI6M{`l{o2
z@}_@^xy({1OwAiNZ91hWf9~8yzaQCFPYOS@jm=zHVSVA3f6G;8F|GIzhArEZ(w>{{
zNcyG5I_jbOq=1=b<@#hEx^0%RlCka81M>2b*S4}t`0f;)ooiikRa8{8Xp*wD9R2?C
z^1%ZKYzOMzXcu@Kv+ANy3ff-8$D1?_M%!#(ogG?mXn7t=XsC`06G!&QXn#_^Yw`UZ
zXMA^7FNTDNYpSTIT)w>N(s{Lq0Rb^r4sG8dY`Z?+;x*IYnYu@hHiwkOWH}luUc9KC
zT%Y21*rUz*_^sA>majr)r0eD3M?=b1&lM>vDaHC!zHJ=rj!21BIQwd5sPaUEuP!^!
zhoV}`6fW+-xU4KyNlbi7D~aw|OiWS#L=1jru{&C3H)FYxA3k~VBs6q;e;28;y5V{o
zMN(fshVm#d&^Ujkp}9h5XD54M%0+|suUYkvAUIOwO~oXH!@ib<2<9&>&iNF}JPru(
zlvY#>usy1YpOk^-Y{6({SJyns%DLBnj=;;ZH9a3cexzd+$eAB4)pDJ=_UTk(x;eAo
z;=-Kq(hdQWCzOGf-0ZeH?zo`2$w4nKFV<t0(U<;y!jSVZA!b=;B?GwBVr({gNxQf#
z%ue)+PybkPx8O#thMghp8akHm`EE;|o=7u_2Fx_s>WPJ3%uXWVj2x#|t}(fz@;kB_
zxyd;Tla2j-eJw6i%?@K-$F|?cm0^Xhq^7!Ca74Gr+e69l%abFzHSsFKc0*0c*X0!z
zYbu|eX*RQ~P1J}zAN0-QmNpZynuYallT6<z6N|#2B9cojZM&Fr#;aGa%v<wtb=hNI
zg)d*a#1b&@?Rn{FXX(;-toUGka`&&cg5KWVg1hufCGxHtEnVz>C%6qkPvUYEv{|y#
z?ida<1PC1WX<km#L7cH~)GGK6s+t-xp4Y8SJM6Qg;jw`DYNFH{oc*inx>cKL14%-+
zP2HvLqpIorw^Jkos&NTiFlil41%A3yQ}g4WNOuw){@J(duU>M9q>9|bZ?33Mk~k<y
z|DV47UwqU5doR6X>0iB7MOIdJ7sCbB!u|;(H=!`O<oXL*=JT%k#qBQ>6YmqyWo#xz
z9B=~J--PuU4=~EqoR4EpUw^;oPI=neoz+Sdq|!Lg8mo%bu#Jp?rCUU7i+)TvuDB`D
z@9@foQAE=$U3l6qdC#6bDu)6ntg9qxsg8T%3`Xu70_ay#`{h%m3KTW&U%|#>TDv~;
z^2TTF&u7gT;tr^mpP*0s_@Fy*#P(ckdo^`8EyG%B>&-Sj6(?*S2L}hw&N?!lm6J=W
zH@RqV<;#^TSAedvn|C6Gm-Jd650hg&d+wa4udlC-)wuQKfJ$=xfQ67wGlOU~4uB$Q
z9UYyMCr{!Iv~CDEvf5>;lcP$#4&yRMeR@80S{k~ftX;eI?Afz#%<hMy=;dgd6Zg|*
zn;rM+mE>+z<zvT=)s#_1m0q~Dl89CVj(-Z!B8e65f2=CbyJxeBL^^W%|GL<We`AKV
zxkbfdAVfd0j!&jFfFuzo_5a7)d|i}9wN7ld=UJepqhX=vmM~#Evv~t#<}923(a}*P
zLdU5gQTty%)^6fdJ{0g`IcGXm1yEa?nVGq`xcJkfa&W6>yng-qg5;%3m)0#Gqhn(z
zZG#p<QNOXsg~qKPQ1+CPG&L?<ut=nq3FPKFbZEPf#Shf@R_C!Yoc*QK)X?12uvV&(
z!nh|DkWc*OSUFCA!<HO}2Dj4BpU-ppm*|lXZ`CxWuV_p&4LNFD6MfcSz_h_Skrz<2
zDcwAx*(`$Zy4>m09u(`r2AjqC878l5C`8x4hTc0(RH5cATei3?&U8CX4&1XjWg(Uz
zD#0df3-lx;CYE>V?gmc(mnFQ}wu2k}o}RiZuwBb9DCpSLckU9q9s~w5`w(C>%W0-F
zFe&4wC#zWodkx+DnUu*R6;&~^Ap)kslT*!htr`jn)fr#>IL=7!-n)0No`am+SCq%p
zGd2X8tGM9kIQ#N)tYkD!_Qs4hDlT<RkHuzYW-p3BQxv;t$-PZ`qb{PxMY8*sJPHch
z)~^`32biUnSd#5CM+R)ODMDdi4K~xBk&?R8I!vFRLqWwG^yQpuXiSmV^<!*Com(lM
zb>kK+V>B~w%<hDz1V#aJ{l;lV#>T09x!aJi#*O{VQZ&3w8$W;k9F=k{wyOSk#H4=J
zeWmg6H&1B7wa`Iz2lF0u-f`Ueb-f9r;5rHu$4Ud=Vh-}Xd-vE&{U1J*-2J3(oQr8g
zek+^vsf*PEThxv{8w%WNy_}}%T~oWdcc+<p84V`P7@7;58hW(VIzP>%{;g%_XHlog
zD~gJ?z27>s`V&KrS?TAyE}Zm4+A(T2P0F2Y03c%(LsqH>25_EkyL*?OE%C*R9#mZ^
zW(|JS_PE5X1Ik@D*X*?KeIA&SFC!z9Dw7XTG1>FX_q5$q<X0N1hBgbFLtrSP!qMxN
zOUWz~?cI74Gh6!c%)}=q*(n-)eXjS~Y|9=L%LLx4J|5v;W04beQMg-)Rp<My&Iw1i
zv5?k7DsOjZbkx1kP0AVl9AciAiEihx;As8zh=u85aa<7%_e@`0Qr>J|oL!5<^&;<W
zbb|NOJ^&_)isrt3`*wuBr{Q5rN(yG{?xnV?Tn#*0MROCcqM|<NCmU*&(G~Psug+H~
zEG{n2I_7xXX)?avM4&G07A7afkF}(v1PQjRQCi1_uh>=c@npNQiPml4%I|i!{xukE
z_)&2C_U(?24rReJg7hN4-Wez5a*_=V4egq3YC8k9Tz`JNH`VI4lw2+rB6K6FynF5|
zSLkt@zO+WOxEC*^<>bQg%VQ<Yphu6!tCh5Je|@0cD;H`%l-Zq_*%gd&P?1zjRL{gb
z0(Xfnj8|>jw(U!3WAHY13FF}t1kfx^O>%rcW$ogHc3Jn}zJ1EMj;WOK;a0cO04|r=
z@yY<Ahe2EBhLu=Q|EVTHO+G@5i;N}2BI=luW&tE_*L0)UooYps*_7~b&a+hpDeMU@
zJ9{kju3fwLsU_FxR6uukx0kf7tt~$?1xZ)!_0^jKo5y?vTWDggGad;kD-GnypZ@7p
zP)`Zihi)(tQRPwUG|`8g5ntlJ7fq8)t>QCcY0<L^oR_p6ZfX9tE+5GbeO_T<q0jbX
zyew$d+~2s&_QsNV9mji0dGbX%uqwyvbbaq6`+W@)_M$Wn)_DBEpJLJWp;pVyWvD6B
zk)Mg{GYyqvMdXQvxv8WVFO=Wt+!0_Y>9pk1%&BW1@!oobEVNOk6`ADSJM>1W*-`2@
z3b)8IZ9SqRDb$z=Oq=@4qSa+e!?ymK7Zca)KytbF@t4MP+P9BXka-dMD#`kTO_^5Z
z;iBj{(0|xuH)~|uhVfoIN90V>v*XUwuP2K$J(LY}#quw==D7$WKQ1jyu^qG2L!+B*
zH_YTWJw1)Eb)26b5nt$)&otFm)4RRXfU);C!cEE^$^w=Kw2h64xz8$yi4Ug@Iwi4u
z(J%g}Y6vyf$wP!4?!LQC)X^FpIl9ul^5OA5Jr<ejNX$+@ZmwCqe(P(OIosR@Z>FjM
z$H>|QHM{ULRaI4Njy+F@n<#n2*s@RbRhR1VuA;ulappe1%8xHWd~K}}$i#~X#q7bf
zM#&EJS&JS1@+}VCVfjl7!<ZQ&Fy7x<6Vw%FdPJJ2r|$PRZrqq<-Ag7%xGg&PuVRs}
ztB!6j_GJ@y&1oCYz<h?GL?Fge@BKO_BV#vE_w<Z^rR6e)maT|s`CL}E5V159QRFK&
zH;~+B((x)Tu@9*RrG-U&b0||5TH-(S$*X^V!2ajGy@!~*DCfgPW_l{0(XlyB)?ZII
zZDjg-8hyK@9&R5cjNYLKp;)nKAX0foXi7+Oed_Zg1ZL-oaa&sS-z%?~YE*r5#TpMf
zR$G(|MBo7KT`k>g;m0ldpY834Q=1>}ZjPCE<joB=<~UfP8l!(t4I048w8rxFVHI3l
z7<Oa!L!+Z^T`Bc`R54CL^1#7^h@56*spqk=#vErWUM-V%!IK|oc|ffo_dZ8gJ2Tq(
zz7wg%xHi$HJIpSRPl71FzF3k-2>q6sEUgQ3o(B`BOSKliSwzt%6dz}xF;Lpht(G=W
zd4{dp)Gcz8hXwA1pNpMcd#tO>ZKhPqEN7&MmOEn?H+LfLVyeYy7?Eb4zc6`N@U8vG
zPvraap+Yx4Jzz)6WShP4P=-4b9X@i<g4^O)#Qy#J>1dabQZ-WyE}}cf2xQyz2l)9h
z-f2D=4T5r4jp@6kB(%GEd71>oFcDCT4|x{r>A5bXOh3I*7qInMn4IYbj=Kenp+~-N
z{TeQ6ED%B`Fy#PJg5Wsn1}71)U8Z{P-<VsQJ`zjG4lbOO%80zKEclovqO|i>Vj`1Y
zbaeFL>y4%df1_FaBuK~fM%42?M9phTW%fApUj=ZK@zVzyqJOp?E1W{M5j3fL;OBSb
z#?Oy5#snDF+4qek>N0D+mO~;iFc7(K0W-9lz{ioY#!uO|-`%idqaS+Nl=C)tx5T4;
zQxe07NDo#s8kIVrhq3sdV>6g-sc68cSJeb5nc!(j@RuOd*lAgL48I``VpVtGg^E}E
z*7TS)XO)Es3r@GjM!2d<vTz$_vBVwSySKYoeDRIZ@;Tz7wny^h{Q7W*mU;D3+^QZA
z8GM$s!poZJ{@ifBR-i5MtLe#TD#y;xkF!#UHq>AfnXcjlQjn<gw8bt4A-6YJt(r+J
zxE%L@*i-H;;J^qxq^!JEu5I{y9X=vq%w*^eu5Z4L-mS+n)7p?|T`N9Iu|!5V_U-2q
zZMAJQ6<?YucN*`}7D%sXrd$3lW<&SI`LS{-y{3#Ctb%6WpNZmA33mkTjMp2d-_Jba
zk=iaoURQBb;>RaH4)W+&#FBdKBuRpn0N=`jRHO<{DgLIYw6vN*TQ6x~L%Bg>#8gu$
z)Yj_iq0MH@er^l>TE|n@;I9VNxG7ZgSjr)-yquVr)3Kb(!>bg{r+-Kz%lg-k_n;ZK
z3fm4~{aPEMc_CF8NgbyJ86>Qaqzqs=*qE+mS+WQDbGSB1tNUx1EazqaPG)ZPt=c|#
zyTp3ju6f=}*N;}Q2y?s)dO}3rVY!)aNReKVV#~-*e5;%!OJrPJoQ)xISrwcXji$th
z|72ttdKvB`#~pk;3`*nnQEI$&WV0c?wY-#L$~OEWtM*p!UWs0?pt)vBt&ye_PyYPd
z5<SO(H^q-?j7+zFAns`&#hRuV!1e3I1qX*20`HpCT~Jmw5D1BI=nC3@IYv^M2%xX+
zz&M&Y_9I|Cp5-cz9pLqlUNO#<-ekE*EiX3Hfnv1u>#q3xY;FG6FJE%H!=1?SjQCtP
zuSwl&5Nculm04L?Sbjxz-eOCtWiLL}MTy<!K$ZuP%3Yl8m+EM{%Qk5*QkjU4MK-JE
zIXe_QE$OdK5)3+zGG1O<nmzta27nWFe0FMhDNbwQ)!VnD>c)m>Plk{ok??E=>PBY!
zleBQx<=w(d2Z(H9mf0EL#htb9$dNY)Qt{bJUjg%$?PbkAJ^@=-nDPdJ0^UtVlOk&I
zLxT1@4Wc#@<TJ)|xHW&tsdcX9-OUyGxV}9y0p2@=)3kSO7#S_)@dZZ<S_t)SpyFu(
z$Yj?ETu{#$aigu-r$nZI+!AcEC}>@rOb3O9Z~)1;IOVo91r$kK)9LH!Ix%-ZUICk=
ziXR8($V*I~#2Cg{|JpTx?S10nZlKOVtEJbQ5X@Dov66o$?YecY$g`fGf{z+w;e+g^
z$9qfm1a@?Fi2;JiCXN;|YZ2dyOy#yT*9uM<Z1@F5#WbD5b&M)G_BY0Dxj0q#5$(=-
z$%p>_$gnlP&#lWFRFkCD8Y|B>n2$odUC=C);yh8UBpVPRY?s(N93g1lf?`Niz46|v
zZa~5<ekec!ndiU`qTv#87^M?zisX8unm{i;Zvj>xw+?J$)B19&=Q#O;g?(-d#*iWn
zl&>0j9HGI_uZLQ2%7b71VOSU!Su-j9hIW2yzD>Vka{c_f74-b-Vu&4p|LKKAiCq=b
zydOc;qr>9TYeV(}NJ}w2oyVMTCYowiz4avjO^HZ!2W5h-ASHREIvR|Vd|nf$dC`Ya
z{IoH7CbWPdZ(&j`y|Fn#E&YLCI;VdxLQc$03ZIOO`~fU<8~_h1z|zV}gIg&%q}d2-
zbnp2m4EiN<VaLnDMTI}~;R_=_w^`iQnw}nMCqr>yu2x0Mk%s#A@ZrY6d;n~Y)7~#$
zzB~pb-xs5rNkmzCqj^?u46;IWY;3v&k!cOmP$(rN(RDqlA+Q6R)4&_756B&E=ycGi
z?&slYO)<Qzs;X)oyiLY$QX&FHU%1q|?|ZL^IFWB6Cwi*m0t<EaA2{&p<;ypXgLG`c
zfPaRDy{zUrl^IIoGu;u)e))?NNq~!n8961|HKrQTRLA-FE-$C?$fFeSvSQBDNIhB(
zosYNi>8}h#%)6BZA4P^Uv^@Vli0`_jUQfkSI_90nEIU5l+f-?0QTj}GE6%=DJnT9p
z%ov2wR5Q=rLbFn`gJ_LV^QH8FOk@*zuf8KOR>ovinE(rcfo<qR&&^da!1PSbeR9UR
zzb1jehlv20!0~qdHKyNRTqOEt@Bno6HuNOY)hkLGHr1#^8r-0Sn?YtZfDy)<m-@Aq
zSa|h5;iVAuYw!P9MIN1Q_njJUUBZ+PRz52R-TO8<nJ0UImoalF3ujWReVEOlFuSq`
z!J~nkJk?}XF2j@iC?o_GrV{ytNMH!)FbpwPd#ls1sKx<eJ<>h_3u0zAgsDcVx9|J@
z65%>OWS3uBS}G$gZSZy;32^lN8a8j1qp#xQ2M|Im{00^1v?lRKk+Cm)=<fmM&3*U-
z3Zf_C&gj&?z2+-A`D8M^*=!hLiL{fTXsoT>1^~onI%&e_fH6pQYWX1o-~skfYWJQ!
zh5E^b=nP-Ic!3U_y_TF}BNtoMYwZbng7S~dT|EEb#eO9mYHA0`w%MY<WxzQ)&E#f-
zs`#R1oD-18YMrUxkWdN%0mpc=y-D@=?^u}MzkgR2l$1A1SDX7NfByV=Il22bR@Uuh
z%`62C3A$$9+Ji#A(mZ4l@IOgz3&xCR<>i+K&E1>;G>@K_wz07xdSgro=ow;#V%b7p
zI<3rsNMShTx4Nf0!mXu$qNuQN0iVk5dlh$br|kLjgqC=hjx}bYn+KHzq!VLJoyQdg
zg_pL!=A6X_eH$7YiUdncs=$Lb4hqy=>3h%W56*Hhc16K2$6U}rFN)(56Tfy^ZtLw*
zZb&S&^|f|OUAe3-6}MN~wZG{co1wnxezBYfgzHSIJep}m<7mcJlyGUSFgDler<D1D
z>*B_t;kOcT`A9Cg$efv(nT(3HX-~@Zt4TYYSMUa%FA5Q9v=`nYl2_Dt6x35r*t%ny
ztLN??AirsTW{{(rBDShMcXMjG<E_C*J&rlvc8(ve0zF7cNs%o(A1j6D*P&!;Igdys
zSj<&=kK}=mX<)r@>C#5sNHy`;s+Tq?yE!>|S7%M2%!7CV-J15s5+(Py7XZJi2i@qp
z@@+P}{pU}ey4B5EP=RFn?eF{g3U+KYxD}>_SrgX0apT5mmKAjT`@aoSv?pI16_WF>
z67`poAicsGvaDoB#L|5CjF;i6EW6>I+RJFq1nDicF^s5g-6a+1n`T}q#7XL5O)x4F
zaO0z5v~DCpfy~d8yMgE+@XjWAkt2{SJ*tdrYipBKQ~OHxMj)mHLI#mb?<3(#^nV|4
zNVSQOU?D|ueIqsJI!#r_DYZKGCjgbAGW^PM7~{_Bi9%KM`~=~W9`#8@<>ljfbTX|;
zTKP0_1gg}LxqFtW*oQS8vr^3FHjlpN@#Du(PI1uD5+uUN&%!1T2E283Wci^^;E+-x
z^>>As4>hFrW5mH0G5J9<B#PDNgGmG;d9r~(GDVSUiK)tm(0_%CIK0(l1r(=n^*>?N
zZC;sj`gAtkM!bi}76wVRpa2^H+>_MO6Wd0S-)8O$4=JY_J+`$YuxDgKl{QT=1Yom_
zHV-WHd!Im6u!vo`bm=W9%>4OI9<U55P{M(Y$w?dV3CdFdyVA~eG2JOC{HD~|kvH0l
z!XPywzeKps_9mpaWI0WlB?q0r!ih9Bm|WRC`~B;o-UNl{8<U`7D!+||9<!2*5F<J^
z0q61b{Dna<jN#6sXBlT|k?L`uztI|l8WtS{I?5+Bpcw|0QJmecl9dmQb(iO(<C4-t
z!}j+nrkX|0BlEh>PmeKI4Mjz50YOP>NJQAk%KA{y4u$4FDQIKx!+@P$hJwpq&aII}
z`GoeaT<iS72Vf}wEuxV8R2jifl%==2IwkU=Sx5JtMezogvRpph>pb}-5C;WJUY(%$
zK$$GuWpk~|c!!i>hBArsaPHoXlELw}w_ES*LAthX`)~)k{xORm@6d7+DnH-@t^mXt
zntbP_#RV*~JBp*=DF~#<VRo39NSWhJ_bk9666^?)Ii&9s%ZKz$P;vxw_3PaVzuHxJ
z>&Ji%RH~AIgVsbp{5uoi+9nXAxtiu!LbH>D!p_sbv=;i5Q023(5Ix+-05N<hh5c*8
zzUxV~{kGWvsN)Snc%|Z84^!k)`RRlj&f#-_m829Kzpn)1cAq@NK!Z-1m1ue=8h$+7
zBXgd!3aetgA#$i^FGgQtY-V8*aR=h1#imx{jhPHB`*K^jB@-kgGmoYSE<;(cBXpUJ
zD1vWh#ZC8R56uS!6_udi;N!nBcW-Lyr}V0=x`cPfTNfh=UOPHkuj@+?2t;o$ul|<Y
zx!Y?uEsprQ@pI1)12Bd=4Q=B$cu>%m0T9LJ{Odia_dAaep<iH@SWUkZ>JT^<UdxV8
zfPSmNI9I~{p%HRC>h0Tadc06ye|W~qN4O%dgBxF$-P{7A>8w8|dZDDY0Hp-gYrH{=
zz`sO;gj3xR+FmX)=Dplwjcn~1HtEYuW1?GDL7xSIt+6T>RZ>uR92Dd*+EIcNw%E7(
zcro*sDeD8HfWe%BHi0o*B0!NBPoDiqAv0l3K{em4+y%G-F`<{ZE38eM!G`MLsbS#n
z54YF8H*p+HtzAU!-{hB(kwKu!VE(gei@mWC8+M8ig$VsyFFG3dBpOnU(TC60sF~m4
zuO75*#U%hyB-fiTt6QA#&)y_=xu8|GV|gM@CtoWKK&!I@U<rYZe1-KkHMfv_y&SM5
zUGWFbo<Cj4qt0FX$0}Oj*ZVs~RW&tT5pLmPE{@-#&mxl_6ae_c5@@+myaO^D&V6A2
zeus7+K}kL67^t*lOVD`vqn{5F4yAE|cm^{|zNDly<sUKM(NW?L4#yti12P~A3eiye
z7V}{MiAAV=Igm$-(DefYl{7UML17+k2|OIa(p6)9f^ah(4#uKrYHHHVb-V{Zl-p>@
zUV_*0{Nyh$_Hv6oer-0JDtDbs-WzUDW@kSkdFk?HR26Y8$KL0l+T1XzDLSYlnUpWV
zN8>Q;4G*1Q;6T>QK+Q)#@Ks?BkXarq5O5BPzR-~)7bO9}x?n)C>AMnGI)N7QsdgKh
zS|r^e0D1A+nhz#MPt>Yb5vH?u1~GDBEZ#Z6$S?P#&CJl3F#Gw6%}Pt2oHdZ;tm5z$
zA(MftY6A?I+Y&UgYuZOvdgTIv;cviO!<j~{SvGML1HW8$Q+MPIw<Zg1C^M)o<;gtS
z0!P0j^!}kjtX^+@wD{~Y=m#p1;KAbKBGCf?!>T0I5bb1i-JU>oGI`nhA&m8bThvA1
zunL%<7_cTn4KPm3<P;X3n@=^q#joiyQ4JnMKwb0lWeze}0^^Je{hZ>=z$N;Z3+jV4
ze+N)ird4+lEkVHGj3_rn-9#opr|?ea&Fx1ZX1yrRbA#y(ZUm`DfYz6>;j<t{QD%^D
zSr`Fu(Bq8&00+l64BDbls7h)gg;{L@UTSFCLAG3!Ss-rX_eqxtfUEonGaqK7vAk>g
z`nE$&P#*#V0%o1_9`LG8nm)-t<GuBbX=56A`lX2^Hy14Jx$)jH2nd=G5&)vK{S}G}
z3(<L3>hWf(C%!wjvMQPKq^#TP**V#TuL-?(2+zP({-wpK{8IstxB?$NQckGZDeCyK
zxEQ#<tbYPa?r?M+)G?`OkYMDWw%J5?JwJVe@vMvttGMg@+iRs8(C1>EcuIe&cA}f;
zt=h{#gScB-s{+r>3?K-<S);@4`tM9I<yE9PSfzW+l#kyqYr4)>>@xYrWa(W<Hr$76
z-1iX+KZUCJM7h{pJVhO|qB=Z$v{s#i3{8So<px+$GOyA1eT?YdOfY;P8nm>u*bZG4
z2tfzO$p}wj_p1dWMk$0Wq&GvrA*1gv>{?xv47ENO8qqH6+goK-RO*Rt*oPj5HF@!c
zarjsy^-#c6Z?2+cg84B4$SHrZKR@N=BPivZa-q{0(aE05cIN!)I}{|emv+M~+=2ZN
zw_dz@^-Yn;_()|=jN70P6tEZ@8HLKlPEKLozg1P$#;c6MHwAT{LMiWdYP-W64lIYt
zjm7s*msN&rPte=(@$o2}vC+{6>4mPKvJ<oV4kSp}+D;;iZe?RzYHVFhBRZGa2^j$W
zs)$8skeR?j)6@0|b)O#}#7YAL!OP1F+fHqqQWB^aq<C|*aP(r+k%~n+Z=#<+cgSrC
z4Gl#zdXY#IaNC5yjsODhx}F~JTRtia&>&hGS0uMQtBt-4+5)M27?*36K2V!ln#neB
z5Us<*-(Oat%|Z|aVC?u0@!9l!&&kPwL^=s*hlF>^WBp7gk6TbgMC)r~2b4J!o$$lj
zCjq;NejiEY7Tf$)SO8JbZkFlOGO59-h!@w<(jw5JxH5ZVLMDfrg#l{;Gg4l>*vG-q
z9t%w*&t;DD{F4gr=C!dexk@(Etkk)5B>2kZ%g^Ze&YeALIGx4s{y&k=gWU%}KKq-$
ztF-!?n?g0dh-vGifeFXd)<q<dc*XuImo-w^MNH3pBu$?2d{`x#9IQTMnF*>yYBxCW
z{rg{DW8I>GE<rO%DO>!Gc1E+=;NC?@F@I`#wD(;$l{S>$unld+LI?FUiMq8~IOM5t
z7aX6f7N2PenoC(FS+%^SB~_#yzD>aCzu9<~e|snNCnqMIU3N6;Hi<4W%xKa;pJY(B
z6S9o5;QMK0ql%3gvkav5x0VHqe`zw(Z^H=;8vY1Zd1FSc%HjCiDGZO&>h7}U2Jh5(
z?!Xeg(W1irNkUyk`d8MFBv?<nuXb1D3Kf3+Bv*`aM2*W&sn|u0^h5Z_?~A?qU}p~j
z9ua@CVkuA<L0qp|f?G5n|AVPV=|n+=tJEIFoJqzYT^6pZ+XgKZG&~Iqw^sHLsO55e
zN(uoPaS$HhOBaaaK}3W&R`%V3r>e^ivIkG8$AQ0ymT2^Xv$ONB;pZ}eOiWAwmll9Y
zkTd~bA;1yT1A3zv`G~|nKX{@@!k7wZeQ2%Elk=PjUm!+J4K`j+Z|wdO<nQOlP9^}S
zxLaPKPCdd1Ov)odmy2}aJ?U<n-3)p3<;$0lU%L0F{cDTpHf@?S@Yn2O=xNFvoACm+
z_|qEN-O(}7ny=-=Kd(bly%&3Eq*X&&QgWU%q<s5cGLu)KiNMJzD$*`aZ-&U!5muZH
zocTP65>SFzxFASSuv@Wquf9QB$yj;%8h<p`u^T^c|7p`JT3hkE+fs~6MJhlyiv#@-
zIHu3QF{wJKsj1{|_AGSS&D><U*vl`Kg+OM(z{)o%EeYkNF=vY!UqrbQvS+`dq!fS)
zj(D}o6rsn=rZh>lak?^5AN?bn0td!sVB6$p*`(&`D?aKoO}2l}FAo-Lh89Mv=Tj|d
zKZWgbe6Tujcv)~@eF%ie22PSHuZEI9tg$(C5GD=Pgc`5~=`D`^@QbSDI!=HGZ336c
zIASkCcp|sHb-Z-xAtghddocgE-)~OuHc1rqodAf?P~YERM)gGCB)oWm`zwYhhXCLp
zZ@3FH?AwuAodjKuy3XzTZ!DA(6sva{M3l!S=)6(>W9;tMTjkw+LFZSq+V=tMx#pwz
zbtz^Wml;2qJZ`dy=j5H!t9S1*)zANKFwE~bpjev!Frg+!{2qvj$x|nc#qxuTho{z1
zQ{8F~m1%9i_RfEEk<M*W*z(bAKQmFQm9JZT|JfN|qa|5+`6Q%n1kU~#F4V@$2bk_d
z(fdWa#<+6zvARLqTMLc~f47%1&R>K1DznYIJ;8G-C%G!rtF#kr28e>J^fPvfG8x0V
zuKBGmZhq;WxJM~`b!r!%VE?L3_do@42EyoTs(0n&$v-ApBen4tP()UP_N<+M;p<{!
zXXoYPlZwvCQSX<jYI*g!wDgpx0!ooIU{30i$8W3V<q-M`xmYXoox}M{uI=wPHcV{8
zs5kJOz)$j96uSpHGcz;fjROT6k_m#bAy$Aba?_?w5Y2u82D-s+!%lD7nx`R&nw+#?
z2T_4cwAmia)QSoUs6IU0+-4o0{Ql;wo*{wTM?1;MsFG$9RCD$a8CDjk2}{scw{Iv0
zqxhU_JHNpNw2g56D#j~6^|=Ke1MLMg2r#%b8|v2mzL32f>XP<AUA5KKxpPC=xCJYr
z?cmpe)7P$~-f$Q6>NdKhZ^*?(D91ohnNSljO_;K*<{>n-azVV$+gATJ&)R-^d-eSv
z%IWNv^phRHeuJNYg^-(vCn+wjzdEiKUH%Lf>D~1zwnE#Vo%XRj9)|NAa~RcARc*j2
z^uLB~<2@?*q%i9JpsjIwqO`R1;kxx_aBoar&t<>#F^KsVKa<_Qla0+S@nwe2SGVzg
z{rKmy!gZXQw<DeuxOAaYw0(1%u$b&tp;}sMO!_2v?VNG1%=0tP1w^&|%Db(yo5?!K
z_3H%=`ZW$t&MmCv5^Fnq1#8r_=UJIPU!9t0h4R6LPv#z_+}Q(fOS$L1>Ojt=7af0~
zxom20Nts!$84@zQa`gOHO<I}~P!}5RbotCMGFrhVleUti%JpZJ_$$ZjpIc47e4t9%
z$64im@u<`6hTfp(#-~WT`u;!J(*Tcr1Khd%nN8hend;n(Z!kI`3HP=*=}V<wBUx<t
zFWYTn-mD;fe7Mw8+ikz=>WG8XN4>OZE?Ju?xP0BWK3zkYeE(L6|9L+RGb43bM_IGc
zsQpioos53d_DR?P!vybJfW&`QYW_>(?|*97C0u``UF6Y6d&FvFK{*~;X^>*~d&W@H
zW&TtM|HXG-CEUp&-v{<v$LpGXPv1joVEwD9eYdq|9|-dGE2+=LO)t0(`u6o1`Il_I
zbGp#GynAeJVW;oj8oIUD;*!iKv);FZMxnFKNzsV&{ZVqU(<far_x_UTjs-EdU+R1m
z_m#UDe!a?VOR;_5-Z?e5LUwzSqY|H1M7XD?zTi3LhSDT*O<9G~uZ7onc`bvsXEKi7
zEznNgc(KB9q|ciS4?atz=uY4A?oEtw)kXrWeN$BVt>ywDrZ5w*Zana?yMB1e_%2+s
ztcN5(Li?9|f5`KJ$)0+}#tPn`61|P5c5q|Wa~|0?AhqxPd3~eV&dPz>>t9+aB0J@G
zF;E=%#-C3-q)-ZV?upE?3>U7y#V&E?qNU~Az5_sXs_t^<&-2$g#TB<B$iUI#raOJS
zj3={p`ktA4_gXc9NoHL5g4%=MnVM1L*elf^Yscpn8fLEXruW$0V){Wl9W51J5U}~z
znt3sl8d?41=l8Iwp*Hqtjk)A9XiVRd$9DPyvP1rn29q>>=Xokqucu-s5q$sNeOs_A
z*hQX~G$=dP!kyv9?!N1<ChvcB3t0N-YSNGFqr<Czd=Py{#S!%{13${faPk@<({bI}
z`5$jCQ<tP68Tg04{U~T&I3pIA$3<1q!Vt;G#Xe?O*VkXBpDf7Dtw`wk6{}7W?fZXa
zr&?7VRmJ7x*sfI|awGrg8VQu_zHeP7R~+R@vsqm0u&lJul4s2k!P5yBN#v9NTT4=y
z@ux6<-d#nZjGIg#Tn_vjTqaep935UG(IzOdYxZB{xpz{xK_PbTmvQE+2_ClUZP`dN
z-7HGjirlI3W&EZW)gL1IYz|@p+x+GC=thPp{xz01brdm3HeXAN@A9Cjkln|{rIBHw
zt*!l(pn*3pA06G@5Um>9EIwut8Ukexwe_X|ZVgpsWte#10q5%L>7mQ#QB5&GX^m4%
z2nMwkueOHtZiat&?-}syHs))uURl|5;UB%a_iI|AaM#X^brWtNw2~1*H*_E&>olL1
zI59S~t<9^w4CyfD2BY>10PB07z}T?fule-l%l5-pC3Z1<DlPTWsKWIyI9c=FId0iO
zp`cs3eW<*oq(0v*e{vrwGS{%h_t<hkk|pjQ;O1uX!``U}oed8+?UlQo{}KZ8fc{GA
zo198hrgKio>GcoE^;2mtatF=T3NIR6rr!QKU`yMaazQnRn+$)@c}6oT$^mNY7Yz-n
zjD{kO9<Yxv##U`xi~;-!S>U_`0<L3J=<({053=LqZ(If}tnK$X>l9Ts2=qUsoE~Ku
ze!J`b_jSa!fK;x8JII1F&amk&&hNanlPABt+q|H-RHb!%F74ZzbF>zLk9*&SK_A9J
zBZzJW1=`w>o`DX-h|D2LY>5ikyJycmO6it%t)ERwKAxs4KGb>y8=3IwHj^U5jGwnA
zP)WLzc0jr&{IH9Y@itrbm8~#MuD{|G=^_?U(|tWvm2PrHKyf>4lcux!{5~@at_d7f
zI&Yqod1URb?Vku<-i!HkAx$@xrH@P0KF52i279llZ<31pMb=NQrD;F%y*|!se0&^@
z0Ja_wV+%U*mtghNQ){XlKv^scn#YdSU(MG9Ktzr;-R-Rnls_1v06UknGat`kqp&|+
zTejViSy{(Ty%m4&Kp>1>X#3#6hRROtUg)$`P*U=e{zr(ad20txdSe2en76O5{IY`S
z-otx^&6^wa1A$LP^c>)prdSWu9a(Kk$h&bT+&6%w-9rYBq*Doxh~OdD*Vi90Amnvb
zI?OJ}+<Sz`!K%rlte}vJtw5lQp|}(!8WOQDinpA&wR!?<MpTpz*uRDe2JS#(C{sIq
zuYaU4diUcj2HhEy*K7IozdV8I6k7Y#K)wIt$2Q|V7hyy|%)?>xF>CqjjyOh!RG|Ot
zEDh8Gw+4s_4NFc|c88!D1Z&ajueYtmSu56$O9*m_iY`38yVH|c<)n(fiWT74{)5dw
z|0xFe=jv`8&+5KZpoc~wMme2)z&p#hL2jFQ?RtkCY&&7m9S$6s{Q7{W@CWfn_oCmf
zJe-@<uzuMN_`k3#=|8<N0-Sg#{B@>c@{%gH;dn{SPIln+g)IjhJ}jOJu#;nU4<0mz
zsTG?o4V1eE(_1z3U0wX+lGHQb!j*V@;FQz1f%YwnH-F53yEqp9C3`q8`_-#HkyQ{R
zGQs%KbSr)TpEkL69dLGi8n#Q~j1>A(sN6URVg5N>BaxB2;M9S>wBhclz`tD(_`WJw
zU8bGI7&CwTL5q|iA9j4XBZYyL@Y>!gBuLn54Fe_eF-`5AG+%jkNDY^ys4lgh+1jo2
z2yUL5+S<EX;pIFjY%6<SD8r5cj{syk^a4SCeizl%bD-%F)?(FEBXDI|@vLXJ&{PiW
z9UG`m29t?gqNL?A0TBXTTNklb(CdVfMQ|#xB*W7*mGZD*&HSIt$6LY)yl*h@qSWY5
z`Y{O$*WMTP@?9`7)O3sjuN%Q)5vXCSg0!EyCsZrT>Tg?cZ+=t7xGY><pRx$qB*VN_
zEW23@SLp4%p0O7q?EK6a7gOovU}Jgcu}w2sephZX_=yBTUpGs4dgo4sB_S;@&&=pm
z$ZOV=p=Or3hk*v>)IG|V=G<X@ayrSu>A(MTw!uEeyy=2t453=bJv&NFoW0@mJ;YRs
z%Y)~Wkf^Sn#qfclHU$gr&nerGV(9<P6BnkzF41oe+L6p_-BaNqE(Lmf!`8#Q7-%{=
zJEii>b}Q_X+%4Y)nM6tt#3lQwDGz$Koz7Pz|6vF3A47xY=6|1}-bFpLy4O!(G6M6Q
zIr&&YD1V&TE(%r;0|<tCpUc9JG1|d(RH|+;P%-*}{zMYoI9N$j>l3~1Hl?Fv+%XrN
z60@eg_9x4bGEQ>CI|D{A8bFk7T6b=ZQ9A7JIP4P2w~ik6cR&b3(v=zsg9mf{3gWo{
zN$BTm)vaVL*t-Ms;Il{@a@+4C-9?;IH*!KrUHuTbf#>q?&A*I)A8@PgucnqT1G#aN
zp~crM#x3N91_x(BdWKSz5gomogM&j@*a@ctZQ^N9-w-H$U+Kx3KgMLv%!&oU*x?ED
z7r)`9f;*vF2`eMC43uapX5$hPw)hokHKCjlJ5fF@-(bZ;$(Qo-){Gm^C{57&apx<<
zc!;CiDAg7|q8G$rfGN2gbAFq!t;y}|ibP4voDU+z<bE3|X##NzvWyT<u&LRj^jN;?
zTCx0!M25%-V$c8XBS((tIsB$^68}8oNC}#DLJh(GfcB|5Vuu3ka0aER^!ER9GqH{-
zx7uxvhwWRsh&<qTU%m@MUzz?Gc!4c%L&xGC|7{|F^FL_O^)e?5XXj6S{-7@S@BqWf
zq`hM>N~>X$60~6K^~PF*!Z!Qk6x#kpXdMxcA8(nj<~?@I0bD+0BUmR0A0+hq@w&&h
z?0-MAro6tUW(%S)DXZ^QQQ&q9*VK(f@V1dpE*RE&)PDbd`%rBF==48BwnJVY*n$tb
z=@-88M*#tgKh}ubgYjp<hKG``U%!@=7>absAGC#VCEd{4@^8c2U(!TKU+PYSsXlB!
z=Z*GE&!5abNuRR#@2lhJ??mRzOL&sOMg%^7eC^sb(ea8?%o^4grxmX2>l-P6veJ*E
zyUedS&%PWhSt3D@e!rHN%ge|Zi`W!6;7$$`vsbYqAwNF;5MzC|o%yHH`FUr?wVCtJ
z9r{mST6I!#RPT7$SEz|mu%zRXVPrPt75)2$<fYn{usf2~MK`uDv2D5bs_H<B!L?4I
zr>7$arM$MJT$NgF!x$u$^1zfi?SjVsG)AM%K2nkuj=I-Qdj?fn`rlFJd};WIGe*(K
zPsO3dtKi@*U4h~tgYU!LM)u*|BIPR1UGqZW$M&A8N)b>1ojJSC!q!eb);F+y%^x$J
zC%Nl`KucBg6M1FlPSbLgFFaH4XJ|#&Q8i2~n?}KpriptP7#P?u&S$2l6JkuLR!tCl
zg$JK7ANGg8eQODR8i}bYV!@<bj~BqT%qUe1zChV(tVGF=I&Xxs=|s=|;~DAM#5Vgu
z`fpS^lHatZn`eCAN`qQjIyx9uAV)!UsMLBsg2Yd#Gt(owpeXT8W~L|#e&OLQvw6ph
z18tmX=Kc7J4<0-yL+r4w8D%9$?fd+*(2I~4&gOq4ld+o>CW0@YKVMR)UVKU2Bk}hu
z`1*whUU_1Bh=`~t+A5c&FG+59#+V(23Kv^7Er<s?;4vK7?w6#Qd)j;JMQo;SY7&9l
z08!tt_>qv#%gV|sD=Xo%I2~y7dfWEO%yxoygQz09C*2rH09zEy9l*=O0mFDvp<0g@
z4m5+{m{)yZ%03hnEG=@{9t;cDy3a<_6-aw*%w-=LSq}_2TpNULn(o$B+nVe+(dXge
z@%{UiIc8DI#gp{KzBEgopn*B}?$v@pp|>g)W}_8rHiW`O0N#GQTRlrp<+_IiYgl6t
zn&chP*VV$q!;w#(K79&j#{n)bQv-v;A!RtpL8Zij+N20Cfv7dT<81lAV?l1{{6dzc
zg~hL7mczg2WxM+F<2Ba*SngWik(Q0yS4qoHMmhX@GxpZsjr0270BiMc%4fy0((6vO
z%zXX(k44I$b-4HlK54gHGRb{salcj)iDVre6-8JqN=pxlh~(_`rX^Xp;ru^ppe8-q
zf0^l!a>c2lRiq5@|KKWq#K7|$zwey0a}EX1IyvAOweK7_371mXHxHpxgVv2&yL&f`
zXfe=Nn#JJJC8TfIY&Wp*ODj~v17qHjGXo>n>Ebf<RCs6xkjCBh+oE0>Sy));>2<)K
z0k5ih+5(AWS9n>7WUU<>GuWIxg+?u5506%!Go)1jTPrIoY`WKJ6cQH~hYEq)N{wEI
zQSXdW;rhYj25S8n-R7qmckFPP8NJr{0q?*8;Krq%aadHe8Gt6UjGB~jRH`b)Eq4Ry
zml~Wv@Ig*ZOeDYt%cZ0+0k`<gn>UF)5bW&uARR3&qyrgQSv;mBiy-GwuWKd|1@Iq1
z)g}{L5h)Z3^ylu*&gU^PfL>nM_xuwkl|`W;WabHdgaw3sn)U0y^Cx6vOv4#6YeC2-
zkn6RYI7sepaw|xJX>Lodj3W~hrk7hFJS@nVfVcJW@kvv__0+~4IB-CfR##V7>SJkX
zKV;K1A9zT)V5WX@J_gR}`5R5>D?C3zdSGQ?F)8iu?_WVpEvbhGoP0ImA-!9;LC^2b
z{iUFw9ln+XZPj$h{R;(HSQGS^7<jJ9#G^^jJTwSwprH}6>3amHz~mtI6YZd%LcRzH
zU~+R?f=~IyD<ie)bn{j?jlHB{7UEM|iC}74K^I9Iazytu_~eiec*sHzTgsC8#cb@|
zva(910OS<p`BNw~$@S`&E{(!|b^q?z$cO{ZFl^M=G(bh_Nkm*{y#^efJ4x@Gwxo7&
z|9(bBOft@DJGXYj25z`mMt<ITT6t8~X=%{~xf3ZuUtb>^F`gDuku)|_u}ClrU<=8a
z;N?e9T)Xx>YTpk2x#^K-9JyPVnBb6^g~kRs4x8Dm^cYDJc2E&3?i`zbmjN>IOmJ#S
z%KZFHRcYKao+9|y2jKR=z6<it!NDC&O!@*LfTFMhB5yekH61&%nwn&7h&9;U`r_g<
z_yBn!Dq}{4tI#-pMaL(#o1dTm$dMcH++Y)XLMn><*F>-g2eFuY7RT3Aqi-|}t{KwU
zKD*_b)Ww0}h&1doc3{NwW6(n40WxNn=f=9_l@bn<-i3{96(%)n+1qDOt|%$B<~UU4
zrKP3W0s`ZqEMP$fGT0WlJdi=t<jaF&APr_3xKNSId?3uyRwOalH2|l>A+V1O<59NT
zQl6=TmDL!mPb{pgTcpq>_S!eSdndu@0J#(mqe&@rb{$>av9Yl*Ed`SM92^N2#<TLX
zVs49}0Rf82%4lV)@5ybJ!gB!>H+RBn1DlfX{rebT>(i%CU%K=Sc0?>QDLsrd#0`8s
zo}zP4`ts$=l9C@EA5_B5ZaC!eG>nrcPs)Xh5Otl=VWLj~v#G25>|okX(vN_}K`#cX
zNL^F2Ea`LSzLid-(-##NH}XVjt!46SYEo|;9T+e{bj&54pr`k(t5aT6tbr~qTNKs^
zl*xotJs73kw6Bv$P1H(}R2iwCK7B$?#1feW)P_(rJ8PR2hvkQx5`pl8EkQ*4r*QPp
zA(O#|)K#li0ab6f4;xqgZ<`+-n^;qE@!nm#u40t%c?{HjPtS&o8@gbrursH>k%5ZT
z>>IVP^h%9rXWc$-A=t=&pP=!_|4v^2-~ZS&{ox*=1al$t+xRZhSt&Wm#FJP5_#bkx
B2I~L-

diff --git a/15-bayesiana-calibracion.md b/15-bayesiana-calibracion.md
index 689b4cf..3f2b6e4 100644
--- a/15-bayesiana-calibracion.md
+++ b/15-bayesiana-calibracion.md
@@ -713,7 +713,8 @@ se muestra como una horizontal punteada.
 
 ## Ejemplo: estimación de proporciones {-}
 
-Ahora repetimos el ejercicio 
+Ahora repetimos el ejercicio de la estimación de la proporción de hogares con ingresos
+superiores a 150 mil.
 
 
 ``` r
diff --git a/16-bayes-mcmc.md b/16-bayes-mcmc.md
index 0a1d31d..0f567a2 100644
--- a/16-bayes-mcmc.md
+++ b/16-bayes-mcmc.md
@@ -189,7 +189,7 @@ c(media_post, momento_2_post)
 ```
 
 ```
-## [1] 0.7155559 0.5372170
+## [1] 0.7147007 0.5364443
 ```
 
 Y podemos aproximar de esta manera cualquier cantidad de interés que esté basada
@@ -202,7 +202,7 @@ mean(exp(theta) > 2)
 ```
 
 ```
-## [1] 0.5958
+## [1] 0.5959
 ```
 y así sucesivamente. 
 
@@ -286,10 +286,10 @@ simular_conjunta(1, datos)
 ## # A tibble: 4 × 2
 ##   sabor     valor_sim
 ##   <chr>         <dbl>
-## 1 fresa         0.755
-## 2 limón         0.783
-## 3 mango         0.819
-## 4 guanábana     0.569
+## 1 fresa         0.886
+## 2 limón         0.729
+## 3 mango         0.701
+## 4 guanábana     0.493
 ```
 
 
@@ -306,16 +306,16 @@ sims_posterior
 ## # A tibble: 20,000 × 3
 ##      rep sabor     valor_sim
 ##    <int> <chr>         <dbl>
-##  1     1 fresa         0.732
-##  2     1 limón         0.831
-##  3     1 mango         0.850
-##  4     1 guanábana     0.397
-##  5     2 fresa         0.670
-##  6     2 limón         0.839
-##  7     2 mango         0.664
-##  8     2 guanábana     0.558
-##  9     3 fresa         0.671
-## 10     3 limón         0.758
+##  1     1 fresa         0.727
+##  2     1 limón         0.823
+##  3     1 mango         0.849
+##  4     1 guanábana     0.474
+##  5     2 fresa         0.659
+##  6     2 limón         0.785
+##  7     2 mango         0.866
+##  8     2 guanábana     0.631
+##  9     3 fresa         0.553
+## 10     3 limón         0.719
 ## # ℹ 19,990 more rows
 ```
 
@@ -338,9 +338,9 @@ sims_posterior %>%
 ##   sabor         n   prop
 ##   <chr>     <int>  <dbl>
 ## 1 fresa      1264 0.0632
-## 2 guanábana     8 0.0004
-## 3 limón      5396 0.270 
-## 4 mango     13332 0.667
+## 2 guanábana    20 0.001 
+## 3 limón      5424 0.271 
+## 4 mango     13292 0.665
 ```
 Y vemos que los mejores sabores son mango y limón. La probabilidad posterior de
 que mango sea el sabor preferido por la población es de 66%. La integral correspondiente
@@ -372,7 +372,7 @@ de una distribución cualquiera $p(\theta) = K f(\theta)$, donde sólo conocemos
 la función $f(\theta)$.
 
 
-## Ejemplo de islas
+## Ejemplo de islas {-}
 
 Comenzamos revisando el ejemplo de las islas en @Kruschke (7.2) para tener más intuición de cómo funciona este algoritmo.
 
@@ -746,8 +746,8 @@ tibble(metodo = c("sim Metrópolis", "sim Independiente", "exacto"),
 ## # A tibble: 3 × 2
 ##   metodo            media_post
 ##   <chr>                  <dbl>
-## 1 sim Metrópolis         0.605
-## 2 sim Independiente      0.602
+## 1 sim Metrópolis         0.613
+## 2 sim Independiente      0.600
 ## 3 exacto                 0.6
 ```
 
@@ -894,7 +894,7 @@ estimaciones_media %>% bind_rows(tibble(tipo = "exacta", media = 20/100)) %>%
 ## # A tibble: 4 × 2
 ##   tipo            media
 ##   <chr>           <dbl>
-## 1 salto chico     0.128
+## 1 salto chico     0.132
 ## 2 salto grande    0.190
 ## 3 salto apropiado 0.203
 ## 4 exacta          0.2
diff --git a/16-bayes-mcmc_files/figure-html/unnamed-chunk-13-1.png b/16-bayes-mcmc_files/figure-html/unnamed-chunk-13-1.png
index 3a154bace33d0f1de5dc28c680c09f1f63ed75f8..91032a2a7207d4e207b34fb7764a4494b8f846c3 100644
GIT binary patch
literal 42056
zcmc$_^;et0+6792I}~?{yHkpX0>ughFK)$)y9X$g7A*uX5ZsCtcPn1pin|wryWI4g
z@7zD(-XD^+lJ$-}GxI*O_nt3cH3eKON-QKKBwQs$IZY%aAPfl!1%iQ$_=hYD{~iem
z843JOTORQf0001iKxAZO6ciLxR8%xHG<0-y3=9lROiU~+ENpCS92^{6TwFXnJbZk7
z0s;a;Lc$j>UJwxx5fc-WkdTm)l9G{;k&}~CP*6}(Qc_V-QBzaX(9qD*($dk<(bLnv
zeEE`rfq{{ck%@_knVFe|g@u)sm5q&!ot>S7gM*WklZ%Uso12@5hliJ!myeH+pPyeq
zKtNDXP)JBfSXlVgt5+f-BClV+78Mm06B82`7nhKbkd%~^l9G~^mIi@9GBPr6-n@~O
zm6el|lb4r&`}VDZf`X!=qLPx5va+&@ipslp?^IP))zs9~)z!gZu!e?)rlzKrmX@}*
zwvLXDuCA`0p5FWS@AdWd4Gatn4GoQqjEs$qO-xKoO-;?r%*@TrEi5c7EiFHM_+Vva
zWo>P3V`F1$Yinm`XK!!s;NbA_<3~qFM<*vIXJ=;@7Z+DoS2s5|cXxLW4-ZdIPcJVo
zZ*OlOA0J;|Uq3%Te}DggfPlcjz@VU@;NV~g1QHSw5*ivB78dsD)2Hz8@Q8?r&!0a>
zMn-=5^5yH-uTfD^(b3T{F)^{Rv2k&6@$vEBzI{tbNJvafghHW7NlD4c$tfu*si~=H
zX=&-{=@}Ur-@kv)%*@Qn%F52phQVMtIXStxxp{ec`T6+;1qFqLg+)b0#l^)XB_*Y$
zrDbJhKYsiuFE6jCsHm*0tg5Q2uCD(1^Jh&>O>J#$U0q#$eSJeiLt|rOQ&UrOb8|~e
zOKWRuTU%RudwWMmM`veeS65eecXv-u&#zy<dV72O`uh6&`v(RF1_uX+hK7cRhkyV6
zJu)&fIyyQwHa0#!J~1&dIXO8sH8njwJu@>iJ3Bi!H}~hypZWRug@uL1#l@wirRC-2
zm6es%)z!7Nwe|J&jg5`X&CRW?t?ljYot>TC-QB&tz5V_DgM)*^!^6LS{~jG39UmXV
z;qa4_lhf1Fv$M1F^Ye>~i_6Q)tE;Q)>+74Fo7>ylySuyl`}>E7hsVdqr>Cdq=jV+o
z<3Pkafa#*B?~a5-^ZDNg5Vw75g@i<pq$DS!?eqO_rneb!?_$3QW*dhxq5Ma-tBOC>
z4c>%uD0U3JK(@QY3L!!aNo`C?EFggW8^b3Cc`$||rfMn3HsurP8|)@;Z=0;Pwney;
z_2~<q!JudQrO}b2k<r~ksdI^hxui0JWRJQI2&bNx8^Duf^8f#@Arc5J)%nWu92uLt
z_hvS)(W%WeG{0BoLD|cmM@g$@-O{}Z<z>DR(%$}PFOz2YzTO?XTG{p1pq8E==V!>*
zl|fV?A<sPk?uceDMXrY44L@F)i0m!D(RwB|kRyPVY&>4u%_@AT!6x&{FEP1+ms<%n
z*?4|8;5qoYVWMkMmH_;jd4Lhi6C}7ApP$<?UKVz(tf{ZAwj_cA7|BNbpU2E@@eRHm
zn~-oJtF|<UDkMj|0kK|UQF<0Mv1L`_wf?qUJfQvj!H+cig#jc>%B{eZrGfJoAVx>*
zhNutQ0ZMvFe(!9t$T~z;dh*BLo#(u^J^`9KdqAngasDe_#RR_1&zsDtS+`@K{60>d
ze_F&h*bjWWgf8`d_ul9k|2~<Q(#X=4RXR-*7h;li_2ei&Ed97mTP@Cf9oTM>^E1wC
zLCM*PTj{<{LAr!#<T8nVm={`vnB=SE{D>ej9{8W?%^!E!=OaC~M<qi+$%R!72SGtc
z1?T-)7SGuCCMrU}JL#vwD&CtSw4hYFS8Hl$A=-2>L%MaP^$}i-CA$sb2c3}m?-C^A
zl1Wg`x1ggAFs)W3vR@uk?;3GIb3VhhZg3U!?0Q9TfjPXbg=s(b&beeRU7DW6=b@)4
zsg&zPBv=}-U@~|+{dtTGydn$;{<EOL$W$_2tg*>`{NY*B10({~;2{k@8+F+pi?aFI
zss;UOG*~oL7JTqiT@Pv5?X&h;eab8qQr+_#-*-8<d*%gL_1(~IWo7ibb7|@Nc&1*v
zxrNf<g2bd}Cdty-)65L$GY>n5lH-gVGR$0)+tW@$Bj;AC4F5a^XCm`&2Ra~KS3bch
zE4w6`evv&gC}}411=RFiQU|oq9bPunl3VA11*=qauC`84HhT^~OMgQ(9;0l$#*!Pw
zG_(GE#HO@r`>X-TEO~_U)@_Ty9vf<y+RC4Vk(5WK46G#lXE%BRAi3i6!T7-16egV=
z<UP71LuPB&qB*90{;R9#hy?AY+&gJha0y@x#9W22N)=(1MS+Q{^5@UA2A$91vioCo
z&uo7s$0|uu@}1|OlNBdr4@{9^;a8{L?3w#cQTXS8um9HWH7tcwSyQ4--Hy0ran&wP
zfKtDEDwg<LKE*+>0aZJ9Qs)!)ws80>Scj5k7uTv#H<GPQ(2<(bMp@BwDj+b*O{xok
zN{2=k{Pxco9|O`jZi2RwIyI-#%n@8j%?@<=D~`FO?yjM3eQ{RM*>fQ{M<Qn&O-N>c
zee&a~tKDI|02;`Su%onDBG0aYV!V>gy$%V4OJ|KX?ZbKx?*h=UD_vfCSdDAUKB;jt
zmlR}GoIF798idhj-UF+7p?bs%kf$=F<{k!a_oL388BPV;GHx+go=+}t@8y*=(jNE@
zgAP`RF9@`6Pkg*ygG^$*U|)Jt^JFUNc*+_2vx9e83EV?!-a(#5n8}SWlVUD4Ijy~M
z6VGw>WHGRH1@5!V4!dBXgYDo8B4HhWAW+<f)Z-y(H{1Fbi7?qEJe4f^hAkSHVhSSJ
zBX&CV4Hej#+ft-tUw3w26<kOhUe*w&zkfRVGzJO%&zf%$DQ`o(;`196C_D$R)Zi49
zDxU27r2pn2*ZlUsoB0xNkF9{`;YZ4qqCrM_?z<@}rfQXE3h;%w80<u?*^C3dHk&E%
zVC3$|qez{(B<ZOSGvT86j_f~E`2L1^y!McJ-*PkjDnXlV<}kC^ETxNA^|=y~i$4Vj
z_nt)65`{&@5M2wlc+=H%OxY^=X$JdqntVhmek2=m$8rU=wIz|dr^C1FEfOxW&aTCK
zpbf*psKkM(cx<hSJgGwfRZb42=J3<c2}0dZYR_!UkC{a@T3<!wAMug?Eu?6;7;KmJ
zTRu?w!<pXVv9?LclwSdMK$xit+mk6kp03bL%cKPe&&MvDRGaKixVX_zZsdBy@jan8
z(?m0Ud}`sDhx2ucc;yrR|5-!>hIc5l@=9JNYPO_2JsC(!&MCAtD<yQZ6hAB77gos`
z?<WDjO1<kUJP{oh-w!Q8+3P4^-tW>G3g(GrA4N;8_ckDUspRbpK5w&Lio!?v?<^%?
zE8GqCDjdw5pB#caKIpCLar;uf^8NXVDG)aOo`$5EGpjcbUu&&&_GgKFt~#4egLS$w
zSgHKNqRt^ny3*wLncCybneqLD>H(e8he+$wD^s>g_6xwJUpmQn$_wYh)4f5Pn<T8M
z=_J+>uKlkF(`OX}17rU6HDiO{yvLG7IdK8qUc40d?N8n)K^4FIE29cQyd|xLMo7hT
zEe)%O4apquuiAI2Qv%2=w_C7M{LN1&D!m`4dX-Dc3iR5+_)>o_^csk{PV?n+HFmVa
z<OA>UomaGP`VPvy{<JQ;u&QsJlVF)#D;61FIHB0;nzA{w>meO*JP)Bx>nINYfAGaQ
z2k;YRWF%twVBI-9<QauyGK@b|6&dF_e-Oqc_}+M|>fvWH#hFXV5%dsAa`V*a6DFBN
z<IS!FE%nvO&s~YuputVi7<i_|dhF7sQZHta+QuQP1g+7bO`>ark<P)Cnu;SU8931h
z^tcT?P;4qKH3G%B0mA2uL62L&^pPPQoo%2!6q~K}7<hmP=^u3~N;$~Dc}AdhPcpD{
z6U4g4O-H8*Lb(mZs%(MWA#wt+$^rHIv$#m1erBwe;7>pEN+TT|9@yJ(HW-Z+Uhyur
zz7yMUrrKi1z?=ULhouA>i%rBFA9YiR6DcW>`12%B`rmQ2F@Th-$C7DC%RIp_sgOKj
z2kHj@Z%O|hkt7rnYc~#)!NuEq?F+kVNTjy8X5ic|Ep-D3@A>SCNZ(TJR%u+t$xyj~
z#D|A#JEmr`mhR+r5nU&J2DC}+G!5~Ng;4Opre=z*&C9_FnQnOFEN7%gaDk@s<)|OT
zemY6gfbyK)7x$FRGN3rp4-@@`AwuHY)G*Jj$Awi-H<cDLG-E|n<trb}@ylsq{MH{v
z4FmD^w=!8DzQ06lkOb+qTKx_EFph4xi_t-kN(z(y5>+tRC<FhV_Sf}WiXgMbaMk5P
zJ7m3Xd-*cma+5X-_POrccg|-gJ|bySLqSI@yGZ5ibW=}GNg4BPe0!n&BvWSw4N^DT
zCg>D*DbRci$==oA)^@Bh+~IJ+rGDR*vP_t=7FcmgA+xHNbr;Dozai0d0nze}K)fb{
zhrVicvr|%!FWT_oe1`c~QiX;MlT>75>Or$e;!(bMtH$%H)0oexc*m<Fb&TH}9`823
zH#jV0%v8`czE8Em5L{lgApe!*4ype}RM#=QMeD`;Q*$q-ufhzc`FzgvX1JT*R;)-y
zP4f~?TXDp}5nQ?Sy-_N}VMe)dHL3~+b*Inu@x}1v)UWkT3u4vi9ZHIVuc0fi;-aJf
zu8f!-iF*r{5sZN;^F0qRc#`Qsr$p7-qksiflbpv)cKm-E{|0|N`rC}t4=Ez+P6%yT
zx%M>jp~edo50=VB7chLa{?NYthIx!W@T1z{!m;5A|1|w9iXFkYgKxy53?)h}mT=R}
zgaOAc5)CW<%#&B*?tI&T@;u(&_JpvBttZkB1Y+HuBY9H^M#He}V>F1+Q~)7bQJkV3
z&;Uzqe8#><0NO$f#=A7$$rgvxPzk!#E^PHZyo))Zu2ZTQ5v+}hi!&XfI!+4y84=J6
zH@pFxan@bkC+=vX-FzrVESu9esxr`bzXq>##PNbILri{DNcC9^gYhu-UO#Q^mOS$6
z3^?h^1V1&UHLpsSkfwNTP?*4;QtX42-~CP{<iOsHjLo6sY7j-OC9%5EzZXDHXq*-f
z-!fLv)S7DI0l$1(b-fy%g~8=1AbYotO^;g{7F23gZ0~BlqA7}A+)%#6+vDG!DDmsx
z3+4psu+-7v7nPKZA$=2jinWryH)d*)+r7$79nk`iqS3JNCL04&=X*Ng^Q>f&zycYX
ze9lGaqvNjCz9fs)e7k$Xa>+x-|M*I}<bx0Pl@O`8lHf7B{;i>r_p2cp5WWcXu)d5t
z*lppzgMEoWgP^^pj9Q~m{t8rHwgYH@s7`PQ|L(d)EOZPf7d%B<-GCaWS1xs!OI+eq
z*aj`^U&BZI?H^IK+$Ky~IcUJnLscww=`rFNGuZxCA}}ja+&Vb`?k>gGNT8qODXHY!
z{n5J7Kx?C=B_AE^-LsJIlVWhq#e+cQo=1FV0%wtPJB@?qQ^Eq5l55f#fm)d{Oyttb
zKjz3938=8oEv>88-0!Lr9W>}*tgod^&yNL4r7yE3`CtLq>dm^u<5a#8q4c=*uo%Za
z6nipt$pF*4p|D#9@UIi{3U;B<m<+*vGS8*qApEb%cF3EbZ&J<h?j3Cx`uYVA8>srL
z(kG{Es&Ll7tPxs(c1t@#s+`Sq{tTcBfu{Z>2BQU%JFASIvi$v_tFP`FkQMu&hBJQe
zi&pyy15j=w$O1{sue2RsnEW1Lb1J0fi<#e*%^ag;Uy1=|4vn)@-|Bpt7P8;DlzL<{
z&B$~%ZDVPIpp`lRw)v%M>6p7lkMU;#WKVN0DkkD~&&M|0qpqc0fy$m|eF}Hbv7ga3
z4>wJmDd5Goi4XI#J`si`lGKM8+xUNl6&(djlHhq0Q(xjR^XqdVni2hSfPx@dkDT^?
zY>A%CDxM#R*f2TUo>{ma*PT}#bih_Sz{`foJ1sJ2e?VLNCagT|DQ<>az^SL8J=3Zf
z_MoVy)e!LMQ4Y^(ezfJ-csj-Tog)3p3Nb45?TS1q;OMR8{hMOQD})0dJk_<aSoAZ-
z>n6sr?Za=7*S^7GkpT%dTD|3cq|!w9<G9@8i3^cxC%DJ&XN~5r?k7hiw%3$vA{Evf
zkGux+_-fb-M`DykYYGnRi1_B#O?%Nu<UcO$yPOQT#OsG;kCWH7i;e{0a4dx;Z&HjO
z@S;F}XAf<%TKs$b&8V3#Y2_jY7)~(Gg2ML4pdVqqjm7W9^hC?eji7=Men*E_8x(#l
zsSU#3$7bcXu9U{#btvz-15&Zc3O{ZE?KKAg&9OZ$nUk#7q<~uH)7b%bSo4tr@9K^=
zJm^5^W3jWl$xM$B3pk~s(RSZ{)M#P$F{%*q<x64##Puu82&9T73gVR>J*e+E-H?BU
zCFu)p@Gdgn>yiv6bj}t>y`YV^7DsDC%kScOvdUUVeURfqY=-6)vY+2gO5jyDu|(9W
zjz;-{!N4dmr5PBO^9ySwp&Ul*2jHGQM9Y=7$7SFa)FDHK3c!}pF!sVEdQ@+e^bvVY
z{zr6=q~;;x4@}D4kf+>rFkZpB=K1oph2bj=x|VdmWqHA!8daTxRGX627tUi=R51KT
z$POt;Yu>~rWQJZ9plvEErWAPX)Q)jMsVeI=iM12)NjPe*^noyFSop@Gq~FO3t+zcY
z7)%WQS**Z`O)yU88&Pzc_!Ej<tTd0iM{c4MCr@z=`STau$#Zs{;uH5lCa<7I*iF7@
ziOn4Rj|jD*;bR`O<i+KmpJJd?xy|r2gA55r`z2?gLYPe7(1e66*+c$+R>KSV4!5#Y
z8HpHU&+=CW8_bbScIpSf>SE$sJ^6C-{Pjn(61|#h1*Q<8k6e@BVemFikY+OC^s%=9
z3d%ie`)~n)-Rv+w>;*mtP+D<+1s%nF9OV#UfU(^;n-lPaQ5Y(?u~Do;hmRErP=y;9
zaA@Ph8|McQi})AKLwq%ox|3Tu+UvfBC@J`g^6098iRogK@6#9HLORkT!u*4cY$Q7u
z96#1E2$v!^ZZc8ar`#^L_@sNl<F~mZ?1&zP9-p$AN~#zoT)t^fti65|TS*Pj;MJzS
z1_-`RAr=bU5s12@=e$O~<E~uHkZJ4T$>N&%PM1z}ykWQ8R)1=N?=@87-C;g4@%!%w
zw~s&SzfxXP>30R8W9<R4h<+eRW>k^7TcPfmuGUh<Xz{_w7~s6;dvZ5>FJZ_&%PZ_y
zdxmO;;AOL0K@Cs|Y9MNU7b*0CPXh3t=3GdRIjKw}A^q~)NRMx5;+Ct45NQ%`N5#za
zE*mn6&^P=Wzq>SkhlTVij&Jx9xYoZrT*#Hn4iAMi(th=!F{0rK;bYlmf`E;Sm0>HK
zShj#H00nCkq#Ff??YjFF5|cTIwF)N=h8(LP5rDh%$^1E^MHnjz*GN{c_Q5+_nliAZ
zTCj4l!RE$Ur0>_eNZMDC<_d!QK^(px?n6dnx=CwkPOrkXKy)uqp@okPA^lqjo9SU`
z4=J>Lda*H70aJadx`qs53SC8<TLpXPY;AT2c7{4a#-Nr@C+}Y1vU7$B{pp=zQu?!W
z?66T@hj12YC2!3_H3J-TOiTkPM`?-DyuFZw*Z<RPMoDRUq%KL?aXVm=9_FD)BrHWR
zo+1ZePHY(SjriF-xcK_9c4us}@aI(e+r$R>hXDEg)fg7%-!AUwL%1!skT=qYUP4J@
z_ffRl#-C0#x<f{!KN{#zj^G)C%z-FS31C`A#E(Mku_&NHcZ7%_Op=jt@Z_)4TlBQy
zOmLT?6g10;!(?ZQ3z<@BGX$5et`EW_Q6*Yb7r8!OAd*GdS909rbVtPeZ++kE^i#+6
z*-okdaYa@oi??F>DC4|Np&a_g%%Skspe6`e%*uve_(%D>U)?KmSH7fLJw57Vv1}rg
z1@E~l90NKQKKqyspc1pJm6i^eHIBMmcuQy6KUgAXkAqno)%FRq%_+JqYWWWX4#es(
zM{cqap!x+SZi?j?1G8_0%l4?a?b9etz>%8JYXvj0-L_xLr}X03M)t>@^}LGRkSw#^
zrC<;~DKcew`g>zRvj0rkNXNg9w7L^7{!^Jfm3z`HvtqZ7JUWBQ7ohTJ)x=Mh?@ZrB
zdL0;%e}4@Er=&rHbNBUnM!puFnJcVQxMJT<c3J%7FsSnpz87}Ny+rd=<fy?xeO2?)
zu5<!hJArliH;Hs{z{txJIWDB-nYHx>vQA?0ZBI}eJKc(A699+6dSbpD9APQnF2Br<
z2Bi*PV#V5{kw6yHr2Ve{@N8?w_z|DuweVjxY^ukyQ=G>xqAm#DTypTLjFQ*TImDz(
zcejv)w9AYhwxp-_XJ~qwc(t0oQ1#HG^0-~&*EI@qW;b8s{*&GJJN3>XvZI(}WKPp7
z+GN8>pfW}20`dWoFUVi&+tT+Ei-Ogr%}ClV*=vvbIA`_FFqHoU%YBmv^o20^mxMJn
z&7E_Dh(k>f<##44so?`Jaz*jLljtsWC_~&EMIa)_$oL|F0cV<6=#8PrHJH{baK#F$
z{pa%rtE&g5Mc^fMIE8D1N~{KKL}j}Tc}x;f&w9%^7!I-4<W<vTO~bdqy#8_R*ogA5
zJC9ovhp>+8p3l+pD}<=E<YQnNV(<`wpA2#64+50w+v?4p;b&K3yW@>dvzdBb#76e*
zWA2N+7m&7jXnvTXc1oTS`o;T~3t3#EWmIHz9k#E0nRfdiVQ5#I$k?(E`011;yjW~U
zI-;nf2#*>sP6qchI1n<(aI`@~^*)8JTg9Upf`PR^*kE%fIO~m&|9I|uz~zP0f-m;F
zY9yg{gX*8S6JPOoE&=_P;QnJWJ89WQ)NSlZP4oW?5cSB-ip$Jg5ZwuGB&%09bFfC?
zs_EU^cmXZ}4tpfkhmIO#Xe3&ZSlHI`v2|7s%KsZ4BuC$gcBh~Q&@4nr%MzXC1x(nZ
z7p;q0x)3_?HhUbqkuF6<-%hW&)gTVM=Xwwaw?WL@TWB5{9>Z=FT0k*spu{3l!5<c@
z{|kF+0LjJ6*>3MRh{$J%8<Ea_(A4i9i=3=bYfQswans_W#|3+OT%fY-kUwr$+eRZ<
zt7y$gSKAA;a~#~$ygF?*<aAfOz5M>fh>C3SO;PO1Y+au)%>69ElRN#mU6gmPQvzNX
zz4KS|KcUOPDJfaoxSyU6wAAhQT4V4{NCX)ad1a1}U20{j%EWqQ{#ngh6}&vCjI1{`
z#M2{|Om>+3ptH&HBoXFtB~jtS@yiz6NmM+&Yhs04EQ;&2u#aU0a#whL{M1_dcK_Ih
z)SWwZ_!7Hvbw5#CXM=)4j!|C=&i2Ln1}gA-r8&w0E>n?LMbI;U?<~$>a1cD7zq@YP
zJ+^-`SCG%({Fjf_=*VSBMt8vT6z)bo{>8YFtdh48g7C!BYOS|Cm-!{?rlfK&IqaFZ
z2i3LTW(3TzQOjj@7d)lJuxpqxM}??rIgSHt{E}zip&L7}!2Y5^M~!mVsZgNzZ;9iC
z(2*{`Efh4E68|bTKlsK1)E0D4KOjT(8qGKL@E%kto-plkCai}Pn%B0qi?G`r@|bN1
zeZB@Rb`rUc9GP{`0HGE=K_$PlFvGKNfB8=HJy%tIy?dLG-Efu*5tAZ=OQ^k2O)x>R
zb4id>Xh?^!ZXkj~D5|aIrdv2&8mc=86jyUYa-CZ(^}UyFM@4Aql|T>KPex&3yN0td
z^Y&Jc$Oelk+wszRXd4RBks>6B8H>#dWOL~lBToxJ^#Fup$^fT(@Va6^qVzEIKZVMu
zP&B}DzqdKdy|1;{DEEP9*t6tfaZ7<+YSq2`^3t(o5d`ZVy&&Y`90uWm@x4@pxSd~i
zbCOdBPbJ*Y4S^j^S}Ad24@KH9uBj_a(VA&?<s^9sad9sj<qmRp4<&kNU@PWi&#hBD
zCfU8%E$!?>f<jmlCe$>NK}>P=TO!E?Z(v~o|Ls|KNBt=f98<-e!}xY!<^{qfTYcb<
zM*{>-vBQM8lAnYY2y4EmomDV~5GIni!ihv+tib!2Er&WBo$<~-L_9&k1gBeeVB5O<
z#i0T!%K#=Y@fI2B)(y0tL@po1b*{Xa9%^<8cHw-uWNNO@{W_mi0`>2WBh=-`FnI+>
zz_=2X-X4EnZ^N0$z%|rB<-j?Ny8A4pHp;`cq|sQhcDiCtYBQT=Y(QJG4-x8xW7f-P
zywAUj%=_S+w<Zr_QUv;Af42LxoEFOI$Y#b?S>aM$hU%M^(H0a=a<~jALgA{oWnUX$
zc1PGMgwC65EC|<&JMQ-?+%Cecvvgw@TA&~g@TBH7K_^Be<?WK5r;fkv%y1Ss1rk`c
zM{<A5v`q6C^R7sc;fk;E6JxLqYmDZ4)<LmtVLKdc#&RlE=#~ImXA_;a>Ew5D1&?LA
zF<M2VIq8=aBonh!N9_d~lF6@7YWc??YAfi2?|-pG0%`x8$TIsyx=6-$b8$z8=E^^z
zXW6j>opk?`f?jO9o7MVYgT@5l=FYk6Bu3U^gMR%db$dNR{^Kfr1ryWLJrbHOfvc0n
z$MSHvm4h$b=_I!yGLGTv$aGxPfyo669^I~?^?2Lf#G6OKTu1_D5N15u$dT=tK9013
zN)se~YkKL}EEA}?K3jFzjW+!I*+Lll2{(DZX@XCPv8!Mw3|Ol;s6s;!^Cek0BRww!
z4HJAlzN91cenlKz-tyD{sX0aOxbeax-HUbVo%ODt*zj9-&jLrIU*c;V2G~u}Zwb$9
z?UwTAT?;kjgiQ4#_td%bc_|Uut%_^FL8#CoPkSyKthtBoKi-_OIJ!M<S;L?1cL?CI
z!RwQDF|!b57(|np(PITCR^~Y_0!q)V=zl7t1Uzlwhc%qyuyJ7^)haN_Ke=%qAa4_l
zM=iWiy?F;k2B!Bh{A>+18k7Oq9hMLj8)C4{l=Hy={-=QhLh6nx|LV*|=5&zbahrs!
zX?np}o|hlRqBQn};6jqkjfUOPwZ?sL^d&m_RnY`ex84EcQp+gud$L)?c>&9+jk!MB
zj`RCH`Oi6@_VQ{Lgxfzx>&;x-bMx%XG_%EyZP@HfZGQ=$_)x|5(Me$?Ivg-#bJ{72
z!gDDQp^hsw6(&85Q~!MT(yz)V&93y-&s%clc?9eV@}T_d2?{GuVIY3Z^V%?T>r|H<
z97sdV6=jI?v9F9Mm_lvhq?4sws~~a_y{ok9kCNEJ)pj8!U#}Vu43OJQ@=7AFeAEGf
zlU*oTM!;?F1R$uYdj$&g7ndC#D?e&a)LwYA8K4S9>T30DP6fS3Ci*e<leXdYaC+tS
zhG3A>Tp$f1=6aasWjNoNi70pJ$g9i_WaJ(JjQVMlu@&N8qOcwkoeqANIXf(gecm~N
zUsSo2i11Yw4;ZT9QP+=-L1@9OIO}{YgcYSn>2b+#GC(gXI1Hq3`A+1jNMO^U40KfW
zw8`N#e`a?n;>TQ~vbPIGR8j&cp5l!`HCjvVIiJ3ll<9b;S;rOGMf8lYa4=7h{^m(v
z|7bKEi2E$V>RwA4RmdlcQ_nEfnAj*2hP>oa%)rWLJA(l<*ihy%xk~(>768C)rD306
zwqL<OhuoZ{r-Le+%6az9T-x|%^1C%%-DEXmHx&XIywJ_cDV}Y4mTPON=N2{I-v}1G
zI8NU=hhLETFz~=$GGk$}{l#fP1#o;cJpS`B^p$yOv2>b7Kxj|QxI)b2?R~iQYizbj
z>a*&Q1~Gk)u@`I-dDym!?0tpawQ=89-XW5ylSnq0M>`?KURU_tu(~b0V~1R^2V(tJ
z589m_q9TMCq=4s>`V~N#o9*jI3_yWus?*xNg(cw_`j9H^a%2M6W0;dPfg33X56s#V
zQ9*u&=G2Rh(2if*0V3~v@3^z*(hd9bI2||p4m>^y;02iC><uQPGVKeKtBbzI``E?<
z(^H~)jl+yPN@&riX?D=H<e`x55WITC2&|1|gF)AW{^ej|zoy9pOsVBUP??)4bSe=Z
z&5aCZLJJ4d9Ea+7hWgxRYk?+4nNgr`aFVX#&Pqq3Em-apd}n^}4%AblVtD~@wRw;)
zqMLb*G34Wh7>F&*Q1T?jryDx}gwfiBu)Rn^6lfnxxD2I{b?K;}Ho7T73IJ}x8Q;=_
zrgKkHF?W>?(Tf-edUhx!de-EId|dR0(&oz3kP^Te5cNU*_P`t;3N%2l4F_u-NLg%z
zW{a!aP)$E*q=S%___g}<^Uu9IJeFi&Y_Tt-X-*T{%&<NcCf?as;0wDx)GE^IZ~v~l
zO!OkxH-Z*Mzyk{zGKM&<dm5&R8|eudNzZ^{G5pDk8M4tQGt>Z%KhUYJFm-3JuAcaa
zH;~3QOs52(pIJ|gK*-P*q|Cv68HJ&B0WHamBW?|l-Kdt9m1L*Z1T062>UCQ+lmGQB
z+8iyb=(2(&j9~EN3SIS!d<2%q;SFep_JB2s!D7#hfgD(DG@Xyq@mF+4s~5Qil`<Mz
zD{TNa*ngLOw{d^mq1`?Hd9pU(d}!oOR8iGIC-@Y_wi1boNd-iM{vn5jjHWX@kx4A^
z0_T6Ns1UH&*>zgJd<8B43Hr=t&at49L#2lUOMjL?cuLv;<bnkOJUF~_1cS)NLEAha
z&j1htZD_}|LRuj{LPYpgJhjE6@3&<gG)r{$vZ`oWvY@RC8aB5(8K>t@S8t6$$O!!V
zRlMiXnP1W5a~%eBw@;{S%9SH%`mHD$0?l^$IHHZeM5{aF>-G0W_#*I>D=7UH_?Au?
z;WMXo03O+MR8}a{=iyu;&}}tR^?{-eQOR&s8{fRoIMcwN#k7q$xq@AjpVa_?SWC2e
zNk#hn`{~X(4g67NKU)@CCp_5Sxgt*4C_p)+cXoJ{y;5xGxhmyA@0iQ`!Y9#SD6Hl#
z+V^|bNyttOSrw|IV(W|>mMyo7e_5$mJ7;aTtG_X*(uhp@o?iZTO75y0;+!H4&K{Xa
z7ml_0c8gqO{Ll~02|>K7Tb2TiZ8|zMi~B%(@G+W4!VfPeGJByjV@g*1=_jv*iPo-S
zg-qa`s1-T!T7}#8kg$iX6A?|`V)>|3qh%v7lGsS6e}#jd#0il21Ibu8=t(qn4RL0_
zqj?06I2~joyNR#Xjh+b{!7sS(Zx#^eO_?Hna<({#^l{g2^N;JrypYNq(oR39fM)H=
zCqdi9Us|lz39#2*lytpXi^8mr)VwvXDH4bwxraDc`k4;4$M0u{+N^6U&h<Ma526tC
z7q?^D(k#4f$F|?YmZvrNxn1Ht{CfHC#?ygZhZE8DoO*4dw)@|kL8B?QKh9k8_r~3I
zA1@DPmV)Fdc_^>1iO+HLY{ivT-iCzEYy5C~4rLl9AG^P{a)`E(r&Y&N>zs%k%T&<_
z5?XUplox+8JT7{S(b|F6NkO;M-JxWjA9f~uGM6UdBZvv~H!<e(f?qE)VeP3926;Qp
z3q&E6^HQ>>dL_^jj(~#5IXu@7W}`UW=;y}QjV$}v8RAy()zXCq$@bD27<u#G54QKt
zZU^wf>RmpVk42Z7PtqhJ_%dM<&576!zg^gsckM%0y2?sz^_h4}>q@$S_*$dRl{5ur
zw<{}D!{MNAGdEm)gX)`k*k^neBR}^bzZV*rF|8A`cZ&!#9OuVbF2%s&`$pRcF$r1M
zb+PQ;AnU|Wf<E=a%_V;}V<Vr=GYTc^#Rk{r>s%O{sg)rW-x%kEuPSW3q#n0TOq?0p
zP#-7Z%YMWvVQO)97u)bPKfgN)b;`z}Yca#H{F<yOfSy`me*OuYWAC%*kAdA6L1m?v
zR1HsUS%>%)NyH<Mr%#Xfms<y;{n#{O5E(f_`xzO7{;RbOhlfXV%f9He*IyjO=se#&
zC5g0Q7w;QC_0ugW1hSvV{Vve;Ocm9p_)7cnOv%GD-oDtFi<CK2QC<&!Mk`1d|Jh2d
z)6zn76u2OdcS2UH?<ll5@MR&&O5o=@ZhX(}$G-j{OH2nhOs4hyZr_oBicaS}QkUkc
zuXVcNcwx0#gWA|dw!`O&FT!&Oy#i%JoIVvrm}*6s_w|5P>Ii$znpB>d+$jSeEyzqe
zIDZZjB*uE#CQ8@?fDh*tRs}o(NKi_(1RlYnAm(o(NaEXUuzq_XJhol_Cf&k@jMg{)
z{93eGmo~3WT;?S{?xCfc@n^Taz9~>he?oR&yZtEbNv^R)21T(s^-ybSwchN`^K&Cg
z36iu$?>6&#Owa;J=iqdgbr2ijPzT}iecK;xr{(sNpxZtt6a7|y$%01B6=_I7r_O`B
z@>zY>Koo0l?CkYbc8#n1d}xnuQ%p0Sn703fYF5gn&8M}mYaz#%=xZco7<>A!b}{d%
z3eu9(1g56;=}r0bq)Vk<LBw#jW=3|$aXvPhQ}IyxrXbt#MRt!!$Z9;;awp|M>bC-$
zL~3$=Q#awHDEPG*ZuN0+6s?pyR&Q7%$Zz{LhR8V?krK|La7*wDBlMAg$QkahTmfC~
zA#Vwb0`oAeI~a6uF7_JOUxqlUKPWk%sOVp-oAf2|11MZob5J*Ph<%)jUs_|(kjNEo
z5`Y=RpgrSE!@_nW4qumC^14U!DKBS4jSQPiBd!BjDP};hGbGqbbfZb)$r1_gG@Z62
z%jdwa2`(kxwnWvT#W8&aEQ|-Pr{^VxS!(U9?9>gb{ComaJUm#o>j;_Pi(7{~=RP0(
ziq#(xMy3vm|DymC3WO6yoRJQEc%>}Q^o*ZF*+~mlR!<7Qdai>gQ_@@T3F1!Q-BZZR
z1Gv9oBhkpof^{5PW-HG<g(^Ft=BoC>ROsBZz$9M4$d#t4cL5~#CdY}M7vUG}r%8Z!
zCiMV=AygqUK>APiK|)xXV8?#c(qF^myOaLwmE|o{s0DtpOTOq&Ww|G9dRVO_;gE3v
z7>bUhdI)~Ual#W_91V~>i(I(q6l7?eZmC<VVtyg>f=3uNk;F;HPT@vO1LWSr#LhN^
z4EY&$>J7^<1Er~0+Yu{Hs+3p0%2ZTU-Bv?_&v}L8jR%#J!gM{;NY{8Bw})2kX7+Rt
zsJV<>4&z5wB_}ev;&@~sL^4VATrhBR)5I0|X6ZYmV1QOa_RHSh%Mc!L2T|nAKC8bo
zQXC>_Y1Mx#{*=}J@_IpeeVAj7)#@KB^^<BG?xVAeb@1wS4uNMdNB0Vn?Wh(_KB7!9
zF6CLJhFuvN??w2ZmOO~?;YzdrtI!ZEvu2nNcJXx?D%*(fFY2d@=x8Pooz2lEer;r>
z=mfb{WQFX6>H^`!|FJ&^;)1ZUv@hgQJSs6CzBaDA`N`QQVxxAZxc~D(EZ}IZG0MA@
z%$Y2QI97(eGXV2{K$?2OA5rL{dI*z*lHd24BYgTVNd|FM{g#pL$Zh^f?3!ku+D1Am
zceL_nA~5^|C|Pjlf>?;n<8Eod4PqrB4<SK}wKMFkiE^}|OZovY1ko?(7^O%Ehk)mG
zA+)-V+C2*}CZhx6f;v)fY{l+^l>zH*wpk(`PnjYJsRv+07~<D9NQEbAP~0KWrse$M
z)?EKUcI3Ep(ueMW&FBENL0MCdvvD3r#kdmzMzFYGeMgW@x2|PV8DP}}<trI<{J#K_
zl>@G=dKeE3O_b}Bwq7TeVRKD$ik!Ijm1~2Tc_?<V2A;|zZ60etI&HGig!9J`TerWw
z*4{sAf5v^UZs}v->q4Do`&>{M*+_4}WMZ95XoVoThqcg*K>yhZGNd-4Lf0NbK6D;W
z<|4@CY@!R9I(PA6FFJ=b&`%S4#RL5-f2fLa;&sPVph{jksl9Qls~z<EtoDaK0$5Gg
zX{xyaZ7YC8oq11FC?ecF2V1l&w-ngXdQIx1$40WdlUbZ{WY6!Kn&+2JeIF_V<}`(^
z*@s<a@9PN0G04DV;72W6EM%{%qdeI*;G`LuX{=2#2Uc-Q;uf$p3V5c3D-S5Z@kX%o
zyU=WOyUZr{&jWy3elY>DJKwDJxBs<*I03>!;|{i${7_@rzkB`7+ceDkyY%e-N+8LR
zm24QDkAEYRu#Pv9ynN`6S5cFa<kw1@fBK>~1mku(-BGO}Mh9qA<E32rlDN`yrO`ab
zY4M;ey86T;CAutg)El)wAxD%mQZP`mIR@~*d?uw~`QOJ?S-LgyadS%S-f?!8_Xp*_
z3=zOquI-0<OE9qmLX>`Xo4tS|Pw0L0uT4Gj&|h&=kSQ53eW-{p?m?9X&au*69%;&F
znX|CNO&?6ZP^gY`dL>;1(IrKN?nn%r`&M`Xp0W|d+wuTeQwzn|&r|xmvg8x`y!7Pi
zZ5n~M3Qv?rvI;^ZsMDuu`2rF>W&c6b);WO9HmEo*W&m5J%0+QRM$AYTN8pSAY8NJD
z$i^NdCv(<jCnhY5#?vL7p{!#k*L<aWl_+1`r2xd^|L`vC@u)D9fc!zVRAl7ld*K4t
zdxE0{Gti*c&PXmqYMU5*2I8n2paioBum<(J#Ee>b9SxI3MZ_CF{z*;M^NlcvL2jwW
zGqrXt`&~#OCe*qOYQ#(vh~+>&aOdTt+lD~0+Cw~JQ~rrxI4c*arWfGsNVeQX#;xv%
zO5kN>w_reQmZB8e#YcB5B)uzE=qG^cOx70nz^M@01@_fi6rJ|i6^dMeX-BVf=RoeH
zR%F=KQ~&CQwt5DCdgXe_hx}J+4#ABL>U&KDK!X$;40MFrrV?X(?6=*OxDqc!2xIRN
z!N#6ooD0H6^i^{de~^P|A#3`V&$Y1ceQL?fa52S)qs{ZA<u6}i$3zn5QWrs#(PZlF
z5#L@CpwtqMfpPM0k%#=Z;q3KF*=$-5Kb0f`(t;m;mOhXsKIra?I7s(O&oWOQ>`M8W
z_20gc`Q>-=`<x%va_jzs($)xZcSX`-4gak(TmxCCpK3Z81Fo!PbB&@{W9-rdK)B5k
z)N)t^tjId$N8}C>HC0K;lg${ZQ>ku7musp7d)b7D8m&ZPmJ=oPTwjy%1jDtXItKTR
z^c#GdxYqefw6|>o{gBZePeZoh`cGw|0E5-<h<N-S7t!(Hi)+VNqLjYLSQ;e-oKu0+
zr}g1?mLS0Cn@6E5Y|tk0l5@U;D_PcC^bGF8h~cCha91oJCUUyy{&wjzx@FFm4Jde%
z4~BrPYo$(Eo>@WuoH4E)#l<U+Gq3!<FKB48Hqk$sXXS9-w7VpDoRS_rqE7V1_v-G#
zd7%~4#Jx?`MlGrHJLe_wXv4jAxWil)s=Jl%GoMw9R0PTG?Rw-{1k8+TT$-NmyaFCt
z8_j<i5+Q)%F&lQVPTY?V3XC}1XPxnEeKmz80orf0QgvYvr`pYbbzhrvb@L+;Q&(wo
z=Vz#S&&`LBr5?mxoU8rZQFcE=q`N9Mm=J&mFu6v4ftt1M3&!|wxfplS%)Q&dID*H<
zm0IocQuUq91T{xQwooLw+LS)fg8*H~g*gGK&6pUx@TtiJaqdyC(EeAO-T~=~z3-V1
zb$5jBd7eyWE%BLPGt8kfvbd+yL0cUFEbTu+PExqswm+)Ru2<(zEG1(Uyx69SOmCSm
zQI`>AACw+OAsu*k;7DR>%a3=LHNax5h(~<en6Q9chdg&-yl)+pBJQCxQ=l>jXY;s%
z$QW$_82dcVw?4$78!hR+^M6)V%B4t1^ZsSu|4FcTj+-bH`ZBUX(JQ$NXv4z~vRPIo
z3Ecf_24gLZt_AIDa3Um1)yAyn@&=p9>2c)cd#v(8dS5aHEk`R$AM(mqh6MleElY57
z-=Z|8Uge3YHXnOKq3NL&S^F-Ia;FV(Ir&7i9y)EuqPRqAzN4Lotdd!YeCyG>#H~RF
zb_Azwt|e2G#7izf9!~><ujt8mlzj&;fF4z1=su0Qnf=owoBA>#@FAgbJF-e-tZuzO
z_ms!jPWa+2iGLxRkHyYiSFo@@B5#&6b;z>G&#`OYy0-Y}EMhBKp=D`2r_g$4dSr&%
z>8r7Qo6?&7Tjgd}$<vlL=f&tmM+k$9(}YQ_?FyyhiuFHl(f!h+@)(kkMSqO`#PMqD
z;avSO`WIk<`Q1~L`chvQ&;m3zjh~pGU~9gLs;#3^GR|@LeDG!*!d!F_{KLrx!lb2>
zsiLmF`mPmIyr&QcKSXHJbNq}RcXVSvU^5R!Kn^|@wF^-!DLJ{t9B*NgrJhRG)InU&
z$x6`KjAQu5hN-KA=3Pq4ANDpYu|$Si(jG>-9{_b;S0Xvp)guP?AHqb=a8UgdSm81k
zB)Ok)&qR*Kal|7|aFXOyjz(-Cq$xN#d90#W0<=clhths@?>cpazt1=ZBRJ+#vSF~A
zs<?BsWWH^j)00QM-*IO1EzwtU6hJ|7(m*SbfneJkiccz;V@TZAC>N|Hjhh)LT55VI
zj7hI*!T1+uE;`1ueb9p0oUu?Hoy@mQ0cK)}3w!*sd4EQ2K6p#73}~^W?<4%g1LjPU
zd`Vn=^YaSnXRxvfO0Dvq$EjY`hd__3ced8%`Z_h~(|?nJT4p+do%_B@wW!=%#N)d(
z$j$14<DB)Xt<!1zo+giUSx*K?e&$bm%zMOLI%!NKCaTZRHC59Kr>dbP@rzo%VN>V%
zY_RZ5L~sx8##te@Em(smWe_!_|NYX`=1;P&Y13MPoA-?h4I89asz8$UCpLi(mXbFV
zo?>rt`k|Z|xEJD3aQNA7xMVjXB?)T-*-Eq^9{GYOR-N1OQ}@a%h{>2BjEI_GT){Ci
zqO8M@?N{%|x?#nWE^dy#{z2tSE~*dLTT^_*@_^(!0#11yA7yOd5UJUaacA7D<{RwG
z8mM}Ep+TFUdPzT3F7?XG0fhSL*_zv;DUbeWdQe4ZOiW&px<KpD)))(Tm#+od7hx3U
zD_ZZZbMuRU?}-!|zQiFPnhjGpEP9Jj8#tWcl-IgmzgdVj%z1T+9c<Y2)!-KQLpWcS
zjcp`ddBLxGF`HF+`SwG5C#Dj_NNq&ZvUJ;|c}$Io9;ssqf4U}cq(c+EiVA6W9xq<n
zU-zBZ(7nVecG5a8VT1Mhpso9yCizAn)S(_7??Y_ZJ|Vi4i;-uk^{Lh@ATx{}Q?ev1
z9!H+`uY!}wo@&^(&WIX1y7!3XGBtvOLt0L4e=n|CqQPs#Up(=XJ3DWf*WbP@WEj@t
z*->3&Sa?iV`n(f>-K~76h+cMpkN`MXCCLFi^8!9?e1(OdSB256IJQw1u4U%yEU;~u
zXyX_iG~q;K|0`Mm2r*ki(dWrr2<eCuND#l$s1cu+bF26a2#a0jrM<cMqKDqOA1+@K
z+@wUYsrHt=ar)PyJA%2y%LN$+ypq4YIaEC&1MfAh(gX#KBPy=8P)+Cgz5MCgi_203
zyd_4JSy_i#OV0x<_=F^R8JmB|O+LoIy9GQbqEAYXfWk#lwHr~HpGgT3Z2Swr#hLri
z?kv^Wp>q!vS?3Cp8<+<cgr!+h(qF;zkJnsLqlXcbGKm|EyCz)x;kjCsR9Aj+5e>C;
z16*pwE2};-x?$Qm^n^<ggFu6esah&{grG7Nk-e6A1rB9iu~vB(gl7HU$bNtCf8v6%
z>fe)-jcYPMOjrZA{U>yrSJt^hIT0##)RO7i$A#Gjq0+^nt79?};rOeRy;dM<tgLR8
zXxne-OQpns%0fkX>J^5@{|^0#oww_BQEpc9r9m?nVr|LxYbxQQ4X#7CkN3F&qXMCp
z|Db=c*-xF+T-p5Mj!z;LT}4%uM}TeM6ddWT7<VZConj`R_g{ZQDUia;%4f`1p;c_2
z-PrP2(vRrM)oh-%&Epqn9&~6bs(AA77pOuKRDeu+r4do$6shEjuAL6if(MK}uaVx(
z#o~a*X@|NijsOZo-DC3m##f~=iuD>W;OHESd~ig{vMgb{-6hMJ=k)C>4}qvtcHj^u
z2wxmoh}%n%`c*E1l9AoC7oO<S#s=}O_La`=I^E|6y+1&xC+(#Z-dr)F3*ED}JGze~
zBd&E`E}b~i2zr}T+?PaVtD`YZS;Dt3JZuv4F+H~YP_miiFY!~ECx)cHKA-})Q)3`2
z-gLNMaQ%AMV?qlMvccfxNz~&^Ac8R1W2g%(YsO?Y2ld+?Eq-Rh1U7${K6}&l`x0;D
zXhD;??p2YwySUgb>qjm(?lF4m*o@Ji*8KvBX`>Q2dm6KMov8@~<88m=e@NJSzJ&QV
zZtK4jo9DZ?5N4_nLBCeEeK$ru-O^@iXD0a>P|#>}2{O1JUCkI(i1SOg@mV$d^}|lr
zXsNcuzNR&@VJb?0csVl&h>)uy+?Y!~^~J<ik)RSDU|Gy>+By{ic#Ln*?O<hb!ibi5
zeXaL2emM{(Xkp3zb;933UiHG=H#h5GI)z|xj4|#IiADmAH9o~jW7Prmd8ax2el<f+
zfblx>174uwqg6oPim4y3cVOigX`7pxrJ>&{L%DV&QfFPGs8Bhl!6WG6Uzws+J=m7`
zGmIE#Z(Db0PvQqEC4&WsQ)3Eg@rBDbF4u6$sPTQ2+8A5_^R8I8Yl0XdY>ga?=Te{h
z$^XXV!tAuuI^F)!l~Ka#GeR?x3nP9aKJ#}fGeX;AJ8DW1MMAi3342|`c$6}Z#J864
z)WhYKkGX7$VUUB*bnka>h&*EjXZD9a5Zr}Pe_gReNR{e>vEA>s=*n`$hEE!8OGpBm
zCH?$W_i#Gx?R^5?f3AMrd+^6~s4xHT8NQ#j<6rG;>QyFN<@K7nIh>GM50PO(QnYt!
zNl)HbrgV68n{&_h)%Iuzel+9zK|6_UKeC*$OFO3{6>A9-8@-A3or!<2IWZsYGgq90
z`06%ZXnyNnW0!bzItdFRu&3`V*6LQ(ILNOyHHnU_JZ=orkIbU_h$1uu1u}<$FbMhh
zZw;%0DpvD*YyLAe`1$PN@)!)-im_)jXd3JMLTF@t6-Hm|jj>mBky*n#gdQUUUrHc%
z!3yl4WK~M)hZEo7wKI`pW7#&%G##xj)kuEZgtXI#HyBlWckuk1z^m~S?$S7NSTlB5
zWhT%-2f?^h-xJ)l`_Nn{IBJhmGdD>cATtAPeM@3C$3|$seYRS^{YbT;`uYclRih~}
zSb(GwwY{=eZL*$}cQzYc3qkuJo+h*_`1f=jV}Kms@$c6bpEU13Z^;G?y-R4ku;7*I
zRWsd~4<a>$0joxqPAM&bat3a<ho<?w6*a(2$`2|r)C(G*)-`7O_vYRBw=CQrYSK-H
zR_NN%156RoTGDiXVxp>o3n1z78+BTfCZ9A`n#{KJcfoDB#DDML+)iEkn=?A;ang!D
zT5gno(*nWK?5d)QYH+e4@F>8euK#ll+aPe9-8-Dj7h6=T7I?8SfPa7&!ejvO^IO;U
z30p$iD>2MBAC;cZXJdqA@qTE^;qsmgste-%Tl=NQC9ytTbee1ytbs@sNBW3qA{(AX
zQHd$U-hc*|ZJ3{LFu#0Ry!ty|`)y>0eW-aBsd5V)63^(4v!{6UIQJJ>iF}-B!D)t@
zP2wjiOpt>k58CHKCxEyKcHAc8cXOC?_o{U>gU=KCELx=Wn4GyCOY5wuqAxgf;%{Ci
zqQA_LP0?3+mX;^kDX?UBw%o?}4X%uynm2J;jg9~`k^(($02N_wyP1=h&7P=<_Qw3z
zWXsJK2<qh_gL_-0oR{@@$25qSWS3l|K}9w1g#*UPdm=VUHna=g`io0|-%b<KxSF=Z
zyIAM$X}PQUcqat)lQI17DiXerd8OqT`!#PQgX!WOB~Ppu+*QZ5!*Ii?mJL}zqGnm?
zjhg71NdR(I4w7ORa>l5(mN7J|K>AXU@sDU@2qOzhpJ{ANUh0NzA=w=y!J+>dZ7-Y>
zEd_sbD`a6+&CRGYiVG?VZ|Slf9&;1&dWb@&v-L$OFa!D0nX-g8D6-mpd*sXRJv1To
z{+aci6zdN4ZG(=GgT0x@;vhf&2d>cXZiu+~FnGHVeln0zj}?rf&U66?)aLl(-tzKA
z_GX5y7=6Qe+a4G1)RUTCi)RO!K6?w9kw9%KWpB>8=zE>rG?h?24Yb!vxk=#;h!o^+
zXT7i$fr$}gztF&_^aX(1{PjndZ3E8h^oizCo5$Mi)PL8fHUyd+I}!I=jyOZp5)a!Z
z08J|OrKOg1B;S<1LX`-{;jeKTz5?&!wLq5y&3(PXBL|Jn(`zO@(^)}W;SO08WFov$
zS;-9GUE(uHRua?S-;-$@&|Cj<e5<A%>TP+NG{@A6L_@Sz#HA(<@X!9l(2Al$pEaxP
zX88#1Vf-_uO`O61hqAYfilb}0b#Vwza3={GELaFm2X}W1!5VkhKmx%Xf;SGqg1ZEF
z8h3Yhw_QBX`<^q-H^$jx?|;x%)vH%6x#yf$5l}i=1$8yYjv_xwKOT?e$`B53CwFl=
zc8}=B(%FTR-wK{i{Vl#aY!egcD55IY^R$aL^>`$Uo4aGv(!F*gX?QpOUoXIlQH4Px
z4$Ot>ukY8JspkJfFw%jWeb0^NW(Y+bnH-Jw9$soS`&`y1wPqWcy=7hgTg<V3Yj7o*
zJ=l_L`q?e;kBk3r3BQ&%$pNN+DR~&(6~Cvc#ELfD<mVu1e9vhGlp!0~yMgAWC2^1J
zTaz}lfKp6Ik->X<H8Lkz{tksbn4>(g_G7wxzq70hewn6VQ?NMkUdR@NplTnV|IhR9
zXUv_v*UDwZuI{4MTGn*`YBmm$euP>(pD)Y~%W?HW*>)n5P0F@2D5?071bS(!>}b#;
zx?L*b@AjdNm_z{;ey>>Vh`s~T39Y6Z`W=nQ4%@6<Lb&;A*ll;El~;9q#y2z(3H;AK
z=zD4R-rt=xLT#OST7oko`l{KR7mZz#P4f5`_MhAJJ#CaSjEq#}(RWNSGT?e=2;uv)
zO@9n&>|0|@t|^x&n>~>TCB9F#@F5Xn^9UNe3>t4YTJEJ-oJ^APGlv|Z>j3+=*qp04
zSWWRYWxDc7-A}}r-om*sz`=gMNHMkZ?`xG31%oVD8~V=S^k??iui4EbFg$r4H~w;1
z+&cw!FZEQH5Kyq>HZ|DQ#~*`iR)fSY#J`{C?@WJk_&{}lKF_j*s9VsQW+vPNM+Mnf
z<_$wE(75FDJ{KY=yIlS<Ihe~cb`6pEInCnIqNdnj&lv{7zVk~!iCRx7=(nP$13we9
zL|%B*_J&VxJMAsbnZvSLRmW(kWwtma(1+HDx*9^eQv)>+%p0hzloTu%xbF;(N__Zo
z5f5b!cRyGdGSqE*S4DHAX>YpEspm(xI?t)o2?d^Uy1lJimYgUqKnuMl)j!V7zIQ0$
zQnBL{v#A^17vL@w_jazDQa?&u6<C&{Si|r#%+HhOYb1yR%`8M{Lu;*J4?#jvPfAs6
z7r7tXE`*lRr$OWID=2)>Bs;p_255yg5>ji0HMvgKm$&ojC`s#c)}lqQcMmu`nsRO2
zX^VK&W$M=IS>MKv=MlPiEHFUuTom^xS9U%pkrnGTY44M9D;*akeN{C(#!^s~V*}UO
z1zLO3W+FGGSJxv(D{xuaaO<UU1bGL?Fby8kF5{^~csGWsk-2tS#glaAt9{bSvTth@
zh^KPYEA7qxqLB)$;wgHM9~UW&e#4>5KmBFNQ5d8LfgALNH&jvm3IA69socg4LVUhD
z0Zuo>SZ?ALq(Rv+AJRy}`-aZ(0b;htDC8ewq!@z_$=4`p`?aAYeC@09cxTt=d7H8r
zaf~fnUH?^B=hyhb5UoW3=O?6GsrtTV&tpl!P%NS!K)58T%olyflc@tP${2R14oH5J
z)X~Op!yXQi(YkTu{IMlle*1H-eZQttR(ta{n51(D2i_Oz%?<{6aG8DM+=i_*LN73K
z<O_N01}_;<={9%GGBdsr%jhvf6~4-*`o1KbvYkT}P5}X)WPm*kUP|3BZ#eZ2h0ujA
z445_dkD5h+UqW1IX~3q&Oss^dd)@E73UscC=Zy8V6Rj4LrqDs3dJVauuq3|E@9Ww|
z!rV9eH?7DgmhFSzhZu+J)&mo)W2RlxjYa|KPm<4!u)KfS*7J}`J9~Oq*M|3`;VVhd
zgni$aG_}%FW5OpdH9KTBinnd?^}sGt&@NW1>S3z+a8=-8BJcC$TqH}!YBG@kD|QvG
z`^(?GuSql}4Utmab0;m8)bQ>bV*6MR!e)L2lkf}e4K@8HO@;Hw82>gj9S2K9x2jzB
zMPO*4J-0<yZU9E^gc(NfY^r3$Fzb*SZl_0i*IOT{;QF&#Vd1zZf?DHO!j&@}3d$l4
zSNgCHi_wIGQZ1%p>99KZc@C?|omW-87<5uxp9I>M4k?^a3tA>;KVKj4M2X|PLuMbi
z1K*8;hX;C_4`9423PN~nQwfh9mn6|B_!>{pPhS>@z><QM$6DkEKa<R1b*;rQIkmvn
z3Bg}A%v5Y47wD*y;c<$z<qY1i4He|keG%}+de1XJCJI6ui>M-1NBlWIozX_tD*Sn5
z67qk(<6yH7`kUsLxA{OI0@$qMP#|5o=5spL*TeDBJizu7KZFJSs`R^=G!PpY8fgnR
z6<K}9iLG9wa{1>0i%W&ny@Yjxqj~-^o4R$Bel0QMmgs_Ro&GDrvNcL%PG8nYW7Ff)
zEH2=zR1jJf);F%O;{Xiiz`j}Ayy{B*u=!o?7^W?b?vJ++%xtf)F%n#G`_IIrUm|mG
zy-J)cT(+O1@9A<Uws#Te4@GjQl>+DwYNqY0jFL%AC92lxay(wn^kV{rlUwM@dn}ke
zJ&j{FcU0NGex?_!K%SpfLaJ_JySn;k;_3lXLIZ1dKDx{~o!dyJYUGh<hQ|>~{>#Q!
z^m4wahm%v-mzCc6pATY%*<0WqFRbQEjlZiMCO^T^(%sV!ORwmAqSNA{uRS_*$DNVt
z!0PcM=o1$nw~0Zm63Br|=4lLbiWM{2h&rQ#%<5Ta_gTO|D-PG9^VL!4(?~IT5Ub*#
zpC6H&VU(m|(%+Xoz!C5@i9`Y0d%93%Td!)-@;yr>R2FCVh)BCsYB)SCJrUtuOhkKF
zW}2Ew_XEc>;fqd^Aa0bJGX_ZTeQKOXvr9KUnfp*@ZcB>B@%Ef{q3;G6N`a^&1z=+e
zw}1u2tYB@9L-NzjXMrc5#95Ht+pFa`N+1>96oz-zYe4@hF*)O1U4kdgG`ygBCU#<l
zw@#}bYqHUGs`!AK2WRK11-<B`a1*ixDEgTKWyg7T&Ob2f3<1KoHnqwc3VZlhQ}_tS
zurN~Po|X(X+RgKI&TL>roje|t&pnX+CRh^)+EMrTQIL=MaE28a!tN+tlAX%^0e$zQ
z0G7J04vq0NEx4-o9CSkwp(!FYdMcXx`L1nWw8ocqjqw?UJyps^TxEd>n0_dQnS<Cz
z>-TA2R<-lI%;$veiT-2F#e??!{-G~cC;bX`sP4z5vlJjpk;vV^c&w=B`>^LF&;yoA
zpZ0{r3-HgxV(I1|B^>dhERv`C52V_GX9V6Mxd*W1d@fcES7jM5bqt_NR{ICP1-17}
z2z;{CWm?{p4<%-Yx7UMN6;Lwx%<XYDTchrXT5weNt+@35WUGu7&42)(E7gWAz1qi^
z<0c4P*c?0Q=swLX0<8Bro*C(XG=C$@o$semltR0vN9^4d@c$hJALm6rb;%t+h3Wj+
zQ_&w*&M&jPmjl=aVNv)QrPWP@&faH$RJcRG4MqPoc=mhOp*<VmF-4A2BT#@=knu&W
z`ELelvVpL$`8d`o5^TxPGV5!vAomampC8<s^7D|_@&+%~zzeWSA`h|mpV}yV+yiAR
zJu29obcqV$Ob+uqxG7x9S}RsCj`yORjSV-bs4_XjT$3#YgM)sJ`wR!K-!BmI858PW
zY=LGA$)+w8oDb;?_rjXrnfOk|9mjpsYq*D-W4A#08p1Vb?iM3@|6ykXOb^fNa^TXF
z@oNZZ$62NE6CC<i!)po^f`aHn_qR-f0j6sCs!KsU<N-dS#KN-no&lTk?YWp3!ZhY_
z>k5OWCPENLp~7i2p|84XLtG0;S=3FK{Sixv!sd2KyE^%PyauLkuS0ulL(vM}0omxJ
z<m>4(QbkNnwMxy&z_Mx2oArr~<K?Q+g$_FOtS#m?>!jbiv6o`k8RoEFVe496j{PSr
z2yq0&ceiW43+)`mubHc-5hp$av6RWZ^hTj?g34@=2kg5vr{N>^%bI)@clyrkZVnR9
zD`<K@JPhS0ska^IPTFu0Ao8++JxpJM-xM{4Fp1XE0s+m><`*b>!sLB#AcK_FN*maS
z8ft3#wc)TBHQ$<~12|E@V(Fk6A+U%MQoua>&`H5GB$`uVdc<6Qb(sOqiJSmF<#@tK
z7OO*Yp9K=JfvtGtS$f8>=j&}Y;X(tB4!h*A@U=t|0`fUrfT7YyNfg+!ujz0C-;e}k
z<N7v|%FIc}B_!o{H3173-x^vzzxumla(LdJHMSU_t$5%$lXx3hhafY@(n3Jig4_oi
z^T!R*h#en2LXGL2j1FQ0TJ#Utp@W>NOdJD9PSa+JRE%IMIN_z-KMbQM=Qs6Fg!WCY
z^@rL0aKif!9T~ln9j|%DN6ItK(5r>te1I@jDC3-}r0E|o1>-8-^fV#U$Nc+lfzQjg
zj)3fj97{9MTR|0G4T0?tr@+?uU2Mk}ZFK)`HdOZH9I<b~b8t1B6V;S3MEwsCw1CyX
zzZTH=g>iT_3b7@hO^B7cJ&+>9CsA&pMy0QYm@omM1Z)0)KaqEL{324T%G>XRg_4c`
zbHZ}#=d&LSuR}SJN-$qyis?1{l4$bSh;xS$6dY#)kb19}stoN%^(~Lu<29g4P{T~m
z`|t$-DUFS`_HryhSkNCu_s=+PKR53QM)wi+2Gs6hg(l4H!+Bh8phGLYz7s78P|->z
z`0-K%=wpq2L6c4p2{dmNls+5}O@$H@aMx=f78V0T0bndv08-eLl?~P#zG@uZl6-ET
z7b#FRDQE~++9?utg{d9Xkzr{Fk=nOhHrfbyHtk@YAA)C@g8cFX^3|=mjPJsWLik#m
zK6ZVP7e%mL%(b;UnypM3sRe6QdTd;O^&3<7`}3DQ{FS0!y&vyW=L*9*8B{8g4nt4K
zwcC+EPd;OzVKeemmm`v{+}GH2d3uR6DZA(s_chW=UMCnrb6Zt7bMpCbp<K83?>c3?
zkE+^9W!rKEU5|=;9N}nJlo`+5l<nGsRlNknZb$tFppAb@6_GA}i}G#dS#%K~*A*BH
za%HreJ-26@8H}73GLdL$+JeM5upx-0-RHeNkBQ(OjoB|exRSeTtdRYrZGps<OQ*<f
zCfVMTRCo>BP0k5K?~;=8-fAigms30fQ8#pHq%&?~gF+4t*7tHOvc0lW9n<kF6m+SA
z!DWnk4<=4PuKl`K!mqav|3hw-Yxa-SHemDrO7{C4J93?`f@XX{4=SH%d3(f(qHIY;
z*+jiFI*C63fhcnHy1Vb%{#C$yi%vo!;2-97KiZEC8zPSb%&>HdC=2ArU96*1+Q*<J
z8C_u!xd>Q+#w*r?L?jLTjPvi$MdHe+c1%q^10UxYa|3M;rTHYj*W~j?XK<q(yF7C?
z948%pP|iUE7xXbegx(A6aY2buOW)7cdyzv-=DX1517d{SHntBfc>XxUVk9UF`aZ;i
zHDz_x#);5zU;O{cU(t@KqXoeBO2Q0yU)k(jU-V>lVGT=5PS?zSz+GNXWLx4hUQ<ih
zP5ki2d^Do(r_^<js9#oZ;NNo7tY2%cexK|9aZq~%yS5$Q%SY^|#3R>`eMmPkGUFk5
zZJUt{DqjvIM2edN6`8^o9JTFxr)LWOS9@N6MF|7Xzbv^u)T!`OB~ZhMrYMW1O0(6J
z0gI|EgJv%G0h7LmbuV%q<xyAOyfr;rVtnX<y7EgJPA{nTUqklDKeP_;ma7(;;cWiE
zb&K#!c0p)9&&PJ;AT<-C{%z=WR|IKZ;d<Pkkrx!jKsvM5X?6#F7b0iQ#;0Nnoqnhz
zlbAt9A?5F1Y|BI(YSDAN4n}o<x-8HNB~tk%aRQXx(A@n4i3kbH&ApVI5K!PezPWd>
zG+^Zb0cuO)Qh^Xbbojm7wN+%n0MSN2)BF|}V4Ch%yMgL507LIWgb!yK(x*Ok?|zJ6
zWm1zxsVh9|(`VfDOl{o--RxK^wbM}Ww|H@!mlEFKwxO*boj>ei*(H1zh|`8qn1gQ8
zvza{N(;nRW((ERiflBFz*kr*;%X@!fw7ze$x(ik|`GN^B*6&d1X9K2?0DlOd8<OEO
zL=8~U>=rgCIdw@<aQ|9SJzN83^-PaB^-q96IABNU_MHfdKD1RuL^7=jd#8|S+N_vj
zgR%1@mU7Mu(VsGrpptk~$Gm#8u(ORHZH;~?J=2mlS9s|GArKtI2wk<Fha<rm{25OE
z9BY3ci)@i;Z!gzzmCp<@j__B%dIQ0u*Q{IKkt9y|vvSnE_!yvbMMHj-(a*5gzhr<E
zuEPfDP<xlO=BW4@iL*YMoxsB?)?`wfJ`pyVCoc@|X+=HGcy!hrEmI5%gHubxr-^&V
z)qq(18>1M$h2k=jE&_=qkX%mMVraLJ3Uwz|9`^AU`vgb+s!u?ZV5D*DO15|ptrW##
zd8~rvEKlKEWKiK7NYY@3Dg&PLU-^1;#j@+czu6H4y2>dU^^YQC6ljs%Pl`zKVvQn<
z+6xI`K)IOdvy@#-r(9O9!6f1Kr@sqa`eQa3mi!jiH4z&ZRE~HkU^F<q8T6+fx6?&;
zN)UX?cD6V|r;%bqLa>*gSXw0S#F4bYD*-yvO+n#GUopFRN`{*yg6wi%&swrD#}$*j
zB#RVMq<QeC(AKs{0WK(!;FiQm#CPATUrd?=xDW$CmRy2KA3#-noy0`tivJ;M8t}B*
z1D@7@Oj{Dk^UF2ebasMY%Ll8JNySvfwq)xj7CMX&myk$b={LGLe{MJ6Y^1)<F&16?
zT}B0QVHpQzqnZzmU!4QPUewBfzN6=xa>Dy|+fbQiJ<Ao;x2#Q9A9ZbEs{C;yQoQc)
zyA?W>Ss=7Os31!me3+l!UoLT%GCYpS$N9_1(P~8kLgc4k#YZQ|fFKCGj0fq$YXK`1
z-mG4hz|Y3|bPd_j5EY5jaRNj6hJ)&TqFN!V#Zjt7I=s{s^G;Eg-|Nv{IZlMQkHYPS
z^(FD!z`o!%mVgbL5=}UkM9*L+bM}OQ2)aDky~>6cG6M&t<VyLgp!6Ck4)H7Vw~v8#
zg?t=DP?y%si9mb=i{<W-cb1|@ltOnw^ju`Fs>^;V&o1JD*oyrtcy$ZNJ2k#(FdpJM
zn=O>&)?I!czi!D;UBAH!v@vZIWdQ~%AHjFGTG{L9%sdvAj*}f{OE4qY!L8>3{UgPh
z^+R6;YqdJF$a}hm-V!Rrsug#iB9*sA?d&L3o%2TVP4B`@<z6tLYkultrXTukgN+#H
zvJ>SAWZT(=S+4;|V{Mz{6cO}##>jpAc<t8d=&l>j-SRnoS~Kt`KCBG)UZ<;m6B|p9
zZcVR5{*+MWml-y;b#rq&VdE8?#-qgsr2|o8VT=96(hrgW`w#;q%Sj+0)%sTl_r(G=
z`t%5FIr?1<azijJ`4eIL6CfQme$)s2VMynTc%M8%RPo6jeO0P{gZmmtBm;EPs<^dZ
z<;>n8i{ZXT0n@#|oOaseyZeT2JR0GB^&`7F1Mf$~juGee*L>_Xv!^|?qg)H`x0Iri
zxNeTVem-jqg@wph_eq}R$)7qUN`!cdPZtn(Ku?<ezwUXLW_P{G`36SQ=3<bxEUK1O
z+{u75WI!J}rwWIin_nI-q`)C37wFz5O7+xC?_7lr<tCO+2ry|FIa#e*->YjEEGyWN
z<x{LOKgG8(y~46un{C)iX&;SX705La&WB^VS~IH3@WD-)`vHW5#y-}ZlsXyppQOL+
zv1&LQc65iJnhg%wqY3&dud#~p2u4<+>#y}Hhyv0Ja!lxdBE5Hu>3m=%5R2A7ZQJ_k
z5_(4eJFIKME{=D=2cxU0_7_R|T|=sqBjt@2!IW%}w$L<8S9Yz&5!?k%zTaMUIc#}(
zD?**G0cQZFBhjKKP@IbtuA%(%o81yjJePSKYIbU=O8W3M8|a}IXJ1U-UZI4*nN_>k
zH|&S9Iyqps4DJh%aUdXKUCtFh(biu<zBK@eU2A%HqUyEZyxj|~R}!^7YR-k%%rr+K
z-<GGpl5<)chYEf$P1yYWxQo&F+4^#tOB1P~hXG=s7<jpb)u9%jC+E?gmp#A!JGuBO
z0^jgF09IhU<|m$#Bc*f&3}+!|vd7clGed$Oh0qaP@_EC4NaM|^j5+^pzD-PvLx62X
zrgVyxol5oC{lw{pmSE~eEQY3<&P<jOBBxJ>Uqx=&e3XhF`p#GDFS^9m=s>FFsq3p)
zx?X^?E)WGQ(`SRUgYXU#i@gW9?(Ct2Zh?V#46#!<acsvEg+6a06D(V?LtGYA^-5Rg
zhJD`!r=SmTIq@FjT?;DV^5dZI@LDfXMrb<|LJuZX^=l2I1v+6S6`WCnKIbsMwu=&!
zG>f_6lx@Jz`RfdIbq_q*<s3?p=zVT+V#)SNYx!IllGc=1ip;MG>tQ@*X9Ww!*iiyj
zEhcfs3QtCl(zMKbsXPq>>$cD$YvMEfoqmgI>#X54Yhsp+#)&^+{hW*mf%O7z0iTjt
z!9_ZObEOx;7NV_2Hwz;iP4uWcmHZ-NO2U+ObGB94=xynMYzMB6_vz|mS7{6}ti1?8
z#yVqzEDdSHUsJO`wWTd;4@qXiSdiJNokB#9QCPJ`jF|A2qSEE}HyY-<tx4Udn*;97
zFkZc?atwa!3lcuI12#cc30nZ-EB#P`pRq;pIV>R)R<eKE&tt^rS$k3ILzhLviFEMS
z$<?i9F&%bsJ+qiqW5imTIc4=5lsE)BsbvQbo7adJZ-A~DYAH9npB@;^F~4`Qyb~7u
zCci<z$V|zs;^#F>==SL&>SNov>1L4l4^0%*UuCe(aGucho>b;KvK02$!@kTCMreD2
z=Z0{y-~2GQoEfc{&Jt9FTXWDefO_r-wE+CsN^c6~`{=m<xCX!W?~1w=fqqBp=YRV8
zdv8VCAYZO&wa>>Q59t1Z?t}uNE7&<;ci2yLfFEBY+k${NSnYR2x^S^bAdqBoteUhR
zb+@<h%hZ`yzRM4h{u+KSt$^9mCoE=#cZUo!;H2DWaTUuh@+;QPyN#+I>`k8yu*00u
zC#eV6k&|~fEeU?K00EWJvfrU{&LO;SFZBNvN-I{{72IJIj}uHsro#gmu%U8ffon<v
za>gLxl-ITqY4R=1+ItfYbN9y*&fj30XG-E!>k_)uORE7MVQlrL55rQJ7MDfAk-|%8
z?UT`-?eEHk(RDcs1}uV`UrjIt9U_i0p9FS(vLak$Ga^+5f(x(95nEG~VA&w7=ES99
z^M)?{Ji_?{hg|;Mea&&cDZMSW4SG+)G;81?5<Bu?gIjC>MWjTAb8kXxnBfx#!utFq
zAoqY=K4}nutY$r@&;ys;MKah~_v1Wu)`QwUBWcEt#Ce78YhfyURV2T`eWTu!WM*OD
zo2Qe1L5yMd0z#W2!mg(c%>o~6Vzb^fBPC(QJH1_+1Eg%6C?q<mDi&uf$w~o2%7&f@
zPoFBS`$ip`gG-7ijH}qRO_RCh-M8Lef(R1>30EC$6?`ntv_hPO<sY872|j%0lj9bj
z$Jytg3Q0@nSQFVsftJId&@Yb#gw|s#qJF0Mmhm&!!6(VQVxePS@4wRIRe?rw@1VBh
zOSCO5aRff?!fiNfe|wyN_oU(Th50YLVMio=jrFJY@8^B_C(e;{h;*rvUDGN8>5d;(
z<B;`CV$)CDq7Y;rS-E~@cw3HfAMz%#j#V^XU`tvq<LGW&F^*wSm53nQMQcO9=GlhM
zlPX8RH=OEenM<wzN4w11`%)1xSnN98$NaeDp_WA+1o?Q<P{M9LmSeAL6zGUS+&h_1
zsFBFhfN%9~)UQ8o)nycYf@+CFYI21C>^w{h5ABXS1XB+A2<*A$t#bM#Q?(A+_Zs9o
zQ{6lU-v)3cHkli3y1KYGbD3;$IutDo9uKiVWMjcG)Y4c#BoZuuwtq8<zMwB{L8XiO
zvE;qyOdK|lNxPFQJx{sIXx<l?`|-3F*Al(te|IcnoCC|yQ9HilF~=}M?50PIoBI8g
z8GruD<#>P8YuX%|jJ9b3fN~iX^VNQF{qHL}=;fzm1E-HgMvd}Yq2WJkT?ft6=A02I
zvlU_hZw}_l+q6(y^cLIqOqaW`n;kM-f)?_B4BYyqgcZd49SnT28dE04E5iqe8Cg^$
z^0gQ!<zwU>TgAE1=QHs76XpY+<dL$H_mf9EbyUucyL<?I8+WLK4;l=_m$?&doGC2d
z*3MD5$(6?M9@3UU^#|jE%pMtlUVS`5g0}gi1A0KWr2PpIPp;ouZ<!ydB^D8qk2JAg
zg(ieW3eY^wgyMs+wawPIDp2PDRdE{hkE#fMLnC~Pqj(%+CbM+2`X&xeretJ|9N7;K
zsp`XsA6)PkVZs93=ab=^Z<>tTFlk0RZ5*WigM!v|{A2rzZ@JZ#crI<M(O|*=0~jxp
zGYOA`)i$2u1{yk?^b@m<r1V)-+DCGm6NqxYT~7JDlqGN;B^z>y^-W6*4jtjBe6bc2
z8ryc(n?A%#=V`_}$}GZwcdk8m>(pYI+F1UzNKo|$&cL6IjWdu0n34lLMV)7efcJf_
z(nj<i0Z`rEB!UnhM?J<ol%H8NF-_a5i<Fje<ph!TJI_xh2=%c}$<AVeD_rpQ)$!0q
zmpaZW<$=&Ua)A;aJeRR+Ohm7^aPv~N-v0D0Z?eXhf5D3)AUIjT9TzKBIO6*}sG|3!
z+`iCU`N`hc*32&fK={37+q8d%+~PquS>p>A$i&C0=$#czIyL@q@ap`Ok*lF>4=O3_
zV+O)vY=qDzG;p!$fIU3wSfGqR++Qdz^{TQ#&nZrXqIA$N#kSFh14AAIIFCSEujvcE
zw!!w$Qmx`1{xQP+?U60a)pxY#0VGI_%}Vf-P`h`1Z2d4FLo+dn+zQdhh~j+U5CL-N
z3ry|}GPp_3(&U)Eq@D)@K*&WtqP`fRret4GI!ko@YyW}$VZMOqK$TRG3>-)}5MA&?
zoeu!u%HdQ)=sQ<myF&i!1#pdyW<5o({VyXB6WZKlQ*?r<gEjSmh1a9AjWUkHlm2pM
zfdOll5{-32h48tQ9#R!cO%|A!I^VUeCEU!^<Z|;h>buP2$ADFyvg@}cljE<vuyhI(
zq<QOlRuY05G0-#0A}ptGBQ0JO@(g)$x$V|ybG$1_H}quRHELi~9r>t1J2WjqH(WPS
zm~8N)zqbJ)G5J0UJ*B8o{ama7n(agMP8HjB`X`(=6p`~7ZtwEw7dzo{d)5SDQ0w#3
z7ft?q#h>l5)IDgIPCxrFeh?M#!-8EB=j=adX(KZO;c{!ly45=P7wJXUA~~)VNTKwc
zt}PO9zL+<G4j3egN`1q}qBD{uf+(MYr$orw6uTiG^MVc5+*4tCd!!f#pOC!8->%0G
zyJzE!1SlN}brXy;{S3T@-(}<~-vj1Z#07#0#0jdPTw33%X5fw+zIon$Y=e!%RNkat
zu!*UBkG%*CW~@Pa!Wa&k7He`dUy|}3+l`>?{35+1Ovp0-x=77!FWvxEA$U|AOxWBV
z1`|8je>R^!2ysY1uV4H}e}sh@mS%DP1BMa;LUoHt%zjtOfhBmavPb|AD*##S9|7vG
zN_GqWf`c3jerVvAnBCdCmmkea>6i#w!3pjHq^Xyz7!EY$lb~GuaGgdC%--;t-plv3
zm}BrohTxM4mu(a48bu4&``rP;(Xehynzf(0U85VJVgOk}GM`(V#ZS0Bpwe=*qRoN$
zZ-d||)&rfxnCUHG4$w(E()R+8kB>md5JFPuE0Mb2DHv0)sMl$TY1$;QI@*a_tj_Xl
z=mw|~;yg^w$)?P&KO%1W<<jo+7*2B^B>}*j?9_xi+5G$WE#cWw$kP-P03pF04k{$k
zP4@j2L9XEV<0fJ2H||>c%e#D<LZ=OyIAcq!_f|IBE~oB9(AWy)a<MN&g1UR}=CX#J
z?*T9IrT$h+<+}z%Ca#jkL|jD9_LshNM_V*!APhWt&eRdGySt%18{1jUYRH#@V9gPT
z?*5|;R!)TuH1O~_uv0D~JOzhz=SLWvyPAUz6lN1qcp4Ci(lmd;!7hr+@IECRjgA7h
zXn~*nQqhph7fDx!9Q!zjjtb<3i{jb=RM(;&uz~kv5xj=s)z;ax35%hEGzW7l9)nPO
z4vC=1?ty!Ai~?@gUoMR$mwOA|T#%B8Y)dSy66<aoegj07>l3cdUmO$96yx{HotLno
zQ5Mhl$w2lu3!ReV6Ri@K00)cBFf~`x{pY=|+p(>ICihU88A;ctAOcZb*HhcX_SU1|
ziwmbC`?3>tMPZ`AE&h9q@0sW>|I#7JJ9q2RPNwPw>NGb)bQG~&Q4>8_JfGp)bBB$)
zDX+89Kgo{`>%?6XrhJPzc%gy1z)P2x*%BaL$OFsba>LfV^IsyDEUs%O_S!|#!-+;_
zg}9j#Zh>rp>du=MH33ylxl6x~vedbDEz{p%;z8ez>28g!a{O2`zF~-iZH_{XJ2=gO
z>=!a5OsAPi2EHmxm=>T&jS6E&Uw6L0S^4|u;pn^#b~c~>Tb%!sJ8G&$&+}Yt@tZZr
zUvzRX@gqhQ5(jk3Lv2)9M~L=*MI;lwU>w|Yb2l00g1-JlQ7=|B_qj+ond0yU53xwl
zVFxOWDYBb1u_yNwyxTLH9Vij0&S^nE4e<%_u-D1{Gra75ou*i2RFPgpOLF*&9S?`u
zIoTv^8yf6FhPU~|vXo+%qFhOn*{INjc}LM<%^>NY+cHQ<-Z62DU1BJ2<o!_aO-b+?
zt)ZaLn<(t<3R5~e9?~*HtvDZM#7iLAku5DKlip6<Pv0t&GdOp(pMmL8u)PP6x`7g6
z6`=t!aI+D`Us0hj&Mkz#POBsfxb4biv`cLjc><7H5;|3@kN6x87kDJvQAL~er;~1h
zr7L^@2l0CoKs$*TX}Nqi2rJVZ=msCec?jFAh4%1U%D^&`WN3QMmsWnevr7pY@$KI?
z@S}tMYpR~9D!BY}mlT@3*YFg9(F~9DisvLvSJ?3L5+Y}Sjw$D;z8<8Ge2!#jR4?6l
zF6b^%onacs*1(h$;?#enbm9#V9f6lRAqBl5^ZVKwI{{^YxB^(mU^6gkOBp_7pN_Rl
z?cdZSCe-wMBt0ZUIS>m#?}{WZ{~^yloFJf7B-tuygu7Z*J8d34AqS8z!dQjnzijlr
zQU0v{7ac3)f(iyyy%u1Mkza5Q!jY5f^TcHPXb#R~reoUlJAzv#g!9-HWELs+{eNc5
zowxD|-{~Q|qnE(J!ZUk7BnNAK8pG>*A<r;#XV0LNE3DuU+IM^z<dcc~BFz6m&ep&u
z-<}(p938>7;vrnTj&!`GsW!|9lio{?Juz#mNE|K+X=6Ro!QREI2>%dZ8vpEV9)*OF
zaJ)vm0PgvQDj>{f!F5Td%~c_ARt^I!1I}?px>U?9%KYxRLs?tBj%<_P>@3uMs)7G4
z7M;(wBl7|xZqM>tk``V6Hx^9>?_qx$>AONf3fJ&W!$#wL6!Kk7#_I{^A{%5>aGLov
zhJEx)j1LpVbs_cyC&J|bh%l3hu4<*jYDdih3}(gjgxUmmL5i(5zfvWPkc`iji0EKS
zeKl7<i|<#EGUZ4NAkG%o?KzeYB+#VB0IoKi^k}t#hOdr}`S-tY495%6PGLxI%)pus
zexy4*)tUg%1{$nWwf%_#NMKT!4jV9LrU(nqD+ZtdAaM@>h9flH1F6#ENuOIN1R$hW
z^XkuPmN`Xd=&^AL_ag|UL&fILMrlI6ArbizfYE9m1{JV<<e7)t$ix`~41SnXuFkI(
zifwumk2Y`T&n73Hm9HB=L{Of_lSG%~y%~&R&zLl=?tbDvV@rpv=koUxTmej0nG|8#
z_Jy?^p_`^VJgYw06r_0&>d+r|i`M$K{-2lT2LOZ==P$ymbMczZ>42&TFy|Hhh#y%8
zP$zrx`v4(l=y)@#?5ROtdElRB2$aPg`z>83+upr#Ey8^Px1o|gvOxZR%)v)R1)$Y~
z91v^~Y-l&X$oDaavh~GdkENRD=0JG^3fVLAmJNxpBOI1$cCLlG1$f-lGQB$BVx)rY
z_qLLRhx0ZEgJ9!*#TA1VGcISk+yYSnf6N=jU-XcRctt#r?jC%y%<*gRS^|8i<-gpy
zGai>Dz$xDPyH0B~lq{qBG<3)BM+17!kQ|bzIK~wHk>QG*>n5<W!O&LH(~9HW-0z`$
zT>Ir#s3#ndmm1<anjq)v3%PQD_-jocdmWiLDS5?vis>k#-mcXlI>YR!EvDPOkoPdn
zFA=D^T>2W~=pnV|=wble)shx(YYdTdB2Gn#(H0*u?k7t>a?v>Whds#aA$4|7c>=YC
zvpQtR76_dkQ_~*h;CG97`B$288>|S5p)DjfEumAiZ=#(l+s2kLKX!|sDl?AX)C>iE
ztwjaI?l=`c-Uw}T3v5qDk&&Z4m^krQ9Ww=)m?L(svSB_X)|f#a-K<{T&cT11dtSFV
z2Cw~DQ??zm;&~g@V~i%OQ_xiq1rVTM<cJrUk%Tm5gloy|VYvq0K#cRwfiTa8WgWj{
zC1An1wwmdhULkH*`@a78N}u;S(l0!tvs1v+#+11K)J1FAjUTwjn2TaH52lyuJFUQ!
z(d;h2;U;-^-_1#SAhNpj`w!kJ+!e0%JD_LU4`iMo7Eetp$<*k6`<mCl`YEHq?g(d9
zDq^IO4Hs-=|EW5u7%@ETz-Z&}BangDk3aeB?b&})&%^wMwFT?mf1}xvB%BF-<TYGF
zz$44RyM0D<`bq)@Z=|Uoz`zFY^ov>L>NC$5wL{YZaTBGIUsr1!aSg}v*}6YgU^h0i
zgD`+_KlVpWmm^Am=AF)&dtqov82z{-!L56bA=tq&p(3#e+q0D~mu>39@YwV$`*VZX
zqNCwY0-)8oOw*INkP3c};A_0yBip)aReH@yAY~H7*`avt_Ax2>59eM&B+cPk_D_vo
zL%M0x$ri^`?M=$HFR#|WlEBXOGe8sohvA<$ge$dvoM0P?0avjvdYZsZLULq!uy}Rs
zcsO_*2%h9;zPRT~JN;dye?`jloF&)UpUJ7aM@o#IYi>00#i#+??UiwV`^l!7OvbX3
z^48s|0(QbahiA@C1SsRYEI#f>sM9E}+>Zr5bhTIZ8kEI1Hsi|z!Juqnl<#kJFz*jv
z8R8OKID`UxA9_}p<i~GKQ~YXw+KgJW0RU@Au%Zf{qPDIG0c#DgsHZ~x25e6~M#y;%
zfQ*+SRE|%gohxi8#N#?UN!OMGf(Xn5KQCUwTrWok5FW)7a<}N+*>X_%F$6gsvBd_)
zU8+O3L*wzv@K*@*+FEuU_8kboV)ROuC^FSYGd@{9o)ZQ<>vG!83?{Bc91`%OI9yt>
ztYb1k+-V9zi}#^K{0M9lO+8H#^i~|>jBWFn`H^TitXHLCY&sCpz|6pMoaM2xix|=q
z=t2|#Aga$N`%#a-0V$XDs^84<2mqVJJ%`i@pa-8JPBB5<CQ%jFN&qOJ&8MR;unnvA
z9|{MBt&q^!9<=m83~0??f%|xz^>eYn>1_?1_uZS{xW>ub9n4ln>s+X1!%|IC&gnlm
ztiKY>(^}y9x+)dBgRuNu>$-KSrF#7&WVLoBGQTv%VpCh)e5rnzRXa9iU^_p5HQ99>
zOGh0Y^BqMLCprX`o@yHWXb-~1$yNUbR#Vk3z-P&-4N;o#z*kgPiq1tr9gL0tevq_W
zJQCw&=0G7kcr7J(EX3|Ooa(qWnCd!QJd!H*6hHEibQEN<b*~}FVeS3EA-ZL0v?}r@
zJ)d!BC5)iT$KHNv!ldcT;f@W+wJ|c+%#KGBRNElY7+Y^}apQmYL|%=YcrL-lyI8S4
z<ucSO2y|!zgAW5U5^3Bw>|w`sVZJdNu0H{g^RvvyWut^2K3W^1HO&4vZwl<0mEMe2
zo!~g?Hs~mx8EY;bol_Jo+WS6Aum+EKJ`~k%`yMSPlp_~iEa4a~o1Ude-~ZUGG9ak$
z8^OM?&YOQYaxB$YuWUw-rF82(c6>b4J)Ox|Kdx=i^u8K+BBa=!(YsA~>d?4;oMKLS
z-MVs)g`$?yH1K<t#5y$$1(ZLo+QI)JeFkTAest0QPJy4r#A+v*dvE0Q{<SGFBe*8f
z(2qtU=^1+FeEs2h?)*u|#rD3||H*#EkGu0PVmi6A6s7z|?VP8FP~7~F1N-dZ$JPg)
z>v85M>%63h%mum*k!Y$NOcKNEH012)jx@Tb9G^4$=u(J}?dGyw(&n}Na3V>J#E|Bi
zi`-M$&(_CnA)!qnoNK;(uA=p_XX>T12ZLkZm7Y1B=5cqL{Z8gN9qZvg)Amm`HaNap
zkAF?zCs$767qa{a6P6z~g?yiJT%?FzBQVpr7XLKw41buxw6~qOCyv;+E10jgJjXni
zyw3D3C=&FzaAxfo8l1Z=x*15jc%0g|y+k%@aT%6WuZ4IL&YLdY5M4JnucTNaPa$8>
zH6PJYju#Flu}t%H`zj02U%wlwGD!J?b^0W=&%pTb<QTsEt>tE6FAn}D5JTs1#M@)t
zb=!GaeO<!QFf6NNxGId+HNCD8@zVG2sf-@z5q*x*=EQav{CrRc9q==7zQT6j5Tw>p
zC|nk_{B2BMdxb`5poN2N8|gC0Qee!n(a28Lx}P0+b!0!=++j4{U8}p~HniSV!TwBf
z&9k7{xgh-alyIaD9a!P<_)so->sS*ScfzqF;N!);A;7aG(sjCt6sATYp}8g{nrFD&
zw+I)<I*E3S7Wh)=5@_Idhn|)deEqSUgmc_*ia!z&O@JWLgUMNICCvr>uzTnTG=o<1
z=`{j5@iC}OCioZ!6fRQ9@^ojCaxp$~>ZP-i$}f0)T02R>sNu%Hx?AIvlelU&S7g`Y
zOIf!?R6`3N#kwQv;2WtYen>%4%ci*v3D@<~2^byhrsN9abpeGcjdXKsnHe4U*PLn7
zR?I%uuYp-aobff@T$`l@0GG>TSKU{VJhJ?WU0)NzOQ^bz=C<~OKZ#8db8)P{I*Xtx
z<}3e%sC%k=sjt5Iu^~;hNfELFMr}Ojv(l%mp$Ap1NLROJ^<tp1d2-Fli01=jmKZ@;
z>0Jsl?avCX>6+;fbro{?W9P=2>Q>*F>ymtZ%|M8uRqk8Qd^N0rS2Os1#;PhNgQS<(
zoN;}e`CV4O#L+m%OjT7tP-AutC>EOeO&nQ+bo;z4c9j73#2oT<L_Z75rozBnw2~Ym
zD7NnlXdFPBp!(CogJ#`OlHm{zP1%1JQi0jA<HCa=48l+w+W(WM$GH(o!n<zrEJ9A@
zCb2^&E77!#!mYd<nFvl8zlZpl>6L4?$<%U9dU>fab#qUS`h7fZD6x_8EMkPOX=oa#
z;SCGJ#k*jpqQ}NSK+O6E5$43k`-$WaH!7@1MS_AV3J>}Q{{KEndyb%&a;?PMcRfhS
zanL`%JT>y?bc*rJr}4u|b!}TUx)HMaS1oah`|8RjyYZ8(81lJ3H2PurA*TE7@R2k-
zR_-Yk7*F6&;(JJUZ3x-x@PzItP0S>^9aU)wXkxZ#7Z0AhEO&>9=&Xh~gN6ZeX!ole
z)}6&&)VVVQ(OY*scYRk3DgUF<fm4gP3JD#ZW;s`U%L<14nYrrZ=iXCs88b<yAX(h1
z(zr34s`o)#aVgBc;i4b+P29L&d$;rzHwSd@J;$bOy_FV$7GCH#2OJGV%M`kq>CC_H
zpFCXI6ZklGe8H}Yp5K!)c(*(bezM-eJO~!os7>CaacY#`%lY>RgRd{@8S3K_AOpMN
z5>?8{g^7l5m@DZnJA%n|f0-HOe82!n;5@&#$HsV<HB2y)7jM4Fky1*bRGL^3mNyeS
z^>ysvvu$?Jq0~Blv`AW>oK00#V5dVb6Nr!y4^MOq$BCfCK93Jvr5{MzCs3o1rBS7!
ztfyKfZUF7>l{}0gPrrPs?_Zu0GI!RLW(DV8v_kuR)N9}USi0@QEiK5T=GN90`O0IQ
z0Qrcxi{fz)cCD(uE|g^L*gr}Pn?G~De;t}s^`zzvx~^}I7pxtfyI^-a2v69d^^TyO
z*(MDatCC{$W{&n`QOv7f^vTZUqD~NDC7jqUzshLE>ft3#fjpik0hnA44A3$piIH|g
zPrxkECXms5AYH;Vjz9ItR3Y+@$dEm!uFUxr_Yq4bak7p;>J`tpiBqn+N9*Vq$?Jhk
zTa*TKBKvi9q!0JM=@+xSdp3oa>H?h*^i0yCObq1lCIzy|<OdF8KlP6bd}(YFbG*Q%
zWUw#2JsxWkukdK6xytcD-%TsK2`AvBR{!7)f(8avF33>Inc`tTsvm2bB{Or2E_03J
zIN^l5mCP3Feb!38rq32tsij<eM(%$Sb|wEfZI&DTV4J8kigBxY(_ng6`%Wv8qIlkm
zT<P1(Y5E6tqUK?t-1UR|$N8_XdF~P=RnM26vM=hA5+IG-4Qc1Pi==3?oPX<a?dcXX
z8!<zt>BFOE4RX9ZbLI#gvKo++^D7Q1xILruz&_cC7{U0mWrS58A4kd4meI^d9+MC6
zeWs}A(>W0?XS3#m!%mJwt;YG69d8xZkR&i6zx18*7<XtFnLG&;8gEy<_jf!>CV<bu
z491tVCA@&rH_(O$eda1_*&TjJE1m5B!*U80zc0_y;OL)fe48X`lZL6H`rL>d^j557
zF1YCDLL|pQcIt&-$q8YD3S-(b*c>Bx6FPucLMnlFqu_8tmc?KD!|6Dqm~vlHl*g#Y
zVsWlar+QAW!zH;uBj2-`!m07<$B(Xc)!!L)Y@rJhk&Rqs?t_nfiz87BJFljR5l`<S
z=IH-?q+Zg}Ax&k8-;#srR?c|aF4@{Vs(AoMWz^C5z^R~|6K@CN-3YRrqZi4wy__SQ
zz`AVW-cc#Cd9=W+n3#vc_eXIVOW;D*XW8MZs-F~L?RckZRc*Cy6t~2uVxAfx2>OvC
zTj^s?E#!qaiNHN}U{SeyDtQDtyz8Xz_w^6nHtE-yxA<5VW@<e>@Q0DHVc5zu|EKy+
zZX~xhFK}U8*no5@*64m3T{>{e(UACv5p3>}H+YR<S8al_9@bDcZJ|S_AVINiQHE==
z$GhTOsk>liiV($Ne5DY<+!R9|C6M4~T5^8ze+HZ-<{gQ+Qi$omZ>e`@r3SycQ=&hb
z>vgr-%lItiAj7pN^(y%+|4Oyo${){Nj$a}UF>m3x-&=Tfezj{h^GV>Dd%Ws76!*Vl
zA^ncZ|L*|&KS$>RvfrlOp1ll&!x%)-e<I;PU4JeI`d9alubkwlVL`ceJLp_)XB?K@
zmm1qUEqx1weh}qe4fuhB-Hf;~Z`W#87|@Rs7vIfC|5`<iu_sfxV`P6+!v^K0`fWyO
zbDy!ymlxQz@D8L{@+j#o#BxO=BE*A4eB%#Cd^IrSx6g;0Nm1Sh3`uthNoqnSJO#7}
zm4p?E`Lib!Y_5oYPsj-9QL+IKFljIpAMM`#g!S)ViB<(AUORbr!hlS^TtZxL&ubyb
zaxed+1uPwmD75i_;ZWYg@4pumR2=@03s_LVwn+Z{5JSI)&I}=}YY?um;u~Fvyp7AY
z1yVt#fDWM)@W7yA|9w-Ib2Dd=+5X#CDDS_!u8r`rDg}%Y(N)#LgK`xMCPc2B(v{ae
z>fXlyjqu4UZ%?Di!X#SRKyH1$rkMieeW100eFh+ua*F{ka41N>EO{fsl3K*~B(@Cm
zznc{Q=Wp_VKArwwzBAkjG@>p@=X*FtQ|{)aTDUz<-!0Q(P-shTBsN#_XV)wDX|uzK
zDYUG|IQQw74VQ;d(z}A8D8#1^ZGSO<Td*&(RSY0Ds}(v5@pflLfS@<xAVot>|8$Rr
zp>1T|5b>0So3=m;rM+`<qeg9LNx+1AP(*nS#N#_k@(ex7dvqD-{bGGhnxjG``jDf@
z^YNGrcR8^esCk1@1R955qnpF>A)=g$qktkeQJ-W(|7$i}ZDVZ=kH5C6o&|9O6F)p_
zK~E1h=MTP#TBEdlRN6(4?>4H(wAlz;q`qIORgJ)b3Zl@(W3Q%D)HekmlV}R+o&A)3
zF^8b#%$BbSIrM1g6(*Gy<FHjz&~Yaos+C(S@>_!qm?jqSZ@YsH?pigPEs;&#67Tc1
z&vE1CEGe4y(5rdV7jUu<%MHAC>sTM$zAe9c`do<NbJ|lB`#6hb)4Y>`l#zA`t|XN$
zhK4PkcCb=h&!Xiw@6%PY*w!RP`luA*vJjQ0&xc&`%y>}iQCC0hR<LH+&0Uwxo761R
z*OJuIp*>0Dw#@WT6>&$I`wRY_i2BIUm)N6lo^0DzZ7KnBO*2GPMl880Oxy}1Tzu;e
zI_Bw<ad|SOOU1s3%=eH%lDBa-0&W0w^7dYD-u@l$#+UtQY^QtICb7pdFeeXax$<-4
z!<cX3gq&Bi2YeFkj(id95*uzyF*&&KmQ{$?_Yb~rnR$!-^D%S4aqBY(nDF*_ELIYI
zE7-K4whx`sx0Aj#q-WnXQr~KUY@H`=pSFuzGu-E8wKLson>U`lMs8b)j%aHp8k7px
zTU~Fl4m}OAs+R*JfG^yM=zTt}z9B{EThhvPdm23ve)ijt*ehSdql+rH=KOLG!#EjY
zP4~WfO!nqXEVtsxP!3Dalz8k5ZAPh<|C8#YMHxp1_Qcv&N1pqM&^^N?c_XVSqW6-v
z^nFRvb4|la23`CYn?Ga^d;98y)%RxVN?4x?nXB&#t?t$DYJ8RUnp{%afCgDb)4o^B
z9o97xo0KibYhIF_r7t6KrN4wPyP=VO)DixtVG7*yl(AAy13e51N9r{$F`hR(65bZP
zKXITzyNe>NZf9RtxI(AfahDEvh>RtTP_dB7LO9=+)>GYOGG|cT<J_+QyCIr?8913a
zoofRZ8&DF&x+9xA>@c1ZshMldK@7Uqe11*Y9cn<NShghG{mP=yv6k4jpq17mW{jBQ
zpt1`O6gj`nEc<#w!PZo#N6WJx<t|1!0Lw!OCs!Jm=Agn!d!PAs$1PWVh9fq1%9#A?
zvO?j9MBq4d;38*5w_yX*%fQ;?Fvyl-?$e4W5&We6ChSWSr$V+FtJ}i3CsGomSo4__
za4s}_tYRQVk9D=hevfrAI8MpRT|N^H#2ilv^qCWT1#9~xMH2{~ygcS|%hs&){~V&A
zTFR_*T&L%lGAR{u-pm2q&pFq@e{Yhs_Y-JDnx~Und)pG;r;>lSTs{PHYAXG@ia*`y
zr*e%)7(__jLa8ytjV4UP!i#@+Y}@n7slxR)n4_v0qKjEWkq)_Lh^I8TXi(1J;sIS>
zs|2LJYj7^9eqW&9{0&%&A5zzX)~1;<oR1Du`L;&5Z2a*+kBp1dpdjl0yBcMMU&Rmu
zjXmEvwNA!23WfEjb=&d0J?YvOi|akk*l`1Yr#f&aS_?PEmp8(4x>f$`1wi2PEttrj
z+JM4b&Tc_GE|J*fpnB@Dv)ozh7;0R%<2E9gEWB1hDhS@aBMI0S`@t1Wl``Vc5yt9j
zJ+Y?O#ho(i_IwLZL+XtBeVG0ZKxcRJmj}}oSbRl%Y*wB<+Gv1OU!VdXN*Bk=wJF|g
z{xLrOA;?kValD3@!N1$UKnBG_I*V!F=~RkC4f?kfaNq9by5OHm_sZ=h99kAu9-0^Y
zr+K)64fcJkxhZ0G!!Yii?D=xM2P2rd*<TiybHGB-Gnu+T?tHndmPOCk%7q0yR==Dd
z=P@LUbMeOkv3W*dRIZ!RFjaZFbDctY@3SY#`+tAW9kSf+>SZ*bw<uVb2pyEMpy<k!
zQFS0JvPl+hUWH97mIoj|CZ_bqk<1jl0YB`yH!Anc)?P7#9+7wU`9Wpsdq_gF&#69l
zGor6T;~GBcr&z{S5dJ{x-vlnVS9eXGt49tSkJpE2Uh>c`Y5ZigxCTZ$;G_p%SM~8`
zdH=0PKIFn<xb*a0eI@-UA_A3psM9@`52x8t(Vv)1hYXa25*_Q;yXOLO63vXl-)6Jz
ztxVj-q7*w&J(-mNKghD=@MmkRSAC;*Dr7TYf<aD55grSvG*=@^$)BQ6{b}AsUu1EQ
zV%l22wlsEu;)0X=t+3+r%A>K+^~4An{;};xtP=$5yRQ#F>xX!qXMWNDZ7Fdq_w0a7
zR%|=EOcS#F(@f9Q=Ecg-TLNM4Ng+t{`EbGE(!ABT@yarHg)#xtV~o92plg;J<+bRn
zX~ijfz0JdNQ|*kWEbm^&IS0GuSZYbA4-NiQ{a@w;!sjD?_}1wVZL5*uDtV-4i8SA;
zEr-K^eKZ60y+kRRUu9aaGw$=a&LcDEghH`!;`)G&OL#$<Pk$+Q{}`RuMd$uZ2OR$O
zh2g&R0N?7X0u5U#c!bBn*^l{@#|w^qR}}2`7|uTy$IKq?74KbkxPKoI6^#Z)P#Ph<
zx<lNq*_}S9kY8SB;xm>=({W46U2EfbtWD~kc{Thrnl!aMLN2SH$uq5gR!Es{qky{D
zx$4jUGJ-Fgdi3*-s+&Ey6+AwlakzS}lg3kns~?f3^GW<)S^x$VOXN`dTxryrx#w_y
zi{^XOQpY)#e!YZjf{=E1LD+P$ym-j$waejy<NNc@zh;rYH}w8*LL7ebR?M5F7?29?
zaX{d!IW}aS$j*C+FW;gL=a;Z+%k>*^n&sx7+V&bj#lVls<Dc#{Tt+}cVNOe8p>m2f
zLgF<QS$a{{X;~`d3GKVWxAudW-VPpnAJuWZjil2qx9Fz1vW&~nF%-wO{a>|Rc|6o>
z+plaPB-yj?dolLVBxB2#C5c3qv4(7AA4T@;S%zVb-5|!&2-TE|$k>-GS+WxbL*f06
zI_Ev-J?D9z&ikJCFQ3otzV7w9zxVgL?(fX>w)pkPs|Bt5QE|Gsv|Q|YyodcjxBK3&
za2cQMD+rAQ94cw77|{1y=cp2N)i9qqT0(*DsWI<MHx43z?zyeH3}Q1kvjQ5%al6%8
zxg@aYqcE=vhUY9Pz^vTGERu_aKt_N>qWkx^*L3!pipqHC09Jc(HUeq|e5*LyNfYuq
zgkuY?r}pG=($p8$f}=a*5^$l#fS=~+|IG<md7uIZJn5x64ij!M5jt|1C(?*)=ivJd
znL5qcg*scaxgDvgfr^?08wMA#9WM*ftfzDRZNUF9e6DyL@V8DUIPgJR8=G9SDwON$
zy80C}vZbjuQMu;C@i&yx7Y&g)aq+A(uHTG;5d-(I6Evc;*PFY~4mS|U!x_uoo%J^@
zC8Sl#o!BDlasUcm0|m|m!3x<E{jbcz<=>ta9VB2vgIq~vb)Kur!Pp**zM<q6CiMwL
zaFHCR0&9?fd5VFg5Gf-K=>r6iGdgJkB#mIv<gkl|K-%cID3CY$nM#=@uz0P4TVBUA
zkHCh3eAkgb*<Z>mJs%q_N{fB?WR3AC4q~V{<LdqXtmux{Ahdlp)T>*pS-NAgoupn4
zk?lT~A%a^booS;aa6U+|182FMk8NY4b;+$>O9f0XC_bT*!+cz*Apo{zQ6z2Jm4pQZ
z6qzj*IX)cye6rEcu@F+vkRn#36!=Z^9jx^Q4`-zR-<i%&&H1M)iE4Lw%yI(}kTVH5
zfb-`T^WQeyC28+Arn&N7dVO%!#NqJW;zk)Mh0-xF4>Vmm^*nf0pQ*PYt@sf%v**T(
zqvSxBjgK%_*2&;Ht9{CYwU70UX*QnwXWot=X94zoSvn^xN!JssX><|A0eb0P5{`TO
zsp<#?;0;!-lQ0<Up%C<zot-UMq~lIu57g8=lmvFMOS0fOgEyC+_9guyX-CXQ)g%fE
z1!7}Y0)Ll4;$YFZ4u311d4@@=umf9sd#Z}mek!4r0L2#HVon*T#-$-pmVr&Umm2aG
zu<99;zAB)k!!j@#_ew+lI#AMULaYPil69YVq5<1QrKKs8*<q<T4jTOto+nK**ReFq
zX5g#W9Farm3>sH0!p;vbUaBZsvkBub7o=F%@{xq218w(@PxPQ*lKJ$NvRSpd!Y)Nl
zkoiXpPvOznV?d;ti01|S-3DdPTV^_C@=fhc`vBINqAIcLI<x^(3T@%};ccIgxB86|
zVk#$MD$mBUSnFT#zPWZ3H_vFzX&7j4h~DbF>{FqxIP=0uncp{h_QS^@jHzL+gztbt
zVx3}`G)6-Fc~}g;CBKizGMUPyAvXs~)SY7n-4|u}pC2Xe+9ah#y>o5j)-nj!JaRXI
zK{&4nd+ZEw41Gm%5N;39FDRvc`Fi<|7~@c5;XQMGw;y}(TJ4n@kT@w<5|5i6QTCMX
za5~3ch&v_&dVb`vRqEk&Zlhq&`|I#{JoX}+sz78#DJ9)m;8YJ6@Jl9*H3+EP?j|vH
zM|>MI8hTye11BRo(I3!YR2=$H^Q6f?i4_0Vx&IdC9JZ7I?m8mR4V4F`{D+bM4#%QE
zY!ib{vyk)x*?1L++{XtL8;Ro3b|X6EXW>YjSA&FwP-q#QE%U3=!8tE%@zkUXVUDlA
z!lAw7_(Qo^3(00V;o;(&?Iq%l$g<(5pWY03D$8)A^xnW(v-<_jII`>1Z5<ug)n#_I
zm_W3&YYMMo{w{j{!)x+mCG^G_(LLl5h*P2eD^xF4Uk-x@?6f+)8Ge@VZBt6JGYeBL
zfV!YbJm~jHLe6ZDeh|)@YEJU`s<ic>iq!ZW!*|@V&|^D<9qgdm?1*TpF;y<j(#rco
z2>O9NxscGK0E*na@s0z*(wW)5`gpc*I!TK7j40pv0y<sgdx$dJH};|yr#zWs8`f3?
z686^U>Jz@sJGu9kv`lS%`+-g0VJY>{z%ZxjX?!&E=)Tu2FMP9RjH>;c+%0bFdtpLz
z-lBI3+n2-Z9f%DZ<}ip;AME%Y+)M+hsklC>{X#33B{e5Jyl}Z45OH%y^^?<0X_08Y
z2d*kva`vECXPpUNr9*R6*y7oC8|2~%WVBRKQe53PwYGHQ^y6rkn|0Y<b$}a{Z+`4N
zaZS%BGU&{Wg5IEyPJ1Z$5dIvg=|X@3Ia~0}wv(o(M4_C^K81@;&wzd)$PYagVU1!g
z7G;WChe(AGn$tL7z%+bo{lW>-iOin&744_J&}GLBtv|25P6oWtu~qia^5d>7+L-it
zu|Q(uDy5I*le9&dR^w*+wFQ{1J2VGaAY|6GTvqgkfpJF3zYVV~v|KZGAwEjo%y`I#
zEApy{50{@sNEwPieInPL(VN%pPNBVb7O;@a+MJfVszHDVivUaOQgYnom*}PSIE0PU
zuU5myrY`HwIV~Sv5p3vZ+uaK@W!A$(JBz%{i0dV9$JqO@Ue2{7hxpaHlEF;3#p%VV
z=G{N`<yAA{vt4G7t=wfKnG-4T_Ms`VU3#8Eu@>HH`amFode?@BCO1j_YTqyPHvy*w
zrTDejYTuoFO{SiTv42^$v6Qaayn91yw*lamO~~pP-NkMyZ&JXM&c+8?l_XZ02LK;`
z;;+0G9Cc-uJX^|_0>IFSbDEn6$ud)v&gTqDlx4kq>4}AF96q(m$tQcylt^~J9nt_M
z$JTnXwiT{@3NPYQ3h9(QMr<{r*l{$2AAGgc81$>1#7t9B20!%4Zv3y6_H}83aT(gj
zcW<i6UgvXUCPF%kskIqNWynA*^{{0cs#o#bV}{XVz?GWksxrFI>bNzXE9^7Sfs|5w
zM+JYXCv@x3NFj3lYti|IC2J^jW|8+^Is%SnCDj9O+b7l+<*q6#O~6=Jp`RCcfpzOc
z1aHW+b6wki&*VVT$oeN*alR2NlZE=Isfi&{j=S-VTgXQRZXtURZU1>IQeYrQb^9j7
z>M6vPcPCm(Lk-5gWr0;$4;*M--3foqvD#MO4B9tw!BL96)LkWZIO8O?_w0_#K83tm
zjT+}&Z|q*nPWhd7)%s*>KVL?`mfX{O#c}ULaPD@^68&8e{=`94ccr{{QhhsdBOmX+
z9=B^gvLDh)4$tx>A?`r6{jmI3fsvj0_MC|<OcqcJEklcH4(biDBvdj7h74}%%FV~7
z_NM+d3x)TzuI?Kg17@fE>1|4k{*|JGNvV4!jYPhR-5Hl@Qbr6w1y-GjPFAqyvh^Xp
z#pl`Pwt>?8VHNeK#w9UhPtL=~41)5Egw&Y<9t;(3F%t@zdp|8!Yx--6wJ~xuV<5$%
zFgY&<AjhDTu>JIxqNy0`9~J!b8`6N;T<oGHx=_w>m>_Zcm5IF?kb;v#5y!|REGLs5
zU$4F-<TyaQm;HH};MrU;bq;jMiVW<jl(@>cCpe!7qjpK*N*kou31pO1QDRKpw@Tpt
zab<=NJ8&e)v(SK@lm+XGD{c;0Fg&VF9LeA*Q|vxG#AfW-X(b_W+^b+mrB6BWA}ah)
zBu-4*t;W0CH;FJ<Q&^i>QK1$>`mnL+R-1I1Wa8Nqc*God4bl{Kr|Z2M#j+DHR>nAg
zu=)rQl1;daDZy$+#IpZeK=A)_rTxX2@DCATUzuQAoZ$4tH^D{HF)<s3k{gg_YT83&
zSoySK2aa=zqKYb1TxL8&gcWqa<G-fDoRB}0;L(c!`{gy%*~p#;Suiq95bduIY8syk
zn>YD<IekQPec})aM_;mo?!3PLtZa6vngj_gW4H!M{z2lX%gPtGm)fK9#ZVUj?B^Zm
zAyi!ez7dEMVk&XSKScSsl@QtH0?Z@s<AO{k?3(DF;gqZQ1y-uY*pQH#o9WSBzDQA$
zYR+?|J%eUt*|c93p6Qf)ZRpv(?)G|HRdrN!j+brCqUKuT{c7J#hZI`H#E}vO@kf`C
zxMcK5y;WMj!WJ6N-`Tahp?QRe_1h3^MrT1rp`!Gd!3}06Eh5+%d^Vb8YIMog8p*!E
zI?BYPm0$Y~TkJMj)d?OV@ST$zRuWKn3HazI+mYpwS1=`{kuplp-qd`Ob`d4NAWQ9A
zQhHOqRFK;RX_VetXQ5n!Ri&-GvZUo<QMg|VxmCm3D#Sa-)8;Ot3xzZICoyI69Hq77
zCliqKOvIlnCcaC4_05EiFF$)jnbOr`bCec1$c^E37l$X_i~j}tb^k;u`q8e>#)*G?
zEGpSaKW6A`9Mj{x@A1xqG>WM87`~KM?_ao>y1$4hdYkI`A5sqY$++w58-gl^GvsZn
z9H~~876aE1e836*+Y#^Ia`1SoE??#2z$&l&IWYbQqVNAXmQzlO3pTxHW`JOY-5UPT
zy*|8j2>(P>vaV^w(>YdY87;?{Y7?bgUf*BbCq7tEojqDy8d~o2JMv=gL)^FiPP=`Z
zq(|b=TknCeodwbbBUm<aoHn6lW`CTj&DqSxC+=y!T)V)9W;j+|{SIr^!xQA68v6)s
zxff?{T1=Wm78kA_XlTn2=}-CBF#Z=AhfEW&@+kDUGvJb^0U!9MG8F$%OT#t2KbSD|
z9B_}SM1R-!Wg=z(C)(ogaVoXpWHHzvNmB>HtmGvGy)QdwEQEH0m~>gz)^W;_fkVm&
zq~9wO6=!rxt%p^A$jE+FT!&W5SQ$(iPi7G#(Kh~JH*b52wm1ZOhJ(RRxVfj#gk{&s
z&Gt#vD~`6(v!WpC&oSG=bZwcI&sv?^lwIBYQ@d}M@#T4Go&3Qf^`mZs%!l1TxpOSt
z!B>cMv#DnP8~t-l7=mv~WzI)r?I;X3_8<@})Gg;|0jV#Lgl?Q&7*Vi=D6)ruS~b}r
z9F2%)X>k>xSTo7yG3cj!R7eq@tD>o_t9Pvp=+2rRgn@a~m8H*BQ8v}p8)XCXT9Y`c
z+(XPiDFt&4u(lfcf@2dT<q43I`Yq4mGk}IK&!-cwZSht}Ve5rmGEgeg7}}yGEdd@1
zdSl5y1iA*K0MIR3>p<Edce&7#UjX`q)V`e(qa__HD4ElwRZYw%DBs{$G(T8A8h5HH
zAr1j~7aH@+70jEAw62~&A~^_1LYdgDq8DrwpNX98y^m07`;~wmZRY`Y#7I^AmrurC
zbD~Z48Iu<b*^lp18V0WxFHLr15U>9}EM$`JBNgIk0MQopIH?2;c%};SSdS)95k<44
z>gEPLE-X~#Uzc5nm?Jk+s8Qs`<c002IZq)wA`oy^bV9hVLps_N7JKqE-s+D!t$+Qn
z%qlSlglBk$)eJR<!PNQ|R94^=>gx2KFV6v${)A{mN7v=2vFEoI)AnBKVPkZB=L#ec
z3d);aOS$r@elMFM-0inI*>h#KUWC#y)(YWt;3pzqC*ffF0zqKw_*0CthSTSors?SI
ztyv2!^#sna2|cPn_!Wsj+e7GQW{sxrqI!w)t+p@U<y(z<m2%(ZTkv}a&1+^0t_tF5
zCzt^F7M={%mZv!5&~D9_*rD`pJnmBNZE2TP#1p4)EmFL0;G*t;O<MRfn)#ZPXurBz
zRK6ujzhdkb-=2OMKKc{zqq5fy0@y}}+Hcba<m38K4tFU>D-sn-QkB?gV+3xI1-a`;
z7Vyk@VQ@yM6o&<VYpwPr?LCjxsSir7v8SD>m7)m)K_pNXdY?LPIn-sot37>ug!;G<
zV~FvRw#zvupS*e;cI%h5=(l;DNrLrPim*0M?8*@_)KpZwyRgoD*-meYxPI+mzNc@t
zwDHj^Y^5>L{c}{qEUYEh%ds_MEv7po!?Wduqsz^RtgeF#i#1B`sGAs)g)EJ*TJ8n<
zkyzVz=3qA}JXuo^p2=}nm2sXwBnZACP*WG!r25)H5EWv2wJXRyy`aMlEr-I>1|S78
zMRJjpTP2WN^lsf+R=U$UlqvIW72`T^+Q`@4S4lVTmhbs}!B%y)2|{!DE#M!63hNNJ
zyWTT0tXJgJ=+=DsO}^FtSwCS3_=K{DM90W@L*so1sN1r~@a*MuT&mK$J+lRGV3Sk@
z=L&7m8uH6aW%=GQEAVuZyHjj<GC;ib_)xrczJMs+x}yjvryNEA@fNy=Q^Zpe5N}~1
z)tK+%Ehl&}0d*R5#!Mkd=TN*A!zFPj-kN>vi1Sj%TNzt-X@Dkc9cp|j?dL~}Vo&Lf
z274cH34bMNVl=nREy&&aO}k~b@H_36zE?{sECNeaBt?0s-NI0PIn-{gb{=ZCB8b{8
zl?}_L04fJLJ&zUeM{U13)N_zEatNc{!7LKeBzu=fsMOE@uIR|h?HcC!jL-E~x;)H3
z;W@x%jtrx?`JhDKkCg-OR~_}L$-0?(9b+6ErC&BuJkeV%|L|GOsmWJ1dS1g319R_g
zuX0>NXK{_y$5#w!FV$a}@G#HAvx4d3@?Fqe)P2u(3zawgFlIt~m{&;l(a0G00^#e6
z)pop3Hx(}%b;Q)EC-UxZubwcyv@$TTZ4|#Eo7Z7Irw_La7V}X$!=!6Xk9I#JG121j
z-1Fi3$F%Ne$<@K->hv?Q9A%!4E$)5J^@L@RQJL!YJ(W4%<;9pcB_=WjPI3;3U-6#x
zW2PVW;VVKW?)ccn`!GUKmwp7+c3uotae1+!+w8(^MVUL@P{%#>!0SA29u7?1K^ClI
zMN#@84Hv#LP5SS#Hxw#{<#-nAZu)kIqu=H%<@2IupZ_*6JlpWcg}mf%^rCfa4QuCg
zW`jib?sN$idn7ycOT|q`Cj7!S3(2%{lT+b*r&64*3NhL@mJydS9ve2RxoZz{Y1Ci0
zx`y*{7*W<hUo6A>i}}a_6Xv2v3y1~BE0_rd^_b*;x@u=s(GPyFdA;iggKGiTOw+B4
zNDJybm2xh$)=dT(E1f@`mo1imI!H%q$15Ken6JF4Z$5?U9r7?254zNqAgluJNHa}_
zOk1@Wex=cl1XWCq|44NXYow3M3(jWl43F+uIl(g@e^#!k?1Z1ao$J0H*Rv5KqqiL*
z{~HFv#^`=hM*AI}6-_#Jw;mZwX16C0hOUMoYzKGkXTEIu1z-dZ6-wR_Z0C0}WT$q^
z?trXoRsq?Q$aGnl{TALpXTHgJR6Z1O(Z8`}J=p$6G5LJ3NeLYw)jDehd>n3ru>jvJ
z-z@}GP_Ty2bkt@SORl`je0vN~QRTI?ZOyinkr&B#T_DP;Zq{5`t8`!z-n?PZuW=f9
z==vbMnVh8(A-Z|RU_kQ(P|$8lRTh9zwGBl``}3Q2&HPYwnXAlM?SGS)Bg(N@0SUMK
z*9`^_THraS?Kvhofo=kHxx~8}Hqsb|kcP4#B-ds?V1v3yPnsNOE>6xGeVTdH8Tz4L
z;}lRJld`V5?_?#_u`q-ILJO)-a6(O|ZZG#9y=#EVNfW9EIx2Psvm&w;k!3_>*8{9$
zY4GcAnnC#^M~+l$UjmzynK6lMn(om_Z3hF7d38argQC49$4+Yl^Ub^0UmtPesfW8#
sI2s@PvzO$5l0pCL>-9fbem)?1@R7w6skL4SjQ)tWhW@3Bi#K8a1#-*m00000

literal 40661
zcmce7^;?@kvo^t<woqIuNN~4ep+Ip6UK|RP0L9&dw`hwN4{pVZdw~`x8X&m4dvN>G
z_dVzQ3Ey=kKP1=l%+Bu4?#$fx%onhlJP|%MJ{lStk)ndECK?(#3=Itvii?5zha!vQ
z77Yyp4XmOqhx!Qs0MOCVF)%PNF)^{Qu&}YQad2>OadGkR@bK~R2?z)X2?>dah=_@a
zNk~YZJbCi;>C<P=o{^H0l97=;fBu}DoScG!f|8Pwii(Pwnwo}&hL)C=j*gC=o}Ph$
z;l+y=FJHcV_3G8@*RL5F8JU=vn3<VbSXh8SAS){?8yg!tJ39vl2PY>d7Z(>dH#ZLt
z4=*n-A0Hn-Kfi#0fS{nDkdTnDu&{`Th^VOOn>TO7#KgqK#U&&pBqb%Kq@+L~khHY)
z+qZ9JWMpJzW##1L-o1M#FE6j4prELzsHCK%tgNh}qN1v*s-~u<uC5LSgEcfXG&MD~
zw6wIfwRLoKbai$0^z`1pf3L5vZ(v|xXlQ6;WMph?Y+_<!YHDg`W@c_~{^7$13kwTN
zOG_&&tB)T)T3cH~AP^fH8(UjjJ3G5ipFY{!+dDWoI668yIXO8yJG;2JxVpN!xw*N!
zyL)(eczSwzd3kwzd;9qK`1<<#`T6<#`v(LB1O^5|q0pe9py1%(kdTnj(9p23u<-Ek
zh=_>KpFe;3@+C4dGAb%6IyyQgCMGsEHZCsi>({T}zI}_2k55QQNK8yjN=iyjPEJWl
z`TqTTYHDg)T3UK~dPYV@W@ctqR#tX)HVg*K$;rvh&CScp%g@g*C@3f_EG#N2DlRTA
zDJdx}EiEf6`|;yPd3kw7MMY(0WmQ#Gb#?X6pFeAAYHDk1>+0(2>+2gD8X6lLo0^)M
zo10r&T3TCM+uGXN+uJ)jI)457)!EtE)z#JA-TnLb?>~S3^z`)f_V)Jm_5J<(x4*xC
zU|?WyaByg7Xn1&dWMl*mhmVeqj*X3tkB=h|h>3}b$;rv7sj2Dd>6w|C+1c5-xw-lI
z`Gtjr#l^*?rKRQN<&~9{)z#ItwYBy2^^J{<&CSiNt*!0t?VX*S-QC^2y}kYY{ey#p
z!^6X)qod>F<CBw<)6>(lvoj<Td47I=adB~Zd3kkpb$xw(b8~ZhdwX|xcYlBX@bK{X
z__$hS?2kGFcuorXu4rg<5&yn`xQ!DlG_)6Jin7w$UK#tDI9_CbZhKQgW6{~j$*`XL
z9&I>E2YTpZK7B4L+lh@!-uz1N3Fw73-jf&T=ztvwvX^X16mMCOa4t7oPN&C(OmAV&
zhNXqURpUnG&0BG5hpo+vk6W%<OWvk`*qNU9XmXuXe2sUB$JA!~!30Rcw^IZabQwGY
zmsBPHfBowzVEdu-XkMpqImvV4`g~9>@SToaD$Vc8_57PiK4wgY^e86W*K%mEUIPj}
zg{hd9X?17AHu+fEwrRsAOl_pvGx#y$@RmqGIoa%UkMVkM<Bvs-gI=d^o-TOB$C->U
z^OERO>Lf=s@y+u>?oTF48u(m|jUzkSO{Z)WjwCW5<K!1{z<EI1aau0fZH&o5v%E&d
z#d?o__Y?SCi4%^1iAL_0WyiczQ|?^`&1Dia{7V-t{D&dghT#&y5ZOi;?azJE)Ln+Z
zgDYBH1aJ#q?u!B0X4E5$<snl0zUI}Fxrf$|D)u*5bzzmh$jrZ6-UeE7iV~w$wYf&u
zUPrDbN7tR)Mq0?>(!lE9m)B%6feD_!F=n?GO<ip+wr&EaN&H4(wL_VAmWt;9<lVHV
zBdf>3xM|Us1=hSh*C~qjn{T=_J3}#KRxO@OTf%JPKb60(QS@u!yMp$BO7<NDTf(2F
zoirbNcW$sC2Rro#9I6&aakp1>6zWHTk@saFme->caL4O(e1Xku>MQerX{gvWZvID%
z<_q&IGM&fhUzy=4o0a%N8|ypJ!i8ev6D-6v!3J-bC`=fr^h#~CRGHOBy#PnaHt<N)
z!Dwu{MMi^d(nqF4_c5Sg*@8i$T*s<2>FypA;>x+D#UP;v_MPInGgrpG&`XTRz#*<{
zN;tdZ0T??McvSx;ZIHJKG!e;K%UpF^=L+04xF7M|9~T5oX=~#zp02tufHFJ4^jcmR
z&GKczbv#x%$zyMi?$k+Ge*m?j?NrY<#x!Q7#u!HTt}X)W0*BcPfvV@S3?KsF1hU{n
zJ7&}OK$64P^7G;PodhnNw1Kr>m7fdpy6+=(?$Wwjk<~2sW_@67q3hnjY|Cnl9W{)b
z)v@YhywdVr*lAi94P3{-c$qmjCKs{%_^AZAMDqv-WSZUWF|=*W+T7LD<+heQo&yro
zikN1-F7k&WtS}<~I})WF>j)0K1Ja;0w4zoC?9baj(64*GF<W!fAplAu@xo>cI-}^f
z!*>L2*`7yg8Hiskt2~UH7z8@p$ygkT29$^9-u-$IRM((KJl>!gc+mu=Or|<iUsrts
zjb>(a;NT$d&LiVhj>f6~cno_gjlX++<~6&R8_8$#DE~U!QJKl<;jOpo3eh%)9BblQ
z`pqrcY?^6{vBuTz4l_H<5j@;03R~`8UR4j2r@z5LYksTYAV3-cRASl^V}cMBr(UJz
z3XDAponF@;88*B@Z9x_owFN4BD{3O`YxA><dN*iFQCFV31N`Cj?*aGWd{R_s#Ul~=
zfb%xg<}O2t!HEVRn}xO;XRp<W$O?i|D^Bhb9~z$GOf+DAFa=fuTLB9>faXr-gRXe`
zbM~Y-Azeci5YDMA5qx3a6b&wkd^!U)kQ}4{@BL@GiMn4EG=@D(bASs&BojSC&(Axk
zVN#=YpZ=N0qNW!{n$GOcK}6#RIg3f2Jrk3o=ceuiz&2JI@Lt3<8vfz>>#zL)dor|Q
zm&p^gO4fdbGp(sJ?1pbm&o(iNQ9EIQ+KILazZOLq;Tko^??MGY%@Xeqk-cI&QW*G>
z95i6JqE9*6#1H~GZ&z65u;}6rU3QPAsZuHIaRQA_ZK@b^7&R-_K>%IxL|FE@X4;3^
z5}Z1l8=Rf}u|RAx|L=If)&$n?{eXi)psAN@fY`zgft(Gb-T8a2c<v(>^i5!nco7c{
zQkF{YdhqhCs~oFe-yU6sdCDK|nA=KdE(t#%%##mGOBCkn&vMw?@M-4AuSh!bH1zr+
zymcB4yv3ckMYaj`5wqjmoZe5_i)8N?n_VK7)LImJd=5sMGsBM63eDJ11Mtf6az;yO
zCO!;%&gLf}J+taf**78g4F4nJ5Sr!lUIYKZFQ~)7v*ki5d(zH$+3s%k=d=IFp%$8=
z)?3b0>C~Q8yiw7h7yJ8+LY7^Qc+;tfqK+B){iO1YdUz^h@(~No9c6)PE5yZ~5tDz}
z)nnkcm-GEc$Z=O=WZ5H|QjqzLl$xKZY)%F813-?U&|GUh@!+3abT@fcIqzkX0|slH
z6f+E!UNO#C-NY8%?Q4^hg(wote^-%cm;`K5P?$!UEya|=>h3<x*k7+<y~OzP%K%wH
zp{|9X+cuXEd7Pw5$lhP<>}kf21O;i6Dhc>DCBp5;>(6JiQ${wuiVl6QE4(T)9wKO{
za`jvPa~uOfZCO58Lj}LnA>uykk5@^1J-h7c;re587?G0uGWYH_>AyeaAU^rRMK`=`
zku37|OAd>F#qt@|IT3tw^a%8SwqzHH25MU@UodHFvdk$vD$jP~y8RIEBmd1V>&Sh6
z&-lUZB>u*B=MOg`ghcE@HSBGQgn7a?yIpCVod4!XWeU5COhGse?c$*;i~7Fn)br-u
z72Wim1zdPWJId5$(b$p#|9|*zIa=6O%}=ghp#=KJKRp(hL#8f5etufD$ZWWMF-Pbz
zRr9IihkdxXku3}A$zF-98lIJC9E!)p-jHtktjmn>sS$7S-s3is*~ZtGUc@V*dGg2S
z-Si|C=Lu9IGO|fG-{&;N6wz(JVIoYQf%PJbOXHz8O6d5N%Ak8j08#a;(o#lPdpxvj
z?-^K15hU~Db155MaU~<{xET|n{tW(vJ-1W|eU}||u($x{377)lGu!=*$2WqFCNaqt
zKV5uQ{^yM@GI%l@4r2Vhj$N5;nZ+=-3s_bs2@|1bj<@|6GYQSUydLQ_PE*{LVY&|c
z?;u@I;OifciZQo+>Ixq)FLcUXe(lJq{Z~acLIB-eCk})53Y586BD$j-!K%jXqWK?@
z@;C+^57$Z=-O@P>ApySt_T^kW53%FEF1Hk-thuFZ!DtK9YY`{Yv7{+CcgxlTkB)qU
zPyn6YdyPbNB8~V27ZP~$F7?_GdR@tQR0bp9T&qfH;iW<MRy!RKFQqy1sq(&9A|_x#
z@4FlW4Z&DOqAzq3+<@{8r9>GfSKEUxVT5s8FPa0kVK^p}6?2v%mbcPbgc&cp4@1O-
zE{ChG|7~0tu6s!Yug~Gz?2pNBEJ&t?%ME>xA)XeKR0f@O``oI5cPCS(QJbzqS)JOc
z){PUV*TZ(gO*jx8_l^_Ka)Xhbu()Bbd>2J4QAtze_NZWZnLSO#=FpGkNb5$TeCdTg
zej|sbOLKWY)NDwGJWlP3C?ww`0P>S?jVHDCE9N$wvO6zsm<G)Ba8s!8o%{|R9dv3w
zF24#>Z!;2*k;}@7-dGgseo--xQTrv$_Rbhq@9Fa`VVx48F3l#0wZ}O%S-M%?`3IPl
zQ#?|4QQhp*<9<M2Y3KD}b(_&xt5791;ZAVmn2ZZnj#8k0_tn}FW?lCGEaz81`LQV)
zQ=KgySNm^%@fZR~**u1T^nt~leVN+7VHjZn_?jgpQGxHt;E=4GMj%Z1>1ZRgzFz_~
zRqT$GcSRoTvSgU7GvXCz2W#&Rh)U|n0XCxE^(W;*$#dROQ^0kbaEcpBW;q9Zo0Dee
zuH}ePX3eTlP|>;kMM-HiaZu1MGDkzPgULBYVLh@)|86K3`Yu8W=4)GaBv@7v5HR?E
zuPhq$LyMmG1C;M+dk<b}fCxgRV2dICd%9&QPJBfa!~X=oFL;nP6>hfRRthe7_`0Uc
zdI})=`QJ%wRqJ1;ZyR=MHf%J;h$nvQsdJ#38dUU)=E;g9e1}S&ltE0eb{G9!rHm{O
zzi3flmW>%IC*5+Gef0V5;%-T_N+jNmk;@55#MDOLeVtO!K77E@jyVF>C4+Z81LGCP
zsTDlz6(Sm(eHJOO5_Mzx6+jqd$<V?U4CEcMkyqY4k_aY7BYtli#}E8di0IG(C%WV<
zx0#ou^LcTZretNcr(-9jYZBczAx)``xztwD^b*%R!hkR=CYTVE{yVQdgGv>KTYu4!
z$8T~WN_h8@QQkZmffNBo^MiSLRz*k=4i7Alnr^py+cR8QB2!iI>j4wjxf8a!9Mb$<
ziMq9mGmWV2lQ7SCpbS78gd;}{du%m1tYF5`KxroRTq`j*QzIg>Z1=9^#-jD3=nV(%
z_6$o5&Lf~74KUQ&+Xwz2p%4Pi0*>s~pFHKuQQ4YJsyCluZmXXA+F$q9aXZ;$B)uVw
zm6ZS^d4VO4R(as?HR%xR=n!pok8`-sjoO#s8j3)yktZ=(2_dft;a_?}SmS)O3AaHP
z?|q<*gb>;RCmQt@5N9u5OjPZ{>?83Lx;cG$1p=-RO|23Zqqh<(ZU#=NLSf2ltL#}h
z@c?=BZ0_F6(ZAc|q!jW_zrhOugZxIsyY~?vr3|<O<QHsS+s(!e1)wp=xYDmEYVu~S
z7<|PBw8dFB4qL^@lf!x|Een3w#3l%(srIfmDHFbVSazaR?foSAom)gYBqVtw{(TE?
z>ylY;ULdVr!*tno$dA_sI-aM5BhcF9;!+6dCiB<HX3@qg`VsJxd}&bnGnlFVI1fyF
zm!vk(6K2|{zMtxLc)>5hV(TK$ETJ6t<?;BVmQ1<C^-N9e<GozRq|T3ko?}9(<{zhh
z=x)T^8&<hG6W>dBT}MlECD!yn)R>8$MNV>0&~ILifa&w=9=>&t#su))udM~HQ#qKq
z!ACx|68D6DODG&<6d~6f7*dU>FnK4nM1%}kthO5x%DWdP@ytNnMi6s!$3XZ=h(CGM
z2wh?)hLa5Wn_D*0sGZZ2=|k1lb#!nMoR~;gbTWa@lwfKj37oMPu&HB0&!Acmh$vx^
zrPy=5Jc@XPYgpxmJd=uA_t^X$`H?z%2um=IP|dtr{Lg4^9WQ#HZ7<tK1qMo~Dq<#c
z&<II;+w!^U$!b{0D*|{-tAnGm(oIq`&2G`H=i%Sk3=@Q$`&hUPCWK!DWOJY241@`E
zcp0IAWn|q16he&e<i%ya_#<}dSVFnYKvGszix`TO@S(v)V*0UKeIqgcgDvl?3>QpM
z2i)fGF>etZUqj0xFN@T#%D>U}^A!Fvux<!xZa+s*wRiE!sumocdhUX()u@szqZ{tZ
zNBXqy{14<$`P~Pls<RbtJ?MV4j`|5Hk-?Ra>=-t!OAvXf{==RHZ(;C@*vwWQfnfiw
zYCH=_mr#^aJ#fQ2%bedU)O)5z0q*`IdD1(HK02NUh7-N=en#R2UIUXDr4${U(@dwF
z+rbhTG1=~yocdD_Bg>7l2n5O=rUC~F<`XjLk2`f}#9M>xZc*<3M%bM8V%f5WIBllf
zWrGsZXI%sC*enR{cfVh@UMZ!n<mOC#Bw*@vhjWV%LP)XEZ}JU6G+^wNZ^D2#291bk
zV45($vNN-xvdD!@Lf7R|gIw-bz1hh47qD6<0~0cM%L2L*dpYW3#~_38%%C_MB!R+2
z{&1LFQ2Pc-uN%ms?<(bBBXMgukHTGkmCow*^f`^NqK?akiFl%Qe1f%lYTxp;pz~+m
z@jV)vsU#xiIjb{|C_~Ut9YIxQq;gvn-3qj~oe5U4P7QA(4VWV$b<%13WzF+g^_O%g
zb8_2m_)IIA^EM(a6RVZ3bl4msZbj=@+B1gU4PSKr6qduSn&oy=dQ8|5RS)xQ#HxL2
zrSO!q&U2wGEOGC)TfYetku;A^oi}72`2cLEP$xJa!JuB|)3V80(fx{@l8QQv?@-$M
z&v3#xS!J3bC2fnYJ2YoGf$(L!vtAOZC93m6A?U-PZ7GWaemp_QTX7kiubU(gib-$1
zH7fcqd&-8NWlDzbnSunn)*!5q@X2>^E4@q(JO+maegw`<qdzNE^z9Fy9Wb&7g^V1F
z;ikOlcvD5xyh-D`cbOxP4H#O*X(!OSOc&@ESfQHElghL`=3RLSswBrs-0yR+a@x<S
zo6JLjg|`<rj|6uHS#~ooK0F0t3N=_bSmY*uNuF)#rld%j#f@i#<seuC$TkTwq5cT7
zPlB^?L<1`6{NukaDo=hJVzBFbb^Z~Axf$e=U$sxuyS2xvzqB;7!q^sFfZj9LikS;n
zod|fZ=U^X47h(z|rj3{b@j%X5BLUmw@W_)BygaWmu`lTsDheO5wFY$a9aicoe2M)B
z_+MaH5^m?{)k)qaYbRQIu6Ng%QPuy%?G53&riA0sVw+oh8{wh`TiyZ;;db8Pv&*+{
zLEKgDLm%kBCa3r#ws1gGPLvMo2?R@RN8g1q&lem6eGHyBE;V5xH8emwRYTA3#E9~w
z52y3<f3LKYX6^i-W0nuC3_(`{VLaD|`ZT?^v6LqV>-g=Z)37|3bc3*D?>U$#vixEm
zH2XcsGXjG6eQR2Hlm!$J%7LhjJ~k&=H{z4MMW-&t#X#g4vOM1jjW^Sw^H>Ph40$g{
z20pQ3uE0+O{O=Qcnd1eI<loWCZ215;g+LhM1->C7VQrxIj@$F&rbo504cP}SA59TJ
zt&$B6q^E1cC7z7RWh(hj3EIwx547?U-XJmfVIV!Ar#yLV-gSuFmm872ctBT<e)&hV
zW3F3q49~+L7tLU@*6<@+A@I2|C%<68dJgxum>f=+;f+UH@JTfW^4B{&;8!t_^)p9U
zuLm|#R4T-zkcyRdCMk5ku(p;mYejhPO<OhC@miPO<FxYETQDvmP>cfn-Y+LW%&>R~
zE&dZ)%!E8!94HTqaQngMPm(65c|0Z=^RH5+my1yKlovy*N2^p>ST+F`;P!E<TOs$8
zdk(Qo3wg%(o^KN5&&r1*kPLsH*-mL)>$24C+RQ=LK0M|qwGwXr{kqQ*4A_WhqnDk0
z^#ndnQoSVcN(~*K^aYIN@1E&a<DnZ{u$s9U=<m+g<5Wf$JEWx=G~S3N3b9TVsu!qg
z@y!wA*=%o*ht2pGweW4~UjeVyd77SQ5-KXCfu!P{wQAJKXl}G_z5TlRVP>D7fxkn=
zT7;)ve_+NBA4z7|NR<ATs3A8#4Aiy}6>c!J6)i=QSC;nFE=cv%)<<n^dKA`Wo?xu>
zbgp6#xN*^E{tSVRDV1HB+oe%U5{=Qe0bt8X|AYc0+&dm9kKVQ=#&q)WlRRK8pnG_%
z_vCrk`%;}58Lu9jiu~v<5o37L%c+)BHq)skC5j%YHY190lPWkw1x65oK!4wz(;Un0
zqb&&eqSB($lI)3WT+&a41QFCr9_LX6X2iI_a%u^z(x)CLZfZ@GsF0|An4b*I<_XyK
z@y8dzA^hTi_t`2x+X*R2_d0&&yhiW&-Mo832pPb{NXyAiY4gVNt0L~dg@*gcRk2l;
zK4M&fK&-$QE`&G$odxvW#d9Kib-zL_g(t%s#D(z~3)A_94y!CjE39I;+ulOD+|d$5
zv&yJMQ5n?Kz+1pwDSM{UcelR&HvwHD!seB!+2yqtah`^4L%l%KSx!*}#;CLl{X5sc
z@*wK{)g8+1p>WqMuSows4*Q7PH(R5E{Gv^GBgg0gwe|Scpc*v*-HDePY_BNRSFLNW
zS76cXZQ+(kYrxo=C?Tnu6|HpkWAVxIk{lH=9XbUGU*6<fcIjD-xQ#j5H?@_tj*A!6
zvrS2M_s#BmlN`#%4IZLX@g3hE)`dz7HV3~C3)qY6jrE0^zT2Ue3@r<kQbh#o-Yk}G
za(2q`f9FuJ8$~x_3}g%Au=2Q=BpZglQxG{E$~Gx6dc1cyAq<?a=uQA;nWwE}>?~$>
zQN61scmC>xeF$Z;6*|H$rg$%qVs{}>p-AXv5W~(&9Xp4t`qXq;K-^1b8P*;pb$_Yi
zM~HA3cQhU_IQVrUP?_0zdL6VuV6s3-{Q6jJjMtycpq51G=Sk5A$`X?=kwMCBu}|;$
znz1<ConzCh2#$|q1Sl-SXyG#rw)gNU9s=yzqKEyct|W9qAWf`9cw`X3%B}{hFo@B}
zHh}DgX^OstS!>+p&Byu^1v#P*cp9L;tcE6@0D2dYUNy_LXg>gOA0;R$=W*jNBxAAE
zz9!onF#CC4C}>0>?V7%6>fBRF=~O_n?L(R^tg}s)z2W8xNwc~=9=9p3S`HIwX@9yX
z#or6dg^sNn;Jv4~h=*c-CZ~%~6K=Rs2_(e8KB0TH)FfO^eV>gjp8_=q8c>-}$j~It
z@3bns^u}JvWSos4lVaPZhVzMcXKp6NA_;b81e5PSVxRU}iVL19oVO3$Btx;c-!$mB
zQ;j%bpzM5%nQ)ixB}}l`6Vrweei5)pu@<n3(T?6$)8;Wwe>-j7f#qY0nEe#+wf&c6
zpathz-~Fker2HYjR|HSI^?U5Bmqs{4mg#wNIRl0`iXaTxi{7J1bfp*)7f})blc`y;
zXDIX`j#{>(tl3$+vXP$G8Ii^EBq~onArRdDL^K9$1S8@q8k&OM^)4~9C85c&B%Sol
zy)WTLGk(94nI97ep{kJSVTwg9MKM=dX;;T>3{fhx>NyRKS}LESU3}Im5^;QXNxPIz
z6=5mw#@n!P9f}!l5CPRhV3|(|7g|5~ObX2Z8E;B5S^hHp!&s&K0;0O@I_ALra2IW2
z@yTO{4h6-)115WWb7-qd>v>5|r*f=bi$uk3@rb_-XR>?{DDHR~{x1FK>ei}apB#RZ
zxI%}DZgtPYPyxwlUqsB?_QBfwdrhn(DhyPRrZoAj-BnO27}~y89~HRB2`dZ)uaGA)
zZ}csELB%^R7Tl5x3n$eHYd-4Tm*LQ7jz9$(OeB2+G%e?Ka}SB)U6^Vt^QpzPYvHi<
z+Viu%BuS>`lsIY6FDVA-k<UGPUYRtE#{PYup9qAJG+`m!0I~NaZn@^p+zz>8Vp%wb
zvcz)s%9-uxw;xiV??hn6qJ+==a_Qcn@4opQe9H?AdrQk@v}=4st{o{5db^FfUZ3zn
z>sNSiBi{NUb9-UgpV9l<wAr-el>Os5liPyTW)wl@Z-1lB#&kQS17UHPO;IGR2uY!2
zkA9j_gWby7Gs!PUf5!1?NZLR&GHcQRCK%7l4`}z9Q}+I^v+66{OG?7Kiwju)j&Cxb
z)u^sJctZT_8=%c`32vj-@$FbxwE(wky9P-KbSV~;`EcS3!tT8n7N<rHj;gTS+lZAW
z_f{h|@f)#_9#c>cp~IC)m<hVAPBGv<YpVRX$!khXakFIjW>R2_a-=dCmD<EOEnq_E
z+BQs8eJ=+;JD#(RNBjoxZN90F7antRMMysbN4+K5Ml)$3d9O}q0s8y=%W<{E<)lh2
zxDp@DwvV-kU!cJkYTbw@#cKFpeQr}r41$^@r-q#<o>3Y4#YDLuC@9}XOxf?O&$uDg
z8VKkL9j|-;AT7Vy`J)Hc-3Osz<%?k4<f4+FgANliuz^h*?f(f|O!eVufo!jn86dP6
zaaYI@v>^eus;=0zd<w3$IC7(XFJOL^N=<<OzTna};q9jqr?_Q?kBtz!?c{6d?IObk
za%SMiN&hVizIrPWBW;RdqYIkr^<YY|If^fsTB>zzesiAP5d+vEfAjtW5|jU0{D&aQ
z6QB-|tzF%!#&h{DEHT1u4u5fvIkv=503BqILK39-1m1}X2z!aRrljXS-XeVo^F@bO
z|M>G)r_S9w=)HYS1b2PiVc&vTIE~g^HyI<`+CWuc2D<zS56{#f#)VLe1^EtXd<{pF
z$n4kqgA=s!ao44ttN-_GJ|nS=22YvGqGIVDIk?t)8FzaMh7j6=@zb}S)~huZIfFw1
z&ExN1ywi;mf>!V!vqsMf69PRWDC*4W{+M_#Vzy`U{pmFk{gT_HcUhyy?C7EEyHQ_x
zO6UpoiKw|U5D%%km7Cqvur(adn3B#GR*`SSu2C%jgaA#n$7I>AKOJI%rP+_Jy;ye8
z8^rer#sok$edc?-1_5&eGPa3%Cvaxl3|&q`zNuqTB+E(f@@rv(SM0&z5Xq=y!If_$
z29soqQIAk-kh5&Y&_P<gy<HmD=+4OZhle|=xis<M>{DT{s;VUsqfW0!tzx;sUMN>#
z>4(_pyF1FxQd>2h^E$zyCeEkyF`~3-c1!Jax4}}^H&OSWt-U5MIcP2^#2gTbGm_Rb
z-#ez8?&QyY#H~3O8t=?aQEtr;pE59BdiAUGN&x6z9tx9Jf9MFMe?#wGIW=Uk6y$Jx
z#tYkI=10RqS5lk)Ehc8*<o((jMCDI*u{|KeQVbC8;Lc#Avk;B^foU`CETjIq+bSLO
zjDFfHTBmpKYYS<5$qKIj&K_r@!N-Qiiz7mj+L0+}`zkq@pX@`+uko0DKMQE@B?58o
z7fRU{T*yUF&6P2za*65VH-!}ToQK{GF6-an-vj+iz5($te9WChnQlb44aPs(dX=Ud
zG?8KsI5;0HJK_zP2iF#R=7|4#75Trl01c?r)xx_#0_Vi^m9y*Yrqy1Zi3@vBW1M+k
z3Pg#vZ?JayO;_ovLkxbNLA5@Pgaq;|yDD2{4rD^(DL;H6D~Vk@E{9<-losCNbsD)r
zLms;7T6;iFnP{i*;g$Od#0G%jb5#ShZ;$@)yHCNDT#g(@AWGQhTVFFHD}%bd<B^yO
zu1HO{*H`T#v%Kj&DWbSzjIf|c4$z1Th4#|fOrIh(gU{HtpNE*L)6~zbjN5x)`|pu6
zUa=*SNU@+)75<&k+1fO;HhD5|KbY}X8xBG4*F;(6h)uLcjFDqJAPfc8f5R&FY<gO7
zAg8V0Sn5^Wv;=xq!8V)p&bjFY4DpTv)5eJ|@l|6YmS+kSo2?7&{9|pvTHDHYezK~L
zV69`hQwF1f=c9lS{4*yG=-ZJfE|4m#>5E5e08(2e1XEysy@E5x>qMhE_uD2!0k7*d
zP6?L(ck5y_{2+kDb+u3|FcoNITZxGr`%4F#<MxSXY)!C?mHvZv0AI7pT2cF#hqyj5
zFpGth2y!0L-;$_&g@(UBM3We%`HgNCH#fs~YNY`x`75_LYfjK_;Q<UoU+~@La`wGB
z`bL8yMPcL5DB(YZ&o4WmtJVr28IF{3Ec$og^Q;2@zHYx?-r&p03Og3nJK^o|fWw~G
zw&Bi%(^5QJp(Xr398&()(|T)#LdP~G_>Mw*ULn@WdD<Usf;3%^-Z+M!fagp}u&Vuq
zPcOV9hOSF1WN!i@itVWjaKYP_?nbt}nr-ORsY@S<pUpA+Rr<<0V~J_}i-!y;B-I8}
z@N;dW`nRxAJXA=ZDNHb(#pCDDTf3v;p)@1e5LjrAQ>Bpvo&Adgw>NRj;D%-Olo}Z7
z7|w{llxi||5+%L#sG~0eN2==BdzF86={9qYy2%^N+3wyxbjo>9Rg0_gY{B7$g)d+H
z-bw^q@h3xPzo18P>WM)r4NB9emG9XbbPau4cZjstmxRGhRj(R7%gRdm21D{-y@v8f
z9x3;|<Oq6ja|k-2<|pO0)ysParOZpTjW@Ses{V166=@5l<$89zvioHQe|oaBE4v*2
z`VjA_N7=t273r6#weH5YNNkxKF-r48<hirrZW}uQ1k26FLzN}As1M{X@b+y^sbH@s
zu4f-dw6A$?HG}(UZ|gOS)-$B`(aO8cLGt|}{EB>SbJbtzEPzU7LhJYb54B;efwpH&
zg7XS?<tH-B4M1ii+r!H>_Z3awGd9><S^dQ0zIM8msoj5=9@3q)ZesY9h(upACc00-
zoZ?tiE+;UIZC6%9viY}RtY@?;<)5}mXDU}Q{Pr5{TEyZYaRXV?p7mAVYijtUPW^6R
zaxw23&BWEu^iiA%oheEi0_{hcAK#uHd?(J)+gWULW#Q^=?6^E31cIFgHkUuR+qy^_
zlzl8Orx*&sVF;leG6;(Lh&`&A?bqe?c)QI@_l`G9Ju~fmvwY+n%bh}ZG*+*4&>KN!
z|6p?PTBp2pgx{`sE+$HO0z1GlHX71<HuEKieJZ&2zNN9=e$$}g<B&)@E;7kny6v}-
zUBRy|eqa?(7@XPD>v;r=6VZ0>L`a!0q28@f+Mp}ugq+1wt?<a#s8Su}%gp|cUbmZ%
z1V7sZ1U9GwNCVWQsE-0sxy#ERxpaYN<gVjV3gt~#8JEAGoR<%oIO-$~F-(=7-!0No
zL48({A_97`q$9tgFBoQ7GIA=uz4VFSzI|j|T=XP>=#LT>c!x*02V0{B?_hRQSLoka
zQ`ueDuV8L}Nb>cYX`Z*ZJE-XKYGX*_EbgE|J0q*EtPL7;c9S@3+`??Fa6geZ?@}D$
z^|}we7&LqPvFE9mx5XB=#JG9?ip=rp=U%0y336t-A&?%)WSnkdSws%w``vCbd{iiO
zX#B>d&OFoGu9xdU^6O+PQ=`JS4!8&w@uxmvEvkWrKl}jAL-g5qM3iBblQcujwxuT@
zf@USR;&59O<EQG*cp`R!Gn*DR^HZ<#oJy{1cX_|MN#1VlT3Ef0Uc<-`H7CI2(|<zT
zt7=D~@sD-V0bS~qkec;3nU;HW&gmLkwtk0Q3|#Y{vcPUP?}(;=&Ltx$Ry;;R{G#5}
zLz92K%$`u)v%twZ)Qyau#;V)<KZUiS!4g;4P>9qIz4yIQk2<Pmk0PR0)Mc<bTT*|z
z+PIV4@;4G&BYgx=<_?cD{`-}NKhP}ug7acw7%pui!ne5B!4iu}fu;I$!_Qo3ysD}|
zfaX8NXpY<GeRJxY2e0I%+@Vt~@5=M?Tu|(o5ZV<R2^zHp0`Ul@i3C{HlmgdNu#<1l
z0JGDaurA#yGV=Ur6s>kjD0h<ZQ^x^4El0ONuyWM)K!;lFbs1u8%Sm;@L4R?)9I=Sb
z@GJnDb=CfMMeOEngZ&e5z*mzmD>$bOqwk6jhYv{(0)rd|n2(z|DVFO8t5SA_IYy-h
zT~H*J=c?*7esZo~>kE7!5o?$Phu$)k(yF{&**vCO)(zb~%IZreOZDW+tpxJ_mwBVd
zw=4XLk@iRJ=@bfi2S;M)QER`RSIli=TWRyUl-#w6$DGG8139iz(PP`tf&&|@xKWJ3
z+JA)BvJgqTb6!{L#vHABW6|#$HW;*pCcI*J<^k1I3C-~nzB9FbBrnPzOe<)By6$A~
z+uipiWE|@j*}1k%)=peo)X)9Q_QnvaV#sE^ZsN>y)eO0xPp<Z=L1D<{F9JDeitCR4
z9Px;TOrk^VmkAel#Rmg1YEJU%bW9qGNhDzFv0GG7b*E^#tQVkbr0;Lb#nBNJ#nBi!
zO-+?4O|hBcpx))iw~dH_8ISjLE1IUoA1tr1t$z8QoFQH02{;&U{)?~to!ka!fRtOX
z5biuP$%z0Ffg0BH#GkVirGZ9`h(%X_`lgE=)0GdYG1Yxv0vOz<@wHGabzL75Y|53g
zKMH+!Hf_fnm91I8>$(aYiX4d~8jVr!wc0vdW<U3UnhcqP*n^+I+2_#>hI0NAVIsb#
zx)t(5nwO^GIf%9Q;YBOeT!awYnlyWlPBJl&u$ulWxBWqNR9a$$u5>nokFvjmf+{(G
zA`0lwIZXomd!>fFnYxjnEqrmANL0otalSUm)RV)A!U`Y91OG&wB-Eu~B3`9H$4W;j
z0`?N}qIU*=1$_yDT&RriWLXAAry1dHv!mR^y;4U6n)_+dsfEF8ENG<7hTd%Agkmnb
zgOCXS1#D|(RZe<+YfAAyitL&jUu2?qRd;w>5CNBDMCzD6Q1*jm!%@5%5Ks{2Q%h!U
zjB`Lqqs+U{e#`xVDnfG==Z52hGfA+P<O&siXWXPNO3wTt<NfW|i+R(KZr;`jQPm|D
z4mdVbXZdiT_kxK`Ocj+xE**Alm<7ery1LviekoGA!%Ez-WwmE=*__b~<>ibp@}|Wo
zsSe97>I?wJO@oPkSOko*m11Pe90(z2KldINM?jz?$K}K5bQt$%8;D>OtP5ey0{j^Z
z`iG#rr-{OpF~==dw)oclDU%f+Yj6#kwc5&y)28gTkX+lips1n(zg&&b#F>k$tNK{G
z=A#P^w&ZwKHL+7U=Y_rQHui5d-V#AL&SMc4moWO4*IA4bGOa-ZTkkX7f7lEg`)z=F
ztrQaqqxmvq)tPWc&!Uat+}`01$*)IjY|8Y;DQ`d?p@|cB!f>Mh%K+ofn&rL`y@WNi
zVAi_I+ddJ)`RzxU_FZItI*OtDgr=9qDiX8Wpcaw3kK1bp*`VX*|3huMD|ZLmX9Cm#
z5m4QA9$apO3}}v5EQg-A&F1unu+Ly#x20r$G&+lqb4gJ(VV_lrYRR}Axn`svmPW09
zv8nNO{noo~GM4`)wCk!*y<5Yei&1Fq5a~Ig9K!N@i#OY8;-|uo--&w^2bD4)%j~`N
zh#htb5orhauF%hbT85%6uAh?3I59G7tIdD5ja~Et7wKIIfY<RL>*Zt0L>w870HSv#
zsO*1b4A9)_eExA-g+&B=Ag&P$f#i5a%bkTLXIzB>K)Oia#5da=Pi%d8U<ix6suxe-
zVzdB3qA42<n4@1HAG!Hr&mikJNjl-?tDS)R7G`>SQJ-`tlw=`%F;Jkj*5=!}?EuN%
zVb4!no%VkIF$w-zRTA4nn8g?7g?pb}QG|V3+oQ#b=8@TLos*|B`mT~W4zOG^SI_}6
zq}|U6BfBP;c+PdM-z^wu<(9Zuh<MbE)a^$GtAqdhsQyUKS<Dmg8(6r{CtTn)p}Zu?
zzS6z5O2pDal8?2mBns_g45NdQ>0loy$$$u2uoC6jf=n#-mD&meO({)HZi-3jx`^Sz
z5JenJ<hCN<@XgJ!?r%=mW{&8JX0{_`xnLth9u5<#{0H^aYzDnfw)AB#t2@grd2O*F
zXh)?gKGo)nW~B0vUkZ@ZIMK&@t}^0--cJvx!>t90+E*xQJDdS(cCn8;+AQ}^ntvYy
z74J(XCd<>W9AP|H?Lq|)x~5UZRf?!fsk}5-HW2+1ML+SZXaa6t7RifzMo&G&uMCmI
zM`IGr-ZiR98T_J>9c67^5>b7rCrj*SRuadL5u^k>u0>H0Mr<JK8h3c^<4qc&-E4MN
zmP2R`uRnLWm9C8dYljgT7=^x=YK5zp+ia?7QY>P7PRyAa+}uRYiSxz?>Rv{~zDO@(
z<%a|XzRYB!!pQne1e5m_-fyImdjAjVaS=1-A%0I5ihb(DV)LSM?w0aJz@o@Ym<|^;
ze1sP5oOEVO0&fh~$#>xJa_SW&oa{bPL7MVt@IP17h;`uOcEQL=DhY0J#QBLv&|cyV
zxpt2OU*6REiWpc%gs@L?#GCu!wFZb4!zB(KP%yE?Hd7=uM0!Q9y^9JrxuC>C|7x=|
z`tC!9C+6bcE@(UxPY_7zhH_+tU;^u4Ba!HA>%Wp$g?1he*v<&!)>y3A@rvbgF01@_
z>K93VvBykd!`&QiPhgTa>HQHQ38~k;m3XghenWh44^$td36B2;sKi3!WooEVc&@Y$
znG;!-;wysCS67NNN~m$Ku8SS*we44|Uf`vqMOgc>AK%Oj4wAMYpD@dp7%mRHdph5u
zh8(LMS_e8lZ~B%YcSvayE?{`;59bA2y3;jD?uO*t;chp17@X^dW(hi%@Mllxdl804
zHLwbr>H~Q)_!SP2j27gT4G6%203h;1qzouZ5p>1{GjZ%W8aNS(GcI8fxV9y1#o`ha
zo$#lDVYOVLI{?};KKh{IvUsQgSu8ZX8R0f)ETZ-dz&8IUB_65-@)vl>o1Jdq;1U!D
ze$J|h5lu-OmC^T<GT$XMC?4^kdU`+(p7~??7DV#N1!sGvnl6GC$|?kPxH)dOo|U0^
z5VC-`xRX0PX9%gV{-q<+04*pKC3*863s7LwVYvM7%Ev;2?w|EPUy||Tft5LoPFcg5
zbBJJPgrs9FUqc*>Y#n}`)(UQrrT@V>#P<8mM`oCooLrx^dE!Sd75GxHgaAS><&WQF
z@Qhf7l+FND^Um+&lygez(FUCmhwN46U$j#%@$H{kA3PVhBJSOVsV=Inj*`Oo_szr<
zM+sa9oy;0?1RP8>eFw-%vo?&BH6P3jMB?l=Yi&peMHHo6>L6;ilRreC#5`F-x8cFT
ziY@yrNo2*-;-#EwA1-b50jky-b=Bdo$pa_Vn#6lyf8{}FEai;Yym=b8g{vZRD2iXn
zFdm-Bvm)DL?o`bm$hUC!YFI^3z`0BI4fuKwx1#fCEN{OXQAVRzL`*kZzXL0&>*!mp
zh1iuD2-#V+UGYv#qT(stA%oz!oG_o~`GSrBt})S)B58xD%}uE}PqW?Wm}rB6_w+A_
z-zq&r*S>`L%*agGikGJIsS)D)QIuEhtuC#qKRkjuu!jz|qd!+UGV=wxIC6~jjZApX
zgBHeJ;~q{zAR4ZW_kM?4R6ZlQCK=7=Qs*Ly&AOUvZHf(pyFGK=c(Ye8;DwHll3-Gt
z>lbA2Xx+c!!uAN*^e%X?L8;Dqta=44t#VGNkZ6qgOe+`K9D8~a=zw<c!;p7ecgrt?
zr$PtYq)bcq)2mylHzaMBVAbTDK1qk8Z%JAu!c|E(t2ols^&dYRqD6Ko2opa8)COr@
zk=$IGh9y-RExu|*`2sN~pV{xEpZ5h*f3NtQAqgBS!)sJ2PqCQy(@Wy#$dO_-SapS{
zxb|kWmEk*V<-r2yAaJ4h`C%`<v{c$9DQU@bdUdQv`)B$wn*2ZR#k~^B5Wi4?%PZ(&
zj^DTYT!2SVvK?xeI9^A0?pJdzB2f(Qx*JZk>tv6suG%GHc8`v>_S2(72%^wuqtTwn
zdKjS~J-YnPFrQh<=BN3c`r`K-m0kZ4m<)baV;Zf^<?0=^%(jnqUf5QWJsh+r^91gC
zNhzp<?p+r#prY@WEiZ-P85GT>US+=2KIB`Hx2Ps@*{prIQoXM^*~ERK&yvO;5bnr7
z68NfC=~@Z}V@(u?0%18w4byoF_){gklm^_2E)p=&sGQK6ODl1+VH~^GiF#t_l=Kv(
z#Bmhl3rWfiJGPAd*HGmHu73HssOjDO5_4x%-Pjj2pY;9J)ZqGwrqwAdFX@?&EH=so
zdkzoJ$VrmlL@~6*(c8TH#CPzuR^QK&8uV4d*?K(pfgXhqVR3TdZ@Y?jhg^M`R{9jL
zz0Q4exg?kPj?3|>(FPvK`0x)m|Al>)^o2?l!sfT3KRa+(UJ?~YD@EK6+>XRWYq=-X
zDUKZpl)mqpdJ=w0NOwY7orDV$s|5EW1(UVZ&sz*wYezRYgUkGIxAo~?=A}Arle8;0
z`46dOkFrP!zxn2M3+IGEhA{vInT*f<Oj`$kduTa!yPI>Pjz`<Cf^5IX`y!<2&plV~
zP8~^1Bnhz$D;j>rA7SQ2Ne_Z;I@IO-_4M_zzcWzQqxtct|9|FD<{G~va~`YQ`Yn+G
zP=NhZE;|ewpjM}&Oa@Ln6X(JQ+Ar!6I@xx;eQA*V&YHPlh3aiJaVe|abP@Nw*BU`K
z@sO&;(_)myN+u0cH@(ueIJKgHSy>brfP*TgWO8k>lTH&Cz#%>5f>E$lH`C#c)oP|Y
zKZ#3aS@DX{i|r{igfRG@|73D~9NXdu9B2I4UcDtFeI)hz*cCnT^Ska?Ws535bM^jx
zdq@LgN)id+{4fAqHUn(@6Rq4f`YY#{?d!T9^WXF8wRlv~CJ`s9DsVXRj>BUvJ*qzr
z>QnPh*uH_Ra5~XPKJ#?L)<K%J5rF{ckzyrIN{5a7i9k-DXz;O)Be4bOMkF?rK{6)0
zdR_+?=Q9r~)ZLrn7}!j3j^enc-$yso&Om^~<Bwd~jwRpS9|7*Q$@elTp1A*0pV?Ft
z@pccXdcPpxwg8DXd^AdH4pzB%SbjYvhY92}`|A!OCEL(A6y`p_lcQQWCGSL$^Xftc
zYL$UgbPV?LD6MLLT+@RFoVn@sjb%W8VccXVtL&^+s@5aEm3k&lsPlxXk7hC29qL(x
z;H|rmq@*!#plXUDCqKS~J@>+HD17${YO=A1^|M5?2W3Mj2cGRzl5Nd1zeOq;t!7uD
zreE!+4}HFP^!GOn{smbEc_K%~Q7=|)5*kcDl8^6p0#Z;sV2G)Qs?YA!WMKAduMS>q
zonZYm6gqg&II`L3;M@~!tFVeIG?P|5_qBGF3clHvfQcqi2+aIx>aM}ojJg7u0OzC}
z=Ul7hL53g(ax6q$6Wuj0cN77H{To+%?&<fY?$q=AvFZO&&#Yv`xU*<F1Y=c*=yTNG
z^D`|Y+PKBU18BwvVLUOCJn6D3>!)v~Y4Mj-y2%P81I4!_wUp*$d;c<RsDt9lr}qmH
z5|o7Y($BaA&p@S#$Sa&6p4tzpzXEnTX0MWl8Izu33FKEe);{f3mC*7yQ2SB_BTHv>
znU{3yCxBKIpUF;~@L*y1wCl0j_h5{CUe?SO;!tpW`{!4I<Y0Y&APD8K+auOI8c-#y
zZ*jNVF4IobYtG{bXt0#M(ZHMc=l&>4$*RycOcSMnyOyqeb_k>HuB)Y)OQ>Lq=}*Fx
z${~}2C0U})E;7lz$Ec6@t_cZ$CkP{((EtTW_h2;J0xi^rdO{0N79A9r&kH}wUelEf
z_@{(6VpNK@o|jNA|5BbpRY#(GFr)frZ_d2xMyr#%<wY~1C`^hw!mL4inoq#|Z&^_Q
z&NJ^oseB{k=z38fPU8MTCH4d*?UeAxe9;wsKfEmI>Gz=q;PvBPe8}ze^sQGo`*Fh0
zq*|Q+q*$vuVf!2}{t>LTOjcKuY2j!GwDOB&hChY8eml-T3%ChROxy|2E)0Z-1;mX0
z4D94bDl5;uJ9rFOe7Qd_gMpi^d9=3h3w)RV>Ay)^yba~4_wkZH|ISS^p8y$#z??>3
z%Znk1^w3v%iD_66`uybQxCXCRG=kx45_7$z`UF`y&;HE;19=sn&pYGt&vTpt=!VL6
z1A`7FYBkoEA3l-oST9Y6S_8Y@eG|pledxm}*TnZcu?&t=o9W~QS^-0|6U#%<4wRR&
z|Ngr_`%n#4Ydu;Vbqpxybmx6IRh)42r^b)GeeZ>8U2)MuQgRPP5L?xu$ee+i-CuIG
z^ve6VG@(;T^+SplW({AFd87HGLAr>0*%$~8L$2c!tXjC-*z3~=o5+rT9y!tv)P}P-
z-eW;hF01I3YExrQbV>ZiDMYz#>^ncIfu^VtK`=9ErQ_!L-j=#HH`Sz~1uf9zi)|HT
z&qM1!-U5JXDd8N~ZJbqnGjX)W%oCepTr$<F-;5X-(1=tmAw6Fhwe6cLa(#leBIBAf
zj|xX@R!_BLip2qtZO+|G)zr@o`SwZ+dQylBLS!kLva0nm7=CMnQ(L?$6z+nX#0@aM
z>z`;4I7V5#rYEY6E^IExcUg19@>o2`1*)0dpYQJZPcm)DV&S0kb;(QZ08F=5jT);b
zIgJP@e)480qtULRs6@HHN_N2J95_?#b~v^Xvi0*j$_a}nAl_ahHj3WHXf%23RjLtH
z=;dutKDgbRh_ExD`veQ?CEX4Gte&|>RV7dF&p*k|9AUnU>QLov6B*z4gOD7S6lPaa
zSg#(~Ky{sHr^nQnUG8K*F`;jd;wzcni5s&HQ-IYibjr(1%L&QuOx9s2=TeEdLUEvb
zsB%!Y!+>v}VzT}fETMYR<SYWn?s1u`1<R@}Hjj0Fcq0e&TR)+Vz@JDxHN+#rt54RU
zzp038B04%H&rjYh)Cxy7+x%~fNNW(Pd&C3{IUmZ^?Sx`>=~h+Az~kQEnV5sIz>EG`
zICnNi47)24;B+xas8nDN@lz5Wl7$nq%P2>hqmGwk8j`K$Di-ZgRGaa;r87eLdlPAI
zFmFqBMV}nMV%Y$IrnV>33SiPeOeA{)G(eWB^vdMG^7roc^Rsd2)fFa28!rN>`qHj2
zP*&N0-C*th^0l=KC0y%3i7jk#E+#&M@$DS=kEz2^r(0oU+&~0oUBpR4(i^5_RP%}-
zNS#NDA8Rzni~weh>U+07*`hCKa5u(8D_0aeSBm8XQp89b5~TR7xz2uHqif^}1598G
z?Oir8utvf!K5G`Q`DBZPR5*Ts6wPwn%GtRGop@#}mfyDZDSjh?&_=8*{(gzbjGwwP
zp$oJ$b|73;^&&LLM#EZJ%It;pwTW?^Cr{OJYvXu$*F$GxS~D5Zx2=?Drn2oOo(X>R
z%cC~+P#!l$=;+^8gY2nMt05-|t3=H$NC7g044vazBwT0uMAFZhk))AZusNEOB{L%4
z0FWpF25#I{&<e_>cj(<J=#;5VnjpY4;X9gj(!U58GDR^HS59bUr%33KzXoRA(+ML!
zN%bWJ@mo&&f9>5|q}mQ?)Wa%(5R|0}V^A<KF3R$MuelnL{!p`-tZY9w+I-8>MtDy5
zOVDk)${cU@G5oP#(Bl<p-L8zRvi;-tj420{J2L>K252bU*nKr{JrO<n@^aTOy=}3S
zt!6{NY?8ohC67O=4?J7-U1Mj}k;Qe@#lF!%z?EFENs0J~-)eYu@AD>%>*47rp}6+<
z(?5ZMei3xlPj^<cH6<H?jQ4f!qH}mV|K&1bHm3mGeq=A9vGCU>^WySX_cY$Ff=oOs
zzql$6(winLM<;qXVdubZ-&|2hJn}mPo8C*-=4vT>DG&#_G>}M{V@&_Dd~)tDGn-g$
zBgmcswa8%$ly0vLYufQ7wxGO_Wb>|TZpVt%+CzJihL!x9#1dsl3dE|uR|cPlTO7Mw
znQHO0V^Pr*3zB!=!E*5Ge$d*6OsDX9xi_;aOm0M<v=AG6ds6n@EwI_6LOs6FJyQ4Z
z3W=jt;ka6KZ(B#O_^%u152_ny>QPRu9}9H($r%HMK1~*~HZ?^|3)cIKRq-K#BTmnk
ziPYk^y~224+2C;5HHsAqG{O`gwxgAAe_1)ssdOV7MlJf?K}l#$?O!Sqv<j;o8+7e+
zx&1GAO=@+Da9aBmU*fne*o7szVB`n~yw~AADGK($GjF}3@y1@ieU5$;jKY%c;^LW&
zcC&88AMhw$$SRF3{Q{a?CP?559*5lgSbgWJZpOvsT>KBC>U7KJ6&icnBRku36Y7{K
z={eYLQ#$V^{vT~`9oAO&eGlSRtY~rCQrz9OK!IXGi&KibLvSf>MT-Y0?(R@Zfffz!
zP&~N1&86@A{meXLKb!fFha@K_``&ZT-fOS5t{Z|}QwjYrw_qlaULSA+Wco-k?nb4$
z34_Ci?AF2WS2L^?w$&f>23*C1^8WTWSfBbEvey6UZ!CQ$764Golk)@^$(*<~2UhE5
zwdzdumRr_;aRGvh)ylUPGG3ubh%>Oe*3m;MTaYJgB;3Q2b*N@+u%~wHgAvkr)HSRp
zOLM0<NykL$Coj`9w8X^BJii1<V$ln9yHUg;&n8QWSGpnthq<=S6f-}M08}1kZ)PW3
zIH{&KP3p|j;RKF#D6S7se7O+<zlum!>AsYHsbz?z#!3Qt<mq3Cr?Q^BV$Jfq{a1a2
z*XE#E1kW;HSp;PJ<&Lk#>~W#HVSF#edD?rq2>u-4dfzv!fSk1y>Vp-AP<bh!&OiKp
zExccU1(*lMSBO3Kmitp*<p`kr%UX>1`5q&FI3|bd1pt_=fi3mQzRV0%C@RWeSyYX7
z`ixJCte19w9ci=YJN;)jf6YH65&qn0^q$exF2@vrTgIHr@I?VGF<)FJ!6sb@Mz8t2
zmC}08S7AF8rkMpB-67L~D|>3eS&I<91Vj!KbesLnGm*Mo=V0%UsMT2X9*k~(UM^1I
z0HzUxkN$3E2m5Uw<O2@%m#Aj|vK84(&89m&s*QqXmGl9gm&DwPPk<`MWq#rGMiV+4
zm&lqM=FSoy{w&qAOR@#&iW6Mq@adkQpCeiIiIpp?g&K=kmTFf9-<7La_ue2`L|t?>
zYK+3S>oxzxPL}E%JljHy0F^D0BnAt$z68@5;!(~A_P%0hv+IO2z$&~im^>x7s-vsv
zjcv`-kf))=#^7fom@Zc|AMng+SjREARw7Ul>^iOim{cc3qTI5jj%?j6RVhz61HJp;
zbc$4cYI@1ra@VKih?r<+`W}5GbY6+r&BYG-Z0kG9fDPS=!8*bVsx+@JllvBNpIFX|
z2M2wSigu|nVq1{auVx(IWUY`~0mwQ6%v`?M)&QmGaxk->?kOAjyr@|?T`zr~YB$Z4
zFZQ`0@dZ>0e5zK&b;qo36|r3I{f7vu`xgF<Uo;<yqe#<DysI?=1@Z5O*+NYb0`vF-
zjWjZQo~3y;&xxnpOBk+*mv?bx%(44W3~LSr->BFf*FL)$nSf<eQrwaO;W|7d;hVFK
z{aH;WfV}}|M=<A<N9pmBrM%ozQGbF0$vJ0^KLE$SC8>ChoOs70iq(OSbcU(>D*_>1
zHwF*>48t_;SaYo%F%k4MdpYrYBn2OT4{Lz1cmh<oJBJ#;UO%V9x3ldRv8E<aTS%3e
z_?BoGXANOa9tqm_L24MTJWu$~za?1H#3knoJ8fxRui5kl=iXY4mye3NHK;xSn{7IV
z%p$0f9grTrh1)Nozu~|_gf2g0V9^IsKtQdBo^oT~jSrge1~7(u=Dfo1u*B=SA~_^w
z%6{9oe)$(FV!K;{2Exb@=%07f@i!6!72S9J4ZgbMgN6$@od9T8Y&P-ca4Ukq7yMGb
zD}U)na<r(VfKjo2UnCo;q<A~inNJWI;YP7_J=vhEXDBs*@6*kL4i^ED-k?b83Be6}
z?Zk)F+1rmuMGkdqDuOtt;7dR0EfAIA-U&)qGEiPhFrKTmfzwU1NBM%7>e1!-4?x0F
z^dDV!=H?!E`(xYZ%|N&rVfTDM2noJ{63~q_{F5E{4-_ygX5Jy}6jCSM*u^NbsEX?=
z$NwUK!>#ZALJpWiI;e2Evwc;T{VN=+E2_*$Ju0?yX=xAv*C+G26P|}rl9b(D%~3me
zEK~ZQRq?<ynkLet%THeY5+Pwd=bSPXi?xGSP%M>6nUTGZdnv^k{doJ7wEt7EHitWV
zP9YUsqZH|9)_fE&Y&VoFz3ik*cwFm9eytTAj)<}yle(?qYRDV(%H+%LTX?MKSn&So
z1@iWcV-+B=1<Ft3lHy8$VeHlDh4Kw34l<*fl?hZZ$_M~?=o6~_1qTKyVZhZF;@@=B
zr(7dZ_frruc@*)5bS$5hcA7Xi(H>I+mpsFA;MXrG<$5?=^M4h#JhOXw(DY&drOH@j
zx6|uU171V<d7ng0Snc6Cv9)s~B>Y94nc&F6g-ok!H8Y(Y(Pv61KH`o$yzEqagFma5
zeaj`82afg*;ZNOR(H%s%@**rnwR<J@!h9A(vKRKZ(Bnb$wr-hIc5e&2&I7+y!RcFJ
z0mb;K*8A@&OgM0Y%F7p#5fXrmthl$t1la{&dq?r-Je?gpVFPMDtaksqdyNorr|}8i
zr6|maG$@xIJ4GG%vBn*?!Z_bJ#FdXJ&uOeH$2`M=(zdtfE#*66GX4yi#E>PkoQEfI
zujazD&|%Ow8@?6ob{=&M0<K#Zd}6_~{Oh~iP-SX9kG-!ypb%T>Ep~1rlV2oI)Rz$b
zE8@L0pQ|wPw4MM!2lK?81tV7K65zul+@P$y#<1Y(hrUCDetkNo$a8pRHWOhPQpM98
zyWeVHu<WO%i=g<(V-iwO=R6!Z71177g_8&5jbQmjdG>mo$ZGM1>XSfcpzGb;JKb#-
zOG_ckkna0ka~pxd>r<a79}k?^1^7Rp*hykud+p$-#!nYE^uuzaq@BMlyvWI7=5*T~
z>R>>r)pfUyMmvW21}=0Afc#T9Utm^z`*0!S!`uR>Kb=4Pa#3+k6ta119n&;H95rL8
zC1q~JD>i`Wql?+a@Q)_ryx?KoP{#MSg9ZiH-2l-Qpvuskgl{N5VZUKVgHe$489zcA
zg>bQ2J-9fj9r{eB?Fa2aTXcJ~Oc{%8jQz6R<xV~cM*VLf3$82X)N+-K<L~bH>qU9A
zLnL<_POeXgtg1EH(z93n5Ns=!hQ=IwD(gGk@%FQ#J{G#8VAB}#Ag&EEJm2?vCWPaR
zRkWSXK$t?x5cc{|51HWixndxLJJ9RZ<Qvt$kNSyKdF&yoAcK`8L8|AFvlej~$PHUO
za0D;_&zV{!e_!R%1)l}4v#kc(&+txw?0x_s%xw$t)lHSx&i*u^#sa`f;|e68wn{s_
z5+Jbc({ya<Im;LC?i}O2*fSQEa-_Np-O7;Pzt1ui_gy+6g;2%T_5U*m%Gcd<X1V^L
zkd!Bt(;I2{d`hy}n;%4plJ?1jEcKUqw)gHAyiUr=({Ryeav)dD#!93p?n}#rC4K9L
z{KE0!K{Haz6YQq%F5^tE%C`rzI2nIyZ={O99{K`S*joPqhag|;vEmB7^ay$NBmV+S
z>>eIa81n{_ldK?)^Md2H{}3L?;M}TstO-&`wFpxqhyjo<Y7{v+^{Z~G%-ww>uwjqc
zDZdjp4ph`Ibt}~qKmm6)7MJiO<tA|gIz@pNc5zgPmykLYs(Kqj>$>T2`GPH#Uc0>f
z%gR3HgW}Q_51=)h^1it$A+cjgQA3qD*4qZq+i+y45ibW!k+~e6Dtd+pN$FUj7g?KT
zA_%S?XNXTLQ(R^QPsipLQU#%>X009FY)a^x{|Htik^L?$SmAMfsabCU&l->a0{L6L
z4GQZNB-=k7WZl-xcrF*aRTT-cKJma_4eky-IuDwQMbQgcno|7C1)-liPNr3EpL+C<
z)NWCuWKloLL1*Q3ndl2st|avJLyj3?$qfh`I*raHSz0}0gXI!D|MrNte}zZJY*=Mc
zJ$kp`K9%OQQny9k)B=SC<Zb$YV;*5}aAnj*^SKs&C*<+!&ks38HfL`GSZ;BiII(?}
zy$fvh{tP1;Jy^kKMY5pJ8ISf4$gs&l)7r^6uzo^Mhd35LhN5e_o|RHWW6MsgCdNBs
zO@W)r`cd?7t-oJeVD7z}P=~ebLc=~I<%<pF?E2T;YMVASxViPO=NRnM-}QY*`5N=V
zZ(XYUqT?!V``|G=Mna(pxegB=BkP?gRZvKJ?@@`QD$+^?_Yhkm?G$&Lit?5JshLMk
z$Ey))^8K&A2p0PNqejh09<T7ARq>XL*d@;qZb(dVW8Ax9wdi>eE8NF)l=6`G<**p-
zjea+@#qgBOCx$1y?^dK;bZ~|<{l%L$Uaw$(+xV@ZP&c<Fb9E68gOwH@i%A4%^a&c|
z{IqhplQL^x5M(eV?5YkzK_v<8Fgs+c&F)&!O0O#6t#KVnSFr_P6&v?bVV#Vg&`r&x
z2`=HF<Y~agD!*_6;fBW)j`pyrM6>4eB_r&jsYDxp4i3;7r@$WkpLLx-$Ao=`*P>Uw
ze24_LGvoXlJ9wYvy;aXp;f&i4k8c<(Z336*VhayEXm_Ks*YV>=lbs2Qivd9&(e!kP
z4?h*rqpg6fF(S;-jruhyd?>srDyMv|cn1&+5NY3j4V>6xNbF-U$Aj;KD-X5!a@`NI
zZ&fASTb@&oPEj0-_QYiK17diN-gm9S-HWeTOr~aK%9OI-{q)-wi}PH_#i`$M^Jm>3
zrMP`ph3s-jgbwqY7Ao_+eMZw&A}HyP5vsxRn`;6WInj;P9>LOq$S}rSJpIyv=n;Oz
z|5cH(c&+HDmh-Z#7dJqC_iL{L9c-P_|3-l(#v0z^Y43UM!{1w9F@J)Cm7yV?EliI_
zIBmR71T@jI&YiQnA_@r~Peyk}hxYHt3GARP$NkJwT_nCsxW0U(4d`kK<|-z9F^#?^
zmRq11quDbaW(<0s=Fi{Tuu{^pogay4jK$q+`5h>rK|THrP(K9KAQe0mmrzQ)OHBfj
zKGo+aMHfvV%@z(K0Sn=4&@I4SE;-|r(Qy5IDWjngvZkg<pi+ZM2xR5{C>0NG6DZF)
z!2Ac4@OfeIcd}~iOXSk;*+b`q<Frm*-5hUs{+zznk~iyG^wH`Vo3<->JrD}}>9jff
z=^=?XkPZP}WB}7v!B!e)O5!<ck*;+CSM`X<Jo<$dZ{{LxeUa0E3cq%s|JWFD#3wBH
z_5^crqG4R<r7hg#hPQ<hzb4&mQ3ux8pYI`U%EEYL4m2gPlyo%g*txVfN``-|>NT=Z
z&t7fT-nuoVvJS4@SMv6C;Cjfo^wH`V<#XkxtEp{=sZun^80j4Bn!i1li*LRa{L<*#
zZjoqx(D7iC^_)jXWs{dqmu1^1>F8Iw4N5jLa1?OT(<6^T5OoC^J)ZqlSLm#}b>&CL
zNJ)6aQC0@kye4mi*ndt*Evar%M374ck;U{idHzo5RY~`KEi}XSX%_tGfYt&E<6BA4
z{#3w19<mAU_s4)y1Ibv07SiKG+sSk~p!zs_zc(o#9($Pw5Wc#!>eE=}NPLNy&+WQX
ze`;8)5TvB__8f*GFs{+eGGM~Dqu~t9>w~oopdyqbtTysKO+Aky*U_l?z^2r5{M_I^
zpQsM7`}h?6dW{B|kr^JVQQUyQQfVe6vKL9~?<jKOJ{vU?d`@6eS3;!37-VXDxMr$-
zet-dYt`;W`<flJw&6SqmSS9M#*g+jgXFt{(#8zyj(?N&#Ct0mK`;t#Bn^p}#t4kX6
zs?-{d-uC5y90+m(2e~~fdaF95ipP4c*)0y4(E;iMMUe*UP_Z8@78VGepL0%LLl!XI
z*x&WD)*rIM<9yZ*sPwnXYaghWq5nBCp{@g$ZxL2!6^zh_jH)6?qDBfpnNM^kA#sL1
zRrF<i@5NbcllNqBQVMAtekO{}*trg&^pN%ON<IJIWgW4d;H2i*?0xW^{R5f*s5?18
z>7(C+vz5JqQ~6_XSC}ZlA4kj+#ukYH#veHn094p%;Yq6ucjA9ZS9@?i(*S`-4$x+-
z;rCTx-8-snzL0tE+&cmpdK4?pGfV~Jj2xlzbH+801wUy2B>gFAar?cO(&e|p234O7
zztR(#!p+}zl?Vz>gnvPIKU0I>D0UMr^9*ry&60v!x@jWw+|zr-P!;6cP&AU_{{R%1
zu?#^D>>qYky&_Y<rU~vkfA+P|>EHiq^^aLSqsI5%VqJ0t7;CHfO13u-I6}q|6cDmP
zBoR&X7mQR0Em7lY?{`y#BH?r%bK%RG^>!$4#w;B_!NUmln{49pC_-~iUP2<#pwLE0
z<+ps>!t)%&g&jd#F~vCV;febm6yy|J#AV{XNb}ZWWYe}QaR8~r7Z5>~hUZC1T>>%)
zq8%z!OQTc<-izl}PP4+LRI@iYfIDauRnfgYvlzCxFDgv-6wUY5Um6Kmys*P#%|yhQ
zfg197(!!!JeqiT<TLf_bkl}CYf#O~cEQJYeu>I{{_v&4ckQ{+h2|LjHcnX@6xKPa!
z0fl6rZsx*N(VKT4^F<H+UhKfx<34klAR&TJv=IxwT>ULh>4_NUb^N3})9+^C`sgz}
zrrOqH`poipfo^H_ot4D>$Qa2n5&n&3rj=2fw6lIm7GWawXSTe|Z=patmaoK#O;p7g
z)k4caqcOlLxG%<VjU+fF^?j%qmCkquywB@-*Cc`9Pop_zpeA~^tPb80=u0CigXtah
zG;_~iN$bA{CxxYm|I_VhaD)Mp$`O3~T>muDdr?B<#KHZm7j-^mBX(+Ab~cUS;rQ<h
z!<Gl!&^ZR8Jb0Omm^5%2r6QYIo?1{y6S*pgH%!r&ziPWwO71!w7eC$wF4br#Tb0Lg
z(V2IS8HOXs$B>l-a?kW6+&c=yZCcp53;Y`|D{*C!?^ae^?}8G()q11cHS<0bl+4_u
zoC>#lbjLr`<^w6Wew^jZ%2}7pd!zeSah`b*|Ie>wgplf%Ux;%lx#sb|7L=}_dNdh3
z3sbuZjbi&v!&jFtk_NP!ME-G|3cl!iMWu3Rwg`9K`+**PWVMw{Ll?Kl`DcDdT9=Bq
zVuLfvNl=3W=jBkM@M0fM+AeO3%THK~Fp`D5hDN!TemSSS-}9|!MTONRQ<{^Syn0s*
zIB+}d?3Is5FEXDag5$@Jz&CE3(T1`^8$he%?n>c9Q4U{B%ax3#N``wpw1yU`PI!b`
zII{4?NInoRjRSb_x0bnm2%q)GB*qr&mE;{-BJ=mls{d(dAc@kl!R<KFSqfL`8gChI
z8VK`$+s1^gmGu=q|IXhYFll}~Zkbr+|A^hOPSV1r+wI3_z1>&qvxf564zk70PHhfh
zBRKo1{1dP}g7MckOQ-~v=-y9j)cT}W{HxX25Fe&Wz<HuDc?Fq%!r5sWR3D#S=Bjm2
z%3_b!+3MkDdE`51$};8zHWux5zcP4N>k+`%%1zb$h2L~3l{L4<HM`IoiGAZc+kB<p
z*CxQ?46?kU#jub<sH@Nw&qlDUiY$!I33UnHy2fvVxks#Jq{*u_)in!CF$0SXaBdRv
zZJWDll`da9KLE5Eg>{<5FZ}w#&yKHgRdANW+UA#+YdJ<z*`4CEJI(V#mHv=moW;Z*
z5?2K*=SSm1=t)!a%)bmskOHl*Auhe}?67htzf{Y_xusqG+}<dA^|b*zOTg%5hxD~+
zz}&=P=9V(zlcqd(w@}%GbHz0pX2QOb2x!VH`tCoY7jng{?oy6OWH?;mcKH5U6F|m~
z0Yfz!JPki#z59{x8y-t90dgx=rq~p$b7vE<E1H#4OQ*V<XxqTiOzn#Sh)hI(UJb40
zF)yPv@_o}ob(x$mnTl5$-8`HW1x=RRyeJ51S$wBca)U}HIT~}CkA2oA6Are?I=a1d
zY+RQPjU^Xg4k)G4TBg*~H))z{u#_4|E3rT_>vzbvKp8)e-k5(#vFMYS-u~G*!7;Zi
zVKvmV)kYTr(D5$7r6q$}n>v$^KLe1sb;;bCJx)nUgM42_;J6;+M4C~XyPCe`I?Df$
zAGa|df)DZYE}LQ}Dr#gkA7*OZRlC6%B096~bT`k08w;`TgP!BdiuA3x+P-%{q9wm1
z|2!RQDJB2a;nf;@$HDQZKfyt$wiRB&{3})au6+4Ra{jy)wyLFOEak+yY9Y7cI;TW`
za%X#j>WGHAdTzrk1{z(l6%*3^DWf$eD;{D#Ywbpx;U@|7>Of+ZoI#=)G@kk?H>xPz
z<sI|c3$vF$EKe;(mZ9%Wq{DSo2-Jl6DU&U89vJn+bvIr?Zg)$trcn*lO7)K#G>A#q
zXp8Qh9ti%jBk)NuVYnQuu4g?65t!Q4kVa@7)nCgm*l)IprL#C?rPN+Vi}Ja*hN$}{
zZ5E|UBiAN?8`qEgO>05Z6xL^9G#@>P<SxQ#I0IOepW7?9BC9tc=ZAeUb%3``^72=@
z*#6Eop_xMR2rI-s0t1d?F6uEI`im%VhW~<=A&JodQf2QdN{vxuedvaBin;O6M|ssB
zMMgHxV>AK4W`^WLezIt&b&1=+NtRuHE-Qn|UyjD%%O$|TA{bjmlv&fR2f0q80zU1W
zXqK%w+p;pG#Ht*uF`*+P)l%4SV^F)Zvvu8*#xUV;@aNcyEt|whBjpp)OFF6Tl~42{
z)G=Ti0@ji0NH*rb{3fX<tqPC$Wl4uWxLv`wOk=Ip!yo|lcB)*H4PF<53}AzF05|cG
zo1A6^G+wOK<$OHrpIdu_3q|z^u?Bv?8<riGenfvTex+(66KXDjp4HrzC(v4*{<2BG
zSt0C3UgV8lbop~t%<-i|v7zZ1!HWX6=(Lx8ePls_Ij`d_RM?|Xknb%@)qs2#Su@@-
z9wA8#-u1)Ius0Uk!n%4(-x0cF;9x8ctoN5oSTP_!8U4q=?CvxrohVTzEZR30F@JU?
zA;`pz?)4Cm=&ie9$~*osK(-Kwk%$B&`@<RQFSu2%pF@`yhF96ZrD`6j#RwxR<iN!T
zXz;+M-{}hVw;ifR&x@BC`xd$>EQW8d9D%R{Ch|+sGmH0+cXFiL=2R7@S|#aU1z=@>
z<GPNx<IA>GUyW8y3N#Z`n{>9RfN#Ks!dcTsxPPsoPSxd7%rnTw{+{49JB2>$n+vO2
zjHURcTM-Mx9X|(f>mS7#J*bRg2!gToVfass5>(M3iDkyB2^kd?P9^Y!yF@=a6afz^
zj0p=Ms??wg6y}^yyfGEsKyRz4Y&;aqNcD^-S`=CMvQ?jv#S3=6VH(OoA+xm{5j}4A
zw4O`QloR?AAe?qF@EQCyE)-th&ZS5>175CbB6ULV-e^A~lTI$ypPL`$jeW$E-c4S#
zdqdCy@&gZA@1+k1u3#>m4_SDA&tdg=kzZ(USUcXA&_rpktA=;1IgSy*1^HFX3cs^8
zdGY8GLfT<qi!b{Eaqj2J_flJDpzrY1-UoywJP>E=PHjKJ_g(&3cR9hJs$SNw!El+~
zyP7ir<s>1S6pj8J86DX$F^K>YPL*P<>hb~3UvFs6hFk)$Ffiut5^bMGVSy-~p;GZY
z70w!YCH)kra$V?s`pY~JZd!5d7RTby?Fz~MK?ykk@fszaQ#@Gm2RiA0Wi8{^GggGw
z_C@f5MI$1-Ht&>|lZgHx2xN61J=BQRj?_33|5}r|&NNk!C(Uuqum^f`QEqzrK=F6~
zf?3C^(%n2o#<5ADE^@M&?TsQzLOtVzzi8L)#Y>e#mQ1Cvyk8;gVSzhDDksrJ-X9ga
z{l^H*$^aP)_G^2R{TELR^J2S7lyhwOd>%VMmw+HYTJOUgO26r4bNMx^7eU2d$B3#-
z*7T3HAaa!4z$a0|R4eW=An|5zfAgStKt82zLV{5knLNh?U#~g?FJJgZJ`%x7N9M*b
zW-_ncB{F;$Kl^*Z&%AhAjvATc=N~^{$koNjU<q9@i;4uv5ZPpOQM`QVKHoG!_Vrkn
z5Kugh;D9fZ9~A>*fht_ByK^Vp7Y;BX)LF-ak_p%m0vH~xRul3)paxHt8m<<^N0<V;
zWb?J`l69iVv|RQ3o{5fvyq0-5>m&h0TFmA39qi!h_rK0AVq5IqFn5bWu~_#+imS<=
zp?l!COk|`wq7i=^wwJ(#!gYsNYM_et+<HvG>GETpe7^Lq{DReF*AmXlZ(>0z6nh*@
zF!>DWjQ)_{x9=Z`jUGZvQZxe-MA#yA`%XtxlKk!Bi9rPU*W~0DJ#!<j>pLA&an4Y4
zj|KA%WtlAekl5qYL+97ZT$+VPlI^?B$LqgQ>|7gisNRrdoQ200I(YJ}$XY)&1nj0I
zwqToGm-mtJM}jbM#L!C>yrrEej*<p##&?1FqH_L$7c&BL$EUy|1%wUnoV89ckyYUV
zFOB+N=SpMY-?G$(y9m<Dhd1h}e3;M+jKoJS$4V|&#8iMqtbx)Ud3Hl^7_3Ywm{{$}
zmN__4AALR#$Pb@deYE0g-S4p03CTadgftu3x!d%Z@#TxUbq@d7hjlv?SFT;D<0DU;
z7Y~2LgSKH7&4bLbN_FWd8+Mu|>>a#su2VT;IeO~=j-t-cSA9esZ3IU6^DhIS%1_RF
zv8vbr`6$TX4MtHNYLfp=zisjSHr!%6)tly}EPsEZ#3zFLdIPd4)`Ac{Dl?B_d_*Jy
zqYoEwGIfmYkyHuPF`NfwWOb;O1*zC#T2Ta#wIo)pp^)OgUHNCjGl1i+NX$)y1s((_
zD5`zkg45|j&lMN?A?*kp06@u!1J|GDeTFPIjBTUurA>6f?1UeU{&YA1ubse)jNn)j
zWgG2Rg<dMWBe!a~{-*JIy29RK`1Rh0Z^EYE0bU|d9sEM&&ZMQm0?x>6xfY;{1BfsP
z_O)-!Fw{knxPv^bfhuzzz?mg>2@(9=CiZ21jZ0mHu06#5?arl*JB*Wd(^~P&4ztjD
zK|L2}t*|q(3rx?5CE<%+)&veI)LwCg`g00r1;K7|f*Ph5&42l$huTzH2TPhZ4I6na
zYP{Uxd{Gi^_Uq3BX_7%FCsr)g_lZ)4t*>8)Gllx@W_~CmS!@$&^4MdurtLzK(}hr}
z0287`8KK^3Hq*K_sTA6|<L63J0lq=pSg`KNy!0|pPvF4;|2!k0pn{`%QPAxVr^$1C
z(GrU~lR1Fp#XVO|!nSl`+f6s;s6x}~0Z=jRbEo95)agu{Wa}aeDR0e2S+mGza-Bhc
zEA4CliZ@H5PS0SO0PsGOCxisIpQ{{ykOEv>xm54Pc5&U#ZWV0*F&a5Q4kNAKBm)=}
z=6j-S3o`>}lM0)dnbz##N1|u$b-PvD-pse`1T1+t-}01PzFVs3>6zaPyXZxU*;nLw
z$4UDbr1eMvHwOU!{(w7Z=?0`6H$*WbUb!qqyX_u=2{gJ;@eNt=Q1%z#K)^z%hNlFN
zQ+!(`{7jV!AO~9zT{46A_$ckv+~<Dt)X6^j`*_^1ivpG&087-Bm@QhG22!wa*rOIi
zPacyVB`qI;0Img$O63LL`TMc}XkOSz7=k=@LGTF8v5bCJWFtKl!j4Z=m&)jw96+?a
zSR>x{I=;ygzNSwt9+~@{(G)34detL_t3M(tHf{kyZN}VwH}L*fVK$}|B0dddR>w9&
zlI;u!NYNh>cycNsy8MEfi7QQr0D(0@v3Odm{icL3hmSHH6gw5}g<7ug+obOR0EQCu
zi^u!u_WjX$^AC#MyOhN*Gpi=*4p_-C0PpPK0e@ES<NlNL%iJf@-2^SiXPrE#Vbg>f
ziEwN#Ieb1Z!jS3M3-y`WFdzj|^|urFg94>7AuM7Rl#NnY46TMff6Av`sW1tabNg>f
zi}jpnXYSv7<XMV2)#1SN4DBBQd~sk^#36!9)(GXR=c*)v{VBJ&;!9enTClj;42PrE
zw;rS@xBLn{ucPF=`^iNjD#ym#ygj_{#7%ORLdB$hraRP`kb;r{m{xK;>8Eo4U-2RW
zxMVytMU4N2%-BzVO+KrVJGnR{c*RXuraj|k4%fo-a!ZK_V&yN?cKOTVV{ZjTIlWG3
z##fqh#Lf)yke4n0OTmbAcJ~;C`uGMgT+JLPElGZ(?&fq-w;W%H8q!axmVWdc`D_)@
zv;DokFw5td_wuBIKeENvA~m;2Li`avQk@SL-z&n*w*gNCX;yfOpQTn&B#dhKhJ=Rv
z%X5!Po`Bh;T{I*X91+QYY4N)Ms2ae$+MN?Eiesy<a&ROLzX85)nU$ajKQvU{r^Nvr
zWLh4JqR2ZGGS}4O`YZHQy#AyiI~=xx?)k@`#H=1dR5HF977-R~+NV*{$vtL%?Ib5+
z3cp2V2}1~54H!yJsUN9yg?lsGD<@x@N-Fr|Zml5BFd=v{g$>UV;#rHD=!<|Hg-rPf
zS_Pd#e`I5%fN^doA(ow35Mmi&_tVl^6#%L-muY+vMv*4)&&Ms5H~Qn?Z&>~OxVik6
zt+S>l9Ykt&J*aEbI{xJ8t$->`4NC$Xira0B7M@&4ADboZb8mBKp5%-zzlhFI02{6b
zVSt--iKY^_DTcC9K%RNkpr}q<J5421n^lc&vm}xQzE1_mJ`zMroEv2C$$RLvilw0Z
zd^hG|!T%`lcXEW;anSn`(sw;3iiy1P1i}&vp!jeMlh!>$kl!Nv!te&^%qJ_yD8}E(
zN6OX((V8@iMwJilmEVeK!L0{MWM}Vib@l%Ib?>{K0pFKNF!%pleHUcib=H8*!qCfO
zlwRS&1~w5NY{PL&V{Um}i&cC6qHB|-8dJaVJ5$l@tJrl{FLQAhL`C#_uREIwKdRL(
z_K-X|5a6czn;*SQ!C3sa=%GX$t$V!Dk2ktS%Ft+cg&_$!pxINbsur758QCb<`Dt+*
z0l-#=w3L>|9(|Hu5mFi(*zu;D3k<!xAOkJ0+MZ!9wT|1rZ6wWPpVX#U1H3!XR9xZ=
zbhx^L6&fd-^Ax)I+ommRiL=kNw_AlIV%c6m2ifX&+EL0?=MrzKR`!lG>(JI;>|e3C
zJk#F60Q$TmuT=4%VY88)XwI!ea$L4kc1FdG`lgsp(|~W<4ypHW?I->3n%y;ipaQyE
z8`teteXx}L)cYR~68Su8Stt{x#fIyMZ+8Y8PdfC-jw+N!>9!tkz@H}GUZHsrIodKi
zA&*dx0Dv_>qYZAD>Z}i3wP6m_W6|e^)D;+e(B^2nUH*Igt|PWYdf*$F<zj9CupZ(y
z6N+;}{5SNvq85jXvc$wUZJ!O4q(VA`yRH9KxLx~aZOq?)b^(V#b$skiNJ_NcyIai)
zet`GU1|V#FG^%TA;^q@QZwP?MENP=iS;qy-`_Lvd7=F&9Xl*DeOwt4z&rdbeX>>iu
zM|B^9gBK|`j4`oCO7Yg>mX-3X6HnF4&UuB!;;Q{ue+5Mf2UmbP(kB;xyn4zDDW2C$
z0{c%H5grDB9Ez8X&)!B4Eg30h^<T!wME@Zx%d-s9Ae#aOg}wi@u7vkKj9wh43E|AU
zlPL)S=e5{NK<oLrnwk>hST=d(gsX#UsPWzSID<GEOX<OTE|Dco0%YzQ_xOLQ$e7yx
zk~kKdRO8hnh(hAi-j@$JzX%qwp=A$^(uR*%XR875w)PV?#&sOQQZJnvF_7twWgML1
zXVG?YaLb*&1Q9QdZEW6C>i|fU#BMntQ@1Y&+glZh)~IhNX>f}9+PzK$T|7T%sbxhP
zMuC#>S*Q(#9N7?~)U}FIGI3A>>z0g^=lmB5u!$~?=w{D1Blqbop^i-XxW-OUXS}<M
z{-p1KV^uC%8@TFiaexjOzvc&>(cds5odF(muktAVHCCB0Ksa<ohtZmNuGOf=1;gui
zp(q^Y=!%O!8j-MGUO&%xjhL%H?|InN*XM5`T)3iZ0`K+4R`8*?Hbq$RmW>@#577!>
zs!Mr6!AM0%#Y#K=OkXAq{H%;a$&Od?@rTwFJ4u$qql(fk4PCw*9)Da6Fu*7uwtwYS
zytqB?5kgyavPX#R{{;NmZgrGs{2^iXYeqWzaG_24M~bfCd=QSS=S<Fm^5I@2wXOaV
zOZcu-BVJX*tmsci8Dn#))W-SDa56X)Wq?iaECy<H?YRp0$%~VLLR$7*QFYQD<qV4+
zf2WAqSt|D-JRrs5J}`UQ01xls^N;nes)Z8fJi?)VMuea<rrST+y4;z$`%^l#Gb!5s
zKBdNTb8`t?<9%|>4Kx?8;Gr0y$>Dk>2-)|!GC7d+D1Js>FZz7Je$Z4POG=GTxavbW
z@mh2y{qIfag1x(;W(QV<(9_uElgPcxj<0;$gv8xdbnyhuXX>Oul)xl`x?Y7`lL?};
zD;pz;!QqaPw=F)JvA%wNTF7!HAW~*oza(%Z2YUXU4CM?-Xj2qTL>q9J`uql8bB(b^
z#LcF$Y4e){TwO&=?^Ua~ZQ5x%o1Iq6k&#$o*YPo8aqTWi6_;X2%nyaBPr46b^bqff
z96dH9i}^j1z90JDYW?iYo>`CA?zY^uVcDN`6%ey*=U`i07Y`uRt{~U;rsOTS@*F$9
z2Ud;y*!C<DOcxa|R@e<-b^=DdSl7)XXYZb7mik@pb!oTUTjwe=U9!QHn#{a;<I{y;
z&?ujt+ESpnPRge<UZk2Qi$m4ReP&*P;}O!=;=W=f`>Eby2@?r}HkUAN3c<8Ufy4^U
z%tB%+?e(aMAX;`~h%SoKud3FAW5_@s#R?j#cxlO=@S{$B*77-5)?)=C)-wkAuAAFn
zu05Xx7&+ZD^BK=V)p|Ax+@ZMMxrHCj`^z=<#M4tlX)?~>h&s#h+|;l^Ign3IqxCDl
zy8GZr3gIqfW3?4HSg6^voBHK*Be<R8GSalFA49j7mlY=qDW!U!wDs%bKxehw$C%D8
zk_u;pk22n-Ino#Th;NT#hV*u)9k%UA22NBfMw{8v26&uFr|s8Ld!yIo+Y%6Wj`Gk`
ze^5bb8FdTvAPtwwO>p0$B}O2-#7bVnUAw82BXVEkHUy9>v(G(Zx`LmfD*xGivIbaz
zJ@>d?hY~rqI?AK8w2Z<Xlmmv{3hjec8(}&~@eajXwLJ@rTM}vd#r4%hO8g-TJ`!XE
zkP&yvln_LS?WA<?GJ1)>Nn%?y@!^;@4c9@p%(xMt^U!M7YrAq*08E$w)FKbccy!<t
zp9s0&cxccl+`8NJMu+tz-uSnIyyAU(FFf&74oT}{9-vPc;-=~qAO|aSV0O(yoS0YN
zPu>t&l;3ALTKeaa7q5t+|HU1~`OK5>^iU3H<Y-riYw4r{u9G39ss@nhY9Jx*NIuul
zz*7P6$Q4%r5~YE>)cogZ8h9!m`T)#_ZV^WK_q|C-1O_SoH*yH1AEnhb07&f({O9R6
z5Z`#fNb8>=HgLq1KYjXBBA6afy5k~)5~cM=Bs*Yu&`Klei?K7tO5BivmP|<W=st%-
z8EE=T6sDd=Dz7VL8_3W>f2W%_U-U&CNQhw-E~(!T5MO_;r-6WsS<s8rwG|K;38()1
z{uLoCzBM_sySYCP_BpvIq)-ieNycKY)!D0cZ!@@Lb9i!r^W%^r&uO1%n6tCT%XO}T
z*!LE-q<ARsgWILWky2g<<%+IH<E4eleQgdxM+D}*pO-nMp|t1WW2oW+>)b_&{zAoM
z!%~UuWNF<^?bQ6mWY6Jmbh&Rd9D%Ax@1>Cm;z-NI7z2ywU5y<Lj{-I6S<9bA=v2Pa
z;N|<oApJ(^;{H>*5SRHTv%qqyaA?+7dnM6BG0s!rQ=cqPv#iTh5w2q;g|!52?0aSC
zc47>qG9+C7l>RKtF5ouIi49w=qXV1h3+EW|S%VAm{uQFCG5lu(Xr@#kMCeSujj~|D
zaSG&|y=&}An{pywu2Fg|72~QSnElfUW*^AQlJA04K9tw2Vb+%|nS%vJ7RCH*Xu(Bh
zpL={G<}|MN))GxHQ`E{4>cf9x%(yr=Ss!uY%ICnsy1RV|MlO*Z&$Bffpt2-?<aWjo
z`mr*ZX{I5t2*J;UT5fye4dnQt-`v%C%PT-PT#{bTfzrygj9#+TqO#;IHi}wTi>haY
zKup5@5@%wkIE>B1x1jxX{i(M2eGXL8M%q&2dlKdhmUfbrg{XTH>$xh@vDM+3$y0RI
z<Tojh$jFEY$!#?7Ymqg~<w8d=az!qz!v~|iKqR3hKHmHcIXZZFx?*-k;IdESJnKQC
zw7dH(-tF6E$kf&u20?2dU97Y;7ESf|*bPmx)Fa+R0X>C;{R`sz<dHu0T&+Jt^}}NP
zE`&RF3e0BXEz{@7i-e0h+s-2GUk%_>GFO}yp<Bg+qDx_k&{mz}=(nRL0X%5~`>r}V
zKwN)0Nr+HxKiu3JmE~1)?)znX#vE{u<DRjkwda-@tuENMAhQ-k3sGihJczeMig4BG
z^VOYmrgaL{0Nw+ZtmhUHz9x1g`elTN|BUzIMPP?egm*_oFUM9qyMC!p&qhKZdige5
zOTQrXK)XMEnb%u^wB&hIM|DVRcgzX)&?Jr>z5DRB{iv{PO+&E4p2WcCZ|-fYu4&73
z)psOlfeoT@_#)?{pERG3+Mi#z7R)-DUP*CLNDz;H<(nJoeUq5y8O`~g+ehK_pp`!5
zk-i;AYyI2vhw0Oz^?pf4Up20Wc9mf%uQXmgCG~H>H2)bFF0g|)27^2HUL-9W?K4m1
zzS!H?G)wdrA0L-BVdcvolB`JrTCeO1(p|QUk5Pt3aePxTrauIwocw4JN-K2QXR$kY
zr8wHAbb3pZEcSuTVv#&I9o#`Qmf6a4d%2du&>&)owo|R@{+hG59k2SssRvvCp+%;z
z14H{BrWSOvMcR!y%W(*?b!v=IXVjqZ+81vi#fG@<BeQ;5uVT*4^dlw3HJN^ahLP=E
zLG*>Kbfalvsk;nkYOd`~#*dprks`Dtm|tppGX0HHh<;+qQYm2;B)xvMO-W%FSbQH?
zEyj^_;)dz^nTn|Aq*|txr=s2IbJ~vA>RI>s%5^9*yNS}urVE4?SYSl^_|GC`EsK2}
zXhL%vDI2Y`PQSJ`${0#HB{AtK6ZBLB6&%YkhcXkpn1WMM1}k2H`$$sLl~4<7BM%u1
z8Z*Cqb}2LnkF~6+<<Nup`9Znq@*OFrxd&h$%&)2rc3M(0ln3nd@Y;ob7!>B8{~hTJ
zGzZ-VZd?h*dNA(^TK?srI%yvyR`s~GVD`kJi2aJG#IYh*z}0L%FXv8VZol{&`y&Qz
zmR9kf9;I3zQ-1SEg8*BBm3p&j)HijKY@$<t7x2qg<o#da6x1=(OryM+7H84|iIt7J
zC-0=7F<#*q3R3D<4R+yvL?GA?_KnW6<t$>$2;Zt6pO^iu^zT;+U7j)1F*4G$d^nBc
z`OkRwQDcM?d)hpv@4E<}jeBs8(U#X<HCC`Ne|3r-5yB8CIcQ@)9Mc`2{k9yx^gP1S
zm)-oM@9dra4f50o^W6~UzM*5=_}JCcs>LV3q^G9G#X>^P4S<Mq;}V3S1R{)zt5B1o
zVZ23@2(b7+zNn5}&-Y;TQ=)M2^!n8b9CBcU@i<5SiV)R>s{#$YM81k+FsbPItT`>F
zX5WtJZ(&@22#8)?YI>%-FKNuj&6OHX;`7mx(Z&O+eKl#p{czeIY9NMj0qCH5%Ua&D
zW6b#J4Gy`2{S>=1hM7w#EMJh3k|cJGl6IrB3qlG{!CQMMf&E57plV_~B8GES!v<*r
zTy>NYM%Q5;*4Ihk{L1;)C}@m98;s$TK8utb;wmJ<UzuSi9d@xVsEaVmSYRwmSM@W2
zy#Wx;su+pN<K6)WOj6Xq{_ld3u|D+XiJ<LvEADjQRSSyT;j>=l@2*gOTy-?MV#N78
zlXIzwUmdoQG}J}7W(*r02BU(8E!Fn(uh7vLgK<7y3XZ6Ek6kHQG#~<jD4X!PaOeQx
z-!w7A13|dDd&g7cGTD!pAEU6g*zf@diLiUqUgq!>MJ$~(o<GE?sP*{vwM+)5^5rct
zBI5r|=>K1@%l|(wjv<Tac#>b0JEy4ko^;E%$I0;6UD@S#bf#c!^L97jC*52WdVL^X
zjwzO3%(mmI$$ofUG!%xsWbS`2RMODE%EPr5XizwDoV{<V;q}tOCWpF+omm1$bYZhp
zibA^EcGP=0W=CbGw75By$Yq?}RkFumYJEBePGapDvczX&bg;Wine}d4H?!2U+~P;S
zgXY5;z*b`M$WH?Ljl?aZM2YjF=GAy5g~Eu)^ho=a<Ta~%!<ZjZeUZpCr^^QicmtIb
z;H@-vAhDu7f&Ofl`R7e4vx=Q~6Hyt)=jt?7--%zrE^r>|4+$kzY=Pr&v*4xty|Pij
zL`PX{_TVRZm8?0l#dl~WiemT^clDson$XVhOlIyyyqklE<tx7wZworJ=E7T4-8EDB
zeeO#=r(hk5XYM~?;C15VnJsIv4}WZt{BkDd4bE>%5Zg!n4BU4*al{0-axN@5MWt#_
z3Tit#MqU;UUO}JLE7|j=<M=2S-z$;U#^^>IyqR&M$)Ir+^<Cn)rMSo&XiQxsW}J&6
z!$a;gEUt8~+3!cWNr&Jt@Z=7>KiU}WUN|6gN=;>iHM}l+TLYOCb2W+a5YTzU2HQt%
zGN!Y07;^-iA7(N+U98Nx`-slu`JADq#m~Ng0&QjWpe%H)2`GjH90C$zHK6&nIltlw
z;2DoK0S~k=%~C{B_O*^S{t3|Oym>bNLVssB_j1c^Y;u?TvQH&&er{{?oPBkkqQkgZ
zbT0d~_RxLs>)yb*<P&oHu<Dj<sE%LfNmeSm)G4o5f9E~)T0bfVMb}C{f1*RI?GN*<
z&Wz7u%FBAB`IxT|zsH&+&+psIq&8T-V{vUNc3jLs_HNddlm!z+9s0B4e)x6Pe)*bd
zj)ol->ce7X#yy_NFU!qVSJK|VlqW0Q1lJrTSSqd;xG(SQ;y`rs^B*;##AU^G^Q70c
z_2>_tl31Jb$w7VXmbSSKkAo#Ulyq|-Yv2)8Z29&N4InxfU8Dk}d2D59?^>eZ2?<lF
zZ(2T#&}>Yyd8kRQm}_Ath5p=7^q;_Qevub3>J{8E$;Xb)nrwXVny96b(D<Uus0zK!
z{lLV`McF9JFQKerW#b}teIY7u8#Qh8pjWJBoxwoSBTuop!1tAO*`nRl7MdPU9Kg`k
z(2eiX0Fsqn8$vCg<4uo(2$g03EUexkzNup(gSiLl>@1i8U#EP`@_Xz<?Ot9Fjl|{b
ziqd=BiAX-@^7}H8Pe?w}#VcM5WpN;dH6EK<7w~B2$!v#wd6jGd?cndZ9^Z@!qYB{X
zX+f2&3y<3PhUQ2O%sB(oUMIB-{ALd+@hS11OTs!6l2BH+3Lx1n>eQuePvaJmgBp4^
zqX$-8oGex2UF^%tKuP42z#pv5hi<IfLh`3V7<ns?<&$Q+i_x70%UdCO=nWJVkf~P`
z&m$Wzmd(>MAdM}ul$#>*8`t%Je+gw$U1Y3wpS6^4t?){NQaq&EP9C?8kw{RIL03cu
znoVC{q${;RO3T2*kZitvD&@sLDWYCSGOMrh&+LVg7^4)yYQ<9Tm7t436YS=(cA3`Y
zC)QfL&}F23LF&1e)y~OM8P-C5hF4<eqy97!?OcmU60$#cZvsHEsb!F5)g87)($y&y
z$V8j=(ldDqYaEt(yUP<0(uRE1z60_(E2AeSuYkl*Z6f#=sXC@=Xx;78)a?W9YFh?7
za4B|+=86qt+}4lcNA_LfA(x&#tAW0B1}R=HztL_q@_AjY!*qiP?;My9?_n21noLu_
z2qc>EY=rbA23pWd=Cmaqi`4C09GaK)Y65{3W+G3K;gYk&JwKKB*yyZrKhQuT7uZW_
z82JpW(#E>$a4g650lK0p(OEpf^fd3d?5Yn9Bl$56p#gzTz&6noXiz>;d8qU)lW=0r
z<zV_#nwaJeMQ8IT<a<RZdFkz*aUg_u->>zDpzD43Y%K)6&({#NgT*BW9nnySj_0Aa
zgg5cLSJ@fUUh<L8{D^~y{7)P*ODBR4tOK4;+!dFf7Y0qrbvXaVxJiW^;;30Qi>$h-
zZ)Axqbt_sYl9G<%a=Tr$Nf-O<DcmMQmaxw1D<JquEo0Aa2UbF3W2dX$6_1;gH>)#v
zfvWz!L)hLbFXVB1-&Okrj-a<_7k_jP05x6>zhQzI1~z=~s;d;3hAhUs%71_<E+`lK
zS)hRnQc?~LimQ=)`iIhB7%dBe;f|F_>BLN~)@>-atW92r3uQB6U^jebdaXd)P$rAK
zdh%-Xc(yQBDmsfvc2|{o5-S+x@>5vkp&@!3BDPy@6RxB8Eko&3o*<WJU~TShSrz-O
z`*jE`CtHBGp?#M9Ct=<GZnem!*)G)B2l$qn;qB39D4EG3p>ydSMt?7An~>aE(hb|C
z*+kk)t0o_MT*IPZsdPhz`cH9!(|)WUB*gGswgT;c+B39JnC?$&bYWY6I|ZyETZIeR
zLL2#C`UYY%RyM?7Rn)g#RXgL{VKZ4*i`0w>zHqU3U9tygW<t{`F{(Q$=saKTwehGP
z-L2QD#FX2Jsi^7IfS*YR6T=4WcwL;nwvEJb@S{8aERTBzzx#eb<j=7suq1Gh?)yi1
zLl1ZLoPWF8b9zy@epwusz4Q3yBXHJQ?-ny@`I=7O;s*VOXRBtrajyk?dF6d+Rc#EL
z=m^LW_GM}OJ@Fj8Y_`w0^Gm2mI2ZTUtXhZ5)XD{fpOl>WX$G=;mIZ}nh_<djm1X8S
zQ?jK~_!?Bu1E*${9ej!>{{{e;JK%Yv=cC)6tGe0)wf*Wy9u8}!eQfJU5NDv0-;K51
zo#p&uc%^qw>1st4oy9ay_d**NFYOSNJ0;4Y5Hp?9kCOWxbOKmGt0O@V=ML^;anBYX
zd1F}AS0^K@lK41`fG|HkKl3;7tEcx>;962WOwMwhILaIDm=VIS>|BKzqt2Pvi3X-=
z)cSeAL}i01qc1=1mvQa)(c|bFhpZ`2__>^4c6k5!EgkFPnw?`gD6m`F-Qjc+Rc>IR
z99WQOH2>x3t<R}`r?B@G-*CqnVrAfr<Qt5ZL95e8l)*PlajY<6nfn$jpSHzOaP2RD
z#J~vnKqqk(4n1rM66&HC!wT)47(mu2NcPxP3G&<BHT#7a*bkX$rk<N{{v!_j{hjCx
z)_BuGhybLX9v}OM{0?9OHh<v%PwDr6N1Au`Ov3t#xOrlVoocDUqTIOlHIL}|#Ia#m
zY30hUWvTJweQkR|NM&Kbqs>PXEJI~kMWK|%4{F`h;}hS__@<A}T_%r~wTg<1%};gj
zrW~;z<@$B9UG>xkHf66|$qNn9tm>*}lVk|fmk0e|^*!t!K_ypNl^jv5teulaJyu+p
z21z;{Tf3M+K5%u5Mm;<1z(2qpkz`mwJk{fWusEZvcsqZaQ;_*)uGYq2bt)N8JCdVe
zNKjD2I^mE$Lx36tR~;xq0wYiYB~^ej1dS1>LnNdTC0-T*l?W0c)_?m=p9K3>;o!&=
z<laiCKYd`oGB8GtmWTW)>*vL-BliCv`3m`OrQJOiJAbN>Oi#&}X7^~jEk4ny^Hj}F
zhKvqx@R8Z-xs@7E)SgdFRL|CjJYP9^?K>;OqN}XWlfG#5qFm26eRQx_H?Zb9T6A~j
zXgBej>w+S5;)pgc^S-$+j=5-(4&pbisKQHog_RSJyl>;oK|0T0#zLRItn%ATSL_C*
zBMZzsKNl#|bVr+)G|=D%cC``VsrPQ^njZgH=P)~kl2%#RkTz7&h;4BNPMh)Qwl}@`
z1cW!fE~XVdjdaQ4?ab@67xBo^0VXQsLH`S3{qKtXf8G?2gyn|)Z^RG(Ke(v>85oy4
zuTA^tgUct9<~=vk+NL_A8_q_pJN#ncUP)6eNMM55h6j?SJSW5-a<~}~>w=SbX@O#?
zBcBBs-_9dKZ6)NgY%%ub47cM!#K#0~G_$C!!VR_e&}ss$T^~}D%ZV%p*eL`F6}Znx
zOdM5JReuq3;|Yq18Z!*Jhj3YFs01PgN<yPR>liJO1wCBmW2ELy3rh*Nbo{{WIYgfY
zu^-(dG)2bWva7{6u}ny44k<u+g9o#4<k->WXN;ET?OdXpKPoLu9;#>v-aH|@vKc_j
z23yV83_Q}ZcK&9YKPoPa7;3n)Uq4BYM&Vsz_ui5FkotSrHCyMyAyqn_jk~-v4K&|?
z&?U*0JIxDT)gwQN9+IiFDVRo(_>TG`TmX$wiV`!`;32#)SOY4}y&3IR>xjN#!!yLL
z!Jn#I#7+MN4vA2X2*8xuS1`9mZN`Pg8Zz62fq%Ol%gxU8m}4rd#718fIC9#$;aS6O
zfM6)aJH#idwmj33Xx29qD0jtE9+Tvv?t%wmE8~c4t5Yv1c*(*&+cD?anm}Jz7=%au
z6j=f-XvIhHyd2v%Aw_5Wq?ytJJ9Pt-e*?I>ze<(2>R!$(OwL%RwW^*rZ`?2pK{hiA
zgnx?fD*Buwo2s#87)B=6c3}@UScg&{M3rF#EMD%spUUw|Ee%yh{UY_FZzm~CoE1`d
z&lBGv|J^d621MN7U!#@4|F?IF|NW-=U)^*5Cr3K?3=y>xhZ?~BUpBk9g7bP&y}Z~8
zD!iRJ@k0&6)?`=Uo!LFB?l6aIMH!})MH`ZzQ)+4h=49R?EHckh7Oh@*1Dy#QHy=K}
z)pPEAl-r+k+~3Mf9<|f^Q66U!jat=S?HzK2e3E#LGUL_K{H0pk<j~~W$QS3&%MZ{B
zY~sA)pzgV+jVj5;Ey0+(2ET-K#>V>D+&{ZqLw$Iu`EGHsNv(A`7IzBO@c^&YEEq6n
z+6%F>^SHK^aki>bE>kq=7rcnAFSa?0RYY$P%Fh%riU3A`!H?3z@!!3F|4+v9f8@AG
z3t+N{rNESU1MCj}PxGQ66`O5IILt5rP9iAf)%pW^p^x<J-sw;V*%XQ&IrAE@;gJ43
zXcklXO|E`N-s<Fq4xCi?o-BT0ANj4a9_jWBn6C!YA6Lf$w+95r<i8;R-$hlKjeqW#
zozSc2Iv+24(779eE+)d9=6ZB^&10W7{+n>3DXQqJg^Uj=ItnnFr`x6wV-~o(^vlmZ
zeViok-@{^i1{D_t`RnL{2?b=EbP2F2?*Fa0>yBzN+x8AclqMo&=r+3a5HJ)45u}AG
zAP6YE3m8C3V2ER<3=m)lA_yWydIHiBq9E1K5eNtfNJ69&N&@c->dc$>*1hk}U2naY
zKeF;AJLlWyoV|a$oZm_~X)w0B=zeWWq-mc?DVMrnhAn!el52D)(pEDR(P-kl*z@rO
zJG><06W~1q?u9#3NsZhQioU-}nY7zqx3uI6WVAO}%a~%d%{>r7&JQ#(Kuc<+f>^vP
zJ~Km-hE%b$n>}_)-yD90LC7%u=@dVz@IV@ra|gkLgSuD8iFt|xKnGLs5;)JAlh|FV
z98m!$6cpV=zdAQlO@V0EO~@B+hixr-WfF+7J9o0}n9v)R<%7dpT)y%I^SktJ7D*Lx
zchiyeap68qgVvOxwj66lHj8^o<J!5Sj~Gt3{*q_MM7HY1xS`4A5oP-uuyhCRb;`)g
zl|YgX9k2*5&v>-yl@>WP?j#p$T#%hU3P}U2VU=MG!Kb5ptax*tC+GC#yQ~q}z;(^?
zgzYc_VBjL-EyGwhm1}w&5gN)qRY4EOx`z9?Okl(tRnMfVb|9JAGUFYrzZ=_kK-m4v
zS0P`^sY4w;oe#5sZiwc1Lwn8Adw~$bQQ+t9GNzI+rfSX6NoRRw0ooJK;LHOt{6i7&
zbFf6cM6l3$u)7V^#QPvluzkkfvNiu271j&<<GGAL5Q4h=K0t)=Z8*X&n={;ZAhyGM
z0P0UOFdb|N60)!C?dy>9ba`jiT`H(P-U$3GYwibNdY=6m2s~^~CH$0XrD<j_1;>JP
zD$832D&}`xlg=DNLjM){150i9a)O<e%@1va3^rXfGDQ>O%4>2B)4v&}2gjKKrQQSv
ztxR(!Z)6t-u$YGxv^98z^f_R;kSjSZ_i4cUa7mhcDvm+RaIl?UREYrqSS^YcThvFs
zM=W7WCzbVnh>bcA;G{J{&WNhg_NxS&J>TR8IzTwu2Qt2=(lcH#FA>zKJkz88#z3g%
zTHcL=v44Nu(yad)!o(|JJF7?=T4QCR>x72hgQp-OeYZIu9RQJRwqW%-O7rXoKNp^|
z)W2pGE~!}>>wFBxr)_Yg=6l`HueoQ^n1Js*gUjxoyWusQ*=Q$xJ)CwczytUrS9mf#
zyUBZbP7d&X0bqSPZlmD%wU~DW*0XAkX|OV7&nE8XO7|l)C#+409IOFFwJLx38s`db
zf|GBt2grV$CofNCq|wy6fSMPYZrI%YHFGvjGAlvvP6#rR8hp%E=`b*jJ<Pd1;kCod
zQ36XLKBFwOt$5$cHz|Phm{0u>9<xMn0EWOl%eR4JNA@*mFPQA~=1Zs4?_Q3QKKkFH
z0xbn5rO8HaL>&hP8l`9JhiY6D+Iu3to~=a7jGHM-317K!hLtS`m(e<nuKQyvWOC)6
zX32?6Pnw~8cFToQ->kysPHt2_ghsOhnYG8sCM%{$LJXHwrGDPU`@p(_e!Eo=(3_xw
zw0&uD5__Pt>5}nHWdnTi_iI6}eojEWBPiO&HfwQ!egamij+PMkb1)bmLNIe-?&u@8
zMiNCKqDu4!GUxtyMU$3-;#_~c!@2N})1yUQERYe2ozEC|mz*-zO>*?{D!XjVtu_pJ
zX1DGoFaI|_1am7l|0F3Q)j4f*SfIN(UPPY<<m4jS4o`vl9BU4D1QmaDsijhc-(0H5
zQYd>uW&xK!N&U0jm}X>KWCGMrip~;aplbisu7Jt^{SxH=AcWD%qVNZ<pDV^ZyS{~(
z@sw>bb3l1aNVSZa>5O8wQ#KzoYFqs09UarjAQmmO9&TaXY(1wcSlX=*hNZZ0;=K-w
z(a?h=)M|dcnCuyYw+L@}{^48tw#l4;+g^EVj#^@M#}$6dh#OjDQ9LGHFO+-yYZ}DK
zCNzi|m9D>xwnlOrJ+148#^f49=lD_lNLr}DSlhT}G|f9)<Z|#7W%kHRs)-Y06y@|N
z4N>qSd<zjP?cfL?M9c-&6_0KGYMTR98+us?kx+GQ40<%kWF?qe``A9;D>FMt*O$Mu
zqqysn*wvw}HbqjY0wJAFh;a4K(^m8#2x~dz;`h8SONZ`wj@&|(pAwJvA{;OPDu%ZA
zKrF#GPWazn^QiKCCNc(f5Jp}B@-Cd*oYlN;c)uzTDjq4r;HIaeOM-!9C<6;Q{GUs{
z!P_<XAzvqr<0PAORBhHL<aQN*F^XHS;FTM1;qJs>XS~0((hV@1BvG|mKRI@<+-NsF
z20SV7klpcN7B_OtHKpHHeC4bRh@MkWRezNQ9FVv!hr@UU!M@sgRKP3Og2m0CCQPEV
zGJ?cb`*^|E-Q^VC6P2F-)<HQaXO;iQuF#GntptSGxkse~q`FD+cJ4SjCHaForxf>(
za%N2G9pY#ULXte7d*W(HOt*~v=zzs(Zor2AR%!ij*|pmeBdH2;FfHKeU#Z~#9U+4k
zIlb@ly+6yx03{aq8Qzz1uwo@y`MTfh?3JTIx_5(%jIRcn`gcSSW5c>1=U7^ovs|6A
zz(m`n{e&(szSh=kTv)Ub?>g$}4O+GnIo6KQm9md9;K44Za|jTt5EQ~xApD8qR5Mxu
zfw*g1IvClM{K*2zb_``~5aFK`^NURhRoB%wi-A^Y+t)jmJ+iN7_&293b`2-Y%a1^D
zth0P8Y5xBp$YZjvf>vTzus5{>Il-o6tE<jC9VG(w<t4!LQ@dU)+~00LTRZ3u0U#dU
zz6)yoiq)9H%}3DC72f_0Jft*fe%9zBRvd7@le)P=8%PwWjg_unve-TZ*x+s=-RJ`A
zaPNy3lq!DSoI<f^KNkBW?c^z5yJY^S=J0rQ0eIcnS4#RW!h%o;E>xDDdlgc!Dvh#f
zX!R*HPXSl!FOaWZ$`a&c_?|B6Zy&##JEpy+6@Gm8a<|)(Ga7w>|A?&lwyLaD@!8!Y
z%oQx#2<Xf19)C9bO!eoT?>rW!^Ff6plsN;<F?rO;(W|jO!Lc`Cf!i{=OsT%qr~D&f
zUc`@!5Uj`i>R1d|whk{(7zrbIj{qSGD9VI1rTgUzKT^aZg;e--mDHXmekf7gf_jz@
z1<(|e$*b&H*0+7nv`BqW&k4LMsTO@fUr|QuV^G`EvmY_n5EUD+3;KcO46aKgv5d3t
z7J!x$aYapE5x7;RhJFRD17%^7C|KlJXOg#7<cuw~*lE`9Gq1(-rS$iIUvM$0a%)?B
zzWM~$5if0=rt1%Y$Xqzsy-c3|FP@>WM?ha!;bcA=1C=hn*4f8zp-=2!d;ey42xYaG
ztC}!$Bz>|uL`(0LqejtjH#=!Z6L;x4sG6M}oUl^-tT?Yzt}dCo$A{eR=j$4!XCG1Z
zk?>lPI#fQ{(RIb~-2yKH5zX>1kq@{YzI0^SacXwoX;$-V@E>@mQlBT#xL5mxIkN7z
zD2BEWo*Xw%UD-3K_}ZW**v6N{Rq}?@6Ny%<2F|*9lMcB$EY7#9XFF-Nb&|N2o&EJ~
zc)^1YH_O+(wepf0{~&}pY{7MEcppeQS{!|Fg>Qj5%npSJrbOqL{h}@BaasIG?WI?5
zh&MZg++|2^4MXZlhM~Dbgn`}d*06#++Gx&N=_!egS(Txn*By&n77`dOaSGOS>mC@r
zBF3NDx&%M&$#TF3`!eNuCH_GH%-eccs$+QcePwk~-lX@Sg9b|2SIKVg%1l_;R`*Rk
z)%x20BB?w95aPu5JJ@MuW~P{U9;tm=5@=~>fDUTm?NbcLDUQOo(d@)a??w@EqI&_3
zBDQV6W{757lJ1un1P8G0V*)~kaG-G9H3Qpp9C6{NaE=-Y%U%l)C%KKS^XF0`rN2yP
zXJ`M`n5ELE)Uyod&pQ#xeVI(M_@azIKah-6&yn^x3v2YNWdBfe;Z-3S6MxM(9KD<9
zmqkVl&BI!r;OUdvva=5-j2LP}vOme4WcIRnA@_xC1`NT*7wCvn05({9S1t2Kl_VJL
zi6_<~zCQM=z|=f*Fp{3Q>0W(ZX*E;|eN6eZ=z57R?2Os^l^d}W(_sw*GxKFOhaYDN
z$HStIMRE+8T}zwO%C1bP>mT6xlMRMSb~F+QQCEFMiN^4RxOvpMxe#4JQ6Vnj=ezrC
zM|&}ye6l%Df?66J2rIX3&e0p;LiBod@gwI`e_gP}OUq?)8F>uy0xh*0DL_ljX6?5?
zkj(4&+5C&QGkw<pE?AteL<kh%f<g2u!1um!!Gf};Hxei3%Mg5~-?(7gy-qYPSkPuD
zkC65K3!n`~dmZNq4;6t_)3}MC4*A}7neMu!?C~f<_nYa*g4B_wXL_R6JmxFuw!KUD
z0)<EWF;jurr!)LZQz@DcxVtWLu2%wVv>={Oy%ZgZq7c{pQ(*%}mvAIvj@?{n*f3Yu
zb*bAWTH{743*B!(lUe6$pU-t#R`>ANBYB6$8q|}$Z{51hyO1{bT}@TQQ1_|Yf;Qr5
ze6?L^`h}h}Czcm7`E9!0vF$&T`QN%V<Y=Z(_CiQ*$iB`Ve@1(NXxKK*H>_UfEo1-u
z#NtNc(tH^mpFF}UCfBg?7RzAM9{J1r$uEi-Ta_zzDzz%SIkDN-i~AGXxqLs&a}B9`
zzB-b;yuaULYRCoO-I49+*AT~qBK(Gqu7wq}0lvb@rIjv*quXO+66ChzAW3#hOA=ev
zkfsw`EBw}*=~l&YR!#TByesF~Ls)7r?^34CYzz53Q3-;NC*b2HAw%eJ?{#h|>zdY$
zi)M=uk7JowCq4;8iII0x-&LI@)ibyIzqnKFBdiq~NHOPLs8upY7$s$prszn950lpA
zKB3(BG>EJsLw5S)zSH4lu?Cp0^x=QlTWa5ZC~w9xQ^5e?)B7{Xo;CBGzfi#p3l;vP
zf`zRlbImy_V^^iP>qk7RN?GT9dXG+~oN(|ChMQ_jW4z)w%uQ(OokRqGej9hZjrjJp
zu{b~6jD6Od^j00iNUBcFF9Ye^yX8GZ`Jl~g|MN>qs`8BE3rSQ0JNcf(QnviP%sv)J
zsHNo<7<%F-@1oSO9YLPXL&w*Z3`|VX&Vxt=M(c9}k=lKyVt+8fOp1dCuZB4M9<oq8
zn!aY&2LUN`ADc7kl1pzgREeN=wM=&5sHE0EP*7|0ztX+_>3P$iT(Fs@t@EbQSxzIc
zn82-`2XH*6{zX$sOTp*&P2ihpNAC`nqvAgvJ909gi+K>Iyty2ZyZ)>|_~(nJr|MN~
zYW7_nitaBoT6!J=xL@)XlH`#!<=piN6~&XkWNE9K3C+d@EtW!;I#b@$0cU1GWgQ!>
zCKlNKUED=5y_Rd4R>6p!(i7xPzTDVpW`O$To(!4iPfajr=o=;-r)NlR4orSa_|E?l
zHsXPWI=<Fp1vq0vJnKBSD7HRYKp&6KrY&pu&Wx+-;aFhE8yWA~G(Uk13Sbk6bUg4W
zhTY6PY)1SO#6u8ZdL<1+lX}Z{PI<XZR$b-<$YE$?pyQrrh8%w6wQm9g19uBkd3c6<
z;msY#vhFZo(vOpCKM40&dufZP(BzpjMG6ilm)Ks*9thZxbLfc#4hX?fN@2VUchn^l
z4>2TP!pBaEC`=<m7b0GW?*7lDynnh&{|g_Y_UT3I1maU={9FzJ5B<wfof7REi2nd2
C%77OD

diff --git a/16-bayes-mcmc_files/figure-html/unnamed-chunk-14-1.png b/16-bayes-mcmc_files/figure-html/unnamed-chunk-14-1.png
index 94371873aff12be41451783dbdf43b752dcb4082..4f13f44f82aa41519679aea472205150f846c646 100644
GIT binary patch
literal 57704
zcmd>mby$>N*X}Tal$4Z6i?o0s4FVD(jWm+dDJcyzN{N&RNJ~pfO2?oeT?z<DOLynM
z%y|aXUwrTTe&;&hiTAtCAB@86{p4P2-D|D;-XW^WvUoVxaX=stp4<azH4x}B@L!BH
ztV_T@#VsCeAQ1RaRZ&9*_!SKe4Gacfx^xL09UTJ$<MQRpn3$MYSXkKD*f=;ixVX4i
zu3W*x!^6kNCm<joBqSsvBD#9@DlsuJ2?+@)DJdBl896!mwQJWXC@8L9zfMU>Nkv6P
zO-)ThLqkhTOGigXPfvg2#tjArhMPBUGBPqUF)=YSGqbR;u(GnUv9YnUv){UPi-UuM
zlarH+i|h97+uYpTJUl#i?%d(!<>lk!<LBoW5D*X)6ciE?5*8K~5fKp;6}@}+u9%pZ
zxVX54gv7mj_ar4HrKF^!rKM$LWbWU;FDoni;K2iVd3gl|1w}<gB_$<gWn~o=6;)N$
zhYugBsi~=}t7~XzXliO|X=!O|YwPIf=<4b|dh|$7PfuT8-@w4Y(9qDx$jI2(`0?Y%
zCMG7Prlw|QX6EMR78VwkmX=mlR@T<mHa0f4wzhV5cJ}u6Po6w+aBy&RbaZlZa&~rh
zadB~Vb$$BushgXdySuxGhli)9r<a$Px3{;CkB_gfub-bE1OkCVq0gQ@d;a{nzrTM#
zKtNz%U{Fv{aB%R87cW9WLSDXn85$ZI78VvB9v%@95g8fz>eZ{LsHoSkUq?qr$Hc_M
z#>T#R^Cm7XE<Qd!At50#F)=AADLFYgB_-wU+qbEyscC6x>FMbi85x<GnORv`+1c6e
z-o4Ao$;r*l&CAQn&(AL?C@3r}eE<G^QBhHGadAmWNoi?mSy@?mdHIJAA1W#;Dl03i
zs;a81t7~g(>+0(2>+2gD8X6lLVK7)zQ`5(fA3uHi)ZE<M($ezz^XJyq*0#2`j*gDb
z&d#o`uI}#co}QlG-rl~xzW)CHFJHcV{rYuaU|?`?aA;`g+qZAS!^0yZBcr3E-@kt!
z8yg!RAD@_*n4FxPnwpxPo}QVR`SIh&?Ck8^+}!;9{KCS*;^N}c($ezs^2*A}>gwv+
z+S>a1`o_k_=H}+s*4Fm+_Rh}E?(Xj1-roNH{=vb);o;%Y(b4ho@yW>v91cG{Jv}=+
zLm&`HBvMTT#~xVv*p3f$T|gk}VAKy<<ovcd2t)^xlfI|nk-XO6U4K*4bJO5O)wRqH
z@4=hQ0(~K#k)F{<zyeN$&+ETVE<adHcpTi^eH(n~#ii%yx9_~5n<3<sC%k7wDl2X%
zarH$AaXr;%?iT_#XChvLAELKu`cC+c#!c2CJPK-U-&^kIuM}<677o>|9-f(ctykn1
zq#}mY_b2M=EFIbaOI4Mp8p8lCv*Z#w@X3e)27#cj!Dt|mgxoXW4{2Evz^AGj9q`G*
zN(g+$kYE9y@;LvOhmU>&S8uP&+Q)%F@h(N)UF($!EQG7z++oV)jU$JM-3lKCR_eSm
z2Z+ZQmniZx%zK5S;S`tcQl2E_bj|3@z0dQwQovUn=_P+}r66pvVc4m+y*!4Zxf0|R
zu}i$XaSWq}hGlgkc9CDOkLNLQbFDYzh)+)RQpqh44BCa{4+ra`5p1)*yTSC(8*;u@
zjR*!piThv8kbQjzTvA*o>qcJhAiJM1U*_8M0uy!pM*5Dj#XXj}5$j3k2jhb7BiAHA
zGOu^~_YfQA_~@^qNJhC}?XXj%ecx_v+}&X(#K0r^r#*bElCVXJ2$%2bi0M+~j9{Bz
z>5Ke@<UO!Q^d&SB`9!JR>CYmNGUO$0^j9Gyqp)Ttzqus;8RpYVTaDwkiQZ}fh_;=p
z9}u<?&CXob2ssWY7uBjLF_`5wZ}7t9Z8qNXX~>LOeWunrsznv&qFi)E+LMLdWYf1-
z+5|*;aa~Z;*rBZoDRK4qV3vr`PwdB$%1?_48hNL6u;_@b%N$O|AO=3$_u7v0z$=Ni
zUjq>Fv#0gfrZ%Ul^dt;ilxhnUn`qjEPRX#+1i#W=?}J;PPlNP3y{({ctnP9$R7DL-
z?2oM9%MLa137!=#hL$ZP4QeMu3in#mpsUgu`5ua#6|~n)ogqjL13_g`OmJB;a+O#k
zpB5)S4kR%679Q=iQMe!b@bgZI@BPSi*_2eHAka(3Gx#o{gcgsFu<*1B)#BW^4J%%P
zBy2Y@!bLCQb4;K{y2I%+m$f)itTY`ukhd8z6uG<2eG~~&kLKUY6hQP)`VBZGhwgIb
zUdh5a!)dE&_Xpuz2A9Y(q4{i$9--0USxdnRkUR$p8{Ktp)m|~}dB2^n=Q)CK5=x}E
zP$h0QD$O>R9tMEQUW<8ehO8romSDwXhCfcD`1hWHl9BLCrkBYAY5|jeHM>#-n3u59
zR88=E7OBIhUNkVHzlt3kt>W}5f_<p7Piw7VuwQBrM#jO;d?DM;#G1Vntdio*PDPsp
zbWq7>ptUG|coMP!wiS)E%4j)e@NLG_pBL;qI%(Iu{d`#p$=r7AkaR|buFB#;&_uIb
z2HAlfS$6GC)+a;mX48#{Y$g@ECGYOAzq_SKeX^jCOpA*=<O4<%LhpV(6JCNW3bw6z
z=R=)4@3r<#(zqsno_pVPTRnJk03znNupjwJ0yNL4bl@&J>|}V*C!?W$^&LS<!<u<&
zv{C(-lOOZc^G*=@?bj3}ATbQH7bK(N^AIGGnZuN{mwq7FFL_p<9wyl5R7sKR+#4^v
zH(9`lc)Gxd=rOq5#n*8K2jl>zydiTyF5J606A3bVe9}MlLl7wwCV(uT)i=rLKEBZ<
zvgtvPrg|#1JNumk-T1|Fxshk7%N4PJ*CD{ozmtL?XAB}-y3g*fB(r}tGN*$+-zMmB
z`FiH3hegLeU<oOhJ%!zkea@C-s>XYY-{v`q+&J>@6lxRh-H_}|bU2@q;h8>L?}cP#
z-M~UKvshW<ir=bZx}CU8*|<~gH`s;<o;psT+cB~w(;N|vK4Akav?t`A_UtB_&f}Fe
za(ZZYmemmg(y+8!5BrqrFDJE9a0OkJx5vfkq;L_&_L&+Op_p>!AO%X08`w%p724M}
z+$$u9t#HBg6qQ_5P{RRtYb^Hv&>43YVi}h+2U8}Q46fR@%U$31GG!JpWsZI5;4+?-
zbE{3|E2543x`>>2A~k>BJKm5dH*eMod1f?_geXrEl*isBy`vPv9gR)Bsx(<jiGBP|
zsA6!kih$H~c231^sJ@rAVwpIe6wS4UN=whA#Z&XjDjs*fqxO|JTN900ebHk#8AW~}
zqqsWqE{<O9=9%l{qi%wmLCrKqxUvPwq4NCmJHm@fx^jA)yS0U8x?%d9?AGn`Z+-Sg
z9~-pJ@hy9*W^j0tT-Og#ROEDFEW@iQlxBLPEl3_^zZL8Fp;bucTU1w!Ria<63+fo&
zPsWlca<l94@1m`YklI)knaSYwu(I1&DeFFza_M||JyBh2uz#?0J6r-!DO|$x340My
z3<KXK4fBHc7NW-nD{~<#2)3NU+~_>b;Ooznua9;NVy(_n-W&D3T7x|uk)0FFTr04l
zC+w)ewGb*u2fbp}O#^OkogIE#Hv6HAh9*(r3N^U>ettha6bqODqHM(4E~gl49{Ff7
z7pn7Xd<!`l#aLA|$9PsZV=z9N*$smwLWzK6^nS6H525{mYoNo%s#F8Y;~QjHbbJZr
z!JsnLsg9x|j|Pa2PIV+xv=s1{S3Is*bdsFlqi&2g?)4jcSR(Z8EAUWU@&3janBWI;
z_@h!F-Rjte($o%mx0>W6I$$_1D=_oM+zlwD`{rcBZ66}xmZIe;8nfUEt{=}pYOkdZ
z2qnmhj8oU9#6Bs%aUl)>u8rOtb#3F5mqzaT5Y@h-UY?q~ek#GgCnHKIahb=00L^Ts
zEP8Wu$}e<1_Rlji&>gROr$M{}Qi^@k`&Z=?_3_!Ca{s8m5j2!|t!A9GxmLf((GwY@
zf1p>z=srbsxLqtTQ}TK2hZrIXFWBCb7D@L)UbvGdNB+8o72l%<H7g~~s`*U#+&=a)
z@2G<sg?)m*TJBUK4-Z9oYVJr3zD^2z)YCa<pNt!25zZunGT&mloKQ<xet(dl6K&P4
z%~$ItsX(d1Y$UYvrH_wI-jnFJpm~Ld{6B=XI^0?0ZRD~|&0{$ETqR=$u}HIt=A8M8
zh^{+@r>!-P8SFY+Wb|w2fmX8Tx0xZfKcdAOHDzU3gG%2E4`O{`iM?+X>40v2II(fG
zDHwXS>8f`I#9LqtxN{l`!zbSwd>%ym_8ml5DGqnVdApaufopha3%4FKAMJ|kkSrea
z`|g(KIm;ps3SF~89?(~+z$HI^04ZZ}pO!GNJ^&G$8&!k$N-S@k`e2moe6(MCLnJHl
z`2)z})}9Z>ZR@3X7`IK+$B3Fj)Y=^S8&#K{-c;fYj;+1swDIzB1=D>_rElyzV)~OX
z`Z)hM=-kE_0`{XlAiB$0=!o?;Rpn8c{zAs<%Chdf*Oy5jg=-a)Jc{o6;GO#15S+*Y
zS7OU#bWz3GC$^V0DRO4yVYGNz$Ps5gGNVp|pL$i(<el_bX^69OY}jTno2JTH8dZrh
zGuQD!&8k`D-tt3Pi9niEuqs>ODz(T9wGR1ZoldrQrXv1uE9y-(DQ0PkDJ)mv{6MxL
zS(J_CM6W7H5C=C(VyJS>yY0y8Y2x~_$R`XPX=(Pd+62@-S{k{p>Nd4p=z)4xizS%w
zD%t1c%R!AV+S-$vQ1`TZ!nE~alzf+|<cf$7v4`O4u_y6}%DjAQk#K&mB4KTUz_W^o
z)Fgpt>OoHyl#Ds{oEx!yu<ObpaBesMgFn{vk0rUaZrPw5CM>|5rE5{<EQ38w=q-yp
zxO#xS9_sPSlqKO;+)@x<ThY@2b9+D`j82%&f+pH@aQQR8+nSxVQOhMXGdfe0U5U}`
zzHI(u7(E1S-0$ei^n#Du(>^_+_M%W9B3$%W&%#i4D;PadMfBq{?K8k-+{yx6%8OSp
zkXzZ<^Q4@El3A{i4#TmqKZa8**_%EeXgFdZL>=G*a3w@I#8B-TMAkYSKad;Gmk9Ic
zfhjx^KO2O;2D|{zD}I#!$V1#1smGyXAcSJjfhZ+F!h{kabV4v%JQ!SphMsl_4P636
zPY7lGH~mrzqhvB>u$rn?lluqYQsiYgU23d{!V?NR<is~?$9<cT+EfXSWnWUMZtW%}
zlN@eGUeQ#`l*`Xzii_<SS($&~wmF~>k$amz_ey8FZfsxXr}<l-3l+bjshbqBck!%k
zu@VGNsS<~`76y?g<?5w(#QUih<n(4>+VH!a!RLs}Z_%cLuSKZE7~}*k$ZPE{nW-!D
z-*{+4o$6tu<kc8%6JWL9`8KiViL)d||2r@X)azw5EIoPAxrRvSxT2rkNjuM)%iz1&
zr)-5$-*O>D9ZdmPr$8ZJAAjQIwd2LgWLHxrcIU6z9r?+7bsw1AX~SBgu$gV0N2kiJ
zC+uZr%ja`PKhfl5kDe_k?znrbLkfz$r{XF2<@<=7Fb*naYGs33P=@N9+xmGeSeD?r
zNWuug7s=H<iT&CX+Lr4(+p+g^TV-@LGs3Ku&C811ZYi|}6rra{>!weJr)qSDcDU#E
zTQRW6<{NI_WYz$2ut9r!ZCay^B5Z85511?ZC$_a&pYk*kwUS*Q!|F^>EsjLMt<XHM
zSwLK9fzyn1>c9^r@I#G72LfX~11X}Rjex-Y&p=>0C?Qr82_Y1Z4)pK-Wrr+ig7T1q
zx?1LJy|pf3?qOe@Ojs&a4JH;H<0ufGd0g7Q4oDd>qhh(m$$9L@f~NM2(6r+7Tg%sA
z-QDnX2nNk<xR6Pp!g7L8Q&6!?2VkcO31Syfy!sXO^yJ||xvh0Rr85DXol8(-ipr-(
z$cCIyLf*L@z!(qUgabwAlPKhf0XIWWfF}A$8O!+?@#7Q0UUtQB3L$Tg+^yxvkM}^!
z?F`;P#+F($E%?sp!rY@cQ757S3;vAw``M8teaFZm&gA$H1Xy&`o@PKK|Is>8=gxJ-
zDD+Qm&l77}(k%<(U2q~Soyttch5e+lne)MSz@tQnmI1e3?tS!#_a1-|EZz~I{8TVV
z*@R%vA+lt*A{LjN2KPy0Cns#ZFnieG%-+6cIF=UxiYow6l(|yC$L(Ch9`v+Ee`W`P
zSF=|2L}8<G7Rer=C^-7V-ia339@Jb}9|<9pW~4SxP-nzr`*8ieA?pXGxG^7Hu`e4z
zaS4Ne0s<)yy8SXt)P;?o7D)r3kZ0}omzgaQo?x=i<4<L}Q%0njH!-ulRxszGPwVb-
zseG{glH1=vdVDmfHo<fkturoTFDbe#<4DLnTUw4|)aaY=WOk79i`UoFGkBlt7q#+9
zG5}a{!oAMci{`$hQ8qX??oC_j>3aY$S{2x&6Z{h}uuJ$042ain6|b!z)*rWy91P#C
zJaSrZcPxLwV0!dewrs<k>FGg3Mrtz6L0Dkwk@L2O)o{kcbhtQaqIs*N(ZP<{*Z$R@
z)cLSL!`;@XU=Z2p7sR@h-PG2nnz6<)m@k*eOxLR6s;}Kxrqn7Sv|)`ND*B~JCQ~vP
zWCwt_FkrDHv%qkDoBL1WMq~XynjT&;gnO~q`B?cI%6^@5*2xPcS}eMzWot6`)vV#h
zX7l_RrL&L+_JfVa_45}}b1Tk272Nzdl&i~}lg*mFE1At&HGRlxUp}3$MzQX&V}>#9
z{AF;UP+6*#$W@nE_9#|vbu;^#Tg~{XJB~*&xmWKneveR!amhFH5!cAOcoo7z?b`S1
z$ii_ZGTU-MF*vJUZTJ})Wop82%3dyy^iGRCYO5RU$9~u<08WXpNtfUMa*hB<POkR1
z_iN_@fNWg(*9$1WK9%1oBonTxjY|zUqYaT1QU@wv{ZbtmBM-3ZT&)m2Ya26oo5L_A
z#w{jC;hlCG2e9S+Ix_6mfCzzSckcKuc@PT%Mp7LxF$z1oR(CH&thPiOPIkyLCbojK
zlK5#@cNmIfOo2zs155z0u@8c_N#vv|u0$nwK%iX8{~mm6-JfBHP{eXjj1KBL7j+l?
zRl<zzmz#eWJPp_bx{ilqyEZbamU9?ftn88GJu-aaPdQR|fBh2pfh-NmO;9wFDP0u?
zCNH1E_MJLtuch{3eXRdl?R4lE0l{Q|`)Ee)Vy0bRgI^x4?g-xwHBhKw6hfKueZ$bp
zvc}H=U(~Hgi*gs_XrdLt)R2$L2jBgKMMmw<Iz&c1O&SjUTHV(5jO!ZtoqxNUuzE7Q
zh$a+N`RSj;AOva7XL4hQ)YqkNa7scL41@5GRzBCg>Gf8*yHG?=DDkMw5_Qr#G9|*_
z7d*}N9K+5W%?uM<0zx;wge8Fq#)5(*=|JidAbu#wl@1ERqC*GY#`yRAvMnF<E$Q|3
zIH@YO`^j}CI(hBpoC5D_xuf}nmKP1;>Ku-cFByoH`A?$#%?+cYQF%+cybZ5h?BHA(
z4~1R>)t3!luMQ&8*2}T?gm?>fJT<D5v*TmsgpK99pPK6iQH#ybY*E@|e2E|0G9iA-
z&Py>sV6#7{yC>0O(FjTPQF}8#s}5>@__AzuuwTC;pC-kwv*Z3v`S-KLD(~g`svd?D
zzBe3IFCSG>&INtf0AZ-AIPD0;?CNt0++c>0c`<IV@vUboBq>(Aver59w?T$YH9@vd
z<zl<i^@tdd<8Pv$beqX2*YfJdM*GEmo7?b@66BdNQ{sf{X)#3mebA8Ov-tYH)7;$I
zXP>WznYI9&#<Rpq!2F0(L{9b@dsIkAJ_*#Lvw21xf2cbCY-{PBDKjyg{*hG*O|D_p
zuD_0ySmRf$p=y=_`55k=JjHKA5!oH_{js1J7WToq?Pz&V?qht?>-OmL4~AA#<^pii
zWH9<ib$dK^x*jU6cOtULYDExi*~+Z~pGG82Denmpuk3#taooD93Ni%ad{mHl*Q6l9
z5B&T$!>oTr*wK-oi-JN|zk@(yV4$dw$t8ONAn2bP)WUhi;VKHRU-%k*^rk)wI|5V8
zq|WiNp-|QF%gquX)((0g+~2_**{Wa>0>b>A0RV_dx66RY%Px(>C;i;s^SDEO%f~ET
zTK*W$ur8%oM|NI*xT%D4W1{F=s2ajNs)mr(5v}8XiW>Z9C}Hq7TtHa&4&wca`i)Bg
zE_1|`#BaY<?R2;?f!jQsfsJyKzmNw2ae&%E{^`~}wcE!dJl@}~apKp=i(3@^T{z%c
z`~y3zKcqz8t9{KB*DP-kCRqI9Jg;^CT*4X0exBH#eY5YFa6g572Jo?7J;%l@?Wp6(
zo5O8_=eb##w_NhG?geUP4zUM*n{!U$>SF#rvBeWDj7n}fP|0ml1#!rziiVv$W?^q^
zyRb80olL@1ezFVw8S!;RFKURBmgdgVG|to8C+_2xGVf&1W0qc2qzcE7{a4~NA69vo
z5G(u3RCO)+n8DX)=I>_>Y?cC8@`%{v8QSM<^ci@6TnAFzs=9^wj;!=zHMBv6u2~ht
zw-#o|2hEE9;0%)H5iM#UVTh4u_|PWsybS)t{nS0#6U#W{?iW}?(t4arFoo%uQS-=;
zHd(uVah~y8KiaSPjYJiCC&;Dv7B4a91m88Cac{T?{D%-W-mb8~#x>O0hVjI^C~zAV
z!WjtAZk1)8jRlX02^|``eTo}y_y!-Tl${j-JoB140oq)M8^?80uNiRsc0Vp@Q+t0X
zSyNWJp76ELxYN>klLGBEZ5tHz<NI1}o~M`58X~ef=1bJb8F~%43%|9rXLoqXOa^zs
z1?HT$uD%-g9Rt#@XO7OgoS5PmLCs`FZ@Tm~zG}S@()^kxS@2~gR?^lv!EH0TSwSMN
zd@76Q5f5gzs`wYTn+t5(H`GnuwSPyw2$Y*-eaL7b|A9J*Wl^K!$S+szqNL;M{Hpjl
z>UwKsBUR8?NWbkzP)p~TWxL)om)TZt|7S%8^(6%f1}UR-0Kqwn9FNH_`A^uUcd-l|
z2mHq}L%Y<^GQ&uAR0aGJb>tWwx&t_eA#!b-$<YG9km%_^vEg>0extxMW~_#IfoH69
z+YaD@U6*~PEylH8#CqEZUQ|V<9eiN~Px9B2r&#|BWbQ*=^0+80a1V+3z|(ALOcq<7
z0Lgb2a2_CluYD7L{r1}LxH?A<#T=PTD`|~L+0mF7Z*KfXJKi)&TK5cSN?l;`%%7Ip
zy@#p=I3F8GES48|Pj=pOGuZmP9!$HBJYyMiDK?srC373v!!!9+`?Hv<uyRB{2me2(
z5;duNk8G2#nH8_+lt)i2AjiDLP3sOgN(_ub?NH^0*How^fT38%pW*hOvVgv+OS$@x
zuLr)tm!kq?o8iB}LhZd<zrh0R!uwn#Oh=){XnMM!(PyArp#O#$VksMR|6Nw=eE<8G
z=RlgxIiX9b00^*3%35ayFD@z(&;HL;B6fjFM0D9dszk8t-Tt3|3kB%}ZpDF|Dt=G>
zzAMD>Cpa#vxlP$cAd654Xyr8<fPik?N7QgIPsDe;Om?*{V`Dz!eBUvF$>4<ksUuPU
zpsMgb)0tJ?Yn$#Yz1XhKxuX?ZXB9Cm3<>jobsFUl{BEx!z}h#6qcuGry~q)z5}Q{I
z4;>|lq~UerDW%N_{I>jT<*A7H3%<Eo_Hf{<vV3WZf#gmzgOF6o2wVefiI6hM21@a~
z81m>%)pP*=M`$!m$^oF~kp?IwWhjgpMwNGQ3zt+E3rgdPj_zO6v$A7gv6O5OJK4;{
zm}X1^)uWR<kAo^ci-X!h<DmaLu;<7fR$CD0f(!8f48bzfqPaSZSGi8VnVEWWdQ_3K
zCb(V}44>80X^&<&1fgZ!^ag5={3a0+&OnUkR|~uV2B&td2e;iy^16K=N0s`DjDAx3
zNdA`!#S0Ped41vm`9<!a-;VFN=?x#P;MM%26VZoU0l_V*`WA8xet{5T?u(>DF$?#i
z7(nxw8NJ2_;xvQlyrFf_>em!JGLL{Umu87y#{B8uZnaQ9Lm`EqEsZ~F6!YVxYtd(8
zvLhXSiMY&6zY`-};OhR=n*aa<y~tNsgXZzGwO_4>{rb%k+}8(a|MV0gcl%*?KlMfy
zPPB0D*B6&{If&RUKh=il>8(9w{8>FWBx<%%3eyLQ=mwnH)%6f5PWbz=*L9mH;7_Pr
z>7h5?CSEpE!m&9^i~dkfR(C-|&L8Lf{Q_;A_Z{>6MsHL;)DeYr=^rF%+}B%3BY8A`
z)*fVfKuntS6|Dg1Wku+}c^n*qA3PN$^nqoeh}~qPg-O^`oMlK|oCULRY+dF@=QdUt
z&cr_i86Ons3EoshxaXx4XJp;aWeKN?D{u<$e_?yxg-CoG9QBuW1uQc_Y@~s#HcTxg
z(tarY{)_v0yprBN3@mk<Zf%gZgkaE{5#qD4x=}yh<w@#p<dePIrgm`KYkH^EZf4cG
zEkw70PK568EKmGzq`};%pFEGcK4e~rj)mx+IT!OgoCd?K@<d)&1<cGpgN??Sz*vGO
zrd3B(o94_uo)fkWk@nuMR@C5<I3PDBc9J#LSR#am1Ia2k5Zs8fBc^WiLWVo+mIppg
zmwI0=CT^@imXc%)NBt}w0aYVZ6Ny*1xE+4pL&7%%zpitDpT8H*uKDSlCdZK<UCs%i
z^56g3K^oZbbq0!)Pf_ng0eB}X%-<tg^ylTx<hx09^*Fmqno$mlF~3DIq|%glKADYT
zia;~}?a4hM;}m859co|n`+*x@fV@mT@M(i7=0i4JR%!*^ho>Yv|Ey9mcW|{|BfVG^
zz<V_*r}m6{A2%xc&0|)4Bn#LUN%$lUI9vf&Rk!J?Rqq7cyvvCH*%-q|+uql$v0(;J
zU9Fno`wr^Z_1MWkXm3ku(tRNgzI0x$C<i)#_1IKEX-EIl#@du$*n~v1tcr=HScbhv
zj6Ll><Tt_&9AO~jVFD6^pZ&jGBxn+VlmWhyy%l}N!y!5S0P2{2A==VQQSrMLNWXxk
zh$=n=#-K374q?+JI|o%A(>7(#d@ovUKmM^_z=O|*mQVs!O98h_b`!<l{Np@-FIZqY
zd`j)u7{l7Gk1V@;c#`dB_ISA6J&PV1GxrKu2O;qn5Q6FJPcNmi{<zJkB)Xw8wO83K
zDRqv2JL!028KJ#tX*b-6QMqc%;0_r0&0qi{DN08+o!3_`99Aj}Q}v?#fp=q>HasW6
zi)g35*s$Jqx&Clab`uT!K;ZzGWhWg06+oeu3&+@GgpNDX32w(nH%7y1x=aYgM<)cK
z1xSG2K|yvPv?U1;8leOhx-rhb=a&(hXoUide4N2gG{Xde)JQF=OaIBEr>0D{&Sfh`
zGIJZ0K;I85AUYf|XPhuRmsK4JG0S~@T}h3RuclD3Rf|g&X!;pCO9z(TGgrm$$@ifl
z<+NnAFn^67f7Qi67jRZUYvD8~(TdlCuYN>&0!OKJ&;QZgr(}0k%M-??m{VVxA0Lr4
zImTqH3G?x4m*vYcCpnkhOAr2^=!fXy`IdNdxO(ms><$>#e&`5Lh%19zxSYD*1zJ2>
zaT+yLwWQoTefC1jt%_Q6SqjmbLd)*fN*iL=LI79%ZLSPv&d0MXs}QSU$~2L$uf^1+
zZ4~XjQ_2Qn^MAYJm{Cz0?MhO8H6gyRb3jrQl~DxS=y&^h?4^}Phnx{|d{8FPC>}e$
zS4dB5WF26mAEU|G8NZ;xv?_l};;g?U50Z}|U70wN%dU_23C0QMf9{;MvagKl>STbI
z)(Qdc!2E-YTcCgn8-+Kxe*XWXYN6n;rw`O1^rV0y#9P3I!&?{#f!08{#(CmF1QJgw
zy&Fbt7m*k`W_;~eZSux8RtxwS>p>qCy_*^~*DZ(t#i&0QCtjq6j{Pn+=spkyZ%h1a
zy#l_F%4B%YfuO8;+r^3Ucfl4nM3#=`xBH`q;*Lfk*57v$7)kxBQPCrp09U+rdQ?-5
z2lg_Mi|%@y&3QjryQSn2F+&_s5sd31<=mWl)#bNZIkzElVO$)OWI&hEPSCCkeuQ~X
z<Lkq}WV<yzar4CMPdtE({fsI0%Nrq<KrK%IL&Tp+e}3oMw4xtU$N==#A7?brxSQc~
zysTM$881@$cNW7GLt5Dl+SEt+V_D1+$7(TtE^1mN;yHy@36eCoY%dz?a^b2EuixX;
zp1qz<Gz`?j>n8Y6@!2;PC(ZB?4P>W^N$gnUkJR^HmmVU4C!PlR0|y+UYA1$YRFnz4
zTR&}j2}OCJYm7fn9SREu>J2R#>BG2wTG2sImW>C`W15WH2MC;Sj9i?&>K@b@t^2wV
z7Twc>KwRV24kb{CtVFgITDM>gwsAhH+7PB*4^oa<@s&CS7S?}SVE`bQL$hP`&+deR
z?mr|lP#J>d&!6{e;CY{fbhevyYZghrYf+4<nCBs1?pJ@(`cET%$`)2`|9QOMua3}6
zbjdYkEr^d;&!uM1bY(>Zu*_0EQ<kkz%{~E|q(BgN_Qtl#|J|23(xN8K?zZs(BK=^0
zb|@RgKj=#E`f)zFgva0j#yrA|n!Gc1;s9QZ^5u2_vqyb_>n2YMC&dXoBaW01{alaT
zeL-dF=}Ky1mEu}ZnhsI%Lq6gP`g5`2Uo8CJasS5oIe&Hi=mB#CSPB<JAHdK+W0WM{
zMoP;`f%o<7^3R$=*((opfFtyab_By+`o&woZLB`q%8aT>=lWeoV7eM90p$7rIFRd^
zJy}9gp?(H^z+nJogb>$rlyU40id|Lt-niLM_j=3fN)H|cMO}Ea26RJ40Yi6^0RABb
zGLpwy0B_;4ue65}g6ph;&(r*oQI1@Vb>y4a4nA|W4;w*c^-SOijW@H*c~>84F`43O
zGUZ*Zsb@Xp73o*k!bxF!m^hx*!5hLB?b@`vf@!XD2gE1IB_?mqlWuzFKE-e;tIB&+
z*W2cD+M*3Wbfd`T5vuI(seU7GqJ<*#HQoNNa;eQLF&u%W@%QfE)UmhqiyS$e4T#Ib
zp<720p+Sx3)nMMU8knfxCS&9krFDzfva!p|#iNjM-QnGIr3U7rQP}R@Av+}SdFRGt
zwZn1~NBN#eI}_998Z&{o`hlLvDYg_>+ALA+%k-}9D*<9-V8NyjQoDgNtq1F7xB_|6
zw8nOD>(I688HMV;6<xcd_WHXU?R$vCQUe9Mnevr4o#~#*7O?8}%G5RSul_6VQpbDU
zHg>x*dS99EUeMaOVd*{=9U}0K=DwJ-Z}^b1{?H|lfFk1+o;#P)`^rGygX7(-RzyaJ
zNQ(~%aB&a6%n6W@iPtSu6q)3#;`ne4M0BGa^gKybGT&jao61Z|3LbUJcdny9EP-Wo
ztZc6>@D;<A-F9%4#2TwT;6d6o0S06oa7j0)_taQlQObgKANsMSxaSc?+0^$Lm1|z~
z)RFzzrf~@rSiWu;TM4)krYGq&C>9)o-c3kpvgrx-liId&=Ac8zzE3pk?a3{*Z+knj
zL(ed4Ui*qZYWcpZoZGK0JA(1Cr&a7-`?1==$7zfG!N5$f(Z~{FFe?O=$uguaD5^;x
zs0WMtqMmVx6jvGWj4cs7rg!#6Nt?p!+&2yT@q2zHue`_i`H#}qMBRq_Tf-Z?#f5y#
zw8)ji#Cmg@Gxa@>I+lLOjw}W>dYzX5trN$I5ow2CKGLgoiT&)=XV$$n$7%OF*~;nj
zvkew^T9R=Sl|1Y=vakyhs{><=yr)j=Qf`Dn;2f24OfY9lWO2FuhKUqcqGw!r9Zb;2
ztnrTH>g~8Pk^5o@;?THUBfp>&(o@x{+PDe!#YbUyerh6;k&K;<=LP!OK))Mr=913_
zXL7nxXM4DW0i2ECwogLo^^<Q`q0>xmd16Dpi*y1VIiq!=>+L?V27Y6@3dMCNV=Hsk
zYlt>+*Ds8;VcuCl_g@{RFO+1kJGWvQR}vZ{EBw|s2jYz@qQnKic_0DwWJSaeth;%I
zEXAK5v6iD=Akp~?)VyL~Q5e+7h~c=2?}ps!jx9DYBPZ~Evf`~^S-2#-@H02nFbWcR
zJ8``}x%G`0xo0xOT$@1UmeE&88^R&R-O>ly$W^`k_LK5pmm!W3XMNIor1$l5rQT5$
zO26-GRc=Sb!G?-$46dY7aTImw4)@RseH_Q8&#ncAy4^WA@s16dG`i*2JP_IKJ&J8)
zc=%}3bSy??f|JK>I=wd<NjruxIE(6?c&BuhDsJhGG$9|`B{koe3ACq0_`w?mZddAY
zc}L+*mU_Ner`)}18IxR|%76M{)M0&#8NVSuLUt*jv@2|)zv?Y$LpXuMSGd-Cag&@Z
z7FoUh*sYvgh$nL5PRXI(ko#d(dHIRMF)Z95dwr*Ct#c-%uiU^LE|5A_lj~=>6*+dv
z$N|FjnX6OmVg50mQ{R)y4k0*evq}oe8ejI~0@kQ}t*%=K-?soe&56-en>u{rXhQz=
z)OR;S!YV%^oEjs|dF&ZI1OZ!d5uV4LI&TjXTDTALfd3$FWp7r1#%BN%=E5wZ%>_^1
zI?G_-E}rfc#V{9td<4|&fj36#wE>tM#$~QMZ=|M7z=B>^MbzP`9z9%$F5@jmx5{Kd
zco90$4=KYx{8TvrXJ0J&cm`=A69}$)G*ZXS`_Qrwc@MNHVz;>U4h_;b`7X})Cml~}
zuck*BIFatr^7)CbdOiMmqDrZ$M_{Yek41U|l|r3wE(_(QzN&@qiFuFq4yeQ-9Ws57
z4nDpiD0Nvv3<ulzJ%HfOTew+ym^`JVJhfbHi*YxpXQnmSm}=w}y&^vvJjd)M&b>m>
z%Xw@eZvK~(Nhumy*abIMpl@x*(>n9JmqXSGioRtILjK5pWy+^a7WYVfQV^`)&%R^&
z-QB!(kcU{89pstpo1`4~Nm`FZ%IL?O0YSAsdTOLp{ft^!FTkffmFINtd@*1*lmLr?
zo{8(u`E8vu|0JVnP2i_jv>i;o7_H<O1c-GWjIVl5MBZA<XDxRy)8?<QNFtAbBl>*}
zDy#F?H*oEiBk&x~W<$qlGCCX_^y&x3PEQQo^ggWZjt<>*n5V0I@_bR)7O-lK8O!#c
zR{c~|lKfhIw+?}?0(M%JQsWOd;hrLD<FzC8s?ynw!ObGtgh@EXScSv0^^-(l_=JJo
zOpJ`%_+s#;H>L<z8I0goNlFf9*=L9bvxB+aA!qC4eSnx)mioOp0%pwYy{>gK@!~nK
z>e)`OoZl!;L)y<9{U$Y~1_qdjZwVTj7gI`HH}Jl1_Tj&%zl$6Sb>~Pqe0#EGHgAx#
zCc67*ViiJs_ONfa(Qke=LS!{@^NPx(%S;jSsj59LTipdG!A91d7e4p)jO6FN=`MYC
z1w0nFXF@5>i5)Hwt|cBz0H6PL8I-&=hAQmGu;JDxBgksmy0AR<rz((SRa1Acp%$9x
z7Hx_c7cyvFULNb>&uhVkYG_nv6pKQOkd<gl(PmFw4o}-x>{wG^tO>9l-QH^)<G9^!
zhL7u0!oACKW6zWC9g#^ak&S@U<J3U};>|+tx4+Mdu5#B^#NY=wpand6_MXs72*WXb
zHh`;hVqP$A?0vQ?oewUZE3RE&e-|s%$jOhBYX9wGeWwlU{G}N^P`aKmdw`)lTA^t3
zC}%5dSv1G8{LR;ATO81SjuJ*mu58eTOY-h#kE&Up0Sk`opHmc2Q+zs|5JyV%fW!n+
zX`qrPadBx^)8o`YRl-u^PXO`EpN}KP>tpt$;%@^R6Os4BxZY~=_5Hf(8+&Rv)9Mks
zOx$9&T1Lr9+Juc5gtZuz(o5MSO}-H>W!E+xlV$C{sR{r2i_X&C&6nF;eJhx*i=-D%
zf`iD56=S;LlPBWDcu@44XT`)LdCZIFu7%Z*46bf>5);2Ru_CBF^0$Xn+=a|KtLrU)
zTcW41{PVVOP)~f4rtMh5(K~!L`ElLvhlC-?l?U?MD(~fn<3oIkXHO+3+<reO+yLMW
zdk`Q_r(ug{KkN`M-yL7Rc<KjITm_)|R#Mo>bMn&@J-y0R7RHPBeuvh}8YS5?$NYB{
z>pcUm?tf7^Scscu=%DBwS1gDF(UE@0NH23B_Wlq#oO+h7y>=qKF(2cfCk08(bn@m7
z?75RC+;BKKQS5{0E5}XkguNa8c%kPTm|V2Lq9gxmw&^MBC#>kVZG~X$5l68a%$PKD
zrnVIxPxir_-$N5E0^YMgVb4L%^h4I$hkcI8rbdX)Mfexwy<%v_LJAf7R^Mk0{<f^;
zJ+8Y0ktc0_vx^bi4!!IdxueGHIL5`ZK7*uLo<x`S<f6}f`&^O@ruOUi-9^WoI5{)&
zg)C3{we<s4XlCu$-<5bW$>eB5zBw5Z<sK{#3-)sF!o3!?NSYi(dIdsrv+c4v5Rp`i
zF&~%iM(o}=#k$S6Z0JtxYP3<m1V1KKCsPaXi=C|(RP?HwJ@wjP?7`7{2;y>jE7Zo-
znZO`(;C_@N<9&Pll%oDx)N?zqAfP7Xggug%dfAVRF+4ghn-z3{OZpq2>60|(A7>7X
z=gkPhR5k7~-q1YA5$;X;d{h+aGnYPIkm#ZosO+<SZ85;-LOq#1JAMs63E@l2kyeyo
zjEggEM70PZ(Nqn25PKF%y||Hq!**#Dz2rclS49IQB?L5cFrnV0PcxSgz6BVAe8eN9
zP1;ju%59z>QCdFhjXn2>-JqUk2gL5@;R${A{F1A$nBdIRi=!SB;+OKS8#*V<jwid^
zGpEyBf~*U|`mbL>esVH=C~a`(Bbg6DlQ?f1@_W-VSG7;(tNhN7d!xW3MbP$nZ+)GB
z93<M(a%H`%VQ%&YO~L6GwFO)_-rdT=YB=C&J1*Zm@*gUQaC&>3cOyR$yD|-={7{_u
zO#Cr)P%-t4zU_GS``ddveT_a1@qoDS3RtkmNi4yZBb#1nDH2&WD|+azd^$kup1g>(
zCz9#8tfvZsZ5sI+?w4k!cuyhKAg=cG=CRM0%OSyBP;I$mgjacd?Ly)n_~-=iJ@a}B
z)(=R%4r3puif=Z8<r7IQR6$d|Nv*-b!^no&WTB%*i02n4za#EDfq**B5+Z{P5Zu0_
zf}Ac_KukYfut}j*<MnCBZ#y>G-tS?wmP8)Vc5J%(os{HB_T2FYAJrp_rXZ#sXGL|5
zw}Lj1^MaOutWGRJ4BmB`U>N{=x!~0%bQwutYy(K$pj;xosT8>moN`1bRgg~D7HqNc
zTNME^eQKAMM=ReNcr+TEZplauE8pGjy-}mtzWRqg+ujkoi%gx#TxtTRNY`WtkLr{@
zkMB}av12mwIwG#EN~pJ#Adz)Y^bQNu`Y3)@l)36rXY1Z5Y)t$iLdnO8Wh|YY)tVAf
z*Cs6f%zZ)~n2QLUZ+lB-bi3`?$TN<Y-OdRO9VvSWMs|a7z!Ir~u%?F{y`H7=SNQrN
z%l=4#dhvO^Eh`?N94wl>)yeh5EjhKbCL`f-DDL!&+4|cTXM^aSZN?l5_TJep-LsY(
z-J%p<6g>J&X`LXpmlENUe)_`?Io9oYre%2<<JA`8G!SwW%)8Cm<{A5pZ7~{vi5{gp
zSNk!k7PHUq)DLG#-D>X&ppF%58)blPLAHFKJ!5-++aLVE!+W6^j--6rwBR7r%Lt5C
zP^+gs7H;zly@pJfUAv<?pB%NLG)+vk`2PM%lEOQJvy_Nkmx3RmF4T(ziZRyy6ogHY
zw(8iy3tvpB7dst&8h{n%4JZ|5-}Oz_ZBq2Qd_t1BM8mw|qJ$S;lBC1!8rz*3P2M$m
znw|8g`bEsv(^El?6C!qL7@j^dzKp(=yy5CDG~g5=G_{YZlCnlyrGU#?K$6uyDFO(D
zj|>ngiI_eFo42wGkC~K~-X8L`dD4ijy-FlWhjqXz*G}JvTN!OLMC^9$o{&MGqIqz+
zCTnksn<sjIcs_-(U1)T)9)~`OaH+>jc)MJF`28a?60*Q<h`54Wh;@N$^Srz%vvf9m
z7n%FYM*h=YdNCI|uxG&OHwc#S8S~=dFrE&1Q^>vEfithYZxnr45IgS;J;$>?iI-~)
zn@H>UvvSVXsx0dV#I=XPN0+)29uS?hhI8SFoQ<5hz=uH=5}*sgqzW7@IC$zbNqoBy
zIC!=SnVOpQgV+XlI5wB}WaQ}&geL+|SLQX--nVk<MI?#3ve1K}<)JP=WLxTfCd_%E
zAuxuQL5CcOP9PBp;N!mz=>C^v@a9_RC6iIj8;RGR3L9|i>&A9bccly|{o(@DEU&!O
z-*Dvzw2mC^!jw@H#ZQGB97m!LbvIqVHP5^&8&8*S*}0#?z#i3=loC;xrO>5umsGwk
zm73p7Q9fP1+t!stSl-Mz!6ZV-RZ3Uu#%hN-<3nK_`Hbj8DgE?80@o%Bv}65P7VX&#
z4PiGa>m@l6H}wty`cCEXJkl|&0*34KUN)?Csc0-vB4&tU!Gimail4H%9ncw;yLPqd
zu7l9i<$ZzbOo1h5p~A|R7rXh_zEPTJ%s2~l@p0!nahlj@x0pq{L@8-=Rtb{JT8D==
z2`7iPc<ED(YWB)7COLbi?7l0^;)wG!&?>uipEE9&cg{IdPl%Rg%IV9}=cg3W!&!^>
zLMrDTS~)WEaMyIlw)4-GT}{w2oY9xA{P@|r^slnbq}pD4Zzc*#;fk2ybY=<aI!U~q
z*uva&A@jeBH--5diE`8Ph?ctTKl1@Cn6TUYuElJ6_I$G)n`+D|K0DgLoD@ZrCeQV2
zyYjpHI<!+j?K}E51C-!)<=F=yePLjzf`VR}?cchL7Eji61hkwbuc7$SA76kPOG}PM
z`HG*Ciulg00zJ*B`nY?bx7KK3fUbchU>jAS48xkf?55y2aQE$9ZX05(G>Hw}kxOVE
zK^U(9+OsM;_Rlflc3a>3JlBdICP{!ke;KJej3$IyYJuc{uQ6Vc{5+eprTF_EgSC(6
zgPerw;GusXG~Fgqw`pc+AE6%rdii5u&432If2u(U?Pd@@+MJ3B(6>Z41y}D>gsz#0
ziB|nbAr#+}GN#ShMT5vS!-T@!ApM03#}QSa9|0(!y!sRM-*t2ztymw5-A5=i9lq(+
z2@&Gp&kYOez9;wL)4Uy^qUw&f^69q8*tti1@~my&|4}Z`w}o4jk07t!zTxVSfb;73
zihChToigycHYk%E7>Z0OZQ5Gz%WV&st>wjzV`((lu?nm8EVQI5_wnuVlS-gv2^~{9
zF+)sSI=D<jV6HXX(WP;mY#Ir6Va0|&1yH~pSqYdH!`wZ3hwbuoBO}?gy&$DWcw7q~
zmPY+{-mvaH;$?G1QZ}oVQ;}=Uqz~JR+t&^L`1Jak4|eYEA#NxEK@ad|vFU?o-EK0W
zZvoX50b+1`=@vj|%<T^cnhug&8XyKK0S+hyuCE39wCARUHokAeX!Hj_nUZKShje5N
zs;JdpD7wa90Q~0qfd)Trz6{wsnm_Jaatjcr)ZB>EORj33VKc&(HGn;Co~bSCrZhov
za_lA+N^5PZ?LV!l@2uO)l5J#oO^YhCY)PtjTi-J;?4<I-^P-8dihhfUqC+7f>qthw
zw@U{%8_lLTdVWR{>D>Re5@If|v5x84M11lQ`B6th@=OA=P}}YxIzX9Y$o30BG!^0T
z6PLe`G8@#~5~&mzY52BX^upO7T#er~5M?4}=}-3uuSRzP+EuDNLk9+?!XHvoX#m}^
z)-&bQHY(Z(x1s1hYmlolE6P$bUgJOI;rq051PCiqNoujtcwt9e{vYE;Ml&n9cbH00
zq;QH!^u}nS$cT~F?xd#XnZEs`rsx2afb#w9;<3`SoTLDtxFi#vL>B|fF&wUDdm?*;
z65QjyUa1Xsi_1BYCq^EG620eqf;-WGAVvB%J+x<H%?R#aFq-~wOac_hbBXp6M-)iX
zV%ju~@v21o6`w04;QnL6vBR{~WhJ}iWt5o&g3<Qe3DDYQgF%3JfB~!{(loN2^Tf1m
zLdY4gMpZf}VuK!ZyhI3S7>yUyC4Q9`eD=tI%^f5_XA@GOoiFr&?7$dkAY4N?CctU}
z|F=+}{bzbPc@`**Kp584-fTo#64gVr>T|JBV{$UUJGmhL^7B(+(WM1|N7f!Zss~nh
z{xwZOhu~%}8O4*SPcM`x^(WiJOPgoP>LZaHCQ=)Xuv?#K6)Qc}CxeGS7wH3PtR(7u
zX-*gA*miTE_bI56TWCo3Jsoq!^;<Lp8_3HCCv6}N=tlCJkNU;Lkbdq%aiQPNOp7(A
zyN}n_M_x7OVCp=&|Eg=KT5|P#VJcg42&IE{(!p#yHTW{DJNBPtMzw?!y5Q_xUVxxV
zdx*tc%jvy*cv`Z$Qg(E#WE(%giB`5Ww-Y3kkif=B1897=fS>;&ov#qb*i@#hd%ol=
zAww^_iO&0ws4jratUd30>JKDNAG$sDHpFzuWSNrfb|V1B1QAX!wI?ksRff@^W{ozW
zwlGUBuv0$&gCdGq9$d8~=fEUu2L8RQ068b$X^4)K*%k4%2Wvhg+~8l_5`e8?cW~35
zGkD0Hdj|q0Y<^@Ac1mn`B}o&IzIS5>a{y;i(4OD$H+M4o)Io}p$B9rC?Zi6H$+ENQ
zYQOQ6()#+ktDcG7`el^D(D3{Y+veBkRuvEpItc9=I=KI%HwdUq{BP3N#M`2c@4iuw
zGlckycwNGxgP<e{JIvew*9VMFc)JgL1Tq~{(KBni%j1lTMdyxE7&|kgv<RrL+rUSy
zpMF>l*MHMo+ndZ)qJ!2aptd7$KrxM9fY8vpOF>VJtI!1|W=DxL`JaJyEKtIwi$NC$
z@L-@kmG+^jvmIn2l$caRiHrnL6hLVx2)v*JGO8&znqS~NnEuBHzwA2I(R=+kDVPpw
za*Yy*)K^SViYkE5<H8ALH>zvrTxfQoRFd--0DAjGnK*IfngSMOt-u<vq{UzD<SFi$
zIE<4yl8QcqZ{?4~vIFV@;p4O?d=_PD>2moui%t?RBI6cJg@(MgN)e;oCkCfL!_7d9
zow3vD_&}Cq-q=%ckLaP=DX-kaTKUG1aEYO=HyI2=c5bH8Y{KWU@uswHq_j|boJmL$
zm2_U9Aq6eabObgRaFPFWcm!NmFNWinC#{B%%`N@Dz0F3lnIyk)A`(B2m$(y~*yY_i
zOI@XMq|{y$Tjj~6{kElY*mO{+5IGcyj`VxHJ(Es>)g<|gZHIaST%bO0fjVc8=ZIS$
z^H1xr8ZMt1TspzURXVA=tba_5WAZgJ0TJ-&dpoRz4sZ!Y+sf7TKbYB&0nd+cacw(|
zbaLCxVYDr`XEvpd&gj^5-*ny8Mr314K|&uZF6}Whd2x7wV;q7JO&Ho`8RCu~SqqBv
z7xt=nPwHs#8x~)(3=gp^6X^FFmAr63*R?V@Q>Qd>7D8>yF|=ym(x&0KM5zJVR}+|}
zy%g2(f$Z3n?$OdyK6&me7kO)d$VeFL3T)Ovq;;;C$O2Xk^q5p)Hx$Wp%_k(~t_>wX
zoC6{bpzd?Ac-;?h_R)a{=!fAL!-4l)=o9Dl3Ow{z=MibH8$}-@0v^>dY16`!gW|J~
zLu>uH_H@wX1Qeysk_r3gS}g|E6O{TAob=5zRqpNm;|2E(|F+v{-q?tfd&D?8xIGC)
zZ@pO5SV8Dpzi7I$4j?oVnawRv=36P={s@>U2_1A?4j<TYAY~4?6D$k#tzWI$@Nqzp
zkNMko5TOmamld!ZZqkyGZBSNV|2dK|fX{dRIG}es1}+ZdEde6Cps!scYWmfZL>bu{
z8k!F1T;*>n4T?+gfglg`>^A>4<?Py$(%RSbVgaq2UBCt&cz@f#;|P<Rir*?>xAG@6
z8eK?noGUO9qH>$_J@q4tb)7GR5U!M6&F;@H40w|VfX<CAvb)v{ut+9g8U75t(co%#
z%sPmb$<Ws3XJXmbt>y?bUuo5#-%<c>W+*@(r92(70E6SP=|H#_`oS>l>7wY-uQ?Sp
z|6Id5>M-%<LWh!o(xJf5#;*i(G~SA<$<2EI7gZ3?`ea+XLz@z?Q`Ci6vF#P;x;Bs7
z&=+pXX(i)7o?<*v)0gOxgvhXozrB$U0F{#Ql;{ao6V3146z-#MIjzs8a!Y^?pudC-
z+&7?iMtv#i|Mm_fK@Ikbjwis112yj{LKZXw98R2M`;{VcLHf;o&VUM78KAtcT!?@v
zDI>7aM0@B!+=&!uc3*QI>g~*ybS`SL*CRVdXH>?tYIZtw=LahHIz}t?7k)^U%`h9u
z9Gzg=#;S~c#k3tOr7Z-wtx;X2w75lVLj#y$%g=<VMUa^OsLeXWs`a(S0ZL8H>)4%-
z=ACQ{{#_{_eS@SntM_Uw7(1&~blL>KzHA~z{a*U0odE{qJQzyov~r;@eToKrE+-2o
zh+QhWwfa6k&`>ioM)z~YN<Z_64CCK|I7;xrNCE{Lg3ACsUm2Pq`0IZLa{ldh3|M#s
ziZt@>j2Phb_Sm~jD1rD>W(N3~w-|r1-wUfgH}Ic9?@zpl@b7=~6NC~?RJhXsWgby3
z+QLZ-*!V@no9jUdb5V{;i0m)+{2zm6_1YX9UNvRnEo%=(yO`&B2Ss;9Il%}Fy6O-E
zVH<2P%S&$Hh%($LhA=AHB7fOqCad*U&^0Y~RBfPgrB(4;H2@CRnjahJF7VaB5oK5q
zt1<|!A!(iSXWvsPX4=8`06CKlP{6im8bu#I1Vki>NAS6*{swb4iG@rJo$RvJ9bMLk
z5nWZ{87!Dx&FPUZUjN9Ed-+v!N68L7->{N|I?tk<GFAO!dQ~0>hmO9}>t^S@MiPbb
zIXZccz0M<+*bacY{#L!y%+zPUb0eu|dum^Y6bcF9dnw3+%J)_~h~}F+F{?`Tf^xjx
zrLKhq8l6_KFu5X=z!!n>-k$>FJ?ui9v|pU^IY<()Gg1DZN#zIgBR@&y#LApq=2Jg<
zev``U>Ek?=v~Dn+Nt%0VG*D{aU7xgM5X#W6M(uiIeb;-ySI{&^W(NEMx4{u@iM~%~
zDG-;Ae=h^$_H24z35Wy`u|DQ4GF!{fj9Pwc7F+!CaVFD!L|@YCOo=X0G$QXNzr7Rp
za$EgX{5l^cPLz9v=mv<#>!0U|DjI8@_WLb7aj_b6Wm5zuIt5+k8=dl230?AU=AHWt
z-Z3{&M8|aW51WfMkO4azR58pRW@Z((IFC6fCunMO0?N1vg5XIPnV40Z7`&HiG|K=k
z4`d7Pe-mCp1#ktUR`zia5th;yfmeIktU(jrrO(?1q`Br=WpUqP`U`sE`+>P84;Ev-
z=1Yasu`R1It1}cvpte9l6*<s$Z$UGeOYhK9)L~4b_X5!@)RQ<~l5-}hspCQ~7Y6LW
z18lY?4eXv{@QL#6Kf-jhw7comgg;T@MF>4dY`6Y!pMPKSF!1?KII{Ma&@SS+&>gv-
z%iRxSi`O4CN6g0vl>>sYGk^ww2n!h5DUR{g-y!AST$27T1@vvnpbZ@_uWmMU<M_Yk
zf*9iLcjtg&hY*l&<&vKQY~{W0BaYp`_GJDbiwB@2YPry=mJ%(}Oq|7aN1fsqa!9>v
zrPlrcgO%A$P#+M{!essjabE!zW%unnlnfyuQc9zANJ}FrB_h%SN{Xa(Gb)lQBHe-l
zA|L|NsibrYNJ|Y3!_YBz52&BtcmC(x|2^kE_xd~z%DnHp<6V2NwSH^;*0yaqv4Ed^
z6?SMnSg$y-f&Gi*fHT+$66=<obP0p-qX*h##gQ_}onB7zI5L2d>2d*y%{GzAH4^^n
zq_D;*DU&d0{NWNpN+qf|4rEWFQ&jQun$c*USn^xxVM9XB+cFALgK2CEHQcj!j;*1m
zF5!8^YrvAtc@V{)Q>K%lc(XG*v-hh?8sLS4vQb6+!Zcc#_vuYZU?OsuHvuP{;ukj@
z*ih&sFp|2vJ80rH1a0$l^D&=RKm#5U8Y;j%Asq7U(A5Fu{RBV@pfJsU%>LISuX{g5
zEFW@GvBPvel8yAY*Z@7SkK-Q&u?@k%TxWnl>ljF2=xqFfOXMIMU*XxI&D)boJ_V72
z=agjCdIx>qSBX)*+jQ!sfB62bY^1$;VWt|(ZH>yEg~nNPx?h%2a=9Nd>_x-w4y50+
z3ik>Gla1nM$0n~qNMN2ni!#A*F=QZfkA6}V(EeW>D7kP_y$L<MyWn{zqnV7BflT|8
zi(ub`P6PJ`|DMmu+Nk4v1ZhHX*h62FG=T`~`7dZt6zY6L`I!MGrCb|&5tJ=L8`hZD
z@lPyZ58yM+m^7Znn7M)Gl&xMulg(j}Dp5)qME;M!pMKwo3kcThcoG4_-NOT35*fxn
za8w2eF(!avA(&LlbeJ#0L4^`*P~bGK&-8z*ay<W1ISEN^xUM;Tk<s{o!tiilaK<-k
zOeC=p<%=k@>ES15*wp06y*2E*mm9^_JUc?K+45v|s#{4Oz>}907C81;gfX>w4MSht
zE*gFwlD*qK0!k>nB+2+54H!du_g~4<{TEJH9cLB<6pY;|h7JS7rLMEMg?m?j0lm++
z@$r$<dsR<b2~DRBnwm19N6HqUtU;ifrCN@2mmQhM4O6ZMju{$-TKxo9*4cYcCHw#Y
zjaB7zM2X~l3#lrmeH4_D>Ych@?#1SRwIF?83v}&aMc0M30K@m2_BpIE<+*Qr&ocw3
z3s584Mx&#<fiB=`n@y=y)9?~tVTnx3-z=djf#cn*p{4c@Ys=M{`MvkY{o4B%ee$kp
zwldRBKnrhXkbW;DjhVi1P4rgvxT2a`V^GaCP=d`h`bJEUskSHP9l<b(TXOeWjSaM9
z^M%c1^TXH%JjJr20AC#t--5*AFWhrQbICM-gsiJ@K+*vZ+TtA^ng{JcHq@P1w<s}=
zowBYznRZ313NN>FTK(=>=<Dl1zJPo(D3a8HC4*jfOfbm441}Z(6dh9lP~RiKuH<_R
zIU{odYMjTw?#2aD!vfa9Ip}L*Am_oP*kDzxt1!SeetS8m7yl<K;NMy{piIURQZ;9x
zG?ew|c!%9DyX3mWknj3I(ZbS{$@{!c)<czbuhD#Lhxd^M18orq@yf&wK$~ItZQ%nl
zaB;%hRzL<0)?BYtYkYfAK%ZJ#pETq^&UlLzzO>Gl^W<}PRU9%FVDV|n<^ILpLto<-
zXU%?&^aQ?9m7z1nd<<#F`pJftfI={@5dS2ao@_D_KaorH)Im0X)>JuJSpUgQt?V#5
zUMkYXclnWX1eqr=A&ztoHSwupn7@7W_FktRytrUW24XCV^Km%hlZPsyjW#IfIZhO`
z6&jc8RQ`unux@eNai1`WG1;-+v@FXITN4&D1y!0rjYqU+!Q<ZO`0l2DHP@&5*%Yzl
zt(jna7;rA&f@rB!*Dpbhl?=J)hk$Yk=cS*DmUTJ8X?3w*ck0^kfl@!{dQ0R6;WnT&
zh^v^*S3L4MeDa$SU9M~@5e}-Jn1%Y^nHJKnL|Y`#e|k?*K_~+_joJ^eXOkXqB<G$N
zD?yiO65}+he*~#=!k59|#n6ZFVhn@73;%QB0MVYm?WzBkME$!rQJs_7Y4TOYwO@EX
z45U1?0dUj~mC+SNQyWeJfnU`>PNG;$qM+=@A6VOCYK-4J-G6-n=}FD7gk^M%4=UhN
z+$13b#e)9k8ZiGZ5r*~}&NpC#-1}3n;RJR=2g-QNmG(<ZUmPzGVzLF}0sjA=cGzQ5
zjHCuS_ypa(`EcOV{L2ft3Rw3`-HVRoHkZ_upM!cd<R=v*ZqM+?ChAPH*dU~|Bg%#(
zeaEsoFFtuBHZkd14vJG=Q15;6ZH8Z6r%4xRO6r|Q#py*8ulX-9;R^MfLUyl+^gz|A
zo<c$)pOd;Yi&Y8le7XbtCLbQXrr=n6vs^RGg<Oja2b|hWOmo$2T2+g|?6mS;-{<g4
z3d66v>&I+pdEeZg&o(n^71{1rvH!~{iPYXeJ0*)5fbwzUIPHp21#t&)dcSNe#=Fh)
z7Ka&}t4X(5E>7p@{-n+a|E|&1U6ns9OBxcsFJ@hH|J(Ez{wOqsR%CjL(hFU>0;P4Y
zpxp!L4Sk?DB(oM-&pJ2B<~N!v0JtWkx5I%df!6?11l;{+tu>B-C-W2^kY&-_8Fq9O
zt4kGP7zJ_>M3+AwuaLSebRSx3^v*fyS%WeJ<b1gUv;2Svs87y}%6hYfrpa7P%5NQO
zvp|g!2XQ);H+r5z%Kcjxg=s*+!z#w@wO)XBs+e!_Xq&{g<sf=|h0#5(R{9UoEA<cZ
zCBi&CM3*a?#Ha3`2CrVVMZNn@jDWEspP-%;ybIWWM5LELxy%YpHh4dn4x$5Y=7;F$
zfRG?4v72U5BUx-_q4Qsdu5$hiT}7sX+kbf-GOW*+%^w5|Y=8=_QtU1O_WD`1>s~+0
zQ^1vpE>)SisDH<^%XV?+QOS5>9htcah+e6fBmF3{N;~1t)%Tkmje%ZbK$?zY&;9Dv
z!o_+|Pco0k-9^=S5-tc%2TTxGJ^=W<Q#GDXaP=F1Z@hyBH0t*_jsYyc6Ab&OcLLrR
zAju{_^Kku@3vkoZthw{szrH%|kFp$NE!+&S^y{5~Lwf@Koq_~+6(PSUu;11lu%1Gy
zl)+*<#nn&kE2{n9gqgTaU}!x1=Z+z;N%<KrnD>>%&1g8|Y`o7vbPrG4{iP3_6`z<>
z4+@t;pHwl$j(p(kqAVYEH5$hHu#BNgH>0l0vQ~j$&=q$Yjy5s|G&fzH-l(=u2gCQq
zqA~R{(ZT&$&HEK9=B9<S91)tloDpBc5|(b6=YZ>RtC+m?l?Bu2AM_J*4--Ex^LvdY
zjCE7hllH-V-LEQA7`k6MSX$3!q_L}Hq#231ze_Vh2QyDY`>gh0%kO~nt2QPb_J8ev
z?}8Y-13ZA?jL1N=k%4dy1vZLBB&atHfk3VhVuR2B`I!gpC1}UVAkp*iyOvv$+i-n-
z<WALWf5za%q!%eVr=feW^y5p`piJqV*ygqBnTWLPF*U?H{-YV8gK+UBFR4%2I}Jhy
z?kUH9vt1;b!>6SpGv@8CAv*mK>jD<F#r3IIkeFh0X*pCC?d>w~B5oL#nz*2ep1JMB
zjS|O5u|15YlGKux7Z+;+4suf(1H;=#reCP7<%9I=1`4Sh`)rzoX6tQ~)tY+!Y}ukX
z%B*gv{^JZI-b=ecyen$wWWQX(_#%fXXBp`E%6Xu7K;ccJNK8Jr!<QzHYy4lnJj5>i
z$k+4*_PpIggfwvrcN)gcybQB=vV(JRgb%@#I$r7EP`Z{!iF&^%rkc76PMk2*&Z=BG
z@Jy5164xHcs?|24AN#QQq4Ji?tl~jD-7TwXi_6I=MM0N$2>EQ?E@WI}8xvEh4Zaxl
zv51)NxUi6ZZT-npx<<P2Zt@>u?ayT3Z@6H4lD9_$_pB>8&4g+Ngj{R6q@<$vHS0M{
zu8S<-S3Q1UXi0EnF<_z0Zmcw)qG#}3NjIl8&udf_rA$#|13!B<Xa90V0ZW)h)~yM{
zPy8bbILY-=vG(`AIcIgNp79)JzP3AzOYBcQ9PYl0As22Qt^|KIL?pmWvJc0f46@44
z+|L|V?RwZZA?M!m!`>@$^QMtps<ladwbsQ?+|FEi24cm<>^sVCy@}$uJy$*L6mP6#
zS&0*eIp2=h+(SN?PxCwMzvnv?-dp@4%0x}Qdgg#%=@nZc{nf+IMPRsCL3FcVxOuz8
z0xWX$AWyU0?`WUhuFofC{D+j3ib(4Z(N9}S!|}+@l6ZomqS8+e@%#OJ$z9GlrO^W6
zG>gksFUH4(w-k0BidSn>h<NzJ$A-SkFMizV4jXIEb1|8JuM`h0q?h>DtzFF>6z;&e
zNbg9W`nITWw+L&sV<hHVPYH2HE-iA>XvJ`DIL2^9G<AIL;IQj$Nlv8n+r(t)#l<3|
zlqRRv_kC5nS|1*%W#<LIorh~wM+#g}(St|z^!x`KMAA3mq^ldyF~+(2XFc7`s*t3m
zg*~tA{R|!Z$A$~DF(nqOnc_9&o^vxswYS1pyKz%$EJwC;Dkb~MwKTFLi~Hb>L5mu8
zStC(a26Cyc^bYB)@Or5jDY(>XmX)h7HM4rlFfq;XOZ0og3ZVN6CQ;i|texttFBgvE
z1joLNx5a!2yO;5Zt5#P>((S&|b}5yGR{ytx8GeMNfvc`kH{J-8x=VEN!ar=wT332E
z9dEV{Q;*b|IMcmv?lD8&>++zdq<fp?;!XVq-PDESkLacz$%R`25#ivYwQLWR3dDS8
zRGscBrFd_|OGVl%_11kqiLJ5>hj>o-_f#Ie+58^zR_#)Dl~Rd2qDYGh#NfU_a)B)8
z(UKB;xp}bXthCmHaJ9U$?Z&+WX~hLE-y!zzVrH9Hl9x7^l%if1(F-2l`za(HV0@Pg
zP8#*YEOTZ~E-!Ze)>@i*tCdpZLxzYRt>Ij@x8e42l9!wR<+0@yL~XF~yF5gei<S2j
zQN`-(BgX?5m7~<p?z&<%#-2}~r4k=drOk!Adb^&Cj_SDk=R|(~*$`onlszWf@>}GW
zdLygj&{s2&oSiN|z7fmP5x;wXXDJ!se|^s-uQ3b$dHV}Zb-nK)%Ta)V-+pUeYSsQQ
z7Lw3+D9d|3Y}<Vg+kGa4?Ep4KoP6XF1f(KuZ3CK2Hhr(j5pKo;t8ebrDt7omzP9g~
zv*Cag)RbVTH67f(Y~#S~k|9y+Bxc8rI;YNE+$AJbIi%B~wP*~zE2)|_V&%F^QGK0$
z+s&)a_fU9jZ11~ba&RFr#qm0N9M@#SZNXbgq9PW^l%waxXit^rwY$GblPQpw)X({r
z$lfVFqc2f7WX-O9?ZDWkGiYJKX|GK}NIExFw*1kmHE+Le_qW)*)LWemdO;~{<>B)l
ziPxyN=^iSq?Hn0T!Oz)mV+Bh4Y){krz3gTl0VF=9q#)|PwD4Gn{VGDKv)5~%ptL|W
z{1$4&)*7q2*={*Awrw#I>7x5Nc2Tz{CEaM<Mn7?Zc(i$`|K1&id`CJz1*=BghP{vU
zZ;LwP55K&(-V#Y_3UDl<zj8Q*7J(-Lbi?4yNdAy$k0bJ}!gl1S_V1CS$WL>23*1_=
zo)6!*#!b7p5O~0q<2BY_k#tKkXq8Lv`Nd3~+MyELgv`b7i+-|J_J*G*>~G|G=u!9=
zGr86oc-hAI6*&)C^%$w=ltdd}m@IyAfze9(=5AYyUsXNvc`tcy3EBAuk2pTMeLcL_
z<+46MN|bdj8nPKVAL3W+!)nTEEE04zE9~t<KU6B^9edFxi}5JvPW03)+acRuZ29_z
z-k?uS&t7}qSRM7jL}@_*fhEqjJTKBMWTnwZ*1S+kD*7MV+X>#hVkUtsI9EsTY3U%X
ze677$z-sxMpY%xDH1cq4J}*&pGiPtxbPh3{sZlMzvp+A6>MzP|c1x-9+HCA9p<I)6
zA4843U!`ywif?;iBwA`!<E-#T13pYla?Jlz$Dr2}dNQeqeLtZ|6%70g;PW~gnx1><
z<fx-_Cpn&YCnqBXoScj#`v`qunhf}Sd6NNr{_S!VQ=FHm<x1m&irU>fUy)bO;e8Q-
z$cogp1f|@@kk}1sn5~tx_qv3KJ#`V1cJVG!PU9Vf2!$cKm`gwE)Z5l#=zKJPChhV1
zSy=bkD`(O`&FrNX&TaAV2?lemTF+}O3_e&4E?Cd5!M~obW9DVH`ebmiB|;;fR`rL>
z?im^Q8LFV}NOqVr-iB0EW#mg*%jQ>CNop`iDlnQDk>>c58ejdl4LWoqcBp7o5Wnvr
zV~2fjYf%~Z6A^uH?Zg#=b=o;DS&g@lRn{4<abt}ZUiBB3AO37z0UmO`ju^Vv%_=#r
z+7hlF&Z_z6H^#k9J8J*<ozRM*W&Xm(CHw#fNPs=804f=Q`ElUTb6m>d_JBn8pSObt
zUvhAu8z7!%0c4%uSC^%}d|p)@`jS=c?;Vj@nVmVRI(ng5E#rRc+UX>Ob(66}KvmEI
zf;^p#*_rBpOvBxajP9Qdyz%yOhYKzNY81MShfv9QOb<05<%6S{7OVEZ+K>ZN4J&^1
zGG9ELlFaGV!tXoyppt2r9t+YD%yZjg5~?jDh|Ircn*kYsU*`qq&dZ-a>bZX`-)H{h
z^quafv1&tfri%QJzYiWa&bMYZM_fOBLyA(!27Dk-uZYlw>Ax>>|A!L#|LgL*&<}@S
z?)Aw*AY2!xF<a&dp72H;jN<ywN7wE=YIB+jXH_PQuCA{obJ8%J+BO?^LMBS>W~e})
zB)A0P`yPKMckteFs7382Ca936;aUcF`zxN8Vrz}0sIbNm#PjT{Bk(;Esaio)mMUI(
z+i>OfHV<l)2iB`pwXzYD`jr2pvgNz(wOu*COJ2=#Re{}41k~UnE@X8xPo-X9?pj$9
zRBiayTBE#Y>@*2E7zw#y&BlE!u<y9b+#88&eql#zOMaxiLo6}(WKkQOEBF1EeEkvl
ztW&s#LIF~c`ha5%BkU`W(!$dqHU8L){+n@QWjkGXp1#7NnwOt2;()^0rLTlBh8&`G
z{aim6u6l|yuG#MMLUYq(d5>azcTs_~wp)+i;CXu2b2A|^s4O*7FV8dyHXlH&G3Dr`
zF$7V;Fa7mg61QqE&mf^^qDcZh4JBO0k!xir)>oJw$I>KDt~py?vNadRcUfLB%A%MU
z^Y<)r_I$<n@<q=ENI~nD7uJ|(Vhya%p&xe`_fl7KshD0ov?}w$e%z}sR~*8|4#Q2^
zVPSZCs>43+dT$bM213UGwFalE`o&9P@o=awN#Jnbgp6@dw@M^cg+2yIXk)46tK*+1
zMKJPhV93^x9e+Ic`O2{2w*dz?k8%!JEz?Y!v4EYs1o;mt)))6=dB1N;H?e+Tnb}EF
z3ZP3$+eTtCI`r8@Qtj{J-DPIX3JK|h4g&-0+-FD3`x^(LkO$j_QxBy!cw|2iTey(5
zUP+UdZ;1%=`Ef(m*oG1Li0Xj`N|ez*KbnQ#r3ka^fYYSV9iHzq`yL}kH1H)^V*_=7
z+EeLV`73JMG}@3_hM9qMvllO&>#0WH@~gqaGjBDAg)BimW1n{8n!b}fGJ)<5_10+k
zUUEg8JseRaqu$?5#=4sLk~la7b*;sp*i-)Nz1=m8cbH`%FQrLoQSvSCEy8Wmsij<E
ziOC9kWZ_u+^6*qf^Nvabf6*#8r6uBzA%wn%*y|`#gvxOQ{7L?$nU%C~cJ=^fSpNHY
z^W~6Q!|gJbgju{vMcEgQHBcDnM{swzR_&SYibY2z^n2<*=47Pjs_ck8PV$&<s3QTl
zhAr;|!OH?)?r?eEA?ql_cHH7L^oKAYKT|OwYm14w*Y>ujT77k^hwY7iM6vKANzd2i
z^Y)gNRHHg$NfAMV(#FSo8TszZhjT+sV%CHa)j=GvlG>-WX~)ro6<2)V-vl^U#*X|4
zlUZcxgMlC!$=<N@MQEvzBIuJj{Z1c9D?iO!OfBV$z2><^9Z<#mA?*fDz4?SJdEGoT
zEq&S`DRQB{J1w@9*3LaI)t!}OUxtu$KiT~_N~CYwUr@x5?85$4{JQUJN5bDyX%79#
z5J7RIUr6g_0}eHM9zOL7q;t79)a~#QrK;jLUpamrbYyDBC3e%m>H;fzKp<S@FpAtk
zIres(wV`Abm?Gw23~(j_(>Ts9c1#^RyM3>$8&Gb1gS`f_RD3R^5jXJjR__ZlD(Qok
zJ@925B^c+Q3dg)9eF5?N;8#B<FWYI1eic)GMEfFmqNb~!`Z92=g7!D<PfC0*f_vf!
zCTRnFI6oC!dhUpQ1uSq>C+l&-%X>9?F${rtM<rn4E`<-^gN|CzS!|Ob#B5~c=&xZQ
zMWsy+R^4kWXNpsqX5g=I0-2L?V7cyOPR|FE{4X<VC<+WI-wW<eEd?h=PC0p>O(Xs?
znU;F+nYi`K?09^WGNCUvPWKwJou0%gXi(z$%Z00oqVWOF=Z5wKk^Y_?`DBQ!o@cJ6
z;MF_h?Oi+yVkg~aAf(&3g6^4(8yBKVYsakBOySH7Fp&oterrMD=Md5cB4)c^Ia}IR
z!>y7cTS$hzKjBSwYX*$2mds7awyXx95oaf)^I9qWN*Zn*#sF5C1d`SGM$3(SUus4r
z>3KmGJ|rngohH_CI_hZKGR||nkm7j41}Z9)aPRVC#=@QZWU$RJABU<DzxE?~*<)47
zqylxWCqXr_FQskXmuZ8|Qf_%6>zq@AzIElsq2dV9d}$~d71+b|@|kC<vuH%R0it9$
z(<1GczbQFc+lLZ7gLOrIoBKP?%_Gi(4n!K6Zgs*--LZ?`Ct0E&$w^l(`D>#`U6U1?
zJn{G{!fws`3E7@fRpbls)Q1=})7TP|4!vtW+Dmzk4b;W=(R=6D@eCu#w<%)p=FACH
z!5UMY*<`oYm~;+!=~o(3jHMPVjO&#A6HHwAAEjNSl-YPDLch0ot+wM}js7Tc5Y?2L
zyytj{S%pZ>V(m3vox%Q%#SW$F4d2+ALkh^_M}VN(6*K^qyehY{5bm=$<{*YR>+gHa
zm?8WSJPoLJh=2_E_!q|G{|nV@aU1()TZm=a7sK*95hAJ<J;`mGb5+ZovoTec+Aguq
z<(Y%IVFS8O<$;5DTDU{4x(2E&1xhE2FO<{Jf4%jAr^0#DM|<x*yffOPfUVLQhwofZ
zHCu|WiyjFdI~MnxxjDjhCaOi={YHqI@@>VY%BAw8XKMC1mo;n1DWc`&M(AhGT*Q%=
zYK6`_tLXf|dFP!wn&!;Ud08bs`s5$wa-5nR-nld(70l}qpH`qgmwer@EmWUF6e2WG
zBDJI1^<JCqz8cX{JGV<|)6!PK1$N??^dU~R5jks0&oDi_282<a9Aq82LuOQM2utIt
zbVAilOXZn(vuDkMsZ)(iW)h9>;XDFOTDDhGOrP?lm_}}#UwXC?!4<NK)6^i&D!^M2
zKjh-%a9x1!)nmAQs8x7$$}n$mTyDA#Eq_np5I#<(d{fxvO!*SFis%*(_vd_c!%Wi-
z6=%%&3q#>CAw)&eGx1bktKcQQKaQ;F%2oBr_b_MZ3?sz_wF+N}3r6aHrA_#nJ44iA
zuHzJB$<J9X*CttV@PfhpemQfCyM6h{i`TZ<qzs&H#rymdht}n+xPF^f@$D+BU!8RA
zMo4E86_>v{De1whnc2XQeK6lSWRk!bF^~ZNy#fOEm{1v3DU3VU7$I7i@MKBsyC6J@
zO#-`N3lg&)U}4|A4)FyzFd>NZLomBTR3T4?hYcAZ0+?`6P4m~acQA*p5s%|w#F5I>
zE;B-~WFR0u>0)}WE)VHMfcfzm^tHn_2@Hb)LIw_f+JH`$$45ReB9e7$1(Ic>w_!dZ
zc=Y3}*&HXv*=xzso<&h}+y`X_TM3)Fot1WWULB$E65^7}1(~66ONAPa0<Vz@QDZgA
zMt-iguyP59Va&;nD!nGD2OAOR8)fV``x~x%^c)k%(kL6XbA0Q+FL=Ouo*N8^3-bhM
z5yHgOTO!@;+dmfsIV=n}f-qCmP=m^?e%j0wg*USOWqbHFF()s2BXJzLt3NMtz;YSt
z<pq+bxV>XZ8(q3h70<7@Uo_FS*^(%DF@JlmouloeOINzsP5oubo8>HtSC8dpVvo{k
z6ct~vj|oYj<bA_bZj3J$uX{I-QjC)v$PMxpTbL|S=*$S>5ZI=Ov{{cAqKJ)l639#H
zDpyd`o*fBbv}Y{xhwp0lbjaBYH!iNo>8IpNYn|)rEUb#-47QJ%>X3a=x(wU9u97a^
z6&10b?%V&~vxCDXcp%KmL74LAcxf;_oEqbA49#_6tT$Adp6kTG{wkH?3|X0)rjfZc
z4>51Kirak_ynP}Jn%A#beLzx2I^*@f-qS<q<RnY1zF4(&pMVsI`1X)o@`l!dR1lC9
zn@r_TNJNjJ-{gn>QJxb7X(*BdUtwWSf;1Ny$U_W_21Vy<FWJ|1o(@#UNG^Nk>sx{G
zgc=tyR@k^j`RutUbJaUl#YK;C)hBxPBo~O}D`~on?1e_W_3SC98?xyPTRo?xc*{_4
z@?s^I;;n@{M_**uR6&X{F0>8{q`E;Fgb`<WeV(7IDL814S)wNTqvbQlnDEc<Mo9@V
zuYfFOv^-zM1qIHpf`BZE%$-x={OdtzY5(JHw3t7F(1275C<D)_#GZ@`j0l_^0#*wy
zrXvg@!2m0RLDWv_bP|!kaIr}!PbP*IS_iUnV2~e}m{(4cVon>?8$AvYDjK2RS_}Hy
zO++1CH!JJ|ZydF_6P+x(Umclpq)qlVwYH`=>NuSHdqeoe#o3fe&B;j1TK&`hDdKz_
zB8iH>Zy8$Es23>WEIO`IB<Mg<SY@GPxRY?X;2Eu*cJBDXT!?2Qe$Ix)a>t60O)gUf
zHNCQt%ZNi&71e!Z7XiL1+q5>BS-6d@aSc4<m{g(Z6^){HLw`%#AR`%fYhGPci${+X
zRZB~Q&RrFAdw1_$eq3=&YW-6Apkex`M-&R~%-mqL$$R-D3>S!Y_%>S@Ck{{c0x?CL
zo8hxoMVBQC;!?lGJ_XNK6j-GcH=WWwP!g1w?;F|g0EfVVdFSJRWn|t>39&3`Fx*D%
zdI-eMb028<jdNKx@^6bb+r)q$#s-u*?iHS<#sw@D$*I^o39dSu2HW1x!p*NOSmN-#
zDui2#^1Whm$e4w1rdM{pM<^s764bq_e-}oRfU_rNo_?9sIfXr9!!&|DLdb{Aj=1(S
zhp~HypKeHtNBUVIO$&b#=7l2ZAti39SFhmqnT-idoyXYwb$x@jBYs;1$3>^|<49D8
zF%OY&Bw{%Kt-CrWyQ6)@K#7s}uKsu{|9I<A^X!Xs-=WE(xhVS#HbV(>i>>WIwXpg2
z;25q<1wQUD>fw&4DfY$>SkaA1aoV};srfDhoHx*ilby<sETh6gOb};KjhgOpMKv?+
z2CgCVwrJvwZ0|c880g}#r$UF%xnL=h6$XF0Li>>5ZbJYBW1R*2H3mi;7A7>Y7zwIF
zGJ45+2T9;EYH$j*Aua_<G~bF~Esk@lkx%J2MffOdyR?yU=MC6cRrem+wmtR|Q%LE5
zWtCTvLdX&NXl|g-rb=(ojsLo+VV|v{*Mc_5A3dCM*>~RAxSW6xAExP2zH1lWtUURU
z0#?Y!2%&^B0CvI0^7aFQ$$MU*6I_2=8?sZ%UZ8?fY_IB}B{-w)k2ORz-<^z;^RT{u
zk@DytQcqNCtd#XawP(P6w-b!-k41WkKbBs^ckyylEnX5f#;9<;*#~1$k)qj5%WzM|
z-xCTJBs7e>p9~`Yy!ULBJ@<YW!C%t|4kNnSzLs_eGT+PH?zMUzMYcvF!}`~<CpjZ6
zF|yZ@5d)#iuu0ZMm_As%Jh8_JNshyIT+0ei-Gi@wLJ^Y4m?Jw*lq$&?-47<UeywD6
zbY2AqWznmf$@D;ZDnU%LTxRjuEA}-fL>XZ3;Hf7nl!5D;H|EvYRd@QG?Oj4BxKIHS
zu94x_|FM;_SG~px&Ik*Jf||u(Cu86$gS^Nft_VL^(68B)g@_eE?}}7^^?dgst!o!7
zB6)e(ss6q98D#&iOyO=#OuUQ1S>SRY*ZXQSe4z5#CPq)wrHCdqdbT>ySBA7L%-%cE
z!h?^wqK}I^lv>}sDn4>djXDHw*?QeWWFl_$dZ<$ZasDOapu&Bp@}256Q_?yRT-vLT
zN8SM2&)+(B3;FS2_e!!&Dx<@`?nk-D4r;a5^iE!P?HZV1skDw0o~lc5IX|#|NJ&93
z8)y<4Jx2^0=3~rVy#p|L#HLk*pFrGgc;`kgiOgFhIO`VJCnOYiAX(n1#$k;vhtxDp
zGIK<gwk3JK1wD(3anYAK1rbh<j`#4A(f~c<1<T>ifh_$55$D2@@mF+D2^}m}+(W7G
zc)uU;(km;wY(8IjVQzebvhjLra!&rh1$56?V9BhRaL?Roop8N)F^p*Hiv(kDLN6a!
zj`F%Sd|5mCdLJnR0h9SLna+{ER^9!eGc2-BtN`!sN$keN`z+>_Sl<~;Dg^O2pbvH<
zZ-hd&yLl_ZM;3N`=br)_=n(o2k(b@iY!~gN^YUJLeCK-1krer<A<_42L%Ud~i)_i`
zhDB#?A))FH*L$!bjAKRH6fSHPX=_^5vIRv%n5np*?UL4<Fjb@p+x@2{v8l|#1D4t(
zrpHS&l-F&5HA9r;uifqx6D8MCk3H?f5qofcce0<+Ra_E2$-?kC8W{BAJfs2xBLIgI
z=1MU(@aci!LdG##o*qJ9d+06^Wq+&f={x+xB7l7)!SwLMGIHJ)lq9V^BL-@(saY9R
zcKcOyIDlfe7aoJm9x4bXlz<V!={#J(Ilqnk-U5s+8caO{4gV=&w;oBqf7%dwofvoI
z=IIIrs#7|5$C>ajgM%R;|EEB+#o39gEm6P8kour(RPz(@Mg4-CABy5A8<U`8JQZV3
z6+%{uT=l&_loj~Ks^1Z6u)EQ|qY3dyvo7fLsp*wYyZ;DGJgJNsL59r>lWV&4>&ZC3
zo(;AR9I(LVIbP4z2n=>66(sI;VPMEBpwP=^l9_=F7^r2;arKu7UQ7SBSpV7x{=54B
z|EbYob{i6Q_F8<1WoC=vGYg<Zr}G4XPs7+|==RAqU;kX&P>)XIVzPZooNn&N3noqD
zqsB!|9O;Q05&a6KM@?;Ek7DFS$m-ibvfJ$!HBED~VaAK}n&w<z@7-2VdtQZhYtl#Y
z)k2<8t~`14OwAVOAy=X9np<fQuk%N<m<U%@m$;*A)b}-AM0!otpEsGJg3~+b(QElN
z=$Tt<!56ZyJ!@ZmIC!Su8^4EML(CL@#s2aU|2-RE#mIf!=9_+p&F~tZ)GM5fMDm8~
zt?7xg4c9}fqGR54g{_L^v{Kq3cp<7pZ+M*UQ_L*A42h9%_e8s83YqfXNXD3Kd3(0k
zFXsN}&@Y<MX5vq|FQ<Ncn7-lq{xb2lW?1;!OBIRM1uXT$h9od*paGMx9P30O#7S7t
zh%8;Z&9#8PDKWaOj`JXMlmR9KkS7>~X$l`7Ll30!kwBq!G$b<TN5zXVFh4SY&76b;
z2GjtQt-W&`^Uq7LmtW(qJ%(_BBupqa_FW<<gO7bitf3J$1Fu}PE`>-NLQKaq-cG@I
zusurEa@Ms&&6wYid`x1-vZ&ZgbgY6X#>BDmwXr5=zD*1zk9##;8QYlz-TaSpjkijd
z9z_$lC8`Td)Z}Pm$J01O>kZosi@WyN#ON!#6jco-S>g0_G$2$&+Cm35O)q!!R>en;
zZO0&6tmH+^HhnIcQXh;pE8RLj_5-%|>@8Q;-uhf4f9oP=v_Y5SPn#MVG;ew1y(^iF
z*pK|ujh9E6XSEh%v}b12YnNYAIZe-b1^%+d*#xCiyhcNBLHyQyp|3@Dq1S&I*Eh>z
z#_zh4%lu_gp-h(8z6YI^vU%aYdBsp+n0qNNT`YAoiF#w1t0U1`Q?1QZ(M`a@L#1Fz
z#)6O4lH8_sp3|r>l>9G?{mO|$q(+*8tyABiJUNLzff((wfQ|4o28ZVtE0DbYhcj@`
z(J``(D|!6W0(B}#e@AZ(e0Y-!sxJ&(zu<)Lc0n-Y6(>N_<o!A&nbk!`$i{LAd7Z|s
zs8031>?PlM7$iO2ADo9GC<9qtWHpBHn?qlV|B;gdtRfkCY;cl6@EK^Gkt8-liextC
z5ISvST+aQsGKbxVK}HAUAU}p9IbdDApaiwjg5sZ=7u{pv+m_v(##)_@bou50j&Wzg
zJl&4bE(qC@#p}Vo&7%sStQWxPkiUF(P_)j0-1l|f&G5af>mH_j(s{fDQr8+Id<7OM
z!Qo!OIQM-VN^pt=0MNjw`tf+bF}^CquQ`TP`b;A$jAuBA1;%%l9&=IyA43Mh=Qyt@
z=A5IGV}(P_lCVJNFubOvskXz0xMW~+&N<Vfx_G~2uYX=mfF)IsxkA6fN|W{-*Y<^X
z6)cICgKvL)_0uF`W^TTO`(FOlg24>$jkDS;38W4LHD26@<9wx0Mb<$9%GWEyJIB#T
z_^H*{a0&M1h6=C5s9B&zGS>Z(j%a5{Vakj%#J9*~TofX78!S-5Q-FgGgGMSQW&eKH
zA7<CjiHnmCG5!vhB)}#i#rOm0P=JUWm^^NI1Yu&5`O7CRfKJ}_LZt`&es{kR6e0{k
z-QW>3TSOu*W^f<_P-AC+*G3Ws`2<pFDcP{=3?C9O{NY@KV@R$*@9N8(T=s_r{9k&J
zpMFCI9Q-<`hyituN5_S=xqD(OXL3UA^&Rqzle<_>pn^kXXPoguoVvuPG!Db3&P-?9
zqfP<iPvHha`T7=T3-+sgdTJ=s8*#9IJI9~i%I&j@(^=M5O!x${ZE4)69&O-tzd9pM
z@W(ZR*seNr9k&unmwj@Kol^!p!Nve=u}aUlJRxoQ(%e#={7$P#9$~TAuGzi55r2$z
zOQ~kVlZ}+^%!4&r(7I*Sebf<?(k;rOP$8iUJ;(b~lnk)WfgP-0zV#JvBAGmmc2O8N
zGFChG@&xd7f4N3zOFr4mdLmKcuk{ODa}wdLY_t7NtiM_HLlUZ8tK!6ebsi|<d(I+b
z8yy`2tZVz);b2|=`P@X`sNQJQb?ZgTyf;adAAtp_-)Up9Jl1M6C*R{YV5*a0$Zjbg
z5ftB*HI|;_rku6MZ|$8=q47N<PP?M<`ixBURt45<pOc4vd6iGc><8<If8h)awjW;G
z_*dMGu2s~P=oya)-yJzC0ZiojuE-_2*dR>|u2#-~*@1@VgtdCZQi8!tufTt-S$GpW
z-HZoD*}mGuTU0L_$Bqi8Iv4QQ#ti#SZ{!r(v=+BN3zs+kWK<lkX8QVaPsfZ!-%}o}
z=r{QxHWhY<yt-j2mqnkwFWViRM>v#|$DPxEBXAh%v{CU+*D$0c$VBUdofUtfeF#yF
z^ry48jjMQgpOo?)W!#1`*rrsJ3cc#;p<ST`#`!b+*L0fvZE9S?6#Gi&sXIOzH^c=_
z?1X*}2C&f&46#9Dy;tU{RoZ3*#%-OYKZGdtj9iKe2k)3`=UbqN-?qeRba_9|g!)>u
z&wnt!QF$KmV|ay4vQ>Y)aH}7g5G~029l)}{9pzlI`xuxM!<_B<VLa2y06-g;{4dKN
zr(YLn^Z(c&r5Hj_A5qI@iu&4>*l0V?`$n)0v7=`0JANb?S+W;pqVv$t{wa!I)o4$W
zjKyC}!A)SVkKF>yibmFXPfl*7RLU{eje2fj78K@9znDxOr}aHNs*NJPR*kx-y)#;J
z(b5=NG198jm%r2b$?f@GWMQ5T5$>0dKDu@cQwKEtRZG)LmwU$4!i1<h#~(w|r>__-
zR6cxoP7ik;LIW@Xta&u_1C8JFov?YndC#k2P3>>!Qb2p}8L#k>A$8uJWV}CtCg3ec
z*q3$Db-HRT9cp+3Q~WQhlSH^9hs{VtL}74jTKez8JmtLOxn$XS_@Jhl=5G!C%|lM9
znJLp|y|W$+oK!S?_e<ozOtG@7%x5n2Z`C;e4HknwJro>^E=ZV&PpFsx8s$&5nD#L?
zWqR(D5=#F9C<X8bAa-+?tGdzGC(0g~i1Rn}i|x+G4UC-)Ww9j@0_?kwE%07*w?lEd
zYb3BIfy)Z8GxAdfbqWvTO_sz{dLd;Yn!Vy6=b35MP=Wq_etRqqSC^q@lc5`2Az%M)
zVIp~zWdyJUlHHUSVz$054(!?XmOITO;<@}Y(bleTy&<*t!TQu!dwnFe<Qk7vLqe;?
zG-?mH20zH^DDW}j&dIC+<$e7`v%MYYWgP2(1r6uZxzxn5xt`@v1R!S__LbToSx5{u
z2C@yP>Br(Cz43-yOxSlR0BTByak#!ba31r1o|HWXvWpHAntG~@G-FENM!8elMa@WZ
z$CruES2?+>&u+&LbRIZ<i%yJd^|IMo1aTgr0k0e5)3?W+1a_;h&34=gQ=>)$);bgS
z1uHL+a{{F<fm!Bsw7;cA0<(~X9D$SHW#`umko`*DosS~xSOCEOLmc-b9p*>1WGpmr
zPa?d@+M{~0+UoC^3O>+d$6rr>M@23p`PNxgBE-+Nc2BLKn>#D6&>?V(sY&qgL21&L
zAHQ_lsf@f<l6=4bquUA&xdrJUk1F~KJUz4^ywJxm|1WfW)Vys@(KEOuLb;NB;8Ck-
z{-gIcg&`~H!92sf?}x3#a1$&MLhjsd1RC~@&pZs*2W+Oo$g^F)x%LrPNxRVB_uekH
z7k%15aB4)mDelt144Oq@6a~pxH<u!vh!{>GXduInfswuYYqS)AqSXK>OPFskZ-P_F
z&tQ{WCV>%vU#?SNcZ2E)XoK7D7oLj)43-V7pD+^I&JLzOmGM=m<z*R2^`l<_4^S53
z^jikuxv<ABF{Oyy$>L?oxXlEd7|JoV_SQ<z`+251T+B;(1En3hT$a6!owG$%dTJv|
z=cQh`%x@0WbGq=DIEMOEI3rsZ!y3oMfe8k9JTiwez|-v{=%pTSEle709jc5TY!694
zdf%8^l^FDm!?Gc%aNNLm#K(;<dF_r=TbQRI#7Ql=>kZ-U>`I_we2|`eR#MD5sa>(&
zz@9#FFo4I3w@&&JoGA6P0kn_+-YD=!e&GIGk-slI*#-Xf45BUO-}Ch>3bHGRej>{w
zi0PX<KhEo?h-^M9E!-X-(EL=2omyOnpLd2huQ>BD9IQkBMIYe>+id4OLf)8OE^ftF
zQ+dy5w1mephJ<g5zDfsBXZGz-gs|<Szk0?rxFVIb_*_tb&{WYRI4E;$;nsMNz{rtQ
z(qT&c1EcB)%UJQm%xFXVLY_vcH@Xf%2|_8-XpMe4lwD{GS|)+|6vnn-Nr-217)EVP
zb=(0F5>)28CF6+KjB**-<pMS1Z*93Bx(4sx*2{C=yO0xX-=uWQ$|7I;m*Q+#FFaM8
z1)h89v1lcy>O`fjLmu&U@bhR3O{f#+6*~Pl4I*(Dj1{=F&N&cxZ;|^-3E>u}tj546
z59CH<!qO`vC~O`GQXD9qw>v(84`|<De$*nui7-#~>fZZdng-n59>OJsXV$v|00zfD
z**}}#L1Gdfd$J@9p96iLp6<AlQo1a?=UEhICyBT;8d9j2$*{C(do4A8IJdxS9svOF
ziWSuIW(3zq&wPtDs$Zb@{=WddH))nzW|r>Sn-gW^deEx18_Zo)JDcBnA>FXQT7IP7
z;iF3IY@b9=V8XjX>EH<DDE+cMi{h<wV;!ILBBz$Qp>;rE5OreQcTK{&|Hv8AbLyl2
zUKOW$;s2(%o!E5$>GJ$HB^%(fH>;iOE!)&P=mcaSnmr;wk@)O`N`?U%6Tl!R`a${|
z2L@>J3H$@=UO<;-6gvWJWwBtTUr?}D$Qp3kZ=Y=N$;Y7FKw6))3-t99J?wQhqorkY
z-pov61kiV<*wsBUbg3y@Oi(P!1sMO^rwR3`zug6BA_P3*<SO(d9C^mpq*k|3ON9UW
z{$Eh<Z_xDg(E;i(^dJBfOX%C=D{)=JI%kAE=<)SJ4N17qL$Iyq{;{opOeY#U*0^@{
znKe9vM_+;7OX_V;5bgf!i>z?Og>oK!Wq_20(z(&oXFA($eZ;=<p@9)nyDUdm7k4fj
zawtoz4lz@Ci@APQk_+P*JID<&3rLjyrj!5M%X>rlS5!3mj&H50&VuDaih}Fw!XfY1
zPU^zdHHGI~x5J#Ms6{hf2G;|x-~P%@CZUPm?hB+{XtfN%c7IQSwj#RCjatp{fl(Nu
z*jy~kF58v+%$Ql<SlHNv0ftik>cQ}J>BpOAtK(Mq<7R92D(OiBuc{Q-XK(qxu@DPe
z<jDvapSB^@QNz5NEHwkt{pSxl3Ui$ycJ#M7<{>kvf*gdiTBeV=*m^T;+MS3&X1h%s
zchbjkY_BK(75p*$10RdDXmqMg8;a^it0)YT#=AxXtClP34EMX}6kZ&48iMCWnO$X{
zhhPTORDnl3_|4MDOp4U{ew<g|OS$DXP(pjH#=0Se%RS)fxcHZM$yh@-h|9t!GAoZP
zfBKO!L)f4S*{KxJHXtQor%>^K@;ws_cOh2*@#XzO;lGw34?hApS4EEXKuNFaUw{J~
z32KN!PlT1Wff@EIahiEw=qB_v?uj9h<qyd8C)dFlG7m7!dBojjM^ot5q@fJ_`<BE%
zm&y79C|}l%j-=)?)-v<+^7QaTH#Fr2m#UUuazngd=84`lHi#X|IM;RKLyZo{#&Uv+
zd&qpX-VgC6+3ZW0+7+r!m#Vz8Z~wS>jk;RwW)HWnv{(yG)T7Q=b8}y^=!lTKZq}a5
z7MxKtXV#M6bsaM(^?Liiyl8u#WRHy*8CEz?w?t;qC0T+p%o%ynjmjGk^|IJvhK<HP
zuHB0b^xTZymmhZ3T|uQRj*?c?WJL|^nB7Jv!!lShfX}}rTK~6lWh_47IF7Uv7|i~{
zd_GaG9U(Ua=*N;v#6adqNy$RuyXLeK#7e53K;=<SNv{`&hb%jN#ctF<BCnZUkwMoT
zpH^gMPSgm=*L%A8Nr}Tweefg5eaM$iDUaQ@R&>5Rl@~V1m)8whxFUd=7G44=xoOvT
z2x<n53MN+=bNeOtMzg+!FK)|~srw4pP4Jt*ODqyn(_F>~llLRUvI?zJVg~*zs?_&s
z3UX%EAv9e{AI)<3EE{eLSPu9PX)P5QkeU>Ahjj1l68^IGO(^J7RgT`>i{^cs=z?Q>
zt(d#$>2cv=;oI~1Z=5x_xL4lfXwT|Mu4bSvg$)ZK0|T{>&W-6sHfCvMuW=j1uOJ_H
z-b+<2y|PrU1;6~HYB_@8aQR74?X%@(@k3aeQXx14N7*(go)qfqg0o&<q^srIx12Fw
zwy>5aZvOCBDC<-H)OhyU?=hk=ZsW=6CBm~(UvGFh&of+L&F^T*K0?~g-+TCi!$~UD
zctxT=JXI{0_&~GbTLdC3J3H3LWbXPog>9@LX%95jVzT?qD4<%9qwY1C!{2biI0y4h
zq%-UienDya5<w-$Y~qpQ!@|z@+_q!3xwTToGB-W#w@RGmBIJ^j2YW_FGHc*c_F1K^
zKhCTXS6?(tyiU9QGqAUGXN!&O+V+!w-5Mll>w9MTPh?Jq!r!f~8a3Q2_I*1ozW51U
zRgQ0)#E6rPIEaAzq@v>c-CPktE=8M4M!qrQtPd<y4E?iLV!IlHd{5K_?VH+Q96)?<
z+?BJG9-qR+m;540mnPoe6UPVA6!Borl9DF8+O_?y>W-Y%qrD~VV=m1iB3(&UOS%s>
zeg&{yORdD*7uj74-7Zw`-aN`)aY;m|*}#idr`3_zzC+EuxMR&n$lMRSW^eI+)3W?w
z-OR&a5>?cKsNbZ8gJV2ST(V^eK7}ruO8Z50I?sO0uoAn`UU*byfA|!Lh<7$yyUcCB
znBT)08!%?GbMG4RHx?B)JJ_-5l-s)M^-OA+dTsj46J(S<uh}Qi)U|qgVKf8G5#F=B
z0X4duS^sRfyZI)hI7jzWXEU3vp~iyJnihUXcotnLlG_=_=%5cL=TN9ZwtP|euTZ&x
zf2M%W{S5yjX!6mnSjCi;%1T5w`C!dee^fJ@Q9iM_f}xcyU8b|7@l5bJr|l<!(zZW^
z1U<e>B`1w)K-}%$3Y5Ba8ti?$t+Mx{sTg@C$E;1l)uC-XRjX??|83%!)-h!dL9zC6
z)T%A-=(R!GPxSVwH?mOzyQ?bAgPi#<H#YClGxr}B*yr1$M~6{QF9t@3p2(q}4NY}7
zt3#e9M;UvREw!6yzVbmQ>x}MSYnjH-j<%`RIwodM{ZPDnUUJAd?u|Q8BX3i0wrQA@
zZ$;`#d&5%Fh!0%We0q*~QYT+XkXiuBPa2x?1Ec~8q{7)0uvCbA+p!&Y^}OD1t5^Ln
zl6DX}exAY-iaVpb{Jq9}CGY8nsCnHRS$VT2>Dd|E<(Zb0Blw;wV+0tvRtqi%xkMA`
zlneFaKQ)-R$;dl0&F#-$=%vta={h`O<tHP{1wMTlwigM{IBS<u?~)lE>Gib#vpP?a
zi&r1ZM)h+v5Yn-CMbe1JKR#<NXiJMtq+4?3=|2AV{lYD+)_AI=ZnDce8#E3AZ1v;J
zAK*td6Pz?gUpoBi)`Pmx`{Ow;bFe?2@O#9N8|UgejE-CMPh*M;nv4F4+;*>b%ASuQ
zo%DTQrL1!7-a6v=uJwt)kK0Dz=+S{kve>g6IC*D%K0J3f^eX-K6PItIktTXVdB=f$
zaU}maCF@4U4zE|s4<VP%&X0bQD~FoWD-15oV~(j7+WCnFv0Cr{N>vz%QSZBLWH_b)
zDusuA`_hq3PTck*U*|26o@#hUuY()W!7m++RrO7id1e_G|CBoVNH+ZbPsUl396cu*
zLe>}+{`%l|;If-?uvBf6LuIf-y~lOu-GE8{8?ELH_@emt*y`sg!v!_@8a)fsT7p|f
z#bvR*TT5)!*ZuCY1|A)rshC@~_WE)4NSKNBKeR{pkk0a1kdzLl$0IjiX~CW3a<{o>
zW8c@T#9fk7Y7eSh1+4spCXCzf_>Cg7W8uaqpZQd;n607CNL`1T-Qn#Y)$<<?Y7Ga*
zWQoL2W4-O}AMPXZBo8pYm?)yx)5q>N==Fr=w@ssU8u%lB=6_04QO<=RQhf()Om-+!
zIpgV+n0~T%_PRl21Q+$P?69<1J^a$6pX4AvIH=??CC#9ey!FKJh8MM`%?H#d!|k8+
zCN$yZF0E&y-~7hK0NX_oUMTe`u?Ns}0I_GrV&<8S4VEZ^y~s?N208(OjzOkLy!de(
zc^JoooZQb9)-UUw!#yQyT@6T4!#HH$>oU&S<3P~8kyEDoeFqcHO>b)buOn+LN<#mn
zL>&brHI6Ld?PYxZO%ggGI_+XG((k`*285%6e^8)?RKM@0{Ns0JnZ9kyLXs)+k4_mw
z6ltutASD{<8|1%o0TQIUM&_TtPhkC%9|amoomuI4vp+TZ_9uVr<jx(k!iZ8`QbfQ%
zI`ytPsVcQ}g6};g{07JA@XScPm~ie^M^<^I4RF*XRSo`QN^F^DUg+5HGNuZN$29!x
z9Z?#q79>1`*Sa>9X7AZQ_*5T-Nf55{8>X^1_$7beal&fjJz{lg7_CXP!t*p7NB*99
zs(_T67<$Cj8CzRb(ib{X?<d&)Bu7aj-{RMC_-uEX9DE1Vy5YZ3>+}GjUj`I_1QV1d
z13v#h<yQaOEWv-rcs1jKan?ExAxR*?#LMdpF3$)UQh5nWC;v&1s=Y8t7=3hD%sIZj
zIN^ud5D|)IDkN4r6qgWgxRQo?#^!S~ij~`?5A*&xMCMe?;r#ZF+ye7>iC0^f*EiVR
zcguhjb)4yeCPghjAv+)nKg)OL7c;5^(_>l4ZY>DVpg#F;jyV*de8+gb`eEjA%l?Nd
z;vExS6RA~S@U}kdoE;{X83j!S2;G6B`V}B#RQqxGulINzF59i)dOKGmatLBvyGHOl
zt@>)uT#_?4+2=l3N1YXKTf|W!Lu>~c%5P(l?10Aaou%C8Vid^~9ml;69&#0bGHo-{
zzj>?BM_K|!4=t}4@RDRS*~-Is%2@cD<>Za!(_zhUa4py~s5#J)iUqkticrlH3`q-r
z1997wZl5`rGL=2PB|NcWh=ng_j#3J^Q#Y~%-TSy1yo}2@M@x6{%e}bnzmc2VmM5t0
zXXi{TuRCoWt?LH(4BKR`yvRCyAOrYB3f{-fyA1;YhbiCzKPSx`jw3ssISBX>56{bl
zP3Vpj3yIG)?=@6H%g3|MT&g!mkuWOpAVS3PJoBC|<1($HTfM-KOeI6?7+5pOiEzRy
zYY~?Z*bveAcZ3=W7Z0W=1F8&M+DV;0Xq>$s`1QEw_M!*hLt4hQ!k0%f4b}B#a(Z%s
z$GibPw$>9_Qm@4J0+VzTTDqGzXk-YsjvUDt{h7p$Ltbi23BBg`+2)~VF@FOZzBG<*
z*dn-FTCOF*z4+M4W97v+lWM?sGKWs)1o*VvVma>2I&(XQ6?PF%^vfO-q>w(?!`;Ic
z8p|jvedFjtopjlmxR!2KISA$O$aR-HaxLCI4`eL#G0XO&^Q0j!kM1^9g1(gj&d<d6
z=A#{-1IO8#oN?Pyo`))+6)67XAf@%45m)7{qwa0@Zz2Z%k6d#wqNj>VQ}nHCIgj1P
zteFlT1-qkPB{dUr;_^)YH&*G4^NgsT@>f=X7j+F|^qSyt;nFKGMa}^?dmeH#qQ*dv
z13PP~R_<5Y994-w0u%geoNHB^Syl0Vl8lLYcxQe2u^mMjQCco{uC?HzyFIiFh(_-6
zCq$#)1D%p{7Y%<|iY6*S@fjFEJEz>D_?PUd7j9DX9LV`w)fyUaWuHq`+{PkZ19Cf6
z*uFGYqu!$VEg(>e8gsXNpF5lkPjm#`uLDu|#HE?w@_7r4VPg3z^fGDV$60@UWRlW!
za&?^a<Cp&PWd7<cCofBFyOdVF#n3-j2mPW?CX-7Ga)9TVER>cO{u&Zk#`eWagmKWL
zV^8-S;;qS9Ho}Ob&21D)`i^}*<L_l4fpnJMy{2C_S{daGXn)r*2p`^P-+R0AlQ~o)
z^coOr=FcHS_8l%U{c%?p38J~G+8o8}SJ@ZW(5)l`ap+g6uc9ZSRcKkUw>jnxSl$Nf
z?{iW9k+A3wK08u#X%fD*^&t|XCqc&zZFs{A#k|Zj@cE>sOMz0Dm{;G6^}hnUz!IT1
z*@T_++%YjDDr40I>(%F?3DuTVA?;C)LV8#kTEcAJb!%Ln2`=h&o8Eg~Lfh#;&Uyvx
zTpx{<TD&zx?SpYfCWr@W3q^*4Zrs^=AZzUE5Psn%)zyQd%Hw>zVPe^K^wic*ru{_)
zDoXV?&(=<=14nmEQn60X9!peo0OAjB#sunN*M*maNxtt7Qaf7wAV0>Z3K-FJ%v2;l
zZ`m8^>V~jLN<bhw-{r9xkxw@_|LW7ulVW6Ye`2`T^a0P;t!#|Xb;Gq?^Uhl4^hraY
z_rZ*E{xe!@eti)RyGxtI?8PAjV`V@YLE<nlArRQV!yf;4^uqrkH~kG0EF*)b<8m>z
zU(c;DRoceuDaQIoa1w)Uowamw!*;3R@LRiWVf;0cZt(!(3Y5AYWfZ3P;#RUk?IQyW
zYI2O*APv0@q@l-%Y~qZ!eiU?wajJ?odsaIGA_5fDh9o)jV-`(|#zmhKjB~!|4cG|l
z*V_zGPc!SE8zXo|-gq5kbhd>>M0s#_G+f6|uVg>v3h67LGq)`Y$WPMFRMkRkHC;F(
zHdZf~Nfau8{O_|(=H}!*++?(t{w;uq<Jn`EF`pN4ZS|IpLJg$6`5R!*ME#9IBY@@)
zp-D2`PT?yx%q_GAt58I;;c5vpn<2*~A|s#H#g16dE`y&aG7p*2*HELpnsyO?gVnE(
zCJyykie7maj%ts%Vee&<f`5@xc&$NO7Y|6YA@+%$X#-^;UNneA@}CTZyVqA$dZI6C
zxopXGV*kuQc2je~1*wR)A%^c*VZ1UB(=Bww7lJ{k;C8k;?Vfh6yogv@(l00_Mrx`^
z5c@T08f0$gCeKM)1m#3{CIbWnZb7O^pqaGfh^JoCo9$ExC>zr`i7p%SR(%(wmdLZf
z=u@f_@G)pqbE@Y9A^)qnFAt}xd;i|YF%>e8nPn*Rl#rQZh>$sl%raCcGY6Hacw`=;
zRAipXd@7P;p6MVmb7Y=**P(iz@ALh=*Y*DXd;PI5*V%hb_qy-3?sf0a=ibCPSfnOh
z<iad1x_N2cpmkVD@Bw#4x8LwXQt;N%`qx{>G-zJ0c|C>c_XsZeUzL#lnBCF6mrCll
zS<?hg__EHFvLId`od!2rJ~cknpYB|TgucQ~svmY(pHSM!s_1dnXy+y(t!_G4=W|wk
zQ^oq3$}{)FJff9(vvwe_GzV~c3Wvzhn}@vd+6j#!e9aNYLv13pNwjEspEYkPmCEOZ
zbZW`8OTIfmH$1}5b7Y6P7|?KJyBGN1dO4u410Y#0)srbHc3Ptwy!ipeTs*>EU#UDJ
ziiz*oy4A+nI*fjeaFHI)8Q}ZuA}v(x-Q`R$Ek0x1<T#YW<JA^!^l4Oh`)W?cNV~M)
zl?-Ar-Ye`iQ}S*ZHSB3<Tjhek9z*eg9dRyxg;yU!k6)LeukEa9afz<eS3dYFE#unT
zL20FDEyi`BD$w2{NyRG5`D^HJqvjINpWdRWwzh=`E$H1Fy7P~y!^lv7?f=Q(QaW<p
zdoW){bEfPRH3-Q9RMF|`%`75=9Y^;jO6LO|B8&Wxv9dYtASiE&2x;e>M1~xTb2%uy
z%6JFH^Rk%_DLq>qy!mmdC!2t8fo>)JFv@}QLfkXf=TrGOyx#nKc++*Z(bNFMSsIS1
zu+-xFC{mk9j^0c^%1De>`II6F$Oq9sE#t$Yf2c^1qM6hbv^*ja#(vp1ioMArAb|TQ
zkMWJf*e~3P#{;jW<{0k?$4|FuUI8Q#M*}L+l0?c@YcECL4cFl}pVcRZjIp5dF2|u1
z)5vpz@wbNB126ElTT7>3rT0_29NlKAYIy}HXycZ8C1iL`5&9^sx(u+?YI3)Xv-GwW
zEc*Su>o0!)Xq1j?py)AH;_JPxjpN4Ld2}g8*_FbGt6%C=-nm(_Q8f#jCXDP08FDS0
z%0w~PHe!xuJY1R3f@d|h;^G7dH?$0cSOq1Bw&!o2VgYaOjqvU!7OYDzpTe>sjPT6>
z!{V+AJFXDb26dVEx2##=Nc)`xoch9JBWYwqVj~qxiljX~2eyx}TFe(*U|79_h&IY~
zy^!}grAh3L<Uqbx8%3@szx4XE0lo+fJ7ue3tc$^TE(N+m4sf&DBW#Qw)0RpC0OeIz
z%Oqk-j+O`)=4xpQgu6h|0$p9n?(Onh{3&hg1enSD?F2HU-Ntqj7xIz?9FOhj9-##`
z@5}R101i1>If{s@5p&Z%Dwa~~3#4CBN-TGapOriPBPYvXnp6blwn=&Zx~@k0>=}Ue
zYgLt(=qA}3BKW?L*}sE$hWlsqYjy18I7S!E(=&RSv+U<<+8fu-&L)>;&4o2ZhGBpm
z7ZbklqJFb5ojF_W<7@W_va}Tz{9Md<$F{rHh9$>UCPI^sBSKQAY-=Nr$JR!;CpVE+
zWzUQ)z06Uk%(MT1AUQ!hwj-pR9^2;tl?;nCYOm720F{qA9%y6RnhTdLf0#>+5t-wK
zFqin5T#DA%CI0sjbG1f~0RbU}60P~<9Cv)N1i<Pe+#{@_M6cVdyPor<x7mnOCj?xx
zek*BhMBUDWohjw-3@xq7hG3!-wQ$D#Myr!V8<Mxusc}M+nlfIXPS+;20jPcbOrKM=
zhH!Hu;18i3r{i4WDRUFhe7xaR?Up)J5u3$n5<NpMD#F*(%E#nOEnleYC^uR1G_$wP
z9BZpqznX<CQ!~k}csTA0(Jl|%U<fBL3K_33z>CJY8BO?4B8tUzFz{oY{3lu_xO?N5
z-k-dT2lO>799&eE)gj^A1NF*7tVR5}1IM9@#TZs^G=Q1<2Yn3nV*BP?{}#~pf7T2*
zcK-zV{6BKlVt9}Mv+O9oJDSe;oBs(ww~6wwUWuM~Xm{-2Ki^c4fvq>lN#bJZCl}{e
zBgN?bRa0-PeKXBM#iq>slpc;<V!HnAvT9Xj<hz=u?>4kls=i0wtQ)Jl-`M8%<G}L=
ziASw8d_`!(XMtzK%WmD*WRyNMsy>-Ex6r_c<t{4}(n5&>nbAye2=4~C9`iz+esENA
zD!Ts^+Uq-*?)h0q5J7}HE_4qEi<IuBkVo#toZu^kXjow7EHVJ|K!htqCO!MU=I=^Q
zFf8)9+7BTtcsdR8UDg)(2d*OzwCX0x-S-aK$7t6lQV%ZsjA~}7M1<YWbd28Mv!&CB
zu+1z@Trw*gyWo0Y7GX)BCN8)-*DicMx5L))d2!Rp%X@&7;L?JQ!_vn!jS}yCjT83l
z_wSLtA?+4H{cFlBuBAfQlv|3B{}#8orbeQZ@Q`IN?MQ!P5cLQR%fy^le=QGgr*PJ=
z=A<BmcnD}5Yx2;`l^@}v&j+SXv7!!I?w$EXz66+px2;Eiz1ceuEZGXURC6SF<8Y>^
zFMr>z*txx87t$V^3hPFvza5$Fo&#h%5-hfv{qsQ!8w5WAc>ye+F*y}*)K-CEOL&Q<
zOy!sHO^aI6bHxjWVp6YuMjX_d{naDDY2YH>>zx#*EHTNEWQSL%Uoj?Wd9L<t0g9<^
zQ~As6)Vwo*I*8#CNK}38p1YQkcFOf+0*__yi)(#tSXG{RR8<~^(LRIq3Qly|D25*i
zZE^o06$e6jr)cGmeoyK#8n*u-(TgL{W#r)1kRusFj-LQV?gs#=JSNAumE`L%TmZBs
z@q(c)^FBaSSg;V+E>OhgvZto=_6wTc4d17DxZ{$*f1^>vMq2#dP*817U>Ffk$!ZS1
zm!FCjU(78G9u?-Na^<B>Pb^*&EHhI$q3UnCPRJv_F*7VMlD1#(_6%y49(edI`w_t0
zL0We#oL>%X>0)w9bJg@x8yt2S;P%14Z{Pf#T~NAX<Uvi%;Z}VxBn2`zpFoN%F5Pz2
z6TB29J=Zq8gu(f?cLrw`d?V)ssD%6uXk+$IH=>pPr+b&uN4dC%T|8*VM?J{o3BV^{
zP5@2B&8OA_vkHeQ4sLI`y{H&L<GtHr!tI3p!c9C0frdSP=#Zfpsy)irq$@=$O+oGZ
z7{3`t%jVN3uwMqrTmVA;U7u=rH$N%(VqsFje))qM=?0&%r+H8i^X$Pq5P{HT4Niq<
zYERJ=fbuy|tg{tB=YMV#(d07+{#=THU^t%f*RYvbG+(P>TzdonmOPto9Sl*Gr08lZ
zoqjdb^Bm3}W_N-%wYX%^-G-^zWcPk~(u}c+7q`6^?_9xF8OAMZzFwuuyMJl3Pra+E
zqv8_M(Q4wp-wSngUB|_A3}c<^*=47S<^gs-LKJ5hj?q^&qxeb@K&&348A?;>f0Ib9
zvgaI4t}=wKNz^mZl+D(S)xJ>e)$jXa!ns@cy3Iya>RcOeyqitDBlfn*WE5)9W=Js7
zeo2T8N`(U;cTf}(FyM)s=hkO1lNw^>z5X>d(7$I>VC@jcb5ODmJJtVnGwpR&O>TW}
z)o=q)6d2}TdX%c<L6i51L@(F8tZW3m(eTenlZTX67w)^yzoRnq3lZ)5qer1Ppvtz|
z@$%h6T?s5n1gs9IB5MITm)%~CS`LkoCCu(9R;5^VVNNecZRd|P&{)37??r%}3cULL
zg4cPJP|f!15P_v)9-kJL_N`SlAeevl*<<e?*+l`7lHib`$xy2&ki;|yXaA&C3;ONU
znR%aXqGRKxoRdUu{JKfTwdKFNofL)Hbe;F%3xeVmw$^SQXnO-B88$fpxjfQ*{xPKZ
zV<EQSGMkE_+`Q(x9E}2`H0hC+d!!~<igVrLI@VAK1)v7=`2xW$0{uN!ip?a9A?ZPw
z7>yJK(WB-pUAfyvQ$%xIb}Y*Qpbg+<UN%I=AdXS4&y8@$x4@v~&n;$3NRsx6M|!{+
zBUq7aD>5C_<F9Ft!B^Zgu<Zj~53@u2^8LOACzHea`{^uwn+I>43sblo4F$7C(o6GY
zQH22!z^lF6{8}bj<oc9EP3aJGdaGcF+uOaNsQX-XqMgNrSRb>)4e!z%$b$eKlQ^*g
zoPj<}+7WyU<j;4rB)xkqhJm%4Fs*-y;X@{1K6;)1Q_aq3+C`c5tS2AYB8C0t6=u%5
zdp&(-LkF;pQKwtL4u}=qhvQ_GX))Pj#rFU7)I7yMR#TN`B<e(zG?g&rdbk{Nh7Yw{
zBB<7$ghlh9OA=}L)5KY1u!d8T=_{Mp*12&xVR^}c4JtkM0$EXl_m!aUCC)N$Sn4`f
z<!HnSF2_FMQ(}(DF;hJIxL~Hr_l|Z-$Neffn=5tETz59BYHYzjnMY{xM<nGlHx}B_
z;yY!vq;}O-viE_J@`GK|C-{QI^4F;g@wc#^IWE~a&^lyMwy)!bjnE??JHil9El%os
z_zWYmf`e;~QT|W%0E49B|HvNjBy6gOvzWMb#nRwGDe;|YFA35X2U<~AlgM*u;js@N
zYG)1R#SLP5Y=obk4rtHt+<f$CG$9!PVGbL5Nnj5<3}@S!n=?2Sg2g>c6KD9Py8xcJ
zebw^4th`s-=}DEnxB6Sj;{EF;n>xAd1|2m^uMCDjl%mgZNH}3Tr<)q<+?V(!ary*|
zxA_KEdiJ+pb^TICr)E;yxef>w8fG}hA_8Lxf-E^$Rss})*w1Xs_5}9>SLX8uF#ry)
zAH8uQWEBhdD0fM5J~%7G52l?vFhSwBViXs5l3Aaw74Nu>XgyQ62*~;q<*5x&tu3JI
zzns}woy#zKJmo3OYmkmT-LANhKf5N(k!Acm3J4%w>iQdmjlQHA1$^q2hhfLufiuT9
z@~_EaFl-`AzRe!zz(^G`l*p`i8ZSys5V$ADxe<(FxR7r))@fa=Xwj6jM!4YggPEz?
zO$UQGA!zr%K831&+%98HdgMKSsC)0oiX0alPts@R1Af^)0Ax~fBR|oo4=^?a6+DnT
zIJQV|diE-)=dL?J-cqt5D;gEakb@$_vPBhsc&{!%xxaEXZn7_7R|&Apo_Mub{|t%U
zutXmLm@nJkcu|04Rp7(v1mD#O0?)$#c4B|+oX05gKReh-Wyd>ix;-xIM<kZ_3i$JX
z;KKg_cKs*%kT2xeQO+@JZ*!{39<s>*nC!oywXHAelPYjgTCd#UZ%iYYDA8fckO(oM
zg*_`W(Kv_jOWxO%D!w~5i^1gsU+p}FTO0H6xZ;bW#u26G;*0fluyHVzy&?>VMCkBw
zgvYrnxQj|YGpHOyn{tq&)zM5m=`^os;rXArb8cV>7XVI*`iXVL{{)0UlK{)k3h^f4
z7-?!4C2(2gm74NDND<x63+5a39#5KY&Ap$g=^-04tT_N`I1efS)WesYNYy!$nb3VK
zhu!%D!SV2)ormy%y9UA)_NTG=f^myN5Q_%CLqdNKTB#I^TH!>3+`W}013QCRzbp9V
zv)rnRL%4Y1Tml()<2QRs%dqG0zj4}+e_(+kH=)0yC$Vcd@!#|LTD|h0_$96+gR`Ya
zfN*YiK9y+3XLDcK^al~H!<^4I8E(sb{2>-&HJ_->^j6n{;7iujPtR5}ndU8nA_2e<
zxlZvT$v)E!L>-eQMCpNn<{1}$j=9SX&fs}@K)~sRH!i$W$wye@3AdGurmDmn&Y0$s
z-pIBqd9MBX$*8GbMAIR86L6A%04^o7Xv?_VgT0=;RSC%KJzO_d`AizWf3WkeS|h2h
zz0yAaNCzh7iZkMt>i)PgwQJ2194{xSy(pYjYB(3@<NIdZ(nLpWylQx!#c{N}UCZwy
z$mvCAVBO^=H>vgAFX~TD0(O%o-oc!>cb{x)A2pwZfdG>EgR1;7GF<7{@t5hOM9r<U
zRaU2`6<&MGr&hu_np2EdXftEFl1!8*P;ETqC{U6!P~y?YZZ0dcES}81{^m_;jkn~I
z?q?s8azy(VhyHcjx38tAO-*vbUEYjZkVx{Y>Q26LGZ37m=Q(i-Z^mkv$r8n~1lZCa
z-OxQhY#P*IVu9DiAFiUq7iy)G-Fx_w^JqIZATzuFIVt~6+5Wazk8rvf)rpeEb9^sU
z+Sx9>VRD!tc;<QwWj{!+o9~~3sTk(=MX6c*n*Dd5dDFdYt@V8E&&!QE@!BeP%_wP%
z0t%<)i0^)kW<JJOwk4}cHEuK{eyP>2F)_AU`noK?>a(8yAG&kt*I9li-<7didTH*S
z3)uB1L7v~rWeg4(x9P%m2|s__Bb-%`T2R(A5;yDT@;zQQ9C7e|u5&i$OMMSgHKFti
z-~gv<2)BswGR)|A7TaOrE~aBs;L?WV)O4gIu4NT?aP1yl0c;PT6_R&hmm2~LBz=<8
zj208IftHii?r+@O*y+`?VOYpFateiKVRHhg+yQapV<UB(z{Rp=>WgIDRbPyOP~zXq
zTo@(FZjK)_`O4kuzwi7`vmNNq(Uk}F=3(tj_v8e033j)@qn?6(T!P}{U}qE$4SzsF
zAMgP7bI+iQU_XH4o|?a$p8t8f3sMlo)Pc<l&~RhF7LgqrZLS_CSC~c9S6KMKA11cE
zwxNHx=d~J9beT6R-(7TPA)rbCJTfMzKaZK+ai5U~M?D5Kdj$EP<7_5$+pzG3gz;fR
z^dRxRsUbPX2WD5<I_A56eF8~MzllLbT)HP<Dk+kNv8#;6u-TJ+5D9rQ0agO|bj(Aw
zTikDGD-LItx6X!XkSN6Idx@q`=JQzrrux@a%XV2B5OSMr!m*+*G$se9hmsN_ndJ&n
z8?b5HtsyiRrJcGtu}Zy|*ukt;-&N(yma7>NnAt>h-=IU-PY_URj~Trd*V3RfEE?K=
zy^aLiPaC_WPJPX~M=|KA0DV8+4*DBUSsb48&c4+Aw?I#HX9D+b>8^t|^1o&+d;L&_
zl=t%;5WvCzouLQpk|5#nH&KTe>L0)7<)oT!+@Cnkl=4Uas0QQ%kyzpAW@`XWxf@s%
z2)U4vGBh}Jsfs2jqe_~XgE;~H(Fws4Kd+rO%6iWc-!G|4t*s6Rb0Y}}lh+sW*1U$V
z2%5V}jftju;{L?L=0&oHfpg$Jom@f@$<YhNC17co&lun<eKu?BRVdj3h#CRWn1|i`
z3rW7szH#aoP3LaPrbqWUr2)@j=W<*I@SJQwfTo)D$_Lfuf;7DadO$)NR17e_Z@iiU
zc3MEy!e0uQ+N>kx2F|lD?rX{@kdEY_-i!1&<4kZxgX`W{2<6F9AmZc&0-%Cni#da0
z4ada*cZvpv<z-^YV1V0B+}q#^|MqvOzf-)grk@nX0fq73{atwRRcyh&s=A&;<-pO=
zX8VR(<oxAHN@)SZH@ZIruj<BW=<CM4e&(t_L3$Y5vAmdPa8{+J^@^p!C&g(?fp}a6
zHYx=|-1nc);XfogD|CM6@o9Lmg7T;0s>Z9VMQtin_`#Rl8WwMsAFzqa0O+MmDOz--
z$IiQBZLi30cd19_j+|d+y6lKd)`*QlV3_5_4vi-^Ucw2K5lL^8J%VnA9mY3GdC@Q5
z8PzYmDNyn(;DCZR+>1-}{M(IG5^EygVv3Hb1D4m;V<WW|&wU==qlkjNOWDX2g$xBH
z8htL^0uch3XIK-2d9@_xY}!X)KYh)xwkq(}k5`Nr{CmB1o~n&8XBd7(LF}l$R+1HF
zEjr{G3kYE;jPyWY()nGoOmE(!L*0Pjp`b>OL7z+c;fsgDT>&?C^PYI9*PwfJcJpd|
z&QD@wd6#9@1o9b{H;i{`zn+^}HceQwET*WG9`Hrjc4(+CSKjPDpkby%Mj3?!U0U+u
z|88P>-6|uIca6eI62V8+WEnrzq+2gHr9LQ1U96?{$uVy#f52tHxTnECrmyMpWKx;Y
zwfq?;R>SWRah~-*j$S+{MrmBDxZvSETgYx{k<VsiBaYyk_R97V$dPW+jU=ckD4M5g
z)y(_YxT{~Rl~f$Am$yOdc5U~4wNc)uQX5G@k(M~$jp{jShQp1e#hu#Eh;zfA<9}$W
z9xO}z8mkkEQ=v1i>Fvtfg{Rf^Azz^HmU#H%oCfeqs-+*>z22Q-6r~K{{PAukv1sFp
z*_cu<?W+O=rfBQ#2S^MK|49)7>y4D_l3PxH$@{o35Y~82wbS|<f{)rWdjxJT-K3sI
zP(!;{dpmB$V#~7OJJmuSCyg6PbxI{|gjWiU>HXU|D^B|Dt(q6=y`O8n99$_hZ|7aR
z9HVZqXkWqZ;jLRISd$>VD3cn$cp&-w3$y;vn<e%0)s67Iyr&9&{rbg5pU{2=SwXWm
zb5|7A+FM86-WwL81tOZ%zhP?DcB)H#<2(xb>V0bR_heFqAQmvu!xVMaw2_^4ALmif
zx^R*YqzD(q!0s+0Ttc_*PH+_+>&6akXQlmIoQsCnrMl~M26*@{Na;|Y?rpg$)rL^w
zzkt&r6O4#otg&`|P*0{oFqr)70U-gq<!bv@PKk!m^m`=Lw3l#Lh*7GV5`pIL?%g_^
zFU%Je5hs>s5keB)-Es{c`*;dY7sFx~YB0@naP`h7@ma)ALSN8ZPmV-9I1=W>=piVv
z4f>B<;c4%BncBC5kn;PoN?c8Q8Cx#_F)9d#f-<n$?Muo{Y;ifV>;41E%fmk3`axBp
z<l7dItOZ0v4B?Z>1(V2Qh>0zPD!XGg*GY!w=-mYykJ}Q0i12|4u+X7Aztygl$@oRY
z%LHQ{(FKy?kHhgT-a{6E00<wXi(2=$nqfFD^_>{S!9dIk1V+C0bw5Pw{Lk-~=#cmp
z(AkMNe)nO~+0jn&?Rk9#VDOv<_taHBscRK~_=0Zmwe!?BSkyMVXF#^_*;8OScYtI%
z=Wv6_ShCQ1;)sz`hZBL}{4x{4PFc^gIW4(B>Wu?0x|ju9<ZOn5SR#XA_7Hv%jDfVR
z(<<Jsv+gCwPipuT9tIESwGP1*hRV^m3qEwalcs0skmd9wFCN6cgW<|C62LnlVhi82
zF|{@At<wbxTaP1A1aKn)@X!wNc*uYYrwhc!d{ViPz?)#@$_>FyvtSGfQ(z${0jz`C
zD?5kSN#HTpGU=0ckIe7%Ro|%XID!7{rEKMykHb@PS4PPed=d0Yekjxob+J*{h*I9#
zPRZxjU$(R&$4BkU@-~n#qm&K^V}A3drZhg;#4l-Cxw;UTZ9dj}02IX%SN0q3CJYQW
z9ky=dPR!c)@tzPwf`UH1dT<`iU*G_G^k+_-UEqW(cNvN0;XpGyZypdtId;~Oar}G_
z`J$k_HfQJ~bEK&$^$p`VS|TsR7hwap@Fqsm>7=(NmX=JPE@&0A_eU3IExC+>HO2XM
z-eT{K$)RVd2}x9b8`Viv1_uixVXI!zYQPlltff|H7kdW6o;2II&({V|0k!eOm8C<H
z0`XOmSPXu==gIjRVF6inC9$;&w2J=LU?IrS?JTM&WYPli2Bn)qeuzL@FMI<3)yZ6<
z+j*ZN9$glkqBp%C-jr(eRk=@O@;$qiewGpWLTuAOMva$Gabl&h4Rux~Lr45n?dNlb
z>wOEEhHv)<Dk-VWiR`@2^F9zWeI{Ku>ZmjU9h)faap$lie0M4jcCzYQK>gDfkz}OJ
z?;$Y=-U%Z{5fP(?zhP(N-BWsGb)gWl69u_L@Sj+ntAsuDymSWt#ovXwB#`6S(J&xk
zvk?BcN-iw*&Z!(tDmNR1C3qG82uIMyIG$Ex=kSJD12b6Ubw1_h&E*f=&M$c!aDZ=0
zz4pYe+giUotu*lFBSC|Soo7eEv5RN(xIDc1jWUYK@yYg2CKO^qLACbcNZZ+vF9WJP
zlG7X$ulghAjn^L(Dc=KjO5$+XF8;``g}u<;b2Z<<`6s6aI-sR{5mEB9Nd!}G{hd~O
zKm5v5+Hh9{;xhGqf<b=C)f-_W2`*r(>RvbBFEgo5uX_aB?z{~-GATR_v#kH1fo6sT
zX~Jt`t&9<TGkbak4`uYI7h<hzKMjw9#W`5~U_ruI>8og(T1sE9cRRkTe!WrHOzR=(
zJQY!Gul4<^o|r}06Ns4gVhJn(H)HClXG=-f2KR;}^M&`PuhGyUX}`gLHcGV0Ajy%i
z#>Zj-Gsz$7t)mUaF7IgBiLY~NXx!8jOMFrzXuRD-*^r?#`3NW@tv24#8~rSH8s2+*
z-A_9tQm5|sPv}p*G0k$5HG0{Cb39KLkPveFy$r-eECyAH3PH|bWrtzyT^@XwRq)Ts
zcD(KXUI>MbC5(dj=s=JCEdaNe|8$8Y;!A0e-j^^ML%u-;$+3{BJw3zJ6@Kc(CtDKB
z#&RkAy8|6|Q<q~TcjQ#nfH=cecWST34)9HH+dZH^^e<!n)0Vsw3hIabSvNajhj>5m
z@Ga<XL(bShTJeqXH<zvWvy9qTETW4u_3}irP+R4O2z{Ce&)~kgjo~G(ToNNW)3`PE
zn43e`EsbjAa6`k(Ce2Q_C7h|v;D^?N|2d}LYHpJWQ{Gg+%GRGKN5gcy0HS!=ycqFW
zAK>G^Afc20n%6>ZGVlA4UKm$1!G#n_I-?t}v??s?4=_cl%amh&17t$fl6=?d<P4sg
zXfsi1Kegj^MOaS|pP5CNM+&`aTzU`NxvYKTj0H!?hq)kTDC#5B*p7nAfRqU^*!38@
zB?5cHZdZ+@{~Nhuowq=kkq|^DzyHVZjwdwm*usF&e(gQKj1{}rO%VJ~4GPoC9~{Q*
z)J5Iz&xYWMRDr`Ep$zWp?sW3&xZ!8uKgpwDMXh1+p__3b{YaQ*xW&>fD4LQ_ikQ_k
zNF@Ze9oh_S_8@Hiic&BPCUAVA^FieyFlDWxSs~=-Q#4HY;~X?_&)|x!n>(*GcPDS@
zTR@cT4>yI7Z%?s-I&Le5V%vy-F~0e@>CVE$0=xzxHbr|<cxW<N2IvXb!&$JW7t^8q
zrEY0Qkt{!d4{<3If<EX_DuO=tJ|sJ7Vo;%QxQEjHDX)J$+gl->uW_DX#f*Z}M76Ot
zz%ZpEswg%!)db>KPUre8h%<;+%$o<q^)&+TvDa7y>pk*P=ucE-90fCg^)$aE=i6Ml
zf@Ra|gJjKD{b(EW3@=*^Tiy$iiUZV)EetmPs9wMP@cUmT9N$YU_c!YFQrcnN#(TrT
zK7K>YvG35#+rVqA;JbIZCT}b6bE9%)^2FL3NcRgFf5kLyzCiMgYvH4JD~+;(?2B*L
za&xBBjxlAiy&Fg{=jU_@aND_j=V}QXk7`brJ!$9d)r?AOoXyIfmg{BKRie2S^~_z3
z(XLH#Zo$X)W$v5R>$xo5G!rM-26gdbUF)+r>4FiOja4fCHfNvqL;2@`_%7EuYMZ}C
z7kG+}m;2o!SC8DOK<O-~Y?yoq^Jl>+YvyGW4|TDDmxOB&U&}DWw-@*7ihvnAyLJ&R
z<j_RGtY=eYWR!RdJa6Ftbz?#Zd23kJ7CL2<fiI-FE)O64q1g*P4rjK8Qc}+gA&DMI
zn9qfK<OH!(HaP_*fQ;9c0^BsBI^u#slJtC`_gA09`K1U78Mubq7?4M(6tVZ8zh6$7
zMAan4C&#0c6n676&tCZU3xRvJ2@euC9Yvm&4BAv?++<3~&@F+9B-~JWV-2Uu=TW_d
z{7S9Th1z88Ih$DLWV{-j(-)S!EYy{stSP^kTXRo6|A(at@6ZeC{GiwCm4oWnBY>2t
z@WsIg`Sf!s>yE^4tXFHY@@|%wy%&@t;-5n4CB%x*ntV~5ef+|y-o)%kmzxIuE(lSY
z#XXyOOaG!f<>38R-KtZ3g~bc2kz9J1q!#JqiSVKTmba>v0TWfHoHj7^LMzggUX<^A
z$dtJzE^xOhb1~e(3F-8=k_gPlNT33iwROt)X(Z@kD%-cOjmNv*N{rhoP2A!t5?A+`
z_2+jV#XZBeJD!3nX}V|~Y#VM+q9A%y*AVYi9N+ySsaO4?z1J<HSdwI~J2<A8$E;_|
zoCsG_OqqJgCTdPGDz3km2yEV6<&^;W-gjy$S|V1Z--ziOD*a{7(Dlql$-Z!mEOUwL
zCrrvgyFo+sv3?*-Gn5Q#?^r84u}I)>H^8^UWn|>#upa#XY>T;N-&$+SYzn0EP7_-w
zKI~-1ClhB=24;Vau<I5iJ3&u0%OKTT%&<3$c`0rI#jqRs(8>S6jW`1jO|>|FLKr#5
z*Hz&-GrLB~KthFIbml$i20OPAFzMsxev<d>L9SbiiDr`3?OUYyLc%!MWkvf1=;!Py
zDq!<gCZmM?4xyHMsgVpM#j5-(!PA+<6)Hu2WD6cYEF}knzWGXg#*-=JqfvtAosXE(
z9*Aoa8KnAnil?E{rKKsjBq1f8x&KEZ6`a~xB5wRrctKbbOhke@X$r<tC=;srdh!eW
zBHrisJopuZYb6SI-Zh<xbv>XtYpirVvX_b#FLbfBf#$GvjQE+ZGa)=mlC&T{y{=@3
zHM0o<e~rY%;7W{#53Y}K;NYqh2>2k#mniTr>beEETJQ*gs}L0(xYC{eug8;I;GVXJ
zO4>gb&1jvgeoPOb-zOYRP3#CVx0uMbyHvdwKcbVYT|O<hH|5+)Fnae9joQyl3jd>@
zn0z11t%98Xnh=kUCGYEAM2|FU{Exouk2YsI5SMf+=lglCNWAtdX(Bzpi)(UO8IXd-
zMDa4TE5T8r<K@o1$M!an-BEAf7gDw%p1sV+=y@hgGrO`R1+chVOC4P)6Vjez$rg`O
zax=IR@a*N#bF{Yiwfx;5d~NyeJ<s@+UbhKwo3Ia8wi8b2Y4i?<&;}?CbDQqp&Ct$h
zQgVOvPD@mebyw`H{1T7XnQQkV&rIdDZ*|Y<iEOvJPpk6hdY@fdH5vE&At~ZBXLvtF
zoO4?!zqYF4L25-#ys3w0(xE>7+Q@8Iy#CRJrPyFYQ#;Wl?sn5ExO=JG(m`2*G0S3R
z&6T00Y*z23j}Z;9r5Jwyt6kjP6KsHY+4GeuRJIQP=SRoNXIq_&uNIs(qPqD-JRdSI
zT$>-6HQS;=XbE~nem#?8ieMNK-3(!T-F%={2I8sQGg_VHbuBy8x?gv=yx&SVAW~&#
zuI-TwkBUx3iOfZXV5V<~oLW+)32?`@Xy3&>1X`p%a=4N;+v5mr2?_{g%7jcCcz37O
zD@fb?l$!VWQjj@g7N^nLs$CtWDeCv~$Sv(bq$_4)>wt(Q)6UGzyf}+s)`9+}tCaWH
zJPW({(ku7DG|a2X!)w~2l%4+WE2;ysLlYr1e$V)sGfY>~ZQaAc*gCDc-0WROy@u23
zRa!;1OJg=ihWcqfZ)l4y(0H$KV>s3jW2WeD${oZ9`yQeMy+bVBX^5<XD$T-Bqqw(Y
zdeUri5m|NVT7x}Sr{>9t*!%shSDJgV4N>?7frhq1cf?-luUpeYGvw7RXLAC|3sc4&
zss~mXlW1q-?<^-BeRx^fH_qw4yC~su>1XE)?csEP)#uJPwL-@3)Vm>FuFE=r<U!gj
zuQX~MmR}ey&7G>Y9e2?bCwG)9*|xf!?zEyx=En|N%58K{?z-U3PBM-?7Bfk1?+~MB
zy<O&mcQnq9a=4|2+8%}RygEGVa+Yb~K>F!)`i5V7$J}4i*Cj2+iW~H*q<c2T2)tu<
zRoq`nEP2h0JreB3DAWj}mGchXPDJHm^`C#&`j`H{GF-=Hke_^+ba+7b=qycG{dPFT
ztnc|XjaDD)KM}GkUh@}~%C^4F?;R<WHf;KL>x-s!1{}fp0~`Y5{WBX8+8w@YuZFXE
z!)WRp?A``6PZQ_omB#3)Hx5T#IDCh-g3b)>Okf3SPFeJI8QdMv@xgj8x4@tKxsLz4
zm22L;t4v8Cz*e)hG><+o9<~}=Z3uQ*a{^yW>}2wcq(Lldr9E$bYluIVzO=Kk=%?$S
zw)yBoav@`xN6$6T4A0>pwi$Beboq!7@U>7a(PS0vWLWGO3iJaPi>ad<vA&Me^J(8k
zWB&T+(08Fyx^mxK@35wSCVfuD!{Sj|Y1UY#VD9(Qt>v1-ms?uh`lDrqcQLfj>3*_&
z%eGS-4xtKg0t$Y3HA8iDf=!D1Q_mCIE=pTy=0T@Tn)yUrz-qw3hokN(c2C{I#`X;k
zM>}3I-sQrZ3I`i<nO~D@47_)0d9R-<VZkjLeI@D9$2|5@y4Hc^1qiFuxHmT3>1lsh
zM#_D-jU9wghISs1mDZCidUOVND|9=d8_lo%+)Zq3K+R8azRB!jt5~ID!Z!NK(l+Hm
zvfb?IszaGb@${8ZMX%V%YIqy{{L$bp)0Ebejl&dbAMfXGbg^Yo;oRN<LQbSv3HIeI
zI{Pc-+7;Wkl+%rXw9ed1vvrS1<AZzp%5&Q}NBg_bp7DCKzS-6vKg(Hj_aP>(t8(`A
zNYL`8tV+$u8miD{>1Ed9+{fDtq}_gXEhA-o{_O2N{&Dd>9;+Ir)4jr}6pCfoDf#DU
zLM~~GR;y^o!){*5YM1^zE@A&B_tqNuD(U=;<S7t6LKNriew=$0L{S%cCO)tK)m%PH
zRezgq=G@S?D2KzOHG{pCuV3lueSNsxx=r?)MEpk5H^trGT8VFVq1iU?4~H7<MK_Ki
zH(O}rVrMCDuaJT7Svp*)nGF*`wm;d98WEs(Ug8{nutMxCGiQeOv6dpG;Lm&Q@SRh9
zcL7h`JoaUNN7aDe<AP1+<pqc2noCMa%en8Zlp~0gJw68vhLr`BDvuOAf7QY5;H0bt
z6z<$1ei;zqOXc+IzeHFTk;Vhx{tMrqMXI@QR5qZGykmPv#<z0jR}_yL{r+1WZJXJH
zgn@zOb&n0c1zLiJqWc}+x4ykk*b>~Ilr1;@T;_JX^_#5SvEL25xkvtMNU5_i4l876
zw)UPU54y5Oz{Uw_68~TJ&;R9iE8Gu-S01+EHEbn~@K&!0QG^!4LJI=3Y0)>}o*PlR
z7Xvmm&`ETJjRTPK0BSwL<t;3A@1QrU&}+U$LoRwZM)9sQ*?BT^IGGvz6dC$D-014T
zX4vR@gKoB&!hMBCd0!HFcan4R?xG~)8``f6<`d#|)4f&i2z-LAh(lfCf8lnui6IZN
zW7o8+X7sX6*QYHRPij*mLu<mfTYFig-n1{jKGbr&F#V^<Q!(TxZr#$_{mN`@5BEPs
z!pP89;g{mn%|5n9WcL>S>xSorxZS-4CfOAAgD8ye6PyW0D8rGJJmahb)T}Fz*7-b2
z=T~nP-8o3c&tf>-zzhB>zZxa%0*w}kkF~T)XyuCEF<`C7XRXGkV5^SC|9RuWeHp>!
zr2c?*=YRfw_d?uoORK4Fw#5Da?4rG|$c7~1Fv7u4-lG)fL=M>cp@l--h7nA{mvl`(
zgMW$WtZw$q?i;J@-h6L2;WH<*bS~5-6Sl9oe|66-Te~h-f_4(Q&48R`xEeHT%YZD8
zRrjltkV#cH%aQ(b2RZwg0o5Gv*KU5?WRzf*Zh%Se^N4$YW=V+z<-HWOC)csJHpiIZ
z;08bLoX|l=%5BJCuD_vvJgDOvqVIV~^hpuTy4g<*Kphj${Ffv5zf{EkuO8nC7?KEr
UzuwXs$J&~jik5PLqD9dE0i4$aYybcN

literal 57231
zcmeFZ2UL@5w=S9lP!N=+B1L);gP?Sz1nGi+f;6cjO+k>}NfZGA=_1mRs?v)f9R-BY
zi}a=vn)HMgN^-t{uH{<)+V}2r&mQOO|BUMx;9BOZZ{GPnb3XHV-r$=W$|p}SoB)A9
zC$Fn0Xn{b-fd3>(BP9a<Ds1w&2m(R2Z{E1A2>dS?42D1;L_|cy#Ka^dB*%^&JAV8)
zDJdx#8QF;wC&<akDJUpTo;-Q#)G10zN-8QUYHDg48X8(!+S8{`pE+~p?Afz)bady=
zouj9xXJBApWMpJwVmg2RJTo&h3kwS?D=Qlt8#_BY2L}fyCnpyd7dJOI4-e0U3m15K
zc`shP2!%rV`1meey2Q`VFCZXr`SRr}SFQ*O3JM7c2@4C0h=_=aii(MeiHnO%NJw10
zdR0<VQc6lnT3T90Mg|6h$;!%JyLL@ZPEKB4UO_=YQBhGzNl95*Sw%%fRaI3@P3^{w
z8|v!n8X6j!nwmFn-n@0|mX?;5wzl@|+qZRebaZuf_4M@g_4N%54DQ^yV`yk-WMpJ)
zY;0m;Vrpt?W@dKx?p<?p^LzL1Sy)(DT3T9JSy@|K+t}FLzkmP1g9o;@wsv-Q4<A0X
zx3_n2aCr3Sk)xxdlarIPv$Knfi>s@vo12@vySs;nho`5fmzS5fx3`avkFT$<pPwHb
z4o4smj~_pN^5luXzyH&xPXhu10s{ksf`XnsdlnoV91;=|8XEfi`SY-_u<-Ekh=_>D
z$jGRusOaeEn3$LsFJ8pP#=d;{GA=Ie)vH(W@$m@>35kh`Nl8h`$;l}xDXFQcX=!P%
zU%yUIPtVB6$jr>l%F4>l&d$lndGqGY+qZAuy?d9No12%Hm!F?sP*6}<SXfk4R9swK
zQc_Y{T3S|C_Wu3*^78VEii!^(K2%jzRaaNn)YR10*4EY4)z{ZIG&Fqt`0>-HPmPU@
zO-)VB&CQ=be{N}MK_ZcDZEfxC?HwH*ot>RsU0vPX-90@$y}iACeSQ7?{a?O(85kHC
z92^`P8X6uR9vK-K9UUDT8yg=VpO~1KoSdARn)>?n>$h*;zJLEdJw1&=p=M@gW@l&T
z=H}+-=NA?he*E~cxVX5qw6wguyt1;gy1Kfywzj^$jz*&~7|h1T#^&bc*4Ebc_V&)s
z4i<~u-QC^W+uPsYKR7tR;c$37UQ6=CLtyHYJyJ1n27%585q`iCGwYTh5G&}q!nNBT
zNlUffNlXKtbH<3`^WTS&)#WjE_LE_l0u1+hqWH*grRi3*#rAV02JR&lxwFqJ$spn?
z;hbZwuei)ch=~LX!}!pS6#8Vu$4Tr8!)DBwK5>DGUlPsIbg19nboHv5>r2427!GL9
zY2~~VaQ*RRA*Enyn{uGPJ84DcAlG_VvOhoG@d5~hko!*IO$gdRB4Xe-2?+%Fw<i!V
z2qb&$G4Ky*X|llYn>Sg3-xn@W0l!b5CIx<<IPt&vaDzt3FdliYaS{SDTdLgRa6joy
zUU|g@d^=a97K%%pma=ezlE_1`*zUr;c(Ixpu_^bWhko+S*Py%B0{hMI?JD=Yy|2*O
zZEa9ls`Z81UdJwx%GfCLVTg@nSH!$A1}+bZ?@`4u;(N)k*cRtu?Q1>3e%R@pr$~H5
zY`dr3gPLA=%4<zO7;z-aD9@}U3<S22<H5o`^6|tQv!?jUp`9SH^T>O80kS;1<om6f
z`;}9kV`RAXMv0~S9ldl7Xt*0IV!ub~%5uGDH4$(!2ewjdPZJNH$S!lmE}ox*U79H=
zgrrd)H1b^SG7!Kv3*j3bHFkBGs~K!JNY<_$hqO{B(_(hAv|*!It8<W6gc5p4qKi_h
z^3LRP3*KPb`ywerFJ1;ta|g%O!uGjBc*TbNYa)+C;bRb5ey(ZhUZK1^h`vV!XMKLo
zN!P(833j<A<5Q%}Zmf0^tqJL(;rTgr!ddet(zZnB^5?FBJl9{Z#qS^2HePggAQvYC
zZmEHGg&(*N2XsqxMa&cdk4t+X6prc7^XsU$hX(nEUGeG`0w<oI>rrM!)bzlvyugVQ
zM}m`4Ebd7$5vAnY5?zeW!+PMJVBFf+q$u)-H`Ck;rgD)k8(6o`N5`&oDp`A}wFOOV
zH%BUNL!DtL0929|v)_wEZm*MzNp=auPw7&(3T}XLnBGGAiJqMdMr{2-8m8ZwN;YM_
zUXl`Q!O@6+erkuPMWS)o<b+c8{Hkxv^tk;ap=Mbqdb)l(X(Qq^q?MM`Oy7bQfv;UB
z+qni(cgcX2;|2I}IxUX;BDJsAm7!?p`8j#uz%zscX&UBCLLERGy<xbzF7gqF+Mx}<
z69^3yW$IE5bjN^>lvSBVqhnj<^%}OxRt_}MqnRR8F$rhhT@9^ta5U--^V0!!;SWd{
z6PW~r*;C;`7~iQ=Cyve%V+t_6Uc*fsd2$)W$l_Itf31!=yCt4IuQXLqh2I~lUxu%p
zV60|>3bVt891B)YQ^|(Qf*R$-u!i_r{9~2_Nk&5^y^SE5ulK^8zq?oYo#u>|LD#z{
z;miBpvLcv1dIJXwVCit7_;{zbpt+9K99au*yzh*3xN~r2=V|{hQH3=F*w<^;g=`3>
z$DqJ~$wGged${wbB{d1V92JXwmJiiT)+P9htT9*7<+}%ug;L<Bk3lY6yo#K{c(r)$
z!BEACa#1o~=&cHPvguCQ^nT#T$uJm|ES=I#(2l9BE->S`@QZkujsDoT;b@s%?elZ0
ze&1+X46N_Q-BbM0I=1?4r(4)zlXBZ&yg(W_$%A(I1Osx_Eotf#3%FHdcOLd#93L<M
z#g8~{4G+(XN6q-U*Bp=Jzy)LPW|4sP*?iID4vrcPaM%^u-T73Prd9ZdkB%B19c?l#
zqz7~Kq^x@Tw3w^4QxD03`QeMMkm$mXy+q!(z8CSi$Kg?uHqO0YK$R{TC7$RsGAN#b
zmm&hA+7p@Pq{q6xd|)SzjQWCHJ+}ts?upotH{HDsTZ_RpI_xV@?F$TGMu&IO4-QXQ
z_^2?w3Z{7c1*s_q3hX(z+eK;SB0iI`<9vCc*DnQ#!YmxFB6&a1-PVX2(yH<U1wRkD
z{Isj5&}Y-Per+~p3cUi`Gm-ITa0b>W!jcP`CjH;M&i=mWD=C7YEuJGLcWd4Db{aZk
znG&JA31xaQ%yg8_7E}?&Nl&LA*3C^<ie{C@J7`}e<7iw*y%rNq`{IW2K=)v3BS7q}
zx8iS9Rz|<&_2s6NTFzicct*t8L-}4Bx%2bsMJJNqeLr6nRwR`&ZbXqwGL%f&US54#
z_tMbAz!@ORk;qFZD2r<1R`Z*T75${C9I))`LdDys6t8n;$g)v5ZaA;VYQ)vRpfDtx
z&f%p(sqVelxtIC|>UM($Ggams!RlAe77U;7_WpQ2E9bS2LP&;SS`@dp`TY-lE19!2
z4(Q@XSrh9)1_?PF{;e*zx%(YLx*6A>awTwGp)43yF^}madKi+^FLREzLAqZ+nKA#M
zv|p04oR*&x47Q26;z=r-_#6qKhR0TdkXVu{wt}O!E?ypV;xlnki~@{)oSbarMr(oB
zk{mk~LX<ux2vmlAB}E(=;GjVxD5=P8ALVr*6a+d=L1u;n-IAeqi%I79b!4pMc4NaH
zV+BbI@=$_pq9r^@S&0UCODYt`Op{)D+!o%mJ_WvTj=Dk5JxTS2W?+T|ckB09WUyW=
zIZ_sM=bQ>Nr00_?no9PVJqOFY-7cu|$@Z~UCdkdp?1+Oz(T54o`~;a;_1wd&TiIsh
z1gQ{bk3|Ddb~zGwvd2(l+|`y06vZSXt23$W$6;?!VtOh?HS$~6dimF};?@8QR?tI1
zZs55Hug^*pm%6S$w_{uqaCD1SHIN}Ps%!o7NP$tLO@+7!@Lfyu?bGK5S_)N4(CIo~
z@3e>7-j!Gt{q$}How(YUWIg7R71q(BCyI*wE@M!tV=jp{$hPXap2ENO@R{A60h*~C
zJc`ElWlRE2RadtyeW5uq<x5^s+nG)$uDg~7cX(HXig`8M<oaP`cB}b=5D|Y}bYD;C
zX=O`U4t2j0;IuZSKGkAL+~ryZ0mIL5hUv3T257QK>TQAVjPb|~ov1RcC5;Y$!-WY8
z>q;Fu&76G2(a;*=%jzkMOe~yZFUV<dL-Qor*E`c<eAWWI9eHmF*>FuUZDp`r%pPy|
zupsi}!xFbr?A@Y_J+Z840(r&Ra<zr0HP7jqc^H`)cT-R23@74oG>Z<{C3M(sLxRk#
z2&Jq5dXVlGEyjlk=+5UjUY!J@mroQ6`({^tnF*ZKW$9DwbS-@X^+n30YO`pqp0VKR
zjL($Sdv|C4ydUW!U)AwrBX{x);u6EYHgwTdW0m%Qq%Xk5j3Pu5x!mStBx;F;Z=%bZ
zE7%~mx@tbD1(L8wF|n~FHeT!5Joy$mJ#bMJwEYaVW9G>13;{0VT275}!w`{!e&p7q
zUHKdOvS`!Uu}Cdpp|o7+^u0LYG^g}yF);ez-0qebI^JW2<jv{%t?GvZny6$7samBF
zJczaDkyRS6e<m<vQ4&k0?m4NY&fQ*~L>SwonQMvY{`;jB&KQ~oMbvuN`lB@(W0ahr
z?KgEZx4K~^C5-ylGGxS;#0Ea|&HQqk4US#zN292qUGAbJo6BiNpti(@UX=2jx*7ag
z0Xynm3#tnge|Y$qqYteIWo3nPjsKE1KPS=eL1`df9-9Og9%Ud*WHtdX6YG#`xOeMB
z+ux@!W-i(`?!_dC5V)RXUr9%3ZfHGhye(fDNkN>Z{u#w~G~q42c0G;rJPITMMD;1@
zGfd?%wGXR`LX8-!BT4^ys1JI6ohrcxLh~4Ox`W&-WSp7FWb}r{*cHXDw=Mjfkl!D4
zw`_>|^N28(nA4OJ@+>+LWV5iFzi3VQbnm&53Ix)_eGmO}_1N%?=Ojym&r%^EIQDbK
zi;z}Q;>*IUpha2G2`W$kNS_QmOadu7MVtl#-y}uEQX!20n|>&Nroky$+s$)9_fz_h
zpy6tRFfm(R-N8?7a4TB9cb^l#sk+s9<>jHrHd4>!7F$hBOTHVT+dKY(#o3(2Cv~%=
zWqYA&-3oaog8#ndTh4_2=^3r=*aCfeVS}dhN(OylFM;(RGf%Vi4bBkhvb+nRYnD+E
zwL0d4i1j%;qu|d?hbR$D@KH=zWiU0MQ+G_=FZMiRG8hNkfap8Uwy!KFp~IteyXIs`
zG`zAg@%xwRbc5?QRo9`|%9#Gb?ZMTXBt^GpUKp8F%@nQ8l?&^a@+HKgVq-I7wMwOA
zW5^}<J-Tz_jPASV8Eo!^zZ(d^foGJBm*kfUQ0u-``##Tvrp2JrdM_1X;5FT<dU+dC
z)WsjV2V#5Yr8NO=;O#iK!^>twaN$JY$MoT#*3TUXU5<N0-dTibkn$`RWW%=h=%qvT
zR*ij<V4{T`?eBK=n3h>r>iZl<tJS$}x8f^q#~?|l3=N`mxnZw2jp*aIR^U<mAOY~L
z8sa9^8e$7N;OBe|apr&biSqwAt*wZAMq!e*7&F5kcI-(x7o=N%&{a)9h|`p}fT)!&
z0pQbOCS@UxBkm_;BD~qy63Irm4pLkfV!j!u!pOjR=667U;$Lhs=S2jLuo5T{7bPya
zEQF4=ilh&m@O<U-;uy~l<Upt~jEoI*iGmYAz`Lvf0y1KbKVv^Sg`4=%iw@$5JpvXF
zP+m$N!WvDg-dWF8M1Wi_M@bQ<`3Ixe5PUwaEpHXh$l^QLKqRyfaoK_uCif`$A3^rq
zrrUVqByk`|l?7dWMS$^-3IL3s-EtT$yU%_R0Tu*ug&U-1ViMyLliYCF__s==^v7?W
zB^iG^LswLh<gzTVOgte2+i@vhZE3WqMn_H1b`u93Z#`3Cdf9h!GAEX@!Z0bJMXiGQ
zEbh5ZV@dpY5_xWArd37k`=TUw#=AQ1*iQHRUB{0DOLS*<i(ZITXY5Lhh#^-wN2+B4
z{W59+G8NdGt#fJJcUP0Y*h=E<LZO9a7GHu%&BFE06S+pce0mB^QC4bUb)3e=b_z*L
zPqX1P!;^;jBqAphk!tb+1cb@&8g90}a$gpRO7IMXit+wu0AhCCzu3I(d<R_)hutXr
z+^%aF#z7kbs<Pr=F>Q)-61)d(50-!RK+vFkCNtLD-JE$`b(LGPJ-^E1>cdO#?kiko
zL)=Fj&F!#`oIPl$kE~6eW|M2(-KG)rwBDEoz}%J$Ct)7Q-TB-N{gG}ZL#5(br;=%U
ze6MER{n;xPit1l9&;{Ua)eDgMgv!^QiJu>p8^7r>c}3O0DKRJ^bImG)#RFfp>KZN<
z9lsZMf4bCSLOUu#9@MDH3^{iPSgR#YgIzkuduG*z_#xuSa1j^BOgOCa!F!L{Zv%A1
zFpYv|d^TbdKU1UCK<HhGsS1PT31IlEY(zc!@owd3Z3kzpIK+!E&GvQF^;I^zQzwV+
zby@{aR7O0GLzKnjT&g=stzUZQ`<}}L_Hklf-V4S4P%-KIT#^B}QZ`eAf@jY-I<pyk
zc^7_gIhIUl<t?OYly6&=uEM*KLRibXle~&TL{(LSGH%z2F7hx4V`HK|?D|M~My121
z%sKK5q7wJ-dWo2bI{T8nv7P;nQ}p5GiOIiV(}sEZKoF}YdmY-m(qzC~#tTap^_CoZ
z3gTyva1RecT1Mf5kp8^^5pPZT$=0}wPH{)KBsLJ{vIE3gks6etJnKQL|M4kkE7DuC
zMY2_<($~(seSQAZJqzE8ZaGrf<Fr6JeK_}9Rq4PRhgmDe8FYl)PJ6meX769YT!wyN
zJ)RhSSr%mloxcG@;=CC<5Q)%RV2Wr7;xsckkaFuwvj-0tvl*Pc4R?$MjYn=U|2VzZ
zdUC>^=qRWVz>@`mN@&hBupS1xA6<`INrgI{Ta>Lox%>G$zT}e}F?Mh2Wkq4%BbqvC
z8LR6!P98<PEIASIj0WEz`qE$`i=#|jEfINNGP_kGSsA8izbd@S7FQHe;lOS&y+k~+
zb8{#Wag<)C1v~~8fV@ZGyp*G0?%0<`^lcgrbu;Tv(C%66kZ(EJG!VtvC)J^Z2A<{y
z%7VNsG=X)1!y8x!bWejRB7yfq)RQU=m4Ql#z!WD*8?FM$0I@6^NLLQz_862$1il~(
zIt?-7A~vJ@@A=^;06dN6UKpEH#TLEUW!tjXo$kC70xUY{)RCsxck5OkqLT}mp}G5_
z2Rxr2<QZ)5g#XxV1`xC*Y54e*b?p)Wq4qTfPR$u=cML8DbK^ZKzc^<>b&si^Ew(8*
zb~bEMt;S2@HLtF2Gh!g<!8gTX%BZk-@Od+?R7Hv2G$ct{-g9o#sJZITdV`AE<2-OX
zlYqF*8Q=zp5BLP(<+t2IlS|SL{&^_Vm9=c~%>o-;ngshkRlgO~#$xTJ<ZUpYjs#a+
zoV^6!e&wZ;gM6V@T%ML(_p`q6QTCudoq|nbesJlHp10MNuFH7f`Wh~ay+yZ2RSos@
zN#%#JeDB!lRPB!C>P4}vmiC>2N^2_ctvI_{3VU_a0*!);705)(D`wwqEAAFu*kUWW
zB9u1JI=pGZ%fmo%LDMAfhKh{fM(ljsfHxh>Ttp`nQ&RN>mc~T2ZMq<~Ji#3!ve#kw
z4vB(1%g<AR#-#Rm6+FM;onbBPLH6uB7ppGtQ~i~TbrJX>Aw|f6@}7d0AYlFDkedk5
z1>y!((guDigexoPzx#)mIS_-mk%i>rV@`{#WS+0vpIo3qFpmHg_C{+8P{+o4XPbtL
z?phs&=P%U9r)+ArrBYlcPyb7qtm_v22DJOLrnWEX&l0fb7*eTCMj>^BV)_4eu;?r$
zl_Kb{YQI7H^+G%u^ENbwwlHK?x9>+S>xm<bKU$S6zU4}4aD{-5N(MFr5$Z>mBb`K(
zcPMSnwh54;A1+%*OP;}S{r&#z;5UC9Jb`*N=aIOPKZLPRUL`p}b+lj|rV){`>cC1z
zSP6iY?q{{#<xLq0wXZkZt*O;?H>B#ZcYMGsFyW=YBcyw>-O3558)u^vCV&LnY4YJy
zoe6*&&yrwa8pfK7!B|Xr6GZf?GzaZshr@kd-nRnah1`dcJyG|4)Aj<avMFIEGrK<e
zSN$jg&2v&LHKL|gPPinp8|~s+Ozl`gn}EA+=I!wQtc}%1OkAKzrj3=XVRCV3w|O6;
zE%+Ew8wsmu-*X-Tt&P<;y;_MnKzP7knS{mVo8WmbO8;k}y^nyw!kz`rK9P3Qwen9R
zH&?kV&|UoGkp05mA)0;XCJJ>zV&X3_t{i1gofuZ9Ae52n)8BEgH6-CpGRl}}CH5w9
zkBIZnTKnefS|C-p?v&^3^P~ek8D=qyvmD*wGuXb3egXnWUtpP+S7U};4q`_X#50ci
z=my4k;s8Vs-8P@}2C#ex1Yo&}2{77DKSz7&q~0f|I)~EHYJlr-?X_FW#xL7l8q90&
zRC;7K7ck+&;pf`B0o3y?37s_6&MZU%;QKt9hQMeSSrJCN3PU7jV89y+kXc@Rs%|F^
ztmW%U4#!$uWLK+iK<&Flb(VfVEZJ_g(<;<T3HFaz9{!)i@<C_889smRY9n1;C86ym
z1Gnav-l8)({Z<Zh2~<aw`sGT#t>P94{q`NICt{CXfh=~Cco4`+J!r+DgqiL0#sSi9
zu;oQ7o`9XNC#wk>ACs88w$&`-_B_vYp^}yOFL0YqYK9<Wy+W|#kBsL&1dVsz!33?`
zIUmB)su)au`>$x8_$8HSKJl-3K1|F12bTZK3BMu@U5)soFgN<J*rScBH)Ty^S<m@8
z8yt4V^%}s%{l<8m;YSS69wp1v;F_}dr1E<qN&TK1+NdovQ!(liN4F<-J{QVasY#E3
z|Bzp6Jwey-gY4WPMb<GzbiugZ^<e4ofONpz6PZT~pBL0*w8I2cfR`r#_tCmkl78aP
z%Do@bdg}k)%H8DueC5uS|Dtpw^FAHp^&~gPm&T5mnM`iVJF4}QhV=mTpcSj_<hk&&
z@l)zppIK*#jc&A_ZnUmyCmN*AzqB?jU1;KzHBoFe0I&|abh`lS;95*sTpbON4rmNp
z*Yl~o4I={@b=<lP)VR10qtNtSMo&*EV_VA6)iuj0vBuym5qF&%<FQb47IsD<qB~*F
zjaOrx(mx{5sOGk(C=Pkk-~yvs_jO0yaE+iquG#@n8`u#a>(@W|P|%>XyJpjf*P33q
z67(Xrw*!FirpRaBb5M@{$pe8t5&*bE;t=}5afqcSafm$sIK*jI&<Su64Wx*i*z7#1
zEEtkT1%k4Hs%1fg0QvOa{KJpWL3j47bKIHjl)bu5oM!_|`SET!GAeQ=^KRsSLG<9K
zv^_8aCH)L^M62^-xAWl6(UBv(f{tpBzQs^AT+Ajh14{Cr3A)Nv@F7e;2MXqqg+989
zC|iZxtBPAMZT3QGX(M*y+@g65&sX`;$o<ZlpQX`gIAj8UFXla0MbVGDGS9Q>?cnXX
z^7)4>|5BhIi~jQ*gwce=3VV>pswz6GYyE;mV`^4#|K7y0za$2%3&a$W?h%m+zq9~H
ztK1>WF1j#Q+}`bx=e*B9BDVmbfBo2B=mQGQ<Ufc2s6?LxGxyj%mg)c_r3E5X$e6$@
z8Hd&Ww&P=$G&-lV3@ORspWodbxnbWGFkqxrI@WeazpFZ|=w0>xY_;&5Kar%s((K$c
zRIzkcbsL$GPcs(P?Zn6DjqE3>I~k>(3)GhzsYd3VPFnq@$Tn32;aZRWn1nOBmFY(n
z=1T_O>s3|5&%{T_)bX4}wH0^Bi;MWU1y0tg^mZW1uim3eZO#P8(w;o_f?U=vlU~y#
zbAIYpMwV!9LD^42ep&y{F-_T_j^=~V9W4zjJ4*>iQg4myB+DL)xc0U-o7t3C$Mx(u
zi;-PUx%?>_hM^A_r~wKjL25`sMy32Mi<bq_di}+#3Zw>j6FKhH=YC)C48E5>!jwCD
zZa0>^0vP+f3w@W&OMBzK6xnf431iNcq;o&u@f<>USEv37>j7Jzo`SfM#a-)Uf{X_Z
zpXLUm{Bi3_g3frshdwX!yf301e@h7D`p6Ec0c%Sml1EqL)|;bnzSfZog~ZK5l!bMy
zJL_HW2&%HgU+ju|V{B2<>R8?u@HOZd3p3U%z$FbsMHZb(shNg7?}|-5Gc^`ISPfh(
zwkmB}8E<J&lO09&pPdRM%`*Vlm8zBf2K8U9kFw}wIs8au;Ssl@9ai_kWiwiQ#GZF>
zLz8FByL4o`ehwACFE|zWB0U~#Y)sEmefSL|>|a+l@|8)cQRz)}D5f6F`$gb?Hi4*K
z1se;|0sRm)iFi9pIqj5+lo-MaG%`~H(4>_-T_w=+@zTH5`u;;Q67O?KvBSl!pIW{M
z#1R99Jy5+*tD6BOd4$+wp-3Ny?n0ErB(W}#WJ#kYyVg&YXbML><eFpAIm^tkSnYdO
z?suB&7>%Z^O2l8&18)L1Z}5eXBuKMpMut}XJ&?W;|4YeV<DRpTJkn-`zL_(uY(MYD
z_?O)I@l)E%Dg<UMYKo*RdFf$)fH7+H>nG!&Clz12c_Nr8UvH3GU^s9p<5I=Ni;{hR
z$rS>ClMiXNzhw$;aKpvn)fo=!y*T^-py=mo(mbLXtZ>Y-NBn-?%3EiC6K*sQQ5L09
zmE&%7_wDMbzmk{iZjkxI)o3Tw%tf7?Rg$Wq2l4Mv<o}J5pwI!eyK0QgWEql$zGE?8
zwGx@k%hER>hvMVajovai1VTqV7m7>T9Qxtrd7l$pBUay-m@cv0^y>)rUGp7mXP(Mg
z{G7<8Csmahn-y(?j)+r})~hlH776Om*#~C5^R-R6akL5MmO2#rzHiQ&oZh1vc}3p?
z)<qk<pgoxYbdcI?&}TgBY<(<^g`0iREwefv2Bs@4H%~p=E4<#H@<4F9a}x8iOslIe
zCfnGK({Z#J$c8)PViKfmcn`cB4o>NS)O0kEzzdurhuf?9i?jnB0TYAkcX&nGiG(YQ
z8Y8xAypXYOEeSfnv)?hE2cG>)vrN;9D)8v--9h3)ex-fRme+-`Eqi$`&8pqB)S-Ju
z#`Iy6n5Kb&*tUmgD+7vGJEITJ@rC$4v+JO+5fOB&)jO=6V(6f>(Ge8BTvbk+lg%yI
zdziMGJpN6}3&)!d^mf&~E07kQGMGW<EMU<@L2d%Myr!QVtvvfma^e5wwf}_ob#nH<
zaAe%ODF{x0zTkj765zBDVypf!;q~`=|15Bf!r?K)>TXpV%@0ut_O+QLw^)Ct^~rrT
zWr0EgAkdk(gg#FDSvq|mK$-MGNdCf6%1+S|nuI_de!O)cd79tpLS<YTkb)f&_eUkW
zV9z!&(3rdcl&<zpADnB{cL$+;YIpux{?pHUL!yCf4M4V^m>vVX8f5hUGl~lm$ZeO$
z4)PZL?_Tc69Pq^HmnMDL&?_m?I;$=)xshoVpi^M;63YHvr8G;)G`fIal>N6RdgpV$
z!zOwsg-x!I+q%IGhMN+-X0nx_CZ&XRlwt_(yvUsgr9V`wrz!h$*GmQ#VT$q6V8J7z
zSg5=tN&Klak9L=!O-VZ-X%%+X?C^(*Mi2klU?-=AyyEV54X2ELhy=zS{GVp!PkBHr
zC?3i{8<90~-go6dF(^QOKpd~W!gYyM{~qM>b3mBE1A7N4@&|+Gd0_02b2Oicfv@1G
zLB9{cHQj3S;VqZBb7t53@`D2K6OXv~=JrH$YZIOC>d<u_bJ(5g8Z_^$8n=5E`+b&i
zJvY=!<i6$q1T;6bJ<Z0~bUMf!((hTZQma>|0B9z(82sWL<W(JS9}{-k^eomIx9+x5
zv_}Um40wzPZ_(VGqGja(dK_YcF$-0Py^ZgDl%p?FH4<x+0Z|!UREuOwBdUZj4+Gw>
zabH$nXHHBOd#uh0({g~cBrZ%nC`EG@1rs{wkI)FEgucC6bN381_ak7McHEpI)=0Wh
z|F({dsl(qU9GBSO-3qiu&Rdu|3EY1S;0~cX^^>_uTP0ID1_(y}pxy~1_E!?>aUBmU
zJ&=w+1%*dm`uOFG=nau*V9o*s$8VH85fDX!{{XfrA)D-UKyUm|baKRs^KD3jgMmJ_
z67gD}mD6ZAuam}HuV&QPz1r3QAX<Oxxcv&v`1<2FtNa#sM~HI0PjoA0($z*1|E2Yp
z4!i{fa`RklJ?Y@=TmXMDy`jOK2b6<SlR1^q<3<bpQYo)TOrlp}VB36*d+?swuX`^v
zKA~v@#@eM5)qE4$J%yW*CU=;NJdG%ZTUjBDvEb#Y%=Ah@{TTkRo}NOL)KCt=crER#
z?ePhKz<~u2O96lQh`|4g(pe=#(Dse`JeQQ(*^ZOe#<*o0%(sXdhh5Dru6;<2wV2NZ
zCoDjEvi0G1Q<KXryKhyN20u!nnV>6}bo)UvR2{Y0H&-wz>bj+lHcg|dwqH(7p7eb5
z)Q@hk;kiDVNC|cdX&vS#jx1cBCkGT+mk`Mc5?7{@I4#^lZj7^w;iDrrt<|HoBeVjV
z$yGL*d;}Mo%J4~%Vf_~0ZM8J7Ug<{9R&XPCh4%M6jp~pB`8?XCxa`d0uhnixYp&P6
z>dr08vS=3JT}Ks}dmHkqeh-yRzusrD1b48owgtb{h^plp#eHvT%1Fl+(#Dt?riVpD
zmFkd8s=S~rjgcDp!HT!vNklGbL!m*fZ=)Rz*(hTo&?Z=6r$w5Fz9C7)tY>8;B|L{5
z?~KaxYYuPzB5gChhB+FZaly<srXzi3=Yyj6N5kuS`emvhJ3qp;gT2fNf~co(gb`c9
zUhBuNJy*k|=xPIkctoK*R?z#mVgurlUCBm@E6;uH7Ha5m$64p^mT-tOrE*2b3vi^8
zp9)<~6hAw*c}tL197|~OR>L}|El|-c7YOg^z0s&bc+a)_EXLrasj~b1Po2yQI8A>r
zMMMiJV2CU>y-U<ueWRM+UoxyaEqw)Gfm>GRpx>_Z`|bq{w*cMWI02&}i4`_e>dww@
z?w+Rl(y%CFmLs9o>z#@4%N%k0vb_=V==cm4RkW?-?i0<u`4g;WF;3?EuxX%lskR8U
z!S&ZB!y?K{q7+PoGV~&k*PoHT(p~sYq|FkQV(*S0M=>w-=^;rzy1RrLwsO=Av07}7
zl{NNB7aOl~2T8V!4fa$tK95rM`#RskhP}~Cc?tXeyzk>2--l!fNAk)W58unYXJDLz
zqK>^wdi7^7d1lrCRcwv_A@3~{_eisvCnbvil95V*-`(|iX_S25L6TC+DidXNsd@FJ
zYPbu~DBNmNzS8#mHkwG8Fd2jWJbq3_TP;w89czT5q+YVnb^fsZinnw*-NrXHJAQU>
zdFxNOGgMykkO;d0H?B>Dvw3sG&270%7U0Pj^697QQcs@B7eX!+8*2s$oCOL*F6odc
zt6wuwenbAo(M&Y;hv@H&R}EC$<CG~2j)ZwCJSfAvpiSQbQ-o7!TcH25HS+2LY(j&k
z1QR?<j%P8zOV21x)$@;HVelK}`wL1X*woC)3Z6k<Tr5^ufZIabZYaiL6ObIV+7%v`
zh<>h%_9w*IqB5)lHRQHIlCt%l2eH@1+HV>x@DR7$l)2%|n95VpHyrFa`%T9aZsg|-
zm#$31gyVTyThc2$V3il0u#6Vd{<v?8+A-)j5_pA7<mN78*4jvc=A2cX_<C=cAV=(e
z17Sk!_EHM{{GRg3(%jClu}HrBu+y&<Ic4mAJ+j=%{9FX`T@u;UI^4+XT4^KDqK#@V
zMKLB%FQd9qTcvyMzH83iQ)-heW1EWKsuhE#3pvB3eqDUDmr~^C#ly(H4(mGAq5VBJ
z9S$%L)+11IlQnw#PmNL5d5wl_i_K$Dx~PH0dR1-?hZUy8VJD+HBd$zLJ!l?daj~SR
zaVlD~$*uM$N6aNv-fQ(Dnnel-9uvvk*WC5x83Jo%g_pn`scHqrT7yg+{kd0=c;4lY
zrIL1R+SaL^aJzfB9^+hSDl1*^qA5!PmTczI3q$1N)|qBki%sSH)TqhbbO9-mAqhWL
zDKc~c<r==*+kHq`{M~acv_E00*($@n{Q5q)5R$ZYNP`gtT$c8dPR#9k5yM}7;`BXZ
zr&}!zLl^O(Zi(nA-vts>d-8NAn%kaT>{w6^^6Wb1o81u5W&Y~BKN`r3t+2{JsdN$h
zID;h!RjLF_`tueTWJZg;0;aGvkLBC<OPpKKgD*p~bwFEL?*1Zyj()obcG3yspSbGy
z1B$*Aaf4sUVJXpKH)Yxd+l6%1d^X#^$i12cof?s|2Nt?tJ?i70c^S<rT(C!kg>G;@
zD2A}mHT2hI6Ze?3m_Cv;bcu7LaT?tljM84!DJ)2NwWa>lE-9InktMo%YD|T@D~U75
zkHcxeyOrxl{=p1!yw?<G{JvzRCnH;;C=HryVc=@rVw`A9zuS!{=fOS!B?|g{)bTJs
zrs!^lgGbS(W4&}9qeCx1CfwI~0?R3l1%(c9zHNowr=HbH;v2FUa|EzAL`L~09?gEi
z4qdo|58V;-+5lqO{RfF2jOC$BjrTNu^h`Wi@#0bU`&ER49g4mfb^Q5XZ=q~RKvXB9
z{~Alm2`J)6Z2iV(u>*9=Twra)_s6IplIV)=H`XS%!A#f)jUe~^#Ught;mG|WS@ae?
zwuAg4;vVc<js0F+y~|2)*BJCbkI0!7d-pMFS!g$&?ly=qd5Gy$IDA>$dfSlv#*?G#
z&0Cw?0Q>AI%pK$oSk0lV1FPfAf8L6rwdNIMcv<|LtrOsbIMX<xRxFEdgdoSPPg(#c
z?LWb2-yM)?_T0W%p>_G_F${>}>9j!PJXaSD4syE|M;pigIOmdui3(WihxnL5Ez&WC
zY1Sl!B%0-{%A|}f8%FNL*Y<GC9t(wM#Poh+(Y7|z&&!b$E>|Hzy~42N7ee&=@rCI;
zv*mYJbHtC{u=K2;KqkmX2v3qLaQ>E11docJKC9<mUO)#qFG4`F>$rjM$7gk-Vky5L
z&g_~bq+k2dF@@o8Ph<9tb56|Q0K>71cW$BOgz-hz_zud=2Al@@=3oM62HyEaV^Y4Y
zWa<CB+A%?aY>*5}d`dsx-Z=>x^tUnGIfbWRo^sLfv;W}Dd-ms(J?;qQ1yzkW;snJ_
zmc{2zc?&c{kd)+{4ynD}HE9r!O`c?0E6Rz=W=Fs5t4p<~$&VgG35I!c0wIIz&xIv#
zmvUek6F$@b{6q$megZLAu-`0Jw;l(ZJhvqia{Oq>GEzM6k^qxa@%Vqu<YWb|^?%e}
z^}FmpkRS2j{mZS3c$CV%J#*qkVg}@%u>Ec!4C}i(?oqM6d^-3`FMROU2WV<on82O>
z$vVgW>dCc!i*2>pZJ~=4hM<Qn*r>9h$}^D5fM4hZGj9FzHuBzV#qHC5%77?O!FF^N
zTX}g|%;^1##uTL|?=K)C<Uq%Ic3<p_(Vm~nbgx-F>9<zj-*s2Jxd8#fv85)5UAo$j
zizr<8$GtYizNOu7<N^#?r}aZ2!{5bC*UnuPxCeR{x1W(C-t~F)R+Z0$?fJPrDK{O`
zKIQta*L@a<t&L?E8BK+=YxiTqF<r7{0`)_V8s~<Fq=wQ#eiM!<9rpy90W;T>bCQr=
zZnt<>;`${7w^w1^l{>4CCig=Cn+#D{D{NNWfy~)tRlMHLM=j&yozo45$wSYV)tt5f
zg9|fW+uZNH5tgim;<dUcxMs=`_j%mp#1kbH{DR3&<&<O672mgaKR|YF?u5)zW>;!;
zI{I-Jo;-6bA2l^6&SJIcD>k+Nblvz2B!8Y{>gvb%E7gZKBzK7}Tv6WMkncsnTm`bd
z3BoUl<2inw_+6ZG_WkSck?J5d*xbGfaxMHbbABs6aoXoHU&PMPrMaR)YO@azkFWiy
zPDx^ZyZvHWm3>p<UCq!Snf;cSxrsYS+k5vSG|&R#UAF60aI3Pw6z9t6@dqCA(-FET
z_=-5JU;h&Ri{oT`OPvxzWQQ36H`mz9(#fb_51Fo9dPkb6iL(vI;BW5kSSI4;Ylox~
z+fz=+1#W{(_GRYi^q$*ipR-Xg@vNTS@5T5P?-bJ}4f#k!7$3fgbsL^#7yb#909*ty
zz3*p<d9YGgYc7%c@XTWhh<LkWorQ}4$u%CdHGI<11kad*|3)d<#fZBap&kYSMxuTh
zhOgNRzt^d3@0Dn39MfKYLUtEbh#^_CFPJnfm@5U`X#$AVB-H-Pk=oP5`K^YlC#(-%
zC_4}L*<5}C*_q@yEyHrK16ZXndi5)*i7rX2@XuXJE#VtT(-EwN_(85f{qFnw_2S)=
zCM^_UuPH~qicht{;rE_tpfvH!_y=<>lp~{W9s-7?shjYn1HE*X`qjEUgM|araOcMX
z#^D?bG4<oCN#}4h34pzic%KQu%^+Xjb`XwHKD#9rXrqAJh`?&@r+SvHZ@iw`oWm97
zJsjW%-b^s=Y8SlC6+W#7yw8>-iy56?JJ2uuEZRc7{2HGcj@c%hpBdpZlN^pQZDjG@
ze+u~8J|u^YS`m{(21+u`xqjZ#;_KtV-^a(GriKHEMl_6G_nzNfO-AjLPI?cq`=W5#
z?Uc10{8TMi)9eaihCD|wG3@nYQ}Wk}yD0o5{>4{1TrXg$iiu*bl847nFI}#UdlD&w
z-j?j*tbdd{z5nXl`7lyjBTx{*G<bBgvi*z~z&B*O#?rUh!?ZgMn-SmHWb^`_v(3>B
zLmq)OwF#e#ljOL7kXh|o#Z63|31eiMn>gzx5e`g&9mC8x1g_6<(&}E~^s`w!WLGfI
zIB_GY`)Ze!-$(rHb{KFVF)-M^lSx=iS9eB13EDQiv;`OysxWu<Zs(+kqTL84>%mX_
zEi|2CkFA;Jrgaa8X<s`WqaC{RY47};yPaLX3<oQMMfZ-cMu#?!M3>H8-`eRKoFa6O
z3D(D>$s*nBX}rU6rDBJrwi~dIQH9O>>P`YNXJPF>t-RZ6>X$ri#5A~F|JhOIo@MGF
zJs4b55W8}5PO^(^OuF7<fdZdr{1TXofH`x+s0E8-+=mXh@19L*voutc<?&KwaZ1u&
zV__EVeYWq9D{cSceqfGkV|kR6%F<l{3gidva<_PPmkae9Wl-<)KZ}9S4oxxW??tRD
zW%4h?<X}E_f3BZRUKQ_!)!5#^%LSekPG7Cr!g^OOWm5qI?aA)(#B;ImAjoVK?qUT0
z*3d)p$ScyR$~0_Je7k8av3v+l+>IZ#;nXz6HeSZY2sG1W?9MA_!s+0F5=0D9EK~T|
zN28jfqS^41X2ird;rKBPKlxGo_}Vfh{zpX8{-9~~Qh-tUZK|^99QV$egBL72BM+zd
z#)|FIXnm*OhhyOA5e@2|ge~+;Jz!+Te-%i=tMD5#kO8j-{AmRL9|TOmewM{8kl23L
zv1Y{R#smI<8P(GBy32N*GubYfmQoSD9U_1a5?O%2WiyDTc#IzuKX{GYu4zro=Z0Mm
zC67-auNirf@ZCbpa3L|jiCc06ef=E0LK$EyOHjul68e16K}eP!QdT+oE{l-8)7Az#
zyeF@Jipfm!V{xKtC>=_w{Fc{Ve}3WpR=38g)bNVa$yC4&$FF9Obe;R;#8rtz>$^-S
z=`(BEmqea5a+l#7<0mhksp2b|L2;{<FQ_tJw<~E8UsYxNMp^S|D~0?HGW7KQjE;99
zizVF~l}%OwBJBn3ZC*M+r?!+c;XOmHR7F$U1M||Zi$iCCb3=+BHLrL~lG~?1dQ?pg
zdB{Es&DfKAqsd>4zN^ceJ?zV>VT38m1nJ_#+bnj~*F8B>iWEvYXFn1yfX+F{_PDz2
z?+Nz#*?JHC;{20oH!dUrTPO<vN?b`Vf<m^Zo!~j+NBlTFVwk=xpm9U$VnQ`iBd+d4
zm{I@c11ec7VL-1Zm=O=?^}wBk#@{sLLzqvRkUBst)VKpu`j0h&66+5ta<75ndv7p9
zTAy11GT39n1k2oW@~=NT?Ec%QB6B5Mj8QlLqXW@G%!^F*YUOK_hYWoF-nP}h-aKF&
z$2flpC1~PsB$)TEn4v;k(E^-q=Qu0@5f504ASOg{A(TZUTXq6|3Cs=3jOK0wHmyIk
ztO-i}aI4%6`@!#>4_34BP5NOXi(GZvUeY-%Waln;Ugg>PxNRVSiTPT2q+*2vT9pi#
zy7n(EK!S-@Od;QTHdL2e3?LSn9dxnDLyHmy9!01_8w)Vk{0sR)?^5-uLZR#qg`e7Y
z1x4D;;zJ3YEE#ryInL5A<rag!C77;iEBr9jNA2wI-+#*2Iobia9C=3rEH_G?Ab@Yq
zQgAlwK_xc&ZOB_bPO#|d2s(HZq#CfrP6r-$KNgP*alyLS&xUd=0tBZ}?gct)FXLvA
zCn<1_vLfig7q|3oc@WVd*RFxKBDwr%Rs#0s8okEsAj!?@kjRMhF_K5jb>E^Z82Yd)
zsYMN#*k|`wJ3cu{RE|*OzFh1In9*`7J)^7MYve-r>7o>69<p1_zYuVQ20|ZZT03_E
z=Ffy>!LjNH>`>-$%@pRjF)H42U|2PWj8{hZ)YNqcFA|)AT)la^8A$i^!rRT3eyOnx
zxkh^9IbTj!)81p#ed#f&UbpSGCoZBMg35YL$Gb(}<EfX>++_6w92-M<>;+}*tI@0?
zw6ejH#smH}ZWi)GmEt6mfch!lkVApO?hUIk+6<lRcG;;nBrs-zBbLR~<xWWWWO(x&
zyw+%W!UEG{{|xYp_S>6)TFRp_5lV)?NcIZblbh;a*QAIo*O(J`j7VX1N`RD3%?}XJ
zb*uCqcX;mzE{e4U+UK=L#y6kKX+HBqfb-2Oo^n&&?C^Fo#a^@?I3uV0-S*T|k&9Pb
zq){scdj6o2bV3Z#cLL&a3Y$aCD_ukKj|x*6f(|~9po3qB=0U&~N<u(z3MBx+N#E0P
z6r6|r#J-yQd9RB$_B`HWE_dlF%NGY&5gj(b)IU0dD1o4n&nD?HQ9ftAN0-yOd~G|c
z6IFIz8<j+T@G$Kj8)$-{Yo7<sn9n$r?A-wg#AMRSs|`6A!@Sd4SY<&5z3M;&hu8z5
zBvVDWUIv=(BP-%A_}^2-{;v)0&}`hAuhQ_6vOw2HRR}l%P<2H(by3>OetEO3VEQ9f
z(@z-VYv(VRgleW^)J0;iooss~Sh(@r>J0>NgXjQ)a=<8C$a4`e#{x@7YuiQB_~D<H
z5TT=h2rIOEs?!;3Z}I@>PCe*rBltnOo$v^B#vfKs+|cX02J18lQL(=~A;l#f2|`*M
z*J6xk^#np=mP7VlB$(t$q<$#jKbAK_cr3n9-`hJfdlYYk@1FFu3?u|Yg3o;O02b+%
zwYN0#hf-wuP$CbfHcdsoP|oGc?+}c4ngy<VHdX99S^Jr7-mVfGRr%<m5`$^?NxLmK
zeiKMKAUlusAZz6Zey)0usr-A1)7V(R<3a~cDD*WNv|NR)EXYbFRf^awaD#wu*HZz}
zR%8UoEqXo}%Sl|Mx&rx^PlG^!T70HyiBq%us>=cAQ_zjH7~lesWFT7X<_X@BS9FIi
z?k}{Zu^W4gOjNRN8w6iUMjQ~&B((%g#6XUu;q$=;1Z&uPR>aag7Z90;KuhjLh$L*y
z@giO?^AsM*U1K81W%*s8k0+D~_(ob)cL|q(LOc*gj(a5ibIZO}dm^kyA08fu$jj^o
z7J?#n#2<3%`I4EdIp2ZK=u^c3%3}9Y28Ti6gm0c^>bT0A-=Ma3iGU$LMI9ep%wRjH
z=CiuF%ij4I9Fuco%oK!?rTkQCX!WVSUEmVIw$a#j0SlbEo56UF_(*|L|8(cYdW@Ge
z9XVyfKEd9Y-HlMC9n6aNlAS^MTiMO;J$Evo@PDRz$AH??n_$a)d;M6a(KuW^VJjtu
zo<6_HY8UxJh>|-7@Mem!Yt%v$V!x<k7s>!_r!Mg$j`&J=yr_)Yp!6AoS$LQXP0>h%
zNz#Jr86x5PrznmDVQESad3I?!A76`cnB5ybW+{#gT@}r@O2POuy!d8e^12on4{K(V
z1HZLtR7k7fpUSVIRL?HT2RBbZL7(qgh&p*@b;e^n7P^9(pWu@X^!a(j67LQ;O;i>c
zsoIx+ddHg(x+13^F!xvl>v?}IHuUbmqpYwEyROPUCH+!kEAMUOjK~5obS+p>g^*>h
z;MIQDIB(H*t(e&NA%Iug>bh|5fDJl%9;)LQKV_Zu8OY(bc|`!{mHx9>t`zr})!GS}
zl+#i9|8r<|lJX!8NEYbUH{(0Xd6aUahi|}uK>lb^(r(USHxrGoIUm}M_o^JNM$W#T
zpwi6o&7OD%3Xj^HI0kZnBmx_40DE&#EW`lDE<z9h`mQ221XkWXKP3W?aZeP0!LNX0
zs1Wph1p6hIFF|Z|5}aqpP3aoqiwd+KLV!9VWjsk6fNKC#Jz9of&$b4SyI$~s;eOLz
zG1C0J#^L2;b^8bo4HO@O*Qpxsu*WSY+NBX0M(J96>}NzC24`gK1|cGJiDegl*{D|j
z>4so9_4DvpP~X3)hw8rweO7VDgCT(z5pFTsuAU2nji2yd=`=4f_LnA~1K#V{;LggT
zccE7PVvU%Fg#DT?J8eq(14aw8Syq6hfji&s<#6&I-!|n1*quKN4Dj;|FV6_PrZ@N{
z7o~-^=SgrVvIvh@`fZi~@&7^d?Lu<`CK$SC-jI-w!=hSBhLr<WOAi-;NIk1_HzlxX
z$RHD=<WxJos_vA|ka5(Z$YkV5u0@!)H$2PQ`r7yaks>p2-NOpS$T9UhUud>4CmA$R
z#BZ(>@Gt;gr+;s$G@APNT}QTdbP)d(5gdqQPA~IcTaCnRQXzN=k@uIV0Rof==<)wi
zeH{R<L9h#eG>3o`c{2G3X%4&OPUhj!@vN`(xOq&L-7fts=0<t}BTVZP6GAY19(A`B
zHky9!BJ$dP=*dsaI$sMw8{fPoAyhHU4cYU-BkHHz!z*0zlNA4mOv69>!T-1{d<+3)
zBb}3{{VfbFMu7bS);)gTFMy$QO{$0+Z}es@R-{~AZ&aoGw`$8bYPJD^?3XB}tn2oQ
zPbKblo$ic=G8^*(KD~|T1Ru*_5$nu(!2H6uqMc-?m$9&J6+7iwn-UsuT;1>15}=#1
zmdzqt?7JbI=7hc`B)Z|NRLK{3O#zyp_pGM&gur6^7fye`kX+BL?%j=i{#@5w=)j^a
z{&)<se*7CFb<X6OPSqdT`!07@^G%_@_5`uBh{ucUsx(-^yd1NKbSQc<B`%vCdXNze
z?~~`%gpVInuZv&3$N1tKr$F6US;8UO0InlIo*V+wAhdl;?83+z*03dP2z%t%*@m>J
z(csBIP!)G-1*l5l>erH*AQ`=%?qkj3<PK%oKU71lC9$H`UDH{TEN97a?pAjzlg8}=
zb=nEM18?QXiQH@({wngjXJ$JSr0~Z~D=%D1yvt}n0UHN)$SIypf<AmIGG<xso@!(g
z!+hD(E6G7VbGRUGN5<H_(z=vS#K0i;SvK86r3_oW++LLAfvS$t?6$$%=0U8O={=UM
z)Q)?-{MGzZ#=&zFC~t>~h<>JX!k+i%6+?WFxlG1dT`A_Zic)@06+gN*?Q>)Xmlw!(
z<4y5#em}C(AvX5xS^3JL+$?cA>)UG<VP{7?yUcwfHFG1ZMY;5qJFyR>ghb<I*wK#!
zWAD$!@NWAfE#jJ$eys0NuVYOGLMOOHN2JOazb}+^ml;wwFgrQSc^n@ZT4n$?A^A#w
ze}ci2)&qgcuOVX!<39kl8(Az25PN)FMJ843vFB~IsJAC!DZA=lZO|D9+-snCYryan
z9k9V&G+(HZEx7!4ia6W^PBBR=BU`Y@v>jPp{U79E;&SE0_jJZC56nA}kTy^e0PqkC
zS)mIAQp4$NgK<#GiBn_9Nm1ir#e!uoz~FlblYcwvbfvWTFRxN*Ih=7mfu^AW$Vz|5
zKsRF9-^#ClNnHsp;uclV0<hf|fS~_&#*x2O5&(!P3c<HzgzUC5=&Y;wLN`V3Zvn{E
z7RcO$pkR0l1lIuu8Ic{|(i6)cb?O7(>T6R#qkHPyXw3p|d5mY4wXpmd*=pq5sv4KY
zyWil_x5(N(E6zG7>FQ$Jnr4mk2NX*`Cf=RQLt+ZoV^7{6p#XN>+4+_LY*8Qrm>H-C
z{}W(nBg9}_+M(NakcQ;P<i<gZU2&bEWK{0AX-dDpG*`zrg`x}z4{2E(abn6k6&%Z*
z$5wdzSXpVbvIyN_U>xpU!s2!1J!T!XXbv&D&*7ts;y!dzcDmsaPO`uTJemx8njM!@
z0f^KWidHnBcs*a(l_<FELptLJSSp(5)J6PY>@xU7{^IHfa@xek)7}n%g`e)<eP@tJ
z|DQ&p=li4e{+y@)d%Y(hinLgu+ydnTm~_L2fJv|DetLm*;=EFyf^?BOAOcnSH+18c
zG~WF#>J;{6?0Slw>-_F@(tqSNfql0~zY%0$iq6o5Z87nG;7mS~Jo;q_a&Y{<;RkRv
zM||gBxtgEC;K?PzRv!Pvy?iDqie*}0qKsLt*#=d9-y+ZM9@}%-{%2zJmpNvA@{esV
zT$cYtPq^1M!E&FVB-D{8f}N8KD9pRwZZ&*qV0h9P*s*2-c18DICs2&knwVx<U5}AU
zrrrCxmlub<zGfsg$yfQkHx{tUsSAugSJD$Gj&*_o8sG0S?ZvK6AqBYw&qbvt^Ux0m
z7qc%VUBm*`dusi|iL9pG!5;CBA^wgZ=Do)uo(@dPo1SfCO^*$%V@GdYtjQD5md|WE
zD)TbwX9<O#mXwg^qyqf?TMbFWUkiG4_5`i<$QkINQ56v8Gtm9omBVxOH@i;BAYcRH
zG%IDP74Nq+$Sb@alfb1;Kw43HVJ*qtxHkq`Tm&;vRUcO$IBrIIgZvBCTS2ih&wgz?
z9?Xg507qaOu1=pTtJ0XYAKU&cJbDU)a)0jdanNz{A+Xs8BD8fndHSxC!P32m>*c5P
z?>o-|R50dl@mIL@*_Z9ePAmQQgl+EPq7(Cr9)!%MS5xU8t|1OUHP}}|Umn%G)(DXA
z=WYM239lR^xFnz!903ECd9?{)gO8b%&XO+{mig=nCS~+t1;+VGF{>LtYr?aJb5lNH
zzQb_A2*v89Dj)Y>bs=+mi}|61dyiti2HI(EPWS>IunE5Z)`C5<WCxktSk{ul&ORue
zh<7NbrhYg9*cVu`#ah&y<x?;Iuiz^q-7kEl+Z(=8Qa1+rs{R$=6A7*$fbE|o?m}1*
z_5ooFN+tSG;5;tBVwL|}0eS2niAoL&waaqvPiilX#1;HeOa4<ly}}#I{sPdG{Qp!|
z{u9aKabHN{VasWW9NOm7l@s_1Us^UsYo6B6K~*N?kk)V$pz&}-&7{Q<c%Q~UO}{4#
zQqHNed~liG;7Wb#R!?)Z;DCs12rP{p6sT?wi_Hj78RGwkS^#RdVJfFD-v$HrxOeS`
zn-2}q0`@p}+jT1o%h<vjc0WCG^oNNkgXu`*%g&-iho>wBm3=Gvfaiem?%_r_|1ZR+
zXhyF{yIs~eWx|qRy;mg-5GOUSK#s~>SD=Spxf+$a0a)t3{@J}ZYrb3bH}8L9K&x8H
zN<uwvI}@_KF6&zw!ZzsFblQ?Ensxl#4L}FN`j=GS4k&s7^G)bZrsL-Spzf{1qH5oL
z@iij|iVBJnf`C#Y-6E|ZIiR8-NGRQ1L(Q0UiPVq+f=Wq9mq7_b4UKfSz|b+jHH5zJ
z`~B|U*=O&4opYT(T$dNK*0Y}V#C?D6_}tI;Q!Ai5)L)QM^G!9G-`-3a8>Y$o&H~=w
z<uP!7be(ZzrE>aQX*!F5Y>Qw8Sc9j~Vd)%MwUSiEcO@z3FG^CK;J^RJs<(*{Cw@ja
zyG?#^9@Hfut^%|`9eTV4EPqt@LjAx(ETTM@8y%vvDp7v*Xyg3dr+LJ@`S&-BLa!?t
zKO@tPj&=QPq6#^T{2jy#6Ql8LS?wdn&HMl8_d_=A_NSD1=wWc=Af_S3kEanA0R^ua
zLE)WS{vS1f87ppDlFrRE;eK4)^ue{vk^RZtRmpO9;q)bASTcTn?l`>ts?nKZ#6K5d
zZn+4?PCF0E)s=r=*Ujd>mzSat-*iA=+hKCeqOpn4UYb%=V1Ku7D5XbsD;DGZB6bPm
z-8V#JyB=uVj=@L!gl3&E6e{6(-zag$fo~LV96bnB=B(V*65C9qrr9!Ie9OHb0=%Vn
z<Rra(&*;D$Z39;%XppkSM~m4cRrcObi_3hoec;|xz%2OY&H06kAI1g)NgltZouy4+
z$P;(Rb8r|TF}qXM(OeDso-kB9huLe=3fsbZypdTUgMNK&^KbY$@s)fyzCKpGzG$?D
zDK{9Ab+jJ#cVB~oVZv`;@_{xcn@2$MOEXA{qJ#}3K&XZw<l4#;$(gtMn!k<G_;Tcb
z#wrzDED0PySAQ}uYP;<wGzkLVj2^DM^6e-u*bUe+_WIx}`5YZ?JXO}d*;vy&H&^{E
z$a3QFHU9kLTJyNDo`PlP!|{?4iFc{paa1HZx!+f;m&*(LYCnK50MAwi2|4+fg(JU%
zbwWQBSH*a6fVKO;3Rn(h0>ao2e0XXsO?W*vCj6m;g1H-5Z_Ip=CC1zNmRQ_cwDTym
z`_{rd?IOdr+uHe8AoB*b-klN{&>d5Ss(%D0zy7$Tur~?)ecfGd4)z#5wgUbM^A~Q;
zC*z4|P$L@g?WTgy@{lGK*Zi>~-Pg|ldsO+t^2F`;1y4S<=B0r}$p0m-hx=y<j2zu7
z|8^h#P3?(yDiQoIblodfmNxJv9;7-xB%Tj?koew)gZq)^8scn5{b&*$O{Y<_!P5*^
zORmTg0ZoQlaiI)<1vFuBTKigro8Y7P{Lpb-bSkS=N`JaSI^zcA7qQELMErwhgM{`#
z^P3Jq!t8VjpNmnZ%gtdoY$Dzn>j-T6W`nM9^CsWa?n^dq_tXMjvO3r5oznfad>%=6
zrS~ov#Z}$Hio~~%84N3wDp@XjYYOd>b5oNo1^y*}fdlV17xOVzAbvbV??in8<mdpO
zP7#3D|9-*yfC6A4+Dmg<%pDy~(^Rt4*E;cy9zKXom{Zy~{l3|H<IX_{DP?z6Qq-y;
zKH1)cd{~URgj>6OzMUk?g|X{hvrf6y%v{Wvq%@P6qZI5Y7r?pypuS-YSQDFL6Y`Jr
z@>Yr}m896H-^GMu5DIl$c`WrG<UV{IAq1U*648$Oy$F@M8Z*8n!4Ir)Q6}(}M`4D$
z)O)*PT!Ukgonap|5o&D@f9aGS$ZPREmb&+RTJ@wCcds)g@NQjZK1j+B&OL}^-A}6<
zIA?{h(;vCpy3^%bpUv4?Q(M>b5o*Kq!?y=-ErlS2chH7wAHu_Ljoq&=$_(+YmxE_}
zTr76-Es&6HdPqisL#z#m>jvdEtsw54?Of&KHn|<wmnAw=!asU_v?<_5MQn$YbC)OD
zLsiAzzHuU!iG`M!><T1|M?vNaUnrTB%oWDSA^Tu*nR?O?CkCJEb+S*%zFs>1jrsmy
zsp_Cgf*NAy<HuR}$C>GwM^3A1{W=Nj^wTz;&1%zW5)b+p9ZE{l^=RoUJ#|&B(^fo;
z%L5Bcx7cT=US|op15}^JL<79ezg`sCl7CIVFlFGGPnrdI<-*zOMUCm>j9ylHJ6}il
z=XaH{aVAHW;l42%u%}$gLn)Rh!#9hZSVgA`UL`zG)miY!`wHo`9B6C~X=t2m`#b>~
z|DKb4I!{8m0mfP>A+dIQ1x+h;oNF^EiT>r7d$5Gxn78Ow6#e{jc4RtuenxG|B?pOq
zvwpf%L6sGbk2qLC=)(G`rYtR|G?&`@HR?hG%{pr;DiZLNVH4pN@r9Q*j|%&yxzdWi
z?y)aUZ0#*Ix{>Nw_FKQ(4RjH$I;FjQNmlY?{W`5DQ$48*mRGWuNCRNOv;~RSphr9B
z9rAye?Ttvxiee~I>sC<F70>Cmid|~Uv^CNr9i|tIU2tt1&RuP)rsbQIIz|`7uelK>
zZ6%lyX|I1bDGxmJuyFg;RP&Q2Dds^tY;uT)YO7qtj;$R1XtHM!K~-^RQ~0BIi|q#F
zKV>6l+k)UB&OdX#mfLljOJi7DN4a%f*48XM7tH#)NwVH8PdhG!{vhd_w(Q#FNRG_x
zD|fv2Ekgq%J-Y^2E%*CL8z;VAyuNi;z9MoeS;?y?Ie4j6;tkZ*@r3xo{G1p(wrQ!%
zY9_p1?Z<LZkE)D+rM$p`*9B|pd44^W<a+n9(#;F=%N&!1kJ4r<r*8-vbg!GQq&h+~
zEu&UbcJ+sE<u}=|A4h#rtiVjkUEZWi3g+1y4w9&m+a^1i=rNzPHBCinqDt;LeODuS
zZG0)s30|wM&_ExBA<@8G>58$g^(~;iMz_hTwPP_YDee$jxPq&YO0N);W%nAM5f>?#
z+Uhc)7kzG8@}zV0zNDP9kFl23m=xc`dY6m4SD8R2%+)abqVvcnO#l}3ZiTrBp5fek
zZ25DwIB<5FrTj^4*)MsW0m>591@6?<PHLS06m$`{0nB4L29?xEsfE{-14<E`8?E|k
zo&MR5FPlxu??r8%8qB0IJbrC6C>dO8&*kWimcL!{3<zF(LN4XZ+Da(rNrdRSFy`Pd
zgfdeSeYH~bD7;~we^0eYw<Uk$n)sMz{9E_TzDHB^Kj$g4d$+<Z^eTdz!0|f1ZQ1&?
zM&le&{8Pn>QT%XKwaTFm=e>@=)=@By#gmMCeMPU+AMp+&yCN}3?i}Czun$MV>8z4%
znYguO(^51!*3_EwU}a(aROl)fuxo_+bt-=YqSMYZJMWm_<jy|7)>H7sIOO%B7&z$?
zw%AY>$)ByY8>aMJkjD6%k?QYw<;-&F!nR>%q?w-|%92}j`2|}wE>xCFER$K;c2+hD
ztz4cy{{xm^o2RWQvl^^bIdo(79X(vgra#_jkzhSz-DiDepw*FtC|p{xV&xy7QAA)H
z(@ZZyg2A$<cQdoH;mis?J)<k|>vt^1cm@eQJYI3%jdK3R^nAsZ+g6T~`*ZhmT`l1g
z=#~5*n;zE_R;9I!;%l3^qHXmAEv{3&9}^A2%ND^c(|y?jPIDxnl3Y7s6+7W1U?B6b
zIP@adG;5`h3xB5T{VpZp0N6X7{yb9?_3QzG0<6<|H$k9A8Lp?<ba?^pxlphq>m56>
z;$O3fp^Vp(E!=2yBcJ~`J)Y@k#@ct|;i>z}m*nu9{-USA{-&4c^;QHv%Z?BhepV?g
zd1>0?>u);kxjeHR^g-vAfFS>EHPX`3%J`%l@*1=0^7sdfse#M5wQ^q6b-UOa#F+I0
zBP#w%N!_Y>T-0hhyJ^qFi5u`Jt^ia1!{s|@fo)|voSkcsvfEywx3lL5i#(Pa``gCz
zU>BR3Bwd@@$7<O528@@N$yroNW!kocX>cu=gkJbe+bU({oRvhb)eO#Zm6p6IBMdAK
zCzfYYV5fhq8Oc2O1TR^#TjjzISJ)>!_af0o3cQl3F@RSeHvG{v-}jHE(ceFshIjuY
zn<yxrj;oAkH!!41f!Dv?{-*mRgTr3!z-Cru5fA;U*a*?<;FC<*@zTv=S<ven-i;%b
zVwYBxM~)x-`DwFQD)jmm(OeNsvSQ_pa1Lo|U-0qar(?;``gD_D{J|f9)b2-Tq6!C1
zEUij&7?eXi?lZtwPkBu&ov#YB(KqQ#zcANLnG~l~CbfL`=~xFpL3FJe&_|W07gm~(
zbE&n7S(jn+z0~py^|QuNHl(Bxt~JZt#vsAR@fz>3dO5<^8qH*ebV`sngWoAuwnpcQ
zwB9)QQp!tQ3n_DZb$R9C$syhvp(Qs|j!9DN?{}PvuDu6H8qYcC<!mw;NCvt|EI|(1
z;=#8yXFz8i{ExqBPxkX}C&jCJ9CmZRiC58;^|KDg+Z&Tl91O>kI<W}o2CsqM@ah)C
z)5=;v`rxDX?+)fIc-5su<u3H2_&W&`{RMel@oyn;4#GQL{M!QAl6;Aa&6i+@<8x~w
z=R9)n?%G<hTev&_wqgFCk78JHJvn4qvyiE=O(iPX`@5}MA#gFmUvD-7*3x=%xMf8n
zT@Po0m@haDz+MROY%VL0%27*-|2+;H(Y5yhn;jzO#>Txqb>6kN%I&}3b6j#I>ZBB6
z<k+g}<Zi2!?pw1rhadARx)$^jukA=8CPi^0UvIXr_|P8=_;K--Zv6RYBWLh9@;|fO
z{x?_W|J~aYEwZ*cTi>Sv0P)j0!jO8yQav`$VH3n;X61TxL)&mOs|11Kz}O&_dpD-Y
zWSl4nF=$e$cIwH!C5AiLye{somOjKbhIJvc80AxO3o5fmGBJ{uB(4=I$BDdXusG}X
zZgKZjxrWzzmV`1J3)s{MeL1ua@nwL9ena*uIgKY&7O_abJed?M_Ql<&LhYH1S6BjY
zn{)Ze`6@}&X63qCrP+@AT>oy=eGXy{WZJ;+S*cC_8m|p>FrsQ>U8nM-nKh$I$P{8r
zopESzt39Etx^eSyx%ZSPi~N}0`q`>FS&0|WqEV`vy$QBt*Rf`5cbAX!d49(rgm<<l
zN%?|-0Pp%o*-@BPZ^_1PISaG~!84$B64Btgg{wGgGhH@Hy=~iJeq}4@gnfRt{@esX
ztrx^>><YaEq<zmPr!t}N#}9;o<t=)a3nb6PnBuRSjJRYBdecJ6wic;M3v|n4a@fDd
z3=)$wY}xRy1)0o@U%I`g5kr`?7|S5Hokrz;+dNQr>>DdGfp4uKa(?T>Xt${1+h#U8
z0=WF#1cFQyS`?L2>G&hqTd*{gMG9MEBR;w0=9kkPQ|MFS@QikT@rlg`#mrWJF)c`}
z73l1`O*tF;t$XtOR}-r3CmR?)i1h2ORf|EJW#d9%&o2j&+wfxPIpxjRQJ-^6Td}-*
zY!$Yg29+iQSGH76IOdyBZ-@M(DX;TZ1uUpUbvR<1x;uZRT{!maL}JroDZw2fy5p)V
zFN1lV4X)<Vm_3uh5g61qt-Rc(c25hMDrIPC-BYUBQ!x*GJl1-0=*yy4?e(b2?S6}r
zhbGJhto|?gscA?MTD%PM%d~}nS9Nz^HK{RGA4SdB?HGnudOCW^yBt5Ox@#R}1__@A
zts#u(0$%Rw?S9?(AiUH}12+22<;-J#gM_mrxVRkL#pQsJo4=YNMTDOsb0p77Zrm_X
za*NkEMNC&-JxTp2nQ<$Nm~2y+>d{)xmfo4gpLcO976gpo#O)vg%EK#Z-F{^TcwWbf
zR6n;Ruzk555Rmh$P`sJ_IF<z&LoHgf!vG{LJ_AA%hW4nWc0IirsWyL|NZd)F-nRU5
zZE(ZEIDzz-4tWs3#YaaoLe#~(<{LkG;h5XGs?Ro=I0$m8^s>|#4()2c`8`w4KUjCg
zE4k#mCHM`;t|ImHnC!~eC)X=|b0o`_roVD?dlpVlth2|=nx18V(-Z<?2qMJbE+wC-
z;;*UIND3P79o)w#=Kb#AxD3)d^fv2*`PUZFE;|AH4Z25v*Hb)J;d}~L-`8w7$M`Yz
zhCYEpMbIOFlJ7RyxzPY+If9@H+qI>d;8c@LLU4Q`S1NIPM6S_1ULD%m!G>est^&JA
zPn`@&_!KVad%}}YGVs$(<`6#g+S3w}q^NTOTQLE8iam=vb-FjPI%GTG!7piES$5!v
zN<kc}=Fk?ng+<lk%#VxumV7%3)z563AV0w!RZY5g*5=2&&*N75%}1jOGgf-&6(jKd
z;uT9X>NP<aIlkrL^T;8Ynx+$0TuWDIAv<lm)2J}&zMm^)`cj`7q;V_;wT)#n0%V?)
z*bOm<7F!!9CPsRDh{uZ@`WHfxu4jV)<KQQAgBxG<_$kJo8|5&{V_oQ*yNM7p3k+La
zQ%mp(z+kZrd=KQjR7~E=IQ{OF9EtNT)S!0KiR^{#0AUIuHfMIuel0UVyepwThys#?
z=-AlBtq|Ifwz{aShuUn<nZ(tMuNBv{`g;I`Eem6f^gPblI`K{2FZ?uvt_;T#2<e^J
zO>o{6ui49&bhyL|m8lHOsSM?Y`A&(ZZP!(P>84+C`?6t$DAf$m9XrN=TvyC=4t~ka
zyEZthSN$um@RP83s&Ra>9CFM`40_$3)?oc%#~cw=Uw<0dpME*LJuBx%wYF<M6(l9W
zJ3Yeu1(+;ugSicjMt$rRdH_P(HGW~2p5OWFSbB-XI%4okx7lvA`YGa&vOqatVs&F+
zJJBv?Z!lmtP4T(|ja^Bw7rg;O-NC@=tY<C3^<GtjxhU9)WLKxaC*BUT=um$#bI8=9
z{#@9s%9MF{L*TC696h3KhxQjpi=%iRbRxv#QZ}H7FcP~X;j;X?kbkW}?q<wSQ;ZWf
z#uK3_i;ctVw(9a^6WYyrORUoI|KjdWyClc8Tu1n+EugR=y=~Fs(e4x5#^~`q<!clK
z3Iku+f!(Red6Rd8I5W%xq<2;_{W{_^wADo!nOtuBBWBJu7<UoTS6?SfR}9#jRN%S`
zUAD~z5Zi5oW!u7gmZNVg8n1GJ%0=cJvf|gbrFHuC0acyX)&<azsT*6Wg7D&ttuBg*
z$-DL0UOS$(6_eS`;v+MG5641DK%FinAO&9kh8X|P>13lj1b2S;@aYuz7dyIja#6Nj
zNYk`5RqscC#BlVCmzk>nL>Jc>qx(681scq{=FC;{+FCf?>Ey*s_C<`+2wcC;HdD=!
zZ9k2L+DvG}besH)J3AV;1xT`tQ*)y2P@B&%jY2{g6^sIJT$7*x<xiN`L#om+o0g%W
zl<QZC-nV853HUEeJi|D$e2vLlNbPP^R-mVfO+nE{XFb1z4mA^q^ZhkZR;sydW~>b}
zarWpcO;G7;A%c{-)}K&Pxv{9An(f%pI2KX<nZ<&tkwxail2y9-#lay$9LwO4U+U8N
ziQSl;hujty2QLW*(swe$M(dk5g*Wf2wQG|vSuI<|c4(8IrCFH!;zf29&MI3TGV(wR
zv+?}KV&8ev#W$%-e35o2-x}jKGvjJTmbc9C=p`#Zxqws=@ZlTYFDDvR#&kN*lRW0`
zp*6elUBT|onLDasB#DW|#*^mivTC#xC<|wO{;=J>%*loNt{K_w#yDl7@j3xi(`XI+
zt366#(zPYlcSEQ;TU2&(I{)-A$&ksY3_rPmb8>oGoJ#&PQ;V!=Ei$P^0x&ghigfT9
zV7bMG<TNBn#HYN<OhN0ETe4Dgw}-pC4OARxtXi=eovq9;Fl6b*@)!&!BD!xjzu`a)
zaubjp&vW<r{aCl0iKZNeuJ^K=w^=cD<Lsz`jliSR(5Yx77-iXgBKX;Jng%)rW=HBO
z-UQ~sv>~ROlvSIKX-@7@ojwC}CEX?cirX=n$01ndDlqY(;tutr9_Ld8EutDQRn_1g
z=oG$c@fAl%Y~h66?3#!cvs>R3(y26ill?52R5Jm{vzj8|#!M6#Fd%KC3J6=I@n17I
zaNLbCj(w)Xv86;@Dh74h8@QR=fN%a6!+ep^?%0RYcpKj1P+}r3<QYD&Pg@{SDzar%
z|H*6JTx<Fyr+DcaHOBz98PB)LW9Z5Gjo0nl&0bfuc4U*5`cml6Cl74Nvu(fXVAKX+
z{K0PNjLiL1QT}_xpx5O}?<_|U=I-no8QS&LX+I#ME0iL#P>Y#h>k9Hr7^xCaew#Tv
zo5k${wCHt&Bu^S;R;p~X8-gbhddIP4{Q=nC;>vAV8ShdKDCB45E0#rj&g}D*-toiK
z8VV;~qpp5VnKC=U3=^%}Y7S#Nn~`@e`y5J@f7&{A=eD1fl8>nByRD_oI+d_4{wA^S
zVrmMk7tuXrt&UdbY`9!xColCBl?>vhbq6i|rgqg;Hv~m5P4}Mo5G2XcbBtT;9^&-a
z%`W#`jvq~8+H$vMkLB+^=~{uzIJ5{;v+Xt4n8wgzM*yXF*Cg()o}!e(G{$YaT$+~-
zO4<u>MAfUqHd}^DGnQPJXy?(AP{TGNh@s0l9ozQ)9x2s_g+$ufDoQeLXR6M$aiAj~
zRvSF&Zh{afR-`!q4LYraB1A*FgC6g~xk+&%K-1{*pf1<(X<Fp<DenzhAWe26+3^f=
zF&83Agm=|AclY(GtZNn)wk;T2NCV~M9B|^qY}$9#uth3xQZ)fi{G@^lZH~>+?7__Z
z+kJ;)P%4;#QhtsHyq6yDbn!xXo8H1mlSy6ghE6dZEVG{zEBfC+M+lvI4>S48nhP^h
zfH?X76C4?d-kJKsIs=Ym1Y~ClfE)F+&=6uBAg$G4MX082emiqzDaKR-#%=RXDJJXp
zdbS0<Bkisi%br+%+$`k*7wHf;#0jj1&v3Q?F!$de+%(P1xh`E8)vuw^l8YvHTt@1i
zc4Q(T699Y3-JW$Bg`!UW;2jhE6!`j-B=I?z`Re<L@Ylsn@zc}w4eH$SJ(v+GB;dwu
z3hvPa*I%(8>H{_M-5S`yJ~k-~6@cmG=bkSIJ|>g8F9n1`AXkZ?Az*=CVq_#4Qi}l?
zmjb}N^)rxP=b(2;2oh<DrN9%sE(5wZ0rRK8n`01puyQa#l@hTOCmGV70XRV{<qf!=
zfV?M!-Z?|S0M@J{1Gp~(pQV5wPXQ*dmKt=3nIMsZSW1Kp2^KfI1(dx2+M$1YDYRts
zCKWGFcUPGs|2<l494;YnP~u!G<yS|@p9Tr<lRR^LtpVq>(9kqkON!1*4_NYe99dj|
zwww)xdrhKdB^(ItT0~)kp;^W%0wjUx08@%{)ORT+dWqUK{@PD&2;0AYap=j%71!#R
zWj<V(`+D6v&slTHrl~jFj%RW%mjfl6rrGM}#R$iM>s0Bsni$3MTC<W?eO3VD&uAZ;
zyhdSp2Xv5dI-7*c^UXELDyhju6+f_AaU;gHvCpWpFnXEEL{<ZxvME1$wps~ioZ4ID
zbz`2xKh#X3jjWRY>|m<HLsK@pge!}hye`kLhPfQfiO+qY=9Uw~P;G^a9X$QP&BShr
z#J$jyCNyq=in@9*Z2j8iRmlFJpZx$(m_PHl`KC8Rhbs7GtWMLWu)nhr*iUE8Rhx*p
zk5dBdT0L2F3~;dTC-n`hvjV;iM9Vj%E<LmbRpaOL38m60+zO5UX4ly|m)Uq#VKI3P
z+HKhf9dAgD3)OeIflk3R8kz^rrtvrB7BxFYXw)z&?yP5}@{3M>vJE^Br>y>I@mU_X
zW`EH8DiPq)B;e}n@ZVTAWJ26rBBj-;m;bsbiKBnGJp+zQ0dQ%RhGo9+u?JsWRdN(k
zUEi6VXjR@4+`Y}>q2VGJ=EWFitfh>poX#*5sD3Uhz+d*m=k$-sN0t4whwJ&W1G6;n
zoB^5tJYwVl-(|O&vj$<dg5+K>6T{rJP}Jh3&7d*#oA$BN47(Iw0Z!9haP{x6Zlo>T
z(2|gVhZt%LRysb~EyUq-A0N2f1jqn_7RdhQ+SMl#$2t%LP$X~zfcl&yPE^H*5G0^c
z3FL-qWgkstOZkFg%1k3-jlXa;?4fgYq9d7e8eSvv^flmVngXbiL>J-NF@df*5T0Cl
zP$7-SFowEs9PU<9QvTkwQa)vTaK#=~pn6YtitFk~;v%<E<@@u{-#e8-%v_HSp+)}~
zQm48>h9o2)JL#fYF`+r2BqWE*gg{>0S_c)g=dfFe!uka%W>$G-cz6e&b2zWihqq=-
z3D*4}%HrpY5i=V&Pvki|!-f_V)!=7G71xBCwKg)hB{dGSU$z2E{ynPO7Qk{6C#nn*
zwt)N@DM?rB*4qwgqzej<U6!w(I0bo8KS%dE{#8K>KeK--=}lsWQ_sM&kFd|T>xO%b
zpC5=G5TvliM84h;$VVx{ZaXj_2TnwY5wwufX<6j>v2<p6e&}J?7bY?r&?0m2ym$~0
z-Uk7KHTYteXJEz>_AGcP+%15Y99f^gZZ>7q`_uX~F#~N0$R?X&#0xelfV3^>QKxEr
z$?WC?$cY=??=1aOPy_<d?t_?%SB%TUcg;387S4Okxl?ZMMoe`^n$6oRgAm{<_F{Nh
z1!^j+<QiVOA%U!vani==m+5Aq>!(2gavvPa=fJ+;yy*+*x~6yuXv>R=eY?gceUJQI
z#CdVfw<%^snm9NgGaX5jVws|;1UjWOdYOi;fb;T#Sh)H6S}a&q#UvK)15bUeB7n-f
zoHX6bG@!lU*Cd*2=TwQqlvnheZ{5y}|7GEPqwO>KJ!(%L<i2RWdKQ?pLzxvEZ`z;K
zQCca(13CQW<j_@4>$R&dfUj@2LC{loaQ}O<s=2+sS-HB)y)BkUqvUgHe|%q@g#6ZT
z|NKoRfzWo_Yo(u67~>}wcx&WU&IiXlK~R%TfwU+FzJr<cq^Kh3*WY<m`pVSYI3ShI
zy5u;hf%-vahYbEdO@b{-8fazfo<|lE=b<5Yt^hiSLtW*%ed+Md_eNhXCrmtmvG;{9
z<?DWG5Xscns5vWS5%~aT+yP3lEYrvfxC!m@A%z&H9#g;P6^6ka7J{XG@w0I~#i$Fd
zv6D0@#m1~c{37f>_U4Y+opzjeOu8WR$N}bk6F9eVo5wSSphX<id%pr7Umh(vz9rO~
zG@YH1tb{uA>W}t9YH5ZjEPYJ5F!y7F`G!TLtW~Gk#GHL+rDlG5?9wk&s%f)cMtGaC
z4iC6vCmF$-23ffvHqV-{{8kq*fr{&Hlu>yz4ypzx@9m3IASi|#mf=IF80=+$`?s!b
zf?$KdpWUbQXPw#SZG=E9@kcF#Ac3`_pcOLfthdsI{Hr?KN36e5?aM#CrT|xc5y0|c
zZ$5^!^~NGHmLc89HCl1`+FP?SWxv!^W-VbgaY6Dd<J`B=PQn-%YUE2B4__q3dRW6}
z75YCq9|ReXV}rB`y~mX3*ETa@XD>TzTA;iL+W&$J@b|piAQ?Kf2rxI-ddhiy^#&L~
zvJY^IPs;x&7vRdV{l1>YzHo9!Hm#nCBEO%1jyumxms*a6o1a>jIiHCeQ-S8l7D$UW
z?-7VEy*DDE0}4J7UYG{I4W$RACV9PRL(|;JvTHu4Q85-v#io~_pO63u$Uta<?}BK*
zN&kRiY0T+M6Jb>gI<La2Zh#Ak48IG8c{nAAep$^ri}tr#Uj2<LkUa1)3sw~`GnaI^
zE8x9gR+d?~$oAZ8-<x@F{~(r|W9~Bksu$DVdB>1~uaee=*d2wcFyL)6DbT3wllCX>
zniH1t1odU;sPDw&feN^9;(rr9v~c^koo*6{JrAB^7?rjZRoIYW!#}Rj<d=m~SP89R
z^o!g>D}T1zb?nuoV8nQ&T=p`#QURO%)B+Bb>_DFW!ts+(FE`5_y^n6gP_4clCR*^D
zZUWg^iAFK3Q)8`V7pJC9B(_bi!<<e^u9f@mR`BTOcuMUuUT6|Jl56uT)6oYB0(pa|
zDI&0ermi7o?nmXW#;X3?&jF?@7piW;LR5Y3#e60i%?ZhC5WyL@W(CyV9N<ddL%-I!
z5N9k4ZrDpqHh#4?r2{g$!|l37$<2BMKzS$QninzRAnb!p87qr=6{gRbaBihNU!=5=
zc^Ac|YOn72*|L|K%{dLDd;_hV+6e-sQ4arYWfQoOSkiaIqWO(HhchHy@}Vww8_U(z
zmI}Kx?<3|-82bku!7A2xtU3ue#vU7D1_fX(_d;G!-b=Ra@@cQXp;0oScZ)4<(_R+L
z(k%7qNfCwYE8D=Fy*hjMgQktVLMMmLtX!2m29!twi)x8iLY{t)VCa>E)Cr70B~z76
z3cqMhv~R{$rA*22cEq2Z_@bdHtah3fW<EmmC5hno-jynf6^CKp$-SEhIl*AaD+LIV
zAt{L&t}!4b?(KZ(-Uu7$)V>>{uxzg0(D$Ipb*j%uS|AS<VP!h;6a9>z&m<D=!p|-Y
zZ!2Rv<7Tp|36F&*<asQhEUuJAI}2AEv*admfHXlG;wlmXvI}HVdYfGHmRosGuAi9e
z`G(iI*c1CN^9tGC<esgrY@3D=N&gj^lU)Sk&t3?kskzic4SXn&hy!1mZG}IWOA`O*
z6LejzjpQ|1m7y$p7B|2m%YR@$^4ELo%UaGOSG3&pR(StrAH0ya!qtq2nJh#|5>5Q=
zp(9&iixF8mPAPGq6#7Qg7(gq`5u~e=QSUe8yGQ((1fK9JALFY|W@WV}d-@;0dl5f&
zCRI?q)o1FTTA3TfWma*9fD9@PW_W>u=T6fd;;NX<Pbc=#1y;Z@DkA2~z7Wm>3F-S7
z<LEuV2yHvCSZvg~ndxB)3cB+d;Kuq8gfxfhbj#t)<KAt7{QcAsROPuN40UKt-MNB1
z63D;d)t{df-0Dj_eLCT<9^BlD@YNj#@qNyqUeNvaZ7J`18r9afEf)8ex6G4NXU_8J
zT{yytUUqdoMw$YFCVgc(5j!!=ppJV+;;f0aiI^qCuTrC&476Y91-Q1SVO;D4%KoKI
z$iC(Q;bI5gD1l7w#VMyp=SL+*kfBdy2TZ!@reajo0+jtjQT%-4eT}^@mpoY4T-!Pv
zS?&$44m6UBNw|oHIz+_v_$IFCqdI1^8%*C2A7}#88vcw&5CM0$sr*4>?UvFLJ|`gp
zN)?YFSP^)2tK2ofgE7kOCaUZj%5Y$sD@?vH;I=QI?l*X=`PZrR6aod<8?AfT{1<8U
zg%9bbr%TmcwpLT~E2yT;Sp!pvJ2w+N)KyC+r{8#3pQtpxc3J3Z__b-z#BXD(0sN%A
z&G*}UA2|py4fkSGO4;N_EVaK*cg5UW)=w$55t7B;dpxj84x2O+CJ%&lsmL2)i8bG!
zuxHbK@cFe4wO3uA5nePcnMl8N0zr070wHMXS@(?jqMYhY{l@&)>Rzjd*Mi1nucXl4
z$02eya|e&-ApZVt%tDfsA-i%T)&ZyD@!3*K`Hs>rzOdo&B}(8xX9XC9M6zjs<2gR3
z-z47Fah_bE!=+r<;cP-_TpeC($+QsWNPpIvl`H{A@tF9Ne!4`5n9SBz5ul<v$qXkk
zyinQ<@u}E7m+rEg)eLdfp!ahCE-?+_Z<vmYKW2*`qsbBpq9GmnN#G*tI#CPtbs~nh
zGC<!4FNN=q%v}BjfhiJv@~*Jzw@wrN1M~l^!{_-w#4>V~SWd#!k-*KQzraGQCQ!JD
z=q&4zDK&S4v+MLa{sS9=_>Tv4PzWFC6}s%>Cn5Ro{?f$zv&i@|%hi!+an3UYhw%Fl
zSbo3b$w)=eLd0-{LJu7Z49dD?V=tIl<JJxR&;DI@)HT=m8nc+r)lA&_n}dQlxRGF9
zapp69P@s6{kp?(KF2gy&O}T4ZbiJ+r{K4)pF%4GxceOw_nF;?{;u)kacnLi2;E#_$
zYU5}3&W#K*q}yMleN(d1l$}i)iZbPO*$*?#(jQcL>lVgye6zxBW~nit^Q+kH3;8(A
zu0@Hlf3y+DHu*DjWzhfhvc&|s+1-iB9fNzv|A3GXHoa1xHAQg`#)ohIxL=4Px4T-#
zbnN*#AdXuB{rx<|NR5*ORe3K#Xf=fdwq*9RpJ;lYc(REFrwlk&;B>I_+%H{DrweV~
z+t8=fJK0=`^L$@%egN`E9r@7ez|t;Ri>IW42uc1Y>r>gv84>gOgi$6Qf}zXzAVpr(
zKk#ml?~4pt>N~IQn{HTrH*FHy6H;^DJ;}JW#{F<$HBRr7w$%zH2k1WVn!P5NrR%uc
zIgYaL**xv5YN#%d&Q>9P7sofId@IGYQG{*cgGgluvrY^b>O;*~dbEI^_%wDFX4Y)P
z*HN>IF*DJmJw2^Fx_e*E!)a85z~o0$PLl6nM%Wdb$*%jy2qDd2mQ+D0{ejJw0)6ke
z2~D~#e{8KUfW%E0!d5`~9+6TKW04T1LF<|8DIxCg$iEO_&wvc+!JD^%$bW}d(m}x^
z;J^B{KYQ6YlZlSxb{<o3ddD@b;t9<J;cA1ocSaf61JfgCVaG;uhLod)18bl0p9K$m
z**H!u;|3pr^PUb(;+=dGm6nR`G2)u;b!=4@WYKF#t)5Y-9C)*)&7x+ZEg5GQ5w?R+
zG@)5ci}<ljPop|<y0wWqxV)y!U;nwe;01(x-0fg=p8&YnUNjamAaC5d0#IHjBD(<Y
zY*Nq<=x{dE>IN~CIGPiyT%R;u@y8$`yO@2Jpz1Cwvg5uvKlJz9g5Eo%?T3I&OW{B8
zWB+gPXrV%yKmGvFb0<|KCNFBODG8s+njEKpLnvz%5jlN6`<&KI0gMFt{w!!6z@xEN
z#&ci01KC`{OlAf0I!`b5tw(=G%vf7!n)DPm(`MVF^6E1O15W*M`3>hL_)s81O?QN*
ziH6JEuRF?2#I24y=XEf*RcPA%aRxGI|2CRi^9JZ&ELrvwp~RmHPdCgPrz=STWK!(=
z(&6j)OSu^rx3@cjv4qmP4NG`$&rOVL^m!OeNk9lRU4GaYHl&^nUXi4R+2#8i_m&KM
zs$%*&+nRD65~y{Jz8}AEGINlM@J|nShe7*zjWL!W;xD$<p6we7;l7?fa36F~zW9N+
zjLAi0Xi$NFabzNklNjR0Ix->((_>SO=5iTOny`i+Fgk^DO)^GVNj;mASJY#h^6N9k
zYP47pUVC*qlP2rVJ-gcuuxMcn-hU&b{-#=M8wt%jOvHodfk5fVGkX4x;&oH6@Kiy8
zesSF;`LWO>n_x#$hrl-)ltkr@iw)(o+sfY&@fCd$^7ng*&d0`>RVUxB0`u$yJts*B
z7+Q%SZ?aI)KXbd1x7H%fUII56?%+|&5ObpK?;JT{dML#8gdWg4EyM0p02<HxcmBpV
zPgaIR)WT@8>jc2qtNaUYq*cBwMo&{NvdC?o8*>tZRr&PCf$)M0`7QtFso~Y79N}2w
z?S}lD2fpzmAHZ9Z$A=gysq2f8)@5Y;iKst_u7&W^K7!KDmT|O=Rwqt0fr16NK<_|e
ztUTdRlf5C~C)63OLL$?!DZz5(^kW1q{E)B%I0s_Rvn2q_H1q7*Wi3`WG#5r|da+~W
zMums2%eP(~&ule6w<5=|viKS6EKe0(JFLQyj|A+p-3{}8I(wav4?XxUy$2EEamayu
znSwrK^CA_$g+=M4etvz9N${?@LM!H6Jj<XyJSA;Ws~?0KZCm;&>2aWQE@x;*RU)0W
zf5lNzqVR<te>fS<8^zD=?kbXhLM_(lo{$$7n6QOZTHido6KQ%jiuL}X3wQ^P(o=Ys
zsz6*?zp{Xkj_Bo%5zmof`}Hx4Os`Y>+6N3I&`g~G^M>R4e&ta_npKPJj~?hI2<9XJ
zy;w`<s+>!jOUlEYg?BQle-Ca8xA8g|-q8;JPPbLyR2sK1l&QjnpSk^T+TO1)w9*ME
z&r@LTYutF<$svYJKGL_K!3(u<vj05^SnvHTZv2+@i3G5){Pfhw!nSu4U+1^cE(r&-
zb>MCbHp52%^jVOH(Tdy))dc-XE{^!74(=c&t|9;rcXWwJW(UOKJu}t(?lXB=lBb>U
za}FM`ig((PN<!XXo*CYJqxFO99xCEIvki-2SM$7|+?C+-o~6?lJI$Ry=l=QtkBY!J
z+onKRQZxnbopc8)b2z9k@qQdf0y1LaM8w94M5-q5o|0Wl9q(^jyVJF~s*2bC>|Zjz
zC&Y<;CdS2s{hz8_Fqg@y?1`&L_bsHoVOAuew+Bi?Fw;_l8OGi{9;?o~ccHcGgBI+A
zUdP9^V3TU;$~zx~%hNUTi)%U>Zi73V+1*exy?(T@r-R~RC%4%O2o@zoli5Ij)2wCg
zmgf-RZ{aLe_rmVnLqF>~D$vrceZ%b3zQKk=0?G_PsjbNgi0e}z2@0K}Cqte9&puz^
zz&nTVF`8YlB$lr(Q2#<|g-sA<v6EBUGb*pU)aZ&SpW9O{_gumWJnA=5T*zJhYy+Nu
zgAcfn@sS}Fu+Xa~Pin>E1$#i6nr^BL#?{Ztl0_Sx&(Bxikdhv4YP#I$8piF9lZZB>
zGWluJY2w5=npZ-l+2|+AP4TF5>Ew~8{1_xkh8rL%BVhWk1b7akN`H;9xay;OjjGL;
zIVsRRZo><a@Vpr@ihxjAc&te7!_^F%2<BuqIuiKyBy(nakz3Ay0M8kBlOIME(x|Aq
z3{!PAkLRZ*`iR;kW6gZd2a2#(iwS1xNA{#xU}X6_qC5n{Cem5syvo_;i<OU}Ey<VT
z%*qnX{j#lh{Dzc!MAY-&METUYt6PRhzo{<;{h^pJKuoiUU<f-44RNL>XbDh^)OdZ4
z$gIjM#!h;Xg`pa>qJr<cILtdN`z5OuG|QbhuuJ6$Kiwf8BTs5Kp8x8~F*ax4wEtQ9
z<+khTWt!~1mkb1ymW#A9lfo_HSN9_@f)crHMQ+6d0#{A5*j9c=VFnqV%j><4qQsD~
zU35_Kg`4-r>P)=Of>EOH8!Ur1N5;Y-Zu$)3CTbIrdR&@g!IGEx2M%1$lHmZ^eC!gX
zl)->3-poirxECMzc!y|_3-AE&O4tN>5B$|yKQg?J^fFRWB}CEhUPHHRO*d0%qq*j&
zUd8mt<eI!or9>MO=NUJNd%EI%MI}987@$|Q&3)a>PA<#qHaSg-@4URfF;^&s5h!e>
zJpU|uB=j1!G~Z=KPAkd@7igbRueqPe2^X~)f8&}>mB#~fk5}Kcf{iH#SW;=w_k5}l
zZ!2->a`v%ON|A~7-8%j7k78^G&x-w@3qGbXLgYiiDqBeWhC4J%j#ThJf|G}9*ON?Y
z`eSrGr@MufJ00lRBz^-1C0Khwb@f7}4<CX3MVNS)QUPlw0dp0;fLdki7TzNIWCj-6
zV>nv@6tq&n5Js1D+EOP8!Tg8E_-u%`_&`Z_K3KvIA2`P6OMJ!WOXSZf97UPA9Ya8k
z$#)4XwkHqjE^)^kr_%7X!vF9FE61vm{>d%T#%xx|Q3geeH29V}$Nd;{37~$!MltUR
z1Ea+InH+jk;Hk3=hBxXCx)Fwb?X*pnY0<;~!L<MKnKtCE(CD)PnAAU2YB^2-R?$6f
zc+0Qi+mok@skeTBSpbqNT&yCAa_(vAN-3VN?ZK}ab!;yYm*MAhXX!urIG^|t(9blO
z3EU>R05wH<6hf&ux2_>mqvgswLX#c>>@Qz7n}RR?<0W`)eA8j~r(kYQkC)*Oo0RT?
z7ey=gKc;vq3D?h3o6zaUe@Gz@oUrsUEfHdCbv87>xTILPV8IJ(3ZEO!4CT=i?LBKs
z6>=;;Jv6XYNMx!S5~29OBWVUP(X?9{v?eXW*N-+#PhET)x@=VvyhcGU)4I)oJakiA
z^@L4L(MsLIz{~G}I-v}aH?5Sx!i-+Eue8CekMDu$2V^{(K}`?=fjEsG!6p6dSz6>u
zlKDDi|EiEDctYc|eJj341RysedY@LWx4><h1;!>Gw9<xvzZec+#uK;g4KKnrX_0WT
z>*y2*G1Dd)kC^ASXO7(2kJgwN4uDe&<qdf{&>}cUQ28Z-JV;`Q5I=E<ov<&@fa~sN
z>V8rHST;LEJ=*(@bG^)4BUMnd>?h!OzI_fZ%8=>(B$^!yh*gq~V0l?MmmsC}AkOku
zGv$%m_Fe-Q=o1XSy4@8INLk6V2&?$Su*|D0QxUJ?LCtiSHZ%_+vahN88AJ3Hqzs4p
zojgC`m6*8RzuPI&o3}stt$fspUrf@H(6jE)u512m-ZZ`vu8!XH$b688qFkTu%NU=0
zB*+f$nsZf4n4cwj0v$F?j;v-(i)!W;5HMNUtO29^3bY79obg+uyocsJjguyigyKv+
zV;tZeP}AVBkwn_WtW@%c#R_?yA2pr4I@qwQXk7m=(}gdg#VskCmOUKzwFD<uHt*fS
zfZp_1(r3XmIiUJ~CRy&sW0D~-BP*rJKQi{ySjtK~R)~JH*EdYBciw@poiZ|Zen0jq
zHt)Xor03(jcYmDOIk8#CVC^#N>8%EV=8`QX%Iuhj?t$Ag?oAnoQ8J8oFhr^18=l~!
zn;gt?Nb7n`@|Z+HbmAzlt8`O5shTwq)Ap@I#z-9Q(pb3TtUOO6qZTHx#+?;!QCoHu
zB}t!FZ4s~y|GKL;P#nc#9H3BqhJ@GRATajGgS9U4h|e`>fIZ86NQg0nG%j2=t2JaX
z)wq|vYil{jo3@GV-l)94`js$WU41B=X+8>XsDAad+MqF==y~~aH^1HJ=W8@0)~QvA
zGdbp?+q7tYjniEmWo@lZLRVPdv{-_sO)@8H@5hajlyo#5aGYcF44&fU9whM^&nA4h
z&nCpDiGf@=FsE#I@$DrQ2jO`^8^wkx=0)7Ey6%%^wHc|exzuzeto!R#N-~E{I@BWl
zZGbk?7ULCTuUFc}5|LSgHrH5#E?GsbKGCwiv57ZTP&N|)pKZO%d62X==#IU+&-&Rh
zrPqdN3yO=2v&=`WOsmYU@nqVeOz!U5<OVucn%Fvbd=xDgP30@dUGiGUb+KQrXFM%g
z`gFmL+oP|fU&TcPePK-xF5I;qY|?oY#4GEA7!SQ(yox9MO-?Z?;0b@5trA60p6CeY
zMz{rCp>;y?1C<rZb<M%N0v_4;u14%7Y)f;w=)`!5R(Y(Fi*V_7WKCgdgxAekbmi2w
z^r^e8>|`A{WznG$R*BhtVJ|F$IOA}{k8}5v$i)vH75)4-W}qq1G3)0w&osK2A+8eb
zNax1q`J)l5VQC9)1N~yYbcm&ryh6QM2i3kBU0cp@)inR$`?}pKCZ!7|)a91_3-7`%
z8kZlMV2kRf_>P5p)V2oSTrQWP*EuBNy**#G?+G)RtRHm@nzT~OR*t_H;nH}J8fUkj
zZrfm*uxWo0Ho!QD&NhuS5vlCIp@W01Q<mB!r=w%qPUPo4=?q)*e+dRDD>JD~kF#&8
z?5_eJiLZwra_cH``Lq8_nt}E~EJY`<H?lUNbNrG6GbiG)3b!`QU)!eS5^Jyhd)Ki9
z)1~ZpIBcWrKof2onz^-nK6B$``ubRR(yU&B=iI5J=~tS;YP+;Omr-V)Lyy%zW*#)d
zQ?q_eB`Y8DF+HF~h|{ovj_dgw&vWd$CWW%nVz93+t$by4g6mC56jw%dJ~G+iP9iD8
zjULO#<auzz#0`J`<mX-3`#dD)57XqXXuBYhu&VfEg5RUiI)wU=I7dq7`R0hp1_OHH
zWGHD@FYV}Pgfw&Znx=BjQ#FlRKj12C-jE!2lg5msV?#{fuNFRA+q^~C7V4Fm&uV0O
zZc5z9d(ufWmnTbfQBQ%^^=RIY$<dn~PRV8Rs@4ui6jr+IXEQ>RdoNst61^R(+6|X5
zu)=ZYcZ-zgKAoAZlpc)_6P2q~N0&zO%XF3k*hsUl{4FCyj(5Lh6k<F(=QH#BO!|(;
z{99TfkNZg-t?(M!clB<{3&QWj^E>q~z%q2$B?@>;acf(3xAMPe;`sU+y;Ib#3b)cV
z+Z&zjkTl<YHrpQQHP%z+<Q`E@{xffYaiG~<4#P=%|2W%b6MokJ@)^VDs;ndsV;stu
z><s<L*p!{Lp06j}y01B7##PoudLL77+?&?PpG_&;o|&Dl?VLZoprCzA10~#yxi+~n
z?9mAZa^ZB6rG{TUJ2HkJAJJ~Rrjqw}AxOK`hF%Y(iz&22b+rbZeKLOAyeK7Mr}rm1
zf8?$XcR!C%xU*ApQk-VF8>YLhtKWcJKzC`+L^8R)baJ)0uZHrHJ*t`eVx_8{tNDfg
z5&qdD8#&H$_A&3BdKX%}aQPdhu${t}@@MCb{uzoSBUY}YckA`(4Wg;er(4?_sbJ!B
z3wv&UH7ZnT@7CSa%VUIo;QH2UOh>-kUBrc!76+TU`AuEvN_W!lJo36fDd9;%A1&X%
zFbm%kfXRUKROJPT(dBx67p2H%A+zhExI?#&YfOEEYSi`la|*MAmOHmp5$2O0l)ChG
zatk->3O?ZSIvkX*=c(uxxGgot_tf~F7CRQ<;l01<Z_~pLSJ)Q{P+1}?AdvqbV&cE1
znTw|Xk~zu+_*YJ&K0C5;liq!@bZ_LV$KNPwMJI=b==uC{%Z}~*zBGr-H3^XIHZ(>5
zX;f%;qe*SxyF`?=<i!_<<i{^~mSdE@jmYBxuY{b>^EJ<S4!Lk3gH8=lt?-T0%vS$o
zpy%VvwFLdmN>qhd8Z>&r7QAYlW&T0O+k3MsSNgF$(DO>$hwrBj$bcY$4^IXpf7Cb|
zoHuHI_Yd{)RignLvX-jv`>a__ys!Q5ckHtjccOBdJ*y|e0(J#vHop`7g;@A07+xg)
zkH6W*xzx`2|AX8}WxU6d-AsLlKbYmO%*omS&zD`ZMA*;8tptCeM~c;SU)1G(GS_o;
z@*MYH_=-}S=s8T;ULCSR(&cZ&$1S;@zR8&f3+TD9fATr=s7niV=MCm|O#iecjpv4{
zjqvv-7}r*upVuBCJRnzIt`h@K>CoU=6JJ@=uTk-J`#larCUk0$vL*U%UjA*HN7dcl
ze0Q6F^G0_X4Xy#ZzNGEoS!Qy_;4f3}9zXb)Dib;l$W+e?ZYOSW$Df4cZZo6TA=eb!
zXuu9GNgvVw3WfGxSL(m6P6mAI|0RO%f1`CHA21YM7<5~l5NLn)obH9<7_`nX_lx?D
za1Gs>z=G!H)6v28p2hQht5R`N_077KQ@3nz^Qs9&g%jpiHo5L1T57#q2z+)Z8>>xZ
z-Se*mYXO2a1C*9%1XhA#S?|+&JCRpwE|Dda8BbFr`YsngB+fL*`GeGLE{Gv?aEujs
z*WbgG@K?L2Sn$hblYk7w$|yb0o@ZCK78YQYUzYSt4;{DhoKYRnzwP#E;MTHE@>XK)
z0`d9HE>8j;`%uykfH`hQdIHWtB|4j_(Kx;Jjox@J`SXN$6-cw$D-1}(Ez&4NG>OLN
zkZNSIyk5N6K_u=hmsqF4>+JB%pb{wnwvi0%^A0)KMT1{z;4T?-b*8)UJIi9|$Oz>3
z`U1*Fqut=%XK@|IPsd8-mZXFG&vGoQ@~#E6Uj&(ccY)H)1fwQ{Vf~uDHa5wDd?EtK
zYaC-u)9C7Xg9KrlP>=Pi5>q*%&t&)L)*?RQpGm$@)w3qOOQVzNG`|fSnDgS;xnO`Z
zj65+)#4RujiD9Up3|LZKdTBwuExtlg@8ZAwoUWAd&X{J61k?I$Zjjq46*uT2PVwV*
z-65MWjBbcPVedpuP~0TRI{%u=rFtT%)q{@e@vXVs1BrXV>Hw+Nx@gTxolB)V4C&M#
z{8Ep1&3WlLbxi?SVO=F;j5Ihn;VOmo<2i#HuSt<J_;)@W!!>qOlWbRRy<AeWQS8e+
z8{k3J3~9HMq*&uyV;v#VGqF?%qq8NBFFW82Zc4ou2R{}o6^eMCV?>{h8~@dGX|>Wr
zj%T@*tO_I?tGICPSb<MW!2n@?=YUOVfQcUX49<x+-PL9CpY+Ez+#3LpD`-kQbj5(w
zmtMQJn-zsc-cp4Yt!1uC0wEq`?R41gOKyqa0d7lWINwLX6!hg2ySJTudoF&dBf0}r
zR>+UI4GQ)L6E^E<4rb$a$GeW5`4ZNpr)4R2Wly7;Bh&*pY620NgPF<R8LOq<UY`(!
z-{t{HZdK*VBix={j*=8yh5_G6vrOhHg{C3iRrQoj5(68p*o@D}v15sA;3{l6ChLXU
zCH=*>WL?ua%Sesyw0+H<_r{m-0MAqf((44UJ(f{n#x+sN+{lolw{nL*0vV>>tDymS
zGTh~b+OOyU&-*(=^wcyKZD3DdOD-4IFplpN{f@?FV8wmF+Gy^&CrI5pji>lg2jeMz
zDMYl-bk1xEQ;Ql;JV<+b)Gc?4D(TU#_!;@z5*MLS;;(WLqa+A)w;;@|>azUuZ*t!z
za4=W#gto{Wmuay7z)6NGnk(2~e&;xyri$>KNz^KmL25bsVkHhyF*JX9bK$N>i-8M)
zuScJW<g5T!;ouJuw@}hV02)imuV=D-NZ~UhC5AHXGx)NDeU7#a)3OcET5XvwkC*?T
zHAOPzCO|v{`a9|jhU@GC`VAiB(Lc6y0k@oEg2!fI53j01I>RP#uybVg=3=P(%(^0Z
zop;!cdBHa#fgpB5EzHWpmSzFxU(ai-wh;Vl3D?l>NQ~$5Hq~|>iYDEEtn~K&9$;F=
zP>C9=eZ8uyW6(YP)6HQFFQ&FiskX;T$e|fio*ZxXest?3h{EgpsNMS*Io+gMoR&c<
zp25xK{^Sk3xH$*^QbzGThkmQwz0qKogk*JMpG&8XT3T~6b9|nhP?!+mg-hcE-|w_M
zx`*In+|0V`xj$58cmiH`5Uv+$6Ql()<``Tiwu3$0L;RT)f(bo`y+;=%7s?Y#nEF$R
zEK5}O)L^>&vdjHRI|ouOOmBB%JIJ$HFTwqL2Te{LBPAXdnIggKzp2#l-%teqi8>Gm
zt^Y_Qe*EN1nzn7E7py!O{W3nr?G~my`q-_+G4HGor25Ny-tsSqOkH7vL|A|<IJtbK
zjz|{B0}~gSOW&Jo4RBHJ#2A%+G}ZdTHMtveq4b&sNGa&d|ELo#kZBs}u!j~Yy=I;T
za>^|5X)!XcYN{*6q-Rw8+%8wA=cg{+y1Lk>yke{fhG-kEqh$>Erv>sf_wzdWE6}vr
z0%(vfff3;lFmb-Jm@6c&ddn&HJkdC?Dlj?!@t?G|ysfGs!y8xSQd=GAgqz4Cz>L7!
z=M&1`3+sc4%BRHQ3Lt+{RR8BZ!+**8SU;ix)^FWB^!-YZR7e4SzoRPWZodXyj_vv2
zhYp*AEC~f&79F(AEr)`7`K3Wt{wDcJ2{0v7MXg~&pt??9{U<!<p*8jCLlR8MZ>No5
z%zjE6T99J{%saX1&p#c}Dh0Sr1!s?g$%0v0fb;SXf{^G$9fk6>Z#g-&0q>4NvQx{?
zB$Jr{uNKl2fcEw@GXtH7E)>jF1E5n+Jsl%H+pM>o(UDYcK>L=1`NpsETL|{UoJX$i
zl5vj`Oo_O%g))h@C8y=P-L{+L2l+L-<W-D%qISzlrSrb|;+Ky@WC#gbjveL4M00P^
zexg{5=H7OtKmq%~%BdKjLM|UHC#mJut!;|9c?wT=U?8wgad7e!yp&IIriI1q3;NkD
z33NJ2y?}l6(|&v{hkcYd=5xbGM4q`u4ZmwE7dw81@x^~pa@M3kYr5o31UTZ!jxr8j
zPoC0aRn#nv?bY9eR$c(v4^xpqJRJ=#@}q7MvJ<81HLvO;4crA?-e6`<-G<U*khT{=
zm*mv;EJnY5@#Hk=PNEHe#;W_dZBj5!&JCGMEuVmyMQ>i88hSxRKz3z)=UeWBKGLq)
z0`3Zs5_Ye`In*gX7<1QRlMF3SV$FcO3Puq@PGH{qJ%aqug~ciSQx`wmkilSW-IY-!
z^Lid2FLG>8PyimOnQ^`nn{b)4=z}3aJwbMx){m?JF)(E5jQY@os%KzS^I}w62sNt=
zRWaN7M(U=T7u3!0{39JmsFJB9_~pNyVaBc8N`5Jh>^X3A^f(Ic{N#xwr(i;KU*!f!
zBHK^<S=Cz;j;bxOuK53&`|7ADzqRcF2SGqZT2Vq$Qc4;|Qba-N91uxKDWwNYKt)24
zMjB}tni)VrltxOrOM2*j_YBtWoU_(@zO}w@t?!=^pV)oh``Y`tu6xh>=74UEfog*<
z;_bsFes9rob2R!fWX=f4$HYpnJ>gY1_QCG<AO|uSAC~&l%U3+xKPx@Aq20&BBXp!6
zYFER<g`VOMeq_O3`@tF~R6vrf=H98YkQKs~_QI<h0jb+PdB%bx*U)#*DWy`FM$Nj!
z!%v|a_ilt<1zTYkP}s6z)pZ#U6?^4|i|tD~F8mo!l-0c*^2Q#T4nDG3i4$w#PU4ab
z+G_#a(3X^n*e9pp*v1>VvPmKdR_$la!H)G+eJgw%qc^dN50A?C@|6R#RSHuw7=!4q
zZ7Unpsm`^!lI?Srs$3X>hzXz&knSmnbP{%tyeV<yd;qEZY%m8r)jL`b`^U;~MO4eY
z(`T7I2fq99VsI*~WNG_po4?aJsVfe5xqoeicTExiVlP$*w~wZ*ePC@FL}A)OC~rZL
zNXekUzxTSNal5;@0MMkPd^IMgZMR-o@5MCBMzx}R83Oebci)EAOh|bIJxEFmProG=
zn<3ImrVqOp^t`ZJ*UOcFr(g0ayFn{{+^;;m$zhArRarEL*V7)j$da=AeKL))YJ5Ky
z;+t4kWxrzY0(3HP=3vO$gn1@S>wNy{%dP%EHQ@AA4<c!5bMY82B1uT^kYE!MJJ^1L
zpNNs0y2E2`kvEY<zmiTWp2F%ZJcS&nO!V&xTJIhI7_D3dw$4B1UQ%RJ&~R4V5z#Ob
zUYLkoAY&FBYjG~Sx-EL;sM*!<b>F2L>(%l0bsQ!8wx1^5%U|`!%;1dZiSs{6xjwPS
zN?%#LNQ-NDDP$e@!WL`!L_sGteb*OGo->}As1w2JK@f~LByuaHGUq(qy=%*R+-Rha
zjOo`#NH@i;U2Ny-8d@9@IDU~_>aW2LSirWu>Ud&A!RYX<0c3&C3s9KW7K~s;`*aa%
zab^9|q<wL~L_Usz%^<03e8R&!!@pnS=#`W;7I_P#M(YaBo5fYjC!mae4$uyz8;~76
zq+q6w!#RVU@}aLkvyr+b<yz(*4)&6*TpdR#aGFP@n~>kZq^4wdX@+r>B#)8f<@GU>
z->;{2)jUzd5ds;6@n(+pR6={^pSWTD2Vhx3r&CrIrkWd(y|aIkRD}%U;p8tdc6a&S
ziToBKxjVmqlz9pWGk>c=pk(ii+j-~QWX|v$Tx{R6^9Uk|7d5^%?#1+KZtfv@)YeAS
zW2<}mA9H74+}L?}!%bY>Svc%uGjjMEJ--9S+>+%`fMK&|#VpGBMH}?bOo{J2m1t~G
zqVo+yIJNe%3pvwEO(T8V@eDb^^7CjO5hfnX>A&Vxe+<I?rE(Yzea3L>lc^FDA=I*R
ze0(Go9{AeDgc$OJbQUn?Twf?pchlkuXw<fGugJ182mU*PBpZMS@UP`U!9aOrPpxaU
z%3_FW;Sb=>>!bG=r*-<M2~qEWFvjd^c+qROk2<qV4&(P%BMwq&r0*{d2Y~fe!_6r}
z=CB-#^7neK!W4TC9Tn__>r`Y-lZnb1ucQywwf>|!Bp^)VebqGOtBsXDZJ(=5x|^At
zR+NM+O(KaOVafY7)lpvli%Ol_ilb5+H?6wqVSqleVapC>5}|Nt(VOG+nfdB~W$#%r
zE=BE_kh-TKW2_8?diSv`=;j+!3VJ7}D<0Yl@bOy<V7EE#rR3?v%dunYH!Xs6RSeQo
z7_#FGnVawNSbIXXmZvqJ+9>hQr(lqpYi*+OZP>lW@fLg9Xn{go-=E{|m)6$*zKGp~
z`NFxxj7a{df&((Ee|uo%rVnxTM2&MQn+W_4gSj2fLjG%c?pb2I()d{>2ni833fWIy
z@E=wj&i`R48{A*+C3a%m?wo&lwt~CJ{@gRGV%I_wtU$5TBGy86Zxh{|Xfz4Xpv8XF
zXD~Z#c~jT^n8a#Iv%O)Q2psz!OV`SPaC)QR%JRz>IxKX7=;W_k3o0-Ye_25OY3C)C
znOJZds?saq8o#ZP*BIqn2-rAjVrg3Q4G$4bedb+NdPQ90cw+zvij@Ek&kMm2F|?hv
z^n);tuy`68k{F5}->k{l%yajqsOWi|l&exbS@>cbxl&^qxjvw#5@Wp@I*=TJWhE8T
ziQ-32UX(DQ9&;Xc{p+x7!|c36uGy|v`RJnZ?ry|pX!9B8jH4J3+3=Z1KlRTZ+yo<r
z@%a~X4t1MZ-OabaiPHje{ggfVH|~uU1V<t}ZXoR~9s<S6!$?($oBRSiiuYQ+ggXct
zQ1v9cSVT;AWxcDoTB>KyWB?g<B7sT7S`oB&PergIAe}gUH8y>OGX%w7@eBlc25L!?
z)tukt0WYn8Vz?=+T#~H%9%lf<`P9iNZes=p&!pLge-Nl$nzcGQI_?%jIPY@BGR(gZ
z%I?obR{hIEP}4r@YT&M{?URmmFQBP&@%||kA>u_4TG#W;v8xTPSC(O!dCkX=8MA@^
z$O-CCV)Dyneg4RTG7|=&6NcK~#4rNtrgr0Zt<T?+pQeU*mAeY|e{VwCpOfJ2_;k(Z
zwZYAv0-Uf0$>Hqmf&7aE4W+4mgWU5ILpcibcxd<3&!%@t$_oy)E3h&=pOM&ObSNHp
zTyQ?l^ba>1lwkz$MsEYzrHl8GD$R-rCiP$Fg#mw5&m(H6_^lf-c=LyE&d`OomY^qq
zv<RnL$FCu%b|2VF@dfmKp7ROLsHEv>XfI_hr+#~m#mjL6Ar;j^lL&LHA0hs{Vi&=B
zXi=AgQ7!yl9@$}7IDQjUv-v~ee3PoJg#ceenN^N&s9T$Q<#jgs)~)6Nw$XcRdr>W4
z%+#2Lhb2A)Pw4f@OlMZ~TZHjR9EO!9g_o|nM5h6b6Hadrm8Gt}tzVhokhYuk1d#tb
zr~ti844hE;OO@xBM{pVIqm0fco}8M7pzU~Ufes*t=pW2>M%yEi=OhRSnZuB*Ws!r8
zagfR-`wxi_(E<69rz$PGW7|-9`0GnYv(DqcBLXmz8A6I|o~K%Twrbds8|k7#+O9%l
zB`0)_txKQ13P`=aIKIsc)ilfUqv;zr?JGYHjU9m+>-7mjJPykQXtm&@BX&Dhv-4Y6
z1R{3j)I-6!lX%GE0Qtv5ek%pM-!Jrx<8VKgHVPUi(E)#}SG2te`f|{U7kTI*NyKOf
zH*|7CS!Ke1OH=pBVHW;wtm}QJ>p#<XNNi)D(&U`W;6-fwc=F3(VHXKWg;|g81-swL
zI)$b`jZgH3>nk<Pb6xhAw*+gDmNbd%{!hr%q}+M{MJ7V>sa)|~68_Gr`qyXP3*J)n
zV5QNkmARNjl5p&=WH$R@4nnz4`{(CYrW*aT*sk+9(0p!xm*1#9XK}ToAUm3%qCuiL
z@yeW#^H`NbZ<9542_7^lFwDOvO=3}~8rnLTP#4fk-+T4+ekF)s(n*42(PEpU6&G7}
zXm4(XbARu1;LM0bYI9z<acA4H2kn#PJcv{1Na|k{D0yRoTWGg43tiE^QbW&)U_Kv%
zbs^aAxl{O7JI%;Ce^?`iGj>Ev9jx!2L{=*!Cz>DIn0-4NX=Ph}1Si>yfa)HQbgVGx
z?j3NK19pA^uEok*GtnLuP+Wh4r3Nm9i9nEraj=dveVFz1N{0FTMazlF9+x<ZEe)OS
zvyn@<@D5WuOZ~!yCvWx?-fHm*q?Aw`VD}wzk#eOX?)qH&F##98mKTvad?WftAthCN
zEy}L@`jyAjKo-Wh0sQvds{sZV>)nh99F=&D;Ps%g%hmnawAr3<G166Klb%;+5n<Im
z9XHF#U>y#rPawx3M2uv>A6Ktljo66yz#>#OL{jD~@{`+O4R(+*tqdf-oI>^DTYVS`
z6s1{+5$3ZDs}$I~RnT~2&O)6cZK>l@8fXhF!%|JKtK3@BA?Oe;q|mvs?dsVc`CN21
zj>efUg3HN@6T2{c%u$3iqGe)V5LmEJ6}Mq~^y0?ZoA1>vlU<qPl54-b7${N1m`_XB
z6|<@*Ha%XkT%z_!HM<knwc^#5vY!TLgar7UFxj9i*OdQRy!#AdlKlZ5zz+a|e=fkX
z=~4_Xvm+W{u>jVUj?ue1El0yR;90kRR$@Srjqwk86oA>J-&A0-kYk7Bx6Ex*3P-jR
zR0xmi&x+`n$#<tih2aKwozIE%pQ9E!YTh@`{ZM-5TMe!x(kBv%uT`GP*It^KTLCJB
ze1bMV51h>0|4(=w5gN(g<%tL1DBYoL2e_w4@vU-k7K31%7m3q6QV<yJB&c+#p;##J
zn=^#czJ&|9D^||J-GbLP`g0aNms+ST*R)6IYSu!2TO@S!y8c0R47@aoT8hiNEoC+h
zzOyMkkwbP?M9;os$;CAQc|L)V2*#j{=K`dGXKxc%@7IE_zz;`5c6`0xyQjFf5*kJM
z{2=hf6C(HnnLS4DuWovQTjE>Qms);DRluq!_&|IalA+V(F3TA-|A>=<k8{0olJa!P
zt?%6a?QY($AiQf=BTh-L|5KVlW}iwGeD3h@<XsI5uej6Y$iq%R?f(rh!I~)jXSmyL
z2K@hYzvgGmm+ypcryud!$*d>JLx!3d5voV-_e}-6abe~nItExI*bDiYFal-^;E0ir
z*a>I?{hzwvy&IG>{P)X7tAIovK$6hU#|{<&e}ck*h?`MyarA`*Z3_`$Fvt&JL_gW3
zWY4Vu@$X-g_Zz{VnQv5{`roF}6!`K)Q@NGU6Uf#VQ}1L?pn~1_v3QWB?fMIui_xrm
z-e>w9js{?WghD>{OUKbMUg^)-_t+W-H=<vLam=;<X>9C#73(b)a1U>73yO^k0T5}T
zDF<}K`x<FFM10pN6ibN3aAum%bKaGV6%u+m(UBe9``IFO2+6#1MZr_dA&}kiuq#3(
zvtsvba}jQ0zsS=t5QE#w0a~SG=m{Wo{Kgv^^>)Ulqs+Od75K-3$!lolt!wB3QGqvP
zTAW(<7+w=IBH*L=GaHS{oiWf2>9eCh#aK;~1lAyd&%6bS1#Xe6HsZ`LjibM1D^4M^
zhOg;#=zX;`({2o~-gz-0;U7%7tF@=*weF?fBo&{Fl|@mK;PW(Qy7=on-S>)iAG(3D
z_;@z*x;QgXMs9uyFLkyw6PSDv9xbWvo}3wf185Yz14H!gPd|~(_m&>4{;La@$zRhq
zMj*cV42MR|Y}u&pE0rH4sVWs$FjtI(G%y=^XMT;Z({E-DxmMJgMwePQ4JqGsA_tHQ
z$KM5^drv()k*-q;Dc}EnRpFLh00cs}p)7w@bNR63fxViO=H`X?@Au0%-#8ukvuKrR
zvT7-FUeaOpS=Na2P14K0(vb7EsMa6C#C)0jt(m_p2R>0FEpr$xx8Zlh8*+5Q7&V=|
z*?GE1T4kc%sf)xUbnP9^y|KNKgRxpNW~phdDQcUxn**wwP1`>3oiztFhxmhy-tBqS
z%?(j9hz-!K)Ii%7o`FHT;|HN+gb?DJ+$2h12Zi2q1XYnCL*PS4V34d6cxzA^SOkMA
zKHdx={w)OLJP|^e34($_qBw}N@NIhC;f37YO)7O>y0_Pm5*ucE=~9y(Hgv^tI_NZC
zPc9c{2{Y%RLfw&1+=`VGFvuxrbrcx_c=MajJ#*_&!$AbmMHuR;6Gj7RgI;>m3KChg
z-Gauq7D+bO!{@T&Y>h~hX=mz#4xHww*Q=-YsMqU4OEZ1$??=Zm9hDgd+`ZdV4E5_T
z_zrmM#L(B2WH3`0r0-9ya8EEo!1;{<`>T^O;CU=z8N7wN<_FQGz;6SQpKvECIx<4K
zA4Asg@vQz?+5ec_fAQ+ctZc^yf^Pldx0Tvm1=N)mu{2!G%1gpFqjga|`nIYbxn-Ta
z1vD1gStWEG>Zm?$Cp=B}R`Vj`3QLpSdztJ8G)d#ntoblg4f-xHIS_N|R|lLF@>S{~
z_PleNPwj#JQt?nmysVUX!g~;Dic|~2cO!p#qMJT0zRwfcUdDxO69k-BM3e-SB>x^f
zbM~1+ZkcTtN_(}{zhS4&JhyAFFyV*r+(IHvK-HH<dsH_!hGv;H*Romv_HL14$u}a~
zDzofBSAKDycWzHa=#=BZ#`>Lf{O^tQ;r4?tAr)lq>a)8r$fNO(-a7V`JFi6gxI%v1
zd;`&SxyubXVR)?V6gY829#1ovjH247&qi!%DPdRq!Bfcbx?m^y)v$2y$|CY?FeUOc
zLKa{UVmXTH(;K!I*$}HjgTzyh3^zWh4l*IG-i}IsRNS!zzv9^!ko1qIKN%F;e)Yf?
zm3=laDJ?#fApyP2@v5C?aq(4KY(7ATEC4LWqM;*=fd2l}Tla^L_$Y6%>wy9#J}u>g
z_QFMsq+L)7E=9!9Yd)t7tieHx@T8oFS0Tju(7ZP-<&6}`3Jo11lUoD%3ok0Xf~HI}
zXz;I*0*mpYDMNFUFzy7>Kob%aAR22q6_N@PUdF~};tg$s)cB_u>;}B-@;?QR$-O4I
zBSpgRN&T?BI5}5EA%&g9aN$zd;6;PeK5DaJ$w75Y!{qy-gY+r@q<|re8manH6yc-g
zY#lo^!vzNi`BLLe%!F1*Dy^b(`5tV`@SOPMt9bE8?9QdE!#mCUm_9}E4n34Ae`~&c
zBRNlIyeYb;+1=&3nzoKe#1H%YPg4h<nrLJ(*YSxGKy;C^sH(BfEge(C6@LgNl4ykZ
zK)@wwY3>pgY=CA*bzOCz^64eSP7>WrI!H9zc%b78A~E=T&;?TNGr)Z>8%^l9z?snG
z<3B52G2{BcJ#jtfag({KaBH-7vyISx-i&ZhH{TWCdz|pXGw>|W>?n>L{YH4^wuZ)f
z6j$@bLp`#(H{JeSy)f-cK9zNI8?CG$zpR40Tm|_hwD4s<hop>YgO{p5>Wm@Q;2RIY
zutLtJRq)pJ7e8C$d_veFroyFEqesXHIweEG?0%+HEMSSPJZ(SPS81Kt@gx{g2p>m2
zCzR|1=`!aZ<&PXnI4|Tr3{blLoZVWeLE#{an5#7^h1^EilFf!%rKa9eDA0VvT&0s_
za%5ca+><BR-wl6VnqzuIi)kUoq`j3|W^<-r_9E%1d|)p>@3angTtqOhQ9W=+#Lae6
z;nUJa6n|TvM!AuK^-cizJacvTV$c1HQCp<97cW`Ha>4gee7$pYA&-0@#ske%fi?3i
zpzXPvi&C}R$IVi3V@xwQWp_Kij2{eK(igl-%FiGk&x|HKo~Qa2{Ln<J(hV6-2kO+S
zm>C}5|43hA91+cx41=`Ti+^|gu{(3{%7}srAyltq3{m^Kv{8GD4|@4>c_sS={F$>p
zA#<@h{kvr)t}3_cx!|m-?VJUvv6k7D<|}M``r2@wg1fdfc0(GS{R+l18EJ+kw)wp1
zunPmHx!i^fNz1DB34CBF>h<wnG*b9m&D?fFg33L~Yt{+h&NF!98LFlANXCeQ1?6$Y
zh|T2FN)XA=6)J>FNYMS%broOGXQf102}+Px`_<h6+stz}phPAxp(}aDa@G|TN+`1G
z>KRO&3sHi|M37;nB1ObFZeT74n6x}RF;X%?0Yd_WeYJf-YL*P9_JyYGLw&q`e77@_
ztDA}s9NCBO_*n&F1dfLfw8yoWEc%P9`&*x>E2lah<>M@w3~>xEj}6*Jv$Cih0i+-`
z4?^bpg?9CwErZQ1_4O2uR{~ifhx~F7@wST)-#qLzahYRmFbvJT*h;qfo^t!-`Rp{R
z<BPi1)Fdl`#iguuTm@Pw(@0G|p&+ksL4K57K9w`M6(uowI7hjMh{}TAog+U~$ob4H
zFSx1~b^nm;IeLr$kK^Yw2s5s&VE4W>R=nUdTG&%XU)Z0=fap48cI*l=%pJHQ=(^fd
zsYmTzNAlb4Bhb$usEI>gv;3K!NguB-DM!Q|FV!{6UX0VGRX01$mW9z_-ayJ0H_+cH
zJ@?T6{Vq~gdbW<4l&*%IvAa*f7iEzh;CiJzs_t}04#r|EcpUK?0+=w?qS!Cc_^pFa
zv%iWd<Khi6O#qs7(8{<8a_9Y==Qk$EKJeAQTXk?K?9|^F0fC;`0R8y~N{C6~ambJx
zLY%xO0N4WEi17J;I0IkPgxI*M_Kw}OX}8hj$(#_jnWN!)s|BypwdT<^T^I}1T1Rug
z$;3J#-H}~GVtcrjNW6tqug8<qW&sxr{9dAznZ+3|z_3P)1p^?x#{BsE{d8XHN|FV|
z750xj1Qyj@V4spSM|7|J#z8zo!j+crAG3S~*E-{z1{;E0*FASiO154lm-lxVP$iA?
zMEf?%zNd^0raqeMzM=_O=qU`ouBH~%G8ozLcr+I+Cq8uH@qKD9$64_BdxzV0dC<eY
z98G&EyNE+>4z>Pu2Xqrd4OiD-+CqUV|7$HbTZJnd$qX+mG)HyKyE&)y8tspt(>63y
zh^AUiw5*|)T0OWJ^q7hSE|=3!kk7lJN3=Gs;5*2RozI5~o1W=quzY}nxRw&{@n63p
zz`yqVz2UKufYEa>%Vp|YrfaMaF>)dk@8i3wHPzCFX^s1}gNE-*?S+1Jd;(YrQV6Sd
zeaN%OuhzGn&o*tZT2$wN?g$4dU>)g#X@YT~<)za2C?oPYFGe@;mp=M@L75gc_}?}F
zBwvE$;Bb+1MAb)dN`)Jbqx~wx+ph%i3zq%?aNrSVM|`pnFz(hbf?)NjTFd8}CGCc6
zsd|}q0<WV?Rvf;vOU-52!%6!qt9w#VggeTf5p(ux#Vu(2RVJ#{D-P5;iO0@y{;?k`
z6Y~&_JUyElPS5O6b1H8Ix99NuCXa^of}#=cvM;#Q&}0LZo!l5@W6b8-61w&QOP9)9
zB=4(73>N4w(LtH*cqgg;*v6WF2?F`Ydazg*i)sF9E`U9Mf$JHt@CMijf_DVU!7j}I
z-B<52LB!4yC8#hWYKf>2qAWxv%l}(h5|klD&!*Eh=<Qq<A$X(#Ux0jy=}e>~^s?hc
z5WJXjji|cZ&+4t!uxvd&S+(PHRdBBEsdj7>KsIrpow^zf+30{by899i_T(#p&a9Dz
zw9B&(M?kzlw+a09U=XCG!-ebJX{X-P60YD>bJ-Ulr}t#c?nEZlwLqG8s}0PoE?<lm
zlg4AqVGubcMGRT>t>+=2BqpXn9>`&Jrurw4>nFvhCFTzeW`($2BZC>5f-sAu?`T2f
z*FW&BU&lDpxq;e2KM0<5PmeWM&pBxRxwk?}OVB#841cq!!h%jNpI%qF!|sN3W@U6y
zHKF=RoKJ8IQZYq+BrK?X)oel7X@=UBdm38_%!Ua+Gs3Q$Qx_nex3;rn-U`B@rhW{m
z=_Nr8kzCY|+Pg>;^5dqDT{^z0QPMl*H2AXHfa=(&J0IrQhn@(z&~I<O<eCuVt^Lsn
z+4lqG?y22k6_;L^D1R43V)|{@l4scKQmMO*KSr6)$Ln6rghHVDAWwnomDdU@$71<_
z$&dux+lma@4`T4JF27zs)^dZ@>B@=yb^Q3p7o8j|IQv(XIeH8fig{CQ#7Y?e7A<Zt
zH6neCG7(Pm6P<YTIvJ`Dy{xcQ-l&26q+?uM{z&Z`{oEOFya(Nrl)Q8y*5PvZG`goT
zdkIOwqUR)7F4=wMQy^8<V3F&)skyfPI{j(9du@V?dHb}F$BRnZ&~Om;&dKGgQCyO{
zG0Fh%1&uhg?b5bA!DaXi8LWOH33_=5tV=Eks_mpxuoB>x`bTdo8l!HHQt}KTufy-Y
z7u(7+^`mKoCvki5y4^Nh)@N1O*|=SY*^#2z##7!ks|#$dhi7CRru908J$PFyx+Th0
zCkcdItx8vU3yxj-p7k!mqc71?S~uLIuhJ?YuDDZ!5}@rK9^Na)quUS0q#8~AwOrCm
z?qax1&tHs2-e^<T<wsf;mR1M7G0A7aw5-ODT{jnI`7mn9yTK=Lv$-ies4g!qYsPan
z%p>rkM5EKr8~tnQ_&oUE6>=*YDO^0$c;(#BrEAV_<<(IOtCSDX=~MQ6PxgCE-7B%H
z)p)%|VTAVU=}Cr~EU!GPAX7cV67s575wbyJ>d~yn$@M8aJ{~<%?|IIO#Ezm<$84!b
z#5l2bCV7mQ4`Y2tPZ<t<72-alFd}bhQJ4;g+BXy4M%H=d)xu<htWiW>2-Wh&rEjCE
zMvOmghNBi}C22473jiAx+GnWN8JNWLNY$VrGC8^f>;nXTxbXt0cL{j#MOB=$t+j?k
z^=3N>aJKX8rvc(oFk`K_wGiRfQS($|SZVXYJRTnDvl(EVqhztb_^c9-+`x!PA=15-
z#YiwY%}*9`A>dyfLB}-S_tUnq?_>&E<iSbG@@Bw%Uq1)o%4tPIu%>CfEuZ&vYt!rC
z#p5CmZ`b`X{k(@_1>cD%=|bunYsU1=?(^Jo(SGW|;NtMyQauY3mZv_gt!t!LQhFY4
z2rnrmui{X3qfuCieR$t4{m{RY4%e@3WB26qnJi$s?RwoSZDH@8Z>b9kehQ43G)<aI
z+;cj2RUdoy4@?ltoTtSdJM|Pxe@DEBfeR)a|Lx^u1F*{8*Um63U21r#GraP&YU@W8
zTzI|r6VvJHio6x=;ho!<(iyI*t?g!K!j{H2Jhko`ntb$q`PS3N-Z~ge56VffmB?K5
z#eYKUx-hz(fl|^~<Ieh?%)R}tJy;I2uNZ#=lXL-9Qz$iv|6QdaVEgtfE1%GU;Kn+@
z!7_|Zx;cH}$z<$=48QRy3I3P9>`{@4^QTjyTugbF$AX=O^0Uk0dnO&W=b<-SL<m`L
zl1Sg1JD#YQ=shP%E}fmswOxwM#al7{*3Vp;@zd#<CTqjlW)l{K&=JjX)OE%d&N7Dg
zIXBzfwERwoIGJvI%e}{oDI8@;+7{t6mR5~s8Sj>ar}&N+eYEf}t;_I!Q>?5;jegxI
zg_`<KIP>cD?WHu-+tq_p<++)kRPipUNL#FpzxEn#XAj-8QgT?cRm+9=<XKW=uustr
zjy?5#yhvI1$pu-IZkPICe|WTOKFi6pSiSjnhT8N%?7&Sm0(7y{pwJ<NXd1d;&C@-l
zEU>z;W@VG_rM01M(m8EiJ~@G#Oc*SPm5ng@i44ysc7vq@CX^&RK3RHt5G*gh<uNE|
z2q+7RbB&m`j^-GWU8wqMpBmvE+vx1vx}(<Qv=KPs6St{au>ZNPx`#fAX7WvLY2&fA
zN{$u}-%0NZZy&ED>7z{s`T@JZ-?_Y!4*8u+n`w}f>Ed^{&qink=fo`#3#C5ED2{2^
z)1oET<wyMi#VXt&+s8hQMttF5Z~)JGfe=}^PBVx(kNRvwmADaa&7Lfk4_c055%ZmB
zU##hx7A7Bij_7WnbUrqdN0z%*ClD*5gE?>S8Qax=c8{ZX38z>k9^9U2$(TEZ=Ap?=
z6{JDC&-p3IZ6IZKUm>-t{7xAe`b<LxITtJiQPaG^-@5#6JEVYp59q=vsX(Ey4-<El
zWKMnEr0{n78tk@qymcdlvf74j_C=M`rm$Ff^Mn4*U5%dZE(+(E=WNn);bFs1`ft?h
zJEXOD2K})2Q6==A3nCn7TsIDtF>;gfMir$SDeFKLr6y5&4+GsEUr_j<tb@6atHpEa
znxC~GB2GA%dAc!48(>gE=Di`D(Y9r{81V`m*PA#SpC$e#x8wM1YXAB6A!iku!$)jB
z*<KydSGKqg_tk_p%g#paXJ~pdSBx4kXsQigL$Y4}alR-wBsOV#{gl9)U*j946mNDk
zzP;C=%a|)s9@5`kXC-=L)_m!P6WL>`1Kg*(iYu4i@GYvfscr@y(#+ByKG-`kxM1KL
zV4*5Z!>2kmVN8OSX%Ahml~doc?bLA;9H#Yfv1ZfyTK~%DeCU$Y4oeur!pzCIg(Nlv
zW>9|9ka%G)Uj4O$U1`~yVkc8tOEWE1Pn|vEh@m!@Z5r`OgZQ+FfyuBn_cE{1QE%gi
zYp5_64U^7_%1i}z|0R~xI89atiJkjc0WHN%A7qlq!{Xp~5MH~Kxi`MG23sqKMRbQM
zBFkgSD-3Sv!~6|2$<61DTovi4{b%8;MgwKxu?As1D)Ng)kQPs&*|{0ag&k9qPQsbw
ziap*7r^Jd~!cr&3alL)sJMm|4x0A!x^~-$~W}O34`}n_#q{Q|u3Mhs3m#q5sr_=kE
z53ZhHAn%tMiQJBHIx%@de9XXj2)5u`<fBx+HfiYP+-@{977KOnRNC6FY;}?Rhz$Kt
zc~omK`A!%P$;v#SBkW@7l4<5`{nk40`Hsge&EZ6&9KR)<#l*q)1B-L>$<qUqcQK6g
z@1N-hl~}9p`HIr5>qXfZ)$=*TgvUn-pG)T4{{Bxl>5YUsO2xug$?6=wc?o`;tiQB;
zO=cZc=Fl;3^Gd^$TS6oJLrL49x#eWQeCp{5w0i^3ZdGrv)sH8_mg^Y<Pl2g!JRMM;
z#_&zed<C8>_0h>Rbns%ZbJpy|UwtjIV-|YU*UYSl&&Lm*_PgCS>QzB*e>jPpO)f9<
zFg+Jc*4$)iJ!R>%-Xj=}5>SeUW9)YD2eW)i9L%gSAI?vr`DlXA*hMn!B(B+CU|4ss
zuwIQ*F1HA$e=!r1Gcm4+HA6aYfnR1wUhQyczE<?*xQLqCpo^}%ms?PE*S3y*bY%Iq
z4Xbav&vL`pxg`zcLCOe&M+@H~+P<SjZfQFxcQJl@+8Zs#q-ZCT=W2RzO;|LqGpu0Z
z?!RQ<LIS3}>NW)S_ZB1gy^LellVU0O5IMj9qs#o-kd%spMGD{Lubwnv^v~QGx07KQ
zumYP(B_5fNH#0mKi%<@(6?*NZ&gj0MEbS~u;5d6)kYRX#ps>e>W!187X2`_x=S#j*
zPl-*~Rx8-`8xI@Iij~$09znmd6NL7t?Ffo!KpebSoHZ(hqqo0He0^2NeU#)2*pvma
zi7(qYiha|G&DpZxE<XuNJmodG4D;)<Nc^q2CI-*!&r9Y9OzrP!kB6#(gCpb5d>Ocg
z*C`VkX|hn&>Qxysq1%Ly%cqsY%sv^SYAH~m=c&TBn#odcN_|cG((6heMJ8uvf@MRt
z)nMc0Elsvn8Rg*m^FK=<JEcfGqWihN<6o~uA5b`R&#t^Xc9iXHY_>wpd#v4Po)imu
zwZC64yD>$Qg%V6zYJSQ7SB{S2VZERG=maB0!p;Zj9W~S0CrvLYVf#suYK0rdDQgD<
zD@ToVRQQ;xuX_jgDvFldo6<rr{C>x;shP4mJF-b-p*#P^RF2u7iZMD15-gJ($n?dx
zc+C&~%Z-sH3xQ!Pp3~+6J#13cg)oe=-^6;@LDNa9>Sw;;+p*i@YR%>XzVPcF+(4Wb
zhW~YreLs?p6EeLnV|Vyq$T$^q;^sK%QT015?LR<bFh;(+0cu{Dq3hqqFP}c@DeKk&
oT`V4(g@Lj6-{Z0WkDtaIlGjdEgwVco1T}&vE2zrn$eQ^74?(G%Z2$lO

diff --git a/16-bayes-mcmc_files/figure-html/unnamed-chunk-19-1.png b/16-bayes-mcmc_files/figure-html/unnamed-chunk-19-1.png
index 6489ec5b414c9d89c6780514cf4e36c9750ba7ad..101036199629352b9548131a197ef08f30e7fce8 100644
GIT binary patch
literal 10179
zcmdsdc|4SD-?u?1YnI3!7b&8WeM==pC@siditO3OHfBl@!quv5lO(cBjCE!-No5(b
zjC}@avW-EuF=pO#n6B!6pXdHO*YiB@eSbdhAIy24a~!{YJHF?{owhXP-6p<`g@uLp
z<cVX}EG%qj78cgSTx`ITU@W~6xEVct>YNF13j%>ySy|cG*x1?Gw`|$M!NI}F$;rjV
zwRP*(ZQHhSb93|X@bL2T^6~NU^YaS`2yEZJT~JU^NJvOnSXe|vWXFyjqN1W=Vq!aY
z?i3dnmynQ<l$4Z`l9G{;k(HI*wQHB0oSeM8{2zb(p`f6ksHnJm_iiO6r9FH0?A^Op
zSy@>{MP=W<eX6ReYHDix_wPS&;K0Fy2M--Oq@kgqt*x!2qoc2{Z(v|xXlQ6;WOVrO
zVPj+CW5<q}n3$NFnw~gu;^fJbW@cvQ=H{nPowBg7u(Y(ava&jT`t+GIXRNKQ&z?Pd
z?%cWa=g(ibaKXmL#@5!>&d%=Q#f$d#_6`mXj*gB_PEO9w&Mq!4uCA_^E?v5O`Ldgv
z+m$O<u3o+B?(Xj4;c@NSHBV1ZFE1}|Z*LzTA75WzKR-WzfB);(uLlGK1O^5M1qB5M
z2Zw}&fWhF<(9p23u<-Ekh=_>D$VdnT5)~B%g+gI4*o_-EZr;2Z9UXn^)~(yOZ^y*M
z#Ky+nxpOBjE-pSkJ|Q9D?%lhIiHS)`Ny*8{fByOBy?giW-@pIh!GnhnA3l2Y=<(yn
zDJdyWo;-Q_^l55pYFb)adU|?BMn-04CL9ib_Uu_!R@U?9&$F|$U%YsclaurE<;&dM
z+`PQJSFc{>=jRs`6ciQ~78Mm07Z;b5l$4g1mX(#ge*L<<y!_3ZHx(5Xm6eqU1fr^{
zs=B(mrlzL0wzjUWuD-q=iA17MC^Q<~(9rPq?OO~6^X}cd#>U3?@835yHGTN-;p4}T
zpFVwRZf<UAX=!b3ZEI`8VzKS*?HwH*pFe-@?Ck97>gw+9?&;~l;c&gZy?uRs{r&v|
z0|SGDgF{0@!^6XPJf1)xjE;_ujg5Wz@@0H{{Oi}RL?ZFqw{H^@6O)sZQ&UsZ)6+9E
zGb9pec6OFbCeO{y&Cky-EG&Hg{(W(AacOC3d3kwdWo31Bl|rGct*ue1)b;gs8jVJ$
z)BA2_*a1<#)&GQDAPb9BEaMN9JU(}Yg+-R-<T2xOq3JV2VKB*YNbcO*kh}GYab+2;
zB+78bfhhy;CLK-J<iZ6$d0rtCS>c8!7*;k3Zk0WM95gR2t9xB_$?w#AyGy4&ybbW+
zRg}0~w!#;}dP3HH*I;a3roG4U&nFh?A<GU1>gVyr`NasRV<(o#?>U^>Q;?6CJhOaA
z+(H1PRD`CmuzWE^ACB9;|MVqxo`+%}HihHA-#%6+5W(L1GjFFbPaD1@vaP52W%z2c
zoPHAtf|lTlbPI49l?Y8PwE*9YK>PQLEQdz|w&>Tr9}3V;rCtX2oC|)_?ZcNXs#%k9
zPm)E+p%@0PHTO!(W-F!$*I|o_3^2&SfrWT+7A2=*m|qC48!<2lIdkY+NEfI5psI#P
zW7)@#M~hn9-(_K(a^b1z-*xHR@ts!Ei#qf(z-KtJ&a}?bNJmG(Fin|&3K5kq@Z~E4
z^W&U$&lb<t6)l<Rgtf<KyK49M^H|(X3d@N4gC}`4o$k^|obI@>|2u_9yw~pK5{n_j
zu&dcwHuaV=F__WvYSzq}4>O^(_Ok<n<K1fk(F14k)dN1((Z@S?o>yx)(c}E!<A*%l
z1hDT8c9s8JWSgPE_Lv8TL?dC#)%2*nJi8;?j2!3mAR2`CBlmM}M$R6w@($E+E;K+a
z*Sb%xNe4)DKRctq{%7-jwUqfSRA|&d5xu^1s7B_z5h-|WYE3$j&GRt&FesjD1q7xX
z?QTV^h07#9S0}>V)y_m87aHrkdwb;cd*}PDe-bA!{w+kRya>EJ*#^Zb23Y3F>;%RY
zzL0ElybjPOL3Q^2GB@BAf^?pr9~hqWJi&3jB=8*SbNvzCiw$YE;(nL*)WubF<prw8
zQVXZ_cK8JH_i%!%B;*cwYyo+(7YdNAV=aT#p*r(>z*lGJOKs~`BN1KCMA!?rmmWbk
zJA6ah;fEa}eCnEtO6qqSRMyPG#NXV6X|P`43Vk|TMO#ZgZb~8<ES61rJQy&PvhcZ%
zJ08%^%G$$ayNfVRD{NYX)ZrF~&r8%*wG~yr!DeU`bMBLMNrZ{8Ce3vg7+vqbt(Tdx
zlgAO|+Kna@v~sL&FFlUl21=9PSfgo*76&P_dnavh?3LUQxm0I^yi4GwkYBlM<y~&W
zT7Tb&_=8Y!GJwtjs0a{jW&HbJp!(ky=ij`s%HQ;$u5gSzFXkPryiL!~?a;EC#bs?C
z{<U=aVIW~l-($$nu&cRmPr^H`Vn(X4tI0eTp&5oevEs3&da?ww-uv%h_GvqmZhYk9
zTJb-Aw%eT$<9Z2gN6gSV_?dcOU!EbG%)Ac$7(gr53yU@+!oTaSH%zLR0y)GFP(o68
z2DPVexf2Rk#BImCTyB@#_y>pmn=SNrV(Wh@s;UpVx>*s3Sq)a-n7<v_zj4=YLm1Ux
zU9W_w%UBI-x1&~jCay+PH!~t7jaC?#yIh;;*A0zMNVg{rKE7_?0oqLa0Ik{^yOX2_
z#x1!EqKkY3#p+8tX6}A@wuPDbr;27EuPW>Nhonr_2K)Qfr<>~yO7e`AnWU?+j_2`w
z2DC_F-WsXJh8+lV-iyW?xmwpELZaPnxPYid)^8T+ZJX8LO562h!*ybcU+pWJZunny
z_rhjss+GSV3P~-Pq=y8z&t|@wjr`zq?yqWy*&l&!jf&37x|;0q@w8adYC81AXyDS^
z$Rr$YNR(MHfM`z#(&j$h!2uPsW885**0C5CkugB+Z6KniFBsTJOqa|?w(#Z|X9JNu
z4MHrffif<?hQVyAxu>8yW2kDZX~)wAY<Wr^2*kEUNsdrK8m5Gl=<SYI4T+eF9wKei
z@Cf=N`>3n-=lw0x{^fWVw?k9Au2@5O!+VKT&b@~2M%Iw+;C3SQ?8po3#BPlLAdzYs
zqgZF@yiO^*9{L6bQwxFG<B_A`$tDAX(TfdIq?i*tzm({f#d0h<xqCQ6JP6m#IWTD8
z+5W?BnL(mB_=l3Bb^c)k-}Q5gV`beC&Yv~f@oTYk0w4Z!V1ILWgCYW=o7z8GocPm|
z2H=`0omyPR(qSL+s4n8`XG8FCl{6p`qljz2Z%=6>)w&SP%cG~X3dO}DEBa)O{tNZ>
zw<!EqCePn*1;2~x&<xU2P6PT+4_DRBf3>3iBZ}|KcXDI0Vq9;4ux;9hgqH_)Tvg&}
z0)`4fN;)L6ibBlAt~Nzl$?&bEL+cKX22=JI9^y-(y1EOJdp$I}-Am0drD&MDd*Qj!
zsDZcV-uk~IQZZTBKGRrz8QKZ$X>d4dA^8H;61DPG#&Q1Aos@Qsx?2}Qw-yEC&A%1?
zhrRN51O1IA`!C$MAnITKI>8t%2r_3sg)Z9VQpCoF{F)lmkboO>_y)SNA<B@Le|uGg
zA35m=zmguqw-Xc?-NrvMi}RrrqwEFQp^c*K&kNM~497KXnOzDEy~e0MW+_{!7?)~n
zxPILshJkXLS))kv`Lb1!4elXf`Zed64>S7)?C9V)M&i^QAS_^)167{B$v(?!c>`9#
z$5SOQ;C~U|aScr;+br)R``fl|-*|8ZjgP=SM`YDY+N!gs?*J(eiILAXak4u|vOear
z{Uk(=k+ofKZG!4zvG+8L(bu+@eoh3zCBIX2o~UqX?*o|aCtNP!C(Q*N+PTTX6)j{;
zmbWr}00QNjpv^=%BEQ`tpdxfrd48<-kkt-*i|ueb%p{i#Rni`xyDDT<PnmzpujnWF
z4IeobRBn9Xhd^4i`T7Pgs1IS8h}^Y@bRbqWNPP|SqicM)3+t3~$&p}lqVP6cPhO2r
z;|(nskUXyD9i+D?tIlef0DI)ivvNz}lJxw8%wToUY@nAHwp49-sqQ&P-=-NZqgxY;
zoT0O)(9yo85G<o*yp|`tVGN4eHr266Eg>?aX%H*>mB+srhinMOE;(knoa3QM;PthF
z;CvA$dBrVqFNyot96fbSn<rUVcSMj;eBqf~n3M1YnQgg|mPR5mLcsV@ypn+w_aF2B
zc-A-)ja4c8zM%FUZlodVk4jtpe$+Qx{80X_Yfm`RC%JA^rkxV<e*n(Pg_bPOW>>=^
zhG>+wra_pv!2Z)PIaXR~twJjQO3Ny-&%EndN3VzYOHi)v+c)`8mqdmDe<j}?=4>8w
zG%A;0IT54aXZnsa5|>^QMKNgbLY96pxWC(Kz3clCv}kSpVtwt9i07Xf*U({n+S~mu
zove#fs8~_#>3S)<vd~y79mJh^r?Q%JDq_}O>)53~r8XR$!YL`WmfV0<b0WmH71_^H
z+6<PTdbB&*;A>llgyNV+8<O@9G8fPnL_n7mzDpyb@@sb54}lLetqr7v4Bt)S*b<gQ
z(kEabH~ZB-(nSNNgQwV#%AsecY1J@i2V##yKe0^$7u~WMR0afj)Q(8)C80o&=~0}Y
zt;YTp&tC#)X2Ex0^`306@R}@sbHf2bBQ$YLG=?f^C$t%6(&Vi&5k1%&8b_I7i5ev2
z!7-jm%;+L<ZF=4N*2V*KwoF?8Dt)A$K8?ff)E^yrV!ewIFcWRCS^Y4QMd31jd@Rpe
zcH`M67`SIRibl?-Ia8`9Il<vN#V>!z0V&}gew|Ijoab{|=phuTt!Nj&JjG`Dc<dCH
zt(?mqE3`y34hklX&7urf+&Y>b#s|OJ8j!vfC^ahMTUTBXsJ9#56WAK1&zSY(D1NxE
z$gZ-z^x<xT))uXOJidIRQ3s=;zkLDi5+maTks0k;Q`SOXw1t>?z81PQL#gShK<P2G
zHBzfAiB21AGSWA!D*blZtg)txU8n69Oo8?BJUewdu3@0KgwFl>ffn@$Ssgk%R+ai%
zRl_(*eV0odEXM(QZ>&w0Yna-MTF^tyf1TvSg#zjpnp^zIR0;2#Q#bXZ8f6xCzr*>R
z9HDE&YS|M;dc=od<if*1WPb*}<xiDOX$0|>wuO@JM>eXf!;h;~R>OeK*{Gt*bz3^z
zQWSdXm2FC@EK~ks7~F6VS8*|fY;H_0XpTaHBYycd6PBC5o*C*jlWi1kBM{m96<<;Q
zQ+$1pvnad0J06mWl;c7=P|xJJIC|b>@(*LBbrX!#crIx0^?&%yC_OIzuKs7qnU|84
z6n?YoT*C}^wBAiE;fP~<?0$-v3ByZyJBfMo4xgXip5yi_3u!LeM=JlZ@DVu26u04v
zpkFL!h2}?TYOmU>%S6o{ZIF;wrR7<d{*;|z`f_7|NdKY7*!0DYNlxea00)a-xFS(z
zt46~ImS=A5j_PRoO*I|VQ`dkdD1``wBTd4-yJYFpPX#Zo_`Z;jN-o^`Q?=27tqG4d
z9U2Gw=T`%NtZu||9PIC2y<%g_v~1M<P<S8f29(qndc;V+{}}Z!2trSN?}kalnrCi1
z#aevpnl0zk^Z5^0FRJ%}Kr&1kFDN{bN<%tC^)quiYa?^}^}hwuPNF(xcre$20xg|N
z)`f`iX#fL_{7fXT!;>%SBVI0XB{y>TsN&jEQ?lh|h8x);N6<X^|Lv6vFp6vkiU2U6
z$OGwI#6T_tLG}#DLxHvOIB<+IXZKD7#*D{Yz2d;!0#q9f#xx`M|7Ih})}%<7fwH{Q
zRinpW+_P~k=+whHLE0WAV|jiV7J9-~Q~VqsGvJn#HOrDfHb^mXY7tM6BlhSZt~l@?
zEN^Q@IW!AK1LL*R)i+olnfRBp(-UDYbo(bYV>`GdhT2Z>{kG_<CfIKM%G|Tsz41%c
z6&3EiDG?D)XQHPI7v|Nl`R{fyIIX7FB9KGGw*2#T*@nHNA+L->7=pN;9~IcZsKR=(
z7iS*$LS};JNfZKeeE4NW9FkOg=Q(47C}jYB4B91+D*TQCfY9!+(24>>AqH}n9Xog&
zIOMs=@lMrI%`15J+`VPeD(t1bF`IvlpfKiaArVw(tZUBosxSPtRFKtBP=0Mi%?yC*
zjWNar#zbLD9CyyY!Xj?5LzoQ42COf48)^sA!j=RzWC`Q?&Yp$G8rk(jDgxo~u&Dap
znIb<h4k>FXeMmPwKG=ztV;}ofaW8|dJ`X=79r}6pJLkqxGBy@CO3K6>O~hoIu-8GO
zYhCcuO^kEoj1-AO<3k(ZxPT%E&+t3zP$93UHdrvOtz`9CZ837_E9oDsSzSA)G570K
z!0?%VfnY4dXJ1*DKXll2cl<m?%BlnAXs%E>z?jo=URnhQwiWUH7!C@c*faCLuW$-T
zU`>iC#+Q1!_%Es4=d`l}EP@QSO9IGiUdl*;{$H`ydk4^5q{5nk)NKctDSPGsyj<h~
z588B@F<{8!43FhKk^_PvkzwF7$|ivCB+A;FPrqkAwQ*A#W3E*``0zTDg2>KKRZP0^
zJXahTkqvM&i8)xj8Sgc6M|ax9Z%F2NFY7LoZ1IPG6WQSBD@=8oKF5tR1(P!09q-JP
zTqQ#(im6&z(Esk@oY93NjI+>OVVaT%T$BOvCb<tCd$m$002C{K{V7GWM&fM-ayw?H
zb!M-j8aq&!B(XJd)^lOIfI3D9?Xf82+2X(j6e{h^#Uj)}VSnax1ntvlG;=GSqI?tt
zP_H0yhhEa>owA2sIZ@Ho?=P0LW531o1NhEZ3;^hq++|>oj1d~z`;=W@_kIk6;ERi%
zK6w`pgF62ZUjh)HFM+gs@%V`XkQQG`o!F?t_iZ-l+XRI-JrK35&NAepO$@8*rjV2#
z=bhWCn0f#ll?Y5#{nj64!MGW?RZ}^zp{9$eHzc|w=oM2#3WxKAwKG#KUgYN-B$JRt
zWR5w6siM}3SqkL`QSu|z#UJVtlpQC3FlhLd0k|mV2Q!)f?E3N^E74S-B^)=_ck}?7
zf!LP0pscMx8xVfid0P?a9A^an?gC^*$OeU#p8gFABlcb!EMokMHd$P?{(GMpxpPN^
zfL``ZKXFr5;<dVVpkwVfd$F>12gp~vS?Osi<QR~|ehRiqG($Ncvk`*2=EIxc?*FC|
zHY*R51_<U^xo{|_NJO@vf1T#AgwhSx*=Lvhx<n+&%y(05%LxOae$QH7<BfIvi0sN*
z1oCF%(Z`kej_qUdm^|)S4e|?4!+g7)=<8nW>;q|qg#i!CHjktAjN|BiVv4=9x=Uuw
z85{47gJ^Bz%%_FsqLQ8M4sUL<4Nr#P5Qy;zULVLc*w^O(a}Go<a%e_#AMIV)xk5GG
zc8A5|QsEjyycuUT2qt4oTLuBN(oZHpLKOO3@TDuTYsZV4roboO&ggnyC}i!4#0R1M
zuq*vyWB^}8r>M~00Vs~VtL?`X^VhFJGYxr60q92(fPR8LEQ}<tWEx5?>Cip2zkn%?
zTJ)@?Es<pd_IOf=%^S^*j{TDR2%+~Afp=ZVaBe9tS{{|}LS&n?6c^@btIQ)Ax?HT=
zvSN?svI`_4wl=FGjxk=`hAl7<Md00=L6OC=4TLrw_oMvdMp*?nUW^bcKl;FvURG9l
z*{8>D0)q%qT(w5ZQuV?-MC&v2g${MIu$PW6{7+xc{8Qw}R9OUdJF$YYZr$&ZGpa?{
z1ur)(xv)2itypqSXMM!1(GQK`thaQ2b@vYTv78Ib9G95V0agyPyRaLA@-ENb{qtL0
zEy70$PKp*@cPPj?HawizSAeaapL_3!sIkM^OP%v9i@liHZG>r!xD@4X_n1zjrCc>K
zdvLgkL(=KXpf0iUu2vr|3)wj}co;sl8jeAFnFVl~%*9@O-tEV=GO?O*#g5kMt4TPw
zHOL!nKlF8fxNG)6pPRG0_)OqH?DGk|h;V7U-s-+})JNSCyrjdOs-yd`lco>Kqef{{
zZ`%S?=mpWfs*LTPXa@8ybC40Ya?8Pa1A|u<ZFHhZ`L{D9248f~xrqekDO_CFi8U&#
z9D(a9Vdm+JIl0x<uug^cS0H`>n4bYq7Shydc;IpmVr|fFVU2Gw{aUcLU89s|+y!V`
zD`|i`+FFI4=$?N4IMy3IA^5EehNu}@oblmRlO0_O&Qr#Qzd=Gi7s2t%g|%JZJPs!4
zbqN;H3f;pI8h6dgsqQTt>8dv@c|#d^J}D&a3zBa+7~Dv0f{qo}tiPpMdDA&DV<-2J
zPB$Yc_7Rz$Vvnw_5?L)%4(D?G$eeLx+Y;Gx`5V>6X0agcJZj2eb|}X5q2*xKW8X!3
zn~$I5<DYw`mT-vVYEX5_0}ICZ73{hqf>`1-ea;4*w1mj5;(}rTa}G?J(#N}q;;}h)
z0F<#|4Y&^7v{fO%nRAU~S4>C7%$+sbIyInn2&q2wbd;~ctw3fS?O5;OkU9?M)^=pg
z5GObK`ax=IM-o0%L8QE<J&A@WBa(XZ<MBP-b6Jl)7Yk@;1L0cKSPE2?5XA7+k8!NJ
z<m?Qd0F}%6&cebT!T2qJYHtd84<bLCzSlVV{n34$lQuK6cF;pKp$sH75R_~F>@1r!
zGe-k}h+K>W&)kQL&QBKmwG3(}S10f$DJAqC-PbXR8yw{jO>AbEoem`&VsMQZq`aEE
z>K%*+OmcLno*MkMESS`^$nWd#?724QCKSPlkkrI-HBI6}_aB!33;5~pCrH$0+E$ii
zI})UIbd^)}s3`v_>7?f$(;_TwjHX`7s~C2DE{ZzexY$P5&%M8o?!X9=-oN0X0_cDV
zycW`jwA6%5Qo)#XiB>7vcbg17$g`#l+T|&XmnUE?pG7Yc7@kV6a%(n{@y}=1)g*X)
zqwp4RaJWYWj7@i__2WJ|1snjGcHpTa6YFq_5Xxvzg=}K%#merI)k~NZw1Pct6?Rsk
zy&A;k5&vIG^Nk`vy1JVk>IUKR;01xRiBZ}M2EnbYnGuq@JAnJ<Xhb*SuUmOmrvqG7
z65Gk~#^|9tFdH6Lk|B{Z7}vs1ZN*cY{8myVGE>(FZ_J>mkA-Nht45C0Dy15Foin{r
z$dXo6J4!FQ@?;l*)G0=Ol;+OSM1Mwmg^FluAy;8l!v{wy8Vd30l?EXy0CkTMh3_aX
z!<BKFp#cf@*@8r8odaAw)Q`?ilm%mQ_4_7ADfU_V+YGS7I!$)vdG`T%hPbz3(PtGs
zWeNfCt{}~dqolS#Gm`Rxkcd5`PjYPFoTMjB^wvg8CybfMv67B+gHTt$5=93^*y)J=
znW^YLsKqKkZ2fkVMg4-5x>1LGzW}wMQbt91V==w<sT5R#o&Q*e3}?e&au>vzEODl#
zwp~V^z;_goco!iVI~`gjO;%_aAWC2NT;HAul8K<{IZtqNRbd4DwJvgnF1eSsvQzDi
zqj87yRr@1GP>`_}UsjR|`(WbZb(5s0Fr$7ke}A`;>u{)#Kpjoqi$*SKQ{14hw2Oe}
zi&9v+zH9>zrGsahE*>We`N!@e_&yXM%dn%gmoZbhxyh>fq!vP=8g`nd4FZb+f^P-k
zMcBcucQJ4uD*i++yG(i?Wcn!kpex`>N9lB1yz17KUejoV{%Hy6fQIW<0D*J@@0!=Z
zNA8;hH>RoJTa#)p&n7gpCLy5?Wc4W&H8Dfx$!dgTU~;WG;3m=JY2;FUL)uN;piR6C
zc}N2wKH~|%H8p*(Y`=QQ5`7mWpks{V>ax?4Q2~8ML(f~q6zw>X50;`YrCkP&LuzX{
zqvdj$zEbeq*mFJlx?VDL8J*qHu`o=1pKvNk?C&+Of8O4h-%iunGHj|GSdIy$0KZdW
OIcZ{ftoVq_&Hn=reaOZD

literal 10040
zcmeHtcT`hbw=ba!gd!?UiXh+-1r!97CZeKpC}IJm1O=s&NJ}UoK?M%dL<Nq3AR?er
zEl7s|(nLB0q$SiKB?&c@kh~qr`8?md_m2C=ci*`0j{8S4_L_UI+178)HP_rb+|<|r
z%q7aj#KZ)?eCfOy6B7%DiHZ3vCksIFX`|NxgHhApuU`O$AP|U|nVE%!g_V_+jg5_+
zot=Y&gOihU&z?P8TwL7T+&nxyU@(}MmzR%^kDs4kKtN#c-o1i?g8TOE+rNMRfddDG
zgoK2Jg+)X}L`6l##KaCBJSZ+ME-5J~B_(y}&>?AQX&D)r!-o&c%F4>g$;r#hA31X5
zx8Hs{di1D*g2J(5$BrLAuBfP}q@<**tbF3ciIXQ!s;a7<I(14zL*w-6(^^_u+S=MW
zIy$<#x@XUx)zi~EfByW13l|Iw3@%-|bouh-D_5=<8XEro`|n0ZM#jd*CMG7Prlwb~
zUNti_yLRo`_3PJf+_-V`=1p^Ra|;U#OH0dJw{BTkSy@|K+t}FH+S)=O5IZ}&+qZAu
zxpT+f-rm8%;qKkL_wL<ubaZrba&mTdc5!iWb#--fb8~lh_wew5LZL7i?Ed}xo}Qjw
zUS8hb-abA)zP`S2I2?gM`1$$y`}+q31Ox^KB9X|Tpdb_qg+`-;gM%MDc<}Jy!$*%E
zg@lBJhK7cPh5hlzAK~HQj~_pdh=_=cjEstkijIzsiHUjg<VkF7?9-=DpFMl_{Q2{^
zxVZTE_=JRnKmYvm#fuk-iHS)`Ny*8{FJHb)Nl8gfO?~z1Ra#ow>({T-)6?I)d6SWm
zk(rs9m6i4O?b~<n-o1bSK07-*CnqO2H#aXYFF!xOprGKxhYue=e*E<5)925h3kwUs
zeECvTR8(ACTvAd}T3T9GR#skKUQtm|Sy@?CRfWM|s;jGOYHDh0YrlT|T31(BUtizQ
z(9qb}*wobY?c2BJ=H~C;zqhosw6?akwY9akw|8`Obar-jb#--jclY%4V6oWV-rl~x
zz8^n+^!N7<3=9ko4h{_s4G$0Fa5y|3KRP-(Ha0dsK0YxqF*!L&AP}aerhfkXIXyi+
zGcz+gJ3BWwM<fzSBodiSo}ZszSXfwGTwGdOT3%jWSy`b_D66ZhYin!k>+2gE8&oQF
zb90kMqtWT~9}iwy0w~`DyJYFb#3UZZ_<>?3=I=5wNikhMuXi1hJU4{2J2-*F&mX#c
zsH9!(@Xux9LQn#~BIW6upUs7XhDyjsZfT{Hk*=mp4g7dfeYXVc$dNtwTTeiZ^3txv
zgql*m2hX#eBc1ymOF82V8qm-_LSj{U>elq}_`pX3eLy_-P26XdDx6B0N|neQWzpbX
zYKMzs%8Rv+*@bFc(oAvtEl5o0Sx(`%%FOIXBhkTJGIp<OPO?)+Cr=R|_pLNeY4|ss
zrI+SD>ts5cb(ed!tiq$|LIshky{w1G@KQ*&(v8<)YQO)2S;BIXs3Rf1X%Fu>@BX<u
zmpeTHTYUWLG}GCJyIhE@vdl`2(|8w?>W#sLb(L9#3hn$%Fq5}!CYt0~RNov$Ogkd#
zoitYMUnJws^lI@W8x=LUKg%V;USY_7xwlDkEN);r|M@cxrZY;{FoNc#ElLsr_tMMq
zu=7WYG;K?|a*f28rpnz&=Op{5rYlWReAzf{q5Odf>>GxZFK9XiS{7FY54M`G?VqzP
z(zPw>1-}6&8Xa8ra&Vh{LMU22lT2Z+Z>Z-bl!xp4L0_qbB^SXQW-kqkHEx=Tw!v8z
zgR-TsVPcB8eR~ezre~1Sj}?>sQYy~4x4Ru-ebUsy%c_N!2v49Cn_TF)p-A(I?l&pz
z%>pOBBzqs@;ad%g$@(yotgb8(8s;_pm7E4#?hS2N9IELcTU3g@wYkCpT!(kWT1(O#
zHb9czL|EXMUoZS<|CB@&2c>i%{kfi|F*e#iAVx}5vdNFd-~hGg(^x1Up*)!7HsUDm
z_(47|TE#$<^1&83cSjn_X{cTFF_5HYo6)A{xC*gmy=ujno=#?Qe(1|zzy)gNEGBYM
zq7Z)On}zcpMIUMRmZmx@=RfOda$1C-@3LA`+<WQdrm3%+6@<CmC~T?GvyQL+aKq<k
zoLE5F`~{I{dlYtTT?aekU)0N?ln-mp`{|VsmBH747IT#Km<;|m-IAYU1jvWAwRhEI
z_w0Bc9*KDN9|BvV`<S;`9N*UDzhd~E8{zj<O-&EO#hRms;r_1^^xv6tv7j;H%z;pj
z#L#fihK|Rx4<l-omiAga0BFKT^Z|q%=7}Mcg`|&s2vj#Md(Jyq79yA~Q=&s{+RLi7
z?~CjS{EEbZw`jloF$8-`jY<J?8%ETI%rhI28hm3_#G<|pb6^`SZo%jqoE8`VceoWh
z{|aw_aqHb5<HRUNLpr+&SW1)m<{>)~P=lQA*dW_x9wF(M9Q;>Q!b^g)j9>PTwDP$=
z1bf&BSA4E_4>#j*m$KW>94`PH`noWjicL^2*+_RvwX#KpA~T8zJ->wgld==Gyt-?5
z_JK=V=Its5V%3~Dwxe|>56;)dt5*DS-qUI`&U4|FIBU*1%t_|Fvaj0nv#F4x`PNmR
zrN7d#swUA{nukvnEXcBg9F(({aS!}dpQe1U%}=m4%ySrXNm!y8&TUW={ulPf(${RS
z$ZPVGPZrGuns?E2`(H7V_#h*RCxS!VFY^>vgs~m98=fit9Yi@(<UxuRsv3_kH<g@?
zF(>-4II}RbpLrz+Qo3*cRrfvBoKkY66gKI%GW}Wa)4YCAeGRPG`A^{jLamyna(BIY
zwwh_cT>PS3DDQYhu=005?(`8L6aNYSZQ}M}9QYvVET(zr;k;)|+2O}|o&lrP#&Bf+
zEAf8y4||)E@J7ejhQD^c#UxpnmsEu(<d+U#>$pg9(kkv9lE~oc*V|5lr30dEn1C6Q
zF}T!ftokD5EWlvLX#xE<`>S3c4L2nb@Ul3;!hv+djruP3xt(>Iy#?=b@}?3~79sDz
z|5o-@vc~%>5`74|W|HOQHOI>`+icY0b#`Hp#e<q3t!yr$em+s`-t#oLy0&u{K33Zj
z3fQ*M^uI$!T>J~Y0h2!{Jijkg-$;gzo;c0}ZlX9h!-N@S1y!9t^fdIg1|9s!FZE}#
zQbstCW`i2rEapa`qU#*Tpre>41f8|_+5zI<{RH(n7KYCXPnL&tz>eX--^-paC_g2!
zM<+<xLCPQ99FhGTnb3ZqpVvY$3B8}w;vrh-f9SgUT*j#522~MjlVy*8B%Ei2&qu{Q
z8cwzq@`!*F8&CgWY(3a05WNi_d1pmdfRU`+<~gY8MOfBqE5~G?TjV@xiql07O<8=j
zlpS~sjp9GH^a3EUg94=OevI<1hRnD;z?WL44qqMTZIQM!VLfI=1)1a?udPxPD3E84
z2UN&57b@Y~!`UYYYA!vDH<MyKG?Rj+)AA(}Z;VT{9FevwnKeFS2WDQ0<0r#$+64Eh
zT@jWEtkmx!AQ1cj$d$h!AM2jS^&kU>SddqFJeV%RIvb9@YqOVpS`Qkl*&)i%$r8>Q
zIK6}qmWIDR#k_JJ^O4B8YRXS7Xng4uh1M1*$kAlh@x4dJjYE;XKYgyLM-QK9ukx6A
zqe{MjNdcT84l-b=J{|Ukq~#lu#HCT*S+c)1SJF@tA+x3SQ*?ulaPr0*@=J(s6k5I5
z=k`*kXpt-uzs=l`nYIq8lt;|vgQ>J7j;FT<CmOUBGMR@7@x7_HAwTE*PPyY_w)8M$
zPN4_mLny*FB@5N>&>TaPh!#c#|JQ9Duapo<amU;`@j6CquocnrWnE!nJXm+jD#bk(
zOHVfLbaNu&2E_CGe_}Ji)GdDtz-BXOJ%U{zN*Q6)r))40NZ>3oqM4xVmG%Z8mf%!9
zg6OvZmfMW*7y-hHccMHzEl^##BN#1@%BPTC|3n82(GU~1>QB_1)b_dj$)3}%{|dHf
z+fc~6TH-f}T+1dBo#gP`jwz1$#$x=+?V$JQE?h{^dbx>lD|=wvfMEX6&pZ>cRTZ#C
zGjG<Yqa8SieI-s7Z!naT^*V~!ftafxt{l1hm!ZQta|)^caB^ZBU(4ono}s;h4K5==
zM~IR53&n<5u71NvZY^yB>!8COcYdm0bH1wWIg+o!3^X!IaZU`7wY@r(==5uJoz=Cm
z+{$)NjaxHnMi&KVhK~Pafo>y380HP<*BW-&^v3*Bc9(Ah)gBYWRvnRxHQ_i2#(i;V
z!r*T^<0H1#N{nS#qWlDX(<yjs$jzcN?^+$k)fXGGi|T-TZUYM=QjJTNkG|~U-U}5q
zCKMX+7fyxAFC>wbjpB*hNRtIsb&fm4mUTt$;6ag%v9u?d6?wO;rLbzP_4Lq*+Vt^W
zB@wNw<3!bMD-+QqO~-Yk9E!V=Cn}MDE6NNpR?|>jl+$Ip(x&xXJ@Nvv{n!pr5V)BK
z>d<I<@weC{+}`quj9+4_Tux(Cv#Y2I?oSS{@y!z1adV(=fxmBo90H}MsUoyPqt@eF
ze^lG(MO1w^W$HpdyPFK&4oMLG;Uv1MKT1;m_xSs2j3YzkEe`5%9%|3aV`FZN^oo?N
z2=}R{WJ2lqNc>|!&I_M+drr27qEB)5na`k8brldZ?(@sOODho0ph*%VvauO*jT&Oi
zSRiBXscO6r;b(=ePvF*GLj11H)weLLn=un~*iqpPBx5jWWhvh*#eFHlf^{|mND6{5
z)iuY2<c(-}azK-L>0&KU(~d+TN%^I?9Yq28^*{Iv*g+dtxK_WutlH>VUr66XB`zK^
za<FlE5cz30D}8e0s5V*_tJ=p}b%ss~5c*o$%_`!kj}c{^&Gx|v76!HT->O<RAn`ol
z`$47mTP7q*?br@P4OXhJ9YOJ1SeIHg3)WO^{@dItQEJ1F>Zg->!8wLTPyXS+csrOc
z%#lod;CoN;7hAEQRDnSi&0=)~5?Gvszc~`OKPO4tjdR$tz4iOuf2iMc1|N)<Y@zDC
zNub>?`1rQsMeW}-h^bK1AS7K=uiB`_FURaJ*8lX2CVPWo$5izA1MTY8O(09L{ZUE>
zO8)l1)1bJ8sw(9A*ds?|TH~hkoqq_9TfH5co<y(JMcR<R;QEx!00@6lJf*km@=evw
z@bdgWuK%*wfywQ=D<W+N+3|1$8o7fuC18vAzn1RT2Q}r8+Ysa|w(s0tp9u6{#NU~9
zUf&UalfU#MGIZomkKq^KsLAr+5oN7p(gEIyir{(WgH1qP&C5erCoj^>9$&`D=!;_Y
z)~}=*35UM7EILF7D|56QSU|1<<wuVF7ANxFTMrZ3EB{S-|L=$zsQK+s6M)f1w-GIk
zr`B!b_X1`uV%HaoLTy`dCj=rWrEEp85_2m=Ydh3fldBLr?)x9fP;8nEk^zo^LJeo~
zsXnV{0iYKXVgJ&cb|w(RID^-HXjMRo>)0`xFryZPxGG*rWuU?(KX)v$E?;^omXa)b
zu)_WoY6yGT8Q%=>xT~oc&Y1WCN;#w6PVle1Df!6!Oi@kAvOTcrVErbPx^7>-$T;X}
z_7T*;2G1zVOIo&K?y#Afsf<$|Kud^x4-kJ+Jl=sJv(%l~zpBAkD)zSnX0Yx|`9i|P
z+c#2x>b*#;wy+YVBw~6$G@4^(DEjl?Zx6vb1m_o=fv(gUKS!;;o2ZOrEwUpZ<Kw}&
zeGD1*AYQ!&9l%jjV%FQWi=CFU_zj}jP|3zP;XZ4KFYQ;kUlpJzG3V*p$=(wE7Vv`Z
z(0x-e(K#|ptNOvvFrd|z2GhcUBf~~6Ra$9y!ZdorbUWUm<%wEt?zzVqwL|$<e-IoZ
z3OK;TM;Lu?QKu{ZG@fKVzApJ5SG;|$1c3q?*F?{T_{iZql9D$DyBB9}c$kzATGE3x
z{Mu*x9)<c`2)*9m7@XGOhRaejD+^p0su>|)sua`M8_EG(f2|&|mJAwBV*SB9T=mVS
zGXH)+dVs3!@Qepg4}8jNG|hZglVrhF9)1IkC@h;;U9xK*5>-+!<>&;ARyg^50ww??
z{@x)RH4#x{b^5AR<aQ~+m>OSu0Z8rE@R9(kgTDJ`tb1al9Jb8rb$77(%wwLrXKYz>
z&SFkIBa}ENU_*CHiMrI&S0oTylE*Fp%`@x8YwGfNr`0bv4B6Sm%uN<B34lo-#Tk%-
z-DjvUw}U{N*fckeNJi`MAcMW;4gk#^jtuPJwNzUF2d^-r^{9c}FY?WI2d;psUsxFU
zQjGrhSYH*%kl2xE|66+Z?$gKaKKI{Y5}grg#<#=!!}F`tyCD*P@{0D3CJkm6-|Z@B
z@$Ja+j^5&i{~s{_ZF$#b!8kMB9fQ=b{<kDY=*zcpf?U1zFj?1gls4=1;5j}npw!v%
zvE$yyv|>r@#@)=VaPBN#@kYwLAwSvl?QP^wCEoY?QPUc?9-w24oGc*g5f7(L6{5No
zQ;l_o&xO^>ta;#phEXG&Tad%R6S?ZAYfn}oOZPUZ)NJ)6RSKe)GpaPa7a|#jkUE3G
zwif)bhyQ%zC0&G35qT6;VuUp5D(8S88MH$NEBVqK9|6@|arHVgWR}V>fian_vTrR_
z04U<{>@(bf9~O0G7}X&dE3|rtfV#-248?Mk$aSssUCH=}|AYY2t?*r8XNdoz>-xZL
z%@PzQWw!0JJ|Ven#KhyZ9Zk)@8*JNKaxT}Nkc>tODEIe%@G{0!athDQ^qrQsgRn*(
zTubp{Ub%oNcmiwP(<cO~Jy)}JgNJ+1_LO6XlmE>JBFzrj>9YR2|Bt5{FI)gNs7gCi
zMe&lij65L3^DX0{xhX*jE5|x$iZ2W%KArI~1pMnL5^*!HckT|?>XbOk=FFFSf%YBi
z$I>^vM_)tEgTi(;s27z3>kh90;(LGy-QAHgNndncOzFHVo)$8}FkU!$kr*$W?ydRG
zJlpe~T<igmA9{F|Ta$bp0Gbzj;{n^E2zfl%!?AoYe(u!g<#!+l?)-r|K{BJAN%VFl
z1dTj@yhGLJ@a~1*tYaG_r9>nEhtm>>FxROZ6L33w&#9E^j!|Z!&&C+z=qYQ*#Jx|L
z9HP>OF>l+fnbOncsDU*uS_Oi`wtDXdp2O!gk{8)s=p!wD+#EG5eo!e&x@t11&_Aab
zoe%UJCxD(~0H60q-w|&+RA*$&czLi4H*Mg@{aWB95);byy8V24cwvXOs1nJisEBfQ
zySKQaJmnFU0315!wE4GNx7xG53i2)y9T~{6zi7xDpCq~T@MU7Wnq=5^`(DS^b|l`W
zi=f!N;{3*PhuO)%f%md358WZ=+k`LMFL>;@|Mf$dr_&KqdXNV2^9HWa1`E8a_;%A_
z5Z{D2x!trZHR@R;VbejyK$e-kT@=K!|03pWh=8)`ZC0M^!vEpW!jEOQk3L$uH^8^K
z&QPgjq-pYol}&BLP0fj{3EIsWRjR!D*t-{TH#GUeuZY2R-fzm~DEw6!VLuZmAKPf;
zt`yzLO8X7(=y>zyoYvu^?m%y&RN3D-YCt+iA^#!LqYI|$2Qg_I`=roYLWiryFLJnQ
zX7n&boHq~)nx!?*etZ8i?EGCY0_efh`KVj3L-Q0qFiNKzldsC|`%WB9pfs#&X4lbf
z&Z_>dNt>Uj{gl2lr`<g|0VtZfePmZGDR$+91jR&4cn;xQ_E6IsKnU=BSu@hr*R*i)
zNlp$B6`ErcUowSUa<OrJU?FV*rBq)9N31p}??iu++pRfhow0tcSA|&8MZoT+Z=$=9
zzt(#L(>u7VQbg|`LbUDQl*BDjLN~_J0SSw$Zc~1l&eyjEFq$+IibZlhHf!2ny0uTJ
z^o=B2EtIt`4?~L$Bme=jCbhUyqn27geTTc?;@-KKgn*&5)1@DiiWjwte0*2$7Ew3k
zU(f6M5VYf{+OnN4sBDDjYkJ@?_wJ<Cniur%ke8)-<CR{qGpM)D)bBC1h_$szc&1?v
zpb=U1JPxkjB4Q3Khk`4{yhMB6)3J;131D;olu;|v1C^bxbv45=--$@ONmT8o1`qUj
zB70hE&jD+&TXET#-`g-RX{N{qHXVQ04dquQL3@n#TWBe9@>Zy{JA~V5AKA5bLhIek
zm07Y2ym{U)+cyw7?f+vY?_{40^f-lq45Ej+hNCs$o!2=apu@8hd5kHv5Tz3)xiH<w
z&_QCQg}m+W_Q3ZsWX87ut2BppPn3eM=rmT<cG)Dm-|x~@EIPf=qsu|{!|;%qWFWHP
z8aFm>c!D_p4BWA_k5-e5IQBStxW`K`4Fyv8RI*{!)MctN?jK`>%bWDObh94~*$L+e
z-Gxp9(Xt?KR<12+jsJTFFkP-{bdZfjoA^M|l9bXqxS2cZ{5GHu={f0ps!ji6C;M*y
zs9SQ-)%vs-Hg3qoCIX#4l}{%>pqbDvr=c);$6Ge6>bguRA(@iF+5QqwpR3z3N&oU4
z!9RUg@Xy2Iq?aPp*k#36+N^_7wyV|M4O4H>S){N|X}pKG4H=90jZF)w`}_b1F2iiA
z%2_NXKz^-<8Ttb)1ZF1cWIya{xduXwsnbVh=xgsCMImCirDmwkA|#jAG`Ml0qc&l4
zT`9${73vP%MCJ3Sunr;<St@T2b>kyR4UqMLA``I5#`ok6afpN}b@I(zrM(0*@!}_?
zG`b}@SZR}7JuHQQBB|4sdIhs4w1vb}z6Wh;I8vq*0!JW~Q8FoP)j~NdBF6<_>WN?t
z4T|YzeWw)S!<W_b<kGq25un=wMdbYqBJm7M(ov1kl?qcI`XGk^+f>EKu9-B4>J_Qg
zszMJFK2bWyH$IMPQrj?eR2-VR?s~N2Jsz^J^3x@K%#f2B9;H$^3Vnc&J+eL~^(RfC
zlmrfvahX?KGT2k>0C=T^o%-S3{+K!tM7-R6U4l?YapGO=j=<48A!jkN+brsFkR-0v
zq6c#M&ZUsggvY4O(04to){21R4^X8Wu45#fc{Man1XQ(EsBs$8$scv8*%909Wm%~M
z)%l;xakYi=!}J^UXRkY4m=#3;SM34O^~e!jsQe($h4I-<MV?{ux9;f}0az11O#KQc
zB7om5RWwX@fZBG6z%KdcE#&k=R_l|1$UN@W#|ewfI?J*6mYD9a25VS<*C$HZVZ<hF
zePL=s4;EcVlNJEM(@i+ChRO&NY#;-i=JC}Ragg=(c#6rUM8nc1k^2E3wQfuk>Ssc(
z$DiW6F-h`=vjJLZChA1(b7eh<Ty&2@>92oH-T<p&Z8vmX=mV}_1}y$Smvu&ww8q@w
zg`66qB`XlxU|SHPC#8fELEI=JX^h#hn4Fha3g{BcQ~hE8`9HnG{KxyxuXpK^iPyfq
W4wedQ1U?R8x_rU-eBL>`2mc4xU`HMR

diff --git a/16-bayes-mcmc_files/figure-html/unnamed-chunk-20-1.png b/16-bayes-mcmc_files/figure-html/unnamed-chunk-20-1.png
index cbedbd82976b9acfe8764e497f011ab899e9a95a..81115c8276e4b74629afac4d6bfbe09c1f708903 100644
GIT binary patch
literal 13377
zcmdUWby$?$+Aj<P3_S?aF#}2}0@58yhX@EtjZ)Ix<p6>~DI!Qphm-<>NQ^-!4bnAq
zcZYDE0n~TD-`RVgFV4BH^9R?po>_7K)*Zifulsp;d_zN#gpiI93k!=xS?StMEG!%p
z78Xbbf&={G)^=P6JZ#=j)42{jVq;^2Kp-3(955J+i;Ihghlh`k4}m}k2nYxX35ke^
zh>3|wNJvOYNy*5_pin3|Ir+JB=O`#BC@Cq=pFdAUMMX_bO+!ONOG`^fM@LUj&%nUI
z$jHdV#Kg?Z41>Yoa5xJK3o9$@g$oxhUcAW0#>URh&cVUK$;rvZ#l_9ded*Gr%a<?n
z@bFx@a)p<dmyeH+pP&Eg)vE#m0)m2q!otELA|j%qqGDoV;^N{G5)zV<l2TGq($dm0
zGBUEVvhwos*REZ=e*L<lqN0+LlCrY0ii(P=s;Zirn!38WhK7cwrsj<sH?*|0Zr;49
zt*x!2qob><d+XLMJv}{ref`_FZ{N9d$H2hA(9qDx$jI2(*u=yHfk527d-vYGd#0wQ
zW@cvQ=H?a_7M7Nl_wV1gva+(awzjdcv9-0ev$M0ew|8)GaCCHZa&mHZc6M=badmZd
zb8~ZdclYq{@bvWb^78Wb_V)4d@%8of^Yiof_YVjNc<|r>5{V293=9ei3Jwl_`0!y!
zNJwaC=%YuE9zTBk<jIq#PoIW`g@uQQM?^$KMn*<OMLm1=?D_NO(b3T_Uc88jiFx_*
zWo&G0TwL6%SFhsZ;}a4R5)%`Xl9G~>lT%VsUcY{wnwpxHmX@BLo{^D}nVFfDm6e^H
zos*N3o12@Lm-pt)oBaIzf`Wp=!os4WqT=G>w{PFRd-txSq@=X8w5+V`{rmUj<>eI>
z6(2r)`1tW-RaI4Wb#+ZmO>J#03WfUg>C@-WpTB(h^7ZT2y1Kgh`uc{3hQ`LmZ{NOs
z|NgzHsj0cSxuvD0wY9aat*yPiy`!U}v$M0StLw*)AKl&EJv}|Wy}f;Xef|CY0|Ntt
zgM&juL&L+vBO@cDqoZgvdTeZLe0+RjVq$V~a%yU7dU|?hW@dJFc5ZHNetv#oVPSD`
zacOC3d3kwdWo31Bb!}~JeSLjnV`Fo3b8BmBdwY9lXJ>bJcW-ZRe}Dhr;NbA^@aNB;
zM@L7;$H(4o)183d6F4f}cEQ4848{CmM^3GoV`0Itl&{I^cqT3-`8=Skt6m|x2vbyl
z5y$Q=ko;A#`CyiBhppH1xsRmsWul7EOGIWZDzHaNg@^~H*G$!*SFm45Q1*V!Q9MTi
zv%W`c=@Q=YRWUH|`R=5`^<ocd_kMGayM0G9?t0Sh)|Hj}d|gJvU88Nxn>y<1TPu3C
zd&QEG$!ua+uySZOHWt=64idYO4@&cZ;s1G~AG%8s92MI%a5a283lf>4+IUPVSbVMV
zm`^Avc!=(z6g0Lh_=Qfxjb1ZF^nzkMe&h!EZke>yMeki^g=e^9?qH-g+YLm@^S~%z
zfQv#r896hu=mI;$JG7N0He`q#tno?)S$7fl?zbl>Qv$DUyH*y(>*Tw1&dPm^{UEZH
z5kHu76buh4a`zp}>W39JVUdE)kBwg;AhM#5ej<wzwn*7KGU#AncFHUDqdF_MC@pCc
zR!6JIHrE5cl&eq?^)Qeu&>w~(qgX#)?2%X>^<RJtm5&dtHTc@LnzRhMKupUDBiYF2
zcKm%eBU?s#r<0sE`7E0c=;>=-`d&Mp3x$Y|9i}k6x13lm$6a<Y##uSo_M$i%`x4Ep
z&e`@EP1Q*0xg#v)`9oa#W9tV^!<$B<{0CJ#3@+2LLX6u9l#s?SRHz1wyP$p*{H~LH
z_X_)+^GsVt)pCf5SF&nI;qY?r(fVX?2&ySxNVF)lrGC@|zM9@J81%&tg8O-#s&RKG
z*TcX@!ixE0Q+5V^!qUwsqSX)DV1CwOf4;raZqKkNJv-zmKRCgAqm8;cC~D=EhiHRA
z`nPwiqsPN0i=A5s5ZaBhl*3IMQ%HYK0w|LHdTY&g1nvrJMhC-fC+RDz2`g-hZM6qA
zE9$Vxqs{0-%OMagC`IXLN?bbO*jP@=W@JLIb1FE|{YyPblP2rr0U=HsV?X&WoPS=h
zry?=2U=U=YqEI6JwJ2Gme(6UOF4NDV#p8`m+R5OCX8Onoc4nTpJ8DXU@bsVc(iH3L
zD?h7HOTiMhBYW#K&RThUJ(Bjw=0M-!!|44=ttfoA<LAk0t+RsVlG#ysf_yX4ixA)Y
z61VH<m|;zJNE1eRo^m`}_@wlvGcvNT)&B5BR6;&EH^GM>Nt2^Bm`$zYg%2nnU|X4H
zVOvLAF$KL@&n%oAiOeV`gL>d^wW3@ORmD6z9}D04Sun|(8=7GH4p1fe-pce+Hi1_M
z)h>Q;l||$(bT84EXVtP``Szs90nYHxhA7d+y>W&Iey_slw};w@d>P^88Bs>qPMC#5
z-A8r!_oD)ojcOps^e8wo;JQM!{gAPWRN)}#3XCF=oZ@zpBKYV@x1W?~FYTK0I_|db
zUiGlCd?QaApp7dG>W%%QUaq|TlTRKQz|vuw1AYp!#_LydzpXCR)|wL{QCf{rDEvs)
zX{{e~lY<c`6BcuZ)L0_l%q>)OV{DQr+cWL{jk-OuBmb_c?+4r?5B+yjnqM{eA1O*&
zncao~7RBvYluC!yu3gEk@f&GqxIGLN0nwh*PqAY1uw$R}0^;-g8+FncqC>_7B>OW9
z%Lul0|7~BdVS8yaDHE=hqC*DH)!2d+!5RPKfXp+EW^^rIC0x%5P%e^w9>tdu@7?WC
zLkK9~B65XYaj~-qZaE);I$rjDH;E?SJ$y!P8NJa5puAasXJfECW=D_EcKvfD5kxdq
z8^0{w5HVxRIx)>O`rMxHXmzB`xTzVl?dPZ<2X>pS$Y&075vWm?6{nghL#|gQ?Q0N;
zrwqb~VX;+r{-=`4J$;dJ%JmEw(^R=w#=6a@2iq!%)ZNqKS#5D2C3emaP_~S6`dG01
z6z)j*?n1FtQsNE^+`vr{t-dUrR%XgArSL9=xt#oqSH1Og#+vUQ3f|+tXE%-QH3b-m
zqA@mK+ym-t&?!pw#2KksUPHgvd7r<(He)8bB!78KQwGgprh-<Ve1tn0G<{hWJ?7H5
zJNl{xuA)$#k<cxb9PW|A9IC;-vLEd=vF7o-i}dy=eVnQPa>aq6oLJS)21wiahmcUq
zz3tERKf;`m-+a5!Ml;2cAwN~bo-)*FsMA&i7x&(PU3hiQ6RqzI-plf8I%4ZFO>Z2u
zIGRQ3TIw_JFX%;U(Z65e^6{UgClIwBG4e3t@9}=H=1Z&CYE<Lj0hT*bcLd_%XK+~c
zre$qSw<pQr!*<IHflx<KbX$JK!TJvQ!l1p>MZtO3p5Zpve9;&N^$5=#J)=o4R&CJ2
zH1$A#LrCj4;@Tuc02=Xgvfg6H3{Mf^*L~mStLMg>*6EF@-iF(ZT?e>3dR33NTC}zY
zQ)iY{IlBUHu&qj)*y$nCw!A)SDi1D>aQn+H7QGGPHB&{KoLi4v4Dk`WyK|n|@8E5U
z^{_oz)$-(9iseag&#~A}X?moTU-H)6Gq~d@eQdIr`=OTyWTnWLdcW;*3PV7%BD-Xq
zne+7wHMGkXdv=n*^f2|%!Vo)8Zq3>Td$?rg((RwCixc%Nn$qJ{&q<IrA)LkYG+0Vx
z5Mr4KBnoSi=2s(5jKZ^PMq2u{7#>z<dMg<3wZ04j%b6IA2W$ARHO#cpV>da!t_YCD
z%Bi_kWNAO7brTuR&F6k$mgJ}v!mT^d7OqAukN{$CFNMxbky*j{hl{N6<rjII%UZUr
zQr-I)@8#L66ES7%1b>*%&3f3MdM`xK&f(GB0VJ9)D87eTV*a%|hv$uo8HE77)wn3!
zVd`<}1u~mEFPJ$)Q9Yz^$21uP9(eiziVdm{2HzqCTgV`=p>a4M1My(gV-S)S(@u~<
zkVEx3p>fzqTp0u@#C8b^Vm}!Qv_Z)0GKfsfDpXiVQCM-AJS33=GZbhOfT^&NF_=~K
zu`ojZlLEI`97|N|V311}(m@9=oxqT0?EmUg{~PQPzJv_p{b;Pe?<qop;1Y`?7vvHa
z$VCQW6ABg4g7M;D+7ysWAP_e$=qjuj3xlPAvF9e@!+U67iXkCGYT7?ll?oFssR2J^
z5tzBowTga#ufDME?yb0m7s*FBKp~N*xj@8ZMybQEqQk#e-EtIt+jWJ6^t<P>hYO>8
zXIaRJ+DPVp%wubPo7@pP(7c7aE>C@G1(lzSat(PLq^YOceint&y(LfrlJ?uGQ;#hy
zyGWzo+f(Z4$C(4iIfJc5feetF%eWoB`I*r|K0$NNRV8-wJN-?}9*G{H7|Ni#4kY2V
zk~up@nza^3PDU!!Hv9B0b6iFPQNJvi^AKe-=k0QpH)i?A7Gr6Dn6roJ7lU^p43Opr
zC=iTS@iMq3@SUmUPv`xd!t$<|{I|jOF9j!0oFwWIDlLst5iKqG`If6E-LO3qyGL=W
zPcy6e@!9-yR25u-?r~bk%Xt1WGb9V%+{5+pjIOEqTCK&Q{_5iBr2?}X+_+@?PXs6>
z=4C8b<L=6%dBuq@JEC<YCE;oO;AwH|TOD)&T3Ra4y+sA@7%vsG=smeUw6d694~Hqo
zvy~!_3$`Dk`+XkC$8I7~VZ=nf=0usX_JZ~Tos#(zq62I;FR+m|gkT{g>Y3<jQ%L=c
zn_tL*hdVM~tNv3pQ>8#b)1ZNJIhBNkL$=c+T8TXFNgcR?n_I|bk5y7}`9;k)k5rC^
zJ%G?q$inT}V(yx%^nbOd@zTDgvs$4iDoFS(jVFaF5EMB`RCT~^iQ5F9m`QJJhoZhI
zjuE#_BN9b?;($M7(2}L!yk|;+jII0!d{}KRrIUT9bN*#4L8xcN!=s18`m%#=e)LJA
z_lK5WNn~pFMB$tTN=}^(I;BHsqI;>`sB}&}y)3GbV=hQ_mhK;|iBGL9wl)5EQ*N8i
zqt{i;t%FkRqu$6Ldw=`bp4k#BEbM1m@UyGONVrVH+Gd5!y79poOf<D<K4TtKq<9so
zFY^+YG>-#3-S`4U@PZ!n6!0l%-vAt}L5cAWpbf_Srt)it|1QMj(AO|SfpAlY6+<BT
zC#wKHh7W%44nb02q7i5Vc4C1Q|90a4roe9*C_!gIx$AJu%E~I}%Zqw7Bt6;*y)8Yc
z%J*4y3#vUvhVpnAvWy^xB;NnfO;{B-04n04);|rXVoR3Z6YXKx^8%}5fVdmjSj~h6
z5vPZ`^DCgL-@$*OSC&JEQX455Ar0n~F1xo|Y=~BTd^=X~<TN4xon0V6_Lh~F3iN@<
z1e<9-q$d@c4PJD0xh_wN0RhnP^B*>7Dg$(F*LBt;RV)0#5mP217ZTAkB3AAMNy`>}
z^f2z~)vK}<x-%!#(=SVXOW-6VG3>8QRMqb(C^5(VLTJQNAbwV(XYWFiK?g^+WeUJn
zn*dw6;YZH(d=49UvqZw6ojy)LzVfz&TW#{Z>c!f4p%YthJArE+3Q7Abt$d8p@YU+2
zpJVFTW3YR3|3z2sR3P1XO(l{qFPJM&3|?_g5v`Z}o`>ECr&^=evLe8hT1VhJ5(T9J
zEzpwpjosc@Su}SQw{dN^;i8)&_Te?|TiV+8ZYKpt(()SP;78JRlHx_=YOMu7+NmfK
z7FmD3Mi;m8>?HwkaOrTbLWUNBM2IX&Apc5wWN|%S<*gZdXTlw}mlbKA6i+cWx#V~}
zjh`=#-PY069O>_17H+=z=`ZTOiww=~%GUR}{4=%CA^j?T3yMv1?#=y^xWu^PbE)8(
z1dY+W1so3{wX5e1mGoY0e!A!yZruOJ?3W$qgjBN21lz3^$<_+zT<6m`UEPba{}}y9
zJ)gZWzQ=X1WxFJAA*)B(Zs})W2;l3dVN%-&|8?9vu*NO)`o^L|`R5qAI04GmvB#W9
z|MrwR38Aok%H4K<vJkF=s_K}{`_-R<Dg_JJJ%5a}qWKO!3iyQ1>5M$8=DwyAU*Etq
zUi$_^i~E$GIfFr|DzW9g;#{6;9+nlkV)AO4R@<aqxmDyo89BCiy1m+9u*HvT6zGW$
zq(zr0wEZkLY2?tG++=@0T>m5tXbZ?9&Ob{E>_bpe|18tMOq7>EcJUZ{(JKA}p|n8z
z@K4X>|414`#7MutF;Olg|K0k($V~nnh<}NGQXNeWbHx3_c7Fp0JmP?bzP&~r_(<Nr
zlU?2=ZtbudSA8c=>!8%n(7=dRadv5Go;0FLv8%FzbT2^fgwhJ(2S5-IF9mi~1~$a?
zMvyXf7r43DxtTNET^VVGKBHId{#Lf5+DmtuzXi(AS}@4@{K@YAwk6<^=}(rH052x{
zOX>$a;{Q(XPLxiG>^`pv1{Y%e&Il`LK?i@y`hds(AS_}4ayXw?q>=2lv~kN6VPw!X
zzLohpgm?<{rmzf15J5;$YnIg&dQQ6pIr?RO`T<{+Z0*X7lhfxEQ92c9B|%QlbuKfU
zgrtq8Ak+12RxMv_VP_@@C-e+~HGEZalh9p%NTaChA2X6O!vkrJ&!(R3#~@>~J5|pG
zU$(bk(v?4*15cC92V-Z|Hbg!I2`=+RzW7P+40F5+-o^&G{8%^5_&6(AIigbjRsr9L
z19Lo|no~<2S+0>)Z!*qS8C{l~;bI=nuj0qFO{?tb!#&rrjZZf4#;W>L`nHv?fm?z4
z{(fa;AnQt3sYYQ-%EM^DKU{innX3F)j2vt+Wa%IA{u`Y+u~omkTY(qLL#FZei+;eE
zdR+famVwPEFVnLO1P@Hga&s0q6rO@2VNVSoqa>R+t4H<^ylID9-!o1V-8)vG4aJw-
zv>xzxaJc9<AhQaaxL+mhsZ&&6uB#WPNLQKV=ClMgaaCo8G{c`7K1BK4nY$>yIM_7H
zYWopc7$Z1|R=-4QH4CIDU~q)YF+*j2Oh(d3<NK!Bppz^QPQr*e#K|j)4&yvAd55hx
zz>m?G6c3n&xk~VuZq2`|FYNZXH4;GJH(eawapxCD^4KXMQ4F2!6A%WSj(8lTsM#}i
z7%wMtX;PxTlZO?H+irZvkYJLAk-)=Ni<P(o;f#}2oxr3y_m_Z~?;4}2c4N(K>xLa}
zY9OjliF2FfK-lbLoG#H~!Pz()?p1Ae8mMz%(^`5XCwW#R0eCtpgh>k_c>W<~9iX52
zaqNmhoD>^0PM*&D8nPQhv5hQ61*jq;E1AX^RzpTl9s~)%9G&f~g!x%Ok}pr?)menp
z2K+)vzGz?2kq(f0dy)VmVW|;JU;{dlposp#O^Dgpgr&Jx8YI86&|gWufm-yM3?ex;
zV51%K8;dsD^>idVe3e-WRc%FkcR%iAA8)}LA5X=BK(+iYWe_n+@{z2L^XJud5#3XS
z3RHd|Q29Cjz23It?u(x{`F`m%E6pH&q|~7seUlyo78sSHOvl=+PXyx)71$I3h33xQ
zvnDcps%nP;zF9`?=}Al=fh?~-k;(@q)#^BMpNIT)fHXYd39sqp6D^oR81y?Z$qFOV
z#yOV+<3KgrATmaIe}=pDcK|ZTNNoTT!L$Y7WeJ__?=e!yUX<Nr^H1`ueBgcw=sK|%
zb(d@49+6@684v=<&;fz-(FXZ+J`=DiiDU#Y0KoC5{}9!tc~Q~-{gQNImmZ56GsRzq
zG5owq@9mL|uijOa6sUE#9!zr6s1U(&k)v01s9$N9CG;kMOPrpwA}s>daiuKPgKf>=
zU`z`$CX-d#mB4g@?|iVQQp|}G09Vu2Qbf<j4z%yM`|u_|8qrwEDMa%b-D&5*sJ;er
zhrRb@!1sufXZeg&s%TyjkYeqR5UeV^>8R_GT?Ae*M>U4uTW11o$e`m&D^hTJ|JY<D
zANCJ6i$Do}_yOF+qpqhkI@}vGx}T{eueonD8*gY0+&vr9LQxVxYJCA|<8-QXDQu*$
z=+JfxSrn`NEV2K!vOmD}(09%zX>P0Me|i{vcxga>{KPVPdu+GFPJDls!k^lEH{iKP
zFf>i&#51f%A8KTtIGkT6GwV-wc-u8UHM5hhJZS}$^s_FI!^O$1oXE$-7)LiQLs^|f
zOq%8$ptv=gL-v^TZ@=rK<x?)EgzZMroFMEbOjH5#>d;I3U~o<ERI+Xg%Z1g&rTH`O
zP4qtP{94<%R#*t>Y5<>v+0j!5&NDgsXw#bA9Y{$7?n%!rv<??0(mL2-tIPay9NIW%
zZk}zdh4D?MY>IU-VSZ&CJgPpj>FKSsO2bd5+7uxI0NvjF1~IHOZKc-mJN7CWAiEwX
zE`@Q=05xP+0dR={%<%|C-{v~9E@G!0{LK_#tBkb<3X!QNatP+C4j^c`Hd?-GYFLXM
zEY*jHlUqa*0h8*PRnetDSo^Iiq&z^0D@3nx_h*Wb=KHgI&Cs8PemygWB;o^6c~2?H
z9GC$9hz;}l@!@A5jk8tooGEZD0)ogzwoU|jU4X65nYAjJ@`$pYxy|kb1f#5CvJr`M
z(B1pa2TBtfH!bxnZWw)rDovaa#Y2>o=Z2CDI)LNK4&#=)SnB+_5E;4v2BdZo2YCBa
zf+7eM010ikes+olgLnE_V}iaZvy1())1_C`clUXTFS>+R7U%HXg|JLfj0e<YaL?m{
z4$v*E*vMml3DGgO?J64m1D^Hwwb$dMh%>Rts^fbWmI^X7#i0`SQ?j;{@iB8V)@<%j
z_wwWx<xoQwgM=e5)HJ?*26Z0#U~;sc9eLG0KZ4iC&=JxC50~zt7VRuNg~wu&IpkWo
zQTMT{YnO!zJPqJndGmr(kMaaSYBa4s|H5IkhFCz4oyaR$?9SbZ<mM>rQ=K@9!O6g_
zARr@jV1mZsgDkO61YZM7xgO{(^Ol>~Rth#JB<lX7LA@@(SKnk)yc?==5`w2FkL`~@
zf#WjZ60++I5(q>_*Dd8=?X@7zPO<)t(jS)uaKlq|tE(ZpGjpjN{0ShwtZ>z(9WK>g
z=z98@6aU2|sxs)dr~rB^H7<t>xZrJB=pU1q09+qk8Yy7U&ishiPYqT~7Wl`|hv%T4
zmp*u8yhSpcKRG3s!GBz2?;>oK(`-%2lv7mn#6_ea<A1yW5Gk~7yt6LqP!-u62CV#?
zTHkH#9!uke2vqK=La$&W%bAp04-4Ej$Wma6cHw98)y980+TrtEmIe<3xTU&@^Vanl
z8exB$yW15q2xjoBulp>D0|2oA4yQW#=7qFmhv?qo?uufZgb-i?S+=zrl~>LumN|Sb
z`syxX>Ucl9!|e0S%dCw$FTgUe=s5PP7jAb{4>Jj{HX}-bONmi@k6`CbqbO_g%W`#G
zn<~9xpZk?UL526V<S}Yd%M+pJ%zd)G<Hei50o+fQh?u(4V<IsV)~3sAopUGID7ReY
zx*tA$42!;(qLEVx>mF|*o52Cd_!1Vz)$x%4Pg6a>U=9M<6~hP2ReQ1sCP*SKAYtSX
z#{0{K{QtO+G^7~^S%QU$A7=105_KC3<Ben(j{m>1tRv_Rk6t(ATrkS3{$EPWDM01D
zxgq~Qu!C%%;-T&^P>$DACV1J{gb-8bkvc<=!iA*PHbB@(Wp;5Ib#glZsod`_I^@A%
zF6-*eO8u_UI90uGKv5Ae6d3N9yf&SETH#<sdH~gpD~rqZA?0R3c>?f_K&VKY=Y}?X
zwG#ML0nY&KUisWE_B1^Q{6ZFzXj!$F>rFCrkErT}d-0Ow^s{#XK-ptkHVOuAy*hHJ
zLo@{B<#uOz`EM}R$b?yq_kPb?V+6m`%x|Vy#lP;o%GDgV{z@nIjkVmRZbSjs9n}*%
z11l))6fe1yssqJAB~_B{sl0foXo!fbGBbuWq`N_Fj>oM$QJ96c+cT|<MvPZ-(0s#5
zYM%y4u&|&rm~R25v~zj%NSs{w6Q|DAo1i7eS~Sa0u^mgcgR(`>fP`Am5oWL9Npi9q
zDo<L!U_0UVGE6ef@F;=e*G4oP%m&RjWpB%C*x}wX%DCo8aborqYh-!HtLZnz$=|g1
zF=`Wm3V-^FYpiK!ENrmWssY#l(BS3bnqSpDFL^>R=~s1LC!}615yQsfi#w2f?>dT4
z30cla5A|$qYW36-pDEz_bhZ$tG)27dw(Vr4E*+lY5jX9HjQ86|z(oxM5HSua?|gRW
ztE4!F{M9LH8$aP7-<ajxtyf}Q?&rj?Y&<4J^uCSGr9D#^D$&L$RsaqJP~mg-S+|{5
zoBjPKu`LP}CEBE3BKNP&Ep5HtQeM`n9w(T{5H&kAX_&ek5X}slp%ng<fjgqcnY4Lc
zg?^<=J~=7PWBlo2$6AYl|11dU@+1M#2^y74V3y9<8CGGhp;=#~gfQSkF5v;fxixKU
zfr1jxWIyNrkuh0(&)m{x!~pjP?X4&5VbK<mz_rJ#NxBS)60K9`EhxGvWpk7i?&D54
z@|EzbGKIsvH5oMB+fWU<2tRj>{*);&Q9VC9dWhovxW-U@=?K`W2$WqNLjiA0J;nT<
z4C|`~&>UkRJc?$+(1EVAa8iwKFaEM^oCO!n#$S7GR)%+4*^#)?9moSF`OB1zuex3F
zJhlD)x9Y?J1>=ur5S&pHQp8H`l#d)?8YdbC&aq#j0Ot8W8v0(#C@R=#BN(-X1Tg-O
zKIOY0b&FiAMF`{(^xe41G{Tw?L+1Z~o~R8G-R4Rp!ltLF(f806#K;94_PY>6;-N6O
zqaXSLG4d%gfe&0rmKY3o<a4|lg3%%`m<P|k9$F#gc&;@Cm5IaTEl|)fXz<OyV6-c{
zk~nwFBHzkrh>+o?ktR9DVa5&#B!0|Y?-GzLvV2jl>PCC>i%0Hh=?w5PkpK`83hoNC
zuXpo;6nMLxN9SYz7SMwQqQ*G|XoK||i{JxP9jZkn2t)?lPGWqjtC>pu4?^a<bsv=A
zU6e=o>t@)HU}D7y4_9cEwurR*%^R%IvM8YVicTE2NR&l};&Icqm-1+4Qy28sqyjQ;
zyw@2XN@I9YC<t!lw6CY7>>~ootmFCE#yI~}J`JcBzm-nrm)a~zV81j}f(Najsi2id
z@1t%9$r0vzuboF1$}tk;d#lHcUjhNUTjr9%bb>LO)_njdwFCmKxG#hD#sN_iLJS=!
zuOYgd63@dNtFe(pxZpxzm}4q13=hmh39T4;fO-f5_0mHtZv2@cgO<ZaE|5SBIsWaW
zNh3t}LagNJk_tvd`QA<?`2W)~v=QBJUYhfBub#a!_=E#@1XB_+RKA9a5Z*q2_NHJH
zFJcqmX62GFG2|C&N{}M_4d_bdQ^B9%A6#|ON)QHYH&Eym4se8=0`7=xLd$WUhp<r9
zc>!Rg%qtW*p)~+L*BR)v&*tuTCm3e<-GlkgLJJw&N&3E;eK)F!N`Fgk8pj-@$Z@)u
zedb`_`yigleO17H1q*una*P{&U%NTkP>)NF1rz0wd2{Ez<<PyNt1&zM(q8FKFeR>i
znt1=VvRLc_bLZiW>ZrZf_c{UtNO<Q0=aV?@>#ja!JC+;)MrHW)ZHo@nZygX3--}8{
z`QvU;(|m<p#T=qvj8R*$q@J@KrA$C&>Y(fhE$bs_7qFU>3?zK`m1Vr*L%rwk<?*N!
z@ma5oK!d`9v<v-|ykxhaTKkjd(&nq-DfE7yf~<FkAbRb~P~4^7Qu{*F8yIjkVfL1T
zq~7Z7?Kf(P><;Io7$ol3W-$^oxxK#X8CIi7RQ82;g_HJ}I+&)ri{f5X)}neo1ee6a
z9M1^2s#vt|ItY0FkiGmhz-%m;{-@C=Q#mSElT|(A%`fpUJA~i9>`?4;$1$P@)O(em
z$1yli4xvy3C|&s768}D;{RX7Du>vp$A(D2F3aE%#F}3cfC3jN)<M1Bp&hr=j90`ml
zM9%$LrWJSOtlAafB=j5#u}!;yUd4Tg3QHu)_Za>mmRlhCbzL5z(1S~Af|2tPl?v46
zn~50ce!VI9A9<msD1^&HLK?DbmHj2?%cmexpj}*eLE2_u9F53f{2uV7@!oyjAEfdU
z)tjF_@rvuT06!bb=}IomD3B1*jR(YxIjT-YsUxwmfI$7rXA{4_pZFJzxJwX;;=%%}
zwf*M-DfeY|bB&F*y1$ha`XvuzlAOq^=Z|Nt)2vf@t|&ZF=`9zi=6ogTdxIb_FiWqU
z7?R(T?JAb3>iu*78qR0Utpf#>!L0U>kInV-?cCh)PkdZ;wPLytx;eCHFJ<{woAT%e
zzJAR~C4log&|*vgp|iQ$>Z#n7o1OI`A(pN#Z;fdX;AC620nVM-yfnIEuXum)n>yYB
z;srS#&6z4tN`bn-JZevTJE3Er{^7pWmlc`<utS}lcJ974Me*v^-SU`_st47M*;A2)
zJX{GuCJxI_&UJrz%9C+^E8>F=2K((BTOi==xb4Sad#atsc)=<2XC>;41$mOId5-|(
zZd(0A%Exb<jX6*1X}jJHhIwFF*{(}D-lw6Ry~vAPpdf)rcb8F1Hod>n@v8F3qcZ17
zPKBZ#bxd7ry?L@1j@8%#qf72NFs^#Z>v`O>$frCJ`F)JXJG}7E%RvTiuH}xhN7hzf
zxPghezxLV!?6u<Yve_W}oT1Z+g>2sgh3|l&gm7wx=o&ez2|uT_aVDdnoFh3s&l!@8
zv=P2oF4-IH0z`SnXobp9yOeahajA|6tJ-{c4k=2#^n(WXHzq<iYYgVTQk$yPNmk-%
z`I(N}jQmZgXUIWHGw4*L8dLQnAxX0l&nc^g25ASx3PK?X(9xEbq(&^!^xtGQ&APu|
z;x*}dbmA153z!<YKbi1qKkC@rT--deP>9Run9A=auLJCtK|&(GXH;U0dX}A^;+;Dw
zj&tEHr$&RM?!_*>kgWWk6Ly=<;?^ODlcNsX-)C2x70OR%Z?&b`Br<<r9zA?e_8NHE
zn5J1LU0In!d=&j&G2(kh=DO(@ebpEk$NhL*-7Bj?VqJ$?2be7Xt6v)SSaxO6BKl9B
z>Bn`{(S6;Vw^vH+$+;Ni>u0WPMWs`X+Lx<L%OK-PwP~)$Nif;mbPCVWHj+ybwM}2}
z=$bQd^Fz0WSRwW1$<+Y)c{#4t(L!Rei;ti&#<y1HUt%mpNfeL$`*P4vK&?Gnoh(&I
z?v&SgZRgUnBHDsMj-;L*nQY1P>=+!^%!?lyFAp)pxvKL=x-b8ZsfE+81Il?sXD%sv
zF^@hH>P{%F8mT-wO>#w>r$(=7IYkz-a0RW17saJrpDOSeI*F(1Knqy}mW&#V7fTcm
zgcD`@pKBC9Fm3eE3B|HjW(3aB*tG%X0t%6!`K2%%+H5R<PR$LKH4uZZy~QVF>N#a(
z;f}PpIUa90{z<F2iQ%)pUH3sRb=5%*S$WO9(ca%*1@Adk)|hX5t%y!HU$GNg6z9I7
z00)@g==*}-(!Do)+fMGiCK{#j=`~H^;(cniJRG^<j1Z1%7?G@shRhc%f6i*wT3?uy
zw>T)6Ab9obyWE&a1C2&2VcVhNLjk#(A}$M9>Iv?!9|?)Iqg8ybqI$zyGR-#GO7`aG
z%!XRuIk87y?OWN}PL!k5`AP&j-3sfL)aOc-dN#SEBCTn@uSYG*mxB)*8Y@0UYm^om
zrk6jD%wW0%$vRv4AV8lnfNj#F@(4%HxJ1$Sk*@}ImDOI|(2tM$l&@byE5e?6o7lS9
znr|7)u!*p;y(6F6V)FmFQ?3U8ko!w}tXuD|Jh1FOP|XNq{a!yeIOFG=*qs#XP5l%^
z30PumZ)h}cM|5h7EmWc2DBuFzR7!9wZ`4;%c#DS$7f45VzGROaYe-#YD6iWM*!SO9
zFW67~4S+bsJK$-RWgn9k7b6p`a8THwotsHw!tcekSAY*W-wh1Bok|f))NFlM87lqk
zcdye$(PTUBc}ZVV+GFt&(QDswVioFJSQMA|%5h?$&RI`9WwA~WOjz?<f7v!l_Fjk#
zXK_f_OUJ*Jbk}cjC2UZs@wSI_*V}0wX93aiR5a1Ypwd(CNodkAzT3Ce6CnQSsB!G3
zy<x(zPu-QU0Q#7RB!?%q5baHpaQ~bX=)Ewk=exPo;GgFvF0C=TyBA%b+_|0a@44+X
xG0pv$Mk)6gNGJYVUlRV;zCZj=%{mOY>=>P0Dgx&Q{wD%U`MSonLODd>{{lXolA!<q

literal 13613
zcmc(GcUY9mlP(Mc40!;_Iim;yg5)3>NdgKgQ9*)8&Y2-25><jE0VPL49EpP>L6VX|
zkdd4jB<K6hfaiF2_wK#>dv@>rgJ-_b-St*=)!Wt8A5k~8RY?fx39+!SNYvGo^sumS
z5Lj3sc?b^h#J%_QGjOwYQ}dQGaEpzN4FZ91aB#q2FfJ}G9v&V(K0X8jAs`?iBqSsv
zA|fUxCLtjqB_$;zBZESr<mBWO6cm({lou{sprWFprlzK$p`oRvrK6*xr>AFNU|?iq
zWMX1sW@ct#VPR!uWn*JwXJ_Z&;Naxs<l^Gu=H}+%;o;@wy?F8BrAwFi`1megzRb_h
zFCZXr<;oR7K|vuQAz@+Rt5>gzh=_=aii(MeiHnO%NJvOZN=ivdNlQ!1$jHdb%F4;f
z$;-=MzkXd&QBg@rNm*H0RaI3@O-)^0{l<+O8X6j!nwnZ#TH4y$IyyQxZ{F0^)z#C}
z)7RI(b?esc+qVr23=9no@7%d__wL<$_wE@P85tWJo0yoGnwpxKnVFlL-@kwV!Gi}D
z78aJ4mJc63w6e0Ywzjshv9Yzaee~#&ot@p|$B*sp?HwE(9335<oSdAUon2g9TwPt=
z+}zyV-90=!JUu<Vyu7@<y?uOqe0_cW{QUg={R0950s{ksf`VW$Sa5J~NJvO%XlPhi
zSa^7NL_|bnWMouS)RQMqqNAgqK7AS!6B8R78y6QBA0MBPkdT;|`0UxU=g*(Nc<~}B
zDJeNQIVB|}H8u6+%a>_sY3b?d85tRwnVDHxS+8Eb%FfQt$;rvh&3*m)^_w?u^78WX
z^YaS|3JMDgi;9Yhi;GK2N=i#h%gV~i%gZY&Dk>{0-@biYRaI48UH$IeyPBGs_wV1s
z;qVV1KGfCK)z{ZIG&D3eHa0aiArOe>=H`zdKYsf3simdm^XJd4t*vcsZSC#t9UUEC
zzI^HI?Ck35>hA9D>FMe1?fv@o>$h*;`uh6%`}+q51_lQQhlYlRhlfW-Mn*?R$HvCK
zfB!x{K0Yxq@#Dvj$;nA168ZDz&#9@Y>FMd2nVH$y*}1v7`T6;Ug@wh%#igaC<>lp-
zm6g@i)wQ*?_4RcW3bnDZ@$1*G&CSiNt*!0t?VX*S-QC^2y}kYY{ey#p!^6X)qod>F
z<CBw<)6-KQ_qjg6?+KjL?z&=OF-4&Nu;XSo9%5m!VyP=#zvY#_n&BJ7sNAs57ev4)
zkG(-B&Z_DixsHRsJ{Zl|kevH&lyfC2?eWwU1*IZnBN54sPY`JX5et|1lfQETrg%Y9
z0ZVzNgogZi1c*S%dyX(P^vA@CgGW`>XvdD+`?^}|THm~)x;*RJVwTBXmc_!gmt_ar
z@8w3lcc4sI@%JK^2(Yj|;=r)ggrKxR7yh4LbU^ngL*tW8RwYz?T}gJ$<l9e^mhN!Q
zoYF=!D>#%r3}C&y47pg8shFo8ix?K~K2{|F2du{F{8~GSAh4_)mxqd5fpDpn2swfG
z){@=$k@CpS_R92kx%tncIq`!LLX^CMa(b*9df-f@_Tt$&V;38KkeRx`bEbA`FTde3
ze^DKZP%MzDrXE;!d17WvuXs;D_zfPcZFo{rKq{VJFp>zuD5OrlcXhM9>G{10%a1w_
z!2)a4D-PwTbQr?<0T>M9;1&YOskNV084+~^9@?uTvu|57xwf>;oetK^WC@n_Z!5Or
z989pt&9u#p4O3_O?2+dS95F06#`-CpE=E8kcG9yLD|Y8b9f+iF`H+sEu2%7!eu~JK
zq0rCxkj)r>5mn~shx+On;Nfq}Ag5;%X!B!qU1F;GioGoB*hOSKEvzk;tGKjv&+=_8
zC(10giXszb(x6~I-Nl-RNQ<i+YK+MMha>n(u1WX?t#;QnC?Hj1Yot0G2yiVg;#(da
z&eWA!NWaUnr%<YG8jKB2kKWCA#3}?@ris@vicux-9;DvmSFE5DJMmkTg6wC0DBSW|
zhQ6Ib?gsACNVy*~D{u#_^2i;JDJfiA%tJ(Iv059uf1lbV?|ZycGO|Xs*X{Z26U~Ho
z?=Cl`i{p{9b*hVeIHD~faKERQx*<8f_ob0iLK)}GjGdUhv9~bWq@no6KmsF-g_Y9O
zB1WT~u9hFAJOkZ{*c;Suzc1zX9uaZGcX(RwbxC4Jce9nFEuM^K)Q<bLMVC#}MsMg*
zp--J5Qb2)OM66LdA)UppAvK>dp00hkTduI20XAM9%^?Fzm$VVU*XGoEG?%WO$blLW
zvLvzd3o5C`4RB}w(6WSrvYlqv;mX=D)NSZ1I8^wQaL-)xXgrz@7z+4l?&7fRQI4QV
zh#FcRPJU-brJA}AOcQxKerZID3aj-K4t~bK@$I1lk6#HR`DqUlIVu8V_t)oE^C8~f
zsQ2(g>#cV8p!?}-)O%G!OCM%<4@o8(RO9zo{4UC19STc4u73n5P)zXQV)&AanglCH
zJj>X!RlH(rOG45<JGd+iOmM^}{L_7U$K{vv1htn?s_~Lg*z5}Ah{|qTVx~kdPABn7
z#pdj;hc7>yRB??jZop3`g`T}0jxd|(PLA22r}CK<!*sBE&e#FVqwUE_kqmj{J+*km
zVI2EhZh?OLdks)v6kJ@*lgKaCy=}T9S>rI}hW3MbjYYXVU#(=m_lEU^$4I$U-^{Ij
z7Lpd-hq1i)mf)Px<)`ui9I>}1Yf#e>cZ=DaNI2EZ{bTpxb_{I2$za+3jB9kRL`O`^
zv2D+W`eg65ACB(-s+|ZxNSj@HZ`ep?y!Y_Xx7)wzv}ua>s@!uNVbEeF08^ShcI#yb
zLF!&1UGnK%F%(Xn$cbv0d6jYSVn(Ea6V;Ub<L)J0E|i3J+|~9`z=;0578(1COJ7!B
zfYRzKPK*^|SHE4Kb|Nx9s7Y{keJ|J-@l>vU_wBs!CYuvKr#kZE_YTQpiy0#yb);j$
z?sM5$<I68uz7|q=2}nF@seQM;YAGP4J~XzQqR7o$Ft$eJ)fQ`NH%$`y_$j=_?<e)1
zitof@m)GKbqcasILrX-g=t0dJV$zS<*>sk<U}DK~<fTk^H?Af62aGEa&(jSH@x*M-
z;#ia%|N7pAKl9>y14xJ4?CmuNo|&WqyDlmRm`<fE!+sx|<xBBC*Y?Gsr41eQ1U*Nj
zbzG0`!0$FnLQGpHS_D{MQ@xmqN_5Ec7g;uwm21^zTR(W_JzeRU_^hp@m!wN<(sSIM
znr&hbkf86fU{I5$z0<-VLi`?KnP7|RxjJg}DMxKSQE}8zUqyg(MdCs{r!FXtQJs{1
zQp9mItHs9+<o!LJW#awZt!1siV?yEk(={1on~IZ4?b$@@@ad20LrhOMX9Zn`zL03}
zQ2ugEUp0lTb6NH&%N=fax4I&)s0DJ6t0Bo#{Hz8ulI!E{WO-03J(_Vl$mn$|mx*H3
zCiOADInmXxk^>&<CabP`PoiOFwYO?{NP@qd$YAuv%!-$wq+Ft9$@%HNuJxwh{<iMW
zlbZWIJ5OjT8zq_pB(_RUQZ9892*6JF38i2chYnYR)udM5C7%S3uvp*=b7~;XD1ODQ
zsO&F?`X>i$t*QnT`KL{6EKzc6HBN9;oDAdgumtQ~Br5fcBZfu%fG@mOZ%g-Q@bsN6
z{u-FM-C+^Cw=i&+ds!e34wlsjd@#2`IqAPP@y7m0ZCfi#E}$=psdou~j>&x_XU{s_
zAd$`KlOV2v3_jKr4hSdy0jnU^MZDh^+!i3mlcRanTfNvpd+jL*tQjH*vHTMl0_!#H
zzjTqpN#OVD&U3irrkCym!)d*`zS&m0k+KytN`fi*X^%Ha8zajytb>KTu7sjahI;uW
z<D+KEiQpf_wud3UF^?!zJ+kuD{VJPy#WJ~_6XKjdY>1>z3a!5QP!O57ny~ODNl5G2
z2PNNiC#Nju4?c%)TKg>nIvmOeyj~8B@``2-jGEzsNtww&5nzy!ytx=O1smoqZ~j8w
zoCsn}0;a}-g|vkr*kA~1=m-heO5U6qdWQ=-0s<N0z;NY}$PHE}HP#ATum=u|4w`}w
zHXs1k1R)0H&9R~4UeXXo3dmz@*o3?}9(WELb{hoaXLS;V-og4Wi@Zw?f2TR<1x-<$
z!llL|gvtDSfE_e@g?s`wA8ayYx)ZjU_%j%uT#tE#23Mv6LzFg-Og~gaX5Ex@cepC!
zSU38WT6auk@$L0fJbX09Vi*EM&xV?O`7*H1jHNa_p!;aLHr?aV0Rzzj{lxobY`{E7
zV6-3=NPVn>APvXW1yYkqZKLiGc=3lt{w6%KKSp`;T*hPQN_O0W4~<DTedhx8KVN4t
zTM<0%ef_<wtz+WxtjQe5L+23u-*$`fWw4#Wcl!x1d*5GLGOCrlCu;mOZj_jIC<gbf
z4g<-W$ZOkA%?&EEt^If*zN@=&b%V@0%Va+W@07QAg){gckq{hoa6%p@;K2-g?DCTB
zT4Sg;qbiyy-C$;)!kq6}2Vudc(;{0q;^09DYFp^u+q=(Ax2m0UQ`fBcGIN;%NNAw&
zY<VPNgH<rjzv=O#0MlXX<kKF39XmKJL-pcS=KG|0edQ!z&oPS`rg1xY2kFu)@L}^H
z#E#d!@|H3m%T~&@OI>%N<j@gzCeZRh+~N(BrV&3b4Y906)5T$A!V*m-2wc<lDf$kh
z-bJT3UjjIdk<D^PqqQ5nCa;(I7Cx0!CX{X3p#XSLI6dU@D{Rnu9{-R|v}b>3m#1&J
zbqv*3z@HXnMeSWapz*MvTJ75|b<=K!%Lv*Z#GCzLQ!mn^t7G@2>zEosYU*vPlfK?o
zZm4)#Ht5g5$XcPigex#c8{zPZR&)KP6U(N*W0c8i0=>?1!UEOSTgz0W35j}4zAf$=
z-f4dWIcR<TwOcE${(|?4#9GfaNsC1{Nyo2EXa{+1ip<>mArcL}3(L)n2<##8rmBU{
zCJrSzmh_hxtdp$jOVx&l&<|SPMH0U7S8qV%;zr-MNDMRg!3#NjYJ;}k!NJ4UejiSs
z?M^)U`r_+_8zS~8b<L-YpQ$8*(Sg|PY99Ss4&ULz|276y@4zHB<dHAcvk`q`3g$x4
zlo))Ni9C`M3kbf7G%$ow!3c604^*|mT0sO!2mjARuR0uYf<}QR4)h#if$-z-VTy?=
zuw2LmP#p|G4C%rF6{AT99aUHmMrmd+DTX{8SOsxFn?Z<6|E4dRJSrF<T|o$G)`}aL
zxW)q!1tFTJMGcV188#<T0%~z_UC0U^=qjrq8KeZmJ?OY*bwYVT#W~R?``0C>tw1gl
zGN1Dvme-;Zqcpe-0{|P<7bWg>PM0{yH<ZpOU`v#pBv#76P|&?>WoPNHsxcp)^RA=6
z0Pmb-$v+@62B`NW-y_D@Z+z!9X;s<~5$DuHhDY#pRT5+9l$=O<Tv-F;*+-*7JFQ|N
zu{y9NM*UUP=V1sAnkyWh6EGT~mT&`{7QL%4J4v~d&U%bJy7U%cqtYj5Yy|MG%4<-0
zmZ+V<Gk%m!|6^-KfRB<o7n1&HFw+aO9tE9d2)G*v<WD^IPTsE=?MGst8#_`9*|s?7
zdFAo`K~ZwIiqymx1CM?#DiSe^Yq(krv_yHHj!s#}FNb(4i)EUpl~Z%g+q0RY3dv~m
zFMEcdlmqa<q^w`+JG~+_lMqHvtv~5EDcEkxsN{yNne%16BDzHH;8P8zd>EUcIU;8b
zVU&OFoxDMsdKu=JQyLeus3_Kd@ye3V)roEcSyd^ZF{=}YdOZT4@Y?aqYsL^uH3ot;
zkpf#^fCrz^Ucty&p@U@BZY)&CH8r1-Crfj|%)D<{blNQr*2>l$sd_pqf9n2mooHJ3
z{<^uKj4!98?Y%gq@*j-|0o^%XT9>%4fkTnfEFj{jg251J`i@hh7<OP>5tok?ani4O
z35=)iZmzF<c4M$}>*wR6mqxKCDG!;o(7zYDLCZw;`W01(6p7tfBhu7@0<eZB-Fke3
zVza)Oj!knC-W$D(C<jS*W3%flRxvSd8Ef`vin~h#2WR3KaV|-qP)Ua+rxcmFG--~<
zroHrh3&&I>%;#SF6b@!~%+-0jBYB``GI?YE!qe|K{dr`!MZC}};R-i3ELT%Pa8HS#
zWK#Wm=FoP+_cW`QI0D(lPkipvnG_pd6UVdzV&efN+>7YVRjHQ?nmZ9*)(jbyZju>d
zcLtFdf&c<6Qp?;V@aR`6fAaW<qo|LgP1e{q(JtFt&!k|uQ3~iB0mkP7K>s5^U88s&
z28=^faL^gUhn8o;at@F#Y;<%QPpPM}h=J#@fONrJVB>XH90SNT2#86+WG}tBHWHWw
zI#)>1kiNJ%9gbjJq7t8zXRDz7_hD{)tMzx|=r3?KL3}i+xa!jVHTETRe>9bJDL?!R
zO2N}}aTU>U{W)`OoU;Hnr*Y2jYEH*Hk6?i*?bHwnlJgKYX~_`Accnk)t-465hoo~N
z=Xr|<x#t(gE?a=K`nW52{y{dxFlDqVBVo1AJ5&14lc2xvLMiXKzvPw#?|}Zwl)@40
zXuK;5?n9-}J7Bc!ObU8xsX)<a162?W-hHoNPL58OJhCO14Ir06Xp0KmvYsIPEo-U7
zh^+#GHbpV%NVEk1xeB-OS3ajxpKDiyv-3ck=v?tZY+s>#Tui#dTCo(=kkTJ~mag4#
zN0@fmUO!{w5=rwbbJcoRP?@p}TFZ+>dHIGp&+=@A>wd-NOKMSC&^PGm4~v{QqfX_Z
zXOc??h@`8BAAF5>aKywF-6@UQ*3$NQ^W&E-12LyLcyNZDeP%nz@8O*Yq(dD(D=u6*
zr~IE@?%js@-3?3L#X@fzbKkfs3Ki&*0A&xrPC?<2qeH73QhEJ(QeELSk$u;0B$FJ7
z0d9@4i%lrvxPP(8v;kgNxJZ}#?jNa}BHD`b9N@-i(=W`_p5Kby{iW<h$)?KY^4=fO
zOS1}ssj*?WA_~a$Z)TOV9zKn>8;^$WsJ!131)}Ts@P-+umyH^g=2POH)!m&I*i3i}
zZ^i+g4?U?;8)90xyz0gW5Cc~L5Vo|KS)}Jh_4gNSB*Ov1E9F@LHHAliV`k6-miYK(
z&UaU;;(#f*<C=WwW<~Z_xY9B%=zX3l@?)v~0-nXFrf1cTYxSq{pAT;7KLOeen8lrC
zZKZ|edkLGioj7wbqq9v_(Z}=xD4LNlQO`mUE(36N9dI-sdNA`H^+nO4$>He5Yc(s^
zvY^on(kMpSUYPO#d&LrQ!@@<Vm#0*NQ=Pvru73)@SfBeJu-Q;3zuFpU51NKde#5Va
z$AFP4@J~M|0{Oo<@h+T&((#RJm-T%#1?{7aM0VmD0j$eTv;wbt87qYDX)F}wnBGwt
zUTP#Eg^q-rj<W$PGGp-6g?m$WDz4|G)0RVtp1-rc6ucE7rvrloN8H<xlHDI)W#8p?
zk}8h_E2b|hOX__4AhXxYbe<b9AI+1VZ&l8PqmzPjjOabexlkYtSZNMiJKvn_Amu#I
zGmrm#Smt4q`9~}DYdg?SYS=J9Zi4{K;>HCu7@trOrDwkV4E#TMWehq^)Fc#o3gl;i
zd<GsTH-It8PW0nckUVjstTQ9RsyeJZ;j}7*7*zLhW3K`v(j2(-4RWNa*X^lD%z+CA
zK<-w6+ERJZSKBr~u~ZmGLBJKRfcm6Df`x|Qdj~eHc?QO~7sdg`Je$tMQN_n}f#7D!
z$+g5PV*L}iq-^_7JP;4|nLxs$UyjU9kl$FBsl;QI*Z>0-s$N8~$RfSJ3AYI9VP>aX
z2kAu-hG0I*XU;~HYqCSd1CHiRlM%vEvID>J{Lk~72RC!+F(Z=`j%&A9o5`IH3$ukM
zfH2rhK%Jap>h`dA%Fk<bYae6s<ug-esk~HN`QU+>x?wISr<_kF)to4ovBE{i%dhN?
z&<Q#k6Ax-niG5G>Egoe)qUh6eMAJtfFsbHN^PBl^iHPsE#kQY+2lIJzQ-zZY4|?C>
z046ZJ6Ni9Bm7V25PG>0)W=n=KJb|<ic;7ivW|$PPo!NaEkOr4zC$*Qvs#RxQm+F@(
zAeTb7BbUWrW6fVBsayt<TIM6Zl4Xes%^WV+_eF!6l{^hJjbJ46qL4i861@>4o~3Bx
zQ4Minni(yS*7yYrpzPp_%az|mN_l8IwEODxn9?KTad!?|IXO`Na&l3ON29T)y#D^m
zQv%sR7%3X&sY)6kvHpq2`;4nVZy_C(umVzTrFNGi*QCR1dONc1*^-(-EwN<tyIONp
z5;)W3M5=D2{3l)@>_h0~ijyXU>Md61<ePtjxsvyN<ZT*yGe*AbYCh<UJ1rOJ$Iw9n
zE=x;902{E!AI>*mK#-9E+jnGb>g@oQJ)aum>+8GdjoI0A%j@R&rr4DP^)GrZ(niB<
z`y<Y+u&>ICiX~J)`!Y3x@09@TAq77{$(8eDWO))uHGqhc>$7|C^R~?!6k+7*Z5#jJ
znaINVYZi2}-Z;Qn@xxc~a3Hqq=PMOabW|O2T<~&@GHTq%cKP~F2IFBwNUEF32q7+@
zE&vRhNmEIv0olJBdH*%Aj(XvWEGd`pQlkBLo~6I|>-6cfdX6q}VPLVY2hs0batn=W
z1(rgaN6k$i;XhOkbQCfPgkp(-Ww$)2wISw$Ne&=2FwB|YNQA9Cf>sp|ayiIQei|Gr
zfR0?C5BP^BDH%G9Fg*wu&pQYaNblVJ)qx=iToj>A3WA&)4bYG_8{wJ!!LB<WLT((8
z9RQ>Nrr7QTjQVGCD*fZ9O&Wfy^%X(pqLhRt0<-&?AeN2KO%b%pq!<LBz_Eh5=qvS5
z(7Fp7n~u#NBLPqnF9c>djkogNcax3}HXvRTDX~5DlPZX<;P%3$<*9xpGUzOcw!w`)
zK9tS0fHimA62vYi?+a&u9=8B`D6VgW#`X3mG}=C=qpo$W3Qx%quwbpGJQ5lZU7t#@
z6v?A8(@D}IjuHY>?3Y7r<(hj&unn$zK$3#IO;LzQ)8=cKasD+24N(i8N6*8+zO2(_
zfDQqXITyTQ_8-jh<kzHccucqEL37)I1sFwGaK<zsuNO+Zs2#<2q2f^;=ass(jF}9_
zP&8?xO>Y8fd(I0uRxzgnj3z!R?zw}UWP<U92ecP+XzsW>vHiS#=gb*!QJ11_BK6z-
z@+*VV>4EoXt-{VsjW!SjS2lfoNyLHY;>kn2!RD|~!~1`ZecH^lYj1lZnZ1Nm_*h(V
z#{X#DvJosF6>Z+3E|q8?QvhN_!#!XhdabJQ{!~2bGV!@De3Ue6dBXL3icXJhTvkZB
zA`;x!*&Pp!&YcsW2Pe&Qn;fyWaVxazTKn&JX32hmJDWu6&Y&HbYH0U*ESi8Za&*6R
z=RcSA^1R{3(h`hd2j1@Mez|l0_UK`FR+r5D-gIxOuSO_~e*~t1uU3ipS%cTR%b^c3
zR!BIlx4}V=4wU-DpyC7=Phh9=S{W)Iw-RU?ajBXa*7^iw?SPj0$lv2%L>1>l-}coz
z<FRE$t?V`cxu<E^HUXnf4%+zvg9^~vvNSZ;Z)Z?K_AM<9uZLjp|D_}Qg*_jIr0?ao
zsmnQMRL`CKpVQSFALvT2KM~Uk!pQ#H!n`Oc_2ho|3qMxOx`bU%xJ)6Q2P|2Dr;-y&
zM~$zFTuW5METDjI{A0R!61DRy)6=2<kqfM`uJfPz^)Re+xSkW%Kl|#iXlfspx^I$Y
zu|~);{reGk<=Qt)6xtW<JIlmRHK$w26rZx=qrJ1Dk<6v#*_osE#p#`F!TXYXV^0Di
zuX9G-07!R`r2M0#+zmS097SJ#Jke6ev&7l<9+~62C7X}U#jkbx@!&7kH)Vzr&8%qe
z9ygsVLWTETI}hpO%Yl?X$nbuHnN_6lHyQtH;qLy;TU$5W`ktN=ix<RlGmhuJvOStc
zp0ehJJ95OROZOELVR~>h#$0gE(Ul1`$WWa?U1}L70N5QU4FS^}iU79P6c~p?Z}QGu
zf{+a;xuA<gK;WP&z}g`6L@S_+EI=LsU|>VnDxP7X6^5b-TOA7>{t)z{@&9+&BHVz2
zS*^pN_Ji1@9+J=K#RG#p{ec%%5jFhT7|1S~*FM*6^i?2GPyyfz27-#Uvtg)9xA;&_
zH2ehhzqb(R3Nc1gZJ~O?+k{8RzD*l-<ZAUGvAv;dzfr($&1X$4Eb@Nzw*Wxm|F@J2
zUdJVOt6n~a)h#@#F+Sf;0DjBn^yKI5f&&TEWh3NlbsGP=Fv~?CIX^S9_-E85%_j(8
zCt(CsTF@!^UjuS0BdKdc_<_2~5xqbwp8bJHX`o<yfITa`pLzO7K<rlAZLxYn(b8ly
zpOh+A^xo1mOr0DK6m~o3Z6eG*eJ-FDDw-}zds#GlEbx|{)rnObc<<%0zrOOIDI7Q>
zSJE0(K#Kglm2v_0SBl^yBx`lmJl)>{w$iTxR5u=8ax@JC=>Cey6jd-#LWYS`0l28P
z?$?Z~4Fh`tksamBgWp7;reR*>e3Y1%0zNWEWv_M*(HjMf`rJ-v6^$0tgDYuMuKBl@
z0~;xf*Ib3^-`4{VC7;)4w;i3%{1K^%yp=)=^Q`$sr;*?99dZKfD$dM8kWNU+@sAYY
zujjXk#<9+A3g9^RX{3BQ#;3={uiZ==?l#~bRtrY!fPM{kHlC-3=h<;y|GSmsA`!<^
z=Z4mQJRB?BJdPYIO8~=goY8Ufma8H}|Ndi8+_i2G`gbNA1R(sWTbLLK-|(y*upRJ-
z;e;?o>SKb1p>=v|sg9<`&wsGfD!1m(fdy-B-pCIK>zng+>#*Zo^j$4JB!-k9PG?)|
zIAt|I#ap_-TmS0$X`mxi{I)n%D>q83db?7{f5A#}RvFzv243;xGKKr$JTXi^3=OF1
z*Ynn-;l5(7ascZr4B^M>^vQeZNQ%!rhO@!VJu5hm1Z;L<Q-Z1D(UA-k-t@YD@hAQS
z?U?`pbiJw`&|P#2m{e8i2k8nM>`4reao+Iv<6=Dy(yCX=w&~TO;?Vq?Xi5-!vT^Nz
zv#v7sq+;kyF9kGzj0>*4w%1ucXkXR3+Kp{jt3m-LZMldk1C!+bU5ol{`LjOq%uRk)
zom^r&Mltu!EKL}8&QLJ%u)k*$(;8qXz?rkp8R|D2fS~}R^PV#lM&#97_!L4wLvc#<
z7~xCu?5zM0YY>3H>n4+03-A>1wF`tWx+0t?THt9qLAnB)prSh~`84np@K0W_RRJj{
zU<o!SZYKZ|@gPj;slH7Ou7Q8L#ixsOpg%-#DgG(K@=p=ZiT@Ol_zw}ne~P#felCI+
z+ymC`fTm<i#z#5>f>ROyEy7&Na1EFgAB%G-0*V8uHiPtq&m@VH{bvu;O<}w1m5YPb
zjWVz;+ri~qZBPd7j+aB9<YaS;Hr)ztn-8iQV)y8(l_M;C0jjb$cVxvGGar;36lw|A
zf19`cD^3T)QckRxMh{U?NofAeW!X>ze>W;5BQ7KzKOH!W<#vh%fqQTO&4D>n#Rkn|
z=Ubv@B^T>A)E?a+gg4q4)}VP_d89O2i>r8;KI+t7)#gHvHQz!tdtdo0ZhyOd1IS-p
z$>N$Q(H_pla~H|X&sY>!rWdOE87M%fUrJAWcG>Bh)vbQKC(T==qhhyMOx_Sbx(D<1
zIw~{6hUtgvyCmWYz~&L6@lErHqBuBVI*=t{#P!?{LeSv=0Otkekx#J!c&@apPOWHg
z62wq(3p6xFJV?3_t5f4yOW^$F@0;iqNXda3cW6EEBhCY;|M#nZwEG{qQL6)VF`EiP
z?1Mno#E|qW0Pyx2;F-g({{ZU$5rWzqL=oA&MD-8^$FkHtNwEvcY3MIW6p%sk!0&5$
zb9?eP_Rj?<WlDkpLxN!;pfKPU1WWGM)(W)1HsKf{h1szJG>S_I{O*Gs^GmJL>t3e)
znDQ*RbbVV$OPByEE_f4SjMG5_k1n(7eCS5hgaa;v0moakyuj}a5wLt$&6?yy7(?NB
z6s%6>uK;bxZcC~){$5%$w?C7o1!ytQJLevxaMNimGyG)t(-N<d-wu{}aeq3Hu7y{9
zaMbL<fLJIku0yue$S;CgL{-48&71?ZJRGf-$Y1i?%;0-8KR(T$s`077(iR<R{9--4
zh71Xbm)K=t$}j`;g%GG$;54FPq^c3#YZaHbIziKJYA+7&587DagAd1#d)*6kD!>0>
zqZ&9gIP~Fc{hn{za71G91!vG?do7{R^Hrvg%oh6mk)q)MIK6+})5;uWLeP5>><AGe
z^b?wBO`6!c5rF?8e<ODchSWZMeL_J6oDGKu`^I#}<dcAR`+T`t`zcZE2Ij9^l3xt^
zMgV7xorB-6h-MCPe6#!gC4lLaS<5}(V4^d|jR~-1jaoDU22(JPehL{^F1PLkPFw?k
zbMgdg7!FPtf|`!pi!Z}MHk~Eqnu${|(5xC`_nP)d!~sQZQ2|ZTw|Bv-zyUBn2p>4D
z=YPwNIzKAC((*pR7&w;#mH@&FAR|j<^M>8<okAv)k8t25_R;xM6(q27D+3iCtpDb(
z64WWYN^ZuB)8i9PH)lKv3vPLr-8$RK*Y2lk>QEo7sgkQVt-gALf7&I&;ntk6smu0S
zGJQERF!f@zP~irlo^|ekm!5SXl{y72)=iH8^4FHR+Q{=S6Q<njNq^P({4RHU_~N|t
zmgy?1-z~E?sIp0=v3vQVS&`rIV8t}R;9EkMd{t(F@bBW;&Fa|U>j8#Mj(OD<lJ>Qq
z9vk!cn|s&&8m}w6`(Y~JUDDlDH$BI3wz%7FoL9eg@f2wSS*Cvf<Q@6p$e|5Mwe_Mv
zNt3~18Zl8(ncTF9r7&_8q+yd+h|ikIaEZ#}%$&(GTZ3BJab><KQty%+j#QYHtnO`e
z49*1#V1lrG+C6@9&}2(*XCXJ*)c)wZHdkOxrc}Y<kL~U+R;R;8zILfEPSAO>@^2U`
zJViH<0<|yslfUsFVU#~$KBh8q$e&+6u<2isw3_|k>0kBsL&k~WY{2ndIlG|}?!R=v
zbjD=LE;4GU5{&fQ^KzX@SH4Nqo_N+)$KLf`Gnp*8>>*NKFy<&&Rqq@jeP*|;f=j<k
zbbHHQOf#Ohr$~L?l*wtAH~QZ195!rVd}NU6EB3&kGtIOR=t}jEp669NBu3#E-RM$2
z%P#fwV(UQp%`rcn6hQ1m(B0#(vZ5dVw(B_^poTXj+t;7y!rcic%rF0OtQkJ?V*Uq7
z%Ai<Yy+OmI`K7;XEeF^dpbVO-lqSqPbH}CY7%>Yv)&semfpMp4bBViQrth#^aofU;
zo`8_TNU531C$W}zFOQXO6PyvkoGbkgNmKlNlf?tXjT_~otBZS&k%86$A#Z($rT9!v
zsr`ei{;kOJS7cSQDz&j^LJ!tFe17gciLF~42=W{`#(TJ1=4#Mt`f~1;@v)ef!P6&j
zMdsoKHLF*nI=3bc5yMrpgg$bCOywa*W#RMv@7@r(k~x0pmCWq@7Hv2AVi)ZfNI6@>
z?y3h}MSjDA1Aim_yLfmuKkjkM*==47{)rQJR0HoEqC>*b0FP*;P32wJ6J2u8b_*(|
z4q}nRs7r^Jw|;Yt)pR5O6@$`Ekw+7QVOyy<*J=ZJ?0xSIt2#Koq}y{9OE>Xe2b|lj
zA>vwcfK_hz-PGsPW$Sm8RfTc7{c^s5e~hYFU(V2;uPisx%=a`z4>W$;`NqBmyV=!y
z?{7+0<wo`R33ruX;uDYTo5qdrb}28^CDsmNe+SvxA)-O7T4#0PZ9qzyo5=Wbeax80
zHn;nH>2jQzY}ITc!=Yl&7uR)3tB0uLKFYi?t2c&iOgm>7d~)FI_<ww6_Sg4j|92Oz
z;=z8Z+ZW_HI-FiXM?TvHRyTIgJ@+hRQ4^Oye{KPP%GjB7-8=aJ7Wym%I9`oroXNQ2
z0tGVTgpm48Wl~`E6>y37Y;3>&3JUM?`v~v(ayUHSc=(m<ebWO+hI{TqLaKg`Ps|#Q
z-Hx-pt()c}`x~qGgPh9c={^ACCp*vk-_9RQ*`ocvl{{TC7rO3qu<FQlZJv6thsmj3
z8Nd{_8Mz!Xv&C|}XxJU<*Z#Q4U4-qb!_j`9C-j$$XVZ(&bidWS7(N~DnZu$vVm~yH
zk+5fC64vG37W8c84AIf;Y-!zd#rg4{Jm3cz^4brFSDC~UL<&=`Kv?mCg^EV|2mW2T
z<jU0eQ@a-Fgb|wcfXMh$rvs{3xt;@DhK3A3&Yg~6bBxkIXO87cr~<iXJZn#I6ilmp
z{UZIe=j(0uKM(x2%g-5q*t;p)q}>O?9|kN4mw)#7o_%EdntR<Z0c}Gf9>>FcR7ri4
zk}>1VQof_n-o~erLE8=~vl&)kGnBLBAIP_Ub;A3ZQ2wMSKBe%6{#Aj~ZbSWNXJ#G=
zV_gsA$(q~oZQlz#Fx|8^-rg_#-B}@i_kFu>>nii-DeU~o$1WoRcH6jrzO_=gW@Nvr
zwp&L!AKza(IGv@Y^<$qMV~6B~+|jq}>l;@)gqp6V<v!Q+>R1#@W~}{g9BZuBZhL8t
z2!yfjXDZ8&t@cvIzlK;c<bPb7SKTd@81(txG~h_dmOr_#xjXi({pZXp{NTW!Of6a%
z7pZ|a)_<(n914^g4e0#IuG2*7yZWX{@WkHWnmsE<x|m^>pzW{K^!n{a4Y`9I4!Oa`
z?(aI2Tl>>I86z7d0p>q+Blz`%Esz>JKpOF{elz*6eQEh$@3ntQ#=%jI%0m1C{{I6@
NU0GYHOu;<(e*s*RL5~0c

diff --git a/16-bayes-mcmc_files/figure-html/unnamed-chunk-21-1.png b/16-bayes-mcmc_files/figure-html/unnamed-chunk-21-1.png
index 754e507ee655067f0de84e73192ffe49a8e2bc3c..09396acaa254c565d6adaaa4cd9fab16756fc426 100644
GIT binary patch
literal 106383
zcmd42^;=X?_da}P7`ld#?p8p$Q@XobL>d9<9(rh`8yuvOQc3A<0g;ey5F`bZdPkq<
z^M1bHKj6Kt_lLPSz&W$my4StdzW3Vu#As<K;$l%?0RVujtR$xc0BFcdurvk=@<$-6
zIST-w09tB#^2lF7AP^V~MnOS=Kp?27sAy<t=;-Je7#NtCm{?d?*x1-OI5@btxOjMY
z`1trxD3pMJfRK=oh=_=onE277M<gU9q@<)|WMt&z<P;PXl$4ZIR8-W|)HF0Sw6wHz
zbaeFe^b8CPjEsy-Oiawo%q%P{tgNhTY;2DoKYsG$2|GJG2L}fyC+E|rPr10bxVgD`
zczAeudHMMG`1$z-1Ox;H1%-r!goTAgL_|bIMa9I##KpxWBqSsyC8eaKq@|^wJ$oi2
zBO@y-D<>x>FE6j4prELzsHCK%tgNh}qN1v*s-~u<uCA`3p`od%simc*t*x!2qob><
ztEZ=@udi=lU|?uyXk=t$Y;0^|Vq$7)YG!6;Zf<U2VPR=$X=P>g{P}ZhYwH&;Uf9^!
zynOl6*4Ea}&d%Q6-oe4a(b3V#$;sK-*~P`>)vH&ouC8uwZtm{xuV26R@bK{T^z`!b
z^7i)j@$vEX_4V`f^Y`};2nYxa3=9ei3Jwkq2?>G0V4<O*VPRq6;o%Vx5s{IRQBhIR
z(a~@?JSHY4Ha0dcE-pSkJ|Q6?F)=YIDJeNQ`OTX*DJdzbsi|pcY3b?d85tRwnVDHx
zS=rgyIXOAGxw(0HdHMPI1qB6#g@r{$MQ`7}MIaEx#l`R5y(=jxDJ?B6D=RB6FR!Sm
zsI084s;a84u73aieN9bGZEbB`U0r>BeM3XThYue<e*F09)2GJ9#-^sG=H}*>mX_Am
z*0#2`&!0cHx3_n6baZxhc6D`icX#*n^z`=j_VxAk_xBGB3=9qq4h;<r4-b!wjEs(s
zj*X3tkB?7GOiWHre);lcYHDhFdU|GN=Iht5v$M0`zI~gUo134XUszaJTwMJA{rl3=
z((>~1%F4>>>gw9s+WPwX#>U3x=H}Mc*7o-H&d$#6?(W{+-v0jn!NI|gA3qKc508$H
ze*XM<e0+Rza&mfldUkepetv#&aq;WduglBJtE;Q)>+9dYfB*UO=jP_-_V)Jf?(Y8n
z{^8+aqs}}CnFE+^O2!@lKpp+}14`OHeF*?`fU?{(J>R@TbIiPiS<2wS4ml+_l+Zz6
zr7cZRz=R1?d+}Y6>>@`xiS|30m&zr#j5;RoB+@U<UpnZTv)B%Zj@=dRO6&%%imcxK
z<ekeO$U46b9&1{0726G(@H31RMRHm=O<7BM6^0Ccxa@Ej!QTs6@<>JinSs%~Sc(8L
z>F{0>Cj!Wnq8eeN0{weLTNsR_nHyWN8w3f@nwG*k#@`^kjqmyY?-wvj1L%ZPTmyf8
zTL*a)Omyb$wWi@M$#`~Ne~7i$(f&Ev{Z^Q6ah*ISsa&<m<|6hspysz2lQbqlm&E<S
zj)^<QT9bTSuSSSeT#fRP3=)t=M(14!w5(0!aGs4GEZk?AuZg^gq>{41Yev%ZWi07y
z$XXrsb1_!f$<Lj-@M1ZPBHk%p*I(9^I>cZy6%?KJWCmDI=e>#LR<5Vp5Vr~(&~dp>
z1V)%tA^^N9#jE{KoNd$l@^rAC7+ow~<lO*z^bv?z4yfoMkj|Eifp%$U>IAemCO5E@
z$pEJJiRT`ErW@u%1h8IjNE{$oi%mYr1f4%M_4Dr}4ai}H@y4Xk!{%ZP%0IF1!G-ui
zTdM&^`T3Wfhv*!Z3h_Ks_NY0u`F_%oiuJ#m*i8&qD}FtaI$PhgU)|fvkn<9gY|+2V
z`D95T-C+<`L;mfp_9sER_A6S6vhZ+BbwF%N)sJ+xNX6x;S?T52ky+;%P3G+$rt=5d
zh)=JbLK;@jb3%G%Zhw-9Pk7~0Q!vW{s9p|fW{;iaiupl-9c|8$#y;{W5N-{QP{4NX
zjVb$U*+=G^-Ypu`-_B|mZcrd67A{_HjvDXq@W>-I0Y1Dm-1;+gBd%tP8vUSW1h83r
z8IzWFsfA-dNb(7PH~=1OdG$fI*am~npMd(Qp(LG2DaIP@)O4$q;O0Pn%+G<R60Gl~
zn=)N}=R86bin$2&TepUXGxZp99?8#@=g1#LX|~x4&3*#kxPXm}lPWELjm7y0G34B=
zN_~{gj@AUm9ga`qcJej8;`|^bnC%E0F;~=?RZ#;%gP}7c;de~K#|~g6@0_2nLO0Et
z7V5QSAl!uw!JFwnD$Ml=y3+MRS;3<w+Oyl*slXGePBd&&6J2wf&CiC>i_XvH=nf3&
zzC8$M?G?}834+8mE=0ZWbz+`gYZlIRgHAAEiR1k}##GSX>LQN6QjNa*m)K9x!K9@y
z8qczVQ?{=7K};g!KmK=QEv;yUa}_>{3Jfb_=2i=uI+(rES0>o&)sr3@MEx0DTW+S=
ziGN}2hw8}8<9ejW%_&cm!{<1Iq8~_;_SdYsSz%|ZXHkZ5iv)#97L&fJNEah3^qXgO
z>4O)kOr2#pAq_L0KG-z7t>cLagbc7fR8OW1Dds=Q1YNyyofV8G)AdL<6EkwTN=-(J
zMO;~1<J$&M4&ps39G{N`=K?>Oa!xG83&LgD0>N5Gzr_&0r|eM<-r;2pfyL(W!`_RR
z7Bc;on7vKSFKshD5uS7Lc1|M(eO5E>jN8uLhm|wHoXX!wZg{?mlLTG7$rYP^{s-RZ
z3HC|wnf((k7cy|zglVPhh45xv`EE3(xw{6%D+K-w<4Q@N?WF8h{buAs`#*h2nfd%*
zno=^$1G!F9veqq+aAzQH6y$8A0+`aD_69s_?i2VbM0{{1jI!e_2$yJ2!2a#1ZQ`E(
zHb)_6S$KFPfnYKQp}MWR_&rRoE;N$)?qI0%HqT#LRO}RP@z=KXQz6m9OGrVY^*bvx
zSoR{H1tvR|p@_-?ice-0ET#D))T%A}lcKZg*Y7Vc(5}!o0m1uPSCm`tUZIE7T~9ae
zbs`wW7vvBbd&)s_0*9EF;a2QP`>e{rngS0@Wq!+?JAaX~Pa^}3mAnp7h@M{2RtI+N
zC$$pDz<YbNQ;DM7puN2`_o1i6OSDXY)o`CK{Gev!R#!xrNkJIoJancf_;3Q3_`f2M
zx8v26wMka<{ji*k2Rv_-Iv{^lW?GyByej!4$`Le5(kDIu?dsQa3O6&cOntg@e^Mm6
z8CdyIz_wEz7|pEB5E~?H*IJs~TZ768gXAcC%=&#3P`&GDA*Oz6Qp#rmmA_fV)P0)P
z+>t}fBZq)rBJiXLf(T|I+}}qrh${%(Q1l@u5X=dy_;m%&R)r5>i-!^hVwG-WqVOYT
zwkD2)I4i<?L!yKsUu6BSusy9qsSnY+4QOUGQ*H|!?}s&JmN|X%eh7*Qsw4uox-Cxj
znjFetHt~pq&yln03u(VRh&dYsX^#n{pRq5df4_`w`(lC0E4RbTH2miWHzM+Hehig9
zu{@u|vF;5=jua;hIg#~S%*EbXW<+TUyQ7;QdGSX%WmR`vYQdqk4K>MQ*>+H2n4l|)
zLN5``T3H?(k}UsdGGwb{>ix?il?%KP|MgrcTrhlsq4IVA#XicDQ6jmya+T<Wz{>#z
zxqR8n?-1_H^0RbZ%WO}qDa^-qY2I!_#2ZlQkF--y4K-;!81z?iuONS(sj=f<{>?{q
ztAAd){g0R2h9#PHMXBBGQ%WL*!zq>?b>ag|u)RI3Gq|YIuke~v70#BobFImDAjXp}
zj_pOa`J9IXZE7m<K1LYs$a%wv$L@%Pw-nrEJM?dP5vPSSmF@q(rCbSkV07%kwh-OR
z%e|~J-Yn`^Ek`+aHffpCB>~BWGzZom%$l0h-)+dv4Lxj*?4<wj%pz%(+C#lCo^15;
zUOXQ2>;L14hqn$_1g7@R^MB|&dfp3Z%LCq07!KRse@W3nWR^0c=M%Rdg=EDL8wfl^
z_C-MD4-FMnENM-;7%}5tW#T;P_I~Lt*<Ug8JRr(x8?PVZuYmc^XU6Wsrtnv@%onVi
zw8-)x_=&m>HSz`1E+mA7iMjs$JI(f=rg57``~Otj$iAs7`mPet8x?ccitF#KIusk;
z>y7u<O!=N7=Ujln@TJ|=$OKIu<%f;6y7peAppqr1`+0f@fR^$k*a+`7&km<}*pXg|
zH8O(d8(uo)*pZY<Y<kaeG&+&I>J^yjT=-a<O<~b_dWd``uy7fr_pWnd?@ZH*Sw5}d
zQVYCV%2d<hMOkBP6kF4B2bF%#bmp2<_A36}@0~t@xp#i!xNrC^e}B8EZIMjbi7{n(
z^J(A0LqCU=M2=zq;MuqL8s~lEZ<H)@%1WbRBmyoEt%<re7mBZs*A7KO)4yshcBr?`
zEk(JecH$n<FFQK=i8vHAJ5bhScL4F#s^@x&fP=*hfi&S^G6cBzh`F|N7<qf;R~}UT
z@7V-KSf3jtnS~0(6v+r<V#5UQ<=zQAm}dpaS}_ZxHM(wM(D6`#-eD>N%Ig?(m?~jF
z>r1ZP&O#eY8Y3R<I!T%{#y5G*Vni>FD)}{u=Wee$J8TQ=uN(?m|GM2uY|bY%B_PD&
zJWh@rX=Unh7lE{nFjy%pg$~Bo@UYXlYI}uUon=Jp?#4#Ilu?PiD`I7;;{Uj7Wu@pL
zrz;j|03LJ1Lj;iGt5^p9Uu$Y6?xP-_j?qvU@zlS3A&%jAE%bAuZz$oN--hTE+KjlD
zi?W%A{+~Is+?@(1gEBws`%CZfaZFPW1q`}uA5Rse^GLOwQpAL@c6sP?eT==Z31+W&
z`!R^p4x$m?U=PvQq0T4#UwC}QZ|Ls>0xmyysW`H^tPT)JTc^o`0YX2;GRXg0)iEJS
z8%*Lfd5}YEq>Ni~^d0Fkjv^56i#KH|G#KvdF5>RPGSE6ydZB^DOlaI`C$mf7oB<g*
zcgmWUC*wpy7zov?oT?dd_jYLM>6LA3Nhkj<>un4D%?jsKLih*Tcw=Quq}@nmO6VL7
zflku<>neZmqaJxOozCEH>OXgweSL0Md-zkxd%S~R$d;KIX#sM1p<G8zC@mgs{}0U#
z<WzW2A2fM*)qD{8P36gdv9J<jf*}n>^*c>wZ?#{_kEv%#$=%8R{(78Wb17=~wu#@|
z!-Xj&^{0U0qd|e0ql|u}Yya5MC<})FfKroAVB`@aKahtQ0^#2Oq9hEqB3`Cn{7?9i
z0j%i$cge=^KhV7q1aV}Gc&wE2*Y0B!`6Ai&m<rU_&sDhjaFh(h{)8SO|1!-C<EkXk
zuKmC6>DAQV3z<5K2YNmi>xMGI-YKO5SV@Fug_mmQIf+0l>3@I{A7i2*UY@|98xs?K
zgoGoOAJd9N`%dL6_qdZNn9brp2%VnrPj&5GDg*M~aBn0z90<F(5p6ytCjZjt6bZP_
ztq?@3dBu#K8p0j9H?Vj44EX*Rfj2B59D08QroPH~PCvU1NfMRFS=^e!e_TCtD&>e5
z`uLuA&WIEFTnt8qH?o9+m0*<c0!MYd<Q0K9{h{irXL6syjs1}c7xiv+r|`T6n{#*<
zgN~$-@E4DPDmqA9#xv+E0SuWW&9QqejdJ@j*&EET<L3pXtJN-5b7Ej6T;<MjZiG-z
zqZk<8uPC=mUWHWt_*6NLmw_?pF*R|bU}Th`nS|q#VfHW27<+4st05<XL|ukErKsr?
z93GHl!j$0LA2GI(E<O-woV=Vw{HzNj?1GKXRbSrG@EzYzR+&m_MIaF%TmvIAlUV;S
zw>y!rgTx0RCyg!pqKI8;(7Onv@s8F)r)0uNBhclR6J}i)U<AezDYRs9pF3kNA5W9}
z5SdeZLDn$n=qDKcH~A5@Gi|o@gk466%1~`bDMUpgp!00y_*;|T>B_HeMl=jwTrrOu
zR@je&#4E#lBCu~^d2W5HI;4&(ee`JMcP*kIrzyq$m?57hZLhMN5l{p$r_%S9U&+yy
z`HWjL!|3JE2_>Gt5{<S`VT2K%+&q0`M_DChGezAk`;<9{!27^k6Nr3dFBNhmMGw1z
zw^CGnmjm9%loBjke${;jypQ!9;FL2|p#B>&xHCe2mxz;JaJxOt(T+m_eBEK*Bq_Ix
z=0EaSZW9?h7%99)z+jcGRoEAvr7l5<{B|Y4ho<9;(RY=vi<UQ<7;EMTzJH*!!~2#5
zXT^u^1s>=ejo<O15#qxC6Czf^AeMnJTZ4`ldl7oI#(@wwT<u2)F?dv#X%Z8~HU=H)
zSn^Crf}h-7STsg(km%u8YgQOnR3PP2+V%G>+o{t5N6?hTy-5zF?sQ!QxvFi7(GW6d
ziUCYwpW#j&$x8?v`yfc<MK?_t^vpc}Mgo+`{O6KFmGV|ZyM>t2*A~+%vV!!epJdJM
zCFNzE)RZ|gCS9H)e`d1eJ)fN<7szQw=9?Ts>PV@ff5*I4_bm6T6Cl<|1r)TtMOP4Y
znaV8;90(FfV<W_3UU42&_;nHpuuqd4LVqRw$H!<Q&X#QZN{ffRmJbnY`MG<<n>pD`
zFx2fTuG2Y<$lKcHtUa^yEL-cnEu_C{pR&)5#gb#Dmah+3V)Fq<o$$5NzExXh<Fxu=
z6^kqu5Zw5<`YWat{glJMaXskoVqQH(j>`_uA=$?{g=hXZfZ;I)IIRA9?>kca8IN;y
zKHIWjt_dO>?iUG~6QV}p8oWsrs{BI)(pbJ>d%Q2au%>=@%eE4#V~Ty6$FyRtB>m^^
ztD8>eI!}6Jedw3tdZz(k!$m3ovL*ojQ=Y%Mylf|?<I6iGJF_(m;BU&^5Z?JoO`$k?
zag2tDa`ZddrT{gJIeWE73Ezd?aHr^Zc+%lD?vP|K3)I}Yb|5WqPv*jAzM@|Pm@R+h
zrn}*KUiJRYLi6jArD`hhhhSVSA6TIiT`*c^0o%)XdX30(Z04xOcympYx4+w7{aYg-
z_R8A~_(Ta>Rab=h5~Y&*OUa7@e9D0Jfao_%Cjcy5Q5U9PX;nP$-SRgl;4Z(Y?Gzh$
zHJJUy|B<OuPptXzX_N8uxDGG5fNLiQLKqp1RG}#7vxX<)Z}NRH!Wny+QNTcdOo~`e
zp_#2+@pNqwo08Z5R~`_L^%KM&=X~py(0+&8%{Ws$NW`gmgp4o1W7RTgS%m`@mNLma
zAd8mkYs5wt4S@dihN238{!Mj!YBhOg5|YkOK6?E1{6oCG;R}IDov1QpAP8Qdnh#ho
z9@wmOH$uC9Zp_SUv_7UkW<~rN#xqg%{MhumbdTe|7J!bZ{4!=Gov*3oE@Up85nQ@$
z$WTjvoLgUsCEFG#mq~(Ss`zc5#~5?}@<c*EFt75^cBEtwRrGtXM}w0Pv^4Wu_XVc_
zxLi&kESE4Ow{W;9zBYK|Q`))%p|mT497*yjxk2*3VaM%v9`l}7$MjQmrofT!yT}w(
zZw(gtTWtZaa*Hi_7ND`qLj3~kqS%#HOw)od72<~2KYtL@9z3>8Qut0pPJ55-p1VA~
zfAWU)fz1@cJ!0?&{}ZDzId$MnXCwmJwQ6ht({qGqy{SYOV!juYxzN{+i)?zS@1Im&
z-^r&AhMtex>9E3n%MDUg<p2tK_p(zpSm5~trFgFzw#Ar@94VdZBPMtV9Pi9PBjtni
zzaSNf9?|$EU|N9<xaj?{{TM9O{zeGKsNTOPBxkpNJZw<V>eKIHTQ;*UHhupx4)|Gr
zPP*!hq)Fcm&BM;8QELrOED*+98#LTf1z@|v@gGxBv_(E)utwHrDATEfg_HL7zhbVv
zM7)UCmj|Sxh38n%Fp4C#ccpz*p9~SXzSU?5zvrd46&o=w`d7HMLQUjWUV?q0Ro1q*
z+Md2t=-u;w(*=+JMdpgWg4&cO=}j~|Sc$cCX5FXwb?zuJXz3-omYFNkDm-S2cB1?z
z8G<IYAK4%$CGROU6MzP5>@P^fR%0e>4#2;N<6aC^KhI|Y<*D*Zc!9GNJ=5wdG1iC#
zRFtXdVE%H0%IM=rN(ZP|w0M<d^IWhYR{ldLi{x^38y=!ArD`8}>o3v7luVsSHd$%z
zf22D_RNG<s9QoYJT2=2QYed(oUnpiXz@9fwKd6YMq`O0>+%6mLnjHEw6OQERVM#+O
z)5#Pds6NIkm0OoS&*Ze~pY$Y(#y`1{)kH*Y1{uaocp|^SBM@9$B@<!(2%=?pYoO#<
zYBO|`QErRGi+hIYQ>Np8VwlIg6aU6F8+bL?nw57S&prLTB0palJU^~ys>#xd9a2l@
zzbs`F);!)jvwt!Rm*muixaGdT=#20!cYPkB@yU}rx+TYfF;)g}CGv_V(U^JMrpo|(
z-yt(7i>x~7*M|pHaiL<%ineU-1Q=_zkM7SF<aBjeVJ(Pd@~T+C0^hGH1_YjuEI0pW
zArXJGZbNvmQ24v=gumkd6?%*ZLTU=*1H}h#V%hDzJiX;KJ{3S9X5!L7fk+~(bB?He
zi<V&^_^Ti_m`#(vR*j$wYP!FgxtF#^?*f1IJa_$v_g6P<j5U2S#WH_Ml*a^JB&P)7
z-w8t~=kN80TYqL%n%pO+0D`Yng4YYZpLBMaip1Hg96;z;-W??%N#}v$elQFILqALf
zxv(y$C9~SsnIKK%lV!S0@1{lemO*`kt8$8p2D~6TAKwdZMBbiQ1)dhK3h-zwkiDtG
ztUvsHgT8g2ipw+|4D7x_Bt{1^Ggw~{4nSK}d7z<3IC3$ZdgI$Xex-mr>iS{3;Q8aB
zucXhZs1KMQsl)(+aY+P+DKVYnWm~AFd9ryb<3r5dkQ1*Xf7MhABWSuM@j4EI6^ZDA
zw_Ga&Y}XlE2@3#yWpE@<$z~)!>$nJdyI~!<PdBTw82VQ<!jmjx5|iYfgaSr8w9+3z
zrR`*8-6u}A$pCcwhff)Mj79&dSan{%wWKsB(qM7-6?3G)dhw==2WY1IAte2!*RM1@
z8O0JlTRXkmZpx-+>nBPi#}IGLwr?;~N7@^cqz#?Vxc(l<FrnoKZj0%Eo*^qv)O9CU
zoND3wdEaiSL<zzj=k=0?7^L<oUse~ukRJ1-jJDZPiW29+kL4*quV3O%;!m9l`G;-y
zd&SQWd{d9#3{B7h-jqvm4Ol#h!jcOuXF23duvr_5MSmjBp4B%7Jvpj<JZ)vqfT}(}
z!((5FmVHV8N4JFh_QQ|KT#FHVqB1V;{YrnChvde{o!XM3m9FJ3uLTvSVOWJ<lXVZv
zBk4-uC!jhRRUd0l8p2A{m4C^^LUl!mdtuN=>Bry&_6;q6k4$!}up`Peurwgh2Vr62
zQ{>$WBC!_7x2_NoS<K(d`zIey>r(dKxyDSu<-Acqd7&-|Z|%&Sl6(lw_`)G{I}BeZ
zOo{H!6>o#OVhzIZm6rK2*6`geuBej$tc^^Jmgxx#?=b4&5&I$K;meo$jTNV<t2^3=
zfRxoj;RE=d{)!DfGuTKm>*b=-t&1eFT{&#CrY{4#m%adV&6^KZMtsROr%{whqCgZh
zW@Da>E)U=@maY>6;Bt+R9!j=}HbbUVNQ*b@QYn8ZYjXcB<vpkE_A#9Dy76*39y<cg
zRdHx~Mv^~=c-_Wv2w}~XN{o@k@cV()XwIasx_vwXNM2a>`3?T`h13m?#M-{?%x|wo
z(T-lDmyyJwb0mr!U#%uNpA8JmRZO#VvlfT<)n^i@r&K*g?nWGc-KRfo^0Ul44K;&+
zmegzoMOa|;kL8ACLibp$gtP6f4<IMv*)QP&pe4#7{ah+Kn4{T`pNCuM--m9P`h02n
zG;OEj7j;s9U?A7|!jkC~g#n5Tb~=oY4vg}{Ih@MA<X<C+0YDh9KYrjMU@gnFabUP4
z9n{kZ2cXJd{RfdfvQyOvGwbiH_}Jw1bM&q8L`Rd2V3W=YC0&nfzhjNykgG1S9WYl_
zo;<)Fj4#D%7#P1csCnxGEx*n%llz1SHIO4F>PoOaml}R`sl#`FDTlyB&gC9EYsd!N
zbK8x`Gyn{pmZ};eu-V9=jV7B6zkU<XPboHBg=|(qpm^kf8mT-(Nc(jh#aP;|W59Td
z6Xmu5pAz7jO1{2o<R`Ai$Jh7FR@;DL?$#{NFaT+CAp7IQNNksUO;e<V)z%I4M;KJE
z2?pvE(6_Q=&cv*w-`+<d7B2K0<eb62oif{8GLy#ak_vTiSnmbxDQZXh`T93-ceo7+
zy9Cl?{!|u}zAYk}Dn=TOQ}D!33qn`Qa8vdNn~)Be@L+#bBmZngk$&KuE*Of!<XY13
zIri_&S`Rbw(<kiOhVr8FMg<l-Ps>=mV`4`tLZ$PyNioj_X`YML{_o9+$5)dGOFWgW
z>d(}PK%Y^ocAj~VbN-|x{o;Y6ob{oF$3r*Y%Ml=QUyd}^^&C2tKYZ-UeJHprE|#DS
z^opBctXYOt)hF~p>XJV*l-R-horcv|M<Cp=4Vy+j_tuI7(ZszsNJ(JK*Yk}L_G7FM
z`!NWqYfS5pi+=_1R@BtAu>EC;Pw87>)mE_BU&O;lc}Q}YG1p8;cptTlOn36Gl`v}O
z{SH|BTS0b*;hsw=k#8N+sK@UXm8H(^SbVM%yG@0UGm#G5e@fXf+H?33c=-_}n=CzI
zX_r(~Lm~bNT*R=+Y=TvS!B>H!CbCVM^wCE<BIyuUCqQH)rQ16-3m}2>*^bE5_w7Gn
zXUf0?em)msX<OJ-M~G$E2O=rXKV>mc$p9Nw{xnTSEPi9Y=M9xc4c@oFI9>L%SLtCm
z?uBmpL04l#@t$@{J_)x>M(!gbWjWa)7T=<$G7xSw^wkD^L#o8?1DRb=>hxdwqz;p{
zZ|!~FIghE?i5y4qud(R5GVR0p3`}}SF0Er&3G`<e8VZyC6p_q{$YLKQ$RFgo@Mibe
z(?g`bUbuUbjJK5-RUZnvTc`2ko$7r(%LLF@4o0plRzic?nPB~$ky1Dfvz7MHXM&!Q
z6?++wFPUNcvx32t=X^)%yQ44qj!HCu8JB2Nd4ROftG7MU8;g@10J9BpEn|;Kvh<6%
z(5S7`r`(qDu-`qhQC}n+cQDw-6|o!;NA^vTp6M0G9v<K>4ak&AQhp15!+9>Fxtp`~
zH$>6HU)-P0XmGe5@S}1;6Y|Eca!uY-V0*Z&<Q-)ISWSK5P%@{DpGyrX<raD7<}rAD
z20w_XiG{(SmfaKV_LTCr&O1y*@7$|}*R^zLHxVjdS$243PH5`n$4F^Ys{E>M&5z&%
z5i5(98!4UWjhOW5ZQJgZ1WQjrM{Ap}Px-B9yaYn*Rvb@hv#QIyCc62#ulZcyNCA?_
zL_s->-EM{fVHXrK@bI`!nnI+D{?Azes|AUkB&oQ)^SUESIm)q(%;2yU&j>Q-lGoTR
z+zXNXQ9GWWJHy_Q|7e7tPUxF9($$rIm~WI!SrZ`|MK*pMEIa~OVLh+4SS8sotsq3w
ze0Y3%4%dTaS+(Lw{m=DNUUhHMi+A+u)5@hJ&lzR+)s6cU7y2(kiH>DL*dD^+ncB5a
z%!eO(TfS%Yd{e=pj8IB<iq0<{!X$E=^oXE-S{}~kVbOb6#PW-$9waHlG%B7xht!7y
zoa~U3t2<HWXsc9GnMrBFdF|Jya`nEF+UZ>cUvTmrUm2-<G(54Rzz*GePXz`mkt1aJ
z6PKS@ye>Z3>@Pqq<IkaP{6V|$LzGO2SE?ic#%$woxbyn!SIf|r+qoeeMDQC7;?n$-
zJA3Pj0}Q%xr&q@inAKQ_HY>C_6SD|+ClRAHTJ*a1OfN+B$=nznXelvZYVsrlPs6-=
zmdG&f`N2ynne$guTtc{=60h5sVefM{pT)<|4VBc^1Pa+6V?6`hHHZnz^Xp9#>>s)a
zoQfDS4(t&%_m`M*>U{#o46U=Gl1b&kV=p7tmVG{u{g9$F^f2@b5Y(e<Wxe<YJ#FHs
zQ`cHs9494eug3e`y(ngy$7Ax;n?=yv@Y3f@;FOLvs|=&HvVq6#P3wksd-|muzQSu+
zS8j;4$=5z3^EUtU2=m(cK;EfW>x0;~K{_9ovW_-H=7uchv|UAU{IGPpOvoDUW0ni#
z6Yd`MVO96sw5FM&=W{f^k7M$k=tD#HSO3Kxm|qW0S{dnAmX`uRRAarzr0HuCiT2O`
z0}AZb+w`st$yaNUwO%F&zL-R!WtF8JO8U;LZ+Q`1frP;~e~`(s(c1)~vfb%2ydIB*
zH@L+fAuLw#Y{1|A`C!G>b?;&}6g^sEPh7s8{(k;TNh6eY3?%nrtMVggYa`>~oo@?M
zot?ZD<#*sI0i?@mw9;;z=Sn8rrxQ7IUmy!(9j>%n*3mAOY;SF7oE}ecasS7cY6pU_
z;l^GUi^M&yZ9$)p4b$0BE^mZdpVf<0MS%K(<ry$p_8QwS3l!1UWt0dMp7D6h(>3JT
zxRrfCs#A;&9~tlWKl;fM)^O43J%~9xw6NTDzD!vpsRkmoh;)c&T;bOdVwW&81ockD
z*6J}}U^o`15F%dfcfN!E`A#?U+1ZJ_@rWK3y0fKlryh~C2a^MqRm(QibzP3#2gJ*b
zPEGQbK{E2?RENRO&O!K-t=Hc9rtdfBBAAO0F&?hl+tr-afYrIBeO2-~QX1u`0VEv~
zFxRY)4w1`(<Q7ssU^_Kd?5BF2l>;|On$HvFhegckx;O_jj4y;LtC=dWGQxhXOs6*X
zzNZ6Dr@v9tG$|D~k1O^df>1%X?3i8WQ;x^|+N=sIxdS*0if`p*e1ANw1)2A0cRMS2
zK<ZGSUGD^yb@@mlfiHrR(L<4pUx+wtG^7X2c}_AdGCkPp<!1k8MPAaYQ?H9vf}w?b
z=#=bdugFHUIYvjVTt{;ae)iZyzN}RXIi?p0T5Ps~VWV#ofs4+PEZ|b>?G#JKKrMk+
zp1@YLtP;2D-fA-bEe)mdCls1TruVT3jy=wAr*G=?_!YaPIz*<}yz=$W&<BRJ1AOU1
zvzoHJ3dq%f4UdVPPuZZt2~wqW_6;1ouQMaD0}wj+koQn%i60^;A!OWL=-_RFiVRRZ
z8X$YkTFqcX2NRTQaoHq1Q9d-Iq&#BBu=>OVOe(6X$<rwLQvS&~3B3`sJvWGpfO6il
zuFORq5{^8!q6QrTkUCSd5Gj0ldLYfnpSD|un?l`=-a~Zi3Ij-$IEqVV*m(XI4!S9>
zAj0;vC-R73{f+d>nVLz$GsXC`noUuVL^D3MZ{9cbW%-!YDl4f^?6Z(BOoAa`UvZ0s
za=dTJv`;#V21Vn$22TU3ZTm|SStojhzK-SN6xqwEQB6O)US4kBXCOj$mv5MoA~MrW
zpO0PeX8}*YuG~*LpcaMEK`1y#BW+4NVn%MnvVlY<Q<)D1!#VOw8>h2!=&`vPGQd$_
z=Z&%fE(KKD2@zsydSA{W>b|`(Z*PfQtT2J7N`<I5-k4f%wQE=yQ=ynvKkHJgR>&Xz
zWxGZY;YpVWcmv@1m=WeUQVuK-UHORse2L2mDlsXea8WslP**MiMwK~f?O8BkxGBfh
zdw)<VAzN15ie8V%Q54M4`1s)NYtxlqeKirJ5ljJV-<=-2JKza&3vtdu>H>ytAEK`X
z5Q!h7ZprYQaHv2cfjE!JN?iJf(n-$Q+ZIG=&iGhZ-r|Da9n|-sX8{Txl&uLHn1Xwy
z;q_E%04a2Q?UNn6hVN>E&J|^cW7+0oDwqgHrtm$LelL3rF{GXC@+B$iyXBH6Az1iJ
z$Xbj6CK%dfAy;(?H${{%8a_?rB>Oyk4i?pOCFJ*T>G}R;4g$I*_~6H<tsinEYfvjO
z*dk#CRS<<L0%8D2PmP8Xc_0fW;ULIiCnfOm;rbKR*E7lK-tw(nfoe3`tH07e{-_Uj
zFYSHaVr2JVQbt(cN#8d9NLHj9ws-s>SX!BmN-6H8LG8E{N4WcvY6xXnM;*YG4g1g;
zXUTZL?KFh{oK}tMjt_Y(6SLSio~%nYAh=3s&%j6__|_nJ)uJI=FnZ^u9hV0`a!p%(
z(~=gZ0Kp}u!yOpr1zpUpTg;&Fo{|UpwfKe^$;k*K*O5SJ&GxwNnj^8&*T(t9kG0MF
zOb|Dit7pDdIvoJ|NcX4h*%Hp3HQ2k>cunI4QQHi7<@&(t+Z>Z34TAi|=DCnD+8cZf
zf%d|6e?APVA7oyTI!0)hLD+;O53=h5ug;$M74p4-^vnd$DWo0XdBPj2N;zUuxTT25
zW`@0(;<N~>+<f#6eRy{n(cmV5#3G0>W<bHiROfdOieE1@v1*CkCc<fhLeg8_fdz(E
zAM4Sqq>uMX*WRujIv<#Wns^QJ$500pTs(xqpW#N|U|26Pe`}b`c;-Y$(PdOx?AWsa
zr!i5lIi6c#y$xMj>K683fl0Q0-1-PZ#mac1|Dih7C}O5_wG&x>`en2H2)W0(<-PTG
zP)<FXK7}IJhv50ixfUtg8&gcMJxi3$r|JrLOfmdtgX)<Q2KZXF!@ag+?AhbesEJ(7
z`b71-_R;3(D_?Jc{GhVa&lo>^AX<Mgf#93-)I1Yn@HK6>+&7E(;v`m+A(<UCo=0m7
z`}~p{Hu}lT!+$k$^yY@H*3(Uqg8iyAsYhF1yqXUMC%Yx~pNtUZw-LGSRmvNQM~f?j
zAXMZEGi<@-q+-qr`xCL<E?8mnxi?Nj&@cRMkj76lT$51!4P05gRp*S!ScTRLqp@(+
zn8A<VI`QBopLdUYA>3iqO8xP?<q?JeWzeJf0f<{i4TqrdZ-fIK(mFG7nwdwu=L~!x
zD~J^{Sr0#BQDI2TSX#k*g7yp~PQUb@Ce80yJOiu`ZcQI;tYt%~p7Ad|9cHA*kS1BX
z49ZxH())oUOQYa2`a;mP?&rEW%bbImE7}vrcfS7tq@!0iB`-%5<h-ANW%Yu)XMrhd
zk+v?GK=+=!+CI~}S+y^M?l;#<WMgzAm_7n?MD^1Zp@Kms-n`K0tMfN|F+m4}#Q4j8
zz4VB}QS+FC5-bNM1?O5MB=Z6`c-O|#$Iq>#stj<Jdc$LipEEjFVXhH}+@SO+2cl&g
zUw2_Ok)d2Eeo6`vV3MVqT#d$*&PnQblhp7R{K%Wm_6Pm_3p?=MdupG86MjiTU8M}6
zqy`@%uSOdhq{=N=Cohxkex#K|Uobzp$HKN*!vy&<@GyMZ81oF-v$uwaVSaES<?=NZ
z{C=7De80~gnrj0QB;Q8G<39w&Bnu_=5YpGM&Pd>BP+M*Ud2Ol`9y%j+<_pNl?zCRd
zi{flnU=KG{-(Ec*AaTe3%)S`*M2;3AN<hMk+}p_Z8On|#>l81z&E!=EZhT+-MlYT+
zNGq}Mq89nssr~cM4^*$Qw|Stf@(wbo3Uy39gWdYp00m^?m_u$*uUOa!*&+;yLU0?q
ztv2lA_pWAYZ<#`F?)p&v<Z}F$2;)QBZ;cObse8S_d%~k1Ba7_2ZoifK?9ZuMgT7ZF
zxmgljQBT;d2DE#s>?((<F=?jk>R^Uw8pR0)Ag31ZLD{VR)uKAAu|z2D-jD7};_CDW
z;PMjnXK!O?Ivp%349{KLv=Of;W9j@m&+~!AL!{z(y4oy6D$+H<Pn{^hxU=1Eqt+d`
zc62lThx1-Z^R*Q^f>o<wo<O=3LppF8C53q_V~8Kr_0CIn$eRe+A7EbsN6Y#2mfKFz
z>I}hoR(#cXuhNc<aTtCo-K!#d{Sp%3W}?=Y%c~UdmaC{q?3x`3eXJ(4YG#sv%c90<
zt!xXPKKTh*LmEL1&T)uV?q2?yh45X!h&$JZR=m+jD>|O}j)dUt0i#DE*=V)VDqrl~
zt@ZG$YAQ$JARfcGg@|m_+aLN<G%T?0K$gR^xrh~tkq?|ZB_G<yg4KZC{c1$gJ`M~k
zDb=n-*nq>k0c}MM_;m`kO<}83gHa78i{LsY>dD7o7Ck~RGY>4>8-xKA7%~I$N4A0G
zI&v}lhje_0YO>I-+OvcJnUIHhO`<rb^4lbMwGrV5bX38hscV+bsKJF8O3b>oOC*bQ
zA#@awp#15IY!;wgJs{dy=nb;VwngIW;%P^2Mi&{s=UH4DS5u3&a*pCSFZ@@T@LytB
zK?m9vF)J#vhG|uIujbkneieEUu(R+~l-%9~-HzsuOl&4)B2!C5jx{<oC~si7X9Nlh
z9QlQ!*kv)V(ka^;P13!NJ7VsiZ~Tl}BVX%)ORoyQibrZ%nZM`4vwUd+;zo3Ha`DcW
z69c63nm6@m{BaiGZ%_JB5!{mhrqLjwR17XZ1`d3WP9*KZAm)b|C^`Hyf{}w{Dcs(O
z5IjFB)Fy7jo(2IOsz4kl*<5A}vT^(B?iKdKUm&r0{d+)y=q}hj923Wr)&E!?OEt1U
zu{7iaY`c$XhKW|{p$B>mNq!8rw~r4cWSoJV;6^?sc0q1xrnaD35Wc57iqIJn=VW=m
z>3QqI|0i)`AmeX`4(3k-G7D`+mk3^t8*v#!l-|khmYiLEm_zwOE5vas^%BE^_~Etn
zm@g(4EX^8R4kE1p=z2HVm~%}CL!N5Kn{wHra$bn>4^;BXTWvdW;!bc<g<wB$3<y8k
zLS0yerd8X<sR48DM_zpgk#=c!CWpV1Gc_(0j*zrD$eNFqD5J&X5e^kOkNJM~MxwBo
zG%$&l1(xqNxpM;=g<!F*Shq*8gndeZb`>ldnG4^5zYj2BSdlV(o|SeR=om3nRr>Yg
zFvpoN49FN3TImsHb|@L#6S2OJ{-a+1(zdJneu*?{)yE1koUu&B!=DK7$^$1UA)F<;
zeCh!A1~s%{S?&PBooH%Rt3E==9681+XNNN*DQ#i$#pW+-4&!q-%=UwzS4Cp*Jr$|g
z#P6o@6dui;-`3h_5!fb#oMh||EN-cyIoQLJMmd;8tyZmT7TwcPyh}2Ssv9pRNZn$$
zhGHd<03D#OK5jB|_`QL@bHe^8Z?GQBaYkTSM|t?ZX%q>DVf_~A^wPIY-{j|eUG+Jb
zOd!uW&`W$&1=>yHw08XMqU#hT3p=vFETDPfC(b5uof@6b09}2_!1q5vM^9<!F+m@{
zG-6#o6O_0|1;Itot4BcnFW8xjzq>6r6c;~3&fxMLmXmV$-wmf<F<IQ3a)+QU*%^8Z
zqh9=&o9fncrC%|$6LFPUGkwee<NSS^Od3<2*v7~6P2p@>oX>4s+_zjje>&S~y4^N!
zw{NV*JF}5Bu<#_O!fb6c^mrG4lVpAyAK`tFBeL;)RQCfr+_rKP0`gS8cd9VwgR}&y
zW9yf%oTC5gMZV%vsQT5!MG~%r6I5Bq$zu6h)}!O5?QC_X=f@pa%#55`LWqb%YMaoR
z>k^|5hCdtTHAHKcJ$C!W^Z4ZRFI_3uS)8cnQ9bVieeUp*el?9Eu(No`pWNoYyJ{z?
z$FI<P>$X>cpWlJ?+Zs&Pg+=u>r#|YGW-XXW*PmA7qF@Or0TR7gsfpkTJ@67v!Z1rb
z^nkG5I<Dy~U>?V&bW|o6gUr5mclV&|8wG+;#~N&3%F0%RH;SOckd8-to`Vr#D`41R
z5{7{WIZNe5s$`DUT02BEHH$;KF3+$z1_^%iWFm|IyeKv8FR#zFsV~vwu1V7w!1EuU
zJY{&JbeoU)<}#+6j{sW}pY%qcS0lgp&%VI(vcRjuku`x%hea#*>*J>4Wy57U^q^HZ
ziKCeZ*NCXP+S#nT^@MxilcwAro^@8VLcSFFfEts?>Q`UgXOHD>Dvpl&>$EO9bR&y0
zDe<=3LfmFaO?-n|D<8)-pl?{J$2TZ+{b0O5iXu5-_d^02Y5&^o7##cK=3^D}>AC`u
znkiaY%JPNS;hJk^FUr9S)EDMle9?}ETj=_w!te$E!qiNmM9eiEnob|Si1!JTeFaT^
z`&R9Am_@mlEdDT$g=<)gU11N)^|QXG#i-XFB2a12>Y41o<K_xjsDUl?>*-V&y{ozm
z5MZBhtS&r-^P3P-mueqA?;NlrWRNqI;mj{3zQc-~D4G7kxtg1T?x!%x;E#NTA+r6q
z#!sp!2eNzW^V1HEG=*dDDIIy>v|gedCLDFMo~AMW5lg8`tS7B9m-5-RV-<|0Po4V{
z<x7|LjF2CDNqV<QguK?^jw)AWbMn&e_-APP(v%`<L@KzV;227;hU!hD0F+7mbZNDn
zx{%OiTB8>it{1$QsOoF1-(wqCp7$BX>Px4?X|R6=aoVQ+c-zQ)`X>9m!WxDzroT6-
ztg(Ca(saJfhm>Xomc-m5gTZn(%j@p1{UJ|wkZm+tQupxZ11TFcMU!SjsyF_RbF+bo
zZv$I5pPRXkUw>rP`F<f2@yJnztbrO*SKzuZCV5xkOwRI-Hne~=MzK?#NP1#Pi^2tA
z_hcG<VQ)GNsC2J(<rZ%XMey?wH4lx6zpGa!Ibi6eTF#NVZ3`1l-4pQr6=>^9qr#t}
zHVOnjSoZW45p~LI2qVwp$wInD+5&V$1!f>ccEEFG+2B-hSAwpi=P~7E0&=Tw3erBY
zSCDx!+cb+X5^%pzS6P3IoFG2sjiZC|KGLRSdTD*v%$Vwk_I>dl`&x1}%pj1JjAzs(
zA=udU__TW~o#e9U$qXC~Ifa&6CApEGzkgd7Ns+NBDjpD4!$PL@Z3^O+)@j~4H%&eg
zW2f4=8L~dbm&WyCmZw-hjjQ@Y8{Hqh5?pKW-YAWe(mc`6W;EpvSmvemWNFW~DHABK
zL|^#|q9N~WLZpV$>i$_c1$OPQFAX^T&Tz#vV=6Mq9Ph|WLGzLiM~-8qw4cF-ze{=+
zrbk1*<Be4C3W5;RBj4ZVdOSxG&Ccr1{*~$t3@qd<e+=H8TlDuEHpMAV2EOYvbv|y^
zS92QH!+CCU#n4bl%j)Gf6ofpwgapsA|K!%-nN1E*V0gzr_VlLs?3Ezq+Q$4)2qrPg
z{fFqEm#_0mBF0od%ikEf#&Xg>3d14nlH!OIv{qXks|9yCR1(xKPLD<np1WTsS-ue<
zd-1GdAL{WAbjmdc^A5&i?pEOerJd`%`je$dk15JrX`J{&%bL+{Yw_;Dw?l{f`M0mQ
z>lhF7rrR01di+1d4risLpIBT_$7ek6q0-W!jAcu|2*oTSwL^$7YIvlJ{3ua$(rsL_
zR47vh3+qlY{oJv!m4xOR(4s6pReC+R?i9=g0u2$5+JZ$cr>;KiCW)K}KvTQ1!fsv}
zK(xqM-{>(06}}#E&ialD(qr1d(OoqugOiSY8l&)7$eE954Fft-o7S`U-t=#bXQ6P?
zDRZ^Q<CBy}OJreCh=o|d-fTQ@Umimf!V0yBkteaDFxKv@7~_)Eh`Mwu3%zj{u28j^
z&BmDky&>#bM_Jc~cX5iqUA;PH_UuVX%@1m0dX^v?j18i$R{CPX_Aq1z`m1-;g?#m-
zP<`K|rXSuMD5^omJT7kV!;uU1VC2Z!gghvl{&dmpH;JQe`+~KcX$~grvzsZc2dA?A
z+mv)xyk}-c`F3(XZ`JkZaJ#=4;J+~yeWDPQZaVz(_0s#}HPCM{GrKa99(4V07%VUk
z5h7gYf!(MT3WhK52EV;BTX~PFp9XP@(vaepAq_L=Vy#6nA5<Qymp;ioEdgPyX-?3;
zj&aA5VY-F1hS<ae>_4Ve!(DK~{HBB4pM^swj{I8<z3bmz2Bfz@a6Q`S0QdIMx<|#w
z#}?lhN-Ham*)$}$Sg0M+FxM#kC@mh5%GF2ciwJbF{LHi(#%~r~Rf_odJs2)*jcRL!
z=JoNjQ!}0@cnAurV~|{`r-vg^-K;De*vU*Mx0<q52isG+>V@h>6jJXJ0ETBXGCOnh
z`5J)}VhvH`O2qi5m{w^y*n`htfPq_WlstG#4^<;jRMy(nuPj~s6Bka-c4GQ5sp6=`
zWqA6u(Ryv{2vI$S073DrDw3jfw1+O;?H@Kz#YsVeRCCYY$XiF)#jjNl`_8n|Q|-`J
zH9Vz$%b{^@TxY>i9QEtgCk>egMR|@JPxbx+`|9>Ixt>$6zNf+X+x`3A&20&IBR43o
z=EY<bYBS1P%ujEB>~+OFn5G(2kwWMeHC+OFx{C!SCC;3h`|}uS)%=UBt0lDvx<nae
zd@G;td{@sbPuqFXtAU)L)$|u$s>gRcR311)v9l>2pqYLC$~IC#hq2~_gf!@-To<cG
z8?*Y72Xy`shHR_G!P}K9K{|sqhtIW)BKf3tQeKc`dayLsxNvw59j`jDsV{}AN(m<y
zu9jv2ePyjxtHQl5YWFDr^NB^%4TzN*;n(JR)$WdRin$hRl+a(BdViLIoU-VvinYU?
zk95cTYy#?6bGgpuRr>GY+Rh2$zx?00O1IY4;+(hMOS3nPYO9{+JZ0;Lz~Y@y!UNXH
zY3@<6hF_a0o<ZIv<d(Ap0Xmy{Uyi@kUE!pd;u%DDjAfKz$Iz8~`ZwQ{B(b4w>urjh
z?L1Ng>|Q_Sm_O)csn!u3&3b=8W&9<Atq`*)?ozxxd>1N7cjH4dAnt`xFZ#vT849!i
zHhu&)<5s(-5v488&SR!kOV3?Z7EfS+35+nc{lXW_O!{g!yL(m)Hav@o8r8DE!^n=y
z3hu%8?08sg={l})ZaGWUbV7<sm6a&F)M#q<DI?|FW;;gKgMiC|D&frt?~>RN5}rqV
z<-IR=VH9(JlOGVllpS2FN3Lt*weR>129zbbddiWnit*Ut<X3h1oW0BkDGblMf$b@p
zzQIGUAzvge(JpyjNl61A7TzkRuoH%M9Di>^_k|QmZYnB3)k8E#s`2B}a!b_cb7L-p
zG}T5_c|dSQxcYmP87<#yTC{|y)s1dPuaBqtYvgGqtM8xJqzRIY4f$U3CvnOL^}B@z
zb{5@iF$~7*M_cJU>m{)7`{2hLm6l)6Oc3fL@!0H5_;-r9Q*k2c#)85nit?REvo!&$
zyjO7?PRYLpUW5Z3M3$>64o{YcgL%Fs?ucOS$Is9eQa%F7r77I|Qk+eXE6|Vniwe{@
zhbwFjk9K2A;+sls5$9F1(+d%5(s{2p#{~To<~n~d9Gv+~=us3S5ZHrl{hH8|v^9@U
zgDq8*JC)XrOpGq{M2qC>KM+G7+4iY{k#o?`80vDrF5oa5i6K?}7l|z+4Vn`(-Cs^4
zuDp4GiLSxf6-gnG80XG7K8|o)F~x*-y$F;F@->b(F#Noqq*pu3tAcNz`M8!9=ur{L
z&BU)a7H60UJ65O>md63NrCWog(c<sB+fT4#IEnD+R6nmb&8rdYabJE(BU41f@hBN-
zYb_fY<jX8Ebw!|U+3mF$n!sC>WCiwEOB^_A-M{IP+~pT*N8m;TAf5?&DDfyvP1e8v
zrO5)DxnRV@RHKQKv67jaimJdlx}tSu#gN9Bz4UHKYjSfch@PlQhO>^M>u$$Y^*oj)
z>?+iu|6tLMJ(*qpbgnSb5q<j8Juw{@I-;P;sB@5Kw(>)F-@pQgW6WEGaga1}sl0pk
z#DZEeFgH`yGA)XOL4*k93#{!6^Tv3pNN88G6@#r-^blb{Gm0WW&H!Ze9Boj$&?7Zb
zn1{^FGc}g2%O(t%Ib9LqDwP^BFxbuNt=(C=s4fhmvKjZ=q>}_4+U*M8ZWhT}@RX!0
z0+AxV7QX|h;XY_nO|{p(%g;;{9Lh4hjyw-|_FF5zPCogNk2*zQgnt`N_lllTgR36`
z>=SQ7a6VZ1;^`?dF!ri(e)3OUx2&*B$LTeD97I!5pTjY>p1t5#ZF5C+bNmJ^Ow1?+
zGe_U@2T@~R;ShpsW3Zgh7HIg5$^k1Wc&>lL%L1Ed#l*R6kW#fev3!L47S%Ap-uaXU
z!>XR_)x7>&5$)VvS}96R-_L>E+w~^KYNiaoRcrlH&#`-6dcXH4wIFex(k$3OZM>W=
z$D;nkge2<~te8mdEO62~vadXk<r7+)c7LfK3FIXEbFAIlg>ab$apPpncZJcX>{jz_
z$$T1JDK#+q?qa27AMRbL_0})Qg|-}UNSi6{C6w#{)+!*k%{IhODpzoU0&34rDt^1q
z`AaRJzlaqkiP5bTstl@p{TqiB#<=M~s)jth6L#gpJSyv+r?&CKuigRqnqgwu-Rnfo
z-3RNs7W>B9lMeGyHWvZYtER#~Tjvyd`U3}KnD2tDw$MH-B!%I1lA1RV69G*A@wr4)
ze%8srgYf^H1sIJ$sTMfVVR3z2X22yBZ*-LlvC3kfO{*#x89ZlDZTdO=La9((=OB75
zR3@DO%I+mgTfclI{3@V)g~pTsTN)gf{fDK{I(+|14JfluY<pZJAv0MOmOcFQaOKOF
z4GCyhbePBCR<4=sqU!Pu$5$Tet?<XxM)EwqiCaFU`B>036t7*!Ur#>+dyAsie+XBC
z;e}W9T<EJP9o|}<W9d6p<y2LxiIE3XIeVFf1gUAip_pL+$!L`?3B}m|L(@0-$N4wW
zZfv_rW1Ed_n~l-f_Qq;#+icu4wr$(CbNBb&d;fy{?96x0oY8sA4}&|Yk<tb1pjpwO
z)<a$$2H}{*6Ynp+hc&vx$aE0(CzAmY(Ev}sH}?Mbo*ja$GOONmI0w9lpLg@Oa6#<9
zqL16yzv@FYQQ8Z+!AXfE3&0bCMT|sB2v+A0d2aIU3EgP(rd@vt5_DALm4%zrNTDos
z!TIZ%k;&Z>ulRfE(hD+i!4uO$Cmut&?l^B4Hs*=7Snk;$weVsUIly1c=_m*#`2G|)
z7_Bthy2U!LPGx_$B5A<jyvi#dsTeTXx<RWg0)?EsB*oc*tw$?+h$8v&q179WXeI!w
zc>C=dvKe7s)5wuKME$}5L<y>zfSe9ICNJQf%jC7HCp>l#4iX_zK}S0ypnDf(oS#SG
z#Fm1Ny@|NI-${t%CKyJeEM>{R44-E{i3e^3PxU-9nsr7CDn*7LpAs+k54ODx6M`)7
z>8Bu&7CO1e3*aRcu1>b5n6NB;K%@)96N{-y2}3q8On4Ak5JLSYF^-wyNfgiA8jyPj
zm&@sb$yp8<NC0LU4rs-jk92Pg0a32x3R{t}n|Px5DIh5}auPHcqOz}7nP#OTMC8`k
zpCu);5DXzVTJeBAG(f6F@yRE2veCI;M-3bxzKi3)Mm_8nnf!GN^J`Gj8n2j6M;mgx
z&{j$#m~4S!AngvFC03vf^pGXSSZH0Tp?bA5=2ARp`O^J<eP`Rn({YJmhU>o8j2DPi
z(?%LH*gm60`KHJ7%*f4e_@E2*9#^&f48|)>Fw&{1)Y^|N2FA=4HWMkYu2?rTE-43~
zI+31q?Nqe4uNyYhg{BCwBAyY*CodDc0RwCA(N8Ujf^^B&2V)yg$ovC==E)vU5&9bc
z-h5s75lUSdCl7|m6r?Pl?JF8%zUr|e1qmj!(swz})PSYXk#<ltPLj-J*`%DTy|9D|
zc^%*OI|WtAVxoR)`n9V{{e+9h7k!`=!S(1j#;6Kdx4H{v+jM7wHtP1_|B5WIMy&q}
z#|{xz?APNyq^Qx*UAD{Dh~S2u79*$YJ&!0x%R?u3icc*d+dg5cs0~m6Lw#B#wi8tL
zJ(Be~z=&x}A}JA1{B8g<&u{{oa`UV*L#O>Dz(SGp{A=C*U_WF;Ke4bBV{l+MH%%)J
zDM_;&S9k9?TZ8{NlM>%MoK$9o&eV5GFx*dC&0z-ZQT=zp6uZ)~XddhP?1b;jMrC}M
z!mQE@Z$7aX15zDhIN>u<4B_C04wEJj&*X}x>TkRt^ICnMTj+er1MO4?3l%!G*2<LS
z4lIUmwIFsVD|`w8Y8|$#(IyeE)J6B60VS6(8L{wT6Le%RFx~l>zgEBxc!Wcfi$Y`I
z;#MSeNL3dX{R{kc9o+SiDb@LG();@CIF`WXZXpZ!mWXYCZJ=z%69dIhS$P^DS?gCL
zI`1-m$#X7o*P_Xb7xV-FDQRk#9?{rC@YP=ieD6Y9tMf$O{m;~K69H^A2S}6bVE4+=
z(4o`b{Iw=<ax3|1tgzc61Z&|(^Z^BO)R81Qhnq|EcmG*?OD%G%pA3+0+$}Z9rnT9H
zQ&|iLxZAm{zy?N^c!nM?{^`6t#bYRJ{2)UH%OXfn!zqSf-ox1|;9W#-VgMz+0F1o=
z==eJNd%l{?!w1}SSqFwA1MEHvOM=b;a)2n=@!LlI$df1bDJw*7>Xt|F`pDyMRe;WE
z&(JwnO&A^I^PhfkT@m2F9Sa(AagU_F9(@yVE@A7xHTR+OKk4#3g&;BpQkF$f0Xj?e
zf%8N=qCx=U17o@jdCDIzO0^qs;(ML9ct!T$yR<do`&VH`5YLW58cq0@&I_&~^_}e<
zcr%<KlIGE9-|duRa1~K!RSd5M1~SjKBRn>f2JEEeKi<;mswo2j0!pbfpLo&3O3|zW
z@cX@qNi29ZPIpxhJ!7yE*K_^7>i3L3sf$@i!AGFI0Z=CV{uekXc!SItZwWrV;+@CU
z&Vu?I$m}ncj0-vaeQ5D8?_EelqHZ83K<Lc@{Mfz?XGoL)oSJ^h-FiMrT>?>YE)Nd4
z(bPf>)6UX<#=uhDv8Cxev9K#uN*O5%&@)LWcJ8oQbmgZ#9KataD+6H$WDIw2^ryo*
zM@Ze*7rGZA0^60iH2O!XZ-dm@FKEy`S&P4wzr6`jmB7&K1?FFII^?i1;qXSBJ^vG;
zSH?;sk0RfF7R(eBA|NmBQS#vMn~_G-lo|MO5!O8f-x~|zKU7+h^0Pc75AdWf9Tx2=
zW?=x^U*H7EOz+C7eCQmNmyv2q0dLuBs&*BNr0xJv>QjK3ij0*(Q9%k|dklp9K7c@Q
z19eo$^o6EKeZ1X75IoN|$3Nxx8kxDX7aIqxcB(Z9z-@NXl7|FquVpM`U{7Os1A9K@
z(0TN`B)^9Q=R91ky=B)fvn^;5ZpUJY@5H|Q5=qw>LlYW=-9|lpFKWUEg78_pP0{Xf
zC!7K<jt=ZmF$lo5U(D)lnrr(tEnfeXrmeb*+U+`$H3#H^tW(&midWF0Noi2fsENn`
zc-XWV{at^fzIqi#UC;JAeF=G;X(}})06#WCe8%MRweFxEVvtAu$Ty>*1W-RzuNxZ+
zy+i%}5Ddu~(($hq9u~S?pd}Pq&}-i$2lV~DP_ZXdWY0A_aVuCakoIWBRKBeYWHAl$
z3X|O7o3%svkg^f*0nnW1^n*9<pxn4PQet#Fu8upVmtKmP4S`#x9%^5Id-qVkv42U_
zF@pv72z3r6BJho@SJNlhFTEx+wHMk}cYe@nB0=V^Ax-GwtU9*GXxKxs>N;xI5m%Hk
z<4M>E7to_Qj5oIRLw2#LlN9>;EK8NgA`}QW1ave$OlT<o!0c-VmCAaqcnnGIWU3J;
z?RdEPe<U&Fdo_J#{F14hSM4~kzzwn5kZ2pR>oIz*`_a08pmmqng_X&Sdp8`}JEY-7
zf{b3?rg(T;e}FVf$nJ>^MldwD&oCVY!*?^8ENFs<)(dAS<)DiWlA5?~X^&K1Uc=X@
z@M+EfEE~C2*GKV>{9BC7@yXnx5|}YRh4Ui-Z{_<@2(Q@Ou-E{#dtyOhr4k7zBt<By
z{B?NY565zk%t{{i^OXCj&Rix*osi7No`S!>W0f3GD6Y3_Yijn)M!&%`!mwKFn1-pW
z@)MGw@xIz8N*aEX$^})^4z7dK!klmjT%x&9TbVN(F0T-~Gu1;{QRoYwe8yq@!axdA
zQ~LeqV=E&e06c9JYYVww${<-@Q~`3!`V&44CFA%A+}wE9Q=7aCtLhlF?yHQo*L449
zd6gcT={N41TY*o}c3MZW+Wl_MT(Z+#<0*J?S1IS<6Yk@InBa68dD=k0=FLA=2rxId
zYEBl~Q=^&&{vb-=CB&lfjrR11ucn{}JW>FBb-Ig{)i=<_Ni<}gka<}oXnn?4wF*Ah
z5)_bYHOKv7(5K#HhlKi$&|r;6PFsKL(x86r$fR0RG$B&~Oi&4l9ASl$`8nnl&n0In
zI!iTXA-j_|$Ydr8+k{=MxTft-x6UM&;19SU9Zag&HllnsQTQ^`@i!ve=s=haBg$m%
zXoq!55j=ZlvcDIyjlV!mUIcz=PdJxgC(v1wTjetBcT2G-$~Q4vUASIwhKrnhs3UzA
zH$Ln9i)0Edt|II)Y|hMMp1hH*su8%|k*j_Y%LYbJIYjrtSIh&Rd0VWPv1g|#bS%=-
z>XK072?_3zMb~+O(n4+~Al@A%NO}w-NTmH)mE~i0z1PK_u!XL1uety4m3Df1dz6j-
zb)Yd?2I8meI7`U-mfr8bM+e=*y)_7`%gXY~8@!Q%i_%`#PM-~w)rugJ$KZ;$F{PX|
zRR0o4QqAm_GcjFA?fab?fnjqw-L&K*@Mm|oVw=b<(~z{12sijLC3xBijoj&UjVAfn
zS`ny{;nH~;+4B5m(DUwknJG&Tz8ssZ?oTqzjLPd%^^RBRp}_^lm#5Aeff@!-Yr`-h
z9MhSP@+GN;1I}fAa*Gqz>OAGB?a&JHe7b)e^GvcKgCTNclaLpkzH$5r?ZVm!aibZ4
zyv`)?6RsOV$Y)0(A>IPP$pdQ#ko!cuU8Oi#Z575~K{RFPHDpsl!Ql?G6pKs3ZU+4V
zIsHUc0ef4W5~6j})CBx2eYIO;^~e2Y_t<)49zt7~`o&&4pxtm2fzQA24?WF_WBQ(5
zs-YLy!6OsKJu@`qb+{qAR(-ANp_#g{NoOZHmzAns=i2FQa!>5nu9P1Fu5!0j*QiP|
z&|J1g$mOR3N865jbGfkVYt1vFYk%5XF#)djvoMi4QsJy3X+PL|Jk?1LZW%uvm0BR6
zQ9-rTaiGd3<00@D)NR?Q6>+R#r}@t)mH|l--6B0*ZUh4?J20f;{i8OBYh*{rI(f+J
zBDXf0Dya$Ra^go*P{-)_J^Q}F#g-*jIMl_14<B+mi?&`>?mIl?uqBoMpq1nf=nrt#
zrMd4Hd9>h%0Su2SEDxNG7UtCZ+DkxW`+HVOWfkP;A+gSbb|;iqw4cjG5MdbLGL|8o
zD}?ugyz{kmW}{4kh7xFibb~N-9kZLRFT`_bFEV8^)q~|FkNo&$XrKc0ug{eRgbuwz
zZAwsHTZ;T@_PLk(D=yiy&KNE0R1b(s9A|C>rU$DtFwD@KG#G&^y+tn}xZgCIf_Btc
zRvjB-&DBGqPH5930~YcOS{A?Wuw*xyf2K`P2l8U?J3~1cu`;_}`~ET%9ZRkG^=}kl
z#9kRi`ZAG+8rWu0UTbqdYaE_FII$(NNA7$W@pev9)>x10JT=loj)R=#pwT*pZ%*=>
z*YWy*?e{W5&07$l2n9wDA}SkL?WR*r=8(ppC3iO{g{k9p1_wI@KZjGPL*>&F`AfR=
z$eJbVuzLRo=0OQDnpX_JfRZRE39ba7J`?hEx_4KyWvAQ{3fExIx*4jF@9GPEcxdD-
zv{hcy5Sai7&+A960$;?MmN(ErYHsQ*VIS*G8oSFu0fp*QPjJA>fAoLlH<L5|xbNxq
zuLg<6>U0c;fY86}2WZgIozl9tY*Q~L7e<?E46dahk^B@J!ILF=PhY^(iOe5cO*?k4
z<G<+he(2eDhd*C4fY@iUu{9ZKk(15AEoV0aXKbJN$&)~bin<oY+0`RsP8je*bXncz
zhaBM6jQ~5Y#gMCVRgx8qpn8F+gO49k>}$)Zq{p!(Dwz(2!BZ_6<?U~lMX50@J4uv|
zkw&2@&D3W*B7%eQX#9eG#wcKiKpG&Qj|#3$ugqaO{tw&ftHl1@y~x=-YVqC}x$1m+
zvrri2ZZd!oL$^*xe^{zaK%RC2L|O9vBqXbeMK+)IAag(j(K{iQC@OB){9MyB+=ro$
zs}_CGEA}|kw!6f(S%m99oWUj0!_X?_GE|c?yJQBLU1#iC647*}Z%wMcdqx2l^yZa<
zzC{Y$v#`g5txN=AudcNHa{%{<ITph+)0v_TH6=j?3uhFt2uX$mek5F`?JuY;LIU|>
z@2lZmH2&Jv?lz)rl4+>R>!3t}I1f8nY;CWW2EmgN)Zpl`ev*15oMHJ3!xlRDtqsXp
z9Od5F^xg93#msxL4VT`|is~}Y`W!j1<g>h^Bbp$YBOU*DM&K#pT+65_|1AHl)@<Vb
zlGZM<Dg?)%uJe_@KCGqov5f8}?z$6<`N7{HxNYi-Ew>z;g{b^DBB%OZ$Cj(B;>(Mn
z8`_iJ%sLJ@N}u|`G+*8A5fwrXZ%WlN>93l8s!Iks>Vok7kofnFd{hyc;;Tg)55<qh
z)E-HCgHj|2Xh_A0(SR{jDs_S}dr?p2fVDc<EA^W>-ZvC}@fb2Fu=W~133E0Py+f*d
zBmAz_QSjUiBf+=aBwZ)1FJfA<wL$RUu0B2w{GZ?B5M*(X(I5nCmLlfCVgSt%0B9NE
z!4p?X0d(GRMDY^1JZ|*_?kvH$MDID&03r0>h&~{0uE;Qucf#;YYkmS0UD>p{t97aK
z<I4gtr^RFMM>>uv43GiK#mF>tQg_ta$#NgYr!oLs>D@;%*fb~t=^sOP&8zAhmU<+V
z$YA4<3~8jtufG;)8$&1@1xaLNn83QFJK8sCUmul4*cF7Z<E|ZS-8i%hmKd@31K@gw
zPSu@0^Dax2tao!)$JT~ZqzL^q__S<QUJmTy3rRhx$2(ZP9@T4TMxFkE&AMGK{GcbH
zzD;!=h6!x9H<KSTqy&-`e+H|&je3G5ANwb>AvFhetf*`Zqo7n?eL#nf86$qGQCK0X
zc&p@u$7u;lB2zMVNsCVkJK2B{r6Z8OAh9Ut{I1b!3P9$bU!dF`_g_fG41$RGOoc$=
z4dabDq#jC{X|cY{opOegE1AA?V~gx`co#_e6$sF06BJQ%lU5w^Vi!53l=9Nkm!>f?
z+r6BrdxpG6d2>8bkP9(%)iU3IuAw{A6t>oqfZI0TU~7&CFs9DU@Lu!aD8kW<e+!kv
z5F)D}scfZBl2(vwVM#f`kOvoXOtKEw*AL`BJ!%`JZ^3%eJntRcfdr=0E-{GD!7Fm(
zr2_i7Oyy#*(ntDQV2(Vsb2z~2g0>`S$E6M~6WH{pI?zgfAa1&lzJbt5$41?whdgm~
zZW5!Aq1=ZgfsB!<1G0Nd<;We|_br?)e9-9g-tySf#`0ai_QAyHA!<Bcj@DO9Q?+a)
zu9)hLF&SDFu32o<s;DhA&MQE)$gdXWOq_W!^_(vkrlZQbHmWh(TjSdL=^2=`Z#2(E
z^KUdF;PG%+u@;-4<|DMtGoKOnnkiX_W>KW0y|gX<YN(=pV)tK*ZywW(v-$vgNv0FC
zdPXJQ3{{u|ThP!ay63KPeI28i%a)WMgj~LmnB;JK29Pl%(r^dNKnXDe;nzrjRrpcK
zQ$wP<Gwy2c&R@tz#&GoI<ni~$#G<EaV&KcExeI!!&c39R?EwKYimdMu-D><Rn1CPV
zYzs8O-D66-N8-3yhuX?~_&-KfV)kycM1Wse)kM>p6riIx7Dq>NAet?Cmjw?rICHg3
zbz5K2kAK)1!h2h8x&liC^xr?ivfO8n{37O_Ue<~+f1wqzu6Wqnfdq)!VfQZNJHyBR
zh?E69!H@i#lwqCfYk|A+NN;8ba|rBsN($A6t6K6QfeKev;BLT`CIEjXraZ+f88y!c
zFMaL+L!XRb{4@dG_5VMKu&6)9=FKoS$hiw+J2ZTv+rIcQDzhsaV)$I|E+puNyxIg_
zglBl23)VOe0~5KX9`rOD(<dcLdvV2GW5RK{XSSxrYvlUtdbTQ8pT8^<uO?HJar2A%
zmguTVMDiFH0>?M{;6oR>r837tMCy=C<;Q>%U+kro$a9wE|Ga|2R@a}k=mrwwv>E23
zeV*c+rWS8qG+Z64+1h*9V@Y#9RbC={o$|z9B4!iuNnhWk{P^Ya2W5?FJ`Bz(-o)6K
zigQ&FgH8PfVAvoC<)innUn%QUb^A33%HRf5cv3!H12IoQp`<KWh|l$`a+Zvs1;!z9
zDCa2=ra9bR{nV@fSoA^rbM$4&b@seHoFN6hsc08=O63QfVQ4C}xe2+y&TV_j&XR85
zgY@<|fNB2&Edfn6l?gGO68LTY8rROV!OTrgbgd(K9y%AGftt;%8}3PX{ekLhTiW-7
zF_|LVZ#&<RpLfNO3#?9+fNx~RaaX)+fCSL*h@}PhYgyfL^(e<blM<jo+Fp+Nk9JrG
zdfx+XYoN7`r>#SQ948DQ7HfkRu8X01hZ9%2eRqoLFci7scY`jpUvYk}zp@e(tz!-i
zMvuYI>kDpJ+R0*78Ao~15OFEOlN&2KW%B2ECG5??>x;;29*U;q!Ump$@PTm$pliU<
zue{HSXCU7s-IUccYw~5|))45D93ac&eBwzMuLi@wejPUVEfX3gFv(OE)v9Bw+j8aA
z_EVZq#rwOJPl;RAR@Q*~7_Eq#n}<f&ogbUTLcWi-tw(uDtA@Ef^8`k!Wo;$BxPzPu
zw+G}^ejsFfl(*#VS)g$8{QTlo#;b;-$tvf(kQvLy@J;xR0wKYVz^8M<^N<Lg*s%jT
zw#W(r(vUV{py#RB#c+V;Te$xS|4Q{eyJsW;<A{?qnR48z<D2b|qa@Kd7iOz23$>SW
znj@rZQQAyM<_Tx~72FeH*V&Vo1VBup1__CP1)$14zt}AZ#E#VpewEgl-yo*<omu(g
zEcx}?EK>nU=oN^>@2O$pd;%}O&|G>w`NaPDyc`9{80<iHiz;a0=F4kg9=TzG$#u?P
zj}myOZ@MssNu>SyJYTR~fb8aG)UTFpnx0E8kmm98`W1!|d@llxtM2xL5KKr~|FU7*
zwmlf^!dts;JX`Eu`v$#xVkL-lMv!spanGp$A5&lg+8kw6pHx%T?)zp-d{tQiYDhoN
zzFGk<^w0-;7Q%#V*j8qCUzhm{Br{u<Av{M-$s_;2E%&=0^>mSqLMf`5Nt?!X?7y=|
zW|XIAT$eN_Xn)tY$eHcil$i~Pu!wa9^B>#Miw3zeSnGJ%e#Vzw<hO+z<LDJ}8Ei7Z
z7&`)Ql779SNmTw>mbL1uZ0Z+vA?3M)cB?ho%y_TJ;38d?!87n4Zi`G?dW_s(0t<>s
z@NT@Utp9W~{ySBQ8EWp_ej&OtxCuns73@(7_vv9?(d4Y~`=bMMh$;+t4SlHkX1B5T
zA^XE*f#Oip-MXn%Zp|I%d3Nfx&w>=o;iFjDyI>J}=ahAWvxE!;lR2nli8wERIVZ;=
zNAlxOIhN{5vFJ_TP@^Zpu&T{x?+Wm!fFdM4ZS~kLk%*@^tcwzwV<PS!^59zZ-`K_%
zq>!W#Sn~COf3}$T%33)Om-^tg4URXIfv`D9_Jfz^Qs7#q2fk5G!0BEKj=B2b7o)?E
z15bAgaKnqQEciw$N-!^;9v;Zr33!?Y7<Q?kc`pKRpbBT4pA&#jmWeE5;8qMV9S4i<
zf6|9j_e*$s#^}&?ms=3-8=T&v=az6gzWC`F`XMuL$_{lN&Oz}p+-U&mz@R*Sp7Rlk
zjjo4&h|HlF4vKGMC-&TDLH78Dnn{ujBc%jU&?9lxb1U3P`mj#MFhc<o%0HS@Bqu{@
zI|al|iZHdajKLztYmMb)IZ?X>!Yoj-=3F)lbvs-3mcxEGePKK*ao#RaCtJ%@HeoO3
z>zTEGi+xFNq3kFa%@7_QZFM-M`g!N{$3vgI3maE0d5~N3BLM`?Hfa1fs<=5LGmkDF
z+q&UH{qzUYgg~5OH97UF^&69eWl2<Er_uM`=*1fKwll3|#N;7lO!w!P!!m_6PVh-I
zFrxlYn}oqUOo<CL<q5O2tG$bQGOHDGfL%*<9@Iv-23jLZiN^%Tc>tWO4FBWF>*6q0
z8Y~_qfQ6YWk5dz`nmOSi-!GJi>+^D4`Aq<<ogo8~z;T&k+yA=T6YjN7d_Bn+<2@Kd
z=zgKFc#VZ}?oQ+pA^|Xoq(R4rf)T58cpSUYf>U*8g<=KHg}nx)O%$0&%+I!C2&I_F
zNZ2}ox1S0dB?|CdzhrRqVBwRk#p(!^Y&MT-0%@cTHJDuv_pGpBeddgYPOr=tyE{m-
zdTdE}6_#n8tc4?^8$5Uwd3Xrk^SOcD@L77-9o$;uxeCkJx)jmOaOv(_Kla|Bql}ga
zZ1TwseMd>kZ!7(i0)sp{D~VHYyMgUV-szEYOUBF`XTrAq$;76HXeck6I$`|yN0aw^
z%d+-btlIv!H|{$8_2rUNiOzNd8mMI5(kHiKJ<~SpoYp$gcauA{->j4rvm>SJ06hCZ
z7Oz+-Kl%3Vom0mL(F(DNG(VRo6=m(ToWh3JK1vS(NCtLu6cSdH30#hLT3q!mEzThB
zL}Gw%A5Sp;4jfAP0bldrIw|jSu}6RhZn}#r7c0+;<reM(*8wUeJnon7z$#h9C~FX5
zu*M4Lfc=@nft9bd_#*v2c-p$gh-oZCgjqIZus*JWz-X8mtO^bNR%2A;wT9n=eLvM9
zgvd-*!~F&FnssJ@S+haV>}3cNOaR3`le=mwdI4{U1km0KgFr+GQ17`jdMk_s_*LS<
z>J4FjzFGJ-czv5%7Yt8XiXQ=WHh(y9Zl3=210Yj!d~TO_;dYVn2e@^CTK7Law=D<g
zb5Uk7Ao6$K{UGT&l&c^eoF7F#-`dL|jOhjU+}>X}-UcnuKoMB(tZ~jjeiT3j5^EZ{
zwmbH8XqvZcdSCe|oLgEcR8Kh)#R%h(_enuz&(*g!?CMF>OrASSyFo7M%G==DT7un1
z|9i<MNs?A?V)D^fxI);}k&sjF7n6znSVU$QhWs1x_gIGnyPXO-rk_|`o#z7CGZ#_b
zAmOr<)oW>Ak`R<;d+&GM6TI%8;Mokl1KM;F@Na$2WG_-zd6ih%B}6JnV$4NcYxpNW
z(^CHP9z-b&fdGr+w?2NmrpDzw{@1AwU=|x3_%vsgXls140lvG_N`#bWP6G>c5cNq8
zJ0{;l-Nz62_)qPQV&#9{^MxK2)E{()Q`DJ|d50p#qaE=OLcMb(RR0dYHP5IW;$aia
zeLkXhQA&iu0kntjuKd8fCV2S99_A2xBfZFvCU4;Yy+nsrZuO;w1Kn2ep{|Kqn`Ogj
zwf}z?U{J#=RuV9UzEMB`vsWBq0-h!l@NKg$^BvpPm`At{XXMCOBabwZ14KYyD~-g~
z-(#M(l}a_96guL|Am9TlQR_V*yUA#hzY7IKWgs1?-E=Ht9Ug#UuybYhw2k>Kv^yvB
zU!CvU-Z`(i{!H~(2aBzvwJ;xSz@Jc@B4RHYyAC-i<(`BOVqu)noFM_=)k#x>c2H-l
z5R&TwH%tb)Esl-0T%)3}aC6HagdYN?(w!&=DEdqp$<NNDv124TQ~MKX?p3;Uv1)#z
zyij)M1f+7;b_1+pYF*%Eo-r%e(Q((iF>XgeJP}8GQ0m3;8@618rdlShWVzwWeZ^e+
z+)X=op=u34j_6V&HhecCA7NO@Ki~*7(r*?FGs}mO^5GF@gd<q9?!p1*R4xU6!~Xl#
zyH#`i!uCd^r!5b=P48ZnSj1FQA9563*q9baGxD3x2qWvt@7hEW<7#gLs@~gjJTeTm
z=+1uv!CBm{$hT_VnOQA|Z0n!|ExIPM%_Q0`p`gYFa*?5+^r0?pq0SAIMF|=i(MJ$$
z$-7ewe>N&b!fo5V5Pl{Q*0UsIs<a<wSdfex{LFs~kmugIwf!B7h|zt;%1i3yTKs`#
ztkQF5b0L()eWa8KP*Q$kD3IMB&M(F#<BQzc+?W~RkY6&5_5k(IU06;MiR2KT?9X&m
z?z$lX6vF1#U-`qeoux#r<kb%M5a<=7uJVAYWm{~enV#m8O61T;gQ2DQzEhCD!OBu-
zPvyoNuJ+iCm#&wc-Tt5pwb-1jJV`$uJjP5l6z`ZRxVb&1`_?y1!RizRYe_*{_WFp`
zgp(X?N=Bj46W|zNo?Rh+2{>?}0FO>MSeiE^rP{(1T-XutpMFs2Fxa3GwTqKGCA8d_
z+O)4F_?%LSk6AH&x@_Br5jxS@=g_BgYSYFh#h<BBIGB9b#ur~Q!9y-R64wa-S`+WY
zu2%@Zdi9CJ-9Mi|*cu_)8TlTDqr0WFF&y|a{B=8yYCcpPsoX0#a}I9Pc8kWVcf~m5
z`!~CTJ~ylu0X-X$%`j)u_)!nM3fPiV#~WVHN~tI3f=AXSmYKQzIg0r3G+6aW!vek}
zYkHq=5f)xrC;U&g#*j3rjdY75O_J;2NUCFk5urp8;Zt6}wYatXaZl2CaTU7USL_?U
z{aaOWBL^R~#ES4R+pH(8nyJ8U5vBp@1%nRai_S26BR;o7%10F8{vLiS*U_Ics{rY-
zS!Cb0TMUAFQ)=+HhUT0VKEcYL*vJW-4Nt*}01?tyY_dUpR&`nV88L~%E!`+~HIftU
zU9@h^K%Jy8UE+7o8n-zUaG;45m9;;<R`25tM5<cjDm-M(ux$q5Tlu3^^g2n$J9kXl
zST<5(PR_?584v>X!pf}c60z`rtC<J+D6QcrG8r&AoMLdWcJB8WW38C#$xUVY^jt@r
zAxQIfFQh9D4@n$EHaLRSV^KZOcqL~s7|i_TjYCt3<RCPy+u=rw!{o{Dl<2iKWmJe<
zZhZcmA<#5KjI`5zT6)<d)ibRqMrz-wK(s%%Q2PTmt2dKH_gNQ8cVymLgmslmiuSIy
zhqa{mNFpWb>nX~I=39jY%fD;k-4&g?-lf6feA;|EcU5b9!~mkMZt|v`hQUCaWf2F0
zAI@;WMp6~T8e0Bt5S0h+I1MG`XxgcjKgm@hEPm~?z9591(D{<sW9by^HKvwF!^N+7
z-4sZPAufzbqMmy2E@?dzqtBqZ`+H3>klz_OozOJd2dqLs4SctHk^DsWg~B^QQ4^(?
zYIql7UWR?M)!5}Ro99Lk_*bzVL$b;6Il=a~0xkp7B#Z(e9+G<;e2Y@zI>C`3;6&`r
z4PI4ovWiZGM+qb{+vktv8CC9J`#@w4QMWky=~PL&IY<S-RKrv~ZAg-QQZTjv)w&&q
zxMMLN51`;pIHn#bbYJz3vj{=TmIGWX3aoDZed(S2*A4MX+?ObMYW^9o3LykKuUpV%
z2o*lMC6fxdz6!Ezcm#rW%cH!ybOx$=`kRBd4Et)P2H8jeIkXz!#}^ZTbFFxk2FcNo
z;@nY17X!C#U%S!hu%`H7EB!c9v-Wk`v}!C?Fi}osDo#zGvJZ$O7O?R->rz%Y#_<Jp
zVsqm?!V<B{2nK|RZ-<AkkPfUrV%Sj%wQpet(q2$+j;NW$TG{LeF&|_(Ae3mbXbs;y
z53c50W<_YjYWi43X4;Ml!kYBSB<j=$5c7Tzk;CR;G5Of*W^nxd-~h45!9HK7uJec1
z9I*HKB-*1}j{>sEeZ21LaxuLD_%|E<5&rxT+#*Gs%s7TxYjR1Y-g8}b@FBKH9~yfs
z)x;Pc=h$et%66IJ4<~1%hEZ{u(ERg)CqKIy##i(kf$-LizW!W}_+gi(D~2YRDGptR
zD0fXV&L^Ll+<y3}jLIk|%VZ<)*0s;^gM&3pB6u}zZq6r-D{SZGB4u@bHYz&^SsRe}
z9Aq9euf4-*zf+bAZM1bw{N=U3vC=4QxZXHs?M=K9__L1Aim9#rYNa66U-S1qmc$1`
z`v*9~GK)RqPp|sgi38kJ43Y_|{!cGHaB6DFYBh!}l1IW;tv|8E7y!TKj;=(i30Ll~
z$Xt<dZ+k+q`g`&vNG0)&JoOV;xYl=c^@6_pu;)3qa1Th-kT)f6HrtJv?W|JmpliNT
zCXWIwxb0IL%cY-b{PU}R){j(=0R~W|us8paswfTeW$AFa8;4Snh)<rkQAiA?28=At
zIqu90F(R2MAr*xgoy^JYDh{H0&gqd&iOIb?pl8}-t!H%>I3Ys^sx+0K`0ndf(x(2w
zq}XdK!@>L8n!1j&eJ>~zqdeZ1yL&i04*ISB@=2PkX<lz-4<pqrf!p5Oak%CV8$CPc
z<Yp)foM*rEy(1-rgmdH;m1y`Q(<vy~ci|o7Vkv~rOP64P-1AVuD#GhaagYnK^lIW7
z%7u^N>>@$bY8VJ;CWty48K<PhqZ4+`W0BK9TEmPuP7>t-LO>&PhybJi@T_(7h;hqS
zAa0%a4zYo)+xwdJPw&{~&_}Og@ehP`Q&Iwg3zFr@m<h<;jJp0py(iQt_{j`&EDe+I
z$fBwD$P`iPL}cTF!gb2D?BM6^9HCQK?aXTu|8+*6B`8*f{Tlkqu5p*)ieo~8pjP#<
z>?Wcm0f85%Wj*oPeBs_W=0fJ+-Hmh9A3bDQ{L4IlZtqmh<_^IZ)Bq?L>A6B^-MOs7
zI7lrGt^`Ihjh<pb+mPjVL=?op#0&U{Fj9;Jmo}~8#%BwVu=K?WYxW7k%CC}`<1M{O
zlzq|{-nVhJ%&;e05TA0H_*b_IXW5zhyaY#^r|F4_I)DC0>mJ@X0#69$nb!&*KqYN$
zD8v2(sbdQ6i~naC`}uEHl)00Ins{yRnCqgF@fYat2bx{lRMQ%aamP2F;@Ewxc95Dp
zVJ2D>s}^ftI|Qh8HOy^E;K!9$UNg*G&o0I|OoMG5Wm!2&Ka?(+$#UW$uoUjs!;%%-
zR;jsm#$hqrnNHBTV2-28=#3q!1$8v9n^;JO8jy5}D-UeMR@-?0immcy)c6klnxqI7
zo~7J^&XYw5q>1<Vm}~xk*Q_7D>|r~7@WC`5A*a}aEgtNc5p1VLnf(W6f9H`-R}nZ`
zGq&u&Dj~qtVT0ZcC<wdS=c^O}C#bnwz<$WEHvO!+BHnMkXep^uV?{6nNuiAdpV1n<
zz{OVR(ki>P|4VV!tL6mVOiUZBsij**&a7B*2TQ6UbDSC-#V>}-0(Onm)uzoJk!LcB
zlR?sHsm%2P4zQB}>m8a|HK*vL%P<g0n_k;waNDx{V0F%_+r@7W{Y&4sUVp-3Wv{Y;
zf$9D_lI92K8?ru(E)fLYE3piLNx5IHHB{F{ls2fpf3|TmjcDB1x{dasU9CwK_qJ;A
zC0NQfmRigW-BTbM?tsQ(4uorVYQj~$2UItx+j?ATEIHBGHL23P9Lq*$df46iFUojL
zzPX@ql=N?M3AMPQnJmTtpuhXKji)4MYA2nviN&=J`;QxKI2ay}LfA%N^LUDQ7cFHe
z-^5J_Dg*dWD??DOgmRCiihQ^%WvvT(Pp!-<?>)Zr=)IOYTD3Q<d>s$xJ5hsWznq|P
zg^w$1S9b)vuliyqSEDSA65z)kdNc>zrNdZ$*Y9^wpt^4>vF^HL3bO;Ua=y?KUvwc{
z(5nUFp!(jxJSEheTT`2>IE1DA$yqY_{BfH}i{&$9;4!_+<`x(5c2KZdh{ZV%r^9w*
zZI%S+pMJP-tZGS@ndet~%I@V@I*uDXQ|qV*lGBL&kM;7-LRR0#X7P{FsUj}SQxl?n
znxJ+=;a+6-RKz0Hf7Q!M6e~zXazHr3ZCl_qic<PArnf09q=hh>fp@nrmC=mx1nqeS
zcI_KtcX&E0&k_iXe3&`>p#)3~@hx>a+($yYrJuX*Yd%Q{0mv?tMmk8~Hg^%K_TSpN
zKuCIaewPo_@z&6+jL@xN7oF+VUtx7#Ydyc|ao+2B`*F(>I_g-2YRV8CXA_&(oyz^B
zI?WL<bWu@3>5Y~XTp+wux%4rA_?Y~uxS86$rsj9mwZ!*pk}s9!E-4+6LKh94HCRR}
zL5@O&mS(0Z>36z{G!1Ei@PH;ou`n&1B&Rs*)z@3qL+6#>mhThy(PrE2KmW5knxpfE
zDekw+h9|GZe~S${O=@FraWidh5Px$uevvh`^O;^r&<z*fCkB-n#P)L-t*3{d4~Bu2
z`u=WpE?#g18&M`Kp@my*=81O`_zCyw_ql!>SG4>`74rNPVIc^G68OnTB9_e07U1GQ
z+u-#;Z5_GAA8YS(yLfSI1*(-7)g;2~hhS$FYb)Zr+H!1l*%(*7LN8C2*`!OLrxTGi
z-y@2r`ID@C`|j@ly?ehXgpTj|p?L&(L6tMy4@grrU7F9nUrL!m{}Lhz8N4srAs*Tg
zT${pS$r6!kWYOfo@Cf|2mdl`Y2k^_v1xzdIN8Xj~h<T+e(-d7%y-kYJcfNq};h3pq
zYC&e<rVIk=G+>z#uxRYI!vr=Q5jkLH-X)0uH9fQ`&=J45M1vhD4pamPItswM7v}dN
zSVc7N!mt3CXtv<#D&T-3CE2&D0(MjPNRwuq(#`cBzsW+lxy5-|0BQHV(o8XCPcZF+
z0N{#@sf0H7GltBN$fAq>3j9>I9z$My1WNX8Otd_CZ_g-}+yi&ovUJ3;Hxq?^i+oP+
z%XjaRXPrU&yCVo|3*{ssH3{-H893$U&ikG?BCQi!&zWz=1;5vu3j|c_yZC1f`xZ|8
z9RHzi+~<8~o$V6N^O2bH$Fa(b{R!!(z*F&i4geUQSva(07HN?r{dcpuR8#vNwVEkY
zwkJ}JZp2g0B}7ac4-Oiz-`bydrW#O3!5&6^CpElya0!1J5=|E0jH#&0#$ccrda&zi
zK%dNZ=HiCpz#u3`T&Qi&RYvs1ia=w}==e?On3jS!+k_Fk-Lwc33p_6;-<XP``>yGZ
z9!Y{BrC{+_#|icy?2i-@Rq$@zhl~_WVQfrsGS8Nd&^ES{LUwk`qPdEnD(jJ5_nq25
z$RM7flA~mkHRkcAriBLbDUb7@G2Y%G0WYdPZGof*F#Dt2t1fLGTb|*(J4;}S)aLFX
z92QKW#VmlT-EV_-pO%uBx0b1ha(ekoClKAqN*S-bI2+g{S`<pD2-z37Sb$c5gG0de
zke3Y^irDFIsb*Bbu(QdZAAjEk(jJIe=K{F|{eAJK*vL5W0No3so?KBlQ;~0+a)<z}
zeFTZ?Oee>tSnmYO<KdOYjwK4Wu0q*e`No`+gkkQ=rU<X5!)z}SmOn?pm<awpimX1w
zt_fSxIsJaXtz7huY36-kv01-16yy!o&p6Rmu4@5!p!z>B<)hl(kOIouh@+ynvpgz-
zgc&A4#<io&cwJMEL<`Ec^p#d1qM7{;eg!m25xUniem%Z*3$K0m?qwW$<{ejJ$S&oB
zZ=P)>Kh&yCKS9X8s=O{eEAl7bwf3#Ins>mm3r?W@$;?6prQqg$*rqK}7zrw&IAPRj
z%>xa}q9y#)R$!877_0`7tm{=nO1p#x5_HeB)oSv;N7IDMBPd=7e@RG&=sjsEdo+|e
z;EVC!LhTbZ@NR;KRo+u9K4@|A&yKq|0p1XDT?jH~uLLlO@ShGQ@?MK6;Db95Lub_0
zWpg=-5b{>QR{%EZ5fxxKclQn{i97yizgU-1kQu7E>A2Xx7q7C(#a;c|sZDv+P#rro
z5R319m!0kYlK*r`3TFCx-494b1k(3|lk`lHAes7ZsmXKBTL=!&lV9l$?vS(B5eeXv
z9tnyU0ZfZ>B`^iu>~6RyZC7{pPF#Ys{p-O!2<CzScp$YF-g(_rW#qC&g8}Gt)!PE|
zY|t38lY}OD03KcZo3(eqqZhk505ii(!u7}VWzL3OE?&vgw}I<5m%gT!P|~A#g0nG6
z3bQ0#!dtUi^qo?lZC|x->hrV}nQma1XYo{uUFLi(tTG+ITaSLW4<!P#R$o9YX|5`k
zI;Gn8ig*F6s(of^U6a4>8iv86SLDZe96Ju;fI6u<B!(Va^_omZj&~H<jniqdGvfYc
z)4Nukv(bS;M_g@Bnx^=^NaGblY>a=Nz`5^JcdNmxcue^*-TEkZ$x+X3zL54>`%Ooj
zC%1vn3PqK)))HOTu~Z#ZL~grXCPc5Nm*mvxd+Bv!notBI)4o}m6o!$*Fq9}E3>W*o
z;Ntdy+pjr`mjWe8m;<Qf`cwiY7~cZ2Ve_m&-b<Vp9dWB;9J%*Aq60_~JuG~MladuC
zbIM5MS`!YK8va3_Oasq50Gg{@onh%qBI$gC=w_AgCy%HcY9BAi@;cx_f5MCOScd`X
z77+(Z^v-^QXTqYp_I9$-0;Z1xdUN>KWNWI}2#(lU7<J$PRQf>`X$yvOp3`_qc;Gl@
zgy!5(2STMMCC1PIOp2KiY|LmyI|d)KOk6<Spe;=&00D!v7*+P+971Rn-!GF9_wC__
znYaA`G@Pf1>@cu}T;{$uu96PlgUD9R%vmPWeTafp>aDgV0@nzS!#3zw);<Rd{*3SB
zUrZC#xT$hja^Z#iNYqd(#9wsv__Z*QN&baG>Z0w){?bKwBLhD$EgH}$&Lv`t++~fw
zzjc}JY3Wof6fHAjBNll5iqXX5+pKn98yUGqn_`JXlSoNM^9tLwE_KX?HCuH<lPY7F
zZ>Rd0(RqP?NrSfzk`($uCQYXV1VqVtpwPmaz{%v?@$ugeP;vgw1Hc0E@NwpOf(||N
zP*ey9f{2|zqh00s$@(fpxFgcF&elJQ(ko7HzCk(Yh%dk>;^w6G=VaVd<#HQ<3F!-L
zT@@6UyGY?zO8)01Xm}Q)$utjjmuRJrduWnH$zRpn-;Opc5djDs=c*F^HTO1N(i_oU
zy@SMwmH^^jWR$<_qnnl;vsArqYx}w1Apm8cjx1U}p-FJP{dNj6B9*$}fVtmQ<j81Y
z*h=}RQH!Pg#mLYs#Z(sI&81)f8aSY|PkCpJ-_=h<!16vyOt4u<3wZ)uaosE(z(r#E
zPvoKuyI1fq(ZD`Mbd1mqtLMXPz*GE}>Z~Is`%~bMa*;mE1qX-0fL3Vis-Uc$wV<QU
zGVybHF>CS$f?4vd)`1a)N<^U552>);4yLX1b}-><KS7~sj7p0N3B1=BfGy@}cJht6
zv0o6~AiZO4$R&ZU;J2CV->+B^U8IaMs(#N|*r~f3ivb5LG=I6g1%eiy;-jqadrnPq
zYP|LruQU>u+{eNWTv&&iPNbvm`{?*lK86Gf0+}=N_Tl`c0@AI5ZxPDshCUX}#IeY3
z)Pmh_sW1DT9$@pTV$v@?(xwvMMJYnV{}HyBqhD$esrT$D?+m%TB{ur1OH&=szrX{W
zLf-D|G23Lq8ms?){45%-q1Z_l&4Ts>J2R&JG@h?{FQy$R*C}BEj2o5IB?#z};5_J<
z-q7l#uZOvof&mq^voJ2mZh2<2*^-*!lF4G=0sXEud*w5X28dg|{n03V#>vgE%`$tw
zFzU}q+g}|;6nEGRUh#!F70~TULfooP@2mhRu8SxRx2D#kgW}%{?14i7V6_wiE^i{x
zJEp|&ap3MA9-wLTa0e}q*9E7|Z#;=3zJ%$wTf83_l!$Xuuh$<tV*VI1gqP<R5r*AB
z8f(TNUed)}d1^jo@J(s=iqO2}hMF%$<OD9Ye}J?QEf#v?R{Z5^_h|#=Gnp03>qp8b
z`)RP~YzH^gYRbcd6kj_v^rDSp6ITQL41KNF`z3YJmUTj>U=>fynrW`U4hHkRr3OQG
z!Ixd^JFRit1GowfZ2r-4QR*T2+25$)p(OY#$efg-m3Qi-yy07ZnG5&E?i9&)5~xr5
zi=WTGr2!ra9VIgNMQS{YWHnO^LUn@InhP3d_N3^4K4*}qcVaVlPzJXznY6fL-;zIn
zh#9;rBvYTc-ugfRf1yJtQW~yMFcs-#`@CFWX6TN~2e6f}y*6rUDq{_e@FLS(5E*yz
zMuTm0%fFY7IR<HJ%p72T!Jthhq64r~h^4d5bXkOoo6Bp6;e_QQTPVKHaREZg;<80a
zO2Xx#XNWWVDz!<&3Bqbg%7^bkIvSs?o|-0|5RM3e+WY~MO|UxjU3t<rh@eFxYtihh
z+38yM^YD0v-(UcPbQXuYy+)^eiXRI`b94amK$!=g36~Sa5zKHZ%oI*{H+FAE5=y_i
ze(`eP%GK<ZOw$6`HP^>Hlj9cw-QI+T<g1?&zB|n%9Aal;qV%8yS-t_{UvhGiv54_s
zK8Ol@vuk${<1hS4<<qB|ZScYf9GFxYTpOK%FpPS^6@PTU`Mu`fCH&sWq>do@V@Hnt
zePNt<$V*BOFaA?L3$g1(8k=UOF6h^1sTMZ0<In-xI8^3ExjwWimZq9<cn_BM&)^j@
z>AU;Blq{1s=zzrwmkYsdYqc*Mzc><~L2|&jT5b7`0M&D&ip>H_Rp}O*!Gc=HxEQ(d
zA&A~~r%ApX{+gr9QHeZvn@9CM&@>RMUZ9%P1XzTtreds^Nf8?m**Zw5C0d8xI1IMj
zh7=}GT#`L_v32offG8thdDHLX(C67%ZEI=eL3(oYSc!_L-)Uwyun493bT{xE&In8N
zpYHz4?h;Ud$O)bu@F}XVqTi&vN2(>>FXBC<r0#+dzhOAKpKFqx<OS3IT|QD|_$)`O
zXtBeKfdSC9^1KYNUmuj|p#kq@X>>OJPFM08`NR1qSb){NT|ME21cCW_s1o1&Q4!N_
z`+sGKleRP6Z`5`pLmWik>fA7_GAs6^*Vz(B(o)n@m$1sB4KOORtGPG%dsZ%f=z`-E
zbOz~&;l)oN#zAfoo?NU1)Stw%`>_arPgUT17{Cz2KnmLE)ap&{a~}?Q9J;h3?Lv)q
zdFowV3*b&wvP0ktq=bE0JmX)w#jM6BGWQf>2nd)qU=WF411S-?G<ZY;&8+wk6z)QH
ztnyS7U05JQkL&th;i(<)N8seu?mV7jQ#%pz_a^@H(=nR$*vrrGHh42c&&^us^gCjw
zPY%Sd9|fqoy0NHs^JIP~aXA<=9PGD<l1f&fU-8d6#Z1f21t`Ik7UJ^YABa2YJ~<Yz
zTOl~oWwCI$Q(e-O8edwntIdCF`DO=7RC5p1#Bm+ypR8wS@GWFw{sc^5d8o408@>MY
zvjN%VbiaR_4(G~#uCM@r88dIBL+R15d3X_gRAh1V1#%2l;Ms0U>Sgr3_z}3DmW<nn
zaIXpHCO7PO08^WP7~?IE`36J-#(-|H(k-%HcfkZ`v3}`Vz%({unPz!x6V<@1p$}$v
zKLfQAW6`*HeD|9Rd3ducsSy$wO<C$Hk6&m->nr8WKH2m^*Y$65WRzustMea_M_fda
zaH!(+G)|}7kwwtK_m2&g*__0f%M32vx0Rn7qvE}P7C1c-%`CWS7(Ckdez|jy;4mP8
zW{DHFs8Us83$|j{)d1YVpIsxcT!zpsN!Bf};DI{wCJY<2;Mbmul_a{VfxJ?~I9e3F
z=Os%VsW7RyF((K~abc?L><TB3@X6E`>DP6oQ)!R@eq;IM<8l$`%M8S<)$PyC*c0gm
zxd&e&jvlGAsK}M!@`{KUxx^y)_}?%E%-?NBY8gqK7PbSZk0=0n48aRwvNBJuwTRHF
zoal$fe>X$fer3l=h+8%z51=Fgwg}$-qY@}5%9__HA>3U9wew^K|IqVSW(?RIX8({W
z;)^`gl$J&q!6hlk|ENU8#sWz7-C<XZWRo&J(CDB(B?!wlxh1h>`zRssTyfa<_7eJv
z(}tBVo%g?MO`a$q0c=h40@-a^=CcZL0oyUw&mnX`txSy9QcXZB!7b~UWdio@1si1F
zOAn85Th72VBVxNEFSzL;EQuYI;<iqc>p3PG``>o0Xj~(NCi6e>{gqofki8>FQpX1p
zH_+931LAfc@sPVx_JVj0>06Gh*<mDSl+?OuKAsCV;Msj;6iG=rsVY2Y*{L#8?Zd@{
zwS>sy<wEa^*x)*t)fUfeYjnSXzC;K21qys5$Q}Wb2s<)%a38*SI_hXO<*uxYC7!(p
zc@$iyFsZk2o1w)c%%nD(HQ!JmWAk^lRUo+K-F2g_;+oUCps;*Z2IH<eFbRE`tjN?S
z;Juil#HGgj4OoofH{wUxcW8-5Jo0=TD!IF_qQcvo!o^(tz=bI0De!?IDT?6aO72<}
zn+8eyy(lyN44(c6sx8ZW$+SL9kN@Uy0hHDcoGyOsb!#(en~j9-hsvgNGrl@O7_7v)
zi0tiAx{b5Q7e?X805eFI`O|HPo*Q;)um8UbK*udJP<?tE_08$xVlqg_e{3nOn*m<G
zlvh?M?#y2g;{gw#q!YHUQ4yR03L?wHRT@=QOrfbf?lch8=^*?e1bs!1ulW)8p!Uu<
zKg0&Vlrykl$#o2er%O8G!EN5orPVL&=SoO&0wAFEL3An?Q=*cls|nV2$M<gmZ=ApJ
ziHX-SPCoQ~)w}!aS?`~0-O7FJ`%W*qh12+~$`h!4!Z;aP<lIFm!r!vJNnNIT*?+y>
zG<=GKt7F!qrtxebFWd)Y`6Qp(B)JgEqPAKiF!RBgUGeUp2JRDa5SB1!2v6a;x1`)k
zmJ!X$7{~tF_V8F!%QtkI-R<#A#jf4pua*jyp2SotBlQl`&amPOclg+2#c@9wHqKD|
zI(cX)4W$PO&QDy!*pxI->K<t9VuX!fF?O}(Q4dF+gGyedNeP}&J#*$TH+M_^X7=Kg
zs6!(Hl_$9xn@Ecm-R#YUKU>*@0A^tW##Wu0rnMPb^siqC^CGbYmN(9dW;R?Tbi(#5
zL=G&C(dN&*pt=_VFx!I?|2Y7*e>Olen<}KV70NV)2D8!udlT*Ve`xv!zrNe|?RU1d
zTDEO0+qPY+R?D_++qSi^v~0I**K*J2{{CLizi^zm&f~aFOj8-X2XR)O*pH99t%`P&
zE(x0g<xD}?KZ2n{{CwJn0f5riMwP;N$PH5QKkF1Cc*3<b2D<pAXD>q;-4G)!#7)`%
zc<hYzaGX~Hc<@NdYFChn=mII1?V5z`yr*TtXPk2yroIW(*N#KRMHZ`)Bp6zzl<#5t
zI-JrCG!Wp++^O4{QCfwvBUfiunCaxHLbllB_W0~L{V?k`)aAg6Y_tDWYc9P9UhmHE
zgi6VDM;bODP<oqq&D(+tu3Cu*HL^u20?%7=3%86pjgGZiO())tJFX;ef3Oz)YV^b7
zWxX*g72ImxR}S~Rmw!tW+&{O0C$BF=%ch-%yf95j>XVH3h%RejIIdeF-Vd>Hc93%2
zEs8P0OjAc$dQ0i2zu^&<I5mpzCD_i@5w!Eyrcr`#9Y+bj^uxa8dVjNHm*oJGg<go>
zyZn@;-<Q=Y!HkF8$+%4qc`Rp=`ke#Wrh>;n0!2fyEz*Qo0J}5elZW<r>nLBfOBWuh
zhW}D&=5J(6f#1g}>byD1Vz?F<1OTdzz@8%pB&#7zb9~D2pNv0QD7i{s@560D5*wpC
z`nRIstIC#27&R~;Qmq|KqWBPV7~J36m2sc#k&XHW?Yab1(15M_i<6(A81HOL<ukkC
z3LAIrz{UPDu-Uu~;fub#JP!NETUEm@ztsXR@&{<SAt0QW3#D1nutiHR(vN8MD;1>5
zfe#m1YnwAcc$j=mgBxl6Nj=FyI&T&#C&)8JYanz5Q;3cu!lSgiN!bR8;2I6>uP!R=
zNG+b^-2i`ru(dVsIh}zIZ^#yc<(|BP;^NL=c0f?Ankn9kX<XVl)=)V0<atcE;U}(<
z;=xZue2FVYc2<MU_d{GmwMfHZd;tO|ty)eld`07z(2)FH?jdxz5t1oF!8U7_mT67b
zS+r+-+*27AC}%rmZlRIX%?_Pax0N89yaRdD0qs9pKQwfPjxn+p8F%F-bMEa^HdMP_
z)Jtbn0qs8^6+O0YM34rqXf8#WhCa_+(TSJs$mb)XePdvH*|_4MXbOC>Mx{bYSb)yN
zS7X6DxIy3;^H*z#eJn@>opeaz?-{6#YcDiEc9VbshVbb+s$RHBvjtcy`!XVfdw8bK
z+^*eRL|QA`;_<OPJm>|gIWythUC1UfZ2<pZKKB{7|AiH}zsaoyGnvCTi^*<Ekwpg(
z(PbYF{7F*MS+HG?azdPd(F^@1^tg-zx#bqbd*+pr7h-b@&Y>;FTbUK~fq0Vh3S3pS
zj|D}Ocf2o=a__|E--lzo^PG?e{n>gEmtq9osTk2?eSVu^)oN$>1GMk^+;pB`RMEE0
z#JnTqKOCIdt~t!<YbegQz99bAgrf~EQg9>t0u^^EK(q%EKVLgW#M<bBJUY@nV1e1v
zsZr@|wZ?P<Zw6<S3ECfsioH+23s0F_jB|Q}lQTVUXeYt_ah`Rp4(S_thY#eVv!~k1
zY1Eig&Kz3wY`B_lka{H_l;W)i)B{zBBHcewx&*SQgRmYuVEi2;Oi8zxf-s04xF}o{
z_1p(uU&n^dRM+ydY=C?%)P%dZ4J6-!lN_0Z02_V0sTHYGC2q6RO9YcXffz8O^w|M=
zh1D|M=u02rPP7<U>=v9#e%J*CLe9DtiQfC)Ggr~bfS1Fu>oko`<ho453gfl1wWB~f
z&GIt-o4C(S*eSy?SC<*@2Q;GV+fsG1_Mj3H=L{gU=fWyxTqe<^1^5rO<05n?T%~a8
z&@%LZ3z@#X%<;7SbvjZkMo%KpS5vbuXv(6y{>HHfdAH0RBFE}2b1#9|ZU8C&p;V+)
z=rc(hkqCru>jeL0vWuAedGG0Ji4$%@Srs=*V4l}?h%nOHNVU1=k<@v~89aeNRdN1L
z+@K#W$TJy*HMGSndGGf&Yko~BQ+W#4UM8c=qE>KE8)!<mQ(Scw!?iOG%l$rCc7s5+
z;|d#0E-Ab!xkUaz>)5~$yEDwZ<O#WGp<gJmvg)|c{erSGI<3G9wu&7N@Epoa3@>yO
z1>j<Ef9&xk;B~FcKWyaKly@8l8bbt88X?DWinjBPp`aZ9wV3-hP0rYvtm1I=1?ZJI
z4_NUtEQ{@Bo$HEN`18Y4$V4U$00mWcK8?S?9mjG?r46abIHKcEUv5*?3o2vGL2WOR
z$8}ISVb7_Pj#W|5z6R@aP-?P4GkBtVswv`V3NkIuq-N08Kc)Z0lll8Te3}|$Z)hsY
zgK+fj$0Rb$BJ7VMdt`GM$_HXqAGIj1%B!uyCn9*D{1<UmW^0B@6chj(Ca>FoKC7^h
z9&Z4aPhu5wCpAMDBeAVKXvbM!7az96NFt?Q<1yYh@LJLI(L<U^6%XHd33p;E)AoB`
zax;2bUt-SU`$8Vq+;$nw?$3C*kK>U``ZDT@zHmI|eknWwN$y8D#qG72=y6n8zuliY
z=24B5r7EG@pzq@~j(zci-`g?8eAAVLeZCmTrObAadzR$;$uW2zKId>4R5v~O4klvq
zk>Sv#zC}_(?Ft!*7zH?n@SB5;0~`3FN{e~a4kgu^QFCHByX3R6>Y0#0O+YR6>H-zY
z=r3@>A59S9`di74uEz=;Q`F|$+*_q`=}>blmuuUWL*Ig*@=#OUY0r3|{f{t+3j9VW
zU+{;XEnMex3$CQ?L8hU%>V`P_=a&s<kM#|aB2ze+oGsfC!bc>ds<y>8FZTTr^f-Y@
zAT))HItF&KE%ql@(?gW96)BOv$(B^@6gK?X&;9g;4N493wbTRRpPEAPBn6YEKE=_S
zGM%%T%N~RomP<xZK$kaWbbdUKrz1G=w))i027vi5tT}b5>h0>#143lT41QDi1?v#!
zk+utOyUqwe8Yv1_*b}2vfw#C>7QN1H&{>QV#e)yDER~9-g)42Y@o$*f+sZP5$5Z+N
z#FLbe&IZxz6~n3)SlM{W1)^77Ls!dHG3%a%d;ktIZ9Ugb@SZHU_kp0XtcPLXKq41P
zdmMq=IA6h@IfgvF&hWB(L9g@YFb^u7y;zwO32jVJxib~|JEcBHu7Gh@M0I}=n__sO
z!Cot!+9;kxYiKS^tnzXFaQBP={ls9-WwXk$B>AGqQWb85gbPUuNxe)GkS|$X5tZvy
zu9X=i8Ts=clyALLX>^nRomt&LqumboWs1doB*reR>94vip}Wbfkn*+g=BFtZAX?Tt
z?L*(@z<-GxIV6aGi;6`-G^Jqy960MAH8kIN&2~91X!oaSe&7xhl(p28Woq=G95;6H
zEF1Cl!h;ZUn0LSSFzhcsLD&&gh5I&;OLC(xv$gvXi2p^=9WN9G?#eJ00cblJexSlI
zml7Llx8BIxI?iW|Jf;Z`{}nb@Bu@;X&VfUondDOg4RQ|ux;IY=A$SICpcyqg``aq{
zUX<2B(Qt2_!90j9m4|)AtPZ(L!XL1WH=E6=AaCGhr7&s>^314O{rg38`Ui#NDE}^F
zvkr6iJ2Ka8!r~y6_QP&$KV*B6kB@tFcXT8BV4~AM;&#Lw8vIxPqKZ1@hI4Z?#^w1x
z`ndQo+02<~KZW#_0*86Iz~*HPE<6vHZ6OjE5|<V1aIX;5x;p7PFjlA^gEcoM;rT%;
zQmIc}>`TN$DusuzvgUu^fH$2R!h)jcd$omRMBTU5c$o?FK5ys*Z(q_&<5N%=fmp13
z2-%nJfKRJ{`x(ya=*XVitlNV;Wh8waQXv2Hi|*z4&K^l>-nuPp-K_&wp@k`JZ=NY|
zoc09u{%T*ywCMX?YnRL`_9`>pJ?)HYIy>6p_E0|Z#Uz{{<;m@K--g8AQ+MmrDX5#%
zKz{f&p&#2&+*go>Y0xm-%8|gXtY7nE6lKWw)i430f=6e2AL`cYw&r&2W5M^EXgKcu
zPB`k~bset7{q<N9>#ak89W1#)(x?4N1aLf-l&8<hc<M$N>7iuGJxM#{#X6)TLts9g
zQ>>yn8Ps`UnX!f=$r?%MLoT+S^bN_1vh@OUSH@G>AlWaQA6dO8Nt|4vWPibm?@jl9
zxj%7?_A^DcZ;c8E-Z9Oa-t^ba{qFYDRCy|N+KV<bAVi?Xlt*PPP^|Q=1iLGmcWQsw
z(1I4?rJ>+*dacwXwVFJrPCH%mRkbhh5kKN<)1pt+`x&TM%(TyF?JMJobC{-PTw1gs
z^)2Z2m{%QWh~f9K<)cj%7=K#d{=4CdHa}U6Se^4dKT`wU<ig~rgl_+z^{-q_=zL`c
zrX2M0LpkAaOLXGFcG$TWMva{ca&)U-@Ua_N)!`uv3h&``oMrMHrAfVOe`W-tDZ+!<
z*da5P1bWy^g;Qs@{*rT-cT{jf_vUsl--owQJ?SzhN86}ttrLf}q_l=Zzq_~9bV|(;
zetn5i!PIFidC~RV8}28ZX1W}Oa^IvBgWwwhW*8DwnexbyAGru~<6*O~%fTk*lO!m6
zj2@)K41VU|NL^8IjmjEJ&~C-4w?1ioqo!gNgVlJwMrhyOPc)LtB(RpL4(g6OKSBpC
zOJQ!;szUrG=REZwmF4L?m1~dF=7SnUk~i;|dJVvVsBgWrX(dd`f6pIAH^y&G>C&Gi
znihJYo{5Wna^$0<pY}=!M_TQO4Zprdq-#(FYrQccPqCC2Mf4;X%E+KJzDj>f%#vDS
z${p6y$d-iw7DO%;M06Z9mS6pc7%-e&LXviID-L(v)NCVi71a`X_{Mh5^xP%gzWCcI
zdQhgLx-g<cDI(t)p23KaS-lGCjoeg6kH$=gIN`@ve3}k4JXEXR{ldkKabf!F;1^X=
z3g3|TfRyyRaR;T6Sa@Kvzionk-1e&T%`+LnX`bd+q!llIO#%MKka+yWZ|ljz>SS0b
zw--B$XCrs*o>58CKMU;wB4Jb_|ClMS8cfxo$yUx%gfUUhueqSt&5C3&hs=t@=E;_l
zcTv?*<JqC&Fs)wUqNpr1wYwD`sITb?_7~iWH)qTgrhN<b!t>-WBw^v4k+cS%@F9fC
z#0U#t%?lL`Z#4{)RxHJ)ht9ww;jLi+rve?WK+J7(^h`;yQOhk{m#eE*=r*MaUyx#?
zViu$1HXQQ?e%WQi8ZOmDCLb|IF>3MJfha368<YcUT8s<E<e`_vd<sixI;4JYR4b)*
z@6hX8$RBfmB5S7iaBF}oM*-SZrY%0E*&OC4bAjNW{Rd@vdgCh#y$%aTR&TOf&1c8<
z4q<dMn#ph<5AXWul@CTi*Q5DKEakY}ZE6O_XU=CBSUzv3kOlfW8ci}1{cK%V+6oz^
zl1O`!N1lfDhtR<nr^M{!(!Xu<=nZ3bycO$Bi>WxTUcL?$?O>fd`4@icQM;@2bML32
z5Gt{Q`tx3yKfW{kEW^O-Wds{fB-WQ+giP3OZlV>P<s|#aj|$e>V~s>_ifpuLk9Y4N
zV$I>H;6t4KH!pF5ycCzm@pVvSRa`vRfzWO^vfq|lJYvJQXB4L3NXP{djh7#<GmHiI
z`m<s1+^vQ~+hEl=`#zfe3PIGZI4rB$KpFQ2mWU~-rQZHI9ERLXCFErikE9MDy(c8`
zJAlhbAugv;IzRY66SpKCA3jseGwU$T0^D=b0fs|XJ|>EP7Il_O=8vX&sZb=F1P>VS
z9)?lZDFi0}4mu@hAK{Wvjv`I$ix96(31Z5n6Rn8xTw&evMTKvKRp2^7_9;oAKDcY0
z*9l@uPyW)u{%~u>-j#iQCM5}-hd2?*+S}H6u_W?*x|Z-LnvG&Tevbd=XXkR-H{U2p
zJx^;_(cP;-l#1hnyOyiDzq}j0EB&Ai_hFiz5Q%notp@FbWc5mvAcY>=yZUz!%RKqZ
z-5JE=>#yUwhe56jZn+xxy<)0N+BP4Jst4aszM|Cx_V$J8Kro;(aN~g;&XbhaaB$w>
zPlhEQgh3QjrCZ9{TP{4yPr*>yXymtbHSF2%l}AZhHV0@X_o$yO!mHvmHgPYNRKrbF
zX5IGTNBN*XcBc!qzbkkpst`D<CH{Ty9G6OCdr*Zxlw2Lr0=nxxYwuLb&I_9lyBN+O
zGHp0d6-3!T1F_x;jWh5T@ZwCM2zP@WXqOS@+|Sx&q3D)*VN*q_3G+nJSG=j6QA}BP
zwiVeaWzL-wsK|@D2Cvkh`C`7AyQ}`k5AYmKV+r$Rq=+cs!-{+j$BHU1TTt%dI`QR~
zdqO83bXnq$j7smGsZNp5y5w&Rp>3U%mjZvYWSau*>dFG!^4%lfzs9%ce(0xi^M>Rl
zu903(i^*~)DeJwlPxSdlSCK~cRJ^MCkToPjC5T$N?t7p|8V8VCD(W9a+gR&C_WmJc
z_&dv*idB%j5d-%`-2T{G`gg&CA+a614VFL3o~>ieCxg<&pm4%$m_b@veNM#CbgJ&X
zEfYw9sre|lVx$H4Q;hlKc;no~)UGCFrl@Z7lL={`n0%8qKWTy0gQ7)N9J5sjZ2~`V
zwVC1=Vrrc~#(He2cR8EaiL<Tou)@~$X?bXDB+Zn6d}EiFe@@bT@RA)yZT><E@#KLL
zX<0H|X?zVf?kM%!&9Q8>C?Mpk#=+whwoT6;0;7&@Axr~}mfD<Dx`Z|hJ|>qjX)6Ek
z)(&dLO{gd6vsd6%9}#Csl{-ihjo^yNRWQB~Kkcu_TFAbNrB^(=4LPLagW|12?l~gz
z3&u@ftVUU8BW%vJsaQWR=)#J054geo6$NWN@4ZVU*_WF+iN(%}OioS@_!6Bx#Zwl!
z-grfn_QfXiArw9274v(83fj%g_un9L02i^*+{u3f*K?v|LP|8N7@@CnOs$FlbZMy-
z`;Q7H@MH<vQ2K>W?g=}$qP=>QI{)r4hZVk+SABw-!4F%#SSbLt!SxU0Z*!iy_Sa!m
z{chg7Z{5e%S!0tn)zSmBgl+}JL)<Iqp4J&E23N?2)l4I?TNJl6V5-{`mZWWND02x9
zNb|t`nWXDF87G^iJBGmlrx+q>L3Wz9QMT1Tt|qc|fD#-n@>~E46$jz0Ga3t96a0w6
zF0H-sR+*}U1KpvMZ)Q+oN^TZ5*6jDfOhKrsfXFkB58FZb2yn6WZBGV@oR37oR<u0f
zEy%x4l5U~o$HrH45GPNy<CkHGQ5|5!UDmal8Zt;@yfi-Vv}r1y^7h8a_G2zHS`$)G
z9b6bQeLsR)EWK(2m%si>#A-$Kw~-j3T8`}g`gEo(-bH#P9qVRttj(&SAMQXcQbzwP
z<u-WV#qB9&s1OAM=ZYN6AA5JutoYwTaj!w88bm|s!}{|>>f1E51-Z<^@4N;5+$+#o
zVz_G0W`D%{a=nipV?Q*X3OhQRe!e3gAoH#+BXH}c3SAq=MAPIrdH_|H0q~Bgbt#co
z{O<16*kSt$!`b?b@g1|{8)*(p_@gs!WhPR(^S|GVb#frZwr;pN*{8B&Jg;JaD5g*7
zqk(Qr?`EftBBV}s*6lZxf10#FZzJkY;+?ptXkA`8T}9}5(?=hp%j>_6isgqxZpT&a
zY^q>@^@?%vuMbRuvPLaHXm%0LgVUfl`u8GlBmqeKr6un5kqhT@zmFId`#~hdfiEd%
zo0;#ySVl~OLb{IW4J~b%^WyL&S@6y-`zQ#aLheZ&-%-~wAxkDP>u=PQzoFW4GVBaZ
zT*++;#wLnhff@Xez6iO+qoXa!ykBEEWI}^UrOSnw<H>5Ci@|BaNS<kHIn|e<+s1CC
zLOpD$62!*G?~|mlEnygRPr(At_YeleWbU<SaIxN&b7kn5#SmSpL(Q@wPl5h{LfX7x
zo&nOfUJBs1iQ)b(*?^GZity{JPO960olpy4<x;IKZkST%rJ3_>w!7=ikgJ2Zi@#{J
zU($+p!O&6QiEmo5;~Rrm*b;lvIX*-`PJ~bY!}H!Lqi?2*)h_LeTW<<>q+JD+|8z$C
z^ssXM-%m>UUgn~5mU1aY*@}n+QL*N-q$<3BwYxvp1s<2nSb+N4HU~cE9>uZEnUo*G
z({579|IBMT@>HS_fQ=t3Qn?U`A-eHO(S8W?<S-<-TMs`sng3>G$~T7On&lhs!UNpe
z90m(gi(`>cK(ZgSC~EMco|y0rNRsLow@OY|7?{R4aD}}Q^D7dl&h^N%c<Vc5k$)es
zOmL?B)-4v>xhncC-~xI2pHZ>(%$lDac`?l2*6lDR-jMtUl|gjzD>(GOH8o)1r?6TL
zLIX~(kbTZ9X;po=J^ip_10bWMLl$!7IvSkjD5g){rg@IS67GGYNUa$|3c#e(q(~!6
z7OrE7=6<qa$6uGk?+Mt_7C^^;o>Eum&=0#6EfSNnB^jb>j?P!pV)MY0>Et^vMF(9v
z`!Xaz4+gSJ0Og4d*0Gx=0>`~&wep4<1<ieI8RktU_<Semt}-{?eNpJ<U@JzIwD;cg
zSqtu*WHSjv{LdOFcGjB$)vbnx2=;{^2k^xuT$HZ5UTb{MI%QQNP^Y5wEkHgyO#1_x
zh>c}o<uB#nHJ17WvnB~KWaCFTI7#aV4dOx8EC)0j&}ZZatW+zJmRFQUMKXeC4C9d@
zHC(;_0kW#SBbYx&nHlvH`8>QUjecM1iwRvVj8DOjC9Z4x%W2IfPxT{K2>;3vpA5X1
z*cCvCWLEEa??S889&rLB=_ZSNEwRu2H0AJ~p!)iXaAC|b`FgL|q=xiJ&x-J%Z!?k9
zu0=+&kFjF^kOg0{PoMcx7gI}(US4Gj@%9JKhKHT3yuMfBrLTj62;OT|mGGR?hJw?s
zu}S?>(vrIwOX=jg2t8b*la1m9i+O{pLCHi3X+N^~&QKS~K#!^=KmDm*eX5jL9ll~)
zq{wX%_`|A2%E&Wgrx!xqqy(Z%eTe;}fh)m+qs4A4zh_MR5}B;Rzda0AOA;1f7uu!N
zFPXaHfG@k5=q+(Fkx@CeQ;Z?ELX<o6CVTi+g|>lm=CJxZI6o)!gU-(mH&VIVbqdVr
z$|Yj+>NwFI!y|?<qM0^tlciyzUm+sLK>j4G7#m0?J5@LeY((J^kEmJf3b}t(nvh!W
zm&m;2XLdSf#?HJ)pcAmT&RzZ)K=NOct{!Mh&G=)LwLIl^oc(xd!gfW%hlgsNRKYM5
zKsd>;yb;G~6yjm7ql-8QKXP5lQQc?y^@fa5*-xF3qc<DanWR=dHPxjD@+sa_oobwR
zsCb#npc6qFA$#}1;KKDd941-i_f_NIAoV}5?0jw{(|hQ{R;btse()4Wv^tzCcL!P2
zf9>f!O-H$R=iPQJV}CVpvdEyN<+`8H{W!wYK6R|8@Py(CCN)zv@l{w1RfMvDJu;^;
zxJl!W#fsx&LZaShBF2_Ntb{&{*85XU7fHypuC8f``)@p`0`APR>!U4YmDoR3mh@Ld
z--b?=OtrI7kC<E%cT?&If8q8*$`cMglf)y$?{%R}ZE$}=CH`MU%D&FNUAOgu@R!?K
z({*&5OSFn`q#R-aJG5}@{*I5won>;~GS!qfDH%^fi)Xg$@Hgj%NBBVZHsRH>NB^BS
zB0SJtN7UVMHY4dPb7NVI4=5m6f3h)bE0rB|?@)ecqJ~FABtTLW)c--xvq`#18YZO~
zK&-9K{NtA&uDsT&7D$D{CA=~DQ*EtT0aqYJ+!{4{AhQKm_6PDgSw|1pDK75LC|z)N
zQl{J}x8#4Wk|J#n&2o>re3}n@hXf{6v#Lo!l5KKzmWr?$w|K=lB+M144=Fh~G6?aw
z^mwRGuw}VTus*Es_u>~8%@dqdcja&xet240N+kxKNTmDQtPx14)b<;9#4o~QcMOn4
zlFBy@wP7exFQ#JlPagxl_(-d=gcF$g@@-(VV=K8D9T$56{N9Y}C{XK9Ehkg5Tn5K+
zjd@Hc1+@2KY!m*~(WswI9CTvhhrcwwh!IR|-*mIK+HJ^0RW-_xb6+PcN~$-D!z@V*
zu}7ddC}A*eB5)L&!rC=Fh0c|P;RF8AFShJQF<LGQ3Zd;kLox%4mSpH2f}dp@%B9bz
z%i<lhCaJNqGSE5)%OFg8R$gUZj(;Kp*2Xk2{%%rWhXhM2hu&YUK?}cl)|_~VjnufT
zF;}D#B(g>e3}^p}AS;bIjw{NoI55$OGQ@!StAG0Nl?3%@YOjv-k?rLZqGiheJ0z>X
z)Qmy__?x5++W%=RtxB>E3UvPYQ-LrhSP*Pq3WlX|W()2gGLXn5+W@M%;8TAeNgkk7
zop?`0fgbfpGQHdZ5f-!F{zQkcc)!`b8D!OdI4=W_RF$8~r3Lq5&wof&M`5VX&Yq!6
zI^n9rT}4Dm66^832?7q$9O~W<<hu>8XvOuX>G8Abe6jA)mPu*DOn~tJ-wOb?jzP=j
zypZz}q7>GQ`tgTusLEb-byEDzV7ucMwNmJSR5<(}ZT~;*#*$kxRFu|@n1zP!&CyqP
z^28`wEXop%Sja;kbG6k}jAkkVtsh3-{p4#5R50>9rXICxdU&nt$dwWw5<tGHS$cf4
z-g|nE`^Oh^Hv{$ZV)77!dA9dc;a_z=hJ||-&%73mN3U{ftF?!kzxJ2);=p~k?vQ}o
zvdCJMz_)<Y<l+Fz!GY5PQu}6`$}{`dtS@d4+ua2?!BY|GRghN)G3j?d8sCFJWiarP
zW2&u2>jM{zVvIFYDHq*zM3qEoMN7{JoR`v_l056PTg|8c)pus^Ua_=>4ls=heketk
z>Bt3kpbiI0=1jnV1Lv^0Cr-1+jY>sNDE~qs=aa3N-y&MjRB(vpCU2!5Cdsa%Y*^lD
z87P;G6c2Qn@ISOD^25?b_$eK151OesFieP22kVGu-Y_asmP(~`+In&o4%hUSw#tm2
z(QSoU^cfJ5VQ#N8ut*N$gYz^uN*fD?y`4LbK;n0WHcnP$0GM@}ZIqK|<Gs*KH0HY8
zD)KI2l8t!DDEwb7jgs%fX>bHu(ZlHH*d*AAL8XHFAIx@HAs3IhlO<88K(xksJiES3
z&L1@rvJfnK7y}I&BJn%Vkh3n4g<pk-<6B~58XfZbir80<vi&%vVDVPcT%hsnpSfWD
z0%B51g>_rj&9SO<PDqhfB6!oVT~!_HuByGP2%97X?ca*pPZdY2&cyx?>Uc>EjOK$9
zup*0o8fo7xG!pQU`<sHZ>@V0jz~ej?h0p8p?v%>VzMogd+*R|w%qrcvhdTcNXc4}u
zqq_PaJvM+E06G(RVzkcTY#;|>7+;Q@8#%6-h1f~>U>%ggx07X)XSEBd+zHdDC94=z
zJj!hjGrDRZ&vm=p56y)e5mGzf%{nSy;Pyu;U?ougGInevd!IFzkW!;dO@W)XmN|Xd
zgq4%?e`B6l4i^RGnZS{kacLg&9|FT3*6wg=Z#paypYQp|<JFJZdvE+N`C@+M<Ign-
z%&0?GS8Xuz6If(*o!dEkb{}bbw!HOUqA1eEa#o8qXTd+|$a=)BrjAujb}p%a8C%L?
z8Jt-!D-*bgLKB|)p%fwg!wT2qwI6Dlqq;E`ce7}jf*u47ndDRCng5Vd=qNoOA3dM}
zr$x=c{FEcaL$pYXsqA6Yk8k?4SetgoUzaLtHRluAn`Ee_f46hWTY!i+BCr}X?;S;(
z$196*{W|tUFM_t^L&hO5SONl@dnU0eX5Agg;1qrOY9><cHCe4{Y8%2)-WDiRpBw!>
z`&PGJoy!~>0Z`ah%DDu!HYI(V^<V-0m<pG3k7Dr2=Gp81Fjxnh$yl7}`d-*RC8nsE
zgv6u+S%|G}WSfDnXn%z9<au>`;M@uywftekN74&n`;&wfNqWJ~+FR*j$@y^h%hIUt
zCq0ACYDE7K<fy7vkI)<TpTRr{8)^GAgcS3Ce03WJ<C}N8L2p-*^1tNkabI~l&<d>t
zWsd8bU@H7%0bYFDD0notE&q0*lA@KKTJ#(FP+Y~naG0SrCPvfeN8>IZDFL=Vbl_MG
zORNG1RwK^#iiHMy(A_}ixZ5S7470*E@O@o^2C|_$cC4?4QX^^Q@EOtg1jeCY#dQq?
z_btq*``onnUS`31ZiJ&<VqXApS9F&hMab9h%jitcqz~gNGl7Xb`}v^S#*Edlu9_eG
z>-3%J@eL|piM4N0*xbUX8i-Bj|E+;@66Y67a$lLD%p)QB+CoheF~BlbHsx)FQnVbr
zl=KQ8UYay%w6LHBDsH0P7-gi4!TfseR6xn(4R-%vU@3adiGXYU+kXE~_pi50e5dAU
z;x4D2rxo^MXVvwPXlS#Zr(bg|g^w`WpS((NX430Df0xP-^_up0Gvj2023?mBI7_Q8
zd?yLkGp4VkzT$s<XQP?I5+h^$uB=5>H3_p-?>d!AfH)%H{w3iospKErx<ra2N)11g
z{P0Lq)5DCjX{(&KrXj3dGg~OFaBI{OuTMj3J2m>3r*uWE;=VG0ZPXii-#^ady{Nn6
zY+etG#$7#<h#S#8<8E6xvu76gb9y0%7s=O<#qBulWQzv3YFv6QLb7y+9FbSYu=1N;
zSPhv&pWY^fGK_7?BGeA6bF$M^fKcKMmE5G4f3C*79_lv^^`yQ6s8$@8HUXxg_s)f-
zuw)%3&AM*5-|<uZVlB;3fWZE28(t*rWBF8jcT)VxJ8a7f$_>@;vYB0QfcH||6}5^r
zsk@VNUwMM4BcSt*@4B?WLZ_i&I;6!wg}rck3w&GTH(!*BOR-r4n7?Fe%)vsNtdvY0
z5RYSb%(}pyV4VL)rWJ>M^pDo3{XJ{cz{{?W(Byh$@T%0t-s0#eP5a#N2H3{<#NqcA
zj<8Y}tX+h7vOGyE51^$X)&fhbh{kv{@j%>@JIVXtVGL#~w13FG+>1Z;<!y1RON2ph
zWufDT&_{|aJ~f)@1dpOP4@cgWDz7@mMVdk)zX@1)RnA;ez>4eqS<g(P(znK;_kNMc
z#Ll2cM740y3zi75eXj})#?M~+<<N`PB{>}fj9G)Pg$qqQMJU+32_|T@Q!Woohh8@Q
zBI?P`OsaqQQrB5kHtU-7Cb8M>RT;+)4(ed~Ci(IJAp9l3FgiKST!XZ>zTMzs-+_4C
zFMpIaEq6LdI!xD&II+?K>6uVtEX11B22oZQ;7+ZM>cFj`svxf^#qj%BoP|B<`Pakf
z(liYkm>fmMe2O2P;zTF8a;K>skVFzPsFP|-^|2)K@iD1q;5rz=<O+Pk{tCF8t&=&m
z1sg^1fLJh;*5TNbt7E=Z)-T`2Fka(n=YTg031z7yJ4zawG1>`bm*V)o4(a0dk7V=w
zEKjfm5(GO7J+dK7*hEXxj_$bAic)0l9+Hc$1%!Qt-0S2r2L&S&@ckS1H7&w+{qPz!
zlwtG2p{pe2kzKff)-G^)4fDtAzM<PYe569MV#8dVIc}=@g><7S>L>=ck<%7EvxL_G
z@uhX5FXgCVnb99tBotP7MPye=wy&WVkEvx;!%UIKzEH=~Q!0R02Wbv?uw~c9^iP3B
zL}lDUv`MU4B7)eu)~X^m&y>&xY4ZDtYp0Qh7qdQRLa=_84T>n`D)XCM<Xcu^z^Gs$
zcwfJaJRyBzlT$A|MXKzt<yScjoU%$yR{%|BmOJ`OYAbMU$-1Om7zsw1fowjV1o-ED
z+Ms`zjTwdkxMLG_X<~FI)}l|<_^Kf^@CX`htzc73`zlX%&G?I_;ztTJGIXMl0LxK^
zKxlQT>B)su12AHo`|sw(J?u*He;fp#Glphj+r)T@h4+_(e6BR?O4K8m<*uykBz!vx
zXWmkw4OU(PFCs2)m&G&R;5NQ`DeE7b=s9zw>I-MZ4W4EEt)R6aBW+E0yx-%mwSKe0
zF`)~x8V>?jg4Nm|mys*q;m0azUsu$3{Bd%K_`Za{o%;3f6ASU=`(d<7fUq-|UUI~-
zxVnJ_e~MBg=AWmZRC8fjL#>X*U*zFTBXWA|Vb?EQ$g-cD#lkL*cd{#xJCJDJoK);n
z`GPE3%Oud%hBozEYd@)UBYh{duy#$^9x}>N_gr*W67OJv@~a!PZ1!5F$HaOWAYTIK
zJ^q{^xo6_seh|`_o~U7gAXb8Y6ZYI=ukM9X+tQx(Kle_rRxC)vL-)oB(=$T{>i4_4
zAzl<zOmDz|=A_7V>9!$_1qg~a<g20Rw0DT_d3u2_YSLAZ%iyttIQG?;U*OWvL2%2N
z1l(I)?7WbI9#Q0Ki;*!|X4I^-w!Akf&!;fk!BwfSQqzNm@e=j4zvog+D0)9638JC7
zFS9u+FQC*8>AhrL=0ATa%fkTY0hoU|FecYM%pfl@%35qQ%S3!!^?e>#Et2ONHy$Ee
zex^`LF(@B_lRkv(1=}Ujc8sX*!hk9g0&*9`K^F(rdaa*Ajk`?ktJa6nx7{^W@p`lb
zvk;cpeIkqgb21(o6vgKdQ5H-HEP?P?$>X4%(s>2OT?|}7oL6zj1opT5Ih(>otQ6V#
z5BJp|#8b=K>v7iKNX>fah0Da*qa<G-4JpJ1h6VH=mh&V5A)L1#`x3cjM^46isen*K
zsy7I5@4sEhT`8ztH=+T${A;4%4!S!GR3?Ptnc)2y)~z#4t48X_J)$S)gXDmJ(sDfq
zjZn$JE(9=;rn9Xy;_%b!b=<2=#q@exLbQj6$9?&+mfN1NEue}uGeo<s^i*hd7XZ-O
z5<*Wg&tDZl=^IOp>3O~voqBqyMO8#_1~$34(Wz}-L@yueQ<C=AUglh1w14Zfe9Cr*
z7%v|D&>0`m$aAnwZj!dscMC<5R)2gRtc<rjDRtqh^(a}buh^LWSLbo*<2C;L(^}zR
zc!m<?3%}c#bomVDz&qqm_56PtGV83k`UzPSJpFrPqWeaI$|Zm($SzUCPshJw%<dy#
zJw?rd8~&)(ksNs(z;>qf($t+P-)~MYj~3}GvhO4tPSOlJ7xMf;HsGoo_$Bc?Has`q
ze+rku#W7=~Sc|zn*ENeM_hG?5<hB(ObBP*Tg~>=eKLqBF8TGXam92F<7U-;L{1+ij
za>%3EpSYrGlz@_Vi}~1=vNU;eSZ1N!+~8--0hLC@o<Lo~=uUfVXQw84W(f^Ij1E_y
z_PbKh><iL{Q4;XuT-*aj6*UD9#7EoRzFw79UK~szNM^8Hn9s6T@)o<JZ?2&Q1W<%}
zZMU(gk5tej14^M^#pAUL)4ixcWR$fiokTQY8T>pC+M}9kJLz)K7gvOlhenJs=ZL8G
zp3zp-k{m@MA;3>R(Zv4L8$XGQ&XcrqG7(`~@z<I+Ah=_HT=`J=hcCW-oT3?pp>s)_
zbxQSq16_d|X%4qkSKVgstmLD&(1wn@shs9f_j%h}Q`1T&AFA`1>y2so`(JRRI%}#H
zax)S1H<aWspH|#|$a(>5;$x#xHvdqx-}(Y8)0ty*$W-r7c*qer{rP3FTmFVNfP9_#
z=MuEUJt9Xm@$9eO=&RHdMNkia@3)$$ujabix?GjGiQ|Xv{2lahdE`lgN0{Qc=nmN+
zSAUU83&8Hs608#i@<meiY2gmU4_|cjM0KSTAaK+)HopDy+^Uq_bTTpBq(h&a!vs*#
zn%%er^0B#tqQ#)V0FT{HNADl3Ahcnzn6_n`(b9D4!1DcT3)65iOg;{B%9*lKY%b*%
zmGPn=faLG*h;hAe*<+wNJ=}x=2{%cjC8?shr6dPw=tD`|8b*E+F<1F(+(MnK(_W}G
zX^jDEkAUXVLqJo#8S1@n!#ZIj*)59!2RDyc-H~aOvP($+icP!EAcIZTM))+1?$WZ+
ze@bh!@J>rq&Biora;U-;%0C48lJ0q%pDWw9z4J~D(;^04JR*RN1G0<&Nr+c|$2TN)
zCURavM6?jA6Iw<LbJ)Ou3G-D`S?14%REOGw42;Zl?3n70&5OrSVT?gnuh~D^XXvx<
zckld$Q%xC>dN)37dZ+VAxwK8m_SSYjX`ytuUcZ(@<|+LSB=2QjYe4zi3foFRP6~9=
zf{oRc0#HJz+UK>NTZN&rPY_L-^~1H=4wbTnmSm@FVuu2%gZR3#a@E*M>e`%iKwAXF
zf91P6>w(}k!f}b5;aw)k&IikH&Q(rUbSsizsa2XUMyp~@qXmJB8GecAMFGmx;AY;%
zbb8wTJ;PZmA3vx#0(L!fyyPSpJB8enY?><j@S2nU6n0B8B<Rz)x_YIi3#q&$OW9ZA
z7}*n@-1NtEd<{;%+F^fHZii3R5_i3>gqb}`+?tr1=q!2S%o*vFF5&F0&*0wa^cyJ}
zV8*e%{9(La#F10BJHYg8$mk)wIU=-+^H9MUr86vCS3bJSbe%!Dgc#s?6Mk9q8j9sm
zRq=e5&97Cs|3z++e2DqeB)!LY^xe2iz>>C=fvob2v_M7Yd`ASw&im}mlrGWyduj<|
zQ)}(4)bGdYv~$~oVHVM87_TFKaG+Cywm>}40u@|K7Y!(1^K^1EP}?sEDYN4dhpPt8
zObK4BZkak*w=f4mz$VMpIdV;EvfA{p!>y@;ra#PjVE1%X|9cU83y$)VkzB%}t$}jr
z#*h)wbW>jy#Y{xiP~XS}>(u{w?)V4Ed5SYU&^}8w3*}SrGc2mK_56<}yDNT=EYs)=
zbJJE!pUEnh76R{AxB7`6mOJ*}oxMQ#<qt*VQ=yd_8wnfY$^?C7L&h26Vvh+VRRqMK
zK9miQ_8yzmF&=70DtU{XgWiQw)>a-9AH+F$!N;Y%(U1_Cczf0PHP!{H0}ZeHHn2h*
zSZJm57E2b8V(<E=k2l$W2mB-xQx_-WRc{HSs$1R|P|Of$m16MGYfLu9bsfL6h$F=q
zEDgQ*myFw_%2$Ys`bDIC)78|r|8?+EC`3}=p<C>y?HMh{*{XN%Ix7{m{%i44qk-QC
z&_7mPXr-Wxudb2_?*F?684{@MA?<iG6IQOzvHzG4Jz=>`Jqb^GGi!7sPG}%IG%ltA
zcD^AYx_J|bvn#?Ze1`<oXJgaZT()tEfciv2_voo@;0Ou{h*97Ci)!IG)Yz-mX9B)g
zo$e*k!VL%sJQtibeRcsq=xH+Fu_N*~U7Qo|Str2fp}!hh3p`W9eSj%k&l;NTSJD(N
z)8@+h;I<onX5Gi%vSp1A1L*W%0P!ZE{Im=LJJjL6JNRlbT+Kn4`NeTghU7(I2qC{e
z`RIOv^qak)E(a}5#5{(#RGY2UaRz%P>ABlQL);<-yidB=h1SzHMs4%`g8zckH*v=o
zA|Hx58cHog!=P9^)Q+FEZ}abL<9>#qO_en0v(Xb*N#Px8vPcnr;g8uR``0Fb5M>9r
z)!%5ub{^-*HwfuYUS-NTMS)ZDK9Q#EQwoa7nO6m853zp~6-VM1iIcL@(uO*UW#kle
z0?7%QfsEh|{9hixqjDG&{D6AVaLy2M!s+J0@e+MEZ~)6HTKX}SztC!?0Hey>!IkAi
z37HLHuJ!V`GuGT1<d>29$`C(tMlsx)8wnr*Bq4fa=a%87CjhkTu%6BNW5baN4=}rb
ze*fsJRyLoF_QRP%?Mv<)T)>6_baskF>DWKrz;=cbtX#R#!5kdV(cl2OjVb-v>avIZ
zBr@u1;&NX#Qi(v<{!PsJp1<o(9M@sa7PoA=@Qz$9qWVYP_+C*W3HekB8X<60cdw)M
z@I>60C&vtuW<NOMNfr0G#vOzEvg6jLHDY+3f)SGOU*hId&IlzXlyMfI$sEaREG*&7
zB}b8hdheSNfH){<l6`e~WWHm%zryk==RK$U?aC`3uEo*uApA3uGPQIuW38|i_F;{s
z1=JyZ{U5!B<&CPaog|XYw<sLj!+Ol!kKuPd$aIv+9QHxWeUyDe!2HQ1{~QN-fwtQ9
zT!j}RVcCDT^Enw!xLBH}k<=Tg&UYGBeHs1sCU6Rc?S*IP3E@ARu}|taBu=`Q@hy`2
zil8RdoDdAK($Kb3Mz!M+8@mL{v_yYkywLLAQEE`KJNV|RPV)!!K*ld3m#5pS7P-~c
z|7VFIEXK={aZn{#yUMabvLAar2h7MrGvY>lK${qyTaN|JzCPd>rdA|Xvw=)OokDPW
z{CNc~<jglr=YNxX4}O*G>Tsjs8o2)x56s_lAJOXHfy(}{edW;F_af(|k&GHT9Arge
zjNC<+y7TngPpo_LIUAM(?`MF7GoZ!^VSHrW`WSlm67oJG*)oEc@X-zln{basQeXOc
zyKI?KKxsYE{${Xapp92xCREn`OLO*-17_=_4T$e}sEy}pPw(I@&NJE7DvZy&4^I0K
z+;V~LGn0%eN!0<Y*?-|4V$>BPs#Nc=YS`o58yV<AuU*>OqCH5z_rdVTjN(381O3r>
zIhS=u9unuTy(FM~e!l>}Bk~nUJ*6S|j>0D+Wp;a%UeuK+_{~s7$T`(8UhvJ(H`@2*
zA1oMfjB0MDEo7G=-m!(pn0)g?|2ZeDGI)G(!ymV}fnFXgKYWyv6m{+cFgZTy`!Sz&
z$^iyc_(3M`?P_}iLUODT=A@fw26(`WYWG8IRo54bw!pT^M#Rs2gfIU=o#e>>zm%>8
zwsEpaKh)`usqpo+bX$T<5_MiM`U_XpZClR5KwAi6>g(sEY7mAweVp8<y!|}w<Zo;b
zWEg>nW>5UqK1wuGGE{TaFA;frsd&)Z2Y-Z0o$&W)H5v={gyJGP7^9_7Y-r}kQIJbJ
z`%7_Zm@3!5Vj94oHz1}Th%9`Uv``D*$a*%!z7mUFei}uoM90e~z9(DGiAHNGgNMY0
zmq=coELcd-s%RdETyw0I(ZyAhwqHG<srhR*cj-;}y-zwf!@Q#@&vRytERpwf6OW=~
zbFXmAOz}nnfwQ@lQ4InzS>U)sU_P+kHsGVO<_HeBZt3c&Kz%JHTev%nV*!j}mV?JM
zYbnAbh2Pn8Nnil0WWCKfvazO>z$|Mxz>>A<gelcH!l~INb=m5vfE64-?=H{nyov_Z
zBzrx>MTHx9k<k8sn{|94Lhe>X2_*~V+LM?jCRzSP+c*M1k04w@U9l?Q@A^wivGd3c
zBFKob?a17Y;sqiZAcUM32s(|hsj~qX1e4P~&}u^sVt9gD#Ii8Bc!);MBJ}*AK{!jC
zw-~0?T0=<jK_q~$Nzi5y8(PqjeQKuaXD5Q+OJYY~cQDpVzO@K*np}KOx9icVBeD@&
zNDe7}Yg$HFKZFh?Ia2e{)w}UfwxX?Sj!~!n3`(A<wXfxbKOmypsbfG@_DBn~Ik%o5
zc*|a1|EXpuY_w^t8_Nb)r~>0Pu{j-b;nv2~_^wR!ke@LEDX3+qi1yr2_bRD4=X3Xb
zv6p`4k*A}0^>e6xY`@U1;iGtZ3Im9bDn#!&GLCrF_nJgv2K*09+`{=_&Wlmi!;TXh
z8kwv@@qv5-`&2W|Ih%f@mS#wQQIfk6g|RdH3e=uE-Dh8LV4TQRdeiMUjQ@y~Zi@XT
zOg0WXaINCnA(%&;855%AuhZe6xM^$ys{N?P%n^Jja=-!1X$jFWGf^S9f(wMDBt;rk
zSk(47_1o21gM#J|-JqsVm;=20<>s+h(|1P#9a__zm-8?&C)?kPVO$g@)gy_8Op*P%
z${Jk+jO<jYl;-K%72rjzhdqM=o8vN%=_N*@FQRAPZzf)}OzW8`HO|@e&0C=Ck1|IK
zlD|sfBw&U`V<Jzf|E%5i#VNm5CKKV=SHG21Bh!$Mt0=SU<s3OjV+vC!%<+WNVtlqX
zXKHKPxD#3NHTBsHRqt7Z2(LU|1vq>};@hj*-W6~iX}EGn2sxhv=i-*OPbuiYEtB1J
z{2rbr?jV6BN52-aH#<9T0Z1sIJR>?^nPiRgs3u*kxn8F&Bjf5j5yk~HFvG^a)W??_
z>!<_X|F0PPYrS4!c{fk(xRHD#36S66UO~Px0^^SnC+rwr*l9h5fCq4Ud%#U%Pyc=+
z=lv-|cNn%nP}<53_(y$HJf=Ir1oGWoVJeTlGC4ZjxVDB5j1+UhUGWA?w99olRJ_d|
zpgY%r%$*~L&(<|&5lg~G1qE_m4fEr?HkdJUSZK^gaKQV;Dv+O<T=~|YLM^guSdAD;
zYDeXhfquU9f~G7{Up}0}741od-WepG2F$)h>y{cUY<SQ%(##<vc^8gFlr|abLnX{4
zdbRhS?3}75&@qNc(HM$xn5e74rO>I{-b^;{AyhWdJCBmBLv2A7mQPOOBpn4`aLp}i
z`A8>)nk_g$zT0Zu`HXZ1&h4c{myFt#74GXl);1;4W&V21oXWr8R|Jd*EvbtUR#xQV
zCfNZXrCk!=kT)`O*i1`}NQCYY#*UaA42XtZ)bRDE+#Ie4xo%tr>$#3&LJw*Q_xs5`
z=+Ub|roA0$cmnz*%=)=<<UzpRJd<J<%8G8Ctos}{D%0v#U{o`MvM}mWC|6MlC_PY5
zU9^{Pqp%OS){8H5CQXayd?*km4LlcG(2~u%MxhFz+#WdXv#9%Y3_o+*r%Ye#hEsKf
zVL@*R!x6F1*y@aa`XG*QL-S|8T{lr=&=h~e+n!Pn|5dAW_j8JWqEn>Xd|v45oKE&W
z-HgMvv!Xt4{?NP1$Rd1#__GjWkg;_k)b(&?l-TbjeutYXx_f|u(}WBQ!b8wO>_Fe9
zvcsC7ckk%kdPLji$zn&>ZemZL`j(F_7zQW!!mVl>@4fiY-njCd`Ln_L6K`?Rx9rPN
z1=oHvkCGzIv(o=z8_Ia=?Uf7kW5x;|XzMQm3_IW90VNN;^YTh05-wzl2_rE@ACPMc
zx)5OOj#~!1!f2(Y0`-xWlI_16wQx+h1319}85@bj3&MF7ctA@L$1N+G1GrZItlsuT
zo3!!|fKUARtBwl;C~eZAy~yxF0o1vh&^6O25bRH}y46%fna!_OwGEmi)2m90<m>N>
z;kN%oFSl9SCS@Yv@KMLkH`(Y1?ie@NSQ#|QFh60zN%okLN>Hzb#UEudAXh8(g2A{t
zigYZ}8U(dprnt^4t;tB?;<MRPxfETNnMFSCuym0mbu1{Hhj1V(=QE!ew-}jN+l&(q
znRGjSNYwCYf@RL2&>pzBRFabTE_RQGrNwu6IO@5sYg!M2VnMSxTrSEhzCk__J<`1W
zBWwC~zZe@1U$=Zy-W6DWtNN+M@-8inl;e!9P*2YmDLu9_Qhoja;mJI;Kg#SZZMWS=
z4G7u)Ok$9a%eUbL6TKS)Yov-p<;57)vCsJby#TzqXcoUG#hO4q>*3j1VyA~HGxG~R
z8UW5tSjQe??Px&R1P12b3zTTjqeo_$udQ#nbU2Z0<P>KTET6*4sF49;IE`=j|53qX
zi9W19dy`;zZ|X4BU|?5Gh_v{ybX%H5$#OkZ9;Pr-T%UMKDM+vk*MfcIZsdKQe}cUo
z$`NY}lTmm{pFRPoup3l%#=!F=+RNSvqL}aM28ElOJIJ=#S4tVh4hA>P58W)7Oz!!J
zZ_gN;#c8EDu+SQi(6yFoK`<+9{&_Yk5)P?Y+I2h0D>(8liQzHrh|lo^V+6_(|1<~@
zW?WWlwiMXpEGv<@sYb4Jz^DiLD%a*yLoeKdm>S%wEMfvz3$ntiK&imtCl>GJHRc3$
zmb_%nq3z-?iKlrI2Hh&bk0X#fb~5LDf`LvF26Dp#u?UIjEfTg5#M3WHs5L~$&0Y+}
zdMeI9dfcuzy{s&duArI%ZbA7O&6))Z#Ot4Aj$g^q01HXJeEI$bVE)UyC`BJ1J|ILN
zxs7&xO=!v*4Og}Bfqs=vr~jqb{)|p2Jk2~MteIv1%k!FRuXs++&SmE>(_@v=;;WgF
z$167o9{T4bnnzi5Ev5M;ZcdVoio;*8CScF(FtKXII9xF_f-(N~$`WsBzWJukQN6F+
z?(@BRA&WaUkku?Hggz+hOlJw*Cx_h>{yo2`N}2iQmeAuE@`GMDhH$S6C#{!crrSWr
z%wfXe<IgyB|1tl8L<O~`7#uaZCt(XnuIhnW^~A$}ipEPc<<f(xCN7QOzQAQm4ZB#L
z*3>#J3FTF3i=CQPYn*XUN8oN_YtRm-Z$+OIEp19U*SOM1*x)>-{oDe9isxjY4V}(b
zoxlJO5a;>O|Dw$l+1YWeY%qA%>L2;0m?=C9M$EX*!T@IEtu?$F%-h?F2G&1oJI)ci
zS)9EA3jY}Cwha~}Ap8F@^$zTL^-r|$j*W(mZQHi(q_LW$Nn_i#t;V*|*tVTER^#`6
z{^xqnb-u#dv)8Pd-^`lm^uJJ-&4-VOfoUx2@14vkVdV^<7~Run`(KPYD2T<wek5C6
z>rB3HNhDxDTRb^OT^QviX=I6>$f@h^oXL^Ye9#cn397D^M%M38x{<#M)CBnkLn2A@
z+4YThqwL*~Q~RV!imJ-+Q}c3UYvjQ|%b{$oSauG6b1n6B!7|C%;#o@8HfhQguQ;n$
z<ZTuw5VVIfsLx!V=%tN>4qbiGl}TtM0CoBN-TB7I&9A6F)5r?R18Z;6GdJ%nA2fcG
zx<^8~Xsa+r8ia#=<+)%`3Pw)+dqhXyyky#;TsF_11D|UNNgUe75w<7GZ(!D-h5O`{
zYhXu+R(4X?UF#Kzk_|GfkTnb0@lok0od1}WL;~7}puTactna+TXFf%ZEwe%V#UiA}
zApkVNoJ@6e!TqbB?I$tb^SH8NOIaW6=}Q0yoVhrH1)%S51l%`?ggxiUnfRN7>PoZh
zWzL5TT04-Ab?KH9GnDxS_#ghBN-t<k(T@L6PyZ3wCH+|Q%djoMq9V!@BT-A8Kdq7?
z+WCc~j|%Uo4PhyvldPRAqD(}MXoC7dK|Mt;|FWG>B5aI$FI8+n>ddd=3(Bb0(}FXv
z-j;31pm~bl(*8)*$<(k|RE9;7Q3(r~O685(YR|AjENnrkOoNx0;->rD(U8RDW46#s
zQ>ZGfx+MSNq|t$8ocv+;D`@zg_pn!)D<7`j(OR^Nj*IF_`wJ6G-BAYDz+>lex`(#X
zQw*4yvr^xXFQ+u3<5*4W>5?yW7KB4wZbP4|n*ZETi;mqSDvuDb>kL|g4WEg6Q5Qx6
z6pmiFYN`KT?L=P|T(Bl*6704rl_V=W1KLdXjp<>DsQ~jRxh4afTg*<J!!p>es|vk4
zi8Zj>agfdiC9%DlCsNo7*QD8dNQ6%dYPbudmFoQOe{WepLT*mP;`Zp7ws0FVAknk2
zcz!$)i9n*JF;wI*c9F;)e2?jAU+f=xn>iuPAW#=MP5mO;6%gMGw@muy8Vbc^OBoEL
zed1Ps6U7Wz`RU?3-s6XkXCbG#_@g?N>dj>B1W{^@7Xmt^tt-C?X&tH%JU7hWQ2wk5
z|63<-PL+j;AMJl0ar7~fq9~rs^QUlb39nGx{Oi$leHQem&4R0c?%+UHZ**zWsft?7
zxk`7^$v)&_Wet0kdPAn<hV@D3kF~6ohw7L~=bRls-u@YBlK`?M|6WpkKtL{X3&VGv
z_#=KWe~9m_7aOc-qgQHR9KhShIMKH+S)||Mj`acv>zQZw#v7N5Ao@ne4tXFz3E#oT
zp^0SD43{fI!IA(_?8&fQFM<YodhES6zHB40O(nD?c_n^C41odhvbteCiHcx=%14aV
zkLPTD5kV(E!^$=~vn%%w?%6CD3eM6^85)h_E=k@t<Q>T4=e8*^+U#x8M4iMDk?WR?
z5bcONFB)~YZIPNo60=}hz54PF_4+|zzUusv2Qh^swkcCzb+Son_nRw`X2#))ia{<V
zcH~aH5uV4zQ(%2tN2c~ah};$7_5A@-L#gQ#zpv@LQ%GpkLlzNb=a80xN3sCo`uaDW
z1Wdkh;wX2mEpc_a=>RWAzX^oj^6=B~XS)ct`VN?lE>OE-Y=Ps6ii{gT>$1@3!WPQo
zx^<P1LNxNGQ)kDTpc;GaUr^!dT`iq(HG!Tu@e3T}WRbF$ms(8=n!0AYdS^{<N>!m`
zg$BI8)NVWy0mAc=;k?0se>9q)Ov{tSJDHyyXO0U13t~*ma7e*og)-L)J6<tfHV*-k
z5|9dw=d~jGpJF<Eh6&uQ%c%&mI43c&5i)FCH_Py1&giCDqKu+N{%D@>Z+;y=3%Uot
z1^eziDK$ngc18Hw$Hbh97~75<xVSXRBLGekR-x&tJ62O5?EtYwZ6FX<(6GHXDGkH_
z{^gIy1dPa3X*2BG2iYytN5Z_m`E~CTAHvEaDrTuyRYDwMhOaS+N<Mnn@f9Ce_sLI1
zIQSDC(h6Rdu#I4Gzx&yx2S&7vx&uUx(QUg)$13t)b5$ZpPT?r^o~9y}y>uVUDCp~%
zYY#dCF%0JXjbBmUv)RaJ3NP8}QiLxZpg29fYWz_*(kfW9Uwa|^h4_Z1V?IsgK*$PX
zQhh)FlPHnAkYxvy+15f&s67}A3_aAUq3&)te0PsT2B`5z_>&hQ`pToqzyu%xa8*iM
zulu;amA`Z303)29X}CfMTMwNphY7jx<$Bi4v{-u(YcjxTfnu};a9Cz^89B5MitiFq
z|8P}CY6@@ft!>04*<nQo@BArJ3MEkxvdG_`JNoH|!_9ViO0%Td!qQI0Q{1(H?BJ~2
zSPdpPuXkM&*AxB=OGu*pko0?Z0#MO-!mQ58Gle)GA#h#y(1l517_k!wJQ=qiuZi){
z?3@0$B-K}A74?=o_%{{);sGUIFr96)rX8sEvd+ybrAp@%ye!<j;!a*m{S(W;;h`Iz
zo{}QY-(4$qyVs_qeXKgU*Ie*S*8W=m>{3(H&xIgqc%m35H$MkgVNUSJ;1i%EC%g)|
zzOnh(LFfxT){q15arf>9)WHvUS`LE7NbIS<#2JVSRTgt?0Ll|TglI!0z8>9asf~2l
zFUAu>0FThHOJnp8KwXv-sUT2g`wv6t)GF+GU9r5$RPxiF?-wA<nD6Ug=1@k$7jNSf
zPOAnF%I)+;uOyT6Sitmp$z}SRU4Hf=icETKMC`s-s$hk4#@V<ns}T!}g{5SOW-`lY
zW%iqk(0JDVU7z?2Tx_RuuAhOU)a@^m+pFLC8tQo!#6wCL*(1ulHKr_!DFa05^g;K9
zo4F1R8S5F)jvLIk{OE)Y-|QH2F4+j-Rz<Y@D&Eq&)lc$enDm0izG}c6)#@|Jvt=Rv
zzKa-uXXM1ddj8v+X)R4{V;QYH9zGAP&eQwFw*zcIe6wdEm>5japd4urBBdZPsg1g5
z?G{bu-j<cke=U>Xo+ypgx(ib2q)8SvET=kcFey=}`v=f?R@&Njy)|_mVf*?mMI1iY
z0O01RHq|gsVQV|#y`Tqqy{$ce5!E4!u!ay8+v0?skm9XF<uD|oK5y9ea|Zf$D|rNx
zMvjJ6k${~pVRZlj_#R15bt<wk`<!$R-e}RAtp*TdIXzp|tg!t4M)6sZcbD?7GgKEJ
z0C0-!Pw74hfsZ7)t@mMR$h96pqpZJO<1P?B8fUBsiGS&0qRxbqU4QaGCAO<47R}7e
zpS&|MR~r`|FfIz;w@R?`{UykXSbTRg4c11E(*gtSxGO0UBdG11t7r_+T-WOBUgTrI
zr`Q$rto34;%M6TA^QFuk|3qzwh=YV`56gtB&pXGzAMw7VzLGm2WKG$VI<=dZVLJRG
zkz72`@y2+N4&xoLnn@yhWS!Znu(>93ie(qP$=o1|lH^_I+<*L<sKP&?GOVIb@`G_g
zL)q~WqBgnFP^_|-H|x?4bw61$*(BHEb)v7V>TX#u_nYBbRAiP0oFS%f&EreN;bReP
z4~AeR;IK|{GMpOpIg*M7xFae^5>DOG+AMvw0~|`*gW5D-*{s%mzh2|8&i@0Ay5gtp
z#M5YEQ8xqEG~=Swdas~&Hk8-zIJ|qx)d<y~r3!ce1Nk^cDwY(&zVKyCn*5Qu;}pcd
zyL|(3`-TvGeop0llzw<tgoY;~bJ1wf$*}pf`wyU~Mq<yxv`9A9t3hm~=uOZ_k$8P4
z6_-0c{KqM?+7k1rld+kAf3qs|PCR#LUTiTgnih?AtweUPJWH!r;fPrlC1%CqTQcO3
z{K)fDKemE=lui_}v00SK55B0-@VA&5H(@ezx>a4^?LJ@jnqJ@^$UyW*LP`t-fjv3?
zG(8M8H2AB-P``(mF?R0gVAzn+CfdR;)<Fd!TVD-qcViSOL{tiCX}s9nEg1I$Ik>jw
z;u^BmGr9B}%j!BV^A)PYJ@pNhB?P7()4~qQC?dY)T6|*Ew`}Lr(!opfdHRSuzDjU-
zVhB|Nb^46KFU{D6pZNhC5;Y~C@$YE=&7(b`x{VBmry<a%u;T%L6_*x*bGOJVcIr7$
zJ6!ul|9~t8%|(S25pm_dgH`|&GX(Gec<YdP%7T4CS}!X&i99s`G=W7eqCgvKM@+>f
z<OS4ygS_<^{s5^FuCxx*+g1_90;;#cr)02lqsvz?TBn>KGbF$HRv(+=&=#uNY>{XG
zkN5y!r~kvjE&iWtYNZDSh?~!Ev`_OcF-&j*|0e#k9q;9j3f-bi#f?=358%U-jYydg
z_$B;|>C#A`CWG4k*3-WX-@UJcGj#Xw$Kfx+c)CaOajQApgDfM?+z4i!e{|*;RM9oF
zli7vAx=dBBnSO0DIx)|@!U$U95{cl$vt;0oGR^;#az#}!S+kjUT6sZWTWl4223szo
zKH@$c`|P(Hdn0KXiFN)f>NsnyJIVOiVU)=BseNF79iiobcnj0sSbxnaNF`8?AuO4u
z1)8uiKm3I}^*kp@`sYHdEoi{}_twql;V0tdZGpODkm<<j;h_Qj@3xD8b8R{j-&i={
zK1eIdm%1DZ_>sb60wIk2UGgrx8h1Z0>?h3_DSxgSFst7Tce4X=Zt$xS2DpZ34ePnB
z1>5ayUnEHCoszr-w=)M6oiWCZlvl2hKqw=m&Cqq;+wEj$yb*Z#C!2sHZ!(ED?uxER
z&nEdy9g1sBga(phJbrO_3*=;2w2c}dQllHE$FSAYmVMBBF6v(;lKkmHH-l&7v)h`i
z!2@y!Ba)GBFg}$%@29frH+SrP+Z9_ZlG~}9&=|)KoPdcu2-=MLa^OcBMqRRuP|4a(
zcH0K8VEo|A%lG?~60F3yteERb{m*rfX#Br;rTW$*tufm9BT8I!m_T7Wikx=_`W&08
z@tJA0800XHp<Yym`!(Kn$U+<bDViyJyKcoX=`y7bQMpjw)Z$dXBM`>w#b1cohPG}$
z{n94>4GFzfaD!On6hT4!t4QgYM-y0n7zcr(LN6h0H<oJ3NZtd@Lrk~<Is^dgZYuvx
zo&jKnbB}`pEYX@^t&q0P_E#&SJ%l40o&Bako2lbpCKvx*iNJciuR#Elk9gQG<Gs-S
zpU0|h@*kXdqX_<}GGNR%r`E?W3bal>U`<%xj#+GOE37Bmv!^$eEB@_qiNe45=<&_6
z1?j3o-SJ%>8Jzw@C8ZCvj0=9LbwjBCs>nq}t9<BV8Pu>g{@p;8mt329vtkYHF}!lC
znD%F4(Ps3#o@+C@%Ehtp5zuV`&N64#W<u+Xj(JPvCK$b$5|_}6)}H3^k}}d5!8%S=
zgJ>o8vnLWu^|g#&K@6Nm%83rwl}QX6?+3^f%6RKL6V@Le63UkG&4`~k&$f>6mDHxq
zQ%=8H4Z1dbDn0}jMFptC1vShgJhFW{({1OWmX~c(W|R#utGk$M&hsjuUiTZ}0UjCI
zPl0dJQ~43}mi-{=h{$9VtCTNFvW4*9QZEtK$=&wyD+a5~xQE4&z={L|?z4WUcjxK?
zMpx}y;6Rh+&kYOn)&!M}IGCF|<a@4cJ>c3sd6(^GCy3haHwb9<X-X}nkarv4pZ)(7
z9fXO8mb!*tDL+%m<Cq_^s_kx<wmQR>;)x-2ZrTs?k^V&1aZ6-bw8DO`0RamQUyVjt
zS0}+W0xMBz2v%^ka~Mk4`cYK7(VsH`RwqZxRoixNb0C<HJwV0m#^XlstZ|M7vX~~0
zM|1CpXY)HaD$ne9VTkd|pfaXrj#+T0T0VM}MnOzG1iwd<_wi`@!tC+LE`04ur=AK#
zAL5i>Fh+DpDs4!S3Dv@w0fkFu97e;I^*xs2IVl>(uszuwHD~s^3c_zZF()_GlZ+?U
z<_}l)>wiux9r1NK@uz!oqNiJ{c@|Rwb=YbeLYAXl|F)gqS9)uL?TU8L`5t4<x{Qsi
zl*W`m0-2{cWD(~W^=kn0Dy~G&LN>L$O7+f|s(u+cn&Jqg|M<C9lm&=7pl=nh0Dn&9
z0K$}_ec|xE>GnZC&seZs?svGmgI*Y!W3L%P!2j!=5lDRC0v}m&TP)8T#8(a+Dqy3Z
zDuA4zah<IGA&i_8%bx+6TIDb=k6PmkGfW)h;mx9lrtjBw6$jdn4#vxbvlOf<?P$qg
zy{GNbkHGsEZ6gfXi6mxSm!*?u{uQazd=(kw#P!O}X<t*)iDR2IACJIDP>{T+lZHNY
zOE{7A^hGoCE}K<=ON4M`gd6^=K6EQJm~I`Pmnx%Q+~Z1uvco!@n`+_n4w4_dRN@Qm
z@5REJ8X8-dDRbN{FNNh-ug*#>6~(Mfyr?2!^DH)SwLSOvV`M6Cd?)3Xp@R7dY)s~!
z&kGwCKOZkn&k}!XCQL2L@zQLUjgxTPY|^lrWjOybTmnV(lr-+Aa|^vT^^Yjl*at&B
zSzs3}q3^d$B=;l+10Eo$P3GI0!S*nkF<BUmh^Tp*CcBXL1p+9e9Fm(cSP<jp1%a;k
z+HDNh_v4u#p_w~1oH9H5)ve+49<)`Bqf0nKm?9q<{ulvbY}m2VkpGvlzjy5Djk>2~
z`cNN}(atKTv{Azfp2+cAd4!@aJ*%Ktq&x4+;U7)$+Q6}Us};+nIIc=X*cwmVMg9G4
z@%uf<{VqPb4f`vRIDK+jzHFWql8x4;)$Wk=%mGQV>jfuVTXMeYp$Bu!*l{BjBO6MB
zyEUB_FBJ%Lxx-?kR&*zQrrlcciw)&;!f9gKvXGI#{@lbw^jklBm3`8t;V6QyxMONX
z1rQD(smdEq<Zyh_b=ov@%E@!i*i~kG1CXESIO=8e_Hw4xwh*5x8FLQ{<eBJXn||KO
zUkE7RPSHC~uITZVF>7(4vEPeAM%`KwTxMEg0zWV?9q3kwXFB|)Xo|Z*ksTSi`Q}gj
z>e;;p4R{REFT@G4Tb{GjrjL?fJ0RT4VWW-kfu7nVl7$s}NT+m`T0BVmMQA|*{dtT4
zS`FFUXB14QmP5KX(+wH}*^gZ>E;P?QEkW-ZI6MMIdEuNj7ypm&+jJ)XCI0Lk*+LT}
zAWy-_7+k4)4Q3!p1g8>WF6vqDm`RF@{Hpg7ump+QL&K2){x&1R$hy9ty$|lV*p|BY
z8)YKhy~kR(#KD_o*(s;0ynW-jsQ#6J-M%^Z<nY%gqzHS^csxdYo%%*D*6iE~vGwKY
z+15Xy1oBVBbAmb>&6gnQo-E2x8d3H05tKX;e_I?PdWuCqO7Q|!JcR-Ln05v`5v;KD
zC2<n!@y8)~uFig*v)Dl!w_m$?=i^z_jc+k~KN^lc$2}^3y<X0}t!;iS%m|<n2<RWt
z^3;8&5jkXs%53>g9$w+Jc*h%Irm6lExg-G6eUiE7Bf_#3Ejt@-5p>+SuAnr@?98-}
znLbp51SD0FLpAfGQT?^P*EOp{Tb;y@1p~4hrR*N%p?AAml00~R=E(v^V21w~Kx2Tz
zD*e2*)My`!<O|sUlx8?U-cp#X!aP?0On3Ob0OBA;7P|WsKP2FD>VNg4u%ME7%r#(M
zHRh*cxafe{iY3xuLWm?i<0voLQ+2!$W-3a-DD)PLy0i@svV<p4l%U}Molwd!MFvlt
zB+>yp`U^j0?o+?j!w84`utQf+F5$0o%~u8aPTt(4ybrwaWErT@5cA&#Yq?ksDMBa8
zGqQVGLF>h_>qt<{sl3WJUrq}xlslZkG}N8`M*IK8$oo>^&P0mZiKHh}X-rSa)l%dc
zB^5L8eAr_)|I4DOvq86*irBNLA}hgrEG#+Yb=ryjO1y?C=NA%Y_qkLz#;6Q}kE~n-
z^mL7*$Ue58Qx>k2%?ID@R$*LGk+j8;OXX9EWh;6cj0-EsGOWQckwJpT-_L{w@Gr7J
zg~d!j#NLGV52HW14(;M5$~W($XA(9Z6GzenxwLD2E5j(z=IggjlUt{EU@Zq-8t&s*
z1C1f(aqX%+S`|=!bzhEtbV33Sg<OC8p;QbJ<q7~m5)*fe|HY0Q_$Sr>q;^1`w-<FF
zL2+kJ!=5@^PfQJ0Q^EjC61=$`Qxx@tO~9YpxgKcF{p9;_d&n1d^|~2fDg3>HSXa{;
zYH%URoE8w?y@ugD=mb=E@R_EadSvp8P_s%#nGBw+HlB*UnVE)-)U3Wg@oTDO);&@@
z%HVr&GO9$ZTnB#8gx0^(jT*wW%u>Eb@{d<km@rc?Lx)-7Iq;-;(=E-Ra)tA;baq3S
z{7I;dUk1n2eaBx~whe$NJDAfg6>?yud#5H^DEtdEqv1Sp9%;*^k@94KOX=wT%~vJ7
zvYR8vK!R?D3Et3%)r`SPa6w>0z1IH0%n1%sb_e}gq^J=ApdzF<m?vAVQpS8}Gy)+m
z0DT5WC3Swm39J_NrUrnFh3VRODq{A3JdFQ5M*0l^rC-T5f_6Fs*J7hEeMwRPV(gx&
z&V>{!kYhe_c{PbCdk>8yvCpoMzzQH7QkQqodlKC>)+6Wz__JT`&uV=L0B`>XbP?t3
zlkq6CE)hF9j7sJ6UC?#pR3Pt8;*1vo;~ROmgv?_g1kYWKndjy8;kkK;i&ozvJ!6Tk
zLv%pj@>g_<o5$!??2msKJtKtIbWE=J1~aK2(9O7^cLY>QU$<m5eR%^!m#V&xI^ZzH
zQJW}Hw5{7Wt5R%EPCApa^O;7$_oo-s2JfiQbCR~Rey=TEMty=VB__waksExMmIbvv
z$lKQkNGXh|SWgeJDKiCxCce1a2+(dvifAVs757>r7~jU_7I{#y+B)TaD}`s;IhPgT
z95+lZ7)c4~tUWo&xlKpZF<l}mK1bK++O5J5IFBOL=HUBd;=t~3lCzz*b$u@MR@esG
zS9<KZO{<<mG}g{T_yOTBFy)im#p&+(08XLpmE~zJWQk-70JV(0cgfLJI6>jMN=0s<
z#);z?s~yD8dqmJ%JyQaTq69%spwL2O8;DtyjaffdIV;?+KqaL=G2<JnJUUBTqu+lu
zeAdc6l27=6fb6-NzF*h~esbcjQ%`=E7okeB#@3HR+_J<}e1?PYC_~I2PwQ%LnW`E5
zLtte^s9MMxvZMw5MS1;CZD@sbDy|n><2(7z2np4_NsQaHix~OuYFM(GmRkCgQC*~@
zwE}5XQA`rQyiGYRcu7`r=I|)>O|fe&_zPaXOPa~5mlx1NqaoBZe4d+;`CbiZny@Gd
zll-BL!|+@V5EDC@4g1aH2)(VFmnk{K{QDIfmDW#EXpFG7rSo89x{;Ii>M?~f%0h&t
zomFdif~PamB6mh$%JO$OqsF@kyZ{~LXXT-;ij9x{<L4{C|AKk{^RwSy1CK7nrKjzc
zc$SP2QSy1>2vT$EsahS`VfHVo<>9ibsu78is<@aj^Y?c7GI#Q_k+Ml%?GfP{-6ySi
zmKQ$Lf+snbLr1@64E(OQ*?O*U9vdHZXFQiTdekIIn|Ahwp!~y<Gp&RWzBXxnwG7#?
zka;kiAWH%V4H}A2pq{w;y08TF+~+|-0q?3)@7dKdK0rS4@;jjz?iBhJl+uj#Ttg?A
zf9^!@^J;lLZA&#Om~d|~6NP?#*dxSscTX?3x?|0ZcwdUeY6Xnos*h3>ot}h->Af=2
z7pkG;gxEP>vmh%v8?tQT^CX;HWwx#OuJ4<gIOQRCsPG*+<!h!}J@@Allj<-o@K{yx
zVuM`0x}QoK-`-QIkDYWBVa9P8EzsFd9r)0nyxDjoQ!q=>9pOs?htGzgZ+CPZWoLU+
z*E!P8TZhdMB}2w31eg@(lgOa>TJ%p+0yyy@x<19UP=^Zw1JBHl$T|rbZmK7kH?=6s
z7>)vbA2w?kDT+upsIshARP_$^&`yLze~4dP)2tsu=h3~sQ|hI^5SN`mb^iae08XvA
z>pclIIX{{Oc`<1L`K;r&n?Ft7sQs-Ys^ERNnnykRBkx)VTCW0x20*&yHK8kCV+}FA
z5ZO6dMbypV&KG1AcpyFlJGl0%gU=Epw#D4Zf2enKwPZHX2m<jjFm|1P%@}?}IAsM1
zxp4t~2((Gw3w<?nOlBnD!bfKM#RSm)26ts)@;i7!mz@c7j`IRvop4M;tKa)ivLAwj
znQh7<quL*s%fUEra_T*;D!Fz%UX0ny4b<46BmSH{+N*~VcPMrId0qt<W}{pLyTOFK
zm~knj-5D#`BiZ|d5jI@y`f0Z_;`47Hh0CQD#uURz>fzg!q*Q&k%%6Q}{pn8j236A5
zQeyOiLQK~w2LVas8==tPp8fff71Ohw+j9H?*uQ*tq>!AWXz&fBJAVn*nvpnk7QaTn
zFJ*KaO!px6dd2divFm+9y1-&9dkt9%rjF$eZb5r1M-!Czlb_dBasEp>vmi%CW(G~*
zk4KJrcFrEo_$5sUJTSgxWjz~ibD}E?xToo02=sT0ejMAd4Ar=R^<BI<q6^p_reLj0
zYLsC8oN{8#0=;svD*4L!nas{B@s1-th6I9U-Dp1!gXdv@L>9orPdJyhfI*9d_T>`v
zSO)pLE|&+2?EKUXs)gqNi*lC7W5`FAJs(@e{ZkF=pr+V0&lQWw>1)O9Eoy9h>~7-P
z(zd%t^VPwZYD%(SOmJ(gl1`L|=3GK5IO@HqD0u!BO@_RPUP!+u5b&RzO7L-)8n3I?
z($6qT<U2$(EDn;okuZPZh#vi58)Oqy{<0w@+cbSJqLPp@z_g{L7?CkR+bN0i!Jr^_
ze@BMqKG$NhPwTkZ14e4tO>X^;PyYCCD=?OW4z=BF7p1_RJh-MmxD>0hL$eNa<<aRd
zA^*b`0fRhyVQDeY=etA^+ip`X_2S$pebKG^iLh_vq0!6+$|gxt!rBVox1lv*x<}`i
z8VLVi16wPuxgZr^XWA$d*}hCKyl&QE@CUr_Pi%NC;qQ}i<&Vgcz4QUy1u{V>KpOWN
zQ|VlH$*l7YYJ1-c0Uki5nW76dYd`@I`C}i_wG=n^vEI1Th&wvL07ZkK!>rK%VE$H!
zm06dA0E9gLFPJwZb?gc@L}KJZS%wqLA!KAP40p^4*Iz2wpM}rSRG68`Z*98aE0&AC
zBp_xyWi}LdBXNC_x*=MJtO}H1g$_XUDwJnv{looT$Hh!;a0r<nv&xwGom&e>xlsw;
zjJGmDs2+5Pq{;t#lpfXQ>%l3trcPPS^>L-UmbfN-cUf*WnlqxIvs&k+R7UcRe>k|R
z{uKR~UsqyMm(MI~K-xbV`8ydt_{^&s@#;;`MLpDX*D2y0mM;kn@7nnH0%V+jz|CPK
zVqp1V&~a6m)ZniZm~P_goV1LQC8DPAEB7krIEM9W6L;ndN>@mH|3pnPmvmrY0p9u=
zS-I)|G-)8V7+QBewz4NbrYH5&oGBa$@@$vU*WoO{Hgj=YFAOvMUiErP)QMhoW~ff=
zwxc^==Pfq$Twqt^=Lr!$W+wIX4x>d=mpko`<(v}?KB%uA+otfcosI{a?%pPjBP($I
zeq|?3Pcy$8mp?wCT|FfEB4-P6nPD9!$J)t$@GY!`R_(G4poL-WX0P$>Wk*ycs1i}?
z*Rw3~EHpK2p2d0k?u1Xu_x`h_bMhpAVn%k3as6S&476W-+vCE|mkrP2mfbW@YU!G;
zEdI{_*R0Q|w5TwdLW6blO}SnJFL;oqnHdqYHQhS;mJfG85Mn)=n?u{k6MOFvwmNx`
z|14P8sE|_bH?60-j+&In-axgn0NrG&%C38ZI709Y?)K@ZY`HIMB0d1$do?axqMMa`
zlko*CntKU<IcDK!UUXq4n}p6CT+<rK3_Ep<9u&9FR>4sdI_>b(#SKS#y3Pd{;m1AR
z^T7S$xys#7qSusgtgF}YnfAO>A|e_5%ChU=iuP6C@8Q0h9B-+}OI@yWc6(EqyBcu=
z;Qo;BMN2Uo9<=MrRmG_p85z!7!am!-8h*^FaW?`@n|wh6K@2-8=j?YkBr#}rjSbBH
zOp^^J=kM*ZolGd=Pl#Fk=uhWL*+j0*6Ic5=1o0*_!OPO-ZwVSwDXzX$JA=M2gJ>m+
z3SBF;CF*=9GvK*wO(&06xZc5PZug9WmZS>@tFKT(CohXi$BKT7ch<PT+wCWQsB<~&
z$TV;u(-=&Cl;b&2>GQXoFvY>}cv)-Q${HMN#Y;nDiAT`_(L$2qr)Fj5{!;v|u3p<j
zRV@)%c$>itJEQplQGJoU3Dve7KJq@uv}=_H$bzRXUo3jVZ}c?OOxEDcST&6&<X+=3
z%&RTLZVy#9U@0m)Fu>HIkz!d;w0-(zP-J}km27{RLG=^=f|jh}Pgi0qaloRK^j;kC
zL7)37qTdA3mpnsqj;rY7rnJQOa@aonSwguL|B%v;TYMM9WZ0~+)HA4MC0*4Dx9ahr
zaj=k*5_r9)^FFjL3`>C{)K^yPZP|cyE29<9(hZ9%(8tIdv0gy7H#)#B-Zt%IFo%U@
zn}zw3pKXoRN9HkE<02+pnQmPY;PjA6z0ei6zLVHi4RIf-U_ciLnaT=D4i$ys{ul2A
z^32ll7KhFXwv6((Fmm8|DiOHJ3*1=c44KaxzE|m|>PhVu0l28`!{Pt9N=#U)5Sc3B
zLK@ZzTkxD@;b}OHd-^Ck_mJ?V3|+cWod2s&)2dulLtx@8?tv_Q86GN2+YX%OK(upa
zG(ofcg{rjECwFw};yMmo_^j_tv|Kj(k-mqYB#(x)(0BohIt1}qY0JuBxsXGEj?R+0
zYD-XIMiE)TG_oA!wariE-t5=dsH)`6Aey9pSHBI8?uQvxhB+mU>VfK^X2k*r*xEx3
z<>c4M)K&a9Gc@~mMz(WHYFLKHvV<vNZK-nNtN?Wxbc<s@GXA%V_TCB!fV41>ch+b{
zujltCr?NIFM<2<%rEU(z2USz|_tRGARX+9u<s6)!nYK$GFMm7mg$s0Kfxb-_G}UdC
zOsH_omrNl>!0aVwfLrHHdND}ov8%qs9zens?d>Wxbf*CmcGNgs7bGNl8;)LBP3w>l
z0n5ig?WVAZ6$G8Eju<O{DcM|G2W4EdLr;nwcH0nGz)ti~d&Knzve8O<50}$lo1BrS
zj7>L1h?<o+>JK3d@9FY|lHvOH7_K<HD`dE&uJww6ZbeK)`BHdEhQgY119-TJ){%&^
zW+%AppV<c{zc`i_=KbxO#Hv^l&$Z|>?9?T4Y6ubSNQ=PatSU0FPDf1HUEz{^dey#E
zl2PcI{!2AL*P+Q!QVWASmx?u5FesWR`H)QO?B%WTAl;G-z5G=;C*@au+l~@pt6FIB
ztoH#Q_pYFp8P$)s*6!$u;GP|Jy-_u^^Bv*rX!(BEJL|bksIl(^C#685=l)Y@=9EXz
z3RUZ;H!%{CZ)_Luk9QGHoU^;{F$N>G+*=jz4|M84n}b%9qyiGqb~w04ILU?}T<EtA
zQZ+_waO%vg-_vBzrO9o<$hPhw{TCYYx|0Fo+ihapn8ZK!wu!k|$t&OF`T=2adS0Us
zYM}1yUy#g{u=(A_b2ZP_769~GE8i*%YK8+2xfMMOdy)TF+Td**(GDI^i#n2NAm>9y
zeP$IDn~z`Y$2T=G7}OwfcxNM4XLynFmWMhsAr6Fwc5DxBRg)<~8jC~zw((ex>o#?t
zoyRJ4w;B>`k*f=%f$G&9vkhv99D$UXmCRmWj_0rBg*$?(`C-vp_oD}#v3?oMf1Z1O
zX#2dBBSG!^Rd3r&#)?xSawyv(=RC1h{L#L4R!m9o{KkX6K2q}9)7G{_5d_!9&aIp9
zj$ZDybKi;-_Z}di*N0JtQ+vQG`(%rN>rEpXUrk{(`qm#pnR1so^H1c_p^MMPe<}K{
z`rOw{NJH}FSB7N}zgLL1)H;VhmshW)k>xBS4dleywA0~?D#$t%T;Mu<U4#fKpmueM
z5rFJ~cNcw{A6)4jvgKi5+_O%_Yh42kyUZ)1i71wH)^D7h+}}WG+++uW`QIy;2;qSN
zdT(pvWIwH?dtyOF5>R8k|1BD>3Kh)gj!cOBTm1rjKm3ylN=6SCr=PAquM`R!S-AV^
z*YQj*&m_Gc=q!VroZEsVAkV`nEu|C|71AAfj|vC(%ybrV@gyNIs)|56CaE_<n6DNG
zU*U&Tuay=#kQ*Ii@Nvgvnv^O-c1s@>^EfKA>LlAr&y3u0q1RrfpI<(UNi_|(n{VI?
zYTy!5_F1`7)Xu-uJLeh{`7PX(78aNXzwvV8>SYWLjczH1HEj^b9nPGMda1$ui0wVn
zMI5`57Sq1po`W5Jo-!SkY&8--SBiA|z$-M*#a-0m+W%|Z7I3XKzN+xC+<JXXo%HL+
zp}oeH+G{zuLw!kv@WGnHALJ$*|Fz8n&C7?)0cigv(+w?#(yq;Vy4E}XGX_kc?xT|<
zlYAqvih&~gl92=`Gx`JlpnxMAU1kxb6`#r@QkW*{^PhZ2fRt+_Dqq%25aA<0ShwLS
zGXm<OkqD<n=|Pkf>T_e&ived$*}R1We9gF=x{3vniHw~k(8uV$C{}tE6C<3Aagm!~
z0r>?54`&2x*GaKQqvqo$rx^<i`5&GkoBYSjOKRhO`K_vARTLJL;P}$M2;vY`+Y8y>
zyH}du*Oe_tebvV2YMsLQOL?GgmteUr5ReljvGfK!VFgOpDUlbT`gL1Zp8d0-TO9kf
z#mZg!oJ)0zqq@&9s^VlepnDPxiVTkxB`Noq)y~Z?ykYvSRA*tRa|(KEbL|nH(LY&^
z52=h7c_)3gAIKdb5@OW&Lj_OBw0tpYHzgk*M-ebBTz;`bRkJ)1Si_bwy?7VK5Ig3e
zK6@aM`BlCct}2-;)Dp<4OgTYVuXFSjU}?9nARyp?y7BmT9~wBOt#{(qAhek|zM&-W
z^#WxPWJy@y{<`0<;-)jNj5nFc0s>1@+X1d)?P(W<pT@L+(b}7N!%9L6cD)qp{PYwI
z6okJU4j&K7GX)qR!-rXfXr;W9I_$~?><%!<^jNPxDg)AYW%liE4z+V}VTa6waX5NO
zW<iS}=%++d6VDx>In#F_{RoP0Qg@xU*>y~C2<+yTisS1()oT-I1r-X#dwiJw1k0~&
zqH2j2&C(dnE8E&3?mw}mbX<s98hI+w?%6vGNcN5ue94jwZX)z9PYqD^W9uS)c|faC
zK@op2@W~O`&b!(v)wdx>^ue!{0Aa7u!6&bn5oH^Kbb0D_D=vm>hku63MUL-cLc1Ym
z3mE6U=Z_A_EKjY*fB!C$c4!;yllnElmpz1c>R|H}O1x=39-$dWrP$;fs0vk5{^o{y
z@RF9rX^P-oc*3^++ABnRf|?Mr;MSMcv5H_Egq$l6-)x7VB2C3c@|2n4MWERR0Co0@
zIQB!^EpgA&t?GdAw$qvN;M2CReFw;)RCP$Vl9i`#hRj?>%KDRRtUxoDYtye3L`67y
zRuqd7Y+WmpCjRlhJlI%>z_)y4f=4!>&n;gYQ?chn;`}M;ztkz<GCunB^HKgk^etn<
zhhBzFOg9ZVr=a}B@rpTp?4!-H2xZRJ7t+imF!*hi6=OwN=3|n`Q-2mJaBtf_6>TBs
zowoQ>Nxf<;%b2UviRr_A8CM$QmZ)#8{~DEG)7BRWL(^lAD%4)CRTtJJcbMm&vJS|K
zdX_a|%1d3pug{?4v~Ejhwu5=7+fFIha3tu4={@t=s~RWaS|lyq0UF3IA7)~>)hA`(
zeT<Q^moq0O!e{c_uGn?aI$Pn=sK4-gV0+j72dba=_b!P)oIAE}CZO#;M8!Y39Sd4S
zFk6F9+7FjSM7**}VM)!SITPhXUh%xI+7)@AdR)gCX-$xP64hCXNt7PrQ-9gPMM&?6
zru)p)5a^Tvb^Kx+zG7d|GFUWhbzTHoMpVK4e`B}0I*G5PvY#Gx>W~23+3gSX@pPRN
zi5{3QqhV2pIwn9}5RauQu-h#<-iUNnTLK9lJnLL)1Q2VyjS_tYjVd}kdscDykh2CU
zeDA~qhF>L~pX&tBfO&y)tszNE&tAkB1((#A*y>f$3AqN5?;wrXb6{TL=vMWKg4U6q
z&?GtPu#Z$I?U?C4;W?Uph77mBvXFxxF^99<b?qC=GE{80xn&Yw)VS%5KfLetu<Dq5
zkl{<=-KLklULkZi3y@G#5dbXK@AEndt*Q71e>@8poLaEe*XD)Y(lK@C)n%JXZ}7J9
zx7XcoOA*2!KWsc=6$i^p6XX;I?^`(sq`$vDvSX9sI@PucvHrN9t`(~aCruE^8dn&t
z_BFz7GpSfy!$+Ssm5N<y|7IY+%<slbgr=^AA4rDLIX=mdt6U?1omigr_V-tsgm2CF
zLmPtf%~~9;7OVCfX17k7LpdMghaM(|!_@fKs#`>fA2vx2OAphH>Hx7?Q&rMC>+iad
z|K@8eg%5kV8wf0-6^c_rMqx_90o4!@T!Co8HT_0yAqYdVOf6=hEh+BQOnC+pD0|U(
zH~NSEEx9Xa5yUz2iSbq+wsDaD>~Sx*t2+M_dfhI%IJLQ%TPPl&%N2I;@WyGbruzL5
z<=B3OgJHFoah9ECGK;Jay42qesWEtkb3#CYuSE4!+3T6-pdM`Cc|fbhko$UIcVSFs
z8U&G7N#<w4WNnAlVJ1rFo-0dDF%A017`L*#M4Y+(j_e{(NC<zCRBY|VcKB;Odys>!
z>sTiBohr1!mGkH<OJ2fu;wf<*?58t^BG<rD3hMgLzlpF5WA66pa#NBnv&zPg1~EGD
z=w|_WJvi2DC<;m65LXPQSXg@8L=ky{8cxjzGK<pEr&-sLgWmKacxZ#lZtEN`mf}bJ
zDpRtZ{c7RX&8Z&_<f>3i)RlKCZ9j8&hZW4PPxvaId?X6M`$*6Vv10E*cf%SQL5vTp
zV57k&1mLuw0@CRha{8HiWE)@?PfvYHWxC!q|0mJ|ESw=U2E+}G*PFGgPY%!#paJx=
zuBkpc2B44BwW)mOm|;3^u&onpw?(=$A-P2gTsTw8$5mbw$<#z7W_5Y`UDZB*_`bVd
zG!s*tG+X{_C?#`WpwucBtM<W26yE%xYhA|yiQB2s%7RUxEDj4?I^tn(j{Gl-g)#98
zgU<da3sGDC{Xz%H>A5>Iuf?TYn44U_?0wxkRFidZj+)Zz2oyNp<|gxsvVbqHf?=|}
zu{kHt4`aibG7(t*`;hjs?lxpO_E%!&F5yM<`Me-=(1rw$nG}&u8J4q(u=YiWZ)C|x
zY~!5J&LpI<7|Q_^_b94<L>`Hyl804UJe0`p@J1z&g_F|Wodg|{jrbX^+{~3`+idm;
zm+X4&k3Y)lUomDP-(i-2hi{!-b!b6PX!1-gJkd@l+3dp0A4t4O?243E!eavCGM^M4
zbXc-$yZ>=LvC|<_oa}roIV9*#z)nCMfgRq1gv2cRHMCEv^-Kl^z>Nb1SJhF$Es$<X
zSb5SnL*wEB((bTT05Sw{7~Xs)t@&^<=9w*J=LXT&m5)r;eUSjfAFK4zwuFDqxQX%$
z4KXTpi{OF?hEjsglm1r;A*6sBVjVF%VHlSCiatYHZ>xFsB&=f^{M-c=emEgAfq~U&
zad)Mfg3^}*+3^a(uj1vNykOQQu+vEQozeBXm+09QN)Dt>PP<<F^#T};+5+4@CYr6L
z2^L~Se>{4g+Thoo3G>KCVo^f{+v=#KAS7jVSDRRSEw*WDJ!3sZS8i_oe#+gmKk8g)
z8b^N_;67GG5k<8RMg@~1DHuVIOr%;q>GS(sP2z=>&z6UwRn0OP4NfKnD+#p2rQ_UY
zZyuSkrp1dGb)i-$61+J2;8=nLH{z^wi8mh0gs(?HwC>JU?d7W9*!SA*$L+P{>+R@J
z2Fp*(98yTxeDZ0zQ@5ta1U4r<W;#6!pOp%iNC4EAHMuh3Xw#}UHK=GVq^+Ho6M1;R
z$szrK9}(W=62%HyiuM6p*9#L7xOe~LJfv8%f&lKzN?k1cO7s#RxZ^-6)_)Vn?TB)R
z)x?V&DE3%{8SdHtj56t$KOI~}QbPX2ad>={WcR=D{~0%q_UCX_N$V>Fy_v^)5;e1C
z<j(85=9$}!TUe5p2uAPZ7wG~x2sPBws54rtOm(=05^T{WTN!~b^!Vgi4|s=8(~ZGF
z0t+fy%$c&+^8MVr_!{+|4h?TZvC&POtPMZJs&PS%r7Z$@x{k#dog%H5&Utqx3X=Op
zIfm+#+H3E;p9e(=l`kfz)bj%1pu<F~dBPY}lB8cJud~9h4|I=A#&62081($JT5buR
z2W#@dQ_SP~WsT=RtOWt{hX4DVWM46)$a=il{!JQpfV<Tz;6CA9Lm_6p93To@%@#Wi
z+r<`QK}p+HRGe&C3q@H0J$y8oP;Yi&zP~5GDYmEs%-_l@UN{M;w<-_lA%R41Pw;&Y
zp&>gRybdj@=RnPJz`yG_m5{fc!&g6(sQ-RyW^Tr_mcho49sqZ?<h{V01htWsqZ_p-
z5g_Efsfk)T!SaRw58-rc$rEPBnTWlcBbd}{w}RDcubd>>3U4HeJ~KKcdNfv=)ssh)
z5Hoga)ItOJSr7hYLPBNO+2-G)7Z=+vcOAVn<s=0!5eW356I=6!2y+^P+em%-lgW8;
zz}Hc-&P8rutq5WT2EW+<W1+KQ8D^n^d+Wx>5!=9c@6C04i(eAPY^AlhC}=9fwTO@(
zeAX$jj#Lf0XWWw!iK?Q9G(?@`=Aat~M^kERzKg;KPVt?y@(9|?ltN=0{3n=?+NtCQ
zcV_7wtEQ*s4_K&Pl-B<wCTJM?X=hUa{&h-RPQuJmyrz>od~|YgU@46a*{I6H4=pAl
zpZM`o6Q=^J=f2*jUfz2M8m5F!YMQBekS({BFgRGsWY6~Qn#(!ItlmHLCQ#e$$6nOS
z6LoYfhS^y{U`^(~#HJDf>GXE9&zkg%<vFXO_3D5FR21v+1vWKnFz5@hK6IfJyt;Ef
ze(Y8?Kb4q)jECb!ISB@Gy&19@*|-PdWL?(h#gHaPYFU>f4<p=ELKp#mXMKA}tR1t*
ziyLo(5?Ek<6&V7Q5xQ;UshKBC0$fL`R&}n9+_NsDT%H+kpqr*LXeKp<G%!_QL;cI(
z*(GWr$?6FX10|8Q&W*eCYg3#$1=u0dbJrsKKVHScQUeYz<PmGP*vz4g$j>E0^Z9P!
zUX9h>3n;#pn|eYQWxP<Zp{t@Uc)?J;e2eAZ^6p)h^~upOOz|G65)?!i^DB}cvqv67
z9PwJ@$xwu0*PWP@H=`9siZoY5wzo5riTmlBuuzdNmrw9wbeI@Qg~Q~41HG1!;R(&m
zp0tnzmSLHW??0)UNvpw+KKOTEzFmCO)Qt@$*l-Pr$3HS7BWzF8oEK0wQH|%cj<C_f
z>hyQU)8u+z?v4J(n8e2R{=ucmcA0N~<mc3+l96EscILG-t>g<}*iY$K{{r`Yu~Lov
zMdyNk^#=5n`e;O`*j#HK3h3!vWJzked4V!E;t8USt46s_TqDS+`UJC`p;n!A+yp)O
zx_NJMIh&0~(D$D>oK@d@X>B*zGN3bq^uk$7s~K-5P4VLmYE%jwcxlFY#6vS@iqF26
z?}=4~1gvetz=)@cwrAX~gXS-Y1QR+P<^G2L#+y~02C6<pnyYyJ#eHMm`CIPE@$Cs&
z2ZfHqq)avL3@V7e`fl34tI$qT(?+tI2$4X)El93`x$BWrv65yoWlTgVSmoS!!KUCr
zK&^ak3!>C2)Rlg3edmZgKx-nwGW}{rq<p5~yu5pxtx{=nVAjtl)%|3!QPfaX>DxUc
zf~Z{V^tdP_{^|+?=*LMEAF7{bGSSF5Z0UqrFNl+1JRg`<`F6}2tausnDo7voHrJ!m
zEsreRGpll4t^)p4n!#=Cerb#xz8{(kyQeia{=KHwSTj|9<-#dMvbehxa&k?xtc5Pl
zVEG%Nb#J-vqZ$qV@5Y9=53UyOm)(@x-gdBqhK8?4D43k4dOk;TmAX1QlxD7S1=B#6
z63f}_N^!HG%dpK^>kgwF6c9Alkf|vs6W4<ZgbSa$^w#HCE@qM#d618{U+}0I<lg2>
zo$@)JzwS0hToirx^cd103)Z+3D6@b|y6~E%h60cWv02Vt)6k%AjU|BiVb?BH`iCOl
zpYn_C*h4i@8+UTt@8z7*CNJCi_S=7T=Y3))V|o6;l#?K?P!$!O9i_qMn75VS3u1g1
zdu7)C@C^3jbap0@V1kK~M&is-jaH1!mlon5UF_(Fg4-OquXM6Not0!oZ!VAH5rCY;
z^!@EEbX<R4&bpJnWN;ctVEf&(wlFo@6!p>&P1iA-`Tl0%)st;d53m2t74#MA#P@}L
zD4#dF@!Kn|+F@wOnr%R|7T?&G=q=0vTK5WFrSM+rNv54DwKnv4Xy1kQrC`}z0}iom
z+_&cr6GtB^gMxo%ysn{#x*|q0{N~F3(R5s8ewbq;L^|vock(QQNC)rSR=&xMzd#cT
zPLvo8@NU6yzDoBV%q)tg2L3Z`%kgXSWO`rTuq+3JcgmttI<{9lqO@Uw;IA%gC!yy=
zEt<2Ayz1rEA3SPc!Y-$WzUAHiOhG{xP^8x88c5uqE|ETBkGa$zjnB0JFkwl}aNdq4
zRbKrcSt-T`N<Wtw;(0}V3K4X&{n#R^|KUNy^&j@D#`&?%H5NG-*7UpgWU5Y}1Fq{6
zjHy4TE!%}URtc~OIp{x6#H7;9cM_w72D(oF?<_z8_PB`PI}hnQ#I6|UK&e`MpTx=%
zTa|a2bEF7UH`O$pVv|fP|BAOz_rnP)X)~t$md>GjxbdX{I$>nNS>=gT0R=&DK^!zb
zgdMu+j_j&}chO?3)x$j!xMZzVO@&val)MsBuukrMEby6>@F|#YGTBchQjDCUq4OJd
zQB`U6SD)m>&5)VtweV0uF9QN8RLGw~#<tBt-*rubC>tpsJJb?Uh06&~UpF)HOp6Rp
z<@As^vegTHOt$($oZPxHZ|L|fY%Rjtdwd=C?%5?_zjT({`&3K)ciRB*YXq}$AfIkY
zQ>xI4JONa`jn0+em5Z}@#&efA59P99{@2QIEMM-Ouk{}<@YRFRz^|SCoPQ(VA;i*Y
z?-gm9n95$W2yn&QCErxMU>t)y9E=2rV=gc_W-;Pk{-)Oj)W$rkwe(`sCnKc|F)Fr+
z@Gd-Aba^yB-F<C0BK0qDipSfd%SG0lpcwqE(v}_D(=uYWn!~xG-q-b&w}aZRhWO5r
zJBwpJe#bM_94si_ta$rFT^-*4$2!~TUWpVRiuD-P43eT*C@%gDSX@=-(_CcVZ$@>d
zaclhJ>Z=eH&p_-mFhq|SCJ-e(Nndyue69FB(B-jr?1a9Pj>&A-8rZSdN~-n1cJeiq
zC%kTAA^kHh#W9Gjugtt8b&=VbU%B&%_$T9?89#KGlmD7W*N3;Dkuu8r#cHh_qXonw
zl$QIZFNHt>7*X=edipe?lb^iCmx3VU!k>B6akKPey-<29mGh+~{kDS{NMpjpwWij!
zrQ;^Uj0r)E`XKV5a;#XUB8CiOfCQADv1tQ_Z{diySDvzbet`j;4mwr#Z}uPb_e}H4
z@-$NrG=A}>E29JxbfrGko0H=&J3iXJenglqFm<Lnf>r<f0g{3gO$>)#Oxf3FyGJNH
zscY^3n)<e}xoQm{&4b$0EeNs##!UMqjKktIqsF6BuAwG>hs1{Fy2l^KGO^E1o`a2_
z!-7cmWoqu@nR?hylflzoc}RziCe!gu7dF0ppC6oB+Db{Z{W+!S{yibpdzq$conC6#
zE3~ljuE0TCO{0nT(~_kJ{02IIG$vs#cj?(MEH#@N@jWflH1^hC)+{I6Q2epN;d%y+
z6?$$hbHjc9F}4r&o&MYTPt_l@1dZ5|+K|PuetU<{M$cy%-G|ebmXPmEyOk(Vh?hz~
z>!H=*G}HN)R**l{z~X)hFl#xcJW0UMKVsvst`<ao>mnL#AbU$t?KJ2IE{k$UU`Q4b
zq1T}0@Pa$_5Zf<4IRX3$9FRx$L|iK>(6w(^CU?-vkHjQqC*+d_4JZPJ>b*&1T>pl`
zqxZimq<GW*b4N5*8OBguYR{m!zw3Wv-S=8yP5)=zgRBP1y;XH!eR!Ci&xTidEG&_i
zO!`2|O?!P$p4WD&6z|hm?2mXRP_S1C3D*T-7Yb3x$OlBTh$0x=sK@<(WW9rbUSZQU
zx?|h6Z8x^luxV_ov2CMC8a7sA+qN1sXlz^Oe)>M|`F`Iy|H9g{=bBk-%}k33=1g@h
zQCH5h1RwXNB}Pa=Bt+{e4Dzy8W_QX0I@1=c@24PU7^C^4gsue<wx%C;c%<yG5odxI
zoJl-=IMeAymXNF^ZRH)SmFMG9dy_(4wyWK6@%J+lw|$DeO4y*YM^uTV5erBJ0;5lo
zu)BhR)XL=(@}hq(r4uWTxpNv!dyZdcrOB8j@BJrEtOmSHyF}7OQ`eo9<*111%Y3*{
z3pu5>=m|wUM<8r_RWU5()jjo(U5tp(atpe+XKVv*gya;iD@|!br4Bp-@3WaYEC<Ui
zHSWltZe-^d=j}fAi*-p)EQ?VW$Ap!Gv;i&$W@K9d<<ub%PuKO8BnjVDtq;`0!R`;j
z?fTHsXf3YiHUKji&hpQTcsJFX$;Mnl*q<SsMKg1Eb>%7YHRR7gJ!84UiZJ@5S6t_5
zTfNhN<9!MUNQr;0_BH8bxj<_V=&qDTbk+ITAc?-iTjyIk*l9t!HhbRC=Jytlx6+gF
zmFFOexu&kUZ#NmNxuKk6>ur+tymxWe_83bk?n2{l8z9tl!o{PxJ78_8n06U+Bm-CS
zX}K6kD9ydEXw%k%m{pOmJs3m?#>jYG0p~!}18Z)JDz_%+M9j-kMo{ikZkCkt2Dv4}
z(z2){|7gkQb23zBn@rkC#EMMO7t3urABPFYf7N)X<W7URpAfse7>e#&rJI4w3x1<&
zHeqpU+28a+cgfZL(hzMJ!L1imF_TU-Mxbje1X&+y%>rW~02Y~cync``LI&%WsQNQJ
zaPKAo$8wXb2zK>Qb?zcBZj^|TQ^TdCRSTjXO7|SbBLC-5QXf<R6->(kOQEbdJz!bB
z5MK#@n&jvT3j{SA8^5M~15LqO&S!x#{o%u2zA#Bx$#hy*&E6|=7>1MAu>q#G8OMQT
zVbtE<FYtDyasmU;h;=dc!unV^7MKYBniC-NP+I-tQ_b4e%5@VQ;IR?De0$`&G9kRf
zGxGh&+XTx9-*Z8IC|x~{X@~SJIK=>^64!dwXx(yeJ;z|*Z_y$KRt~qpbl@qmA9j&N
z65O5ynK?oeC3M=CmAeyxxf?o)jje)&iA{YZxs#oeH!-p_PX!$qq)(G8AHuJHCOi%?
z?nA_EC1OstDJ^;8YM`Wc#&Jjq<Q3QM>s>9IVi6;?;QCxd;whSKr{3(UQB$w@sJgi=
zdHItX_H!Xd7g?aD#N&}(P(XKIa>8r!o<tjX*yzw+z<hB^puk9SS6?}kIRZ#oQF14b
zDlV&zCE1CZi%rnnZ@7>8N#uu!;KDx;{w3OL{yj|SHWNlw-di|^%-x}p4y|{N<A*ee
zIO^Jz5B+cCi}-3-L*Ri~P=6%?2YgR#;M}*G9+_ZBzi7id33^Yd0|Ct?LyD-W6^ue0
zf)}6tVCV)PKwF{h=6_C38Jw7b)(-a;f4$sGlQ*`3V_$ss9e<=4Y=-@p_K_iaetG^A
z!>FQ34w?G=aVr@vJHuJgVCgz*{`Li;A9oF=!F66%y=dg#NLcRwnVAix`@qmm*nkx!
zNDL4s+>;!#3?qvWu^$uy@O{r2lyRagYo+FVAWV)<#6jWimcjDF=um2a{a&^4wA7C{
z+>G*s^>Omj56(bU#?&-4?OW==`QJKJxr_?*j1w!egd2>{MY#ffXsRg32ZjfK-SFm)
zSSYdtRUymPg1?I8=}Bqt3PPOUxlceIx?&}yq4t*zAtu`_hGP9pS0b8w@(qx*3qSEf
z?+~|=dkj%rj*k7-;Gf{<xlqsKyLT~*m-p-QSwpQA>r)e2<Kclm(k*`tyG1(Fc?nsI
z3YbWI=AYcRTbIrM@<!@fST3V5GV4i#kbtMfl9#@z(s|q_I$lKwqO6yv6<7cizg}T{
zR+(tpl73jT7SmA<(ZbaMw9!^PV>BjZ;a3IKjh{QgkUA`9bC}AkTC%49s(f6)ZL)xv
z-Pgg^FfWqRY52E6qz@@4dv8i1k7m{;iQJv(Nks7T^}~71We26tkFfe_F%F&n0Hhxg
zGQZH-QCZlmSi!}a+bJf&S)}5_wX*1uCKW9-+AgDsxTE39x~;>=+~_xS7oR38iTi*T
zRbsfG*2T9s$cST}s~Ikt1K++6oa^qP6~dn!=$@Rul0!QCnImKr`PHX*WIS#LWk1x7
zCn$Z8&<gUvu9Pwbsdr3`AF(;-fcXu}K?iYy>~QInr1G)BVHevmE5y{#d8n_v5@o(b
zzfR~zHsvC=Wq<s_X`lF8i*A@saO=IhX)^v-p!RxOWghR0B6<Fs8Y?<;rb8F@ukU_h
z`yi%>^NLFqVICxJ1#$1~9m$!J0RfS#qgjtkcZ?^gRbW3jk${<AUVPi!A$oE0NQW63
z)*R`fG*j<7{Wpe}<(;tz_-}LdS6>~ca*{?T+@H(FnV30GeYDYl!rK4iWZ=kGSELGr
z%B{vH>MptUvJA)$EiOICQQV`A@j@^0;#x3xgK9EjA@Ia74S5QEowM?Im2?3L4#zP$
ziTRW>%1(WGQQ)VPG|Rh%BV-5O&$fFbC-826h&au)J8n5!cMDG(8xb<-dW)sx+Ii1u
z2d{H~5rIzn)Ja+Uq67tbP-d^HAX!;)6Rm<%8%6}#)p(|YDd{S{ufQ(1W+qe84Glg_
zNk%{viy%VO;&A9z@Io;L$(cEVd9qlJxIT}{i65`YBaOb2DQ`fR(s*Qd!gusIkIIxB
zv}j@5w()V)<RWOouMGJ{j7N{JMTCf<$=}RcOC}W35%$62RxRaCr3{VeuG|ixj6)Bk
z#5qPIfJ85VzNZu+J>d5a73fAlM{K#M_uQPXNghiMG$14Un}ls@KkfY$J<>L<;?lQn
zvrqar&VIJ#-B{!JogQ;>^aE_?mX@VO=pgcEo&Ou@aU?zoYtvbvk$OJr&@|L%cSR)3
zWeUCN_n#*F+gKIGH>zv&g6AS~oL2&hd-tFa@=S<`u#P^)v!X2+HW5!@TtQl10ohgE
zMYeDePP0f3@QlIe)QIG|Y*j0bX|HD-mrB8*0^2e#75*+5H!8n6(CALh1~GvnQ(44s
zMA>yJ^?^2lZrGh37R-eA-X?KXzg)S^B^ZI1XQLKthk$-Qov6?+TfU~vD-L+LCV6mR
z_Zm4h?3<Njp*E}uMbzMB@#J-lsEe|dizfZ=Ns^!7P%-*BMb#~_L)xojx!Is_d%U_1
z*ORZ{YC76JS;ke)+uQAPGeBDQLk$jBhzLLjly&z6i9rQeANQXLdWkYZ0W%gQ_||{Z
zZk+(qh80L5zP!>OSP>G1U;y>j_+Jq_JBmV!Cs+<g?J%;okRj&fWBQi^((%a|-fm8u
z_;pB6w+8ox>mRQ)$<hC5j#z=yOlcugP+juSmLY8}tpBu{^(4LoR(at!ft^ieS98%w
zJ{rB;*5AiaSR^r6H=ywZ5f_~aH;bm4+ADKMZzW!E$ymR$M9AvFsq%GI7MhBoJ?Qj2
z{FDFebeY&EFhAM>ijGe8U{~|Eq7)a)6Ita>=}tR(Q<EA8Ck9Cv0crI13$37iewVr;
z^?~(FKf`4QV(>oU6*t;32eqxWFvEjuH@v^#(Y<F~lFpNp2~ZZ#Zcqmluw!*tkJif1
zWQ=7*x1rEF*67Brb`iE!A%>yl&`5YNuqU@%h--QxTV=z(`i`a=STHQzmo^Dsg=!LE
z?^-mM31d1yXLLJ{GM|!ZRC`ad`dKS)v!1>H69u=0wJKm+DUJn4Vs)T3$w=!^F5GZ8
zpO68{>ok!@SiL(xB)M_W4lZMtK?5@qR%S0kM&nkxaB7^ZiGBO`fjizyRzv&;;Zey5
z_ZG;|TZQ{AcZ@Vfng((}#zX)SXy5;5%?Z1$`Xk(6VCw$oO$Ig;A=qaZAngyxLD%Tm
zjH2v5Yiae6NldLo<7<Nr`(-wE?}v#NJt8X9MV!FV4&F;P6+tqf9TR18W<=g$)QHJ8
z{kV(%A*h*1l#KP5zlG#UwE(ta{<ilmO|=Fha{@zOKkS8Rkb%BR-M4)!f`#hWhY?l{
zQ>6`76GxT-B;-qWcA&pYO7PG4)geO*gL*Wd^5#o=<;1TppD+2I;{1nO5IB%XcRy%@
zW1lUB&FzI;To;%|mPq7H=abzEC7Bb2kW<(XdDwE?Axh-6sae*F?~RvvJ3bKhj*Z~O
zVa3F@S2b}nO6Pq45}Mg(NkEBLB_*~u2a)7G9B=QXf<hZ-u{Ts#UtAs;u<-`)@D$}m
zPOs2`fCb?$g6^U&u!m~+F}OJ(po+#Hr#cFCa)z~kq3$71^RrlTYpWucx2rncQxHU>
zR${dKwE}l!c%91I!S$v^?f)>w|B8|OY8j+)*a2<!wv(lUnwlAmFJq*n;@$SwsGm&t
z*3(}1kP)hbx3`AvDkJI5)55I|`e(9J(MzD$!2~B~(+L^|icx}l3x+GB1?ZtJmcW7Q
zvRs%y<4k4HWsDOl`aP`L3PhH^=WG)-O*fSC&nXAoR~qKX-OZb+Qs%Si$Ua|@Mj@lf
zQnk=rz<}iM!llG1d*8lq;y?XS+A9007e?}iT)tTx)zug1h&j5cGrGV_#2vx+aSAtm
zfAcIlgXa=fhP$?w{4p=p)@E#HQ<N#gy)P7VAf&Ri9PrYCk9~#bkV@Yen@|xBOO$%J
z1rc)r*pr%kD<R+pZ7sLq`NJ1;oP|~@Z9G(HVOx!5Pt=oIE}ml1e#68ETHW)ep8uyS
z#W9!{@QAYQg4}EToZ9xH%dv7Z2OdC$;Sc`@rHm_UR5dr1(Iki(OcldR3XVdoj=VoD
zF+Yt1r=xub$wOUj(Sm|5RhR)hfU{}G6=>i67vq}bwgD}K5fEIIWSG@_LqB^}7Wmo`
zUWUo`$(gTTe%NIGVy=-q{i4m{{Ee)%n!@71sZ>VB3sDWmp~i8GUIc<y0QWDm;h>Xu
ze&I4#xQLs38OzuBoEFXz8pZ?;#L;<*?7P-sw*DwS?yhN<Fx2fe-c#MdBYs34B(f@H
z8&Y}Ep7K2M{km?nQG@X=Lw~P_6(shbeTuq4*$iuD2@79?YNk15xj1`l13FK;5AZTR
zq;EZ9X+m4a@u91EAJg0e20h7JL%fIs8eqRRo7{5h%Se-$2ww2%(@}hR3d?qk)VrAZ
zBt5za5yW;Hj=B~yKZBLA-FaRdpE<;(1b^XzT+oJ%0xc#{rYsH<+y@pAX}nC57e(ZL
z?z&&?<BW}9JqiYhHWD*>{${A{lOqIVkS#r5)AAU7um&*afq-mdmT(RVC~kL^uGo#K
zx6v_l2~n*l&b*t@*L~R)A7=Y~qp-rZYI+{Nxmo_ue;p7Q@H;i$YXl^b2eDC@Sx4sI
z^$b3SEE8G{+^G0FFCB@w=UVJt3a`E{g@`=7E4neojO|5D=y11EDtl_J#a%io;3Ss{
zBcqC9_kFFw=ORQDRvIes+D9GXWJLLP##Eu^PXSFX67#MVAGmbTl{SmcYJKUe&w8ER
zRa>vGSh9yGM<6ou+JXI8NGYtLhIX>^N-j}dHv;!6kPyePCpj@H_`A*@{<R~1;DRXW
z&l<`TfZQ#tViZ?q{`K5|^s&-T(gZZ`OF6CQcrL*4wx$#n*kd*K{X~U~$T#xp6>;Dx
zbos$td*lN;{4SV+iSkBnTZ~>kLT{7cvQKCVWJq=-pb5=?Zn3Ab`);zI!ND2V{aFS;
z?L;$P^DO5~BLuc&Z*SP6(cCd~1c8LGdVW9ow*iHBUXQ#vpr@PK=ZWGFLkuPHz0=bJ
zVpEY?QR1(ERlgm~-CdD-nSp7mUu@2J=VxLkY!{N!sMs+Ffpkz#w-Eq<_Ur$Ncbg~v
zS=N5eEz&UR%gpA_mlMKT<qu_I0ym4u-7^p7Wgi-w<_bS$cMo;H^niUNe@%iiBpd6A
zp1A&zz^}T@s42DzS>(eT@b_oa-@GNB@Sd%vzj7af*g#hT86w1jj#rD&w|X7nz-La{
z-*{x*r0In{zk0h|B_9Hyi6juc3>Jqf24gH!5A6<5UVlErbKvKVEs5~kU$j)<pML$r
z3=z94t|EMqBJ-I^S$Uw0pfhS$``$l%Y^XXX!CjLJYY82(HcjU+#|^sRgQ!RWGqU4e
zj)?iHdJlJcI_q>Z%pE*d<+q}Kk1iC^p`_L0xOW_QU}P|A7Imm^LW36&7&!)=KQtyJ
z2NR_{eL}aQXZ^ty3JL6Cth(jU$5O!nrXwtdbyTw7V?6-UAKapzReszwjBv-Jpv-Qu
z|I>!eJ#8-fB=k?D^eqqCH+fTs6P=grbnCuPtnq2X{0UF3QoT{%lJ^y0C@C-`{}JrS
zK+i9q|1_GDA~)unWMfta*4;a(`|p>T;<Ns+ZECmtc#x9fqQSmZ7mB1__sn{8;y=Tu
z*VGW{)Fe{$!C`(f@kr&*>oO7%$D?w}{20NBFWe0wh6t7@J_y{n@$C_Bn7UWP>+3LE
zrFxStGyh1x<8l5Kmx#7`;2L=W>Be+#SE{Gi4qSls@sHn2?jh@BENToq3kbuCrrzz2
z?dov7dR!-e$2t8v&)nZ|V4{+Ho)osZn4neo(V0m)_<k`=znd;;Yc-i4*0>vaJ8L`h
z2;<=hqnxl&uf3*PMgjTO1*1`lvf)5Z-}i8Sdw|W7$3#Z(T6gPWB*X)3hZ)V#%yn0L
z9EGZVLDaDN+b@5<c&NjD=wE}%v)g``@nOZKB3MF^0Uu%gjwcU*Jb{oVfqo;yOz`UK
zb6^2S^Yw7IOmqy%#QKOd_P+|oC|8t)hx&<mu9SEU?tSn-5sz|7`Vl6^=g-;r5sYJk
z2o{EKTr*;<z7Cb9yZyv#*1kmt-@`e$W0wh|>~M^54`(AAMDRKc8SXmaB&IiZB2q-M
z3n7UOS)=v)O&C*UBYL_g)&86$nnGG(h{lay)*xFkRoN1xzoEgtu~~N8pOgbau&Vi!
z*~A~2YkA|QVdEuU>|)GK8W~H6{Y1N#lzB}^KNzbKocPj&7S+*QteidW>e#AijlBI!
z^DLQa0=##tK`?(d?Yc;dl#=AVKx})H88Iry<46{Ug?RqxhhXM0{)`e7>j~!jku0$c
zLuco!`(-R-r-bi0w~ICed$JD2R}7<j5|d@*OM%D*);6}PG(mvlPU?s&x7zXbWFL$<
z8@F%J;Q1nNfe-0ZNw+2*V!($LATrq2h4P`0k?Y}04)jo9ZDQnv##KWu5*+~|^nDu4
zVjuIZyUwFNjWg2*`V}6}wQQw6$I?%I8V^-_jFW0`x2KM7=Uh`Jp5(gUsuk~u2>(5!
z;ot!pUOF;IW{n`HFq6!Ck4>C))%In5J4yU)VS)R5K^KqkV}BnEVPag2u&AuW__k^n
z+>_^3N3V;2xM?dvf~l$HomEJ~os3KGtQ^^rQAsoHo*u`NaJ;LR<_IcuRW5RARfFhv
zUD?_|h4oayNSw4~il;NyGy63tRKoJ=kB_ELa|?{#<yGWvaW^iGEl68qDfi%KGtc@9
zQ7-~7RlQ_+SA6Is9A-aNi3@~GEQK+gXBy(snj3Q<OY3yCX_S6F7hKM>?9cea>w0n&
zIH4yUR=&Ox1uT%L=iRhS({Srj6xq!S3R|fnHsomS!Kg^0tZd6-BnVjbd4{5*vcc^@
zZbLj+-)_((yn08vB8MUZ^TmArul#3;g#_mbpo9`(q^#6v0}k+x%mPjd6dw@>5E~%L
zJYmYCh)<(EQ9f+^bM@vRyr&*fV|Y#e7l3FB1=3>6DfwxRTTxlqg(_<`5uq*ryi*&z
zP|ouIL!%V@au5A<?@Cwn<tla92Dy9+_Hhk{2m0<0x_84{`GsO0^c7m_5ISb+_n;7A
za3T4;PmGEXU#^v9x)jBdYKfX&Fl*5)k*$W!JK(QJBOE@k8w^-1u>D!YamgjB?x?G5
zXUyo6F-v=UIsf`PbwHR<O()tKu-s?R7-e7J=hb<zh1+7)_D)TU(m${aq~8Ar_#yoT
za1CLCF=NTW?-4ZX49hi~gueX0ZE%apiwZj56Gr{-2Fg-Lbv%BKT=G6fXkCYAGYemQ
z>D*UO5YuUIB3_9Pqzz68TFk>i?w%y_gOS#N$AYMi&JrGfN+3-4VR(#GY;FFgDncJf
z1#Qr#wNoi43e)gi=T&b6KphKZQk+Zo4F^?>mhrp44b;vu0pQDk)?h#CyC-}4auGzp
zgHj;pXDX$5DqN4Qj?~rI($w!AsRcvPvfV0@KbciOsOw+t-~4vvRQFVvw<<_*dKAq6
zZ_Fz#U)jlsSy(ED)$2YkGy4@uj_|gX3mdW50!igJe~8%JV9>`l#tMsCaY{-^98p-}
zN6!fhu0PNcbK=Pk-LGE`!n_YEYzqnr7JQ^?l!AA2+*p4Ia<c?gM;Q_OD%A%HQXe|Q
z+!PrcbNxOrX?uxF)LvYLCGAmJkExk@KoZ3_I~5@0R9ni&7N>Wom>Wt}?Ej|iQ}KK5
zeX45?d;i8@+iNP#Mw<{#TmQF#h&Z}6Y{w+y{FRlBO>#Y})|;$^!|+YGYjHW?KY7%0
z%&7MpR)inM$rmjMk^MFXauT+fagR+=n^HIc>l>|A`x7zYgUWNXnvPpb>y#Y1uQ~s0
zNz@U`o|Cbf)iQX`=c|;8Y?D#2yo}9I+DMRh$0U0<-lVkDKZa*Ehl2_j3907})a;Xj
z2x;clo=&(4qdJ3+3Di$A*61Mh86cnf;at=lCP=RS6zy)1^pCpwKZCFHU|-+pp{O^H
zXA7CyG5>OfSNN30-tB;03J}}PJnXys$2`TznVjt?t6-zQm#)1Lnz2T_Xnn#m8hJSK
zTMwc%k@h$7z3i2)foSH0Uuu%|*T1Mpe+nU(5JLSqrg5C->KHV6`3<LjL8RQqQ>m={
z?(UC$PuH;Q9tKjfrN><bJ{j~7$wAVAmHh!#p)(KJfbJhoh44n1+lp8sQyh@2V6=J@
zuoW|AQ-CSGYQT5EOF$rMX(9Bnc(MlFfuWF}BU$)PeGhXkhf9l)brOV!X(gen+Qo&d
z-#KyzgXi_PUkJ=tLu|0;$2G)AE|m2;Xl&Y>S5$hpk}Y?Kf@~6dqO#b*RbdmfNOQ9B
z64=@)U%ln?sYppA-DvvE$Oq)b%8lt^06meS=;K_DOZ$+iT1TLJq(=}kZfO$g4jJ|O
z#S2EB1?Xc)y|jt)@Tyj2pSZ^x!yWa0eM?3oXR&}7E_{a%{8x{o0l!24ryf81)HLGV
zEhfx&HloyP6E|#@LdUcqo_z3>9ngH}jZv!aCt6gC@mE;|8U6jj<#N7jWMx$~Hyxt<
zFU&w19V|Db;__XvpQE6&#s7K%;{5h&H=d~sTWoPJa);>|49{y}N$%BNszWhsDc8-(
zk2DBX6@7AxavYD}oDHSalvT!9n4~xaJ?@0LhvY@MIvkfoLL~#xR96vo!uj<{iT?~m
ze^F69okmvq+(|f}1H(r)2hsFy{jJ7-C##_P$(ml*xSJ{gwdOC!qw%zOq7i;8p^%`1
zs1)7Oj+^)j!HN6LrSDX~u-Kf=wm(c#pgEL53%}rdv*bkbipvC%y)XKpB@-+^`OU%<
z1|4;(EWnkwRBb*Z@yb>1k4X%KtSYrPJ09pd1HYH!1*mHwftkeyzNqw^6xUol8Qwye
z0Jm#MU#qe#z0mpslE8wSA?Uhww01C*j7PSQzRxvnh0n<Glc{jsm~^boOI%l?fB)5p
zaiGbZe>Gxk!?l=KDn9w9#=<AQ`hav><0Owz#rwZjZ@>N8{qH(Q@4HXbG$(Qi{ewDt
zjYZ^)pezsHZh3uekfrM4X-W98*b)G4R>>s}`%<*EByh|{dvq>#-Y4$5s8$$?JS_F9
zO*T!ukK0c(DYUX7QKb$c+IS-`GY|Zd^!b(^*?9<8H=?*ao;W@kur@eQ-C<?xOz9cR
z&2rN8vja4x)b{o`y%{gqr4I?MEB+)!n8~!=;?y#C{ZiQ;smduYvuu_O5ee(^EsPH}
zsRFbGxz)Rw#DI!Ad&=%ZD?o=jiUj?Juf57GP`dpaCo;(W2_lNkGf1eUBgVKv6OxFI
zF~AflI%fImIiu?a!kWRL{?b(~6dnceBQ8-aa|(=A{(E~~Lj@hMiBpZT>O@r!0-uJA
zEC~E{UT4B)Is+O=;1Gxk<aI+FEQmHHe4{4Hb@!e5!DmbvEywp{Kx?XRnMU!EHBM`5
z^)#doBgpdqLtjzLv({$3w7jdjQk{kxb@l_QHfgr0mfSD<%<~K312w9UZkQ8YVvQ}-
zK%>e<1AP7{SkdiHZgAqMNfTE$QNcpw`w(ijfO>l?J%|T-z7bx&A(U~|WRE8Mm{2!$
z)yQ~`qCS2HxMpg=!ID$aHBfeaNA?fN2i)-&n6?%_<}d^L7NmLYD#4eC5zE&70X!HP
zPv$cT7Hd&1C;4{Rg_4WQ0+Q{$nVZbC=Z42inCwQd-?02i1f;kjty7VKC;?xeDz=UV
zN0hqapnLFmg-{C_haV7B*+{8mD&8(sG=9ZcJiWdYnA*aD<@1_Y!2bAikIf1Hg+qZ_
z4cjo;nYkuMc=)wA@CD8j8??K<s+}E1#WD>Q;9gW2bf7;QPof0HyQBTzIwPc#;Xh-e
z=|UFxKtTIO&SHYv+ALV;--!+PsrC$w-pEsS$ne6%8wX5)eY{s>Qj$hfK*;Fx`~dzk
z+&}ScY7*=eWYsRN@AwV#qxlU+=xf%*Gs!|Hshmfu*~=E!#~31ck&}b%8xyoz%N?9~
zGFJXpj8#AW%nVHGqOu)wRO%<A(7kUlYBWy-CmCWN;u1a+LsMu!M>0lSNTH32TvdpB
z)J^p?+>o*iG4ecM3FVdO@_d?BWp19KRC!Dkm8CbYb(K4wH<1ICCypk2>5Xv&gkg-O
zbiwdoBC)BN_9dA>(Qbz$n)&1n%l?V_8J=FAg>3hi*-WIUKH?jJj!%C1U*YjhfV4Ys
z`3~+L#%+LZYM=u_v@%M;AH{c8qQ{n$#LpY4m7ScT@=WAXbkoAuNtQh#0CC=rxnxMx
z(+{14L5?1Swmw@iGpVfXQ;B|_OH+Ch5vY8`(NFjLLDURdxGwk%hUWxzjs|LrD9}Bj
zOgS4`5^3%K6X<#KuJoOxWwo{taR9>Q)I@uB<eP*W8m_b2Mm@UA6v~9NZ~N4Ak+f;s
znhAAjG1VuTj(?fn2>8wQFVnM0st%`1X;$1hI^p9{ZjFeCT4Y&#`6aGDQ;f~9wPAZV
z+=_LoyTR}dnM%49iFSb^GmW^kGh4vtXeMk15r~Idq;$m-c3esvC#uE<9yY_M<w<TD
z-x9S~nJ*aN)eci`Ec~;-&FPv%{A=cudX{Od!cRpC_CaUuscB!0tP4TBg=ONfhmf*(
zRtFiV%n8=YiC*IH$WIVe;W&*V!M*N~y}o)-s-we`jK>eya<%b_)IVZXNC5E<YNrkB
z?F!v@zVl(-(krhxd)@9W1&iR$cmtK#qYI<z=VJ`S!_?t{;GKtpoD5IP2=TcGacsm?
za_6A$UN1u^#M7cdga@V~Ro$&q$I3&cg&Fbu?*7!vRt`z~&znq;{ZI@iAy??S-~etm
zZ$l2~`uw;{uuCymfa?Acy;(!;ml4j_6GT_DjS1811&@)9E6T|`s3Zyb$?Jz(L5}Zc
z`^D(naGqw@R{yCQ??H`d1_QZIf5LmbSIeOSIqxHM-VXy4$e~5<>6>W(imi>^=bd#q
zC7S6dVNtAw3G$*E;@jODdip!{DRpAPr@eNk-Oq2a`TVOA+L7GhY0<2MlxE>h2mtIS
zGHI}lzs-(8iVJUW4w7tFKiGyorYOtCE};VL2Qw4t`=>YkSi)6w!XYks*-3tp?{kAo
zpxtyoKEO(2^=m|9RBT_^{_w{w`*1<aN!3}%+?Si}o%(6bF+S<Y@@`(Qpn;QLt~)Um
z@Q2Fqq`MEe4A@e5V+9lS%SXN=mtwq!;`>c?gPWBG@8H7JY4Fj<7iE+MHNv^o7nt^K
zSPC(KXg4cl{$|ou)1QrAQ`)WT63Aca+c2g*?M48^mgR<%WdQ&Ky$c~vVgV21+kyh<
z<grE&k;<arV|dTo%>4qA@ZO|sCY2z<bX^CBJgUrBvoxEJ{fV#KaO}R)xYs`a@6$;P
z573N%_RAQ3dn+`!wZ{w%$c16w^N@cSaIu#hv$@)MOo9=v!Sn8Z#88)%rG#FJRUvK>
zK^bJ(HMD&Xf<NJ4lgr+BsI-9h#jik|P^hX@A-1M3N`_mvX80Akh0#jOL}%b$uBF;F
zoSgG;tN>u8PS+Lk_&6DCgug~lj1xaHK5qP7tQH10h{RH60*OOmc;PqfoEFyhfqzK5
ziq!H%7Dk!DiH<B_U#6su8k;npwZGLYlR;DMnWtCVnm;vKN01AS<A9y|1w_v3ntBQ}
zjzVnP7^hS#vz<h5-di<hO(XSNjk0Irm_mK|Y5w?G3Vke4><eK-;Spp2#%>m(ks|l9
zj$L6yg=#ULo8gvPmw3~XKMwGY(;6Ph1O}|*{~~#)3R^v>Bc=O63~a%Jg`fNY9aFaz
zk73I2jE6BJN2Ab^;p{MwY>Zs=3UJh9##)I@pXYzcJQ)AyOUwsErbgfAdGOpFCjI2@
zC__>Gi07&?5SbDjPNvOK%f`KTZWy%cySMT{2ut%pKri_xwBi5M+u?CO-|C$eJN*=+
zCJIZ$EJvY1IQ>gGcVb4rJmi$E{EL$!WbTM0Us_*F$y$RAsU5;zy>L-Nej&?+$-zV}
z%%(2%ov<!x;^c^yJXS43R&H=UE6UYtQ`4VVPsx7lL)MooCYiU7puF?XKU)jC=6m`X
zx^i@2lGuzpR20T}L>DjB>p3j7bJv-=J=4?<pe^`e)`=8OPznuas5VoZ_z<QJ%uTS8
zCzQXqm8jVfr#-w&VlG)ZG9vB83%~(1qkJ|DuCJAL`}`G(w$AO{r-F5ncFK~g=b|5G
z0IN&KRGGUX0C4oJvu)y+qbt2>dI7&ZLKH+$QXeatbBhYC0S{`_=-A(~j?Xx15XU-#
zf9}sK7p63s%k@i-p@P?LZH}u^|Ft|qm_U!vcNMPaAivpnV+Uxjt%Kf&9LvXd+hI%H
zUY^~~TC8fSx)ptTu!2hH$D#<&U9syAa-llLQPHFa1_|+nx0U<{gt~bo4cU<|#OB|z
z=7Uoj>cl!z?D*l%NTN90=Jb(~jplxVxx$ZG5ELaCv22nxLzJ`F2k?LeKN~P5qMFC-
z9?}9aZhli}w(>rht$1C`Ir13j6(<*`1q7Tq{=w-V9MmyQU%MB-)qE5l;;2G)eQ1VN
zaf0XIQ68$q-$AYSvgoW`q!ed|Paw#JqZALVn`M0bu-<TYv-L*e)TYkJP5jBPEpYkX
zZ5U_;rJy+${GC4-a4g4+Txt$R;7WV9kM}1Sohm>^_R415RyYGCiQ)woogFL_IAAM~
zG8UU_{{f{iWe(K8za%bRr8g(iD>!bLwn7Ks^smqjuWrk(&D|ZokGvA%F4S5boP~Vt
zkmT+S%Mu%E;xy=O{*S2XO$#Jo_o#4<5Bo2^8<VI)UCZ#P@{FmgrVel~MGHneNU2pf
z!U!|R1d(0QltSg2(RuTUT^k7d<)Q;=brw5>Ej^-brwBz;mU;JLadjDbnqnFQF<*4w
zV)A;3Z(S+NwmF2O#nT&3qTg$KzO9glA{L$e-hOnepL5Pwn9jjOaNvi30^0WNegdpR
z2&XcC4Q;}pd{TEGmwfKtvfEcARr=l`AY%^@#XsbQlID6hdfrV57q(v<OezRnA+Our
z)3Zf!^r9GEiMRx_Pkg{kcTxGQrve7|GBTR%u(n@?QqXJbjKEwH8ZcDQU$LwLv1WJc
z(2qIKseyrzqK)755w!}290ei8-Mfe4p#o6e@efBJm|b>hXwrYTr2#;W?3_p!(q7F5
z!c+z_5R*;WE2rK$0*(YYQ2J~dRu!5KQ|vUvvE&9-&eXOpE^atHS)A=PRLa}@d+Ahw
zJa6fJ$_g?6<#%IYspsBA>4tsk&zF)`@M|UKtQ!Q|TI^pTnuRv7?yYy8UNR+*l<8e-
ziSfFpL_OYZkS9^>+=gX%_k~U><`U@cQ;X!w>5IQ_zYnJRzO%8Rvj6l8A%XbZSEy%j
z6(9PYM;ab)?D5&&O+?QeCqN^%;rITM%&b~wi8v<k_2)NvF&$m=Uj|JCU-u-L5|S??
zRUHR8$POMmA_x6f)#><u`?C!wL2L8)SuL59gl2;2vWHMk@B}pE%7T+%XIatC#j4y`
zN)EO#KmepIK7j`Pgb?7<VGQ2zf@@~#&ZUTz<#DE4psOwS40$D9o&MR2*h7VgLT+C<
zwv~E6k&5sCCwc?L`Ye?ULD73LHT;@xIgfE<;DesUkqgXxR<Gz#)BLIe(bF!Qc%(94
zgr6y*FX_L8D5q7i4&S?J%CM&kUkXx%S3pl2CW$!`OL&1Wvbr^CIPIgS2b4W`YW^YW
z-iqd>^z{a8chUyG)S3KW<r!k1j6KXN&o2&nvt}B|bs6KMPJWOD>hAs~LZ1&+wNgdn
z=1MkF)fV%~3#*J|rS-cDNA>Sit)YiRF!nqrg_rsfmueF7vq%za>$JO+kRg>4#okJg
zdID-Q-c^B?RfI|`0k8&>^<#gx;42Wt#|M{-7&P92T;+q4DL59sT2vg|nP_Ek%<+?6
z;CrOaBK9nU{H+wC{T@UJs>GZWqy;{7<_O<9vGI}#II#8SE1VwL{@aXDe)$Nwt6vZg
z5Za+}rD0R>ZbS{1ClDPQ7@7QbQu)r8?7s5)n<7L2>LmfHx}6AUS{G<pJ{+bfA9DD*
z(*Ds1_k2xV8v+<+($d82LxQ3>NyokEN6lk~KY&ZA!Q$zs7z7;<k1;!K$*u89+Sg14
zq4AUHnR@?xZ>e!oj&>`3;J<HN53^q7_MT&dX1si`&Qs_C=F@Lo#X**mq^2i(n+<~e
zWx_x#8Ag3#|DXvNzk+ffv*HCQQFVTp(rYB|s!w66<t5#0&&a10V?xI9a12#2zM72X
zQr=}EWQ`41k<Y3tc3Bob1*2b0gs6#jRYz31x`^Vcq|bQRAODz07}yCQo5Bs5gCD7y
zEPDhJN#;Kwn_6^R^F&zeKslp-Vk_?qnwZ`rHIIlmZja>>BcrX4w6bW+fPLRprPXzn
zNZPfp=O{L;0BQ-!wsr>mqyws}53tH1sKke%fu8m!8ybOcCid!D>TF(O&)d4#*P7uS
zdO{0U6Xd{c691yf>n%i`pf1cc^EgVThiMPPYlLknXmdygg>h|H?Qh(N(2s{9?Mmfp
z?Sti{b1V?tjwY`V2|e%taN4Q5DyH!cX2>hh;sp5}3Li*o`}KLmb^(TdPM7FYs2_Re
z>yPadMz3#RNrV*t!#0V71+Y2LA%%FSFRGW_=sZCQqXbW3avXmSA71in-04K?{vkl|
zOOSe0N0Zh2#g)yq`jdy-ln|kR7%iqGU9-=rO;OvDsyBu75rb{oBJntppyJM)l7FhR
z)U|oQJN<}j2CeeU<}I8RNtjp|yr0C7*Mev8vHUYQBA==6M|Nr41V-Em(H=Rl#s4VH
zf$j#55$tCZLiaDG{nO&KC!jCt-{Sw6YBxG?s}0f)T=JOd;x366;G+R5G+<;%20w0o
z<$~3IIFs*D5;6DENx67qKI;<!Kz!#mB?xF*TXM-+3(<kU0-9k<Kw}3P{MI=YkVW?l
zPDF)HxvaCNBBniHwxC#r%82n2k~#%ioA4_tcIws-5C6w8qc<h@V|zyjRBg@^Q3E=7
zz!j9eTurA56aWa`^zL$h0Hlh6JRh<>p;o}-6qQPNWt5S8ci}<Q$(x=aJ&iql+4m2f
z$%n`}-T%1LK$XRaSE8||jP<X1iaylm(h5Pmr4oEQ+~dpNtJN4D4y>p5+2A~x8lj_K
zi&Fo#b${4yY}1?AARjU^pmu8Bu*%)tH@`>)%Q7`in=k`emCs)Do}hJP<%Z8+dTVB<
z^QC$%n(lVcKeUy{QlLT+vECvk^m^UVX^5t^2;$m=9XS3D&lebrU}PE0W!5;OkGpKt
z5GXwS#2%4-9H8A@o!setHn@8R^NI>$=*n;pcuMBLZ8px0Ba<=EUc@bw^Yd1}zC;ig
z@y4Tv6<HF>&YWEa-4Ax&AY8z939B73X(MoHdRIaEGRj}_h1S|8s4n6K7O;r+xE$@8
zLOdRLFS2#Cc@k}IN?YJN{Vav*#sM(KXNPLNJ^acJ-U$21TEyxVx+wnb@+2rH2H;W&
zKixyF31|X9j}K=$<~|8Pz1=jyP;{~AR3`$_qYSs`Ze$jlTX=D?m8;i!*Qt@sD4ip>
z23q*!^S@{xuGCU6g~!}I(-BCa$vk@AJiW@h!#Od;G`*K~@Z6=6FnF-DtGC^EZsOU<
zU3KYCbrSj~<oKeOAZoj9(l53amq!p5u@tx5Mf>Y46fN}brItlJj2N18*fhD(NrjsK
zhIy~e0UHeM?9eEZa^vk<GdNSljMmeLXG7Hbyz6){iUy?Po`%?stJE4L9K{CXJ}IzR
zhd&FvNFdeNP4d!3-hL-Iydu={K{zlpy?8K>YpqrrUT;O4C91d&^;}+UokJ#jNuZz7
z^@f5ijL>i|8N2m>0mclQf>w<OT1vCMD%GJ-!XgGyDtF2uRCqDX&>k>=XZe>a<^|-S
z{4R1F4n)u)<Cz8%N>}8Un9SLBJivU|6q=hF+cCf(uJ$Il`TT_If6%5De%e@Q#|8je
zo0Es{6T-lHBrnAYgraIDbU@?i-v`e@g%oM!nW@*Is`|3>@?Uf^3&OVlEj0}7S_+d<
zFW>50iDPIffVkIsc;sIVU%!(x7PPO2J`@{WdKs$7QX9UB&up31ck#$D`b;;USb_SG
z*;n|p5lyXBwhYHVm>0}B*E8_QfI6G)x5A{veRkCxLrWheRD@2b@<%dHi&A}gdpiAR
z6Dr*;xHYv&USV*+Ui^2{shlUj)ysmWINfAc3(5M#<_e7z8Lk7reC@jy)4D<Ybb|3q
zxH!Ykv&B`#Ee^|7QeG;I8r4az3S3B}bg*y;orXgK`ywO=SBjgUId*WwqzJG5AOSbB
z^JD9|#|LrE0=_YFVp0GxDe(i*K!TVa-*ck3;XBE<gW4qia9^eOOMB>m3ghU<4<8!n
zZ+fS9;KWp90AgBhRxyta7!WYMjxfb{0|!J5Y>1n=P_$Hl8H(>xT}@778eV=3xYk%#
zbZSv&^AtAzw=8BAW1Lrm^x?7meyLXXwk??d+bRrI4iRS4^5o%fXYYvhK%okkr%O_O
zrJJ?#ZD)oe3}(i-#VpH~tv`|+R^#^|D`^II|NI@UwPCAT-9I(6;pxqN5}u!K9&~68
zaHpI!giATb`DmkSE)9tf+%NYB4t_;$Uv;lbPbm$v*DQn<5`}#IE%r%Y=>*o_As06^
z=%nge)%kdO;d2qyuvb>M5s8#1bj0d3EzB48lxLeRIZ_2d?4uoN518Kl+(V$}6Kuy(
zbvdtdfX{~Fz^`!efxk3!k0(Y-^YCV9F0eqbS<A`EvU;%o$(ztUVNs!SM%w|RhIu_-
z|93(~P)b0Ik^KmktPQAg(LWv6K#6U?@7*L{+ux()c)R`;uWQ*>G^N~zb9OONRmKAw
zlQV%b#aDmRyvg9z{F$%n)M&c(%;5~3YF)d#L;9+pK6X0#)^yR}v%y)B!%ErHTNBz^
zN^7=&{h~;7=fTA46w#e{tC4{FpB8%o_w}UfEP-uR7RY%(#6iP3bp}~V@pO;zO){3Z
zG|n9%A5yVqf*bxbGyvf4Q;)$Lg-{-cL{#x0ypM^F_Rvf4E`WzPKZ5{SR<;05dCnFV
ztiN4s)1g%Fa;{c{at2wLAAOhAWij*4@6k`{fS7|>{IR<IJ9~no7Hq(t=E!*Zu;h=O
z?g74ck1K4P32rd~1-C6}<l$VhlklU=zc0~~<OwSw>FuYdkv8#aN4tbR+FBne8NBQF
z6%Q7MLhJG&d5p>tw;ewjaiVJ~-)%1z^e;h=5~M;2zQZ_s9O{34?jY~*2SgUT4{>D-
z1!VwUgDu;di;^X5?lb427~4nR?yqOw%Lc0B-aHT@)#(k*&$QeN21^3m-4+VE)jWM`
z$N8nCZf&o-o=|HHp}om;%}rY`Z){TCqPu*zx8JKKRgf_3D=+rhUN5uNnD*&hbGnNY
z$2`Q9HxT^Z$=8h5HmoGcKx1nB*hv+Tl8{`r1$aeL_F4j{5kG1|)Y;{mSYW{berqX3
zRM%Q}O1|;wd}vud(uNOMxsERpDoUhe55a0QT54RR4gL-7AB;KI5%el~tW7NQ>4l<d
znBxm&erFdq!ft2A-wOB1+<4stELm#<XX}FLoLE;@yen-#oer$!tn6;Ixt+?bB(FPi
zCqVjqB6$ebeS*2XX&f2a=eN*wnYBrvBuLAd^*xhk#Aj49)}p(s`J+3H+4SiQ*7VVh
zsRlM=0fDZ~HyHw06yXJ{O=;YKjmu{Fn!~_W{gTQ!1udi0EUvdM4jDjZXrAb188-E%
z5VZT(1ZNz@c}Gva%(dYMSe}gn{uX02l3UuK6w6x2{<(H@uJ{@@D!@Oc?B`b!3l6(S
z*rDg0rrCJpl}*G*Uvw|ua#Zphchk?})mO6FnE0MqeEK%;-QFdVc$WV50R^yY7Gi)M
zx4qjJENG{_|Lqh@-_bY1%p@L^a@iIH2duS`uAp5)+mo`)V&`!sA$kZyVIfPYSjba2
zdJO8nc&EISesQ$aZGWkE3gxfEUko2S&U7x6uQ3mrkW?pC6G@`x9<nV+W4RJu<D<nO
zfHnW7-t;9W35TP;vM=$cf+Q@!&)T^D16(wsm80uD3d9;Kk5rS3&#mYKzdTFhaufn$
zGRL>zOW@x!Og?GtZu}~a=uU@@TDPDGWLt%tAOa-&L6Fu%09=U|5Va}J{-r;N>9+Hn
z5@KIsW&0pdGx@>rWl%7HFjeja?{;Gsmvk?`APIJy{D4PxsZCJYxV^CeL_H;%f3qf?
zIm9X=+x@uFVqeQMm~9Ci24Tu^q~9RS!Be<F$~qux(wiGOl`(MtnM$^sSNVNn>h+}B
zS3Nf5$ML}y<`c-5`lEOg*<TUC*S)bnb<;%(@nD@<-ox(3bE;D4Fw5WE{Fidu^@o+4
z2n4`UNX{G(v$@|Sr`JDcu<Nt&T{XEp7u}DIsu>3jKs|mGi-XjBt)#Sg8Z4>qb>4z`
zo5+8`B4k$i{H}_k>0Plpq8f%@HN8)#yQos0sMpX<CvChL9XYosnV|z_SF7Gik@I=@
z`(yVeA8oi%p_%ff_aw%HxhKhn#@IRNY#cx|8YUDCAz*wCzne3tzYuo?w#DCm_$}AQ
zl-&AZd8q_yoIPG~tBj97AUW|y68{{!<is5+73}HA&B-D&q7?!^OiwCuUY7_c4F-^+
z#(rOcHOo%Z@gI+_yQ0=U1~X1UfGL#Zu|A`L_YBSkXhIZ#?JWlgpI3gJ!cX`E5=Vw5
zvb|I%x|cR|*w8}1;Ey{DuZ74O2JbMaU3>|kWc3Ln{hatgS)<O-r=VmdZ=RzS%-QUr
zd)!Q0URrLjS6s@OCzK@9!R@b^0;y!;ZHb=#ikFBAE&H0ga|_$cN%ae&4<5yXvLd~A
znU*Mix08<_*9Av%NNc%CQ$wk~o=`@pQncCnjvg6@wj_A<tNvLup;;;1@SEV?EEmyU
z5`@GU9;b{O8>iAcy?<_VQ5hm_(F=ET{}sf!@-TRznE>4jHQGZ04XuqkF9^)x`r=j=
z7`LW*ktEkhoJJF0prw~NimqbMp4__L`gorFklK!+tcx#!vH-zU>RxZ2`#oJ<2c;k3
z&w0iYwb67cd@Z%3XE86Jg2OKFdn;BRZJe@TIV)@O6gxs^o-{BIxyb;HR5%O^d=JXl
z`$8csZjUe0{&?wb_w^YB=JL>Q!c}j6>;+Ce!sK76HYR#E=;xRgaA0|O2nSz^`6R#q
zVC6m_>MJV1iigXfhyav^uNZ^BEYK`1susm@W#3rRsPIq)8f$g^Q}sZ61>tN_6K30!
z?wU<}W?!!d87p4MSn~gR0ZIieO>@mIt5{I}?sf27enXlE|C3BC=azXv@(2Aj5n^tP
z<@S$(bOagfpP%gxIvO~9W2ib$U^#V)XL(wJkJ6Z2G?H)TxgK~SKHp{j;&Cavz7eya
z@WYtN63o@nJSRmE?j+D$Mh8wu0X=ZGa@+wiT5hP&s+3*n#n~Vg8TM!Wg`9kEc3XW|
zj)J;a+#Tk}<fri2QnD-l_{H1t_n({Ni?yS;&%OF>VSf;d`GFeY>yyK9po-UA+J2sc
zG1jh`{A+U@PcvQJz@OQ7BWAEIXW=r3%1AV@TS<`|*Jbmr+~zefLpKFA<>etCzapML
z6F?eY<n{tcLq%4{i`5HV831(`B4tk>uifzAz(K~rXi5<TfP7n$V&rYm=IND&ht`4N
z)x<+r)u5{OAA{Zf2M*9=r(o`HX09gfLiqt+!lsL&znlpDT%jL?*V9SBGe==RVTZW=
zejU4_L%<W%`~HIv?n}x8jZnM|0qI?hGeuGli~T~u0^lD@9I=|J>Dmi!y>rf%Tw<#y
zT^4*wF2w2^{Gp@k**_EuTJ{gB*(%jt-Znc>feGjMC@6y44=-pHjkE-)Z{OdyAbK(g
zcz#Vg-ye<G-Iek??MTgu%zD2Bx8%YaTev9wuy2S~TBay_o8U?=FCQb|0x;J`@Oo$~
z533h82zYX;7iLDlY)YL8a%Vrkv~c{%TficK0|>xmnn4+^xdNY_%hH4dLhYENa~D|%
zqBQR+i7_vhKzQJ80hoaxE95&_K4K!Cq-|&X_8Oc*&y*X3`k3rwu&pfQHpJA>#J3mR
z$^mRuX!~=@gC_aIQG7SXeh(6+IG`Mf7uJOUDq5*{-+vJ)1OmYFrT0g55D=hI$}S-Q
z^yK%=N7OgJ#)+n+GU1i(9(fB<Cwkbu|BLAl1K^QHAfsw}!jvIZ{s{+jv2N7{2>7BH
z>FNciVO&?wM@l?kr=W3mc<QRNvVwb9w<}}#b3)cDXNHgL?$VC?QGmsiByT~)f0AeQ
zhNb8@A2~0!J$y!F?bN0_xT#o<Y4`&&fS|r!G2K73dz!o`Zs7w=;rIz)ofo19XFIw3
zSgojLme+=HEO}SiICs=^yKhsl$lXaTPs^?vK9XHoImL|@Yl#2rM{W)Tr-?4!P`ArK
zh*-U3V3gZbl+v$Qj-)FBEr<-afefY=HVzKjy8%S%<EJI5#~AL^fQVb#JKavbFo-Cq
ztGLa^&qM<9AXjL-BBAG$o?|{IM8DDr=(a~G)_c<~BDg=jyRKhlSN9DI|NdAlW;q0*
zFDfj4@)!Gy{F~G%CR3hj?T?SUcRY*&<NiLVh1Cvnw|QuXHIr;_YC>`hZ84y<xFqZd
zMFjwUR}Sw4pz5OnE%Sm2(Ywlkh`eu&gt2=yft8L>fn?*`dv7L3Syp<bzHs(`l<Sud
z1Yi`j%CK%+Vh|>XztxL36Z{BF?n0c%;umcnF7nUB*4D&_eTiH#W$b4TxM{S;tX6nF
z8HtO{FxxLPn{x4NXL99XBdWv)=VTMF_jkeiSFlet-~V`2wg!v6)ogh_DO|&*r*q!Q
z^5thv8H+iEe+jbU5GiVh5nMYEVG~A!2g_E_X-6n!80$)c6|IVUiIh1({sY}e!5|Ft
z$(-*c<B>hTt{hoO11|C+u0QG~Jr38%g+0*D!(2%J^bwhKK0S^WVvrir`TU!h2uwjL
zSjx|`y|6HHWja?c#n{7<Wm{9^W;|mF2(mbS2a{LeF!4ZRsg?KFR#W32Cg=y7P3?<u
z?*Zm5giVSe%E4)0<Lv1p{1d#~TxDZWzV0aVm9+zLM7{-tZ*M^${-X7nYBCiXFnrFY
z&AL$|#fgUVX*Zk%_#vzs&XT_TmmAVvWq^c-Jtid=LPErAG8Kgx87^`!Rb}I>P1E&o
zP%@Iq3bSC!LCqa;PDs}qKYVL9ynrlNCGpx~`GxEuDf!48f1N^v583%w)kc2P7ChT6
zmkO<8z+j9`<XE`J;rg!Bd*LrNlXqK{6NWKr#5vX(@_{6<g}H=-V&nw*A}1Ek)Kx)W
z-W6siURBTXY4&r9kLAF3r&P`~b!qc}=a1cJeUYTuK{$c`hpD#?i>iCWh4&0YcY_F`
zbf<K8gP^2zgLL-*qBJ7iAX3sH9V1AmfOJYpmxScE@%^3ioxkV0*6jOv?x*gx_FgRB
z_NA8JQ!-Vpcay&~q9`CQZvY5WxUCIcwKjA>&DZ94YtCI*6fp~Rd$%ZMaS*N4#3>;X
zB3?{hyRQ_TQ5{s+T0JIEJt;&C+Cw3L)1u+lpGF(IIs$<vIvAWla#0b31r1OUpKJ<4
zhQJ#(b_}REL2YvtADu6Nol*S8nnU~vAe^(||2HP`zcc+{r5jSf(Axkmzt26*j)gIi
zGcd-aTR-XumncFnS{U(j^UYNQf-VdEvv{nj49~bQu3*DAf4}tN+0L8oNUSr?xA=+l
zXi+VX#n(76d0?z4eboJ-AGbL$J^!cgh4@l}T{#j{eQ3D4LS5e5yhg9{SA}V$a-SMM
zBEvCh0B2^uaV#4C<N&%NMz6WmH-+z<qzgigeZ~fV+je;iR?f(qE2@5Ylgw*BV8(}w
zG4@J#9&g!F)8PEer6o^lTm|Ymb2ItE8X)0@b|yI!{wY}j8N5P+H0D#>Gw-jn()Sr8
z65&rv&z9A}vsU^33k0Hgb{Js}4M>QQDjA?^*hN57R}!H19q~s(U^HE&q>DoU?>>U(
zEE!IP300eUf#B|s5dG(HUFv^F#I2zKVUCr)H+AC<7m+)YR!-PX4uhy|6Z_J8m#>Gv
z+O{J%5oTmK#>3Ns=)i6<nEpimt~?$2g)|2<&A>UJa`PEbi^<fS$yd$TQ`F^Awh*$^
z<vEX$Amh%D=^v0`>w0}r>_`62JVnABXFhn~<t!8-iw_9I)P~_NL<Z0uqUXgvw?TW|
zx=*li6zfYN3aX!MlYJQA{m$t9R>^r@Sa~6f1?UF5wDo@8hw0B;k;Ma(kpTIS@yH|G
z3f>~7D>L_r3va50{vgJ-NrHq}e6&BJS!BT6r7VFQ+RbgF>Rl~xq{H3xwQ3){XYdu7
zdKzdCn<pvZ6Zgeh0o($#LEPEnbhU(g2~YlK4|rWsb5hXrVD$xLmnbZ)G}yuNMB$Cj
zB)Lk=L_jOQD0DUOWdi%aw4fW_=xbbZo-=%rzwne+wHz5~47&lLW<QZ{TG^8`@kDkU
zQjSZ*Czgm-`A=r09ivJw)MRolLi+|}HqX#6iVIYr`Su?eyI3J7n6FrY=x_kB{gSNt
zt;Wgo)o1RS_Tf9mHQ3n=Puzug!jlz|*7CwRr;`tzb9}7dYaY2JmSFV74NLLj=p4V;
zv_J3wRv3r8KfISvhoKAw4rpvWE}P`Ri(Qi2h|$*AXve}{$yUSydQk{To1PtKPTzyK
zRYv%Vc`ky7f*E9=Q<yw?4gh_ITeLH-IGE<&vo+~bknJ7K4W%EsOnlaU_$u2d4F1!<
zoQ{2l2!Gp=Od~2oOsDTg)LxFYWLQY8Cm7Yo$rp82pNOz6>*(=|K?o1SJVntMRm)e3
z-XW{gkD>ATB1?<rRO&thl4_Ng*5^qgOvs*3R;ELYDK(Z{(=Dm*vZp0V3W??+NgmmL
zljBOvw#}#n(&5imZ;P!UD<9DSg+?19d-sk!<-(V8{&{#xmF)ALVVHu}f@UM$X+N7b
z{^Xq}9XfZcT4Ck?J|_-rQ}^|&)a@C3+vT%o#W@dw^EG<l4VtG$I~@*5ic?igcG9^P
zXd_K>Oxc8@IFHh!z|RAdG7~Em88$!MT0<m^a|lnXcPk)pBj(@q8~~?681$xn*wt^c
z?c!^#9P`dr_u)s5{}VwJc^QS;@1HS_6C&9ut}#k{H3|L^gy+<UW-+|cwk`df^O5F<
z!;RfcK|H^C39D1<Ty7Wvdr(clYF^YUnY!1Ectw<Zbr|(MfycNmxnC@GMbCMqR$Amr
zq%X6$BvTI$^m~q6ePI}cbTOkwi%<l({RmovwVNB_vqRURG4NcWDrW8->-(%)>4@5Q
zw?lI+ca|>=ZOT`LMNH;Ku&iVM4um*=74MN1$2J$;bj1|_oLS9W6@cZCiE@7BxR*OY
z8nk_us~!~JGkQ!%w+N6VdZ}GyB^kH7NNPZ)u&TfGA$4j2FFqwXL%rt!fKPrAbDiqy
zUj`Tte}sHznfR|_5dvg=oDq3eXbPFZ5D$*Q3l>Pfnr+ikET&pw{ITLvMxdm!mj=4H
zlJh(7!Qh#BW3sfLq{V62WE>R56|-1^18N6U<NjQp;S)^bbNfn}o}wDa&6XAAeH!u-
zRr27QfLR$e^c}M-8mWwyK<+zaI13FhmXz*o%)ea~A^5U@#2U7qjaaELtj3mP@2IG+
zROcgZF5ijiQS}J3IL%e;id&I0M=Y~`$b!g75qE{+0ClK?7!fu`G{Td=nWmo=llc(Y
zIz(;vc&zxcDPRFvH{7-*#@fr@9ZqtPLD|-dx!++L06yP^FyI4lr2!s-H$EAf7qntM
zEUlp&q(}Dp|Dywf)SpqNr(6(9U~PuQ>QHbtSVferxlWLrq8&LGgJ>fsvhleqUhrfs
zVMR4_qL?keV8L2Qxw7Z8b22RHiqFRyZaa09e4jM(KI~;d_d((|8t)?<h2PPBT?50k
zHm^KxU33K#mPKg#ENJ4e3N^sNPe&HW&Ct>f6^!#Os<TZZ=MLwHUV0}O%b`A|j{MnG
zTPx@7UzGqzo|kHw4XvV_PFOL;-)c$c^Z1dV83=q0zcTn2!S(ix{HefgmN=2Aklml{
zghk-DZYmj|rhRvqVC2lQ#`kv=iXhM_=4{%31oV{`M<PGB007Y!FBb{k2nv=kPG7f%
zMq!+@1^gc?F!(dBN7x2`$q_TTOF`Cpq8&_yWT+i6@ZG|~KjLC@0sg&7u4!LaQx6B@
z1%LJ1U3>3TNv=V9&oUusmiDt6>r{4kQ4t7BsA%%0i`n>OrC#KtmK!q-T=cAchtByn
zq<Ui?NkSH%fnWg`u$Yz2n|fMIcimhdI`tu^#g2k{Bqfq+B|%m7x+a+{@PeQEPxH_q
z$5W+fp3h@9^JmO!6q=q-JOBxrA+I;mu_T#d;z~3Js80r8+p=l0T@5TnKU*dxKtk9`
z?nu#Uu9Yr5l#zF?&#@BbDj>lA<4lqp2!Ieay}A|ipELN@yqmD;lX8mFV^6vX&bQzv
z5z+vOSAJdEZ^Ga*#%~7(O9)P0>g4$nH|22CdnLpldn{#OhOS;Wyj=~xw?Aw}`b^)B
zddKKVR6U(z#$oJt*p{*OS>f<ivnGtSIGnn;m>MH|1`m^&nvq1Nnp6W{28R3sPej7E
zNw|;0iGFrSg?__zekLGCym8UOZPxsw|Hm}`!)?ZIFL^2Sg7=J%=l0%-IKl)*ng@Hi
z26i_bj`a0URaB+?lj&_lT6%I&nwQAlS{igYvf6$*G4ec=3*c`iVt|{ruSq|KEEJ*T
z)@ytOy4Y7UO3o6$XrUFFw3J>#&hITIR2XfBxSgM1JJXQ7s$hQOg<AVvRO+!YFQ51o
z4(6vYmk{6z@ib%B;h7w>=?p{HcxPFfC-COmh<O$HyOr9m7GDUFz}MU?v)m=-T|pK&
z(-pc8vn>I<NoEEcW96x{+=`MHG0X48e3_P}qi}Jj$2z<Uz|*}{s0q2lo=x}ZF~J<n
zDe=O?Q3uy76w*<z_V|s*Ws(buJ=mGF?R3l?Kq%>SMO}F^x~?gfOwU!yW{23Lw)~L~
zW@@m-*eAsdN{Aw=^tqgBvrF9GD)^Bn<K5p&^t<c*#2I?Cf<kJ_kDo2_UJ&{lv_P!J
z#})Pa+7fgz5hC+H!$l*XHl?=kOO3R!F<HQ=So6a1jhUfw{W5o`<$@6*tjeLsl`k6*
zIGtmXj|eUkzhALBLZrD%AJ49ewgHYVy%j#OZ=4)6Gl>xyQj!h8I30jb1dPOONn6?%
zM2mY}ZCb37x3zU${-B1utun_z_>PqCEfm-2i4PwO3I6eJp*VIndW`oz(#Z;QPioe<
zx_<i?Nr-XDKGoO@?y$uV=jJ`dPeGqU1jd{D*2p$uPWUbxop+m*3KZTZVWfVquqO?v
z5T-I+)cz@n4xgDnShQJx0Q_UOP;@g`<TKieb{;#2+-Mi;K2G_3z;A=-gh_(_83T+n
zR=u<INZr2|x_bJeGv=MQQslwh@Pu<CQWydeY!lt=I`ASrCa-f@JS0Qu&J=E4wO_b{
ztdY@8$K@UQ@{l~O7D`51&cdz*WeoV9#@V5gYTkK;U-Y|SZlvc}dT0Jp3W{dhON4j{
zZ@ztZtZ^Q799Tyeb?qV$X&v+}?Y9Vrqj{iZX8+tK4*&dY(0?9|cV^LM2o|r>T5KDf
z!!R-Q<+z+;0WYa5_R7{|BJiQ=7$WkF>3EG+muNaWRkdcQq+J(x`?u#{hTD_SwD|YE
zAk<qIU&1ht9y``o;B1VD6M6o&d+KY_Pc!Q}f^5n?f2A>W;m~vZ`rZDJXeyImOgsX5
z?ybzg<M1Aa3K{rb$eg+vNxqDk(5|>MG`Z8?m;>JIctkMx2yIDOdjw<Uug$ZOMBped
z<GmG!!pnV0p)2)%YbLei?+Xm*)Nd)ZfP}C})|Ay-8+)dr(c_tU^BNJJus4!Cf&1yh
zbA|5>X#KP5{UlHcCAbEE)skJmT=>vS0GBQk_EA}OPG|+0%-bmOz+{=c!)oBxUKcGU
zR_X@NvsIlW<-mS<fzRbViF1`->uthu_%pb#<9EEn7eC|t%rWXwBT+*+y^6l(#`5eh
zFHvk6VO}_1{xQlTK8Q@H#xK2WEx>?d%Y+h8U)OT#YcAZ!Ywn3$X>h<%4uY7K(J?7R
zp7iTMqTQ|fP!qTyWuA`Vfl#mL_XYmflh!<*uPeiw)CnZegbRTsl5rQK;EQ~R?oYC*
zT4Y-nW<mTMymOJpErFH}ZYdP{9WUlZbMs*nqA4|8Wpv)MsumTU{^qFA*U`a>{Dbo=
zYf|}9iAwQk%0yR?-O_f&@Y?uqWlUtWwBnBrt&Qi?n?#l6k#LDc$MF$(vH*5fcrtb=
z6}4KO`!{9?E{Qn3*wZ@VNz<*0`U4L2k1gfC)q$Vx9DDWkW-M-d(7ymxhgjCyhpg<G
zG`OO*MGbL(pm?8jd?F{=vg6jvfOmgKia++H0#kx<6=ctkoy-h*?!k+uoxjv)(MbHq
z)?F#~ccn+JlZr}z{e%vCdJbZR-be}u?q6kn9ky<v+_P9%{PYl-Azq*7bssf!xP+8Z
zoVLfv?ixp)DPs9h)X63&Kj5QS$?Vx&B2khyeOAUzKy{<((Q&J1^n_1jS47tYTNH~}
zbO0Zb5D7fOphHh~!{2Ef)<4xaUYqKOP+{GF%a(<)te$IpL0Z4t6%vgi({lRq8%gSC
z4B$x?=ULoKv2#+(#>e(g)h93|!i_eVG?hgk{C7F}4;`QI56maQx;ACOsXS6B6Tc1~
zd*#r=@jO)^(ZsH}#!)Hx92cVM0y(3r4TRp+YqigWPwc0STOFoo?QTizTTtB-O`5m%
zHzg26NvfZN`A8_-$gPsKF^==t(!qD#A|!RwbGKPVwzQwhXuXqV&w?nm?;!2j8z@h+
zTqeTtiR6W$-b(lqG*6{D2lJaHTBOUq^HzBmpB?&J|8S6Ew=WX=X7d<_q?hf$3<(*D
zFzJ2(w&{>wh5-oARcVh>{p$2_?Iql8zt-h|lQ?!dB|ohUxu5|{N^m>R7g}z5IKh_5
z_lV~Qv_o?<lBK9!tLL;3Sb{aw+lVZzZAm!V_vw?~?-G*PUY{{LJlS;StunxflSkK%
zvsh)Aa<a`Xc8ht5h!)i`Nh$pJ7TMv)jcb7#R*3AfOX4SX3F)!nekt8o?yMHn!y)2*
zu|rN9zQ{QQRXcyVVu9=zurb{}g_wH=DPIzNR^(IXW6tBAvd|9$e=taV!tjfVnkD)A
z7Fqjj@aRuCgmtSo3||2@BYH^$f?0y6N#0CYxS18C682tKP-U~Wgnyt=KUt0pr2=IW
z8?bYniGr5JyXQ9y>n3Q+?X+-{6BS@o=!{yGL?!6MN${ZeyC-{ejzc8wYdSqYv+GN8
zcOObb%%TF08D&a&!J`6VF6*sVy85{qODGOMPLv*S`Gi#^3<%X_VBVf`v$_*s8;aqj
zP`&XY|LPLN7CZ9mb+9=lBw7<TCes*ge)L&=okic(nXAy`azbzN2UCgF@th=^UVR=D
z8LoR()nXgG=#Q4v(`s%wMxZ(|aMycVRXV<Lnjzd072MUE^BZ|A7Ay|B?^+jIALzJ1
z$mV9X3DcMjT<<Pfa`f-Q4s4)ny+6kzkl^&8hp%>)<!5XR;=4XS)cCFzQuyXu*3Cd;
z{{&x>$0CxDl9<(3GA=nQI;xMIpNj@kzuxZ%t|+>O)Y}An)nq_PduzPOGIcyF_uC@=
z+rV!DRSBFp;qPj@V^gYPG5NIuIx7P5z$g>chTh_{(Qr$ry9r9Re3VP_iKg`n2Einz
z4BI0noa*p+<$zg*<DGPmPso$Zx*f*>uYk*r8*}1<LF>F{zK_MS^GGF7y-dOcm3gLx
ztdX4=f*xbhxIt%r1`2hl<T+Nil-{gWb4mJm{0E6auS&QG#ruoNGQyAanBZHK`pVF=
z;+NC&d@X(P^mmCW)-NLBs<y&$JfB&NsOdKB>SuarYb~Kvnmbg!+WAgRwuZA@Gvaeh
zWq`uN10*9SFsDDUizBadbBi+q{3{R<v4D{VZ0z+6su*z-sW&<+GS6o+ry&9^FT-^e
z&o_yt^kS~z_iGokxLmtB8C=CS<Z8q~a@kEux@+%!3jt@#?A4gHFT~w5;Yl@2F}-OD
z46FI1V#pJ22B64B^!&7GG&&{xsU$a<iWd}}I#q-ak>44<#^(7tS^^Sh@%`zN_8Up#
z!23gojKy2c>XJnN&gPHD<8>q)oS>7ks$CDyQft2?w_?xzs^3lGhV=Q}NJR$lG}wO(
zbLHegGI^D1`)j;wPp@r%andbiW^`OMn4zOr%tR4jdmBATCcMUT205&w;t0f9;Jr!x
zt)*h|orvej#9;)8X+BTt7~~RnCNof$!g&pB^gG6nkC2J~X)*#DS_o71LO$QNJ<!fQ
zTzj3@4O(o4J8P=n_$|U?>ssE|3^Jjm2z*;l15gBhny8Q{m&lDT4@fUo2Q`5gQ@FWu
zrBoF;eCb1qxl8mP=p5%#3KTUmL1`?ldPB1SFQ~U0oX=$wS}4W`lwIzSb~BoaK6!=1
zCy2z~%0fQ3I=n}EBE7p}Mb!-5Yl12pI76bfU}KseuvjL(bWHD9miyFP`XHK7`5Fd&
zSk>0zaJTyI*KYTW%Dka7vsXy3ASQ>1(f7>io>7ZqR46^T0U+zVyr4WR7<r!f>4L2A
zDSrYBvNS@(o#!Z3>>wS=$x+F8rRdQc!S^scm2C9f9K5~P@_o9tr<?4gL~EmwBmP{`
zeU)!HChJR>>7JwITA&QdzPl)-IEJ29wM5cowxXY^O~yA!8n`zrmJo)Tpz+GJ>5=Ze
zf6`*p92lo7f?H9Z`&ml7Gdulff0kD?*e~CfHvvXEurVbt|4A3+itAQkU03}|kto(&
zR?Y4i+6*c0zA61xa49L=$CwSU1_XY>0BGVjALqS3x_6iNB){~(C@=!Nn)~Bss=}<k
z7>-~dnw>6%ItxKm-_HHT9?yI1?%a+J-xP(=v9i<u%jk_dXr2{CugUPPG3paMP)Ceb
zoLQrgU(Rn*t6#+9(&bc}5M<JKj}v3o8tHW^ng{Cum#ZW13guW>Ncd6=)1JJVrn8B7
z(p>S8TlP+&rK6x$cpnQB^{5Hz&;%^KgpJAl?)%m>D(1G#i8Oj-mRaz6k9CKA)U2gX
zUnh^K=q=?_)_o^b@@_(S!6-S{Sr~OLm?wJPDyMP3hkQ2PMjq>pDYr!b$g8#Qstugo
zy59C%r~&Wi+bU+d)B3i|sCsk8JQ8(_j+6e&<ZTcm*QMu2S;|qQq++IIhuwOsY+oI+
z#R4tr0+Rhlm>W#;I?DwL_s)%KLRdAB2yvOd9vpQU+I#POPu(nP>=yB~Jdew=ZaAIR
z(a!!|I)BS%xS)s1AIhV%!p9Crj|Ap0C<O)FF|P8lP?fTodh`Y&LXJf@^xn+O)(KO1
zkZ=qm?NR50j6+&+dj-twUJph{(QB?dZHRgv1(cH_L_NjRY`1y^rmo{Ni6zQyN)mV&
z;5(iRvTWP2+X@b*A~+;@)c^eQdqg>x^AQs%N-*<?G`%Ntt;Nal{3a$J7je)F^jKqb
zuLIUqi>JwRUs#+%pR})$sHXuF-%sLGSEhV1s5V=A`7ZM)yf$rP2p(HrG%OsTWa&p%
z+k1PgK)vPhC4`?oWx~n2`crR)l8xV8IiF=2Exe(KH4TXxYSPd?kPz19%?8Tu+c;iA
zqGiOhnQ5!#l9@<P_NU&r|HRC#%;-cr1;3^c{TlR%T-{`U8X?@RG7dRz0YmBi<xg4e
z+1K|{|K;imsl4y>?fpcYgk+;+$3!zpq-zbsI%Nbx^hz`DQk0(VLzD&d+_=2IxU9Wz
znLe?I*z~D#Zg%i1s%pbY?)Y73SWaQ#vdwkAe@^K*gukS8+;5iE2rRvVIcJO|#EEWI
z%}xfBd&?2Wkwy_+d^Fd^UdnQ<)-W5ln5ftN@InOg68Tbz7<l*T&bQEyV)Rlzi^n10
zw{ROk<!n7&-S=u(B?_D<J$^9123f^0Ye~h9>;kt9@)>{s*5hW1?w=!UI4t^|DgwS+
z?`<VIexFrktya0I|C?^va)O2Zf6fBXOC(g8s^{&#qsuIdDUsg=TX0>g(Yb8aAD&se
zvU{-siHB74wRvHSVJED8)#v-3UKfAK-Q<q?#D&+ouR&H}qAON!E$pwQAh1O8qm~J@
zj{d|;r`jDyZPT6bOjG}h_b(9td<;PP#a0dj?I&sNVy_eM>%!BRxg9M`gK>Fjw(+&l
zMrgvTumnL!W@J_|iM6s-#TU$<H9oLhu771l!amR02i)5d62MN>Ip{P{A~_`QaA-H3
z4bL&7q($Onu!7RV4XP)#ksdniLjzGoDOJL#-WB)PyRiY$hg5n&YB8dTyiL7xI`dQq
zrzmRVn{LRCThP=(wA@gw(5`jStz7X77psxfU+8928!hXQ=&PH;eNDZl(0#5=FerMP
z<8+WYd*((%qTXW3^>swzarW3J8v#g*%9w+3t4sr#UC>5G{gBfk^c7a&`)8Y+BG0NF
z`32XN&7Mh6{Ku$GDR+PSf>MbU4Y=4|H>Qd4)USr&hBCZ^i8o?L)J$7fV%{+&QB6e@
z1M?=F$~nG|ASU}jW~}{Ub;s`(`KcVO#-eBNwfzqx*%glIv@nw*k@O0*+yu_VDA6r3
z9Zswy-}XyNE~KtS*F3+>bmwwgBle3CaXTH+SO)kl#Oeza5pslCfk&WAEi0v_C1+_2
z!SY=ErZnrve6pSiEVx}h?%)bc0|AH#V_@&LHfq(eT%WI{Z>Fz#<|shX=tp5=3S^HQ
zN-5XPZ@`Dyu&mlm?RCmHoL#-b%D_1<=Jg&VP@C%2G>BPu;H;cm#1?t*f|h)j_+IQS
zv4<gaD#uOv6<ygi?AmUAaFIgo{+H0+H)Kv}d<ko=r?$v0A9SD$>>au{6{)Y!jC_Fn
z=ZBlkQh}SkLpa=#48ShaJ}Q`)USn3`Z*tI-CNe;=PHrVR;JpibUJNTieRNB4>&ZZK
z<5|2~RPr*P8`PQV1?#Ca@)n<ChLf7i=gY&HYubMjOKyd;-kVBK?8e6%>-)Vka(rtn
zf%8KaUOhCrygw3vuEkg0v|U-ccBR#=*s&rD6MYG^liyG3-v~MKt5ocv?v9qcilM`H
z73^=HJ4dlG<FG>-vEUNkS|p7Gn!~F_h`-Lp<3}e2G+m0iKpl;n2FTBqJj^0mzFSv3
z^?V5J$FQRWX1I=X^o^Y9)uS4HpJxcc@8a9YM{loR?AaVytqnehyj*MG?~4c&_2&jF
zpyl3DZ?>4$B30}hD{a|X4cncV;y-n)m}$s!yZ8YKrn9&o7uDWFs<k4y#-`EZ{ZVG<
zMY!|Dyl0g@qv}e9(1#-0YHxl9#KODvtGXi+q}6syukg&wNxMj`M@YgxPD~4sJ{!e(
zm6Iavhq1zG;+O~<dlkJ=@&@(0?>8U&N}?d@^PwdM(k4lU#nD2lsL|KzGsaBW01hOz
z90LfrmB$vu4r`JdQHhMWF{w8Kx)xlP6Ck`i$F3>>MOaaD{_PDr7~-$HYkxRa0zcJN
zh#<P9Ge+-VcuLvm2t14ouXEMnXfJ6I`z=@%Pe_t7Fy%zn?S{!gYX8#UqKXrvmvMh#
z0$pzPtfoW`NEN_wGhQT2Y^oNgFsP^EXeUj%8_mw}Ye@K=h)Z%b+gUKG)+zOInF-ps
zs7eX$UHP9Zn@;XD{<1X-A4NJ@!xS^sjT4*h8{Ji(laxD#nULGZg1uvEeR2J`SBJK0
zMnkCw-f!fiy4^^vqo5yz?=#Gy2wqw>o5NLC%ID(T4DelD;PED{V7}z96{C`p%zaSM
zq8o)dw`qtMm)=jTF{w?d4Rs9p)`Nuu_YKZut&*iHj0Y#qw4gqHfQ9qC*w{@y*(R{_
zm&RhJ8zx`LE<xG$Ytq~Uh*J5a)*jWqUzSYKO}hw}0U45q6%+I~J&N1GC)Kz(ld*C8
z?ul-kyTFr_UJLxnS}ol@cM7|p0&nQ0INbOpWCdydm#@-t!1kR|P1_=y$8TiLaY`{0
z;%?FWBSCtgbeJGd9QF4{W551P!0{KIVmU=&lmXpIChO1h3BeD5R6h^oHC-axUg0@*
zOXdNsLk5PYQs1S+DPgHRw|>HFDjuAUJ|&0e2|Bt@-rcc*xz@**WqJKQ`xD5dxPs!o
z>le}7dD7gNHnjJFgwnraf(ts6VPl3v@(nDFx82+VR=0A;SjRhexi=fv$E%yW8)^+j
z|5^`$aaZtz9c8KK*5+Rm=N1^&L;Z2#O}>z5Phlc8Yw2w;z!L%Iv4xUf^$kT5Uy_H=
zb3a&0e%o7`_D~doaE#NcxIBft{b&tsd|CWtQhJ}lQ*f|<_7r_aIs=RfKCM2RqY>Tj
z(Os2Qng(p^!~!|C687q`0u$fs@bS%_7eaXJ_D=E}_js?hx5X4-;I*S$-`|)HJr&Zy
zm%u&Vo5Q{#$OXF&{wQ(6C&gET+iX>R$B+3UjQ8CA2HG`y2_-z>%P%1&af`;o-LWi`
z$-w3|STeA0#X~(l`-p$kf|3nUWL3XszbHuopPcWoJ<HzbZ-va?f#r1@BxC{!MsVoG
zeEa2<7xQWebW4zlq8&+?!5OUBX~ds>)wB%PIj6Wn3hMO-Se(Tsx*SB9LUgUn@An$m
zrxF6Xq<tJRU`6vO-Pa&%-UuMF*rK0Gb}&PKGs&GE=9R~VtbFL6?=}v<&%*7m&EQhA
z5iIRw^=uY(5C#J{7Dcxh15mcQTa-(XH~18Jma>C<^of$i{v@P-2PR7U9;TtWK)j$^
zo#oGCt_U_9XO}$tTT@7<(h(6K+;}Mq*lFP!HM&5C{MWLH-B(S2u0B0wv#US2!1TNQ
zVoKQb`Hk%lK`FKCrS|7vIoDH*-XMHq-V~bN)Tec2Kg~uxB;>Pt9GH+arwKC&AVX=G
zKm!<JWYMNFAY+&Jx~o76&;H#_AV9e<JJkjFZ<qBd3@;Nw7MM2k=e15rosWVorH-c$
zx)T;pqDa+ujXI>bl*O4bR>=SL9<fN5(!>ZNA_-I<g7lO&U~?Cla|)Zf^DUB{JIP03
zsIGfope}Z^ue7Ip^<t2Aw9F=KEPQalcVPEU#!PG&H(N&k*#kNqKsF0IKi>2Y$$iZz
z26nzBTfPZvyN}2_K&`j=HJ_J#(G?H7ht=LMwhR2rXMyuKoy-L7xl9}$pU(yvYBUYX
zLBic%{y0yU_DlQvOp5F3lQeU*pj5b*)JRR}qiZDhYEskTNsw@@<Tr-+5>8w=RvZO&
z6_{6cn#W#dB0bbTuJepyP)+K%=IFX7O25!oyy4y8*kQ3bUNQHZGPsTHCs;wgWZWhG
ztg>CbRX9LHWZzo9hdQBS6?=h!o_;!?WDP+P%U~?~6EiiA^B<;wXd~Ip@$()$uOI39
zJ-M%}w&#(NXIg#?UOMoYw;Ud7FtQ0L{+8;{X|+shB3NNMhf2HaqO5_x#)(R|{w|jo
z<;GD56Po@YOI72PIpuRvJQjqPP=&(r$EMv<mhHvS`opO5qOJLTStN;6lJ`lAq%6?i
zG@exl`)qk{Lea_SNk@r_uyK<|2T8pNavz-(4>Dfo(Pjf`FkNs&i~`tq^J13pi&3g1
zIN<yXXyd0x<1;<r2!b!M{WKPkTl?e@3^`*5+jzs;ao@(rqU<k-PIyMwYSGZC6U&GZ
zBKcx>s)LxA3sQAXWTJX;V>bC6XHGoQIaxC~<Db^vQYBs|qDw=kg|?IVgDDrOIvfHm
zNAWmpcQC1tbw60<Ur2?1Baibs|BUuaY);TWXNFI{rOPkcZT}z>qE=`x2D^*_-)pM!
z;rj?25O5UyY$fIms}wG{M0MFa!xyZvecVyYo+xja{Z)8(4SmI8_xcO;a_ox!{?Cu~
zu@5t-CV57HQ<FN^)>=Vz8z}hEz)MEXo*aCW{q_$O!a>AEk$<<^NAnkT_GS;S4Y)=0
z@hg<f;b@$xEk-fI5($gy=J}|NW4Zz6BYS?V+`3|+LW^1-iHyA{7bJk=@@v~iLtbjI
zxv{paGMaiV#<g}NKg8v*L9A<d++UGRB*|JDyzKrO|4oLuPrx;7Yj(Eai6^(=sZcjM
z#H9F+TNdz&I`z3oO8ARu0)2N8nR1Zv+rTWp-Rz>@;D+=kh)iz&p>{fBAeO%Xg{vl~
zZNY0H^w-LP)0ifogu_;(T6sOs18J>cCZdebx1c8PXKN5}--rB1SZVZwXeBBRGI#t+
z>ubz9sISx6gF|!bNMQTZKaex~k>2N|k5pCMB7R6W7~&PkzBdmr%DB6`5J5rI-u%q=
zgn}GBrP!5uR9YhL!@3*Pt!c2oPV6m9>Dj^{i#KS8XW33)KYT1=C-m*_`Qi;VTzFqr
zI*bf&@+pdfBD7JyHDzhEA#lWgs_>AMblBIy-s}h7-~JoQgHU}}{w}f>9cGY)wQmc+
z@0#$H|1OHiNvZ-r$nOhL@tnvpBR6mC`o}_^U8y<3+V~^#Ur|U%JBDCaFkju3IE<Qc
zulj4jyEh>B=l_5Wt(d1y#2jRMn8#eJo*LxeU&eYkQv7ML{>o7xi+Zwm7Y%V;b&c<o
zNu8DSysNo%sD*nxYx`6p9>GPHlV<Sklt_3)pSLwb4%mgf6-SW1$si099PL8}q0s<s
z`?9GXLh&4Es4#6X=d-AV2_YK%J1M3LpO7e%uJvxX#N+@ddg42MT?eD{ZJMCR-g(qk
zVgUB^_re;4VsxgrbWvwe?@`L$Mi@toC@7SXLx@%&Ax%7&E0g|$ehD0Dgc7(|X(j<_
zrHd=04GUd*NdF>Cf-;n}_<H2X{rjEnL{&D7p8TDQvQ`&kW^vh4^GMuR*=t05gx8el
zp{&@GIv8h2htPQQ-s<SR_SM~#$rLM0z;3a5ixpB_ssT()nKr1|`yh`XdADLpXn9~h
zhAjzqibX|g=Kx2cje8>SjN>h7Ky*)5*Vp@9EtPc0oqmJk*GsSSdUlKme$<(JLL+A^
zlW~3foP$r_rr0_z<ExP(%Yv#g6?6rlEBZH0LR0tq+i&o^zt&-)29xRqnPwr_$~)Fm
zewMo1S7G)A_Gbgle&RK9i$w}mk7|?wQgb7{=!`H7f#maAt6xpAAZ>9+;od|qmkvhm
z1BR(O$E5$UR>S6q+`tXnGBIcn^Enplp4fQt{i}$kRlWQk=wGCa$eUSA9F^AgCo57p
zae7d)5o106aR%5;JKBvSdTy}|5rS*y5fS3OY$|qLK`Yb^B05E$o7Rjxg)_r2=UC2h
zSAMzF?9xmgq)&R~c>atC!D}sA_`%@wO|0;~0~snbvv`?UuDqR%qb8dn-A~-+wjrCX
z^=r071@;|Hhydn_y2otG6n|PA6m4Og01<cfeCl?+`&^kF&K=0GZ!R0MM}Sb&T|Fh_
zLxXp(Slbn~`GED}U_)^1;%%s|7<IcHD+1|`;v?7?>O*+dL?G@-$w5mC#XQ8uH(Uw)
zKTJpnXz^&_hO1TIcLS#g;iJ8F2I|1g&)N2rw&EMp#p_!8tK*$VtU_llKRS#cChWPr
zEp`=o0=Om}HZ7tNuT<Oxm)lbfNUx~QKbgL{=SwX&jW`kZg^+<hg=CgJU{gyZ#r|`n
z8h!?Ijy*cJ-4$d1t&X;N7ir|TA_8AVflOj9oJeRZO7y~Z(Gh_@Tt?)6Nyozb3Qz$u
zx9b3~C*LN}c5JpyROJEZ$cTWl2+=|Y^Yu*8ymcQ2hkO#S_tIm5v!mde=Qu8Kuo?=F
zaZ(vQ4REf%3a={v9#<}Fp8X`p`+^KXEcWD$8KwQr&m^E|WfYqwAjqE6fCTBWjwbhr
zkmJu+sjheL3Cil1jEsWwzvTNO)g9eHrXoYTE}CYAP}FPL!+zB$m&yCz2*PqJ+tAi7
zI=rUL<r=W*=ytWy=XFOFJP4d{YSn~p59swX)Eds4Mhq>Q3PLYJ&mio0OZwORM_Y5R
zmg3}AIzFn7i?OxWwJZKW+x@}h+$9Z=R2r5W@KCNOU2hP3Tl$Y($Hz>$9A0N^*zPjV
z(!mY6SRJV1kea<IMyQ-9o3qjsA-d|GqqTo>QqtyGgt`ml1@&6$9-4oB$e-ulg})lG
z^1^>bTFJfl89>z`Ymmu8r*9ZexKr%X2U@=wlb^TAft_i>aCMM$xw6ek;;EZndAHh0
zB!4-v`N5Q`OG$2;`X=)HrV3pAyJgC(biOp?YBp+&C*rM3>S&$}IV@STnere*5J}ib
zECDq*2jwX}uOyA&!45IlwJlLsb!&UzkvZ8+iyWrum7A2jDG9{8#N>||`{`}^c>T5y
zhZpU2b3g8&S&#whdZB8^tQLf7o)<Z{{BR3-ir69ggbQ0%w$$9QDS3ZF=+S$pxj$?Y
ztA(nsR!rIlgd$L{of@k}uA`e)-?{F&S>6QPr5prQ;Lcg=S=-ox)ztzkAciTzl471*
zok+NnL3MNbw&~rb=9-(F0=RcM<jFBX|H7p-Th!-yJCIz(%((WvsfcUi#($!09o;f;
zQ2)5It=XVP$<ywWi3VVOd_s=S-oT@b<(8B7<G}!z0#%0i2x&9ENE&}0>y^Y$92Ssh
z=Pe3M-sH6kqpzXS9~t0!Q~Y|@HP%L9Aq6!-f1sChY*0^ZBhuy5Knqj?#8gr{k%uV7
z%Eu%Y(iPgo_^JFOrBW`()BDyc%NL321!jY&5Guo_6NlZujvE6=CLHZ{aLa83+uA{$
z$THtuVN=a$qjeo(RE|((uBzUO?c#9`qdTgDlf}^|iLU6?o%+B&sJFhsNt_D>0(SSM
z;2+XEGoO>?$@T1>#M6#i{iw)J&hlI3i*0WFz0YZCE6B*sID>mDu3@5m{Etm?+t3mi
z7zlb|kcw@Xp=8vXe?-QI6Y`IO&-~gyB6Dg|WG~9F|2#>j97+*^+}H#pcvK>*;XKzP
zXy5RK_-~<f8KZKZV-CxEQbR6lDKX7WCUMW1#1QXZGBln_L;GKoj%g%AM(h?_xQNR{
z^G;vD@$V=!Ni`Vt=8-Kt`B9S64SXroS2Fq&I46rjk0<yjb|Wo$uYL&e8L`{oiTt6b
zZW!jd)1wiFv?h3WANYY2`Z6o}ok99b7BGa>Jzu2sHq9pa?A>PnHKk`6+~A>5hRtt^
z!0CR~u<VoPbS=ilcuDk;{z0ky%Z_f6n;P?}`LHe`u59@<h86x|j_j(?teILH5id(c
zuMUW>O7vu8vp+{x2JUx|IR`~)D^PWduN7j!F$c2d_R-@A+tL<pr2Kgro!$+s4w`@T
z8yuu{51Dd21UR|O<x#;xyGM+uk&|1b`Wjuo$5Cn--Ee1bFYs3!dv~evk6@S0SPjDF
zC}74(3B<)11er;|QPt)Rvy6^f<{M--*&3Zs$01S?yytHzW?7^Y_#$*5nFV+@Fwmoh
zI=DeN^0HR!{EX5H+T{0SwPYbGD~j~A5D`|Ze1W+vx>ES5Q%4dqr@o)~?CXl3&&9T(
zsiB}0FMLPs{J9bcbUigDWxfx>5qt)3CtI^<`(h_*PqM}~pnNc+K@DnG3ZCmHn!uK@
z?cz10b+6;;KQQnnJDXYq{n$-w3PObVJ4RsQ=6ngoqdA6A8gtO$G&5*!C4#t=ZLRWZ
zmV9mC2OlyqYL$I6!|CI~OydzdHfAQ&s)~Ml+-+a(^XSB{l=GX<^3KU!V<<PR7d^$T
zIj+#&zID;^{)ny!gm(A$^<L3Oef!(6JP7qx6PBzEyJ1<XbK|2`2Xxy_*K~sOR_T>H
zbv~0K9$$R8yrc~f*NjDj|9s4?r#IejMJGU#bnE<I;b6fvk&UEJ-F!82y59y=UO9oP
zC|wlt2mkeYpCz@pLi703<&)Gbt_l$cbHnxH^n-qp_e@(+Glp&B#<2lBPFQ1f`DM&0
zuPz|-n`ZEWUp4lS^jeRLg>9@KYgvBJOc$<2`rp#R<-dO(>HUkM8enystycq-C7iJz
zq8iK+o4i@7`Ij4v%bjZaACJ>qmNjHsYJ-&GSou)Zj}I}KYkfg)|8umQ=E%9lo-T6;
z9WI_<XLYt>FOXJoMcet@{UN`bLt4p_`KicFFZ@#)9F~6k$u%GNd#_tAOSg!xHumJc
z%An<k<r7uj60pTo)^k-)*=^fUJ?hLt`*NE){U#oERGd2b7q-lG%E6iWk_3S{M}Adq
z`M%Wwm7Go;Ap1#173_(HuxkSXaR+I$c%X=*C#|m8!ENws1Z0y?4LuqU?2%o706oi<
z*}_?b@1{kmFQ9U=`~g$}sW4YI7`H2pYLIjW=J^tyi~de^4HVs-dU29&W^!u~x;PiZ
zehtaQZ&!blk&*k!eQ$XsCiWf!bT$ug&O;0lu@ZmIsrHAG``a>)0SO{%&*GIW3+j?u
z-9NL1Q4`%9Z|Dw!XAu*P+La~VLuuq;8ZVQqr#rBo5a9-k9@{#-eDE~<7%=Q%Z%ap3
ztE=BlHGSIRb;}Pb#~*va18spVz0FMQT{#;L8B8aa3~SIS)8MZpN#W!}Hk_+33^@mb
z@51(ws~p-pY>Vsm<eXk^H~3RQw^)o$bO;c5$0e(@%Ln*3qT++ls#c$li_rab*ZqXO
zzR7j;6<p^}(jfvkCNV~Uw(xt0Octm>k~-ACrju@dIN!wj@t!Z&z{7qh3?8X*%L@0{
zYaY$#6@vQz{ufjtBnW%68gucbhF-S8dA`Ge#bvA^e(fPJT0kFUDoVQ-u;ZLpnzQEG
zXJoWW%GU0Qm_#4?^cyPzaZ%vq&0F*4?@pbz2B4|M`!Cv64`h9Nak!PT0g}`EA9!kj
z?k)pd!&MeIGiC5psGJ|6I;i*f_G1MN6oV3GGdJw2COoxEW7U5;?TTR7Jg9KJP;Z(t
zSE7@qfK_&<`BsMI8*cW^1gBCTDSEEb<Mm$A2_%)|ZiF>oa(IP<<2efPaJ^?j-{&iu
zN<Aj@6-5{B6DA<+YB5h?vmeTMFs|aZ>G2^dR;B)*E9pdXKz-=U(QpaVBoOz)HwZ-{
z$1Uuage3SlYS@Ff<<q!AqbzKI`^4EB1t4#|8d!?{2nk3ISXB!FM37V1R}-8rj^C>M
znw;~>8MD;76u*DQ2W_tBYL&)t`>SFx@>siRoCau^|NRoKf3>)JGyk`Hr(0KAo_~lc
z%}l-!%<4g*8mF_Q$^gZQ5%6m{2+~o*etz33u?gcGCILS?rmO-uvc7zN_Mi#JAs>0_
zZcA9SFWSFYs!wA!z9vgK!8p9uzc4<*D02|=C}z*-jBl^4mT`B`TA|dey5r+FUGO)m
zQ6WHZR?q)-hZ~%=n5z?j=Zh4xeU2+AEfPf_tZSn$pJrlJ#lxsu+cJ&yq<h6bfO|KD
z27mxp!YBrxUPZ-|oBz=^b=^;vP`5+)5QW2O-anUI>g$MbnMP}eJRl&w8@eh4c}!X2
z-caFy(I$=25e2;r9py@f9VIftdAf(32`*`?Il-NczI$}7ed_#mbQ4)x$jxZ;R6Ovu
zVY(L6{;ht#6+QA%6B`gMyOz+YqHX0Hd;bp9mritngf6>Y?)Jb;N+~d4aGy&T>i|>h
zh(0RIaU!w3z{_ZBUpP50g|f^vUxW91qC41A^WVn)vDLKj)nBjcYW<bvOf~<Ce?rS;
zK@4}!<AN&jI(ly375!>P7DiPXh@8!WR9k)6jtvY8{(9hukM;$6%{$F2{f9jTFq;^1
zO<R(I{~X85Rity=q=Q<0^X{2|tk)cxq1f~LS>0#N^(EVi0>NLOF9a?J0e9BSP;a@+
zD9_kJ1yR(Nk0kQ=2+A5{@&EG2D}cC@Uzn6R<pT)ZV>aw3kCve_=<f*<U@1Yxb#r8a
z8q3`0f7};0BYr98GjEknzxl^b>|zIRv`lYrKeIx$*r&#d@mo}dgPo6GV)E&m&zayX
zkroNag(LNDMNWj!Icsm{fSy*(_PvM~<!AGOUdG{;zUN$BhtQ)-c+V#q>UJ@7KU7xr
zdVtgHm-N=u`lD66^ZyYoICYmh`lNoiH~j<6?uuxi1yqCJ_!A%&pq-*|O4LNyjcPbQ
zd4Bc1d9#I_WBMor`3tb%5_;-RJU}LpeB?z0aZjn_p7>{I#1##*?0h0<WrsobvUp}G
zD-EDy28wvInIr5r$)UrY-S=NTQ90e87}EfmU4bsxjI+X-(0zW9q%%~7b2{2{<ln|!
zB(G?k=a6C0CZ<P&VKN^)1fDUMi-<_#IBJtw`BPc|qSeY);#_23OcH^Vp%!T<kkJCG
z4y2x@KP?U?J_tf?tY#@Y7_7=tH)LpE51lYwN&br}PQxJ&a?r?|gu~&dXK;@uuKIg1
zA7m@r>tD!*avMDPLq=2es)m*7Q=zHcSJS56lpw3i1f2cjoeE?#1JBeSMC73BFk&A@
z`4d=Gyw2AM;K9TR=qnKl8CGh*?iYz#QTWzf+Lnm2{%8TDBk3QlOx}s&G7*5ZQh!b}
z!hh5kMI!=5z#A%icA4EX2`x|3>m7D{LR5!`b3aG(fmMlNM7BjYB|ANMqB0pg!r)|i
zaK@0YyYvnWC7v8Td<X&j=-h}N?h?;8oCyrga=Z~DFd-GGBNhIJH%0jZVJiP$4r%(I
zYq6@EKMkt_y%}XwpFRp;pRxqimMDI6+9C1oA{6tIoaZ|et(LPw_ZHTmgGA%gyl4Mu
z$zTcEIVc}3H!hId1%&#uSxn}iX9KqFv`7JfqCinb@?|*sN<N5N1E6n#m=W&i2>dKi
zM<2TT@9(QDifm+$9)f9<7YT<SDbu0``br78l%ss&Sh&=o^}3Z)^Y@nuhey6Vf1G+N
zC@j-x;alA$XeocS^}lcHu;inyu-~34pTIOaclG7qAgcnfD(_ohScMoC+5kVaind`l
ztE`a$W#oqDU-{tdprXhY9M5ZYIBhEI=wJD17<>HZ%pbdKDNGI^m5+Qq{#BJm_1&FP
z8EpaBHIaT-0xz^@AeaG>{(J#*juZPU2O^CbEsuS0>UH8z__Q2rc@}s8?0uH*q%~8A
zS7!o)&14SfQZdY2*2D&f_OrsZUcG4D#SL9J>ISjEmP@iWkRto2aq6q8IrvY8fHq?k
zEu@w4U93?_M%s|BWx(S7C$tr#095T0m<e=`%0UbCr6;fVv-p7iGJp^3Yl8S8*3a2d
z8fc6<6fF#f7jsgtTJ4g^U~l>qAxN)$IA&}VOspL^iLo-Eutw5cZ~KWUsT~#3G4M<g
z;E2l0*V)~q<GibFKbT{LHmc{SF-Aa(zcbIvKiKc<oS?p}H_rrAp^7v^DioPJM3XEY
zka`M{p&@o~!&E;AYE777{^o_i$gzVtCpA>%qZrQfo{3mj>09iWR$w4vWZHTt<Sc5Z
zu9|mgFeLax;$<K&-!cE-ql2J1^%`Icu>LC4dj=29Wxj3_-y^I7dv_uN30}<~e9wIt
zyCSF{CK2U=`jFSV6ww@?>@%zY+>*g_h=$iwbdxgg;SOzsEy{0yVHBX}Aps{bv=}DV
z$wnMdFtqoDwUJ^Y8f}x&a+^h0&7HLSCUHRxh8Aw~`ICd)>B0yD1S?U;WUkb>4%)b6
zFs&R~;gIfk-bwJHbeIF(B$G_Yx2=9exaP-Ia|TH29D0%|l`nv2OzLDRjQZ|YH|W<{
zc!pI3pjM;YvHy*J;u_2s55|94r%DMgRZY$|+J<*Oit4aKc202OYzG?<eY<8KTojGU
zwvT`jvD1~)*p!cay{@sBJrU&3bT22}3!r6Q6e;f^^9!8%D}7h3l>w|FRniOOSW}<|
z_`k&s#Y#6`W-lwPpzglStR$8QuFz}+^KBiJtc+b)L<Tqv&(h+zMZw=&)H7}nhy+l6
z*K&YM`*kk;XLFY!rrf$_nPjvT0%3UhGZyo8aQ6G3$<Mrl%O*rptW^A4Bv{~5hIuBq
zAYG7-ikr)X;J%cJu&RQz)e%BY-cYnm&E4H(Iedgf0G>SovPRM?-SCGH5Y8ACj(AvH
z-)P_&*Aofc|9@>CxzY&;RlLs4cl+mdFpWWeGUM;^KwQaSEJWj8obQqf7r8wzL?GYW
zj|d`C#=-OfBuZ#NzJn4JZ|TnJ`AX|QbK4y`wT9_bABDp3Fwm(VmDpsw6XBmrqk`NF
z{aaREhpq?o<eZ>o9kLp06oBzOKmp8;LyS<eOjpR>pwgbq1FyT?E0mu(rqe--dkdA!
zfVo-7gZB6zttZ0OqeTZN1bN=4?2M!3X0yxJfd*bcIr`kZ7X<8hosuHh(&f=I1~j-n
z0#rZL&6Pg^71nfJ;PHL%Nimn4ImK&a05(9ZA+T&y%k~D`P?3U$%Q4Rta^Xga_qthT
z4YGW-fsG+DCbS+Nb>6m42K=fM6YK_AL-`cVw3g>(EG@iukNF5gkqFVVjbAwe@|kU9
zeV)K&f)!r^6^^T#{QhVhj;#oQZ21%)S#LWKYBqvQ_N?+UjA3UQE!XAqXSU}0ZM37r
z=DkNV`7(Oz9PqhEp)-3J!m{9q5b&y>7K57mAxftDl6Y(7fJF#>S{YGy(*J9Q=!@{o
z!=|s<y{EkPu(2V~UtkNB1!QZd4DTLQH1`<;4N$BLy|1Wy1Ym!^TdMRC7>{`d)S6BH
zb<?AP`vFEgdZwwnP(E<3`JVv<JE)9SwzKYubrZoqW%Ca`mi32mKI@!5Bb6|`H`YM*
zjG~1z@G`wnC6Ks!fvkay6VnVvunC}UR1K_b{=bO7oeI7FjgRm*(VOFi+HCYwe3qW(
zGX&)1E}QQ_uMWEG7GwCy5Fiq+ROD#k?)ezGT<(z%i><iaiCkeym?9q9M+M-Ag6zDm
zj+@9K`-ywSeNR3&xEckMQ)PuiGgO3Pb@vK07lLpTEB*2<J08Y~m78S+9v53IhH~b=
zLXrmkOyjfnNBxRG`eC#4zt!qce%saP?JKb_AtMTZgK06tdIXLGGUy=amgJaJLK?mm
zpuLaziUxg!DS%m+K%)5=jampyf)T9}+bRKpdZ_<5Pt6_Kwx#&~atlP|lUAmI$7Y8=
zrEIQmqS5a$4ubHac=ucXysPp8g%s>DquNy12S-nT$W7-&v443<9W%Ho^uL@TY-GOc
zs_s4f1Yu=k3mYR}i8u<_nJW}VVB)t|bO7nl=rcVJu}pGBfCdn<AxQwnhhsV5Y)QH#
zjiVu$wEs8ZPh?LUfHMnRHW}cvL3Z~0Jgy3|402@lwsMyPZUxE}W%@#YDixq)CF)kH
zPz8SBc3_Ccq{wXIMxap)1(anT6G?27i1X?7wjB<bOqGM`sN62uAHAO>D~`Lp#8a|{
zM!}(oDN;j&_1VuXh%xW$Jrqv*&S}HlYcD>CH=TEmeOKV$32+0z>Nx!setT*xHvai`
zSNhMA8Eco7U_~hSl28^<fud`Zeu!Em&j1$rP-qqVsW3`8;HxLjj^zE)DEJIWw&UN1
z9qG1ad7zNtJ1J+pOO_ls!;FliLbcHvF^*q^8NX5lxGqWv5e~p%+NK3u_#cVEwPuWV
zdbM}Gj||C&-X(GI*PS;_S-+Bpqn1CJY8&D?!PtY&z1DADY8e%V$h|vsi}Db$srRP?
z0dx#FsVc)8LN1!b_EN+#$7E~07kAK%ReCqkuKzqe7vo3X=ee9y_zwV;d9C%*W|vLi
zL+>jIxlFKhm<U5XGu}!wJtNiD{SgaUAZALIpXLj=T=>g*(+kWU?Xf_vvG(Jh<a1a{
zn*-bQsIVg(MA}xiho!Tt%>U(6^NGk9zc;LEk^sS14jjY~4d&Ztv_N{~AA-gPl~d3=
zCy#WE^bhVH7%~Y+&`YW}2beRUZ1TTcE9_0uNFnOxo}$<LwS*WE0C2oN)SCYPihJ^S
zDA%^18Dxl<M%f}!v{<s1WritBj<qO4S&|`RO{r`{lF5?dwaXf^GxlO6QCY^4tZ9*D
z4igy-$Ee<$)bV}4-}%1xzjyvK^IZFKU)Octw?}2S`LOHB<#-tUeOd{Y$<-i0Msypk
z`=hrOMBfEl<k^J7T_cz7MCe(#69gShwt{J}AC~Y>AFlpOu@^~`BW#0qv(70lI;>VS
zX6)n0kj9JlR~VOC7@*tv`y)*cGy}oPT98c{n^FOF+kE89g5pE^Lh&r-Dpl}7mC0fR
zQ(A9id8?h6`7&@CAX7PnR*KMvE@tGD?3ENE$Nt1tRb}6^fLqcKV$Mz)zn3ZW^{Jp>
zoSiRQQ&;X8kQ|6mH3q%UaSesn%AMA|1cmOl>C=v~ZXWpq4fk{dIqV802pPnCzoUQj
zW$CK2#@RRD*#`SNRA8ZVig%zJE+D3|6e;<4Z}%-UhXg&GP~0jskpOk1;bE)~kN%w>
z`Atf%Mo3lUA?$n=+1cik8*|VcVrtn@dP9`4FuvHw)%vl%bVJB*1Pngxj{i364VEVk
z-}0$1dr8|f?j(W9X_}PgDX^I3-IY)K=70P%%by<pOF9oY-XFGx%vk=;$CI*z0i!c+
zV58s?0o}QaugQAFMSU+}T1hd`Ni$E_L8IUNLM?Si6Jxt(QL7d0u|i^$j`g^kGmTCj
zWJ)t+HM({-<-vsAqsSC)9r+_IJ}h2KbqbSVZ?JVLlese%&qqBkzsJOS=I0TlQB5gj
zbdA{4))w1oKiEzCJFqZmb<Y9HnpJupU66LtBrAr#;xw449LYj9y)n$w{o_EMD*VMF
z|DGo;4R>-9rBMV``jOR|5}fyd5aE*3HqNHeoQB!^P?ni<k)7t0kPNE*kGcJ+9+PR1
zY#G~O#Xj56`D5@CyklsKyF6@&BDTSS4jtI^nqIs<a%8wV3>j<@)jSW2E23X0zH(=!
zgv-gOXE@Qi_1#pEPV%jRLHM%7we&-=b(H;a<3AYWT3Zzq)5graa@hA`eDpTQJV*$N
zj7=zF8GN@kxw{RYcIdDNYFko|*!gBUj2=RF<E@RBA#}ood&NBWD<O78@l_igH+fY2
z6=F{Y4zpki1#my${3IhG54#lhp+aof<0TtRX?)<vE9|dw<dz#a8k&EdrRk|8NS_*f
zP0r)`3{vU6#Ds6>gmqEbXHrxmvcJXK@M^Kk5-x293)Q>HiDy0f`X_J8L(Ub?SIWUY
z%vU)|jckT`9kh*Scq6h8;8Su%Vfc@2%HRg61|MPQt$_I@cdDC4u!p3HH`9N?2;%XC
zrSet&s?G;hI?;VYHP9~mwH;eMn6#_oYud?n=k{#{R$<nkq~5X4@8@s2zL}k7vqINH
zT32G%S|QFR8|BI&85uqLTnq&+@RZH-O82kT*dCO!*UZkJ;_{5s2R*4ATMU;s7P?Re
zao+spxSS)o%7jxB4_V&}0CXTVN!E!GYKjYgj6qSGAL4ErS-5|SFmAXWD2bm!APj51
zkCK+(oDly;?3B3;!wh2*@GM84{)|F>S;Tt-LP6f21|9o03O7KmK@Xa2^dtg?z=|4c
z!wX)}eY}f(ntzi*L?Fq75g1zJ*8y;BHU@IIzZJeL^1ik(#I-pVYF?`>fhm#iMQEJh
z7VBQZ8aHR4)RQcB=9@(K*^*dv4m1+M>r}L;!QJ`6Qg^esHqz4|C~$I}Ex@WE25Z^N
z&uS;@A9(H%*{ltWqSi7ruww~84SdJpJ_=rTjTO98u}Si0^~E{jwBJ0n@HcjFxT%)H
zvNpV1tIfc#Bd&^zhrosSHQo=w5gP@zpv&6Ha)1;N1!E^IcsN+bLAy{7{s7}3z$o{+
zDt&%-DB#AH(0VGQ{$z<2^m1T08UyumuY&=}o!2qt0JNYP;{_$tx|&ZMSqB5uv(_7_
z825FsV<#FQmAjYD<_#1+{NC=f8{w$u`ac+o@rLnTQHj(I>+wr_r|_0NWT<BtkOmr)
z{E0Bf2Ks#fjeN15hn19F1$YtIu4TAEp$eK5{QAYi>b&KAX<*W4mOZo1Za^A<1@)81
zQ!<2}*6dlF$A$a<^gE{2he7AgjL@^nWr9n}RwFS`$R1cr=(=4(&9`PGeNkkpzMU?y
z&1FK_KBV2k#p7eX7Z1;zX&v=Ez8X*>(}o#n8TBzzWz?yB##lUmhnpoUe#Q*2e@nAw
z;c?ln)e!(f?S&5Pwfs&}`a40|SD+0${6X>XNM8TbouaTr>uSEXQSXuXME-H_S(1-I
z2Y1b!9h0ju7%_^|uV1PI2f8IrZe+@J;5hd$7w9Wrg##?MkX!;fxQhea?>we(%O4rt
z{$p-bz&CwjncVpTGwv=h_~0ona3=>rtRl6QD7;#-0`$nM4aR|D;rTxDlV$0BQOq@z
zV&Sv5ShaXhQeEf&sPhnfwz%m#K{t(;fehu`=>#f9i+5-%z?QW@>S1mBFz}|R3QIOz
zf`z;|e*$7fCVg(rICa^;W1PcZ@p><uqCn-wI;!u5Lc%+K?N8#vj>}wlf=gNq6RQx}
z4q>%%{oPJAmAg8p-j2Xy;$M0o#M}44cv-QZ4gff@aE&##SnuV<K!^~rg}5T5R#F6b
z6RuTSkPp-FEx)Bwj!x&BZ-M$V7~I%EfpMPwp}u+($CI6p`|2nb+$rR~8p@|*Z2_k;
z#~UM=9Xnxh4T)FrHw1q;Gm`4%=Feu@R~B95i!>U?v<bAz6mN|hruWvGIJ)ZHOh>p3
zPmO;h#<Y?ZUAn!}&8KVSW-ufj=%QS%NNiL#_muX2obp_F(W8r8Ru|(N&za7f4Kq~9
z8S~i=m!Cb;P3f(q#me`VnS?6QNC{trOK+5;1DJ*feMY5FYM<MmP(VKz=`dWrT&Cyt
zOcnXcx>H>o96}tT4?1TD0Yn|wb_C%|B8G^4l_ri(ScyRK9M|;D$WG#b%c^?p7Lg_W
zThUB~IYk3j-t5Gy(z?p{!LlaIPdqG<zC)Djy7WpKDevin;7<3}DVLGF=fjdlV-r%x
zk{NYX8RbmLfFRsL#yI9tx0mf?0GNk3@byI-9cOCq_W)uox?)nk**Gidk6_42Pjj=}
zJqT1PU?t*vtFP@mYL5}q#$nO(9r!o<hN3{E9UCGMJmhJ7?^6*y1=ppb$}J`K7Bfdz
z<9fXOgqBqM1>A?%sQ*>m6Xf5fR>Qr5j#^hp9t5DvfI>DyN9=YLBK6f%g59=8`IX=C
zuV>U%8b6llBF5wf7%Om%$LEHfJSxATQ4xTk@_!+?EhG2ZZrsdebm}cPcXr=h>IjZ1
za*J*>rG26gqbKy0d{F*~hU;s~m@$UoP9VZB99n7-F6vUpgbJ^(A7fJPWGZzUdF<r-
ze!n`P(ea6cq%p84{ce)Z?37)k(h#zc)aR8B2j1}v1I9j7>0DYW!liJs;cAAx#uM=d
z<|mB{T!sS}eo$DP3}O&tCdrpo#^z?j`-iBtCLBJadSkeFgjm}1$~4rTbNx>&loZuE
zNVa*n^qI!Q=eGVXKE?wWoPg}s-g5^6Zh5scL@LzOMV4NMMl&nctRK_mFwBkm((UqS
zvY|Zfi*`9(63w)+e?8Q7-mr~%e{q)nSS3<Yw^+q(XP!&CFgUIfG-Z?B9(^_~Et=^F
zLStwtX&?LqPTmbD#!PvR`0<(L4Mtr<<`)(}`4<9dlLZruY|rHHUAnDHv(FD-SpG-I
zZLKCF_}*#?e^mTs!^<<4H}???asA%RHGQ`UMg=N$cFOT9Ez7wv4P}2v0UoI(6yYLe
z99n5kmo2_GEsl>Z2}pk4z?JP@D5OgheDF4BS#www*-N7IPH|;p+u_+ymw+j~GsiJW
zKRq`B>a{df_Y%)}n7pJJ<|Cf-DC8P7$NlTfD5mHHDdqKdJz&my#$U7s?mvBm%O#8d
zj>39M5fZ|J{IZt5u)NnfL2g5coBVSE;sYs*SNj3TpNNnbX8#vZ4Dt+IBZWGr#Y+u~
zX5w$M#voHsA4*Hfltqg@;g#67fNCuK1)_`Sd1N+$T6&Dv<>tm?)pE9t96`NAtxqVQ
zQJnBzLgX<i_`SOz2g$@h3V+NKP3xO+I$v68!Q`uNCkNk!i=E;TSsDThxCGuVAd^P;
za@1O3dNrURb@;PIW21tha$EYDzXa#vPunf5obhY!lB%){^+;feoH`^4Nl(t;Y%O60
zspwXv0fw4;%}v?#`(fnUVp&=n96msOxczK-hLflK>l@<XjvIMs<au?FeiP)3XvA(Z
zr|mmj&%M>>-9Fz=vm&_wBcRNXz?)|6`Ch!m)VIYIT>$55TKZR}$HU{IHv*D;+Yfs3
zr5;_KIiL&`qzt1nx4%wboj|K&Z{5u7XDX^$5YPj86CUE>ho@^PDkmU8EE=hXo#0r$
z-1DVOShVY3nG2^~E3K_+R1->{@higu2Oc2i#Tse2w{SF3ZkWYC<(1|wG4bOY+`l50
zIpd~VEeu&|+gnhXA9sC=f0Nc_!}y3MELBN0dp%sfD%wQFx8S!l%^YB=BeJA{p51_+
zL%hW`*ya-&CM|16TTmair^^%`bq$0i7NC(<q?(x~OwD)bn(~eyI5tjD?K26~Qo5$b
zp(URAq}(mmrv#H%@Mqbq7Wm04X$W6+KRc7=Ne4A#I6ip8$Z@C-PQ3ZfABmndl=#zw
z*%g8l+5)ZTyNOzk+#6QUNJUJAOTpv7%qN2q-xX{u1uq}}<u|X0o`-D2O7D7!Yv=**
zp#9DDC=0vV<xvBJnxu>B$g%fxb2+Hs7gv8NQx#d%Hpt<x)AQi!QJY1SJ;LvCm0B7?
zyZV>HJ`+v3gYcT{jK5fI@yZ#^$4FdK&D94z&v1MX6V=*G>azGK`;M_>Qil?`2vS$R
zeq$j9wd|ibEfsPd)wB;4w96N5zM1w@Rivb<emO?%b@Z*7mN$Bu$oN^0HdC&wP5gmp
zz?Rai_jLQR!xf4dW0>CI#L!*_IO_^VT>K#O0Q-HLxMlmkQhv10Hdw58o5<2c*yd6J
zRJtnkE(uQaPvI|ErfW?=9N}|C_s^zLJ+yfr-Hv8fV*mdj-c)-0=aK(k0gC&<US!5`
UyNrjY76KoA9b@g1V|F3`2Hqispa1{>

literal 104955
zcmdqIWmHsA*fxA-7+~n`ZjhGlp+iIvm5}aEX@sG>k!}=Fx<L>|kdO|gB?UoR5Rm+i
zKCzznd%y4R`{QLT)|px7?7i>nx^tg>PmH#fDjqg9HUI#4>S{`l0RRL12t~lrz)K*9
zF*^XD0os~Rl)<kM2m}g+qM@O|U@&xabPNm(OiWBT9FB#Bg^i7kgM)*Mi;IVchmVg>
zKtMo9NJvCPL`+OfLPA1HN=imXMovynK|w)DNl8UTMNLgjLqkJLOG`&bM^8`Bz`(%B
z$jHRR#LUdh!otGJ%F4#Z#?H>p!NI}F$;rjVb?@FiZf<TK9v)s^UOqlPetv!d0Rcfl
z!Ta~`3keAc3k!>gh=_`ciiwGdi;GK0NJvUbN=ZpcOH0eh$jHjd%E`$g5D0mBc?AUp
zMMXs=B_(BLWfc__RaI3rH8pj0bqx)T2M-=-YHB`w_)tqr>(Qe}+S=MWIy#RZKi1XN
zee&eV)2C0LJ$t67r>C#4Z(v|xXlVHS`Ew&9BV%J@6B83tQ&TfDGjnru3kwTNOG_&&
zD{E_O8yg#2TU$FjJ9~S32L}g7M@J_oCue787Z(>-S64STH+Oe;4-XGdPtO-GUU+$V
zd3$^N`1ttx`uh3#`TP3^1Ox;I2EKgxGAJl0I5-%IM23WfgocKOg@uKOhet$2L`FtN
zMMXtNN5{m(#Ky+P#l^+P$0sBtBqk;%B_$;%C#R&Oq^72(rKP2(r@wmjDkCE!Gcz+Q
zD=RxYJ0~Y6H#avgFE2kozo4L?u&}VGsHnKOxTK_{w6yf~>(_7IyeTUyD=#mvsHk}R
z_U*fO?<y-RtE#H1tE+2jYHDk1>+0(2>+2gD8X6lLo0^)Mo10r&T3TCM+uGXN+uPs2
zf8Wv3(b?Jg;lqcnuCDIxZWIdj@#Dvyo}S*`-cO%C_4W1j_xBGB3=9qqe*XM<XlQ77
zcz9%FWOQ_NY;0_Ne0*YJVsdhFYHDhFdiu+kFJHfYotc@Lot>SVo134XUszaJTwGjQ
zTKe|w+w$`A%F4>>>gw9s+WPwX#>U3?@835!H@CL7e*E~cy}iA&v-9)k&)wbKy}iBt
z{r!W3gTuqaqobqa<Kthyex010oSvTk{{8#x?Ckvf{Nm!`&!0b+mzP&pSJ&6qH#ax8
zx3_n9ck6Y=FF_x`a#J(#1OU3IzyFY=&0}i-U<A~a<e&KF?HOZck&4jV9&A5;{Qy(v
zJ^UWyR7WTe-Oj_<7dzM`NbF}YbZ+K3ohPIpBe6wK_3BGG$)8+Wgp{thpZvH=9bs)0
zk_y<R!*nt!h*}&?AJH1h4J6|M{S7TF9A<Tn2D%O<r4<1#<nNW0^Z&MMLkRE_isBzS
z581mKrfLfkM!n9`(nM>^y_v^6vQsG-M=^4vF3mHyZAqiDvkh6kjlT#E;~Zp%w)Ot;
z@2n$$T>ts}!z42}9Y$hNIY@~7$Q7=e2Lwjue#}mAVn$w$&5EI-02>AOq$$g0GoyGi
zDyV>Xgj*~L^4v{Gyf=jxmW;V7cD3s^m`eoBLbJ4}RwG1Q9(;=ux=zNx$c_i+(C_GG
zLWV1~5F&b0-7>E;o<SvTk-$`@bOF7iawvd{tA%bw0{Mik2w=rHF(KJ!QEz@N`l}hz
z<3L{TAGaH8<d`Z)0=QaCb}+Z=<N-<cq452c;f)T{@79q3G4Z@2unPoHtn@3Cwz5HJ
zI~&a>p%jTrp@phY^J44GaIq?ipWg%F5AJB6%LIq~v<!YdP0@=!^Xnea_nc3P>7a~`
z0_LWTEr0i+-EY3p!Tain^+pQ~jVu+wM&+cNqci!8RtA%(n(Vu26=<*oHE}q3(fa=O
z*!P{d**jqPec!gNeO#z_D_-S{5aHmSHQxO?GvZGw(8V(lU_K>1tPI4>bFd;a<9<yU
zebJep$m;7>Jh^!xcGDII$nr32C5}daK{4vC5u!Z3{8UF8iMq?RwpM<9V1*-SFzG)S
z$<e=%%LPhI$_SCW(n3ySc2)!s{od94x7I8u*HGZBp<rfr?MGPwo*NB)YQR*U0Jmzk
z-}qrul^)=cq~8*#<br%kj-L>yu6>xn!I4g`IP7J_fiz}gy1TG7ZLj7*RdU>^(y&U?
zLMw1Z=r63CZIPbGR{}`pQij4=%3qu8P`8AzXp;ka4#;^LjPOM^txL_xNS#Du%0t)@
z2J4NwDZ8S~145MepgK*+$~*Zk=L`)%;`v=1!_+%Pcf*Z9eNu<AwF!VSXQ=D24YnGb
zGT;T($F9RCxJ{*r%=@z9Xohry1BSFHf^UONzdEccT3)@B5=`{6Wd&x%-OhhSyA!*~
zsRNMA`QA~l0xD5k(S{h{aa<4d(t~e_55C&EB#>Qrgpa}l*ZsUQ8MO(-T~M7s3SJlM
ztckz8<HBr8lCB{E<r^h45yD*YxF1v&5`qy0csP}X&O(-(s_22jiP_T}Ve^iGkd92%
z_%!5I%mKq+pa=g2)C@1A|H6+1^-LZ(N4w@4tAO(IyaIB5t*>e8>kF5YB2$Eg4~X<m
zBLjDcKzLa=M}wa){4g*I+~A1QABT|Tx~j($DtUF`>`v$ZjcM{Gj4!Qs^yPnGiUUl-
zUhnUn;8s{-L9+nsO{)C2{E{$%Xo}YG$Y1esMZ9<k1fR|WAl@ppvskO;*T8R=n<G2&
zJ4rM&S8<p?I)&56>~mnQB}Z;b^?xpc#@85?d-d0JoUAAywEVc|_2XYxZPWoedHIGn
z#K`T*&EnQhRlN#LMMb`6CVe{VvIh-NLm&EV=_Wt7JM}-=7#P&F(6$hYiIFRGl;wIj
z%m0fiw``du=COZ1;7x=SA@nI)fdukn6?H9~Fg|5NKK~Q3S)A0V`iN6e(X!*>7Gp^8
z+rYOVK-HAi2>Z}@>s#6{DLmXK;xI;XVc`=O41M--ah#;K|EY}su|f+GsxrBAItNXJ
zM;*{pAJ&e6L+y6*bO>@tAZmx{eU!$7FP9zy(S;dJn4DewC&8~m<X`xbpi)?$TQz1r
z+dFVyJ=i4LU|kI=xxzHEA!v<0f!oaGy&V*ViW|bopHuWCWUwM>)@XM!3Nj*PlZapf
zwAc)5h4>VqYr^ygWO}&aOB~FJyRbUvMbqn5W@Ki=r5H73+uB3m8PyX#qU92-MH&31
z57HKSu%$;Y?uv%LyLlGcMvj8<lm)1i6JEhSdc}sdYAwaA-BvV=`-Bi>i4qU1OO_O&
zq5e@eb#CfWjQIQ#^wEZi6o%`MDXG+ydBPWj2nLsL;}enp{|1x<m_&ZZa+3{Ohb5aZ
z#JZqr2NaS|8zW>I;Q6j32nP{P`IlE47TO}OItJrEF&cExK&$?G1_J&o=s|0H0Vx3n
z1_y;Ktd0%yOA*Y^$8CW3f11(%wACaaku=<@4POrH(Q1Nvvg2x)Bl(jbiT7rl)ckik
z^06WMmB6pWmzO(&ei!IYcHhGyTWhln$&11U-17kw^g*G6m{|sAf(g;B=qWHspff=0
zBB7)mXFC!5jF0dn*QqgyJ}k*-K@T{dn#pH=UfYr!$ITU@nO6nc^diQO!f-Lpy$#F*
z`*nk=`^RyfLhH!Cf43#=)0Pt#HsK$^xn%eye()G-n#rn3Lwy_25cDo21Wyhs&dkGq
zBD<2yFgYM^#qW~<`uV&lA>{A;vVZ5No>v3-p-+kkyv4V>%fIqHIa>d?5<~Li*@*5L
zR{~JQTUxm_ZtBd_8fU4vv6|@q>V(w!_Vo|<R<a3-8@_vnMc%W`2A8|IxSeaNKt^+e
z;^!r6%phAV=)^#XD*x<&I_id&r42d!$t{jD5;VEbgBZ|3;nj)316u)sLBlK)m*fZm
zn3<W|z1~xgFrIKG04+?Hsw|Vnis^6Fk~bvqVTlpP9F9V*K8f7f#4w1S&IU|ernl6*
z<-$VbXL|mQ`n017fO`xKj5`RJGr>P%(*IM}aze=KH~G;gL#M*JKe+U4FM}(DZle-9
z#`b#cXBZlaof}HWNbZWqPNyrzbfbW&-nHQM&Qm+_F|mQ4*J5LmN#PrTn;h%4CvM_n
zIMY3AH^(#W{9`z}y=!RhCjQ&C;{4L|_~oDc>|54eIOh89UK2Qb9?6|X`8K|Nb4~Xn
zV&GOW|M(GdP0`)N0nRuki3fq92oX-=-xm4_5icmA(Ik-o)Ev$TL;@CJOh^PJRD50l
zMhI5~AmD`t7Xs@cM179@cL$DSLe81N83`i+Ezy5>5P%joq;uLc)5Oi{^F+aHt=h>2
zF^*Piiy(}UE)rM||BrDfbtdGlq9i1IAWGpzv3LiQx$9H-bhR@sL@A01DGn|VP6>4*
zhObUcCC33a;WY%$rBtHGl>%e1%AYTSQsOHD?mdJEMsQWeVJx$6U+9@AAlHOH!@50v
zGPMGXO)YwNW<(=_l3qeY-hawJ=mFKAsx|G_l7#T!jh1&$D^~u0J}1s4i42m2jKHhQ
zySa7tN?*p{{qKuP!Oh&ZpI>&!3ith1R$kDNqqVtNJ~w+o3m@=XW17L-=zJ0E{iU7%
zR`PtnZ>KxB&u{IcZuQuPPCz4wa90>ch*V-i78E<5o;Oa|%~Fs#<%kIk;wu$C#e;k@
zC}%<ftk{qjSo_7@xr&O$rcWrL0)y-za6|AG64Y877*ru=CBdP)ac5STH_62Kcl!YX
zlDBx^@*x+y*Br}|=Z(j7uK8E%okxgkuUqir-I~wsjoz~_eyk3zqA)_+e30a1PMDkX
zw$xuB1Hais^IV)Q(@~S=>8OA7-p>afi_+p%O%oz<_@G&2{)M+)q=tK!#55Kh$cNqS
z*dYJ2_Haf?Tu|sARPw;XB3m<XTBGeBXJ~Cnw`17VULbvfBq$-b8PN})2oY}J-G55)
zD!_udNmC7B&e+3I=oFAZ8tUl9ikolm8D9Ly2qmP5e+(0H6$JawDt4tw;Hy5!)DSOf
zsM|+&q)27!dLdV_)O1zpwBor6Xd2)$zR0Kl`ydizBO1u~YUo4Yn<D>eVg8Nchca&h
zV^7BQ2b$5@KtHtV%Y**~Nj3uDhkVEShuZ~OcWWrzwmb300_I?9AwL2WpkB5A@ae1Q
z+WBKk@R7e(nCC<Zg)7;Rpl$$l`EJlM$yscmlvqb?CMB&84VB#_7%u+9T>tw)O+cLd
z38>5}WpG01s?G-`*IXIkp~e3)VI&r#mv;5DLW8cf5p5MfJ}rp}$)f@ez+GkfJHQyo
z?AHe;`7bw(-iEnt_eQq`A2v|QTK&<wzdyfL^ocDJXrKhi`3juxu=y|U!P+?cCBqhM
zywW(X@u{GH-T0r+O=1-Z;<`~(TL@iSXB@KXX42iVRA?yz260(fW`rmi|CnXbMTkS6
zzODA+7i3RR6v!DEJpTDtFjQRfYRppIt)}CGOSIS23|yRTeyv&_)YhQkCr$8O&%wll
z)0RJIMGo;^7|jpSgExTx2r3-lG(Cj?J%&DpTVlrrk!Y`8-8Fmm#^=2wpZ^|y18n^S
zWxvKx;RYCB%@6&6Zpu{9e!=jg1yUa_2U&s%K#Gbwe3^3B;`D>kMhf>hov`15>X~7-
zfo`J-Jm8;Pw4p)FQ|`ux0M|JvLS$v^*Y0if{|rkPBq&zo7$C9KcQS`tIQ=Wvs!JI<
zNBLhWn!s7b_{+GdC47&G@Qdgs;C(@RA68_EP5QMG3^4x*LIp4ywDI(DvlTv#>IH=u
z@O|-88#MU_atO?=SArn!kqo44X8_~@{rYJu9|*(%S@;rE!o=jqY84@H9xWm0cENe}
z1>XoUbQ?z-&V0U~<{rDUFZQ{rx3m`XYc?A!S+@8q`P~CNi1;9$NMHE@K=Rwtf6)nj
zkijy4a)JHj$GlQ&pzF7U6C9x$fMjS64QEBNJcvB&3{uW{*7EXpo8FS5@8JCle<G9%
z^8K+fIOPNuJotUADx~$~xWLM!r<DKSLUmYXS$A(D0a_rBU)BqqvYcUhaf<`FBsfD=
z?Ys>5)(NA^Wux#Q+ilQQ%xcD6tkqm9({_!9T%cz|vG}%l;zhIW%P(wSywozalNh;4
zXv)DL4Ns<B*|5QaG<}C1l~(#Pk_WAceU?sCEyQ^z^Z;%~hT6Q%q=wd|;JX!E45|ga
z4$^kb<ptFLtyutTYbQlr{ae!xv(A8f&>z|cyzBV8<bP)gapmB1Y#GlbNsnMVHM1%C
zkm)#Jj+op+>+&_z19*`v2a?~oPp0ljrdxe0Nc!jQFRopXYhS{m#J^bcN)w%JRguM)
z@*kXo%&Mu@a1QVaTl`#l@Q3}OW@8e7Yiz?Q)y-MO4@yvJGrYVkQ1Rmh*W)i#i!gm=
zh1s-G*pXrdfp_^gV@U~j^qHu#^+FOw#d;Tr7C+QCDWPR|<NbY53=9rrl2IRB-O*D}
z3~%Z8tQBKCOh^NIEgXE}BtUk4-&2^ZDD`5H+B5kt1EW<UVdWtN{XrYEq8*+n!Uc*k
z*#RQFpzQ`3&<acOP_u>*B9s#+t!0UA_1GLc%0e6UG<6hKCoEiet3^=VGF;G*R@z0h
zVGN3)Kdkb+w5<k2%{h*>%B$y&!xPw_1PrR!>1gL0C;%*w)SIwb4qC`a6tV0{pWwmg
zs5WR(?O!<#^&K&iiXKm(YMm*=np&$ifkq8z6=8&<aVUo(AZgff{!Iz^1}u48>i>*)
zhVnrF7_rT30zwMAJBJ#|7)^Y_<fE)fc({F+%zybvOlaw>-&&bSW`k*mlOSs*jZKrW
z8Gf-Kv3yStiX<Du0m-oLH-1oJ#JJi{K1!!<TC^2y`z^zl#jRXQg)4(+_!Zl;qNZp~
z4OIhx%RNY)CTWfLz7#HG+ivIV4J&jT`^cTvt6vmCo~V6hD+IyCcjubBoFM&?(jT?u
zDtES9HKF;<NOnJ(0Y9oY_=-M0G5u&wyf(gLENU=Lw1E4r_fVSfOe!^w1hrwf1~>6t
zOks(V-lYvy{Q>*h&R@VGqu_PB5kTlg&TvKljlne^ldAoaKK?0{S850hb#df*SQ-Tx
zib-;+p8x8log3dBtMQH)cyj|cBSBaW`;)HrTL1*HJx>9>w;W&)_$WGGXpgQ07>Ck1
zb^eS`0I=m7Dp_-}ifrv@Zf<yl7q~RD$rP40sJ-`o=Mf<op8E=P6e_G<(FMMXiZFC{
zH4*PSY~a&Lvf<LU`n{)KBeZP58W#d{V?%K)s}exci5TV$8%sT`hFAEY7pWhnS3(`6
zAlf6Ky`fjA!UWQYP=PuZ%mkbV(#cE!-YDd@oQx16pc2fRC~3(8wAA}Z6x#@aWr@2~
z$S6~v##(iy0vg9}yx+4cV^oTO*y?u)_Kw2PQn42#Ft^-id>XaK9W}i4beWq;<z|ja
zl|HL0AKDE@&kII!ddr0&O>tjVa9#uH#0<HPWsOOQw>OfIz>mv2kK~)*JpRP#3Y}9c
zwq$^`6<3h>xUeB;ezSrWmeU>~qXd9Er1;XYIXI&n^1(&kC`<sg3=x-57Yn~!AKU(R
zVB9r1_E2gsO3^7A)J7uO-*T4<|Bh}YLf8G56W#c$sP&qYquV;Yo$hu<bwbqWLI_`{
z+tFGtrp&IB0xP&r>k?<k`1&WQW6uU@d&Ix5Z(|#j&qeRGa1`g7J&FJ~(~dsDD8Wnz
z4QBtCTcB_Ub|e8C@K(-OFp-1$G87HvK%IWemBJU%qkG6$DLl-;P)s7`Ljnm03S$2W
zZmLu@<HU}1jClLwa)P$Q$<7Y#<|We$<9Hv559z24H*9XA@mJY3uv==UrPp-ZgE+On
zbfKo<s;cN;v$HkK&Xig=Ru1hB*8=Q&xI$>IV&m_;{HdhgW_uR#qf|aP%bvKwVo>G?
zNP`%p6Gyw4qGjZ?2S`(1n+7N<|CyTMCR#NTUSwJKvRN;`{vpy2V$^*v7t8EfG{30{
zHgAUVU8KTw%q`q2NziU#M;TC@L&Wy|<Ri}30v<`h1Pr7m5?|$^yG1ZSo_u@|2?M2q
zX=@OUN$Goz4?4n#69H<V>O}UE0O5Xoe81q^!y>Uo;ce#=xg&yAGZu91pw)+fD<pj{
zme=!U9(bI8&WgbE>j?>1Ndjxmel6LuAUQ>>A|ez9DCFu;^l84}SE|LpR4L{(pO!0#
zT5j;7*qQzh!b9`!NakOpm#e4D$O6Eruu~|JTPPnWlY$lT|Hy4j^|Z%4d<eWbPVGwB
z^v};$1A+mAoJRUE5wWwjXqGiNa=$Nd$@&(`c~90vL(&kd<w>xfYwn{tsT+ZcmK)@<
z#WL)O8Z~cMh%-wDZ8Ie0^O2sMYY{t_%?FnKH3@(+vy_boh(WtWPNDO+nm~NU%2(`Y
zKoVG@QvFM+1WSo5=rMnb+ZNfY4lkeOvpGTnp+D)&<gtlhURprQV+fm!zyu5Jw%1W#
zfUqXCDpv(XB+d;SxdeL;>N*C7ocv)Yp<cg>eeqt0l@?;ON~6a*B6P|4({W9x#v(Ve
z_v(vzH<3l1%r%YM%Fl;DCFECRopo7#!K5mwY}tjA4`i;22<4;hR%`mu(#X>cW6?${
zwNY5&6&4b&uC*D;42a5nl01C-H%x9tADs^#=p%F$e98<|%cClUstgj2>*e6T1Am0S
zy41)V%SGeM5dQ!#;zGSi`8aDKUq*spV@4AFP*Qu!XTwJ4VaY?qj6BU;o@QXBnloG<
zDv}gR0o-amp1rre$;BLVA!4I4=pjOQA@w?w4y-FZj*l&PcLE5IvjnEEs#s&L;l8j!
zl1UQ?3#n*;FA1>m<Mr8GW*IrK&5WH1kfw6f?SJ&4hE&BwYwxdOBe{N_<F-8XQfm>T
zD8CO8cDq_3ffy0#6=hLkA*+9T@37531f;H_{2%VqzmJrBdOm@3VQWyXeQs{7U>w?m
zw+JFZIxo!Y|6#xQ1h*1}(5|v0aL#0R3q~F5*XbiBj-%jaT&ypP>9ru7I}OJt?}SLO
zkizi+YE=SeTIrmmSgODeNay#LFIt7}X1=C&`C$6~HfzwpVyMgPi2PP7??HOALGL}~
z%>eZ+pewCw(ETMvUQPh{{tt&3HKb0aw>e1UDTJ~jXxwZu9j%Vx_z64m(q)H^2(_tW
zqM0tak9Blde!a0>uBfzhzQ+#i)+&c};2RGW$`c}+H_vor`p52AwgU1@VA|-ZSC{Wy
zRg}SPh$!E!sl`5ES4=0*c8FTvzEFe-Oj_r(kQ8AcrvIsAkneXexr&CLS3Zk%Gy>F!
zAM(|*s!s^rpIrq|A$rKC!H_4O8q2o)K4ib6@Nb*zRiaG&7WWGMlY;iBikx{--y4CP
zxE@SYz2qVMs96Q51p2RTs+O#RRvE*bbu|sP3Wi|X<NLpqVH1hYpEoL|3xMZ6Z@O&r
z&3GFg@(+1K2XX&OO^uCv{Chbjc7z>SC=QX1!(hQ{Vj<H%9=zWc$_k~#JwopGlpQ>f
z{yCxqEn0oeU1f>BR@IB?fsl~9stad9KAsCIm|J`O=RFDw$BHPs#%fQyk82)C|F%Yf
z^uBF&w&ieMa~7Jd77Bv;TIgh@kMB^#p$uY8v5wxT0Q{FUC$|`USCgO>$+B^8f-amn
z(eeO7gm^xC#?k+meOlE6@{lKIHpVgAc#ymoP)_)d_!ql(mWm+Ft$Bb+A1Z~yPE}(~
zza~0S4$Ms9L|v$qHNdzua`T6m4Kt*vA9M6-NHW&m?QQp{21vA^X*ZLo`S3)N2wk~Z
z-1p=0+XJgObA5srnT3#r%h||C{gq!W+28T$FDR3((NLc{foA131-@n$0q7kRi_!Uy
zzeydjS^OhqyPfCHcn9?tUvqGoXFRfow!zJspCMoq>^ZL*gY(Z|`m{IAFH!yN&;QuL
zSFK18{`a4B&CGi9-@Bg+h`K}dT@Bic8oXhO;bT=$2^geq^A>k=yL4~6j`eBes<ACQ
zF+$gWijTyeyFq~Y<EsRaYXu-FfZ--g+6U_^SA3EZeD!moa514)`!J`7o8#Srn|W7E
z>nN;sX@@AUu|9(RyL$`yIVUuV8<16n>I0vp@z${iOi_591}<K4ne{CaBYZ2I-Qeah
zz&O&H13aXAZ=Vh?YNl`HngN-wQ%gS`Q=PB;v;5Qw&UhfXksD9Yz&7>7bULA8&<T2x
zMFKqM%df5><c+0*p3J=$Odvs_1D3B!TZ`IN=?4ZEYsWNMkr==@ny{TJch)zF;)*0D
zJ5T|#$brtJ+XL3iYV3DLMJy<fmOqu->7Aq0r{)^Z7N)vITjY>Qev9!w!};_u4;M(A
zXe&wBDW6;FvI+hX6*0=U{nW3&@Kfyo&D%7){skKb&Mx<QVwCP@k>8Y9*$;sZZ)MW2
z?z@LEfT6Or`O04?jCq>@Gv;m_OhPcXDJ(NFL~nLiPRCyub0{n`xu0w=KALMi>rAfJ
z#`lDBmJYL(L!R6fIE9zYf(ZbHF`Uw=-Nol)?Mluxl&&TYHRJsqeLlHD%aV73$S13T
z`MyXiBlC?DIk(WjTUzQC{X+R+V9};Zih(os>cq^Hx^Q^jX$g}4GzXP+6|4qWzF9xv
zLOKTCE79M0o*o+CJX%;(Rb|PBG&#_x%+>@<zmHx;bnGc0DVm_n$PD*4aUSqho`4mK
z7)3&&U%lW9Kz5)d*+GV*2Kfjq2V5twlQw=!55A>W>-|MR!ze}w`R*CtK~y>s_U84z
z?a$|bO$iT4@<ZX^abR%3E7$EDB^G4x$C&QJo?0ZcbWeM%Ji8yW>}fr_O>o+`O?nMH
z+P;QeOhW_R#dvR5DRLyLEf0UzY{}DpcY=JCgcUhC<+|@4Y?%r1WfAEy?q7)9w_LUV
z!}biiz4naT`aXo`Ou*RY>gNwIxgp8<@O<OF7D&wK2RDMrm9iM9!B?8FBQiw!A|#{V
z-5zw~N?{v8H3Xdu2hZZYg2rtnOQZl1a6tQX9tFh9o65#k=`694U=`iZ{Ybqwo%-EB
zODZ-t0fZo6+7Ym3Th|!;1g7SJZ}gFGew7u+%s2nrCjRiHtYI>q2JJPqYp$c3#p_i*
zb->b7vbp57_WSoO7<17ufmGs$LOf`cTe>=Oohk1{$MhG?ylDb-Nx2~}Jf*B}Y@y<!
zucztD_hh|rOd0;G7XXQVaN^L;Y~lvBt5MP(LV#|b#lpOki59=LhTY)$*U!PD=$GUd
ze2XQ_LRf%uD*sg0zPV12!&li02888$24y`Q3s4j=EumHE4%!V2c)9aWKN6fsd`$rF
zc+sl$Cj2irW6sJG+4ryLh#3T95=({c2l)JS8*%&-=(=B(T@3S>1}c`k`cf!HUl)nC
zR7=<KkQkLkyC;|O*&cszep%u3nW`GVVl~W*PU+wm)<E>uW`N3(=8MyOnsXIw;uS5^
zspK6Er0(liXRUdKB;0o*TstjLbwGYdq*ihhmW6+5W76AAXddkH57Bd{9nbE=iiB9u
zp#|N()+mloW#v2Qgqw*_f2YT0Si(1e{KdrmY@;pj3T>7j5|bd5252C3(is}Rkpx-l
z4yP?=vQtBfoVNlGOF6NLT7>P2Jm+L5GK?Fji(Y&Fq%QJZU6Ci{U^X<6bG^~c8MRXU
zc5_ay4wR;SOnebQBL0rACF=sfGHVkK)-O>>yD9TfWLtgmsB@BM^8f@UY~Rpp^hpEA
zC#z4XI_V}tVRHzED&KFAqIWFP)9K}fvqSC5{^`pDs)@c5F)uWHF0XV3zE<bR1?DDb
zPB{15gyKPTsa4$R?U>&UcMZRBY!bwtOb|^gF+qw-D;Bu;`vikx*4++uL3aY)<^5Qn
zPF*q!2|#yXnS`%W9O2Q&+z%ikDHpCVN6vd6CQ{_<Pm*V%YjJcZbB?{V6TT+BU7`Ql
zbm{L69^L6s8Y?Ni&&Nuv$*Oi-QH4_8(_0{b;I{C0%kKuICu;pB9@bxE%$-*mt|vwe
zRzJ8KhaFYfNXjsnnW)cBb-jQ7N((kYTH@>RXO@ctRhrZEsBagDV`APeWJ0Dr1jZqn
zLD1C<alTP$s9O1xBfZJ2zPU=z&-hFI<D@hdG#dWNfC*%v9Iit`aW{&lXlTn0`EDJC
z`61z=$f9t|&(t4-7!|V-WESRwgYDXxR5G|Fa2Gv!>phP-=WxAM4#d(E98$1V&K;`{
zBn3m<VqC2UbebBzUvK~v3=FF^UlO@xY45ILC`_U%aonM~Gt07dhZo24PPz#6H~@pz
zRIg=&==#1v1~KRAF0gqJ)q9rK4=l|lrkU1_>ASHllG&qIjiJ@gzBRuyy^y2Xn9ggE
z0&FP4J)2J60Xzv+je>{+@t@bv0})(F=UnHFF_=*uCjOiYhw)u~Vlcwlh+&9}uht@c
zfBXmw5>41}aq`a2?L4}RZ60^xX;wcIgr|2pQ7USfg|3dNQKaZ^qOz*cc&EYV@B2~A
zVx3A2h$^+_?sw{%{y<<J-b#0EsdIK5aD^a2?#@B4_eA_OT(;OKKb!lCU(=%^Zl8V#
znJPP41gv}Y+PUfYv5-8zw*xqWzrPu|gPV!WV^`OfsiM%u`Af9J6WJgtw4*G~#^Gan
zx{{lp6)0<$yLgjN-pXR1j7vG_zognU@=i517G-&>SR=2PYaLPa!9LetZ-@9+HvGPe
zA|173GedHf{~c6#G12(9c8$DOq|LOUhg{2-{(g+i(}vrgW&Tb#GlibATNXZw0o0zn
z9@&b}38FQZpfj-DB0{<23Lq8i4v<Y_vk9lTtjJSgIob`FTiUtK$hlv1uo866hQY2s
zgTcxFo-o>~vPtL+z%o503i)ze&9krQJ@vt!0rFjaAIGiK+bNY5S$W*;>dRCk_9K>L
z?!CA1P*;=iOnE_RC^~R6<qr*vG-Yq__Ht2ifAgHv$5tvI?RI<<qrYgExBK-q*`l&u
z-G?{NP(L+eK4^s;e|^??M{u1`IIrN^;2A#JC3GS$2=*I5=x4AK&GNHiH&$?GM|Q0c
zjdhj<<x8J2@T4x@nfQx?dY2WWhDPry%$ywh^VTHSf;&xVDmO1tgl)>~waT~YU6M<G
zkp&eN$Wk=N-A~%F7<a20Z(HGnDxniNBl>8v0KH&mnCabs4Z-1qaRl@T(=qqaJa_x#
z6=zHR#5MWMLJ|?`R2e|8ZS-nV0~C38aWo*`F_&~k%D+H@e@`lgCF83z0Yovj%;Im$
zXds;llE~YsGde7w`cRI-_THI56PmX3n$V1Bo6!zy_osJgZA<wDjP!J<6=@~OS%@^g
zz-Nh1a5EM9%pW6uE|!K<5JpH_YRf0!IVRzZ*e>=KBpw+}2vfmyYrsM__m=pC3cU+p
zI)R&^CCp5nwJpBGTzyGkCO+@f{QxF_@#uD_ouV&<X1p15^5=*&bS~;3g~N@W<<S{0
zz;yV4Bmoc&b>v%q+Wpbx7E4Ui!ihrRW9Lf^UF2IIwC{x)Ob8-=K=78`M-zcprHgV_
zLdy+sUlhc&WQE}N--Xu_4MlvncQOpX!78e4dRdqx_@M)PqY*1dM)P^+z@MRSAWS=M
zGfds+SHQyio2hjsr14TpA_?j~5^Ib}^k?lbdPMW|xED<r2}IZ`=`KLU(HjozhbH+I
z+{Xl7R{h>43BU)n*)ev(zL5*i-%P0I<q%oXL#m+Sxl>);gWZR-csH{;dxghY?ATaE
z{E2D<f$`O!a`k88QEud*>ZLxD;5|(U_%v(tFYT;WRjHFOG>sZ)OR@$%hU7Kx>ryv1
zbLFy~JZzv#5h#4XFnmc{R2J-f2xaG1imUVSe?L1ss|KZvzI)k&GwGxknQSq=kgEoK
zuqG{*(ixgFxL`||-7_~_lhNSx8P{K20(zdqh?D>h+4UJ6XW-Hi6U3Y^EC7Q@m$6Bk
z9R_INXZk&2YvrXO0?M8a`o2e~1J5(U_M_II0sJ=~Hv>C){)B?Z6zkPOU7xe6I|{D9
z=!FxJWvc>7Dc6{^s`;G`c9c!8bCnLB)e=L(zX*L+sgS9*&p`i@?FT!e+00(hga6X@
zq=Q!MzO*kP0wUdYo5i->SuOC3=F5B&Z0)7mRGb_vhLfd9_V89c_k4YZFXMk>|Iq3%
zqROs&m)+zOSe8pqn_ML&;TD@BPo@&GvDa7DJeva4q<WPQQzr1o@bufid@F%48O;~5
zbPl_OQ>92I1MCRAU$5+xZ3z7HsTh%R1wCm#M*^~W0R7kHN#6MTX?rjkTC#6zOwj<f
zopW=;m!m9t1R=`B$-Zw>5k1#eptgX#uT~1>pa>k#P_0;v%qVkrCvtE3uoI(Xn8O&X
zLzUBK71TZ`*{n^wcSpk}Fvefx>34mqj~m<XilI`@2MLn%s6R-18sTZJC6ZMitHc5{
z#pe`d@B(c~!I!wrG1n~p_E#gaoA;5ERnDFXt?2yY9`%399dhbM|9JeAED*DA47oxE
zyl+$V!_R;ra{GRF61Jk>mkD@@oJ|J&ha`Wj5pFX=#ge`%lnOJgy}lo3UFEFO;IO~C
zB@c6Z>-OAn$Eu-eH$ESj8lDbykvf091*=mMdeW9X329q@@eFJR;bw%Ixnj$<#2Nl{
z*LleVM-H6tOMc@vooWWQ&5QpO)JNu7!x@+V?yu%8>_-WJ{EaqY(6h_*E9aTK_;h3<
z+!V4N_Pirg0uqijRgM10^6q6g)EpFefiXXD!(ybwAuxa}Mg^!1brh6cDR`<a#}lc{
z=ls|_Z5iI8IVL!BR^A_<xUb%NARfp+n+6;s<w@)q1gg`d5*35h0n&bXm?b^NSi^hz
z-QQ}!H|zXXGu%{eC96=u92)Y);xEv-ebQ~`MrYO2*86bM);>mP8*IbaplNqvW7Rqu
zc-%I1`RG7>6l>Kyfz>t*uZ+IDF;7aCkO&CS)o7{cUx<tT1?zkxM?88IGr;Af82S9q
zrh|bkGVwp{y^Vo}vHSfjO+6USkO?<$zvx<R#h>Loz1#S1=f#cxj1~FdNORy3a#28r
zVS6|^26oi`z{b1A6qN>QyG3t$AM2pO_SK^~rejeLJIOo@K(CQY%sxH;!4_AI*p1Rd
zl9ne}6$k{u;v}iOx0xzZexZIj6&P&QwtBDps#KWQ%-BwZ7@=qIgSLfU`73u!++3RU
z_ocnZ8shH;KQ_1c@!NKd&U}2TQ6xffMC4@ffDrP9%<zp(&w-s|7@w+jemw~)i0{Q^
zSioaXsIQtX_c${emyDLJfe$GGGeGO)xL-1ye%(LG2wzp4#NCzFG7lVq>m@0C&gyJp
zLi+B;yWY21tRwJN8tq(*8({(Z6{PV$vFiwEe=bo9GWSU<aN#0C@Jf3ozD)0^|FD%R
z<NMF)Q>Z2SOWb$AIfKDm+rKuuc1b=YpNJEmkQaGOwOs$m+;JpiyFMU+*Dsh@t@5@$
zhutY=@%c-UB4^6?VaCfJQ}*3XlVJ!3rJ^qR)vOI8+#_sagl#)4KcXXSrYGI^bs2uO
z>Vv0jqOWE|Ti0~Y<B0{+>Pc81P^l3@j2N%ybK$=SaJk4v2_e_RwYT|EGn=p@B>BOq
z%CD5+H^xsfR{6je$H)&{_BrN6R2WXa;xI@jLj*8KIPmG6;U}CxJyRNJbR!tyEk%80
z2?k!BhQL>A?H^$lQesg+jIioL{bNT6gw^9}+oQi<kv5Wo+rNSxp7YEp?^T&Q->!x+
zu(0znR`_8P<jzT)0J!bfP^Evf;pf(&Md1xiKXoRk?!Q|59@zYW)=ElNIPZ_zLz1()
zp+E3=3c+ZYJcY5mAtSmmD@2=?!Q6h1f!2~&GZ+jo>ZxN1oaoLmTsAyY<eZnU<Md^;
z8109@ih54^cSX`|aT2G?8afb$!ja#$A6+l1u?}Xl;%y2k0e5bVU!4PKK4Pr~Rh!$S
zzR)C?ezrhI6c0z<_}k|@PP*)m>lc8rFQ9~K`ImUa*YU^%avQQAA;l|Tq62AwBfTzN
z369<<sD!`-h*E<RHy%?!>TbubVw>;_%immNJ>t7}UMfq31z5&>^g5f2Xfjz;TJF)=
z9pQmyF}xq%l%SZ>Rgq8=hU(+<*^d}WY0((&dCFsLp1YS+*NfQjr_a~1t(A7d4n452
z2x*_!Vg8QJTY6rV)SPi+TA9CSQ&!3fSU?4Xixu=EGy6}Y)BrlRoe@N{r%-V?b<JDF
z`F4`Nd%k~Qq)0nwPaJkc#g^@!OUf4|_*u0XwBla;0Y6F12>Xzk7x?<QTf+0Uzil!V
zR;QNW9VOS=A*K18tCc(fj?{UxPNv7NQVz%iG)h99JMP#UHi05Jn;6w9fBSB117Q<{
zX1h9l+Pr88==R3LLwPqM1fJSu3}_9|IwhQY5=o*zS>X!MBB$EKwR2vp<{L}1Uq45f
zk@@P$<C}h8_?hb?5MTy$6#I(2DUA5A7pde9WHK=Ho_Y3sYvd{m&|{3FV<JEa1W*uL
zi`(}c_;gT~HX?kmX$?OlED{}JgYj_i;hxUqwswuDzzq{I0#jXbc7rzu9#?GMvV`pp
zL5&8={0bm;zHVx1r+P*E6gVaA!v6X0Mlu?>heNURAR_Y2kSq8z=ofv{yVn}@siDBZ
zblz^cCtj6JzV1}DH51hAv|RtEQ5DC}qX<T%kFrqb8pwmm69Gr}&P6n*`!GV+{9__u
zw{USft;m5klNoY)gcrd;#s}p@kMMr{C?z!g=>o1)?np%D1~xhtBY(lpT#jbxycH%Q
z4k;-V1IR**Vb(e}^0l%BzOW;Jci1_ZAtRaFsV?^%o*{{kaEjd-IlHZ>BYQXd2BF}j
zLZww;+Y<Ugz7o$mk_m^U_nG}0sl+q}GE8Uma0ja25U)>I&5&oz5P>p%1@viw)jewk
z5y2-o{KjFty6yJ@NW^ZZ7HmXsQA0tJjB(_@weDXsLMu4q87402BQ!SN1XxDG%^1dO
zj&KOxe8e&<?jM<T$Ey;ZPj_=Phz9sxm=q~D_SHVaC4?O%+F|fB+zcBy0w1<}SGcWB
z!IrC5&^wB#S@$1~4=aY<5hg$GLkBxWy08!>J<5Pz@w(g>-R~C{i0KBca<2&?PW1lH
z#tOwJkA0KaA^W!o!Dlu^sC)iXxnG4`n%_|Hv4SuCy~-RI7ojWZb<0NMvM%M1vT%%u
zxa&S&Xyny*%tOtWRP$7ctvL33KYYh2S+z&mGtiWJc+#KtKjkk$yXDBi_zjtAY$Hc`
zssZ~}*SVkfjEsEJDuScg@oKTvA9<jED*s+s);h|gY4VIq<36@#u+5bWCJ{oy+$G%6
zvdXfG)C|m;XHe4mK}BLkdpD?l5!X=pUq+VWI+}TufGRIid!Ev;EY_}2U?#@KcXT5?
zqe{fiFWImczicdL<H5gj2|V{h2ClsrarJPzBJD<E$bt@2q_JTcr3FMmeoi%L_hY*$
z&t_{fZ@g;5uu2<ok(^$`)WC9j!}ivt6s#SHA=1Fm`-~0sZbCb-c<|X!I!G_dc!Ztd
zE;2|K%{A1p#7s22Ioj5yGPf)ZWxZJ`D#LoddSGXF?+f-zSh0mJa%-WUJrz(RXFVg!
z|9H%8raO+!ifs9{t#m3~!N2zu!*S~9gz1OF!AfV!op`mgi^4Z0^u+;Q;=W^|E)vsF
zco8YU8#Q@BjCU}grZ<mEgraxxeNR@D!G>klRmPG#BvO~NUXt~ESOJDIRCP)vZ2roq
zr30o>-bq|X87=z}L$rymt4GL|l6a8;J+K$4{(6dx$l9F*fyGIz#Se)vk*4sF;6BPi
z!X{>{38`3KKv8oJJ6WCM__S@0VhR;@i&}Y>_8;}C+t940NnXv@1P11#Z$f{LutJN>
zUccY_^-{+FZ5UW&X!ciLkWMM_z=pRQ`pBZMURk#*S02w~{`&fU0xk<nXZHO7rq?d1
z)kP1Kwiu;Yjgh6#*+pb3z}Xkx#vL2>+Q1&K1zjF$$Lm8kTf#lkr(i6hmKIrv;};%7
z<+#O{QR!@BaoXXbqk^xjc%c%A#f&8GA;fof)t<Sp;UI8d3Fv6YSmP)d&nH2B%X#y~
zswU(sDTVS<B_%V^)b1S`J`*b_kj#h_Qx>W^7LTZRW?N9di&Pr>C>jc8S-28b5(*8Z
zMV^yRi=P{Fgo3IaOsD8<yF&smpmP!2;<TJC20byf4%dA|Rt2gVw@#^lAVr)A{wq7F
znyh>3i_3O@%>}nq>`iv<&4Ed%2IxdfxQ4-Y;jTZQ-BRg=8HHRVHSo0B-FI;(Dy)jw
z9}PoH&Mu>~nH?L~Fe<+0azVUXoyu(YRGmq?$xes#kl^*ndn9P*Kde#%csFapss$H{
z!*icVpb2A{qKof7;G6T$R0Sp%wlI)fb_$0>?mv;_qNFL8jgJ$~FixXN)}N#&vXI25
z-)jM_hR<c~4~q}xD$I@Jz2sunCy}}Zqvvyf<7~y--@mC7C0Xu)wFh6$+S4T6xkpyD
z3C@y1?oB?`bG2X-W^^FEt})o=_j2CizVR5=tKOYgn@B&d?fYbEfYb3)ImO=|+c1Pg
z@I3w#_lG=h``^(irJ&6bd*Eu_H=UgNNF*%0UV3PhY5Kq_KU)dVo64P%rkwV8|EuyZ
zM`r6@$_x*<-HQzZNnS&b`K=9HCasoW<=M~9p1hPxCi^Z`^deMhiAYXIM?3S$#~()T
z26ibj#88L!9b)E2!EbddhS<UKcgyZm3}u3Go#>*E_rgPZw8s40R}oNY4lWN$!yJ2O
zAv)gB2Y<kyJHGi*WTZ?o|5&_8j+EF@;K_#)<9Rq2_b?%(NWKp%wY0<;lLILwJii13
z?Z?HhMXHt#Os>_{@7|glm4da2V+qx>fV`d@TA`3}CG}UNGD5PqV_<uw$NNy<Xh|XP
zgmJL;j8_!t5HamI2=RP$6oH=%{Hz?4tXJefnrZ`a9fTaR&JWc*eyN?7nIFsg2Q{c~
zSt4<s=8oh1dDhZmx@6EXrpb(q+j+Rt^5_*~W;us8n5tC<U3Sxp1mlb}Hyn;AE}QfY
z?VTRm`^qM38<JX(_c6a#cp>@8`QWuO0a|REpp*9TrHjzkLW5i@>MlH%XybC2PMHkJ
zkqqQ~*7stWe9)F5-ciVo%Y|p_x>5hwMDbyLrE|>yl9cTw$z#``*?EH`h%Zx-+;<9{
zU)sMCiF_j;s5XOEgwJ?<OJMW^EZ{`>x`^Azfq~AlcE0apIBGDYjz3*_^t;5`GR&ZT
zk9+|hGFY98a4$DKM<e?{IQ8s?igLe<#c1<9A!2nOlFB?bBmikrwSA_UhD{MQy*mg=
zEB<7r26+9x-m6#OK=QUV_e-FN{y20sr{mMJpyBqY+sHf@eD^W#-U&iF)83x>!`~3Z
zC?_nFC{e>|JQ*ti7PtZ>y4>aAN`e-G36=8nD=|$MiMg1$@w>%uHCYxfh@E4wr$ZO9
z^rb1x@}ANBT(8ugD`L(>a+8LAD6%gTAUwdpjwR8zeVz!LP#SY#Icnc&py8=Z0(rtx
zZarZnu6-OY`N6qH@i)TVe)W4lLHmWLwpNY~z@2xS+%lt>Q=lrI&PZ0g8kn`@U=t18
zevi94^y6XE0#4F9tw@ik2XaNNNfG85*vv==YJ<YV3L$BMu?!|0sP9wZ-eGzUJ;3Y*
zrfy34Ym9QLtys|Ukd_A@aw~)sVMn(3$gRTEzy^2m{OT{MJD!oDq0Epk9<?kMVOqdS
zi633f0&RDpi^HQ0tQE0hCK>{XCHZ`t@TLmq<RlGtB7`A+lJY1}7@iJhF%raZf};7A
z&GC2}^>=dC@4wdU4CY@yKBs;5c*Q0n;E&DGLlLuMUQXaiJ>2airUenfHznjvRPL5c
zZdkQNrM~$5$I09fHC9o|mn{NidExB6N9A<o%(z?_Y~G%9<M+A_CKhb&@jZyWlo+fb
z9hXlDuYvdu?GUsgLI&}(bw8mL?kVgaew~7GfWR;d7sXD*zvKZfUuoF}8hkKoCdU3W
zzfQ?u1b=_66<svs;DLnWHYmE{K=pZpZto?fJ%E_$AE{qtP04p~p~w{hd2u-Xz4CZs
zl<^)ro4NwFbI*=bUOJN_dSNH~G6IUaB~s;q=L1~`5)o1g@u7XkFK*_}_AhVn-qp{s
zHpZ@J3D+lKpA^<4*5^M0OG?yuloxcB-jQ!5BfAE*MGO*_6<2>uzsODP)geuAs1-{(
zx?$}!@Oa>R-otwoMFOkLi!I)eiG@2nkTqJLg~wJgQ>3Dta*L>+_p&_MAi~7DlA)t~
z-lsI)J&#$Dksf|huyS7`($ac+{)d#0DaYmqDBg0n2R?Aj<!a-Rpjj0g*ZWoFi|^0U
zJKxf~L+~MgA__nClLB=1+w94}I=eWyd06>_drTM13$nJh&K)vogjSgl-MF+y8UrLh
z$QNIFRk9Fs+8QPchx!^^2l9Wl5CmRqQzZ>|KTxL1jS2!A`(eT*(aB;~@c@2tR+^B2
z%Bq&@=mPMnu7;-=0&-T|B6#sc>;?3K0R=Qr`SAR4)mGFNdBb`Cxm#Gn`TnEAH%?Ox
z<N6%oXoz55i&DLOcOn#;TTrYDkn39*q$jl-lxbX{1jenhKQ#tZ8C1!3L@&$4w)osQ
z)NoSethg|D6-3;A>nyf(4Gh-Vrp%z~v9mVEQsa1ZXm!6SEHC_ACekq>2GG}uwn~*S
zEPU%@_H@8w@~I?uLhKUmxW)j7o@52>Ez{+dg3VOMYG?N15N-MHrxR)G5}a3%MQa}S
zOGSPugZf_>^5LbuUQeIj>PsIhgB3PSGVuF(P2mPiKy`X73CannaXS3d^v~Q_`~19h
zl#`;oCl*C7G}~V?-n-U|DZDEg_>BIiSa4=Gkxt!G5fCIq(H{x)|7Jy6KE(EA*Er0S
zJ}17t)H#@D$y*M&F-^q2-;e+@SlP)M2${e_dJvbU^hbD4D%cl<v}H+u(tdi4`TSmk
z$$e$PtK^hD74aRbz@PDPfWmfe2{E$H=|Wq|@vT&9$>051|I+x*+;-}H^1hwclQySZ
zUPpx!SNEXDRy<DkYeFrWsauTT-#-7E%=O0YaPd$L=u5-a`4Vj;@-#_M$zC1UHf5Y`
zwn}B)gUx}bc*Ka8Tr1m20UGJCMfLUsqwAb<>2iafPvk!LmN_~L5oNOdxw$UQi1|?@
zLL4ALAbC?<J`lQ@Wn{406gRnwp&|4YOaQKots|S&H(`;<_nk!-+7~E4i_{sChjku&
z%25J-gite;bmk24LS-%t@6Lrucg^-DDf-5PKW<ocU52M*O+I1=Y-w-On2?OC_>u~Q
zKhy!5H&^H8qu$&iRQKSweR#`AH%Xd2lJ+Ix*jNsz8o=3UOdt4;MiNprFo-WMFaD9X
z;4$y<yR@0BM=iI+_rK<nH74cvVB8rVCbMXOMu>QTOcI9rhRB7=;um=IMJ9uLeM*#R
zVmyYtjffD><F%6Cdp?{QZI^mHUKw*_z`f`$P8;g<M-9z)hM@Oq>gCxBJ4JP7{UlGV
zti(r)CQbzeii`7Rwi|ulIxz=OVyzOt_1TV7pkENzd@XGrE23ml@giK6$Z)J9XGWAn
zyD4>?pL&m2I|Ii_BDipx!QYp4ND`Ns5?!mKV(KTqv%pLE>gd&n(1|si+rFMRUJ|Eo
z8F_mi8n(p~fj-RU^SPQWp$P_Atr?_>vA!vxCw^}>ChJJugb2gK%{dOu1rlFcj=QQ~
zPY5hMHf0^6RK0V8Gs3$=qQ3N{0R`lXT_h9^vO(Wt)S=?rgJ;XVy*o}){sdzAab3|f
zAuw&yC#oN_ib+t<#nrBmKga)+y*uH4G=yk5u|I<yir}Q?In~Qy&6e(lbyBXg0JsE2
zL`Yv)K9}D6d^;GJUX||E9F&BQIB3Iyg_4Gj@M_fko%<f!Mp{qv#W0LFtDoW*HQ!N-
zm=T4gLjS85;24)TuU^?ZAz4$0C0P>NVIWow+7`}#AIqwX8-snZO*R^4woB(<%H9y~
zTUWDSC@%IyYg3+#ev{gz+PS~9ian{&?>cN<LalCm&OKzB;e9~7MN97H;}?%DI@K`V
zE@^zu1%kA^UvSC!c*#PErt~ODpV)kF*AW`0#z$=h$h1G;^RcmJ1_D!a_e!Q5_{QPk
zoP4fyNJSby<iU@=aZ|2wJG}dEuif44;Ouc3f|qlV`xIbOL7_e$w&6HK#eWr(tZqCb
z(l*cRt%uWVM)miuz5_OTcB&k^5Tc7z#6K7zt6I#AR2>vAp!#vUeChP!exF6(c6BLF
z@{^Y4i3NtA=R|o(`L%Qt+K%av^aadNQ9!;}9fI8pQGZyrxx2o)kVuAr-wv)fUszVg
zYv;;XDQPhU+IW^P)w(y%Zm9BjZRc0t<LYws3wY}%HL<|(wci`HzWZ%b)TPhy@mrC5
z?anDU{|`~$z@BH<bbZFQ8nZE*CXH>gX>8kS(Ac(}G`5q*Mq}Hy^<BN6<2}B=u-BfM
zwb#sEBLz;vP(Oqtv##%X#ziauUP69{%xhng(8D?~BlTS!;lP&`Ti1@K&l;t^5jVKa
zPs@A%D!k7zMlXAO-xU<>WttIY_IE^3a%BNH<COj?=L83`0DgkeP%j~sw$b7ReL_Wo
zSXKe=gbahmxk4u}8`rk$^iP_&l+0+5rHN%*!at+wfmyjfdkewG<fHQLv}QfycqB#-
zhBk20Zpk!t{96?S_*LWb_*pJO{{4RSXfiH?u$O(u@Wr~@9esVc0<0)-hN=d6HA9M(
zY(fxJgYd;U6hJf7VrAh|Ii<hf6>#X?htj~HY96k2Zrr@7REJ&+P*;y|VrME@8No|{
z%Lp+5;fegUDD$ppMM|4w3N{DBhf~}9p(SQTHBWF{c9rTxy8iU&86t_xQ5Kg4E1@wt
z?FPVkz&u`ZsY)*j^5JqD)L2yk2ns*5r%&8vj6G3+#bA4&G8HhKBezZTO8^apL3P}v
z?o+H-3-6f?4b01wy9J?^ERz|<4db8aN4HD~TLgGutjm|I_$H4~w8`j$iPM6lUE$LO
zzSL~hvux#$rUUEWr!Q`u3RwB4$_5IjIH`Z;>v9O!O?IU5Cz-UR_5OR333P<ySJ51G
zYalPjjilAQH=?vGbi-fWSbHM-!f(&;Z!74oV$P!n-YqtRbTG)qI>&nzsmu)Ex_ik*
zpKj~r9f`x9o#6|L;g3L#6nCpNFJ%s|Nw$Yr^!dT0ei#X$q@#`Dzw%B3He#fpt%^*i
z<Xr>CKM;0~N?2NrLH)=^D))mvlgsM_RG7lP=zvBy*xzA2EM}3;=i)(yNdV^Efh+#j
zld-8xgggVEx`ylLuXqnwHRdn@T??!5l_EjME4puF^=sY5C0h}u0@qb_mj&dX*uaJ1
zDa>1B9VEeXB>#+9En^xs!yVD~e6J0U0T^R)!&>ub;3YwWE7XR~UR7Ta$;NjInRdA?
zDy+qJ0yj+w-ILvrr%!dDm<NwgGW$Eb29gu7Q@K-OM8#lQ1L%Htz-=N0VsUdjvf9+}
z6>2+4?n*uf&Zv%>N4VnEgh@L5yR5Xnc!ALkQ!mS$7e9ahG?xS~qXp9WH+5hpL$aCt
z)v`8IEl@rCa&^qW$?~0adpM3j-G_Q|Zw`Y0{ajS)35J;L-Yu0Tmom;Awt}D>r3KsV
z{*z*3AMXt@FEMl$zSOh%UB@&;%G6|2f9Ya1cwsf=mnGzV@SJV5VTV;Tigso#xPp}T
zDv|<9nY3Rit9zAuE(d0;xH?rE@<;94(`tz`(PHw|0gnU4H-XMC#2-7DW2lfhnf>c_
zK2>E8gQ`@$f?}1OF!{mlL+s4pW}~`tRI$Cur?iUf?3tQI-cd?hbbtDl#IqcuQs6W6
zjG8tVFpNaZK4v8?I1JgDyhQ}nQQ{*CLq;$|C?tV~N8;s$I39~K?Sl>pfUR)51w(@*
z=2(xSAF1Ip9d&u7H5Q3({1eY%h8U2)$nHOG$v41a(9p%!#=EFs4@+X?DGOJ<L_q#M
zo~(?1S!u4&n0G<Z4WkS1^V<Yg6u2!yjQ5sfeS9VZq~(G+nljAQ;B;J@??181H_V<2
zURN<W=~8m&spF=CW@^qo!fSm;2Sz?T`=gv=ivtQ2$~4G#T@3ACKRQ2cM$ZHRizmZG
zIfUD5-Y_hPL&4iOjPs$J^7ml=MOAhaq#x}``t@Ny?qOFNu9QZyk;uQ?)Bhlp8CqnE
z)eC>X@A;-@Yd1nGP}#9fp3a=Mxv{Enr6C9u=KAu)%7U?8UG!3qjX3>9F_J&fEC0b`
zUfeM{M0Z2?Al$wZMh{Kyi~YJ6Ao1o-j$@qfdE~rGa==<&(Ee&k?|dDUh^Z_H;+)|H
z+`lnTHrO3fbAnh++U6cYLr~V}D(hz|$>oNgh(%;z;aLPtY&KS$8AZTL7_lY^r>dJ#
zBY|rk*P<SB4h)=u%h?hIQEs+KU@@<Q68h&2dsK@k4@@Dh9=TQPj1%En8V@`Dihp$q
z4N~C5Kuc}gs;(zS*QnHU=6AVmkA#|<wSH4GDUbL?HCe}3YldlYa+QNXr~mY0JErZ&
zV66T0)Z1~tl&TqeCcIcuCd*qS{ALcdR>7fDesMzieBHiW#%9l?Cs?MQr(-8lz{jWt
z7IxtVf#;Y44OQc@y_>zL-*KGEkrZ8ZGvXu1kuhj*q$O2p*IFXe^W+z^b#}0VC*nv~
zZx-rJe(^IamlN9G{E`xbQ98B;0k0JU2+_3N{w(4VIPk}uxE=CVi4gDaP=!UI(vY{`
z=Dz8mf!Tu({cf5uan*46C4vDKDnm~=%x1#{I&jH1%*yCDVs__rrs~;Yd`fHfn`I;p
z{69qGbPk^EpQ?gbZHB&2HX8=`Qy7Ws?|bhFMlC@qM1GL6z^8uqwq9^YK^IW{?RM_#
zf-XTp4P#p+7q|<dkkO+uzK4CHu@V84_}#hV500Tv3KsEd14UeKTT-Sy`)<EXkI3BJ
zX{faz;-W}R?KWU-!QBMmu9g-hrNh27x>EmAM^kx^gY^m9=9b(qgq?E-Q&*P6Glv83
z`V<-Eabh*0p={I`MUxik1L2WwHCgQf@xX<O?bzr%K<fLty@PE}9U$0PXwr<_<?@el
z@L1lg2F3vj^^F1b0$A=+CST=+wIkx>qbtfs_@soxtZ<_2z(2{J#m4FhL;xX%12Eag
zL2>{JSw5*WO=%Od+{3F9{d-s!C%IeIQ#(8@9H$Ki&B^O+wOr;ED&StxqYwx*zmFyU
z+&3McS{KYi>nd#b;zDJT-^;Vo;+M;Vt-OT078;#=8j;USNr87*?lDPYkB-A?c|#v4
z-HKT1c_;j0AR$<+?_jE;TVKz-<;_UaN1b>%L)kX*Bmt-zmE_UcWy((}R2DJ4ou;3O
zllstTv`lQxJ87&SLO5VtPDQShWce5E%bjX-L@#C4641B>ok*HBXrYHS3Zi+b(c}q%
zFM3WcRIXGh%$nRyeKoS;`#Ss*weZN#zP7%Cv~i3uXPZ^tZ+$UJEX`)zYaJ2ybs8Pg
zm)2vDqLgG~OpC%*3!|YV1Sft5rp;i_<J|eO`SIzCj1;_p&D9ltjP{;&we0f`;w60$
zX{H7blMtYW1A2YO@??dsNpIsQ${1cZZ_SPx0tQ?TDX3xfzBGv6Ao9?Y14ekscXl#=
z;WqgkUOFU^xcro)%qib;7x5AT$ROOSmb7MSP+A0UyC2rX5Ye0mgn<oZuMRBmL*dWK
z$Hi~M{*WRdHsTvRGLVw`usyAdH=hr{I9KDdj0o4s*UW~>d{>cylDyZ63xA-AOY5j6
zu5f^EyU?7~MOkGi_FyCW@J<INc6>^7W78L?Tryor!7`{AqL=gVxj2iFDdFPDbDv?@
z{ut9R9!0JiXj`V|0>&|RPHQE7&fA<rjdkQ*KZ^Oaw#pYdlj_2Djnu)zkosy?)S+dC
zzz?DOwO9tTm8LJM=Wa%+9Ab_jY1MTa{&vPOPnE)O4sgCVCY|=>3|$N}XV=qj%$^Nk
z(TV_t4PaT>wkC`*V>~*Mi+(o`nA27CKFQHAN3aZr#TTwT+<Z8}%ego)5xXks;}Rg+
z=R0qYri3F2@K50d@mzN11!Cew<Q1J7n8G~4ZeZ-ZBaV|pf?f7Y5k5rb`=SBk(Q5y@
z+czz&OO_}!oKjQ>o*|ROI*s2S_Y(YH)vk1Iq_xGlepY<a$u_4rP&MU&6o|)iQi07u
zNNk4W^JK$EfB^|NWT+oln3JXi-!8koaMP(vr!9<#1a{>6Xut!{jx@ybDO5iM9ms28
zrX}XYA{gd$2fOkwFEX!{*a&WP?~z1RIF^}}m$~)b)gd?KD103Y_Y^Hv?&7Zg3K&2P
z!UckNt5U@7AljrAH6X2^CUsE^p782OJP&w-L<kzjm06vg+hCV-t)pnW4FV}Z8Bt1R
zFi@7NMOYBLq!^)<ejD+g=wVu+1o={kw&`VH?v6Gr8@V0K0t|8C=dXa?EIeMcy&v^c
zL`+Sf*7+g+R*J~p&iVbEG3+Kwl-~jwitm*X&287pYViTf6{LI!hC&W4`jjnZutxs1
z=5o*%gC)?w3@k@mv<r?DcvqwCC?RS#KB9r(FH5fy>nfu2JutC#?Du#-V5nxmChSKS
ztnaP+)Pc4<=y!fw%y#q0-|~-q3Wo!VEj#0(aoI#YQ!i6`M69x^i~0m{oqu=y<acKh
zJeZxMI`3DOWDK`|e7ja_UcV$OcC`y;*b`AV!pC<yGYJ2wE16@36v(xvc(Z|TL0h`8
z4X8Em>Oa;<(CguE8kP1@CNhfl>J58E!4-;2PY&4{lsIsq@C}>3F8Q2>rz~>&o{-iU
z>``Z8RqU%6T4e8ORAIeHLKndu1-l><4%4thvg6&jPrQE{+-PH))eoe=1c?cuQMbTa
z*K&`H^PzWrM$2tg$T#Rn$mhYXQ~R?`jZ?f+@^d}aP+Z?`s(X1`j=n(Ug3X2DVY3n9
z3LQ)jgNv@cJ)>bFS^kUCO=r0nnkH_`dPCGGIuVxqWitm@By@j|wLix!!Dl?BoDrAM
zHjqCFnDCewg+Vf;FWvm{YQ(ll*nI@)oK2pZb?{YNWHIiuj#URF*jRBqe2XmsR0Lo~
zzQFn)18?rn)Emv~^v8EyZKps`j<krG4SGoXgU^|3<q3y>s_WJmMg8xvc?w0n0s-p5
z5KTBo6(pnLf2Yl1o87pZ{4L_dHI`2Ns^!tk>9FF1l=u-nkN_NYnPQA?0G5cmbxl5q
zMk9%tyV?WMIl%pTx?;rtwAz-s$1jc`vpjOW{-hlVE~i~3icQ$iiK~gp8YZKmPtGZ~
zLM$&_^1o~%2t3*AmyO<VJ}*uRrmHvZUmQuStu!m|t@k?5?~ecYB-PPJxqeIcS&$S=
zYZrb~zH*p>MUSA;cM<yFcP~#Kmj^trY@Tut8}-q#d|PrR@4z&CI~9N4F{VPS;LGw>
zk69=N(8)GPLHo_LLdwCCT?@bEAohu#{C3IqieC2%W}FjKzm}Z=bp=LnS`I^*6LJYy
z^#ZB@s)EzZiZ(UeCPbIQvC;483!bHi;=jw=Hty$<tE@W$=*dPPW!-m~e3-Q^_uD71
zzczfr>P8$+;3<2}VT(U{0|$GeqE<;Q!w;jZu%>$8jOG;0!d;<gn77FQqQ4zQLcm=;
zBSTCjB$ZOE8Dva)lRW-%H&P>SnVo7VdJ0krTl=FIy|t0<_b$-o?xz}l;@=fB8Z2wq
zh{en8Jsr)zEcmlOBCTqFe{qjAdd*J(`l*N(e(nc2#sVXf-xeqVw&puC;$>tGFON0}
z>{PKHwE7<l%O-#Tqt@c__0OoX(}NsXx@Hf_Mso5d2P&-AOzCp!+V7;)Z~XV}LY|Te
zpl*q?YLjSskV$i5mlqQGiZRR+Hqm%F(SakF0kw^j$JDuTW6G-N5b|93NescT@Ml?c
zFp^-DYWb1jo0p_`d4~q_^%^sNHs58(<@D7P0TCNCa|ZK>k?iX41g(dQ77;(X6*E~n
zL2@1aT^dNqMy)x&2>j5M$w$lN5L0}W@r2<OH8SsP{Hjg>3yyl-ia7QI#HFG|b=^Ec
zEgW1TwBA3vD|VG>LBZ9B2JG*Lnz^@)a#8`<J#-Jc=Y<%B6K=4+Rr<8FkX>Hk^DPsj
zJDFH~-LRV|ETx)?9Kr96rdX95ir{rb&MnRo<@@!j-oZeCo;BV@$f20gi3<$fRs_rE
z-f$z-MtQtt%kC%d8e8i2h~e3ICnU0OSm^~eqTtlzvhqtwwRq9;^Wm*3L!C<(KkNw|
zB=N@?<{KLI=}svAcholbjxjYqrJva8uVm#DoLLXbU@jd6sXYrQHS2Uqm1ju6W<w=B
zT2Vu9!ENM6*x96!FZ*C0y@{M*^-#}|TNRhbhVzuJj4Z_P##lxFc!U_Sk^!qTbD4B6
zK3_~Ll&&SA&82f4Ztt&aTz>j{67o=F3BbF#axl;;Lv^)8`=8BEvTJ~P^ON08%M+j*
z?JI)qQ#~h&e2+r@p8FcbHuLhdW!2J5T%k#V0N&MLqo3NuO_@OgLrwgWeE1hmuT+l$
z%hMePZ^RSsj1|8wOK5{Kk587XMvGczI;w@$Uwjr91?+N&zaI;H)4;~{a!}}=R;-ss
z1`tm|Jad18ckW$aGBPYPu87I*s9W<OVR+5_G|M1aIGY~+P*r1A7{WTty_kj5?qBG_
z2V?VdnAE{LHqbQX`fWKh7lw8#;tF~aB|m<DNP4d2RQx5VXEcT0OJbHH0vq>eH3IbE
zIxVZV!gwy<$wpp^$Q*w`Z@2vZQ~oB&5vf97WdjmOPbpKyIk0nT%=VxEhG{(-^=Z1N
z<2ZFVzMg)6Qf-|p2~l-L(i4TkcexUgvB0bav)D|l<^am=Qnxcjtozh|RUdW=1?q`(
z8bz1x^uFb@|9)pgn$`21|L1fxFgtSXBZQ1TxPs<{@w<h3*FuSqKFUneDo1WkTtw&J
z9ALT$umTSmZ#I?AtNvvIwWPjXj?Bg`0mGpbK<&p<W22p4Rc~M_FVJaXxzC=rf;=z$
z3?leLbaNfkvrm3#UFn;5%-tXm{KTn0Zh7j{u{>2Z1RDg!Mq?qaPrGtI3lk^B<@!1{
zqxXvagTZdjjV}<8g1`%J#%~UwBkSn;Ee&j4<tltAn6+MyVuC+LD(VtIUdDc!VHh&$
zgoz#rRbJQ2^4+!0U@4tpoFhGHM+OVoW3hg@k0buA?e6oLi^RP#erH?f$h^>f+osW$
zI%7;CR*D-bx7Mg^mQw=wya%aCoCEJld?-hXL_l6x!NH^>r7@ZNFS+9<+!)@<YG?G?
z7rqf-wjor!5-RVRob%AkJaV@t^Xo$WWWidI8l|T2MbN_4lCV&@_<^2Xy|n#vN|%Rq
z!c>6N<6Fpx=MMeEJ-X^py5AT(6kA6n<lioZ+4^K+YA9n7K(wC$_Q-D(ew6BHN9G*+
zN=O`@)^$$92GjmeIMDcu&|8QLj&=T^#TNBI7-Zw~>maA`9^EyU0?=aX4z^M{4IzfU
zRP;aD+tUl*_BiU{7X+*ppc&TILlwJWeJR|hJ~L&34nIPz;XtSoL(2wUh+2|T0Lp`M
zob+?j;!W{Ef(-?fm3ADcvxf>ir3=MPX9eaO#hp!><!c$CZO?x8#mznGosXI#?Jw2}
z!du1^%6z%Y&3;^P+VE_S!S>dUI5H_%IO^$tF%#j48MJgmz^{a_z_VCP-m`}lppT~7
zI#??>t#!=4SM9;gcYm6}oG^KayjGe+JFx^s!f*uUZ;`J(M1fvefshK|RExw1uL07p
zH%f=Al?AAFjf?zwZ#3VyKJ|@L);9|JORYY-?d3K1cq1>neI1Z*6qQvT4<5-gMzsk#
z-<ni|YxrHLA^i^x4NxT1^ZX7sXkuIaVj-9^S(tsjQjU}O(3SbAZN}ZGLD)c7UlvA{
zaL2CEGbiCeqTwgM9HfHCs$Wy}W-Yi<oFk~5bWRDrKK9(Me<F3(>h(9F$DSKv?)Vh^
zgMu{yJI^RbLxg}I#Q|i!!~?^yVbO~487yXN4v(E?8rhvs{$`z^X}W3giCxG;N>UJL
z);~ST>==iM0H{wKgT4$q9zrtr5Pm}I7Tc;M#2BUSP5+AD-^&5}yeuaBBG6ixT>ldL
z4-(&y*ZjWsF^EJd@%P;uEwE51Id$+y(pi{=_qH~t<r6-qpDM&ZkGn~$jFhKhu#?~f
z(&7rc6SZ3X?dP6rv@x$ZPCOGsDKpm62wP`#h~h&(<gUrrC?}6A0$9abRxvQg^^%<$
z<_r?7Yi+5&^u~~}HGRQinQnoD%lR~d!Q{)8LZ9>Hf%rs*wSj#8m54JJciipGEl#__
zT45@j45!MT{5#Wd&%iHq%$0jj+0`VJ{9)X*6^mUlOQQK)P04>y;Dw1q7_zTO=GMBg
z;cRU*1P?BE=W#O-!TC~0T6?F({)(wzKBCtB+iGnYpwBm8I`FlRqw&!t2-err9G19!
z$8Yp{gz6|t>cm1JX2l3cBu*^O?1?t{YK_W5oh(a@f>uD{4)7$`gJG^jYymONEPszG
zN8V%5?8AgHcU91?%#8=%mzvLIPv5LAe$9I*xr5NK((}9!H}Lr(1|;OM_(l4GT0W>%
zeb-h-UVH*B=*u;T1M+)4i%ZZ;-D=*|ST0zZj{@ueu}q;jp_EUAk86TxOIlb9Mp(u4
zARD?Xj++@_wB_VtAu|<+@)_Xa7fC^2;gT-KHC=lKQ>*>>3XFu^Jy)90_X6wueAdY$
zRNo&&%6-LWhx@yIW$jXEY<ZhLoT99YoV{7Y2{2Eyow=7RB+|qYT8zR5EMHPWAr&Ci
z7NK%bE#hden&nBSVYz>YcJ|-d(xHppflCSJHXz}tctf7g@rX<7coNl|Qerd;UXzsV
zrlxDqwNm7s=T&31*tO^oO<0gR#*>1nEUc~3E8))y{Q0Z8`gGQ)z1jf7+qLhY3ily;
z3Wta0j)W$qsmb9iu82zk?ZZ2zWBWUk5`5EEp+rPMy4LTHR42u;-#xCe@%Hx5l6Vjt
z)oDx!qR+^lN`iY45*Z{dzxpu0WNUsNGK3|$B&a`T+Eb)L1JrNouv9`)HiO|)G_g#J
zRlXu)4~$1TS%M1!heGVS;RB$LNr2I3T<zO)xJ=;LQ*<VC2%Ejan2J<bKJ2>_A6l2)
z*4y1gukFeV^dtenKXN4SNrroBU72Us7F$&O>bMz%#{UfeRB=)DHIvmJ52F97(D_Rs
zOyZ2sg;+gl-iUsyboabMRH@oBsVBgt<>*du0J*Uy4engtWK$f~?5N%=(2FaI)F<5g
zb5|xOmr4~x-XjG@gQ-s|M3}{X@r#|<8|^Ww<ID0Ij`aphCw>@ism$$Iar03>Ry!nF
ziHa+}-4tWjSkaRZ)v};Nu3>K;g@ykkaVH=dYOT=`A#;=Pm~PO(8vQ4pmIBR|QO#Nc
znA4(Zmk?0LN}gtNKp53@xzw|Kd@+qLrpkrr+<q3(fe7U3?D#2|yfXi5h(yG*qrfE%
z*7r~M=o9B5-6f(L)Fl;M?wpeO@2TNHKVqH(KecAv2#G8ss>pukLW6=87M>gwzFYaP
zo~i*q|4}HxZ>g|843dBm&N#c9T0T3+(dFk|7EL&Fy(V72<Ku2eD6q|YD^UdzStL-h
zTGbV64~B9AKio@H2)&DeVT#WTn8z=`FBgSmPV=OD&xjDD6zw6)=3M;EN(7Ed-aovy
zi*cK)9}thcmGIfOoAey2m2WqS_DeoSMJqqjxR<`wq@t@y7cD&ly=rZ8EQ60@XEU+U
zl{vb?_wjXQ^EmuvIIoP&xBKgqU7J7g9d(3NwdoAUJC=K&MaEZ(ho$Et$6o$J!~wVI
z8fWPw#EFO`x1U3+mdL=e0DkpXGcUuOJN`!o#Q7_@HUw9UyB3a(QU1RGF>qq6aw>>)
zI};694Fu&Yuy6QEN_>N>3DO*a3l3kF(`!vF=&^P{I^At%n<(_!!4|hq)`^<w8(clY
z0O7FA;6f>(+@f=6`*Y}Ab44b#?aM;0tG46oU*(ay5WvkvmK05x?MYb+y)U}}LW3dq
z*9GHLoU;VB^Es^;c+w91Ta25IN>WHi=UF#Qx$Ukdys_O6HQzdrZ(dffMkc{bMW!ov
z<tk_tmd1rI9kT`_VIb&TU1Csvae3Op<DEhfSOJUsR$|W&-VLkwn=i=_q-?IWJ6P0>
z_m8#K#&i6sAVTYE_&NNL#8rY%=M5Up1-*)8-QOmjuU$!LoO9g+{G#keKo(r6$(26X
z4x(63Uqon(D&}$(N=|YVO2Sv#_uRsWkprm0_Di$ldNfM(;rqWIjv|Mu#L_7r0|_}i
zilS$u8BfH|mvLK<?vC)wF+XW04J^l{cTK+<v?trN82h!~;_eB~FMq?NEZR{g#kpJ^
zo6Ld1PmPNs-o(@5{urTLsO6u*wN_GEl<}g;9Jr^2Scf>woc@&+5Lpyg?&QPq>Kb;f
z<@>TJ&Z7hSfGeU*!6o%XQxylJ{yE}MgWD9L{}kypQ>E}`3HT%gqn`PpGLW37kC@L8
zA=k$98LGpl*vf<zf^Hzy$3SX?Zu@h6Pqq$3j*66&t8buNZbyTOql>0D6XNL#EWZ=9
zo6!7|EBwQ>S<hs*)wZ;f(qE7d6RjD9ovW1OCQ1-jf&~rt(#K~ta23<f4oSMySvYTE
z8=J(|dq|9GH`Ur*9^Zc;!)+2re;c-63=0Bz*v7@=?HAD`)KWP}0?MLgd|?yJk#?F-
zEO=d86AfQ8`X?Ef`0Zf9_3$_AI?=X35*az5v?pR}zHRt_vj7Vn9-wz&3{h%^Kf?}m
zzU5DwC&#C}ToyFnOM;w+f?H85j2Z!lIS@+@xlD=*vkk{f7!%|c)&0E*iZw=3o53c|
zuPXe^HLM$}{OHFw_NC63Qwj$a&v3!{My0p|pLy-<ImYhIk^_#`w6PkMMfq?|7&H<L
zM6eSUIDUDa1_%_mJ@G`;ynh>+S^QpJ`MT;y*z9lfH22B<D1XGTyxx|YK0++e*dqJr
z713iPYRitVQBAC+@4(7|K{j<H$CyJ4C!4UlaVm2|4-)lU!aPX>4Ab$$Xq7Cl_k^s&
zZ4mihmr%#k+&+Q6u5Fl=k?bXTG(%d1y@3V<DpK%?qE*QIJzOHt8RE8KY{&AU;Kts6
zwgG7-L9Em3Q>!xkP59>J<`Y_AuT{Ys<3^5%v^UdFB&5sUM5{R0(oM`|UmwA^A7dYv
zRQe}b?wB7tj#oi+fr@c4K2UGImsX1XncF7#iw8JQgP|^OR)@PL(p_pl^v#DZy#LVV
zO_st5gWOs-Canoi4}If+f$*khu=Dxnyn3gGL^*sZ%-GABa77mcd_Q0js=u*pxk?}-
z$4P6`8o(vw0@K=>sEJw}@L&vb=q`|RFA0eAsiy=-?d;__D7ox$#r-Ba4+2ePuWH8L
z^;Ibn^Qf!%S~&@TUgY+L(|8&r&4qk+#<0z66LP$x%SRKCCw2ZqP0$IQGoj<*%DV(c
zB#4sLC3$MK%)TGLaQb~e@4P*Po`)YH9}qNY>gCp==Fg|!V{hHV$=@wRp^2{P>CA<d
zwMU}c)frJff_&FIrNI?M)MtRlUv$j8!AfK=4t*ixik8i_)-}YBXQsR?JgO{tq1t&3
z&)-}p{-Y?x&(U^RymblHet+!;Q1hzexWgN!=87bu8z^P<ZAjkPHKRc(99w29{F<zT
zhpsCgK%GVjkW^xYCj{<;dP|IO&rt=)GpgxO-JQGO(sNU<r!_A<_4PKMAAcz>_sU^C
z9TH3g>uagjX8Sq5Qqr3F;Tlh=ZYo~|jRq-BrN1K5p7b<ITq4`Esz~z>@<%nCnAWGo
zcWfM}>(iKOZ$b$JkY6rWdyrI4Rf{QrSQe<Bl8)40_8Uptvr)89_YP59;a0d-u;3iA
zkj5yc)I?v&0rcy=us3uNuh0-r`L?-sBY>-=nG@4Mk;oH~s4WS}9B#ShRxh~K=8~99
zDOmBZlyw`OHzDv{hxD1o#=PK1Z-d~fM6!2mzsu#+c#0%V=6c6}z}^xmC63yn*ACqe
zpH*S+vUFTn<~xm91S^+e7@U7kpkZ%nKMwhVbGI!z+v_7C%0FGTYkrFCylbkZz9HPS
zN{ho&t>TW#JqExc<kys65<fff`Y1hXk+CJB_1R@StsAmh(=dWtq^6A|N#RVnTP~3q
zL99>|-0LfAS<s>beD)me)=!LzICv`C{&EZ{C)x*a`*l6DDzUMTKN|}MgvfmFPwsD&
zACYq!9V1f`stk#jILZ^#Mj9cF0qE-{Ok`rc$;bTn0Tyg$;peORIo$BF$Nb;FOA<OM
z*_<QY;F8HUYH)NTvhHo#s|!IWUU_b@=l*&Wl~;hve}?#-0X&{a^yAlL`zaB<Q^;O<
zA6nI<dI_>Q{GL5?d*3-#t3yvkZxPJqKXSqZ>lm{&{zl+IO;IQKPDV$M6!_y$^x+Tm
zshxz+h+u!853QB-IFYsE0Fi-&?RppO9z%;DccUWGp5=AIW*Jmp)r=_-zD)c)f3z+_
z70pZ5wp<K>I#?WW5hMwrnQqAiRlaP{LJIqHIY>J?oL4)|FZy^aL!&7w{{!>*q8?%W
zKBV6|)@7WI;gLofs1xGxAC}JTx46oZgV2s_p_*m;E?-@LX$(+2bf#dou({h2j#ZF3
z9Wmt{%&FJg#)k?fyhggr>(=xMKIMUENe8+umgDHEu*aD0;IK5YDp4P)uhBb<|Ja*!
zcC)!Mo|ha~gbN+H-PQ-RCP{X$Fl++}9o+e0_xBGR+b%t9h@rV=ns0UdQFOYgdN@e5
zQ_D5K&rJMN#4DT0$n9nIVJ9KfC8-R>fL;N@(4K95JRd5YxEgOXUv}HV8!nXZDinpb
zYrobEnGFk|eYZXhr5b$QSLYev?FzaXQw;IouuUdYw1$Z#stdfH26dI8I54N%^CK=L
zrS#5=8Fb$G-t%ICX1!{bXNFB=!0?I`471Sx6162$y=f`YZUx;*0aebh@f7(|z18k!
zT1jD%`WM6`Da4_1LXT|FH&<AJ6!fn}>du6e(WwzVn6sL4I0GZWT9uk{!2JE%5OzCm
z%mK&{x%(6AG0QHc?<kL~S%A_Q?sv@nJSMcR%+-uYaeo}RQgL?}W7p#dkZhSeE$h-<
zCZ~sqrRGTm#bXSoXC8ttG67$q8lDXYSM?@bws5Ia!|7yQ{vd;}KoQDVNQ7$*v(U{9
zE(BSRuEyZ*h6%J}aF>pW_G~54Z~rNl_;L?_;9#RB)WzK7swnzr@g^khCM^!mvWvu#
zM>^uNY_Fd`kL|CWR#Imd_R!}>C4E+@%KkIFCFGoA?x9CvYL5}))z_#7S-2G^!f#|L
ziA9|<48ke0Ab3cP3H3e5`$EY~B(xoVl^$w2!tqCZS5F8b>L@pR>xfx!uWlzdS<Q-{
zI6<*v*J%t$XN&9_AJabvWB&oy9O~%6;?QJ4b=saz$5h<X{23PF0C`#$MCvUS1Ikdo
z43qQ7)B3~(HM94KPB(3vfrULlxH_tR9-IxfC7<F`HA2PLC+^{yCUnpIoI8{%1yqT)
zwtA3agOj{8X;QlcW5`K?{@}iNp#<)pJr2O+F_C**uhDHsoGKe$X&{4<Sgg#LBXW>7
z@Vxj>Itx<ydPtqC!3fDZWoru%dQhWvf!k<^g(Aa;_r|Z{vadQ1fCMEP7r6@f<BX4X
z_P^GT)svv;*$9F*5WcvW*B)i#(~t-%zkMe?x9H%w75E2FGL<vYxDGiOEpZXFNMh(^
zs2340gu|swjl*0{m~Q|SIL)tb%DxWVUsW~xw@_7uo8n1FCmWIVVy{W)x}jF<>?6NM
zf1`CFJo#&9PYrs7C7ZOF5D8bp_2ZO_9<Ver)ggP;1kIec@F|!`%(IJHw@}%LFu#de
zkNbIWaXN+Ga@++F?moR*@_ECT8ReJ=Qq}&JH6ljA2)@Unz3XnD@ZZY-e&9dZ{Qf53
zU7z3?vNEj`X>p2q4mKZ6R*GS<6QCWkTA>k6aU+|UJd?12PIVsM$Fxn$L<Urf{vMfJ
z5_VRsSIt_?;d__6#p=z(yOy!36%XXhShi`P+mg&@7VcJ-!pHc_f9`NjrvUCuZ%P4H
z9#9a?GwBRMDM!MaJjB*~$5XlU1k{F-W@wyT;oQRga&8-4lgqqv<CsvJ`q&of!4KNH
zBJP$h4n_{xjA6|^Ph)U_tWydI(IompzsXge<%a^;js^nZKe)&4kjVCvY%}dqyEZ<}
z*)c?$4PP~X5@M9OM23xktTtWbi6#ddy8i!ukY=*gCTts17YNR=>YUj~cOPMQk;Pnh
zk?!BpvH9?&ln`GL%mdZ7NZ3gB5iI&w97MU&OXA#$e%#+6=h`R7cdyCSbAICYk<RV2
zYSFMH3eWa+3BHkt=(xjkKjIzn=Ye6&VR*AilXI^ptvUG-27Rq&XcX3eP|fyb@?$mR
z3&ypAXxUQj6OxP!OBnkK4%pIUO(Tufg`Qp47IjHavCS_@qe-UgSZ&ICDnFFCC%ZD)
zShWe4%HIPhhKIra^`DX*jfiy)x@+a;{heR6?|`B3_A3FdOb%3klDARbE{|a&vC}3!
z*sPyM<x^UxCF#^m!CFM#8aMc`lJeZRTk){Iw!Z?wI5o!jx(Gd<;<XU%@SnnRiz%*U
z(4+f#Ghbv@x?2NLXt~VjcbblEiF+7i=sGzVqQZ_@<AD9XLj!ySuo!$-n5?^P#_8aF
zM_AtmW~s9R!`DBD=}lP=FLwq+D|s#1hme0&Co%8{BLKJQ6T<F;)*K80S$Vb|{B|FX
zGLI2H)==0@G@CbhXhpJNjOQ+f==N|WQ;@JtB{t)!E8BIG!?z%zbr5wv0s700OG4%#
zLQ<OnxZI=Mr7iP1A&Ekq5pBCvUId(0$SoV@Ds1Ft3VHEAv<Om6_t>`AC)MiB?GI0z
z$4fb6Vz)D77w+-NV_ZkyX`J6Y+>o%NlZ`{s)OHqG+1;mGnllNzqYrCJm_E<gxgTYM
z07PKa`?DcP?oW5wUulSa3=VB*HMJrQU~YAn<13B8=q!v{{i(&?>5t4$&~13S&TNn+
zg-3}Lki@JM9vRI+3aHr)qoapSgCqBo+BSb&KLD4SjvH(d+hrPPBgpw-L7q4aGGeRW
zVZ4Elp6HtQYf&Jr4WVOu&Jmo^U4<yepTfz7r-YxBoLm>#{bW<#PJjvXs){Nb*Bnz(
zp;CehX2;fgc_+f#{l>ElN|ap91VY9X_vi(v98rngU(b)d>a^h|asAR6(6IB8y<k<J
zWj`j@oP8w~8P?jQqv9MVrXa&oB>a5oIsN_oawNnQ1fWFwd|~ta$iTxm!#Vv7?rESA
zCjm+kakpb~iPMj)($-v04;PoKT>jYynmXPY46aq>ty!9+F6xRAf+B!<j<_s7at{)q
zesZS>&yq)R*+D$31MI-jI<HrnJAIM+&-Hm%oxikS32Ryz-hg!>E{?~V10-u=REu$j
z2oM6=iB$V+G9x8hanuyo9Fc$?F2u1yM{exX{kNbSU^v?y04|}ru3uzSyU|fAog>dW
zwXs=e>Gc4!Y`Q?+8h?Hiz0sar-XMm(wsD|~M;Y9=!*QvLFW%;j!Dhp4CH?O~5KkNB
zZfBCAGuy5+#h%?aRReX3M;v-P%O63DDIAm(VV7Wuj~icX*GJ+B1WkTYa@J&{d~17F
zoL~DCyoKM2|A)$$#HKH?WQSmODkz82OOmEg>Zb^PG*xf_Kf>JZOKeiM0GJNBPW4?2
zjlrBs%qn<cO$IR~nekWcB2<j=K-#mp-IdGjzabua9Z5335XM9vGdqJwN1V1MSHE{7
zU3CQCNwpn*6EM~X!(n^&#jU(3KX_C~6RXxBSl(=FN`Ne~u2Eand*d@TpdOx|2x&qC
zE`KHH<49mI?Bd71^$mnL{iTCuvnkEBg1~^wGl<#GH0KXU)QU|<7t&^W!E9DyaJSHu
zyrm^b0+`pcePjQ@wM>%4Fg&_NJ#DANO%D*r?v=!!hL@H4wN>p9?7B`*y+tiVBj14O
z)yh<jg7EqzED}CB^M~C)A1Fbi+Y!TYZsN6p(AHB!1<H378mo{=Dx`X^A9o{O#vlOS
zhppEO&Z!wZeAMjI>@!N;7igday;5O+40$|M?|BtbwLJ!Lijj?#X{#%2uqCdw^9QL3
z#t<LVw1KH?t^23iymcxXmeCWg{(Vx3XB}-Z%SF?J6m3&?-Cagx|IA#-me_$_JCtA%
z!Z$?(4g_ns9cgOSSBT}sj6<&S#t@5ZYl(Bm9ev&+o{OS;YB*2DFo91>tJ#@RpG@G7
zc6zxaNz1~1;>*>tQ=Es3=fo^Q!OnjOk6s(OdwIT(S?>|fxxvM|D)<U)+Cd!Dd*q<#
zu?|jT%A^<E!{RPou0>-Q&tkOuL$E!#a$pEcHnI=&lLC4~i85A#ai(CdBv;?eeP2*J
z3}Q?Q3VrDM6?fDt$K)(gaZ#MkCl8vAEwY7<%Q$_i&`V$S;eq8PngonmBROD`A`iwk
zKVS%;T`#M^25ya=gLI2d%V*m#nyp>ZegA4G$K!zDkTdI%$D09ieAH^(pnu+DL;H>-
zDWiRH!}eYk;lINJ_kH0kIofVQ<+oD(rSx+j4Cu1t`0lxX#L<v9<f5wP{vP7KwdzKw
ziUgY9yga_`U2-(ki5Wpw2S}jhg|D~6S+0ssL#uG3Z{&^#TBC@35cX&5g0hAwYrpn6
zdPW!Y(Kj4ZP?1QzHXRjIJRK?2ogjLj%U3IZ|Eqe0o3w9B%|35-I|uV?Nd8uc){GfE
zZmT4=Hte=0A+k#D;vUs!#XMw_IG4gZYcJi-20mrVeMKQmpf8Sp_xoM3AS0Pr=I633
ziS*Gdq(eB{+8F9&A#_i=VH$0+DqAdOAi*(`@n&gQ`0-blK4bDqT{Wedo{;L!*TJBQ
zSH6$*V268^QoeO=4%Bm>coiy+tGN*ldPtyrmSX%J|E?E5WqkN`wN=6yEAB=lb2c5$
zdcgi7Ds!IXOjLlf1)-dP&?6xhI!z)trw4>fv)rEuFZ)FYa9}WAZ1$kc{*udq@{kGv
zU{if3?TqG<_9XTSl!?u@%}qW%<YgxYZ0<}3iyORV!*#l)ZVq~S00l<x3~WYgrt*Eg
z12J%4arYYA6W17CTho!E$A2+ms+h1MY%bfm>m$SRuBZR=T@nJQ4^>>mC!}&qnEr=|
zhb}}G5EG;Miuhe<UJxD1zRnnzKng0WVqJp*(;ZP=@j~ICNmioKDbIiRk9=W}|D9>7
z_R_)93S`T!u~ec`=P&Vm{#GI*o=)l<qS$(6wE{nA!Q|RMF6?0-Ojvf0>VD98m@@@d
z+0!VN&;-wXI-tzX7;{g(j4^&}TOeKc_dtOR^q49&XBBe5Jw!Pw{|QB7(~?KqUsd0k
zj!xTd3JBm1hGilRdO7;k>yeH%Ckdy?<*0iPA^{a{c6jCOLx!fPcOiWrQ8&oQ?L7@5
zhTDA$G+&Zaz<hPdk?~50Z+=*v`}1yE&w^iL$J~gZ%|6_IOwEQFk&ex(T=wr@YG#^b
z3;k|KV{Hg43k#+Ja&}|mSEY~vJ#j;?4vV3}8a#%_65DD(K@m$=b8R&JaIW)X*dbL_
zZ=#zuhs|vtqN@~E^sDBIxa^Db^sR3HM+hpA5U~(1>^_P^n=`%U0Pahbnu<HU`IA*P
zb58)+)DI}~MDGhEH~PukiGrwl#IZXI0Mq0WKOk5zcn4jEz`Lkh)>y#0hF<HDM6j)K
zIYU{Gt`3Icc}yOktn#>Ipn94lq^Bp`&9^WM%%f=vrf)+tKXrF+pzJ;~HuU=T2SN6E
znT3so@cGYB93O@_JCZbqY2O}erWjZ}mO3XOCPQSJ(a$%hQuxa-{?y4x=h;?fsg~r*
zFXqO~40Weaa(WZ3s?PX{f1iAe{xmfs5L*jHHV$d;Sedxsp8*F@pZ%zr*$S@qK|-D&
z5Uanpd_w9@<{NGhPGGrSe2J~2s#IdV*xWo@sNRMDyd)PF2L%Sqgd;lN(RCs*b04Q#
z^siH9%@p=wdVSth>^ru6{&-qF-^ASb)ZQ%}*GaSx>YJ|-onN0<It_}DoP&b^#1tZ+
zv5RTmW*z}^bB_EgXw45mO;Zau@SZNNS|!?x^}7Lk=(71RAO8kybFx$_kjsDDO078U
z>n~M^E`hN1@5L+R!=O-^RSteYboc+gzVWhU%_|LD61ch&9o>j8taOXg@ZA8wrW|I=
zeatFWm0Y|`x>27}wJ;kX(1EYMx^BLV*#iOwXxN|Mg9G@78P=E0j@JasAea!cb97!&
zU6b)JGF6YA-2F{77v%<b#`0dXk!~B~CCtxu1WnXzuHKRycBCbqx>z}&vz44rzp8{3
zJJ<^~^tLV$>lQcM8z5Y9EhXTMyKO=riU?Pm*dK7_k9d38Bt*>Z2dXqL5H(d27OIPe
z9?IAw8}SvIg)ihDSC?g<C7Q-cJf}WSMWpe0paVdmRyi+`rX*I}?YRkSUIW<pwWSTf
zp@T=&kjQ+!=o#yp2WL3w0Ryv?nrx1=kJuer^Z*TXku6W(F$d)#Rqn~<e{scnC6-a#
zC4d#+7xsolz+84LT1MF-10R`AzGmW>;)BvEAC^->Q{7{R0<_2!x9&KkQ2>d2<3r=a
zP|v64II?Rl1RyZ~_*L8=t@{}{FPXsF_!<o0%&GJHTEtTPVE6|OxXo1*o-F(yC?xJZ
zxQOOLe;UO_f1$tuI{G%QgxP$znfyac=d36IfKBSL5l;f`s;?)DBA?mS^zv$>Z;A&f
z90|{tlv(?Yg%Z{R0?%X4es>)18^a9V79_3fP4R6#c<(2u7;A%X`2)MJQ@6eg<|ZMT
zSZgxadm>ASX`jFRN|@)vGjlOU2Q85FS{*4=H2tB%Uz9cTeG^`b-n>}3+UCADC3N2d
zjqWi`-NrH%3ZYb#7dTVH@X1uWTx*l;PM9tq{F^lY>S7baGfgdAF7Fc+qf$sg65yxN
zzHyUWv90YgRHyg$A;NDMYxUdTe%ynFach}u0%{1f)JF1|?hz)lRzpz{@paNrn_o+j
z72n~TLhy9KcT;WqnVXx8u;PXtNaYeJ0G2-14yJd1t{qyIg8Fv|IGp$X4Oo-=ZxlGO
zV8>v8!UbU@C#}R7u1skmQcbnMo#LL-V^G|XFhoazD$6tkAYu8$q<9RLR+<ql{bs8U
z3<W&wX`=9K_z}5mL7yka8nZzHG_z5{V#CGCu)yu{yv}b`&PQW?5NR&TcY6s^v7@fa
zBaGqUM*}2C+v0S2-=ob^3ts;fmli#dQ_0IW@X0d{G7YMg{HuJYuYg{pfbY!Ieb;Uu
zsJi6*=51)6cu8E{E>2-wSTUn9K>BdC_f~Uo{ml|vcr+~j!nIdZ6@?}#aOS1->_(3s
zV4>|(M;ATumc}?HdOFGF@7faO!h^QejVZ~U>FZpF68HVi*)BC2MuVNA$J^Y4+p8=`
z{K!HJMp`D-<>egGLx)Deq}z0Bw%$2-+wEsjf&8pQ>J7E3PO7$Y{NhDv=!mp(s%tuK
zXpWg}p0NIYks5*ywTh3h!3x?~V*rw0(>WxYsokeJ^xy|pm-=)ED2VTu?(UaVe78Ig
z9ZRTsJiWgkX|LDZr3e7M=yurO5cAfVy=Vhwv}(z{esaHByuI17yrcDQtr+na?Rmej
z;Io82c7qG{a|Ekqz2L$Bw)j_Ij<ezw>5jmFj6gCH!gV2-%RdF>4Ff<IW4?hw<T4y5
zw>$PsY4XUO8HiZmTamFhD4L}n$OB!tyQI7sHqjIaU|l*h?!+uk?(3(vB}EDp;6VGh
zz8Bmdx9V@bcVPqG{}|cckNo=(4@RyI+9FhG4=$x2XfpudLj0rBTGoP1{J&%dJ-?f~
z)5EVu1(g@R6sjdOf8mZnv5vK>Z4t4weOg}*m-E%ke>A_j9QGm}{Q^f}=M=_kyizOm
zDJ*W|o1yj{lGo1*4E}QtSQ(k5V<%QX(3V+}u5Dqyc#QodT)?{;_x0haFJfzp`p*UB
zo+mn+q23Nhml^F)(-}rezhk`$9};UQME7x@_u#cSMXTl5LB8!d41Cae62o3@#=V_^
zru_o;6yZ&W*wV#MnqbB8DX@FE8-D!epu<-&tn>YmCu3Z$X>JZP5AV2mlLRA68!>t?
zK!h!HvGiw=$<0f2&X2;c)59TxK=4=p$H=+I4|RHU{w1#wn|5*%NgIq7vgPZ)S<?h>
zSPF9}%0L&`q$Jf34*+1x>EkKHzG-f_7_xSs0J<(Db(mW;34TTxw)lxcZ}kfSm+@G^
zQr4qK@$!G1n10uzL}dr3bp!%>rtl_6^Kr)SLC*!*$iQQYAcGIJYoDX7lFvUxFzH`B
z(c81u3K#Y;yWv{{rM=SH-CvS2`|@A>Yrz*N{*Wa8OJ>T;mdwMpnX9i1k&Kipu*cF~
zSgBcs<s^nOlvi;%b95AUrs6ehSXd2I#`gT#3G?e0)*A^rcedMO^!1jtpB_0c&<PLB
zoN%<L&5MfwKb=GE`yctvJ}H_bV_KqMe~{;U(?fRqd(1RlsQ#jgI<C9p#_PKJabqHp
z&gmJQQ%by5g}ja<kuRtJ(oB(G^=L~I-rn*oHC`d$D^Yc&XvA`M3cr_7QYWNfi_E;q
z2Vp{Ls&TBK5l25$52>sl(2Di*XTZWOM`Xno#Y#=yCn-R$Uc;W{S|U^ZwLeIyGzXH$
z!%7q#$gcdZ9P=&YWFUUEI7a7FXRIBy$YUPwq!sl<{dX^Z;v&E|q+TA`MHU4(>|iMC
zNO_VedO=<L^69w;h~cS8@?c5vaORX})Ydc8F(3sN@miyH&P+IqZKop&Vy_q!mNa33
zcZt}xZYIQr1tMAH3_BVxX`<dNyF&%i&djd=Vgzc)=3FwS6Rh=!JqX-u(xoYcLC|)i
z2d`HbL!Y8jDyHWjLAH=$JN~XF#IM3Ib~aJuKTSq`abFOXPuyX;yZrh4_$URX6)s$4
zKb?=v^$U3k<4si-sNh>aWt3(kxLG>~X<k@QubW5oO;bp6vvvL9MYVNv47H3OLOm$Y
z)lfmEx$58jT0`HnzXs=pYxf6{>5c4qW{faQ!_0z}3`g1MZ`CD_v5<}Gd54k^a$TO-
zP?xuKWT(j*V?<I5KKe?S#2#l3x&>TQcx#GcJzQKJG4n@VYjKEvOaNdNLLh5-X-1UW
zotR<7J#c6N0Ei*_g86&|PNxQh#1fwhe>tMhg$mkJ)C{aAxZ!nyWS6yKs{wAZB0<R7
zon9}8F|HaTS5iKYCzXm`OrW%sV0Yd>0|1n`5^n=J50rYw*(h4xe8GH6{-%!^n^>bv
zNj0<6XAwdBo{_<&Wa4@(8;17L%-#_$fG7S74{7*w8eht<vG*crXOO^SBr`es#jJAb
z5@81fm~U?L!9B&}_jKuoU`)1aVqpZ|nlsJ$zm>K#URcHCng2m??V)PoUX_Tj$3Y%2
z%%0+Oo-IzaW0Hi!OqmP@rNtDNQq91`hFSGWs#7|9iTC*7xi(%Ye5jhOq&tH&OG6I5
z4?i8a7Uc7PvjDWVW>r}6%(rcFF*HUEv9pIZFa>pU1%;vR(L+5rUD}S^?FPd~AJyDo
zd_0_rh*r~E&IR0<iLsJwyc08lFVr_QpWiFTXgQ=*AS;SQaR?2Fr9&4FjHHR`$F#E+
zzb~SOQ1`1WDyw9TSvo?{cVDo4)WxMH7h4-29*$1cLtlSDXoWM7UyQdJ13>;<*h`}2
zGi%G&>2!~|TeUO+=k284rqEkR!Y6MT0tZK8cPFfXuBmwYc0BFGZ{ho%Ixc(HrOTjt
zvYB=k<F7pf0a_+_yqTdTkyttSdtsXq0{C#*PBHlE6Rn9V!oM{F8_xm>yhNZHu)CyY
z%wUD~!99oX(l#+*gZqxtOv^P-=v;oQpGhW$=;Fh(z!+9z%U6*;o*d%^$~ob!5_v<1
z_ox5JW2zd`|2b2+vwFdn+vSs#TX-5~k}wXY2V!~^1F7V2?<E~U(99>d!hqJB{pq5o
zv-)BjxC?T|g#F6}APv!Nd|ZltIeN3w5L`dQ)>!b94|#X!_`Gk1-w-zwXm&hNWkHHe
z#AIGv8p7Awk{qX6?81a?z7y(*SCm8^mXz@xwDuJvmI6%uO1E{Q9lw~KOmt?8IX=*4
z%nB}akZwr`>WEq0eOFG4{*Xdcb#XlxRdDtD-@E>~esD_`VcBi}A22*GO`M|26LYAu
zXYUvIB;qr8Lmv-u3g}czX?O!%K$l%SiEd3Sk@f)RCTL#>-ps(TzB+BBX+UU4%PIGS
z7d)Igc_Dj<Lnl%pU1hKrm>o5SKLv_3Kwat(qlp4~$(M-FPuX84Oi{2l`{k8?h6ea#
zM!qUO4o3AVuc|ql2Sk4Y2SoIbItzwR8FE_Y7WzgFO>9YlM?M-W{oXx-wAwh)ol=_g
zABayQRgohknkT2ul8z_<9QDZ0-(bF;U#KwXs1td-p@#8;tzMEWC4cGRuGN{tHBS7?
zRlg^tR`<z0Dm`Q?x0nQZ4`X<c#eH<TN*fKvzW+a(zJWW>u5J2^jXOr8#<p$Sw%OQd
zY$uHy+iq;zW@9vV@?Foj*83Y~&4D>)&o<2nw3L?LmZ@4MrM4&-SL!~ZtJvsHHg`Ag
zT8=Nzb|v_mHpK9pa(Bjqfgcoxe#so0+@uP{HLcyv5Z7Wee+327aC8t?+|uSyKb$$I
zx{AdcIP4R{YS*unFeLd0Q1vXMl|uh3JCHLH*X1Slm~3IQ75q=Ij@6B$!G1*%S<#M`
znDhhOL3+c{En)FF_FfT5&2nc~$5z_?_o%~dfT1CG=JIZ9RZ-P9$~C;KG%O;Z8$s+6
z9m5%rk<@8uJ{HgzhvbAkk3fzT7|IArZd9$pmj6%PFlv)lRyCx04;6wMqVN!egwP|C
zy#xO!97vS13K1yVJCva(o>~os2DTL{*hTU620I?0mP~d(cCyix8IU1_m`_{BhGXO|
zG1-p2e;}vUh=K^=*ni%}jynGN*KxMzi5e`BMhjV>g;Z?x&z~UN`n(KV;CCcm78{_H
z4~6{hDY`G)B0*#7F+Y5%J3QZ3=r#Cyg4w7G&ua!2c#Djw?YK==ACFw|DyxL8G7|tB
z_t|WrZJZoJNN%2+Z<j>=Yl(y*m9!g?0{ga_b#mw8$ydt%l)b8<m*dqXEn|6a^u=7w
zS_xs;c~~qp>|k@KRLq2=z`a(^yF=G&0nKxiHRk7RX$Lw^m(&DTEB@%NbS*?FP=Kqt
z-K}9@2amMbsIl#=F6}+V!InLoD1p6CU;klQ*S&(iaob_w+;-lvDVJKm^G`Dz<r8lq
z0RBgRs9nw1+sXH;bNk*EgT7m@!+tgvwjV%X#xl@n$Y6O2i_^%nktA03DP34B$E2yE
zfFVS}ga962HvZR^SQ5p?l|dt$Qx}i1IM4On$rBvNnb#_j9RzJ6-||Qemk9fa&s2K$
z<piF=0o$b25>(if$c{EUb%AG+{55i8fQl63#BQ=_M^T1symH?oJS_ic7Bp~)N4>(D
z6^qcmhSJ>8f(1$tr-|=6C_VPSwZj0JR!<$!0HLun#Y5RA?NGF09KKzrpR6Jc_%v~R
zt?8{n`9%U0AOoqnE1;<A%g6H1Ds7Q#1M=^O!l6XK0Y9la1`UazxXmpJRL|W{rRaun
zhD-jy5X|kSOcOrYBZzVGnLW%1`+Mxo!b<*pgbJHh;qb}A;wvS)!cW6vr63(i#*3_Z
z{`y~>TS2)Mr}MDI0_!<}!~)HJYEMs9R`HNW+ia~%vG{Fw&Us#2INtKyx_l)qw3YEY
zVWhD&DKjP;Rde~KCb|YII7s2n8>mnC?FR+m$j>(UUu2$P9Fs^#DkAD-m-a&GCj=>B
zmQgMcV_N?KzeG6Lu)WNVdLP)zgblBT@<t-@$b;sZZ@rAr!48=B2arDhc5?$Z-8DXL
zKpF|uf1KFRa0116-*N$=ZqXJpxB7bkpb>VV;(3~HjaNaJ)_jwJ0A@1caBNXoCXDQf
zS3tJ7uP7!!?LTlxySp*U5ky-UdEP=*FtOsEL)GT{t~ve{)jd&A_6qzC&R9_FXV{Jf
z{~t|Y7ZSFi2MsHlm+>vX7lh6q>L(juH?@jV*~?%0X5Zn*&B!CYcl_VQQXL903`I*&
z??A=v<SSK|!)%aNnlt>lPUS}TN~=J*vPXN3n<!WMK1HLF9I$)YZ=UEcghm}Nhd^;I
zdw!2!n$^?c&*=yOhsj9>dQkn=&ur+dkJu`~f4LSvCq!&r`V+`gSW+9Maq7WB3fwio
zZ_XZq5pUZ6*n*S7`!A0lV=bj#5?nj6(9X^Uy_{mT1(<&;VhQSnAy2krI8-yB|C0fw
zxzW0==#a6RHnQE8;<jefC~@YSYoU!N{VP56^Ul4lpYKwPI~7*rnH|@u)bU^tFSYv|
z*IFG26yr}3HIf%gJ*F?aKRxGR{9!+nPB*s%)*-D{0K;&71Hq9dytjSyfWKt{W8xL;
z=j}5vfDTxl#-{Q{Xz2jNZT69y;mH2%4k)D9Cn^v?ytu1>+nmav-ji0thHZW;BAAf9
zbU;6r1YZ3hkI<~rn~s;vnbuufJd3lqlO;&P|FR3%PqVaIj0aBu3tadwRR9xjV5I27
zPJnd<2vpG)_S;0;lvyKzm3d)(?{|0K)Ox<urKu3HjVRLM3nwL6{a-f`{?)K~(xb=F
zI;;T`YZFVUim=F1b(iink=-E768kn~hfJRFW}%!lHq9%WNEoEc=$%~Ge@-yDlM^3a
zHs!N8${wv1Wn$W6k#Sm@iidJZbx9GEgh_?IyD;a++2Aa<QtFyZHghe)eI{}mqV9OD
zV1xzJPMAMBycGEPo5=q6rAAs4ONw{7W_rs{@uE1>%6v+fqPYRu_<%Lda*0U$j5KuN
zaTGz`Yp+2|U%S269^(5vpJGGFtd)53q>XMBXmD)hoZs^v1ggaV@aqS^mByPMA)`bz
z2agvVkEBcj6Fw&Ttm+MnlUKSd7HzpJYtVaFK*tpF+!A33QtX#Cu-a~>l^1=zfxrEH
zEB-N^nW*p0Z8rw|%i*3qrK1OwUe46~tlb<cQ@Trhl#|cQAb_zXwzi8&DsoZ@6W(!<
z=s3Hoz)MbF%5}zL4X=E-O}O@!<t*U@pMe65HjQwM3BS|-fyV=83AZ~GC(T#JSLp%*
zohB`humE>rT@5^-mnV_355_L{Y$<tv9Acvxdo@L}9&=I~k?Ma7{9pQUrhyNR_D#IZ
zIV$fhf#mEF4XLWu!jOhmJe=>va;VVhU%9wyWabYMAKPSWgehVMw(6mZj9?9~ykue?
z?bq6I$!O@1W)GBK$KNiS@OHk*>V*G00o22i(b>w(NgV&csK6bB!qh5!YsOuBA@LNZ
zYM?fYhcpTGcJJ#9z7r2qrrvX)9!rDeT}76%9y>Vq@&1mRck=JI4LCwIn}n+KDGZGB
zz{-N4c<-B+#)+V2q-2eK%olz6+>WAteuhCp9NUf!ZZ#7SnBVaI<|Tu#8+Xyg6>Zkr
zxdsxhG4MNo+b=mA0^Y78=yKP_&7rKkwyAAez^P(P5D|#1*VYbhY$$nk?eEewH%Gci
zUV0cx@e!xP1$}`TztkBIqS0^D6<Nb5d-m0{^Ev{!%Q$ABj!)aQDS=$GZE}a_3(6Sa
zS_a`$`q=sF+6hA<;lGh;u(T@z>09*wHlBRZxJO_BHLVc|Lm`cR2vTMUV8AL$N$aPJ
zzi+RJAZsfi<R`ZT09Dpb7;T3TDqTG8WH*nNsb^0}TW%RLY%k7cpxS0Pm`F1v!x-kL
zB3O6({>2@CD@f2=Vk5d&p-o{dAG`w7Wkd;ENY{R=!c*00?zn3cwk1@hTHDMX&{}x3
z-x?9k8oW3<*qhK-0P~;T2S*kgcr2S6jOXp;U3o&PNdq#ysNTEglj-=CRlcx|4`iXK
z)JM<Ho>GLIy^<xw<%!1*E;`N!HC!K670{P}__md6^!^k>o!i@G*-t)a|Go!>VkPm$
z6A0$Tt#LwJQ09HqL${w*0OQ??(hnYOOK2+YPAoC79{%f<Rd4xZ)mp*j(;INWq(|U{
zmH&*0+Oe`k$#B~q8krt)+Xdy81s6T9RO2+x1AKk=X(dtHltF6XJemBLG4LzR?Z@}$
zA8Ki}sdX&2tLbk~f{9DKiO#Kc<PbotBV9IVfdTKe<$1>YXg2WEB`jpL=z>q8ScU};
zoN=cZ<AZdK9sf<v2}4Dug9;sp&fnaqa5LsLzq=XjtMX&KFGWt3GdF+SA=>U1OlYU(
zG5~hJ%Dv%V9O&|3dh)f-d(jlBHNg&`R$t-(z?*+_qPqOr6h1G#e@FLW?K7qW0ZT9U
z53v6!mc&oZWVTb2$;oP*RK_)(f(*amov@{lTfZN#Jg3%y7KG|U+o4L{Qq>uH{l7M?
z6Di3*dkkcCvU+~l??8ssNQ+rgCNl+X*z~FNv=tg~u!qGkjp0gRZ+M`hk8wDIi6^`W
zYveUAMWs3+Sq1uW`I%jVVyl;gD3~kWw7{l{7Q}C_U^fH<m&<xv3e`>g<qS(vDz-0b
z)Q)Dftny=i#EdIetQ)qqpNR^6)l}XRpMOEh;*`5@f<rcnqF<E0*{o6Nk7j(In>kCV
z&&(@@0fgXp7Jt6i^W~zrf#Hj0lj8iqjaxR#7u!R0xcjqFb}-;ODeKrVMX2mb`sr~f
zhydi7BlSk^Y<5-BK74Th=8aWt=Q+Po>-hh?!X7~qh}Y2pJhK@n{H(uiDV6hQR3PU>
zH2%v_T}pg&kje!mj*@H9neWs7k6cy<iKp?AwYnNbcBNv|jfw2z&#Xu0%RHI?%}@IR
zILRuCugg)I5}dfeFX|q%$fr%yy%%O>5KbqkuY;<2Fn|%Hk%2ncx7*Zm0|NCmij5lV
zf^c_J^WZN{VlD6a1?G<7RT2as<%FtF1wwN|y29RVcJZ-(k!BIW$_rjyCDstmwoEis
z#65)1?nUiP_ZVY7k3qGDC$Y5sRqc+GDwTAuTtDY1p)J(Bpeci|XwkLn&HH2SIP(ug
z$n8MEe?>9?slC27Hc`+Hdv^GB-<6n5f(o^{NXHe6Bsifrl<a};uSP!iXXe%p%vLSY
zdNCi&7Jou}EB0GaL*x8V772*cy0=yMz4E^G(yNt(BzddXqBF;--p7FUs5Z1N@*a1v
z)<L!>Q}QwyTv|V^78`)+Zhbb;2(Txlpjtg#QR<73oh38XK^KDf9H?D$$w&}4$<sdP
z@G<!h!-D8z2E2EkZn-Zj^Y487jKz;k<B8H62b%^Eq0Yg`76y|+9g42>+sr58((jV-
zTEA-+^B0@Byfu~b>tixo<W{GNf}Gr&uCJd=fUDUwqxR5qlN&RQB`ny!#vgZ1v!@)z
zC=7z1sK9$`PcpZnBQy}>a(X`NXZ#RiN>2cE{wTnwe&5)5D>oDOkA@CxpivZ8$7uR}
z!dU-kqE~wT*aF)MZ|#xbkjx*tU^Y?pKV=hfh_r3!6rNkUxe7V+)djqs2*yO5D{}C(
zE2OT1=x{;P&8cC-c|?4a@WQy{Bd+(+uY7^}>tB=~weL-uR#~Oe<@i{F{A<ud*42ta
z#pmkNy$R3eBEj!m+-0Ppdqfua;M(Jn_~f{d;0nyyPRCH}5K6^$*06{kHr6`KX3-O_
z7yXUaq2eo(BnE5)1zk*zB0PELGp`iMoe@k(2Bg=o`X#u%@}RGIUbdy_yn^Y-zpH5;
zGiY~Byf#tdD<aJZt^retBcp8^B??e}S{$|j)7Iqg6<qtCb|q`J1^Y$Rz6fj7YgTH%
zlk86kZZpw_oXB|du?0%G@HUo4l2NKE?xQ@YKx^&V*&Y;c?%xxQ4F_1m+AM&RteHCx
zl!-N%5!i7%Ld6-*KP}O}g$#oZ&y%Ljs!XKc^RQ4YM@GMNTd4-C;UI)O(uvp8IXP_W
z1yR&f{|<Zo%EU2L^v`H485=+)05*es-p()FQ8K`QgRN)LmPgDn4(wE9fTsg&pu@)4
znx#P6=ePHILGJiwK$YN67L)7nu>L504Y9vF1Su;d=<4_W%85bt)R(hRKwI<Ba|<V_
zv8@&XTeDAtin_F{GY3U?tZO8cRu=&VRfo0m+2dG*b#bwc?(S=c0;!zmn$4$CJzYxx
zQsnLl)OhUo_JpB_zS!(?{`nPG9KOj#f?h$`__%rrY$~5B8_aRxVD3Z&-7qZivva!t
z)<+ERrT)Wg&FNGVe#H2Ne=%wE$jwy$`yJzAUM3WY6ftW2K>c~GvVOLov6^&Wz)_oo
zZ%NT~ikP65H1$h}cxi`74#^<20O57p{}Oz!^Asa6UMDdL(Jl2=86q%aB@AavjBN$b
zx0mxM={)@28A;9Ax?9M+h-q48|MSO@@d8QB)D&W&6QWJe7a4#!_4@16k9;Ebnuqhe
z5Xr-U#dJuq++C)@40P9`r-5M)X#$+sCs5PW{{9YCY;Jjnm3*)~t9xsv9xDugYpa_5
zd!+Ss)I*F2d~6<CS83yleiD4Z^R9*`>N+A#eRxrIT<Z-)@KEqHC7bR7fexou2d*}(
zfKdrSXo+%eFa0?+9RaZTqvsp-&tfWtgri#Mz&yg8itAc?@<!<c$CSu0S_OAj_Ifk6
zp{bPi_;J|0%sp@G|G{R49o6uh_2arCV=qGHvg21tKun12>d%vUckLyOW6N>TR#)9L
zxt}AX>H~t{6R&Dj9QR32D)w21B+>_CMM~rx+8BjQFsI;`_jx6YYb95}QWGq~BF>9c
z60R4X?GC_(TSq^jBku2ebLeLTMDz!l@!UGQ#7+EB3qU?MHio7rNAskVJi-a4wduSF
zS0?uShCtsi+g$c$uyescv<yR^I6@88`)5TiY0b0JYyD`<Xwm-i%j*8)S1MS!-UmB{
zO2)Gf62QV8^o{=R?O14kgleN)_%>-wY9ccR85XZ6$=1pQ+$PH?yWJ!8&8b#A$SD5>
z)Vur|+HOv^=r~_G43~qu8k-)xxZl_Hbye5W#P(PIVKI#OCOM-xuCOuAPEVnIC_<?>
z1y;1Af*bj<C~p5@Pn_l4>Uwn&#8?x}3o*Z)Mb24Gd1Zw;iBKiHjp(d8(VV?;LHLN@
zjCQo!s`&A7cG2x3{Q|9f&)`y+2j|FmoyOL)&dS|lvHL*#>B6($$@!3L9e2t2>h3E-
zNGeg?uHM1waToV2*R|-InSWNY`Ni8su{$KL=GP$~eumrF0?ylA%#LUiyJP-FznLCu
zAe#Gu_1?n<4RC#lx6N5~W*5<XFN#_6<5?0IA-44W_Jdm91WW_LZP_oVblT+{5;iA-
z;mFaTAjCzGu|=ud?+Gnb3y;lW&0f{BmqFgXTPe?^3lEt+*+%FWV}S|MnPm_LbFS!X
zhoX^fx-9#&lXH`m6;T$c1gD+=>5;~+m(8b}cDi%Cu_y^GiSTU)y?#pwq8o4H1b^Gg
zRli8IKsw%{ZdD=C%H^gi48EvN(iv)$+Bd?0##KH78W$y28Er?1QEF&L4@gMl12||7
zP<NXubfT4)IhRrLuv459@){#s+?L-B%^2uOun0VbD)d6iM(<2^H26bNr8sU?758Qd
z_S*+5`1!TUc)O)4dgc1rSXm=p>F66-`8WK#$epPn!g4#hNGFPl%1&=)871{r0fH7W
z5y|D~dz&v$GZep(E;^KI**fkA7*EjW?g}H4hWTCj9Itr>?;3d&XzP18d0QUKwCDXq
zf(M08%J<Tc%*BlxyxX~nhPP=w%Dh6V&R-fPZJnB<rgDo^DO{3;<v!g6FH|GN{=x6R
zEYoLYwWvEz4T`7b6RaM?Q&6-(!r5+MPk!@ie#%!TGqAhOIQZ>9mpe^(vvO6h`b7-Y
zGZEKi4pPCD1IP-zXH$<|dadZ;d{oi`Xx%O`xSpfXgA_-612UBfNX>v@%n=vpwA0vG
z;l=Yq)Z{728Zyx#Jo=*~yZ~lm;dahRmsZmt%FC7hy?9PZaMCuGHfkNlI!2Z10q(6K
zBC)H9KeJZ)g$1@{c+6z5RUz9fvuBPV)R@hdq`-v?V_V#W+yeB&*-M=^8~d$R{md=k
zX#+m7rvgzhYv;O|<Ov-Qj>p-)szM}Ck$@6xFp_TEYo{StN5>^Y0~W$!?1R~Znq}Ej
zBfM}42{8;jg@Ux(B0O+K#p0i%m0wIiwlygrNqL&co;iBr(%An|(+&nqlXpL^IDUi(
zJU*QVNKH3O|FoK;kKoUeI-$Kd>t5H7ytw%&j-C=_C1Z2sAY#%hhdCr!Xp4{O-e<>~
zNHU_Xh{;2PyIgskIU<;AW+~J;N<z7@mz2nZ@+nl{@F}fZ`CICH;kfvu|2b6tC}qjT
zh47boj%qgWtI0<|*05bBC-)VJrUJZ&Y?k#1`I;qccv_~q7?DzDr4r>Hf#)p%2^i(X
zFbfMD`_f=r_@NF)h7V7}jzoCCya?uiPP<-lKy`T!9afn&639qqW6=Z6@t2XbC)e?)
zR{X5;xZI{#_`tj4a^JB1ADnbRms5>Z&7yY6Rh?9EljBXa=Sh+`ZY-a6U*%Ezh7w=N
zsWz18o3C*k<Tu!6$^Eo_#{c+xtJ`0i44qmgqUs~`SlzR#a|j^aqx!31eQSn9qTSk}
z$l5gA>KUhhF1zWv_ohQtjaGd6MSJnE5x2t$Ofec1ems5mdJp$<NJ3WXUIclcXVkQP
zWeI2aMmBUGQZsT*6TMi2uqDiMAHHE+#e8Rrd|(!!GvO<R*bmf6^znWN)z5>5U$xIN
z%YWTB0_iNS{E||W^SkQ~@qkdf+?Z7#e3gsL7HFoQ3)_p3OGytcjY4?~dF9M+;{t4G
z45=A744a9lHDGe?co*5sGW%eWkof72zgceNAFF%YoH|IXVL6?s99`H0&&J1lj|b(D
zc44lgnp-+DxVY-mN}vNd+*DsFfmJKyqg=6l6dJ23b{9z#P!>7A$o`7O_4Ph@XI#ZL
zSsVmkYr1Mr@Rus`8XizE56<^}C}JsP$JXJtUkc(le>C759=uPocW`q>*|-Bjr@G%B
zR00nMeXwEYA!^6OOqisg|0i9d4x81cRjd5gIQB0C!<rE)^^fvkr2HS<GS>M2q6Y=_
z2YxoEbvo0|FOlVB3pT}*^DeM$6R<?JlCl5S#^DRk2Mty86~h)1m!>^3%+x+dx5?SE
z{FN2l{M;8<^!wD4?}BFEQt12-MCq<0=l7DltT8{DB6l%Wgigl7Td2Gb_k2DH=S298
z!uzS$c4&aL9{0i$9ZcUlt+n-x6!Ue?e^=ZO-wO-LNgWJNl_e(txwNbdyAPeFSo}65
z01L1%yBQ3XY}kxR{-I4~PVk<z<QH(EAf1K?XwIoSmb1{R@5RpLtj>c6BD$9%TH&Wv
zeuIJw_{dnt&&dAUPBIqunZkRkZBJa!1n7zIfwKsBEEQhgqCLUfwGaOb%voy%FDzJJ
z%>L`pIRys12eLN24H>fn#Fyc?djrk<jvNb!^8lzQOeKso`F5`04glZ~Kw+ew0P_R>
zJvk38hL?TKKMp+7F#WvZ4j9S%y~XR@wR{ACIOo_jE%J1O-c};I(j-YeBlvTrgiPKV
z4WVv4f%%R9pWYtSv8bg=2mwdqf#@*Yon_qP%J=3^YhOq)svJkbOTkU}+SW<Qydr}v
z=#2h5cADY`$^V2rt8DZC&I@>M^R>dt^1=xoB3+ttL1s6+>Mzz!)Z`;_tt!WIUpE4N
z23{8qi<f06!_=QX!pze#_8HM-HXO?)o0*v8wa$6==gI^@%sK6~j~b*VujK>~s1+^?
za_6L-NM2YKG!*3!gy@(pdxy0vfDu{s+@V9k)hR?*mpUVE4HG<>=)m+*V{&Wi0!jbm
zQ=ly5%H0@=v<=G7({Za8#V4>$sjul#XSujgG_a~C_<=QD7C=uTmGS&7=NfA*q8JLe
zk%?Sg;(jhRJT^w>5bZY%C>g$R*|3}lej!-5AlLdB(8d93Tz~R>>r0RAo<QGR>DOb}
z7r}_Y4+BYG%(8gN^*hLb5bEU6Zwv@vUr3xcv@y#f0UAh9W{<+a7!z2~kqJTscx0{^
z2J{auMRYcQ+-WO1fho#nmCQWZbdp2kZ0Eib@bS8W0p)$2orgHNCf2Ol8XZ$|RT<&z
z^~^7!2D)Vxp7``4+ZFx;QYUgUSr;X}-0yl8{yZt$8bti!>4Ptys8y#u3)x1aTbpfq
zFvi|nxK{8p4ZW5XIKftpDO5WlEuvYw)_05auxcz$D<Mx!CWk2@Uk}RLI(zE#R;A2_
zdThB!hb|o!`#E?G{y*7^@bShD*8V+N7}HBofF}0ks|J~}ieUF~2{-epaNU@ujS++?
z(9n(s7yP2j;+-6UkL&Q?AI+dGB5QyNJ!{VCC6lkatrkQol8O~l$9dZ57pV+RrR~&E
z-96I`EFPh~fB)wNaA!wQOi;DzDO?Oze76MKJ*a9F3VH3aWGvPHr@qXklY(s~E{>%F
z+ScUq5tccE9%ZrYP-PK-{e9Js(o#FxN}2PS!H2o0!$s$zXOkCoOtA!G`AiVMj{n%=
zo<u|c*P+SxHyu@5rPOK6Ztb%2V?z9}n+~`TMPlgq+=|c!dHY2}M;FW+_pD36fK7Yo
z$Wi<Db1u)qAh1B~6i`HpwogmoG|e(bCA>G%Y7qwA)+hLu?KtA5f@L@Z(91hD04b4N
z#9(diBqQs%G$rNCSp)}1(I{7X=dBac`5*n=eI~~nxHeq?tQ}c-+xS1Vw%GFz^}`I?
z**cZcvm`;SK1E44PEp$WFck>I!W>n!J*at{h-bYfHAm7|uPB1~8ixW&E$aS`Rn)vB
z#w_!1BB{+{9^zfki0&Z_L3c-(c_e(Aisr8Wb`VSok5;C8Uwye7a!-5~H3i&U8Qf|e
zVX6yq>4I8kBOLs#7L01PtHe^c1`W<AOgk|c3B{uO!S8~gfn=K@w;l;X#Z%mMT`}1i
z@w3rcIWY&G0YZ~2!2`Jk%<QLpmTt|^e9hc@6k9O;Uv8J%yEEn5i(s#Z=qX^olH49Z
zDvi&d$L;I^PHE^s<|a?A>V0H7WHeK2uw6Fy8@OPYi58mm^R>Lde&g_?wbm>o4RmGg
z3iNdx=)h2i!P-a78S1nuqaU=;W&;Tc<Qe%g5gn@Q=;N$?ga`mxLQ-@P3l^|CQiksJ
z$A%+Hb`7FeLF~?g0Vy#(k?yxJJR-oUw08Enc>RxxJfiM%?W!qbQ`#xoU^JIS(#Xbr
zr+*z(2Ay75xJK+W%Fr~ROS4DMH{&W>1>6+09FTICVdOq8=<jT<n2Iej?Z!N|nOqcD
z_}8Rf9<L767Wh_Tzbl>1+w)mT<n;`_o{puao>{`rB)uX>^572;dyzDE!kiAI^%WkK
zhc2$%>7rhsT;GEJq)LH%$sJ>;cs);<lkDP(R%J|1;hEaOj7(praisj}N-$>^@?^eh
z!u{8Dja<eXh~pVujaFe3tO3zQAw=j9H23&!B)mDZpX3tpwmv3)V&z^?!r%Gig9l71
z;Um#5CwJ6U_IPOG)(Bk=O9V`p$Kw68j^3@lHkW!YPHCs;A~x?xCD(Qx2Wj?M(3THb
zo3yrM)dgJ*Wj&{FLE8E=r@2c$ZjT*2ve1Dp<v)hTvDg%;X)>^Y!J*Qk6WBHAA1xnj
zrT5Pi){vC`rE!B2`S!*@LKAycmkNHonc_&RY%H;YExE|2Tf-9fA6+ksn-D0#1ozOS
ztP=_-0V4z<l{<MG-FN{O$nsy(Z<Dz(Yj!g>E|5DW#*WN*hqMCX_u3hh{$N?HN6rfs
z?aA<YCB+5^B{eSdyg$v^_xm7}ep1oxgtnaiSuC(+nW$r-O0t9JYyDhr>~a9bNz>`Z
zF>AamOr?6CE1xytAB82JVp_o%Qds5t8t{&<2VL31&-5!J_C(@TsLzkT3kzA3E%=r>
zkd6O&gn0_{@G?Liw9pyvNkr&*%o_9)nr^^{IH2+?zPx*c_^W1UXZd^`>f0#oDc?Kf
z565S(r%i6Y`!|-s+?6IX7RGtnK3BQ^Q`0vR|ISd)lA|ie2Nstp3LDj>Ge^qT5zESq
zPVPTn5zByU2K*O8o(K!Xdi)et)lbkeNN{t`<s8CzWbc|#W!C&JRN<ra7|c$at7)XJ
z_?-X}_(|WIy>R{$|5JwY9{_G&iKRC|&;B2%;!XWB6-mY#9^>e$lUvII8uPmar^!Qe
zQ1v+m0H@MqEdl9Z`RGHAt;XLS2m`oe7HWtxo>)ig%mN^B-FVw)-p|q!M4^rtn<eun
z`34w(CR_Svl=_{-FJCkWz$3mp@0Gwm--MJ$t(xjV3+tbLv1#zzT4mHibVQ>lHV`q{
z_wJdj&0s?|bbmImIS7;3TP1%yW0q!IoH+O?@Ne_AvbmDA@BJn=!z?wCMBp5qB&<;9
zjY(QMo8E@ojMd$_Sf-ZeDUe=U<e({kAX4u~TZRxKD`W`a*MCLtaesw%>s}zCcFr3y
zwudl~xg|eCi}KI9#M{L~nE8MCm-iUT67eQjl~qKAkoic!1krmL2Er_G(lE>rkl4?z
zwAM2UXpf|$&+KP!xJkHwZQ<|xYase9(Fz223oWrj5;>BgabSr~GWsFe=uv5cLJyiF
zQQdm%s?72@24@Q&B7I#D>F?0F>BS<+cw1C_B%Tc`1Mq-PY|v<D9|1J1zH9$fn>YX|
zyKF~IkPlj!X5d;bwMwL_FXLm>Z)MGHcVMgifK!hg75pgJ+EsLLR5MnRYapvs)i10y
zsY9+}VGagmEr3|Uy@I)e*#a!E*{N{gFg|`%?>UVhKt<_~d7${?HPerZg@6cLlxA^4
z`5?6PFp1sn9&M079XG`6+jK2%ikqCu+<%|{UoAUibzWYhGYX9@h!N<WR1N^|e9au3
zX}a*XCx2TYagQW)OI?wx=w=C)gFpO_AZO(L^yt-S;HLC>70YfZMj!ElC(aWxiCaA7
z%pSxN%NY3j#AXD<C2c>nSs$DyjV+<`O=!Q4Fb`%LIz-=da-#VYf_-_?<DKX%u=`rO
zBU2bFf>gDet52VNXnVZ<uwX29Ozv$=Fih)6B%kK<DBH+L^8v17_h#j%0H53_*6(g1
z72t#)7@Qo;Sx$O}vTY;{zgTX6@)2$gZ)m%{uD=bKjb$&-s^2;$3*+TIWGo8b;8;M<
zIT58-n3B&ehf@M<Y8iLT?=Lg9O6Vae{(YV@);d7hkO_;cdFCQ_TBRf!@o#LZOC$<;
zWBmeu7aV}lLzMiyGUg+Pbbam~^Y%8zIHka1R~W<z_ekE6R7BY&@W5`&EDKsP3}ytN
zig}*%_P!h=V?_56nEi$%V^K!*msnwXmrx!IKt^cLO+7E4)K|;SwQI>9JIe<RSZ#}R
zX4H~Vp%)<8?>3Ix){%+_>QY2SRqQM;*k9cZt%SK)eD%N4pT6jxnXsD3x5Ln%AWt#I
z1Y_R6GQlmq^7Of>R+R7$s1Exh2M7Q8AE3L7mDWnGjiRb2miUJQ&9GClRKTL8l7+*e
z1f4+uV*q~X>dfQsZ?JJx%aJ&`3r|9Cr74Pe&PR8dSGSKotUzn2Sy@Zezin6kY$%;Y
z49|6`?-+X$x^AJ#&(=X5?qsD)mRZcujs~^+F$L?hsNdq<I!I70(2?)Y1}WYkm+`Fa
zR0~CXmli(qrxnM06m0sL?D1_DebjkhSeq(mri?NC`F9`5g{*u$!G++|-Kim+j*}&&
z&ZKX~ovgux&Z-9uS=+t2n4T!8MAJDx7Rv>`3tG`50y8wJ++joo1?yQOLEbHlJwH-}
zB}iW`SVNeI91kw28D^(OsdWKQ?o#O=AROcvc`0vQ!_A2Ns!5K`5da^A3V0;-Fht_Y
zl3UB)eqrIy@WAe`<C&|D<^`f|?AG{8oRq?+3>AIs3~#NdRYnugO!n?xVcwRt=?%7^
zSaCK=^9LGjN}U?_C92gQV<7XfIQ>*p6EPtb?6F<Bp3YPSCqAT^FHdsI4TUdx$$jIm
z4|7~lTRx+;WpnalcV-s1+E!OFrH#@Q275<<se08hbk$_>#b>E-9ewA-mNW6M;n$i&
zf%q3}DseWnZld$9)5@l9BECQ(1Bukr%O31sKT59Ki4yKb=$cS}(vh*f$>!L&yC7Eg
z1gih4WzspVol*ZGAzCf^fnYEO)zEnl3`M1*?QBpo!Hh+C=3=Yt+FsAxvsI(jEUfxQ
z{_)=OGKUn0Jg%b@TOUN}@T2~*e8mJAxU;>yGM4CJYPAl@6UwG#qJL2Xf&?eMi2g^k
zT`)vrf7!vJ{L^Jv8=SG7F`!j_!4mGYKS>7lPGkoOcBz?pUsAh?S^gJpn#0ob#3c#+
z2{2M`=~5p9)v34Hf<m&6K7E8WRT|M%FJAxHY4SU}O7}vYbz>{ZvRiUu4L6n#a^PQq
z4C@MR!)nvwTANsc@{Z&sR-Kk(69Qu@!5?53j!1Uc;4iD4Gl5FtezZ3NcH$!o{`<wN
zZ8PT3{7ZYP&@4w|9_P%VYGJ9;TAdc*WB3Nm1L&^Qq2vq--lu*1@l@H3Xt-l0b~qT~
z;nE$&9N1GA(j7LT$6KeOv!ZhJO1cqeHHq=!DGvJX=xRz3(EN$@9z=It!hfRU|1WS|
zSX#Y1@5`W#Hl}ooz>zDF%;xD*<TcFyL0p0TO{*(TVIdAHsgHFECwQeM`b|5V6#0SC
z)AOwuS2XzB96}`0BOGERn{~5`D0}ik_rFgjt1G-IIg)~sI7b6K2?AnE=ivX^ooX}&
ztbEt^V1N#rz5y<RQdsW226YWp;8pE^aQsH?*imcHqu9{URJM^8K{j6xB0Uye8E<6*
z60Fu3-n2=Rn%;MWgk+(=B7VtB6>Gfm4;_EZt!MAm&Z3T}Q6T^ZV<eIwM6QQV6HED-
zZR^)K$a@NXENqZJTVL1D&7jdQCe+-h+vQaAKrO_2w8l(kLGwOXjH#&~$39~Z%ylkz
z7H!|atb}{OE}R8e6<}cXWQu<v!-vYg|9~ar&=}xT{5Ai*iiw5i9D=ddMVsiTDA~{C
z{^=a==u>n_#KfU;H}iT>efKrm5{k~UKzYwlm$ct1={>K&1ksulm$ng1eE~0MX}R5k
z`}2#@(S|Y0`gge^`J85Gs7pVO+NX6N(d5wF(d(?;)1&BFSz85LZvf-JLHVQ}){6au
zJ!?bN)<NEvqGpc*m%(^%rE*z{dd6DE()PuT779IIRtbVh5~xnRt37P6a2m52vhJq%
zL@bV%9AlD?*p^Qv%ixZd8&zBe8}-YquHV91M1!B3tMajiitT=px{IIxAOhpz6v74I
zlqkLTj6~*jO+DmcfH!hR`3I^AX~}_dWp7@Ft=23I@`4bsKyR~qzxN%^C$kfL%g&Xa
zVfVahfkOE(r<k*+a-E{=x(Uai94%`w;;C;tWQ{<tZ}3UCC|Ca3np#)Gj{M~B9hQb4
zyMl9dT)ioU0V_{bn7cUCP{2=1&n7jQ1;QQh0wO}k7Ad)U*`3T;k^@K~jt6JUuK+Nk
zrxvxx9WfKy&t^Y83iNE_<4=z5;U*Nnw)i!PiB&4aZw8oygpQ7tdo6mKQ+1^$xYkBC
zCVZ2rsi}F>-6%-CJ$C2$Dd!ArcDf}Mg*~ZZiBJ9yxM$;Of{g0S8k*OnY;!ZM_Ow+M
z4vo~Q+-lqUT~jU6kyGgq%Bp`CYPMBhQ#h~7z;^BZNb0MOgldef6bmk9w*6)A1;o`-
zKR<n9S2pC5df0vM>v3{Qu*b%{#H3t`Tn6#a3whiMRA6%`95*<!B#`)6Wrwu=v&(0(
z9Qjt?HZCxX7BO>N1SPJ`t?GjD-+c1+D_=6Ax>cpwQ%|DSyS(j)!Gt)GPa_!Le{H18
zOn|1Y<XiBKjLXoMFHB&SHBTn9Zt(t6s`6=cReu4jS3#M#kh9s8){T#ev9I2Z#=pE#
zb}PonYX%iaP<x%}`y8g4{}%ng`rW`KR{uOPlthH;<~}a=i%aujyvGn)2bvK0EiAaj
z`8Wj*SmI+#<Pv6J)G3wx04F5jO;(|jh7qvLnMu9|r|M}coJiEYL<AtQ*}XYBu1kVD
zsWIGFSastpLLZ2GDwtJa1hdzlDIR#a-o6lkfRY?zBd}fy@Tx<X<m$=T#cjKT>XaGW
z&xlt4LrI5#DGkL=I~8n3+q*N3u@6p8?jv)|R7Gn;Fv^dPYHd8AD&A*onXjVdQ{$4o
z({b#+VL>LUJgywA;mn(849V}0o2|0?k8``b#cybto)+iJwf!A><S+$MG)0Vq|9P(&
zlzo8yj2}@#vPs7qeSgM#o%KDw^a$QA4Xgn9s(*NAMFKkC$?%ZCNC_HZrV^xCMD=f4
zx<5s*<xQ42yAJ*yx(INPURjp&y~YM0fsW9G-R*AYijVPcXbuK<QeElRC#OXF+qw58
zPvKy|2n$Q<NyF>2;H+l(#y$9`uO<zO;MrqwjK<r=k&&68%FP$ZE{1j?p_IT~Vk^wH
z^=u}>1FX`JZc#b3$V)pFqN@~)E2JmeAtT6hBR~L&f4ar<|6trUtw*=fc78#|jFO7F
zO2T0uJq&BI(?NZs;g81+i-7|ElO3(s26k}*ybe!)ymlixk~rg1EAL^&#CL%ACnJ<(
z{g#2T=ZCGtEv5&GnEfabOKuC6bZmN~n1!j;-7{5CZV0kS=g+?k0V(^VO|2o0bC`9W
zz502D4gxPP9na~zBZvm7nI3Qypu`=0p7g@6uzPKIG;2$9Zi3@vA-ua_hl%G?Rrjbf
z>h>uqjurzRr|3PSUTp{muf#0u%t9@c2AZQ(AnTdG65H?_Sm61gkeb>MxmS}2<EQ;f
z;c3vfa~H19Sq6{9ovx4HyuW)6mA<-|^49-dX7#z?K&@P`Ij-MuE8x}!8D!P|>brkK
zI4z+mSVVA9tAg^+ggK%%_Pt=uB9v~kw#V_X?|cA7&|Xr5N*}BBNO@6adrGn5%K5NR
z#jE>aHX*^ol7k`7R1R_v25S76FDiyv8h4c-`qxZEb!G`fpga%ZU{BM#*IB{YZAe6q
zyL=bx2KW_=D;6|RvEoPQ^m6Q4gzkuN9Z-`q5aA*#?cyGLg|t5@J|>xXg<z>p2k52Q
z^k9J6(L%B^^yv^E>i}y-m9pzkbt|Kar=^v@LwE9u4c)fkXp}+U!@ofUx}UAP4%2A#
z+Lbe)R91-KNDL9`$4ST$jRQui=xNAi%TR9Fu)!A-mDuJs(FDd4Z(2T-PJTsa1$0TD
z$54OyRXh0x{!L!f*UET8P5C5bt-;gu9xub>6e>f!a#D(zQ}GSzvrWKnmzyB^dg1<f
zLm=+V&Oj+yq}zYe%TnX-mO&QNR@Y&hy;|1veVh6HA))Z(L_wcAE6`~u(8fcYImh(X
zrR0W*#HWt3bf2hYwb90GYf~2uoSpsq8rQy<P$VN#gfX0%(x`9YT<<+{3-vnzAtBT0
zNEEzkVbTmna<(MBt#sU3vU*UYfYe5qUw?k%FNR@1(i2@FY`Zmm@4`dm5A-5!+Y0{t
z1FP@!cNk4gkg(O32YA-;&Z8=~naH66BQM&_ZJBe8)Lw}R1q)<YbORgkM#1Pl3V$M!
zsu*D+%PFgkzVp4w?KE)DB9brFG@R!}C}X)d#sV1bS+u!9^D|%M{nF?HYw~)ObV3YU
z!H;|m>4)x#W<w9o#;u4ernolQ^w~UasKBj^3V+d5<q?cIW~eB&1Bfonq}gw;M`f3L
zO3;C`kRk?}lfmfV6|aBEe}b7ugz*#j-DGV)>#@HeqNN|3&GSkKm9}~f@tdnt&y@Bc
zv}YLR|9W=B_tTSy(#uq9nF_GP3CAnG(qIl_rF5O6ADnMWURc6Bv^!O5lk`lVcyjS8
z4XGfWVtnc<bzGgvrxmV7CG|Xo<g&kc%o45;L7I8kLAZw|ZC5G}``l@7^H+2uZBY}{
zjIQnyDV+IM(-lYh;1%ntXjpDyuh|qdB|3}ZYq9VFrztbyi$_{JR~vPEA?D+Z@23Ap
z9TcJVt}AA`Qs|fBtvrHQ#eakPlZ`PzEH#uNoCt8zJy<ameN7mre}O)y4KnQ{lZ64!
zhe9cen&IrN^vG*0>c>b<ySHTIiXK7Bj2e-*EE^t-ae_h7?AfohHsUgdf!BjMU%m~P
z*Po|F%8s0uu2EJXBiQ3EaZa9S{OT{LK>pW}$80oy9ZgpC^~U9A8%e&-NJ&NW3+bC~
zgw#JbP;sshOzIZ%e#3(Nr2uM*rZ_^m%#>b-7IPx4y#ja~V`HaH-2HD9bKsxEYt$Og
zp0peVBN~P#rnZOYN%9<aEmC$uVCG_E5ly@AXWEV43=4DrRxt>dApNl4_t?*RhHI0v
z?ETrckUJ}+zxto;D>rLR5G#J}?|muw>9{-p$&`kiQ5Mnrp-~uU6HVI?QSrUj&nm}Y
z!73Z4czMR<y+nT7UzLh_!K~=ZV0ReTK4KrJ*aHHSJIe~g(dDayywX>g8)8a8Mvcej
zgXrEenah}fu=!*LPB+E=)OlqNHu2qU%y)f-I?<dC%DI9RMDIlnCDRqrt`u+i>%RNt
z5vToaw_37muJbyGKtESMi;LeE{K+^KbH<UFeTp!^wKP^Ra+CV!3g=SW;9oup#_qO@
zHQTvj9z?+B#!oZ)YNVZ+ZEE2HM$nuGnk%H6=kb84x+z^f?-#=s0v@1&sXWB{>8gQd
zPwCSL{7#RLIW%S8ag&UY7AUy#Ps@=L@0(LtXmFzglXm`d7l|iTP{$1ez?&%Y^!dpc
zVC_J`pN87e*B9HBaOv`<fDtcJidnFKDVp;KB^@fGXM9bngD`;WgtF*A9i*rh4(K$#
zT_Cj`4NfIj$8}`5zH}1jD71XbLbKRtes!58(8lM#H-4}hVyt*dh+>Yg0V(7D>f&x<
zMq?tZ5gS}|WQ7)pCh$;d^5wLG6`Ge<h9*~F#Tt_B7cbC`dCZenIDdO)Vgm7fjWS^M
zro&48el9W2Uwe@~EcI8=FZJ9(KLRt9`~rTkd2QdXEE7E%NI!9-E(s7$YC@7v13ioT
z#&SQ!jNLk`|2@gr#3(78lhh8^!2DT=+KCy{R0J#Rikwv=I`v)!1CDzqe(D+o<mFSa
zy$2#|HWWhyHqXDqv%{J5>dq-d?_Edps1Qlv>2i2zo?J5lkRBS6B&qGP2MG<aRGJ;`
z<>^JWdV0YVCW1&yORUwer!ECV6`=QA!kQEwc$SXHfQ`-`^o&F5wwWc8`SVFn`(L)y
zv>CO_ymu=uwe?^fVTNzK=iJbM3@4I6=#lY&`vf4+R`j~;7tXF!^7pVB*&J{}k{Pzs
z)xZ}(z_b!3|D66kQpo5}PIdi>vy$7XeMN0xdlHQvL+Pgpr#@W?&IpZgG9nZ_EYTsQ
zU#bs<;VMlru_Qmn<;iuXjA{8MMjTb-P<hfdb?E9wU91ZuY2jI)f5WJ{gpT})u8UBK
z&?9k9K6Ft?5HM^xgHk&ARu4M+_xahe$^jK5|C-{_$G)we%G)6|-=Y;C$?UAb;_2rL
z-_=j3SAC={_FY_D5<9hnHtdyRq?yr(JvYX^Yv*jF>(}9~dMlGk#ICcQ(rd5S2>Eu;
z8$KH$qIOLc@1hN37=X@w6*02U8Jg&>`wnTUACv|J`~x2?Vrt`mlmf#-!=`cj+J;h9
zvB@uPf(7crm?56ne2vUdGtoDF2g*u+a7w;M1dq6e_~hz8_@^tdBLR!k{{E$A)%kW)
zV7*Cym}~Cg6(eJ-Gijufldtv;MQNJmICoq+Sk1a5jDJe)Jk8~SFen%>*of=s0ly8m
zGQp#ndp)i>_{Ap~{1=1)Ec-^%8c9QG{i80spALNEckQo{0!DvkCm9wSJ6@A=Cc``R
zgB%6c-24A<D}HLnH7UBXCf)6bj=b|Ce0P~M#dI?CVdEWIyH1l9hF!}O^Hkd8jlQ;h
z8)C?hF07qogWIqqL)(&~aaM#mzj{x0c#OTb&Hi^EzAX@ZGR)<W-5NRQQN;n-eVpNx
z7XJ`uo+W{$O=~k^nM<XJNI8dgImFMx`e@nb_vZM{iV;hBK_rW=pj_^Q3OSrH10S1g
zZ4u66OF84{XxfOkK0@W@f&D#%eV6M3Q>k45RFhP+;s>CC7MT@r#yI}n`Xf$)8zg4q
zlf$<E?IARTBT_DVEEj*O2P*m^nH6H0Jg}ewafy$Ze9mRYcS8&fCTJ7_j7F<2sMKun
zm5qr-1^fzRebY6?|5@{*b=>VWTrEX~D;=(W0|s@Rhh=u225WN%g{Rh^KW2p=C6rW(
z?p>X}Xqx?4irGehI_+9jf)<aI$bGPBmYEL0@XghSyD7h<Z4d|%H_@><)3;x4)Fh%y
z8v@c-PcA{}CV{tRU|8*QpEY~6hNslenk;PY1WKE*jL*6eEFVVauq%ZW_(~?*<>w()
z`YHsO7bt&FF;MJl%i$r@0=aYE-yB-vds`LD2E7;Mk6NuZB6j2p#?Z58n+4(9DepQc
zJ+l_8dPilVk$#r?rN}`H#Phf3`=vYMc+$;(SfXPVq{UesyaBx%&-|abfB5yn*X`j(
zX8Ywl3d6i5q~u<4vWAJiY3kM46TX)2gc`j0R=E46CQ1fiDI`*oi6!dI3_t+}lZKI@
z@h`80l5i&IbUDaHG1Dw0p#F@4V<BUIdP=ddEtWSJK?d)OOb-vZ5VQ0R=6YEnUJ?0w
zd8H-KKen@xua^4<n{4YFrmZljXAIvhd)$`zMW;I|`YPzZP?q8vr<qv;yFV18R5#3_
zEHRXp%I%hEerueTs$Q(3e>-`lNsb!qx+hw!yNBjaHw9onf4Pq34GQlGa?TGX2R<Kq
zH^1E?@3TsfSD9WaK?9{K8D=>zI@U2>$8V9etLJ+swz-dEnT0xuwlm;^jQGm!=UoPi
zG$oWm0}>vvHV%P2mlU$(IWQSHfcLvwcALoPA3p3>Ok53xRwpQuT2=OJiICa2#d-W@
zw?jit>@DK&T(e?c1@wx}I~zp7vytyt$&@eG!9pC1xXVPj?&yU^6#*AF-UO)_-naxW
zn^cYJ)8DZb!lh93N=z<G+jOd5G*}8E?Ueg5PZDdH?A8=WsM`zaQ;=>p1Y*|^4GD@h
zh&eU2)EBHO23lH`*FT55lEC{7!a=Gm+og&gcZRG^BE>qve-|?`n)V<VJTmJKB~804
znEy?pi6&<nxmYO!zeQ*)X|WEfNpqmYFE<|y@8BzU$Zjj3##)46j^2ox=@Eh-#NE>c
z&=MGU1XY)8g*s0~Dxd$)3vigNiwAhdg)YV_yZ>gkw#43%^bL7axmX257#aIc5t~aw
zgIH5~?2HVU@n5Pv&rI6P{JGr`kjj67<Mv7x#RYUqbqsc;l6pMpN(8P7qO_ih;4f$#
zP>p!w2jEIJI;})b#J6WhE6Ave5KGaksl79H^al6>f|0=|RQOBawm=W)pNiY6wWz}9
zAvVi#*ezY|u~j2mWATg6*L*If5I%BvEV?DnqSY`dI<7x{wkq1{xa)?JMb4@+&F;*i
zNY>=X1P{8oR(&csAEyoyYwm3z798Py^uL*4GEfO-OfIWEzVUd^%fqGT1_#z<G;Wd<
zk+~KTocSs=4Io-Di;><Ej7|~c)2L~PKa6=3NnYzd$#}192ap2qbAEcS*aQ(EGasuh
zQeGO&Xf>N^re*Me3bn(ED_|6zP5GAXPKJyA5j&8n8x0PGkc+{6C^XH4QU3CrZN6>^
z+WfG^{~jRa|HQ4(gj|x7{5LsO=hv+Ay}Y?vJ|ABJ!$=x5V1U%jeVuQ6=E*SXr||LC
zTD)3M&eu)22tStWl<VF@nGVd(aESLP&iNVf{PmyzxA<K(46&&GiyrvaB+GAhv>9>~
z8d8mKE1whS`3rqc8%4tK&U3ekYzbCjy-w}(&Z1rG>aRgXP-<LJ$d8d8G2}Ds4vra=
z@m|eWMtKw1sRkAfSGNm~4-Jaic`jTry5kH>hVNk?(h8T;jjh&fo7DUCs>3&uR#M8x
z)8YzI)0RCgyDNGZJt$mmmaq7)fimhdQd;NNE@}j+noW3dvkuGjDutJdHjF%KZO3Ss
zwmYw(+YwzC9u5prTuu^i<+wXI0zL9{sb8B#xEbqqSkf>yt+q2~BMlX<-0CvnB1DDo
z&!&fe{k|jsPl_zC{eUq55S{unQq!wjl#7A|yIhnitV9wwkVFVz{qhRV;TyvJ<Bi7e
z!md5hNCnV^zx8cTi^9q)!t-}*VQ{bmWN1AZGnPJqpPf{-oX_t`|A^iEO!X=>N6*m*
zn$X2?=k#JBq(Bod&Ich&jb-!R;mKCYQeKwH2!&MwE;X|cQfl6D%0IkpSpX--LZe|t
zzCSPdOLu$rxI6?sYIORu6>yqmcu2@_LFYA!uC^2!no(7b{Tt$cOosMJ8dNIJl->B1
z{p`PtnCIP_Wvl{`EBMG%5o}1j@KyPDV-jUNsA?G7IizjIecSejZr|}9CeFIQci*}#
z+^^(F!-dT}ACpVBKLqQeHgXup@}rO{mVeb?^!$&`r4-QR?{;ANn?>J|*oI0t;z5hM
z=z<PSUKvQ=AM_EP@BV*8y#srm(egcf$Bo^nNgCTpV;hap*tTt_vDMf%8{4+s*w+6!
z=Y4<I`3igOSu<<x*)vX4_=Rd|$F=<MP-P--ECuv?A`vzcEslYr-7;RtSc<qSCjPf2
z*>nabLO`labK|0pJei8t7Q{<~Pl(3VR9wm<_b^c7s<T9HfXL!VIgw=v0o<_%IaN*A
z!Uc>Zc4kDVJ}oIuDuDDt(@bfMX6W;-tW#}TsFnr>Z~>9;Na=lE9G>;n;{i*pvkA~E
z?GumxwTi0aLY-Rj2jAAUgB*zmY*utt`n9T2%dxeF6}XDay-Bz}C*nIKB{u1IR1t01
zL4#eZks#SiPeGQU2ivxynZ1rr`z72Xo~7-9I$Rd_E{|jAiw}GxOY^>kJJ?isnAY9z
z1lRCpI=SF+ZDM&(1o{s#bB1#6X+?RjCL*Z0N8B~$B>2_Umh$>W*x8iRHPcT|s={WC
zpg?MG_{fUC1h6Y-R#MQaQ4&HW*3|{lkt?`2mNIzSS|f(bI-_i*c^tBzY8Eln#ussK
z-<S{)7i{Llh)^0sM69iRgj`%TSf=7YzYthxDPhJgw_n|RI44<nFq|~a5IUKDk<awT
zgzW0)7dAX+U*$g}*ZR%+;%1XFq)%@xTz&|6Xi{=Rdd~1r7R=`<-ACwZW71ZXo#@tr
znLSzVAG{;e181igFAeLQ*_=NB0oOe%#VHAY?|F-YYfpM+_bsR%g*<eeS|@;UP69vE
z@C!Mk-7s45LA<;<C{Zg*7!NG6^>iYjDY-WFl<a!~+fw{jHlOl}(q!B*Fa$Vr?83jL
zV(DkWUis$hM+;6SqMAW?*dEe^*}K&7>jqXz;#qn{;7j|T*7|wwKbV|sMaKSV|K<P>
zkW~DdO$fP_{Lt&zp5n6DVIuHm$9o+Et{$~ch**w_yq!?5E^K9kGyN(3PkqlX!#fW<
zbHZK?DH-m+#qya;1TwfgO>#+;=OF1NI`Z{Io@|{tCE=AG0iw$+Zjp^l<zHDBP4_u8
zP2;axHt!7@=g=4_W)-#h3-S!lj7P(=h*oYN?Rk8fJS<bVtnZ9B`BdDpQ7mE7He)&D
z)vk)mP9Vg58y9J+b)*XaRCGH_ZO|JnHW^m^P5n|fRy_DY`{j$syN%@(wEwAmJ9%v{
zpH0)}jB8|-rddHine85h^43jB*ota^OredKS&{aZSJ?ya^V)e|X;gAPGBd!F$K>=6
zXSJHc_GzZ<U*NT)`)3^(z{MOVZQ|;G-M2?-vtY`#rL&tqfEX+T1N6^?Mq!3aZ%T)H
zg)8;=@u7gS2BxC>vl~_dV_9g{d;fkTFJOm$IGy=h^9}3ug@x+#pdyUAk|FpQ_Ua_3
zae=~B_Xi^2MVRh7@;tO7)5#TZQT-1=((rk@-V>%+r}?{NloLzyHRw<EoxwdgzsGbW
zcfM!6ILg@*JK8=|QatJZkRksuGx;No^!X0csZEiGG{-s~WopB}_=+beMTTJFK4=Et
zGtgtgt+o_qicq`2#QyWk#+I*@uMBIzSLh@rv3tXtsB=3e>%u7LNl4w9%@uA|5iuqh
z0x6*;a_+2n8a5$nuEDHgyqq<q!*)oFoH0b^K2iGL@3ulc7EAI+7k|6&a>Jedi?v4(
zYL8QMPt=rxMZM^@*rYagI5H-}*geHdinh<@^gd6XOv6ZzO#V*EawKw{m?RC=B;p3S
zlv=I@?`Q>o5voz^7=pJpTl~J<gTFU3S?dpj1x(7{`-I|MW$F|%gh}kaT_OO}tC2rj
z6?OIIy2*Dcwl_50w2$>hcK=h@s$RIYU7lPliL=i1>bf`@DgeI%9VSaLh^s^E^uwp?
zZCmZZkpL~0Ltl5N-QtJ)j)EGC$YMDeN#$u9oD#9P!3`%+7Vjm4ccCBJiMJtT5%(Or
zY)(L{rD^>P!FWW;kcy{E0T_GYe_pbZjG3Jg-MV@n(keZsfNfv4&I_AN{uED|o!q@I
z<WDbB7cN>D{bzZBz%%qBqsp`AacC&Up@(1S`HQ%LFMihyfrV_&IaRQMj`7k#NnR8)
zS0_!-MAWn%-zZ_JltYt7v93Q5x@bAafaiG;)9}$Fy{4!+Obb0~ypWULR-cgm>VZne
zki#RWf9-pZTUJ*t3(xsY0ZehFC4owF1T}Ty9~t}&de^^s;H<(s55En%ays#qSiex-
z&~ISW(4a}5?TJ-7hY`nE2D_Bjukr|mD*9m7E_DpuI&0nFQDNs+$vW+r?RDn_!UOlE
zO%6210j`UqxAtvIgsEVFEaOVZ7y^D6Lpa65QUSAZg!6ZN0HK)FP+-8X)5g(9g~ncn
z9_M~B54ie~{A3ue+Uy@h>-sv?xjzew1dP_RD95%qhT-YfS2sI;9r%T8?*+cW`CYpB
zWL!!gBq+M(ovb4<>x!)x=w~@vZImLVpMw2_+h`^1VAV`$e3(HH3{ZE__k1(MKEQrr
z9?pH5(S`QwI#NVV<7EG%EXAEcOv45lMDz6vvlou(7zDel2%^CNOOXPvLAq?0S=6Uv
z&NNCi=eksW;RD|<EpBjh@pT#PzhD!4U#=}A4AEU`tIaOYFwq%v&=t3)K97yFaHNBN
z&Q+qwdR7zPECK3K{6p7v1!L++z?*wi+AA!pkfS*xITb0=JZeJ_Gc&vB{baF)n&N;A
zv##OLpX#S-o5FeP+5~#A*iv(i=G0ysD=~vOZ!M-S{2ZJezJ<E%-?ec*ijl+(&3_SZ
z(=CzXsJ}QR?7q8*29)%GLLfpO{6OWxT0?+%38n`C#PQUBl*>;oz0&j)iC@9zr=fx}
z#q^fEw+-B}&sUWs^r%cY`wQCtWQxja=1Ii?R}>lBn{4miJK|SBJ|P&>64e=e={o=6
zzF17dJFSA9T0vkzAVI6MMM~la=>9(hk)PYK^r^hwl#H^E-KPcw$!S!QYNA3C`^F8q
z2<)Q)MVbRgqGH5Z%N~EBv$N~#9{+dzwaR!+`s2(zSe<Ow74zXVdDZH;Zs1c8shm@e
zg#^syBaaPV?blbn_%NS@qaTbs%Im!*EGS|uHk?p@mT5E0O5&e*%VLrg^SrZQnoi-+
zK}+<X7`sZl-i~9BN~9?|`%mk{)WMKdLMbOXx?ilwQZOW5uHfj|T(NzSW8@D$;c-<g
ze$to^{z9V7u~1bO$F@&Pld43pPrC!tVC8%;J%%E6E`M-RNT+Bgt!2C`c)mFh(zN_*
zML=}0Lv?E(+@J8O?AWN;`Ui+SNdYw$ppW4Eakm4@`L=<%#R7C#IsE+ul}@aTX=q~s
z`bL;k@5;=>$3V{SV`gprg5M)jrFG$})VJN8HW)xjtcE(~^Sc<rw|_V@Q>{x(=HGyC
z#SwYTPRy_o!Xg`z1mc?BX)sK0re6&H3a+1NfgUlx+|1rlFzKLug~Xq~qSyzw#OPXs
zQI2#=?`wiN7NazA)NY0JOLGA&#~eR2yua{5m`ZRgTE6}t0i`6K{>JU88_w^k)(_D8
zZ9KEIen=7jYJW+KHBG=6l%QLRHcFLAF5yhlY-4tL=~q|J&8U?pis6(c$I^-_s9A2p
zvpUAZ3qGZqmd<p-VMbTebqd~AgCHa6uLkZ96MpOx-7<h@$CG*0y8t@;{+V^S${YWs
z=Zx+Lq;QBv&tU%Fzlnw6OHU3|VCUQlx6ZRrQc44TLg5X4llk~YEW8INlhbO9=Fo11
z<#3;mW4~sW*{uCT;q|=t<YT<_#H*UQtVu|()!b?MUf948&w}e<2ez%8cX&pVO31#)
zrE&%EQ%8XTWf@MlW@YeDLLH;+>N<}4Zz8Fn7+8JUiwWspYG8;h@yOw)fCIL2V>N`5
z_PWGE)Oof8v}us>fr|4NfWE9V@zp<nzaSS66amk>csKaedz!xs3yh2EoB$My8bjQT
zS-r=h*Ky}m1LAR(g8WGxVyD={JSB(>JrDpZ5{3yAdm34t#Inu%D`N@=GPb@rz(>}9
zHJnjtq6^;^G`hV-D#-AS{%83WN=G9HSv>MB&vR*b5x?Mah*&_Jm>1HbZ6xQfWP!J4
zs>ZkPkFLAIq!Ry37c6mWC&qX|01*6e$V6clajgAm9ff<@C2MrFzjpy2u)y__vPv<-
zoO!p}2m+vi>obgWCvE99Lx30=1VE3FX8%9rDB8TTye`4Tso!@np)E2m?eqwe(j1;F
zW~iS*c?hLrq1uVxXSm^q?hm$lbG6G>SNkHntD#4quTN<l{l-Pm$cv!os~*L}l!Za1
zYNY(V9SxzV?DIzLEYv3i1dted0qa{33+`%KqNui-yUUUwrWObCw@n7KH<igRZ*3f%
zRTxn#7Y~>706-lLKX_@kt0)w&Rf{Ga832pxx{WNv#s14gF|5Q}_fWIOCNLySaiX_3
zqIE0)c?_0Qv18NQrewa$_pD%4d+rKpskXvra2`%a)JRsf4XTE!8o{O;ZFb*1r6FIA
zOjuHv>$?yD-zgBgING$8CH_dykT2eC(kpwbm4tA&y&2i>a9coy4(7<fd8Hcbr@2%Z
zR{CCUJz7@XcRZca;C9$Ni;;S$wcXYInz<_W>sd-Jrn~+%@;m=OuYdZDcZ;`!FVzWP
z;mlT6WYP~(6+)ZohHc}{%#yixlS^MhLPOrp2Ry`edW<`)5gkMYS%%zXm{f3BF#GYA
zh(;hftD{_Mq0m(-DiPBM_;V%b`e0dgQ@>B-A3if<$3-oQaaby?G<o<7?6``_C#~aQ
zmzqGN!HSt%pH)s7p@^Y7KlDHW-WIs)VPiSRQ$2@+sBwXxpbzrVj9hxcl|sTW3J&D$
z^g2nP{5#*QB>!<*F9B+_%oZMrS^tFr$V<E{8D*Fky-$49tF=m=5dqV#-r>jd@Cb4I
z#ouu?;viUv>W{?DV)y}2^`xS*Iq(ppI`1KvF7sMQnu_D=RV{A;#gxfdWD4!?sDzL}
zJVpkr0~E<;Wd+x)qz!TCyL10z{tXo=ydl_YoL}K-pYHSV|153pK%3`lu2|SPH^?l1
z5+O1Fy1K<yj_cuJv%SN;i2u%Lxn~WQ9uTuTydYOL&qi>QNvNE;VV;A)t6rSe3t?3d
zLsCbKM;38_Ase0Ig>Zed{*>e{8%Zy6{T5^~FCS5Rfb0gVHGtVPL&b12_r{qJEsrT(
zS|?Pdgd0cjGVCjksLdV({)IULYY~!(xX7Rog$_3(HfQy<hGFqKC@Zzo)H<mmk>Ku>
z2ZsV0LGDsr8%z)pe8Gl;{+-E!h?K69|LnUfIC2hYH-)=$?jV#><iYzXG(f2rqK8Q8
zO}U&U*AeW3s$3Qinku0Tolm!tK5Z2I%m4W9D-?jC=-cRRLPNr0QlyZc02+t<GOU^r
z$ala?*~A6b+8b8^uSC`MSK6DiwHd@)5}uf<sCyDOK{~lZ4YA(@_udVxNRv$X^+vZy
z)0MoyJ@o~vz_!UxA9V@#&bJ1#6uDkc|8)C~{rAcfCpdU#3_r(~2V9d^&h0mW4t^Qb
z_!h+(I&QB^)HYGL!e-htWip*1;<4$Jc)(dq-?%cS(GWx~$J&-Q8Ig@yYAQyq)2=?S
za$b~fuOt997PaWj&(4a_Df%zFus<P{_P_%7&wfetE&eAt>FpK$tQd)aM*Xuhejv2W
ze)ZTp5(UvhygBrW99`+C@PRUE%pI0Tzpe!lj;lANf~eT#hl3M+x-B<2mPF*w*!=;!
z$Uc9}sEz!9bAEx%=7vXvPR$}+y=+)XUg`>^$Eh^z_QMgM=}>$lVA<Q*IjmjL3_@dB
zkO>jcWDYuw*uoR<X`f?_2nHScH`HoU#|lNc;{BFlkYAJ&`%-CL;;~rNyE4crd4K{S
z%snzgpE@K>#a};~iu6zYS}5!*rxMF6tak!xly&#s#|c%GcQ~o}AGK#H)hJZzwXo9n
zYqK)eTijd!Bq2?Z0WGCQDdayXi*_e4UWjPw{sg}*fo<FVf2F=$(k%I{1qW86-68gm
zHnx}a*3JT>CEU9yi<eJkbC=3q!X7#N&|S(8m)WF3ruUgJsV&+5`nDx>?AQ6HkdQqG
z??_fzawuEVVQ!Shd%SXg79Q4)M`fKHGQk4=ef<tT_~+p0P|L)wnbNqBf-Kz*CC{;=
zJ-_ABdt3=4NOfTy63vL>37le_LX|WUWvKpAiIL25U!Ri14K*10TwG?H0T~8JChEn(
zPB<;Kbtu>qH@+pi=7ct{k<k@%xUj@G^SGx}9$hy)`tp1=FN=Yk)+STK0;`t1ib>ul
z13T8qlL3~sOi#T%B?moRpaw2k7Z%)G&DO2mG^SrJ7-ST=@80szmDbzItScSU!2%Iw
zqo=K(OgRcCx&A;CIuSU)^@Z0S$BVlS={zr+^~uj2+gj&KY16^_ESSJmwgGia+UqY7
zl}*pj*SFM{C%GDkbCSWI7`hIw8+x-&<8`-~4zPghGVYVICX$PZN!vyVGanM#n9}{l
zyIlV^(#L-<zejnZ3sW;3orrn3(&jyusIKFL&1DRIrkhEqN9pj!ykP*JoEtv@J&i|$
z7Kt)+(4TRQ1`LJVGNz1puTCCW!?}PT{Ob6vN8q?$CX974&KpCo3M>C0-HwQqv3U;8
zlG(9Pb?Xj}nUVhid-Sw%0I@#fu{p=2ArP$Ehm-~J8-i^Ia{PK$a}dVJg;NDCVUjVS
zqDE2Jc^a46_|*6zUA{0NayQsC&_hzL#&~zr<?qBTnp2CI#8YPPD4#Kb|JAmVh#<oF
zQQxdn!>7_e;kPzaDa<=le$f3VzXg8DL-6?wJtc1)5l1)v!UdLhq`Tc0Bf|954nE3z
z<AIzC>c+*YsjZiEbTH$uq=lH!0DhH=7fK{79MM?nM3=~di2A}ofeO*aVWy@AAn+Ku
zU*fVNBqlIm6`$nG;8J`e5C0vr+RtJ6^f**%NR=K;uMqKqK2rbfFA?Im7moGz`f!SW
z7)`?N1OkLfvqiUP_B0YG-!kS~$zIaSKkh#MjjzBE<g|!DGQ{kD)viL$1_W7I-JW@i
zzY|D|z95adDR(9gBYZ?A7Kz2V>)%k``bZDw9%LPcLOkqMd|6!~ip%=(CYZPGLc*S?
zH06MrC40mgZ);DDlWPJydLJ3OoPPW>M+ndLj9@^a$3EU-XLQIJQ#m}33&Wz4vNqp4
znc`1maUP<hkh96h$kUdRU}XWt@@Qc9ThqIvNI(P%2~}<Stkjx+IxbTn+$IZ$&U4qz
z51Gm;xLOzvj3Hx;GR|DETi{P19B_N&>1&nVb@QL{5+p(6rcz^}UIfN17s;bENTnfw
zt^$cW*PP7Vx^-WbUub}`N^tQnj+xZCVcW$M>hmZ;NT6z3zqnIms4m)p;FjfwScYxd
zhs$~(40cTcjGrZkFK#TtWvL4bXL_Jh1Ag;Vl}!BK*Kp^lwFHwFW);!oZ7uNb_a-F+
zuOLCFR|yAQ55zQng+e2ncVpnm^S_=nNdkglkb|Ax_D^~kwPe%erN2}!>FV4a7pu{K
zH|>dcR27h{N2sm;(l$cYV$jsdaRj|IPP3#aq!+9kl?6VXtBK?Ew~)EVTl*5wPn@zT
zX07Uf&M(S_h8S_JsTBK??l8wV-5#MBNXE}$HezMl5QgyR2GmJXM*PYuK~e+f6B@dD
zWk;9G@Ldd$Ds$6UNlL#MRfN2|Bfmtf{Wgw-m`0J0GbjhSkdURv{WVyU=CV43EKJ;q
z0CP@JLvqeQxWh8(ujw?0>b9kApD4;f&I@cIsdJEGe<qpX5up|!|M8noQWWm9ya>#X
zL2SNFE?Ui76N@1bBuG1DnWQ`68$WpQ9+f?$!Ts8(>+43J#6GC=l$~wgEl+3v5dkea
z?RV9JR#DY+bag8S@%<bXyeCyxXM^7~J%C?UqM!U%W5iBr(X#fjKO9o)gIh{@zL9_F
zWTvqLS<$E@5Ks)CasTSKhN<Sk2Hl_ETC?>3XRe9OuGzEorH!L?!XT*24k_zmdrus!
z@fPW8x=^vBrPIb2{B&J3KQulr;D<XAF1<c<yyRsZ($1VYFhfu!8<e-aw7O&)8fL|G
zE{Em(=Ih+uai`5${V64|9xQjpQC(j_PkF}rRXkyPS3?m<leY433|&jvOpcNYNw3rT
zoFeYm_CGeVokV(NLaVV$bGFA+X^1V8J#HFJ^!4~ra9x3FuNqb^B8EDVFF5`UuTF{8
z0g65rH~qUj_m-3Gpe8K$9cpCr=q5`J59STJ5|+ih0rA&TQkh=&2ukXx7d)_S{ZR}8
z$T^_f9^-4$-d;F`ZaSZTO(B>cvXZigXT0In4w%={cWB_>AlJYflxq(jThXO%{YfYw
z;JJ*>?<aujX-<v^8a=Tz)vVXIrX+o9YIzK{P4GG?_iFwlGwuz`ktf@zn&mU0SggnJ
zZO->c3Q#lp@b8bfQ(nzGt)P}G{nLPK@RyIJ|9gEQiLgZ7injX7P=*{xGZvXi%`T?v
zxYAVeHvlOwoWPxBM2waD)KQkyv28z*e-s_DE;idrpgok9jz)f40W{n3>ZqKKm7V_f
zoLqb_u@r_^=3Q5-UX)y3oQ-f*WH>GG{c_R2Ye?(E66yG3l0qt4pJe%$*)a}v<IjOB
zfpYesb)x2`#%+YRI1u5K9pf)3>z9;%e^d#~IenHVGYo>BjZH*Z!K*wKE-nz7MYlS)
z3#j<p#l<3s!c09}<S0{W@KEA^t-9a|4HeUi*MW%6l(r+}E>|{;g-jG8wDSyVa`rTC
zpi&vs7(wuULy7>=O|mG8;LQ8C+w=*kg>Z#lIS8Oy`hnKMYwX?YyY3HA%_*;S_<3*-
z&MuEr<W2(#411xMT`zyhcX|wn>t`Q4k9Ph_pf^Iy3-CZSajxK-_%S0)@C}&MnkE^s
z*Ur(425N{9+b2r{Zi_zt`VuW-<1)%DRXP#&Uwol}kHP=S)-Wpvt)gdYderJ()FTAZ
zf#;benNN~ctFn29`N7)Q^g+L>`Mv?pXI0<%{LWQ0yuV8%_2klHfA+f$FQtO<yH*~a
zbVZx>((GplYo-tv!S<*qCiND{F%+DXZ#zLL^+r^5;dmSgjq8;5BB{X6$nP7rbhi3`
zZh0GI6zY#AfecDDNm-*=S@i1$&KeDMVa)Mr3;(XjUVGC*9RwBTNi_?xunMUr7}40*
zKsO<Ns4pzw#ald#f>y=w5`Xz%pqmo2wS_#&O16yHk$wsm6i6L!N!w+Wg=`8%ahd^F
z$x=-a73oDRh=9yWz!d^?c7I1vTw&TPh+h;FqgRMVQ+WL;3(3HMr^%FMYgdd0C+yUv
za1?Mq!t$h-D+<z$=a3&#t}!KH7)(oVWfN@Fu&oVuz;+o;eVWYBgK3BkT6GNXzn-d-
zOkA`{i&hB}z7*gpKFJ){z(|0Hg(V5GlxN`J!}m-ZcKxTy<0b>z2#@h7Pl3O(L;`Dh
zIcYqpD3z#)XjmfjPh7L)*&x=qJpcI7MB-;^MCzZ9Z2Kw-_8+h6mgn9NI2Ya&YFn?D
znqAdX-c!7DynQ~4I;ZMW)Cd62f=ulH^#Ys?iu39nZie}O!{X>nGS|E9PC7vPy>J0n
z&2pwzUan`2j_Bz+_Z|M>n=x7|&3LWmU4e28^}+@zWdt(W^flyuiSptOMuf9y>YB+K
zn^Dli8f(^2A|BGq_ao;j#j#Y!YVQjIFr=dqne(`}elwx8PzmK{tyadSw$p1?(OC=D
z2R#^T`B-(PJIbPScC2(7O9~m=9I0gaC<!_P-`$+f`}TI0sH%K$i`g1lh3_R8A3_!A
zMZEB5*`^f7%(}90tnM{KJi^$<!nL3^Dw$9a)a7NFp@flRc~gZ39E41ahQQT^gtdp>
zt+kAiN9S)D&?FDGlS{lZkXt{Ynyad5)CzQuyWLQ_Mu(j<XT$sky2JgOG<?u;>l`cy
zZ<zpKQ+Ewj$UL~)SY7wpA>uT+ujS1&8_&Ac>Vwu5_+`w!i13AD8#AfwUV!Ip-Bwvh
zI^Im9#f2^#*k&){XD`SXQB_^&<_}S<WAdQ@wHt3dmB5Uz<IQOCX^Eb@R1IkOiBNI3
zt6Y4kc3dc`*-TZ`H1E;g=M`{_)KYqo=C+#2-!glzYn7C*neK1Slv0OLi3x+X8En;V
z3R~kbzAM(PX+64C-;JkWI(=fNJFK;cRNQE9Jz9)wj4^S{F(rMpE?h4*&(w#SD#{}0
zAYB|Z05>2D=qA{fp34$lPKSj_p4kr;D-*Flm--&(z{vJRbGy!T;X7{voH9+fl%j^d
zMa$FZs(Oh+?QT|?bbR~<#O#k}Cxk>(*VWKg9nR*_7v{t&1%fPtf<5_JHU1iS=K-5>
zFr#tMTcQz~kHjD4_m=Wp1?^=AFu$R`zEbRoqFb5@b6RkqL%9}nGCSW|H1UYLrsg|v
zmGaXME6RId%+s6m+mC2$Ror75sr@kr7$CcORYl{bmQAT_2fcR5Ruew~Z&P#OCQqC3
z78bbFEWD_HPlw&Is3Dm7xat4IQ1`c5Is{x|(mYA@BU=>q=qIf@L&Jz3)k@tw-Tn~a
zwGLHVu2J5*!yLfNr^@bFXNll98#k<qw}a#+U#L7Y#@3BwwBDuU*gBt<xCwmakvfUQ
zZG6<i-Jt}B;{~e5dFA{okN$L(^pL>XtI~5Q?RNh?;|UmB(yqeMSa5BCsa&DLi2UC7
zXd}-?ztQo8{5=k+HIYLK1jYLF5`)z0yl@#jV&o2Vv*9XzmtO+b;7uu%t%<O{L@r2L
zj%)pDf8t0f2~&&u`hw|CVa$JvO>TXGQnk~6j(1pn&L#aW3T30>AvDd{(kGY(<tObT
zZmA3auB`&EmnWsEKKe<lby|QPae_4obaj~<pQoid=t?(3fS-XT6V!4InWGVc0kgYj
za#jHp?@@&C6}>*=o5I@4BcTUKKn5KK0gjpRw9Zssc0DBi68VY-R7rVG@imeECFrQf
z3a8^7ZhsT&vMUbQcSM!nS-;7`Ap-433dpIg?0F6-N9UWuxGNT%Ep>0+t!urf2&2*D
zv98krYi1&lK?xxEuV6J@Xy4336k)9F7KdmXvv86N{26gYd3GxtotlP;rx<aQ>*=2V
zft%Rd;exS$3Kec1tRK_u-2S{P@=dPI=1=4$hx5kEYqqw3mPsczcZSzN^kkb>41iFj
zP?VKr^O)3rX3HaB3*4p^`}{HO=1AjQz4+Nq=@oRt`j1b_594^VDvX<}`}Xu0O>)#3
zK_>swqyF@IFWAL${OX(X+l;y$6EzkkvK;eoSMU;HHWRgrETNh=HsYyja{xuZ%sPxm
z`L}L4ER||7Ef&x+5M{v~(mxuSvDY~gC<=Txa&3wD!N1I>ZV$){Cv|)ZlnJxvK76?h
z8B>=+8qT2g>34wf!%q!~J<eT3#ar-M?WtaUtFY9ziihx~hTxB*%?4^_Z$7yc3|C?s
z_gna*$UPNd{&1`r2kWJI&qJ78MAhtjmX!P?Ma8&>br9AvRyka};(~}TS{>1_KTO$@
z>LVGg2$vovok*~=y}LlNfBjAJjk+heIO>g8if8>0;=JBC`sVKpNvdJsy?4tq=^hH)
z%e~$U(vd6)OR_9RGs6c7@@-)0`V(h_7d9BuslD7O+UBq47u#kc8>_M1ZbFlgGpcXW
zmW+KH{}A&?*@|zHk>*Lad7I;K<B5eK?ni5=NyWzAzxMLbEQ@}B*R*C|V3N$OqL~fz
zTtD%6OKoU*4*(NL+HhCj>X`5RwnUt@$qSqmQR@jaBLQ2HX1ax`H!hLCG`sh(*RuZ1
z#%D{9JRhyBc$TJL(;Ra@2|JA$ope~tWqpv?kFjlDK1;9C&HbVu8Yigs-~7ns%}4)3
zC!7(%pL3LDXe>$i&J3x#Qx`}3ZZ}D}y>jT#wsN4X@==pgc1U|sYzkeU$i>jsQ;W;{
ztYoc1&FxuApmF0Vv$VNXjpREEF<v*kw5t9ZIjLA`;9;hbXuxnBoknaMTPw~RV*&^m
zC6nyM5O)E{-ACJ_%mYPE{rMN4^)v4Es-yXO;}Z?^CT*lnBmUt^0N_!44r)S<?T_2;
zWy!)&14}CAI&~UOPCSz?`gYI8P5p3y3#K8<$@6|UZtC3n#q!wcT{Ri|Js#l}qJ@G|
zlurilN5*WHeni2P3n{vc$cO9|kF_sri+Ch-@y+wnpF@H>>rn(kfp;~zyr5SYFyjM-
z)ik7enm&s#W$R=|<)s%A&x6Ngeoje<&+nS6SdDAiF@&j5TFeRX{l{f33$R#cr6(+s
zL0pBh;PQCd#4BcIMIaY((EQRii7n3$2JO|;l6p)seUmfWW<x`0Ax8I9uh1u&_Khtq
z!J&DTe(u9*-GTSmDt0~AW{kCo^PQ=Avv{84bx7!)IBkk^<ccZyiAk)~(l|UzWD>c8
zn6v0YYJz5X<iyZs+!&KLVy?^MW~EoXLhfYP^{yOiLjZZ?^hpUq#o%;NRLU9FtBdT>
zu&qg|i|b9Y-GorQbFrjG;}fFG_OrfW?`s#j_xMk$nAqkY>FpN^Q~5YRtUpDWhydC@
zJqcfkstU2L^g@-9jK0&m*Aa^|JeskKHx7%8=<k#VQfW%=)W9}o6orki{WL>Ar}v_h
z0B9Kzj_IGh`rvVvwZ#0%`kt5Hzj{MVped<^^=^L>uuz0dHQ!1kXKhzAkU6CbgXh2R
z_b&Ii6dluSO$l=JjdtOcS^SuUcvhYK<z#?(%r(^pqb*Xs>i<ut7)=?YgEe*yKOg1-
zUS;?scFCKtv~o~E5QfNFztDRgR#IrVCwUo<m8ZwG4T(n<r<5+?c96BQjHbc$gP@G)
zXom{@u|~1cie9GXu?ztMU`C1J-l?RN?x~Dio-D|H^$B3Udf#2*BCZnQNI{X`YN^N*
zM{m#X+eziov?0nVt+G4zLJKwc9pbmI;yG@sHrU1$E!pEU)bdY)fc6mur9ihN`t(+B
z<<5xL^_f*GMP@PHBdjkyKM7-jr9F8HWtywp$KL8`sIM)Shn7l5XO*UtrJ<QdNTB`T
zzx0phNq}+k0OOe$fwK^^o&GIPxmv93spX0b>s`-jvqEK8{u<Hn?$dY&@w<wgEOBOt
zCyUyaWeQYa9gMwfu{&`uzXn6bMTMhcQ)Fy4E?6MDbxr3M`<<M+Tan6j25b6+Au@NK
zj>jPf==3>e2i75D2YKfj797%b7~f{bPM+8BYTA+Kl}>&0e?z5wbvkrtSfzI>B~$iZ
z1UC}Gh+z}c88CL;%(-kO`PFmm51lDW>Z~|v2EziT7T9W0=&c+|x=^WYUe6`icaXXb
z_dD-?+JBT`S3qv}#Kz>1=PFquA$T5*5P~%A#zL^#k{KSbtW%zAP9=(yDKVI7c1!_*
zK&01kY2BwC!FEpIL#Un#4=iV05w)w8Ix>Ss;8|JHGEAfMK?5Q?Cn{(tMBMb5hZye5
z`9hT;w-)YzA4vi75Vv3}+4%}-r-J<-1>AB22dFP_I`A<m&yWDGPb1fh9|Lc1UkREO
z12vuKg6!h@Z$&fob8ujpH^!rTyaV-NB$hEv7c~Poo<S5)N1X4-D67EhJJo90yyT-)
z{X+t^6r*T?(E%v`?2kK!yqLhlMOyy>Ze~jHnridwEcV`aQ~fB>3Uw^)NV2;88MFfm
zHbG4)AH$EsSZx_lC<t6pcfF?dAKHR11-$s8Jd3~*&*N|6?F(TzN0jy;wqjh^n-Jjh
zsR&PKRY}WsmZ&T>1+_ag(9G+({f8s`(2jmeX>YE5CiwnRN>^T&go5w_7mOtE-!-N=
z&ySW=Y}`fRE{4ZV_uXjkst3XwzTQ9~*k5_Czw8r8XexPNjLzr<EZob~h?tP}F>j#N
zy)9+Q?>Di-b9%IoyO2j6&Y!x?Aph1q#c&FM*Q5vZJbOdWh{pdgLjer!=b_mKauJ|b
z2igZ+huZ4TX^qmK{wf8`HKL<;rXFJufK#Hj9I~R%^f?*MA~{*Pq+(9ZkLlJh`QRH^
zfaiFi;P|RVbDb~-3WF$}K~%S5Y0tF|UPI{crsVV5F1xBP3$bC}>j%+;cLe5aMl1X7
zhZZKfajFz89jGtUV%4&vh7*cLuhVJ$c?;(cyJcf*=*YjF5JdsRZBAxyu8Rrquvm^0
z$0(kRm`oyViE#DH7#pcS=@a<P^BLs2Z%s)IZLAsDM(3S8a#NRjg*y7|FZP&qRme4W
zke};1tXlNzK~rKE(=L)}-c?@MC$YLRqQk;wZsu#X0{+oTi>8iM`7cTrf@%u(!Zg$`
zr=lNkLtsQr*HZVcZ!RG@@9M>2@nl}XWW017cDzz%Z%uIGWabx4tjkg>xM-y48=K6O
z9{v6OZ|h7iUFkL!op`g0cljl_N9yDAk#D8**2q(|z_;!Rj#LtFQ~$KvR!|_AJG}>_
z;B)VHPky(4q05H+dI;J8uJ`cGkVfijb`leidIaanXI8kl3-}!rQ**=FMLR%d)PUpm
zPx#*Gfnwnz@42E_w2JsJTRho^bM}v4#otP6Zn%U{s`d5BW-oAn|0cGLJMni|fD0Sh
zQ)dy!OcIj@szXySl;1kR{eHJdo+&ZY`nxf@77(UM)BzF#3l&TKX3@Mq2Z&GP?2H@N
zHXa2E#ctS>{qjA0@JflRrnC{nwXHQW6ga8inVkk<=|vWjy5M*HVZ(SRBKVXO9wX}K
zL>{m|YPt*HzpiqQ8F$U{4_MjRl_-J}_}gPIp4)?HNu;-xb?DWvi9M1<bAeIZVt?K)
zl#aw+i4EtEwbds^%UmQIZ)%YGz2w0%F0~CuBDjS&WlYWJxZ&`{bUn6sUGyyioqB0v
zcQ}j=tn3Isdwx<jzobQqRg7;HGwuZu#OC_tCMAY%o=K0L8|r%beR;i*X+dQ;kzAu7
z3zHJdd$_TbleGz04#L>&jV2SfO6XPZs;pvwA3yVgrmY8r_?cB9-%S%6+}GnHaDeU%
z0;i5)Nd$TilplbrCOOkD53DiZK=?&@n=F5=)DA&%-Y>;qO!ixA5F$QT$2>P^+4i7?
zD-%f;sJHD@gfPD%Ff{kkfnQqKbDzonLXfqZ$8Zim41PswQEKLbeQy?jU+u3cXzujw
zwU*(cy?!P%;#R7N-2^ceYTZsQE+O`0X#lmD){2Fc_PS8R71b)S%Ora=EuSVj0fm^>
zNG>X>l*vdqeGB_mvb^ED2Mv8)%D;Y@nY)5PFULQBva%)=;Vp90Xld)r=f~&@C5*tT
zYoFDL_MT>LFIZak6Y#I<sf^n)aeI)H;SCL0s&R_F5?KbA%H(56-VEVQk32%^oe0!o
zR6Ke+h%LkMJg`N`e&3T6dphU9xdwE$96kFwD6H!jsPM&`uxouInkIFGd3VU(eLjG;
zj_>DhOLv;c5Ao*v*MyaIY46_-<`e5;*}h&ehk{%z@wE}h7nY@Q@#YUHx|qpVfZBn=
z#fOG`AZhmkZY>Qm{49wHMPeaqQ}WvKUTCXSO{inW5zKFSmx<G45D?HPtheuTYN7-f
z#c|0vX)FCQ*Z6u=NLAF(?ES!h5dhm3aaAppr2RQy0?5A=wD<XC9%XIbLQ!Ahy5T^n
z;v-ms{~HR$^ds%D3_V3@%?#Dp4oIc7@|;+r9c{C}JBd*gQ>b6|6CSHc*n*rNgzGQt
ze6%4qMdVj0OY$X8@v5qC-B>XfMO$fo0U2yZ7MnY2sqxlQ-|mVo+h#l5A9v0{6;p|w
zq|>f@etT%^ZN{YR)c9BZut!8yFTMIkh>Ds?mk;Z{O4SmK{TDOGNh?rj*jK}=<oL?0
zWek_9Y-KKK@<lez&aNIOPcI~t@NF6;6SEGBpG^!vU_#UuI~N4bKqDOzq|7#Ip;PwG
z&TVx@n@}AIug+kYp)e(Ku+T@8qQ$nsP%-n%`IyDa@8TjO^5ln@&ymn8lKct2{+soN
z*n8Be`3F;W5Hb@;91S}B2a!R|OV;X#03l#$V>@gsTCIG%(z`mj&)`YjxUU};cuJ*B
z>{Hp<V&E-I*9o#JZgzI&{{x0z@SHK0;BWOqM+~54y6)UX=3XzglpU!mz(dr>9j-t-
z1qNtQpiV(2LP7#^%F3|S`88KMwTgwGEtTDrCn*4wiFA5(rKCopn{yat$~XE#_*Oan
zOiWzB!F`_ar`^ueRS*MuL&u!^vD)weC^SBfO0Y5Azu%5G;p#5UDouYf?h^Elu4UDm
zdYiu~S<<znp4J7SrV-~F$^Oi;zWnL#3&G<9lP2sifP;D0d46_RVHG}4&s0udyRz!v
z6jO~c8`aR`i+XU!nR3)wTB5`*t~a@T$TV-TZNbS@?*18~Z?ZYxYg08DJAcb^nHyU0
zF^zpre9fKq<W6+NH7aLcJ?sJZt?~ZH$*CJBnL9uH-Fcg?t!vdM#ookBf8@V<f^5!u
zS_Kl$3>bO3qZGVvnWWzc$T=TtdQ7vu!s-E)%NBd`F0fuJZMD1$j`?MDzxYr#WOMV<
zStVw4>?C!dCCM=CPL+0<MCVuKRT0eIi&Ui-^z*C=m1m<R|8W%lE2Hs>h6nJ(O@mp?
zngG-c9>}(d%u@nach``%$Y?5tQ$&>I0>*J#P?|Xj6A%FQb8erIlqKNT#uTc3q!ic|
zkioj#j}v)-1yFuPT(&KF2m|kqNyJPSr~ci%a$&4b!wT%*l;~!p%6(4S3Gr-^S4njV
z(Sx-1T^>GuKD24izGg(6&gNo?5J?!(p3zaG*wMslE^QW^(H~nM`~^IVQ=7{u;lTbx
zS+PbeM0P}|j?U$dEOR6HdY@Kvx4Q8Q{aEWgN}P(*ip}~0)HejzsmGFjDdv%4`%k@e
z==9$<Y?oWi1NP0F+so?FT;8=lRNw#2LH%rHbT>AABQ8Qp`cv`kHJIngMGxnKH<o-}
z7xKmBIN|)f=FwfuIWsh*(lo(}n&K;|6h;1((OucqN)Zg%NEac&S;4Vy<!8pO>_WHM
zU={)!hFA8BjdMt6cqK3R^pKtUrclXbgFu3B^%!W2Q)Q(T!aITOz$6w;>8g80$$0f~
z&#$el#`-&q$MEA$YPVsOsh|rnbys~4&<JZGn%bdp48ZH)=2)B90t0Xh*+^q^=6Qfx
z3w2Qw-Eyf=sd>@_5fv<N-(Z)s%1Adds24H@b^9|>v$^N{3;ag4UahDPXs<3N{yzjM
zaWFkobIQpMk&#AEQe50dAqWBb7>~<->K9j^UQ;(Z$w<$Oo2r!PPv!-<R*=y&`|bMW
zVMefzm`1uEVSX>zE~M<ZmC^U)t`#J|U0Pf#o?Bvn6#pVQ!kw&r9LyCo(4cGg9ND9%
zRh~YjHqNF~IX)qg3#mo6ZgQznQB4s4=TS(iM`{kq-{-w4PHD!RG(@3P9$Foq;8B^G
zs_hCIP}^?htIxX$e>I^6$5l!&+D@l{;B+RZ-CO(C0sb}@<s=y0AJq1y7cj^G;oW{W
zdc>&>AI*AHRDE7LUvSC;$0K<o%*s?mT(VEm`Y1<0+0}bqV2;SIZ((WLqZq>tlX|s?
z@vaNe+~dd{sXM=n(3?BF&fZ-2FTRv8VMnM2nQ5dvKfC1QC9fmy^#zKV7h7)F;(iHy
zJ;k7tTyI#wT-b^ERL&HH%oDY!xCBE!a23cMmVJVvkPp;k+88iYM>lZw-)>Shw68Mv
z$YG94*@7}U7TJ>qHO3_j@C|n^+q#rt-3N?RW%c<d!^#ajfFTL3xTNSQ4%Tg1M)8um
zp*vJ@_T-Jxi9-9Hzsc6ayOvD;=EKWCffhb_M1f}dH_-s9j$iS&UYJ*Kjb9a_nMvJG
zmU4%}b296n)L8SVI1sYG&Kd@dxp~?<o}Z|#Wfu&!>|CmN|K@*pjr~yiwp(1s2~)$a
zYVQba#0$Bpr_%LZgO4qpWm|I;)?Mx>YeJ!?8}YVhU%Yr(9;;vjtEQ%>OxBoW8ll63
zplRWFLhqHb{|XX@lVukO?4ejoj|1v63!C>EYzYa5r8h2h;+$FSnkE)c)r`KsG2#Jw
z4s*DblP71@lYM95D=VvOoL*FXx?W4Ao|<iykRt>NWui^KspOJ32i4<~1ce3&CC@n!
zMi!lX#(!5$$OEX9c-0VO<tSH;<{AzKwUU?ajEL68?;!)l3K9e)`Fjt3!tEY`6+PtM
zzq+sH03c5L$^UsymjI!S(Y1)pD_XF2a0ojG;g{9rEayEg1J(`AFDC0PLyO7=2dsm6
z(2lwc&xm2%tZ!4792Xo_>hYG{fOb+>{N?=S4G#2Qn_)dQe~Dnw5mou&Ciow$@C8uv
z(#udF01Ev!zO*tT!i4nUl5=pTmWG%Y99W+-Rf?BoV`Q>PchI!3Q<Ctxt}tEyvqvKv
za)6OoOvEij{Rr>nc9x_h5>|E0=j4juc&**VX*OVlXEpPV#O=Z^i??b{yQc6%Nk~AB
z%-yX#QspZL_jy)kAK_OrO%}v^d~S#qTB^)hgStyp((1rqMXJ-3GUp{0KQ0eUByYmB
zziK3x>7;;BiItTzp^m?JkgbiY5Jj={%>$LUgG#)|p3QdP<l-6={6FH-Dtk|ylFMVb
zMrm@qvXD$ibwSkE6NgREMBllYtizpq<+ckLg|)}u#~$JJ^ai#umLd`_q2{Rtgzm=7
zoMHpm#9qj)L^;j(R3wl_Cd6UtY@oJa08w2seUT?%>?Q=-uYjmWD$_=!4Tk2I0!Gu{
zGJhX7`)Tt~XS{oZ7lmrZrUbbK*3wQ|pURA%7sgTbjJ&|#1GVMBZ!Ahc&4QYr4p&in
z-)1;CF#3`Hb2WsDEy~NSrq|Rfhq9-5)TE%92j~vX^j}E%obPr3{0I_=N6o6Aym+0E
zKA9BTQgnguiuh7-)TL!ifE^Ph!8qH(+&LIU<~OYHbux<_bAtQZia`EhU2jgyg!N=>
ztS)f%3l<>PUk8iC=QBV(X()Isdbf$l&1zn!mw@Ymmm7!CMe8B_aL;Qs(@4<${q}qZ
zpTLXey*S7(Yvi!D@Q14Uzq5u=|G<IiVCZ18I7DLYEV^xVtp{)XYn2`q%v$K|LC31B
ziTQ<cM$(3NG)<qlDE}vmm1UO>56pzG7#RE~My=dqRaxHRrMN3E7#@DeFA<y3>H;au
zt_>mp;6pCLL(J`Y5$wph-^7eZ;()&sz*q5js`aTfQ8?Q@AB#qmGTSdQXpt7on{Q9I
z#xJu$`qLt0G>FD3RfT8Q1>Q$3o4Lrw1P2Zax6QzAVmdg~5on=>Ph91H#dgGe%2r6r
z0y{$jX$%Jb^U<fq3(}YKEdS_J(39OQ6>fk#4wPS)x|H^soSVz<wMG!{xqiLgMyaiQ
zZXB9LQG-zfQ;39v*K~)f1;L_$l-Plvd<O<(V&)4l{mGE&2{D7Q?HuzU8)E>bUy%fN
z9m>QIfOSYT$dMZ%<NOMPPLpAmv~Nyu?ZiEpVYlc@;+>kX#rzv^RG{ZOBE)FWT$sVH
z%Squkj`y%d43Qp{drFF^cIy@_dMjDruv17Tgma6Dv1EeZo&7u5WjDaU?a;D(MEfI4
z0i!5K`i}a-6AL<vV5W!aTBkqdQ4K}MNZxoV75=6TL0wnkoqSzX)TBH1fALH4ZHTN7
zYj@CJfmEoEX{@2;i3QGh+U$1RIfM7ByLXZ6l5Up7T!fN=b(0_J>-oV{`&!=hS=3pl
zzY)@WO4?oSmxlH&K6_H%l2zE`c|sIQvwk*7uYRy4jy$y<dMDhG>S9?Hm>`;7tt(7|
zyVY^P+^*TZOsxN4`)NNI)YCtwgQ!{}=u)bI*X;QMX0!+iG{G})4y5wNh1nqnBqrPe
zjGV^(-xGjsEGfD>LM~80ad&@T348=tL@6tZQLyo6wfxW0ND5JZC;DU<Kmu)tvjVaQ
zKq>AF*_C(>`#b&~2i^7z1(eZ<TAP`xP(TAkLWr)k_v1UX`@{dZaA5pqV80T1@Y(C`
zw`ZgOaV^p&3I+*!IQV99-1t-s`bg+yeTsxcoHM!Enb1aE(GBwv((Lt>dZCx`v%(<C
zYTw)r_Gmv80stpZH_kfz%WJSBJ(h`d*J`x_b;xm_sj7^AYMkAVf?93~YQ!F1RLiTm
zd2d&YeAQW=Y}=@>bSKCb;D&8e>gMf*IXo%~WwWvw^NTEBp5F#>?TYYd7LXZ(OYqsr
zJH^Zz6W)F*OU6KZ&^PaYd5>XD{^3yw{(h16(wlWq9V01XtI--lE#0L@c~5FFr5m&3
zdni=3a+#(?BRTSWrbo<T(Jm&-)h{wWFpMqj7>|<v)jfJs3nKH{^GQVcW@;Ns>ZK<v
zbSu#wc=|Yp>cqT~Z%qaK9&K#X6Lqk<zUuxnlpq`cRmv?aoyD!+SPTKZ@?xF&s{U7>
zN_hlep6uBU*__6y+V;1YDAB9u&Y|X8D0^2j+P=oz|MddE0Lzms$*?;CIzTs>BH7uh
zx6jh(j0Y|c<1{!<vM#V3qpNTwcsc|C6MMfFLbk3rB+InHfMbGtarCzgaf*ts1GWrP
zhcNoiyAjKn4lItg%PeZHgI~Tp)obUi>w1<s-+ZF^6Bb5mm<P`Ov--ll#=<o^pQ|l`
zmYTV|w22oA<69suB~!ym;Mv=s{W{8-^1Dv7^hUE~?n{4lqk7OzQAsz3j%-c<6wojA
zu?yLr{rg@IVFm`NxPdF%7qLWdFs=quV{FG_NyNHjj-kGIU+T~oT6O$GN;@QN#2}`n
zw?!iG`PDPiachhu%;@C?TjP#pqB$dW^h;qGH}{5VcWQ^gWFfXQcYB1Lq|ZRzEu{l%
zbz~U{T0xC>eE$-Gq4Kd%$VS)lraJI|EifE;Oxb33LCw10{{cIZb2vP~F>mKYmQ`#s
z$q4+#nopzK&UL8$^Fw>bQdc~Rh35!5wDb<J(>~#gNQ3kn6*}p6+@2IL0CLDI2YO67
zM;wEbA;kW?E6xuc7`Ow#x!1x&Q<qoMOcWC|fW0dZYSp%0luvb5WTe?UAOw-YH5PR*
zE(nZ9GCUE69xn}+D&@adK<tNtZqh%fxIo>hGXo_qZclOv;gNO*CMuWUbr5k;;k9{6
zI8FbuQY)%8HtH4g{P>YoVeuHJNbW(TkegK@p-VefU?+hUN*;eg>}i{MvYO93m5ALz
zeJ&F=@=)nW|6UiI-$`)6YO|{LL({X8ikcCz<C18RN{Bf72CMk__1YzNn-1MTLytIi
zfEO4aGbkpfk-t~fEmin&lUtb*)V!$<f8rMs=l=o`Ug+q#8K+2d?vh6}#l+;}kXn#d
z;O@#2r<qw9ry`F*&!HnBd)RC+_3ggnr`5?lKS8)wm%otG)>?IPLBZ2~WOLkCgR2`I
zmoyy-+&^bpxKA%IZe(wm76KLlsR51$CJhhcqiQ$&%^(Y=6G1W&>pvlb4BTZ#$LRhp
z=LxY-urxdd)5|Fhis1!m{q8d3P23922r$2g3e3j=*a6Wvb{d<ld8K?!f7u@lKurww
z(11ia_-`=4@XcmmF>_QG!vr5%blMN3EmL<NY2qnf2Y_yHjwmc89xv2N_wGzb;yh61
z{f{Y`8-DHUxfVh({GR&Ge5aHxdG{v4k0j@EQcLw$C<_D(?7_OHHArS9@i8MYWa}5&
zxy-)7FS-PiLE;p)z_-kJReGaT`zth}!Am~YKMux`b(=9<pD*!ju2B6NuvYLfZ*rP>
zhsE6flIdkQH%ubBHb<@gv*0^a7@v$@qa%jyeYPV~l>LS@GuHCoXI0iuLyp^5BJf?t
zg2@!Z_J1^2i+2U?S4$3s%lz1MEY!_d#>TQ%j8vIQ4S4eY2#5%9R^|*OMD8k)>k?yk
zg-J205gyV=B(2J6K9$CChwSAg)LDe)#8PO+%&}2T80v;CWYrujIK*;$u4!Hqv$PEr
z_M|Umekp=DVeBAWu4QUs&~YrhOuz)>rY60v`(i5o%V)@d@*N|Q58WZU974oYF^u2*
zyD&2xfJm9O^GwZc0IHfo{KG)cN-oMnDbZ_0yw*op|BP<IfuG8YNH=`}oixC9Rq7~>
z_{Ofy`@K)o%iJ%J$}~jg7c%<SjSc)lI2gSh#xp6h8?YcByZHZ2%FwLv`s$PT?VWmQ
zaAICxIayMVyW_-vwX}8r#l<~^VlWZe+f1#fW;m6;sF8b9$L-Feg+t~wjF;4oG32u~
z^%a3|X3%VS7JG1ll%eteG4+n&b^cA%=p84GZQHhOJB`gIX>7AmV>@Zwu(547w$s?o
z+5hLg-gC~^{cW#%euFh@W~h&y=6;qXzU}MBd)--?Vm6DPTdWj<j#H-sy=RV$23E<l
zm#h|<kMn^MB<RvVqgT?C+1${4*WlD2cI(ZHvh2*%D_LmqG%KF9E;ZU)+w!?TUwiom
zK(51Yk*`&aL7!Et=GU5l7jZoJ8==kPy1D~;SQ`?`&8UFj-!YLG;{_QDJl-5R*8sZ9
z)oCi@t{sx;ug6j69s-r=#S$ZkXH~0qW(b^!zjL=4bf(*z=bMp#kl*f35R4VkMP@pv
zNwC5Bw`LdHAb5NQQOOyn`aPXT!@eL5xlMGf;QXlJko7i5D|>7%mJi|paDX!_SDzQi
z*R%nk2a4fIM!lDlq))oULm+C~5bsg3CKVf)5qc}_)GS^=03`~EJCrxwsIl|EpAFo|
z@>7g}?YHeT!#K5SQ^1&VK{P$CajpYzv7i1we#~!ydws3lB~SX<0i@yTpj^5?=G*U`
zoW;B`Kirg|Ir@-k_vqgVQthy`=T1^&EX6LEliO<6!o!-j*b<PB{>^m6WF|OVVW~f&
zNU1*nrTKo|>0Fi~!y=pMp6gnEEZRNFn+CR`#@~S|N~R#u)tp1~n|=31O7fmGGo`0`
zd*Yg?Yc8|Ei~k1mKo@)!<5_q1Z87q&<MS11ny+h>sY9Y14eu%SRFF7qzlTL8xaP|I
zNo7wWviHFB?_tYdn?@fj&i9nSnlKFA?($y(XParzgjh>I@ilT8)uT`;(&&=nFAIWd
z7T);0hg^OO%Cu6*Hf|n!F_+2wfIhsL6dcvPN*caHRu6i+CZZX2=g^Y}3y&gHHV^E!
zs}L&P^409hIDMD^7onXJGGF!ubi%eUN_!)nUDh#0|9E=?Jekbjqs=T40DB7Ig)LA7
z3CAtn0>1o7qvSFlum;ZyCd~SA00rFfOHh8E2ZLN1yP1rx;QLJDL&&#kYc`_4pvrP0
zSm(e&t-=if@QuTMgdEGrQ2SY?=TZRLpe%3-#{MV;T?LbqfB$il`?|qAV$faCK3hP0
zQ8!>^AsEg$tAq-J6umW&g(&sVWm(<ascfz$eMh^yRyGd$5K_i>`xwDv|A^*QX(2Bv
z=8ZtLNpWHUdeK3=mWo#l^-18D`b%k=<B)pl8sp`L=8brp)SVISg`Eq)^1?9*-A*K-
zP`0Cg_L5)5=VW!sunx?b?p!Q>fD}j}AngcqAxyk6uAFveoGNm*r-|@C>dlp&@DcRl
zi4w7V!;`QGL_Kr9|Jkxi!^dcH`Y}NF?zpt%^4@+fIz~Yo!!uRH^%K=bSVDKp>BqZ@
zWbnB4ad%!a_~h`?vlN^@cWq<F-oDpX*Hyduv1U>G(-uFd8>|WcH<Byw5CZUmAKzUn
zY|pxvcgmJS24Hl(Zd(5P4y&0phSE@AG$5V&zva9=>c@o7J^FK)7)UVqzhwf<fK%!5
zpF0F5XrKd?{r32*e>KTlflbW4>%Z6nRfpqf+KvOvAP+hRXscL!VDv#(nR}LqLG)aB
z{ZAJ(a6{}q5&|J<GNk-_-dgZNpBw_NOHj(^+>^Lgr`(W$hlVsMvT`=jyv>D1&D>0H
zLIf1sQE_^m`)l^!!0P4%w6i^({tQ_DF+G-$IkJ*p?l!FLoiZjaTBquX#_vjb;H{ww
zUE5@^4wzHW0yy{C=-;1l6IWf02*Rf%y-f}((oy4V9+7f~W??l7^utnqTqKod)R28O
zn~H7-$Mm56i0w0vNNB58yQFgiKj~yF;GmasL()XtjbUktdTEr%afPs=-?AyRGWA#q
zNiO_usgQwMZT(9u-0AOEC7K__*kJZjH?EV0Jb7$EvEEngm%>?Sa{LlhL>?AcfYNn)
zhWQN0M_o{>%6hUzf2Kb3|H#zF0UdJ*(E3vx`ak@u{Kanbom3G3ovaY<3uSV6LN>YB
zViIXkt_23Xlj)O8w}43F60|Q)=Msm^J*DVtuRSwgK!pjGUUodik}4~ZZ#3gRp5E8I
z*@d$+^#c^vH*WtY<#g7h7C!8L43RXObF`M5FV6D8okoL=O!xOIdGIkR-z4B!-b327
zhlD)LbwQGR(~Z+Yu74adDR#a4^hH1Qm%A&_x2Y#xN#uNjunX+WXdEc7D{YsvSb5x|
z$d8eQqJOybE8m#TzuK``hbdanP+2~^!zPjWDSnJ$bB|x-{QJ+f*j*qGmN?I$oUsBI
zpB?!TUvieZ<Ao}0w5B(=#=a4l0$<GdbxK$GFlw(QF7H8)shEh9!>BlDw=JcdZ}<0q
zL1B0(9_u#9^)OZn-z@pmBv#;z|7>$ufqfIQ^qxc4^E1Q}nC^rCaU{Q%FYTGG3y~6i
zaualc^rDApYcIqsWPo4Znn01YL(Q`+Q~PU*YkDH+9+h)xa|z#nc~xA(VlZ0n3CP7)
zc>F1Qs>y=Cn!g-*jOx1q2SOyw&E%Xww?uDig+g;9xOPVA&m?!f35fK_yMdmaxRv}?
zH_$q+yONF^FqSnC^27BrK&=naeITd^9NnT-!oDpiWV>vZp#tBn#>P2e!O$mFn&pt^
z^izi5!QM?xrkTUJ^4Qq4kx1Cb9Wn-4eJDln-XC-dFNjDl70vdh%6f~bD{62B8ogy>
zxL0m(GFv0bhj$80MZ(`lXX|q{>p**6?@qt+kP!)nVxpt{-M&1+ZZhoUH*W2;^(u`J
zDXPw2(nnc=tbq_uk<IAbh+GrAk0G9VE>xF!hq7zsm{#~UvK4hjf?h}CC(e*qE@h`!
zs=-~k%eHD2*Z9pup4g<&;#^~DgUiCLVMaRQ%gx~$)BH+IAsnU&SzQxvt2f(Tsqpr$
zq`zo<m4Z1KdBn$%FQ%LbdpJ**7h8X8%G`nh=lNzKBt)^2kq2&Yk;uQG0wMY)9=2K8
zp@C!{rOn&HDHRE7Lcn}a_YGIK8mCBYK-PI{-xg%|g4#^I)BZgGAnX#G&@SHM4IO->
zQv4XYjrR;4xOqpTEL|`j#R}AlE59i1eQ9NU7?cUuqVpx}`SyR@j`_E|^poyKZTX^J
z$Fs2k)sE_TS*F76FAy6alo~}IDX0DZd%0*hEpTy_S@R^p6lrQmb&rSUpPmW%DPUe*
z!)V#-M=Ry>gsF_z5rmsCpjEagt)plU!x>CJ7tObn)e$6+gDH8UHJocV;$<yYe2)>{
zG?kz4kmZ$Mpm#iqJp`pIyFU$H$j*?%MAF?wzDbsX_ts-ZgO~3nA>VVWMqG)zr@C-{
z6rN_e->5JyO_3q4&|BAoF&rWdqL;v3>+mr?W3FW+cgcwNX`LiR#V~4OlzrvZKfg5`
z47~lJ-$|VuNyhCL_S}@h0-KC%zYwpE;m;(427UAce6TYEJP?27n{{zC;l`u)bEUCr
zr2ZsWAhfEWubHbGvPA6j<VPQ9f>d-GMjdG}0}^O#nJ4<{K*70j0D1~mw3@@O;SJ8H
zMZs^wUz{quK+B9x!iDhk5P)SYxGR$MnVz?6=w~mJ7X?rX`&C*|wA^PH2e^_eZ{yUU
zy~C4`C`5S9+zS6o3kjS}{ZGRWYj7a@3*yyD)$qaX&nQ&ao+(qW=0Jy7$FGCjiVrbC
zWV7zNWJu_5w?e|jMf@x%!r|fy;-FaEhZ-3xQ@sctuL+%=xR*?6kah0o=A~Y-dXt`B
zJNU=8dsKxRQP2+WWGk<RnsWW7wSJ3o_D;T*iEZ(;=Qj(PvJ_jx0F~$DK<IbZM5cBl
zWjJI}b~T-V^E(JLq4_+~jP7xE8{KABNBns3a|5qWipy4roS5xR6tNS<_+rTOVb_wN
zL4pcz{X>0jwVmL%=$v)$?RPFbNXqR{scZckmfZC5u^<Mnswbf(JQkZr<<{y_;HsT2
zT~Qms-W6B6>`gzxRhEc`F#d8J0|=eYtqzPB8zca%G0nZ&q;h!Ib`N<u>uABYv-YOm
z83vtf<1$!*<EUaqH^enHMKjP5@?&h{c*4&pcL<WAWz&kx!#hygn$H%}_Hz&b1fMSO
z^}XZ#X_hjCH%U{5uLwbWp*_i$aSE1P<#>p|FBRaG!=C(2;;UDevDEFo&VR&!o(tdq
zMP(I_T>pOI7ub`|`pp^z<`09kM3I=?VESTW<PG`r$|DnO7mv-k?5&v8xrBU#cwF?e
zPTmb+B2#iUQqQ%$^t<=QIB_|GsH1{**XRPA5iNPQjzJa=+mWukBIjEMKRM70!Qy42
z^=g^!o2Z?{+$SqjWEon{3&X*1xj<cW01*U70-=BCL$i&6;DmAy>6J*;ZXVNwukdlG
zbFPY_UQ^9Be$4Ef^?UXn^Lx#oBHB^9FZI9=uTnPk7m+yem@1xMb`L0q=P=Tz<z?%F
zjpwhSLhD27?n$dn&>e;oqxOgPCdXnjTLTsTWpTK&B|TEwEU4oF>%2n0q+~msOiKad
zOC5R;3Bn2k@P%TI`WBhbId49wRYVH|zaRrqCAI|FH~#ZdFS=i!zG7jfK}K?d4(bTY
z3q!ZT@k_;9k6CN!jl%=Rd_D@v5aNJ};6fC`ug2AJ;*B&EM9K;DF;G!HGk)&0-8KLQ
z*eHF^E~l_raCL*bu@3_RH=h3&oy$+W{>~b9UHG`8`Jv&)j(HeKfwO+8u9b0=Zrb>&
zxnWe!dd8i}r1NuEFtMRGUG*(D@2t}~7$Y(kR{WYU9+b=uofK32?!TSsKknQqc;5I^
z*S&a$&dzaoBs-z=+aUaQ<n>8Vt>&kuJx}(9h5CW=_Y#pE^3?N!o^1APq=lsT2kJ~a
zmnOBxLQ+-=W)*g~_!QII`6*a>l^3nH5Rsfh9C@e4G>Jz%KU$kFjhR&VJ-eih7Xmg`
zz74rNuD*fP;}!zn7Ft?5iJHP(LgTord_4G<D(|#OA>izAsT-nBW!Y)-CuO1<-t{38
z<2AwY5wESG?^8OH=E#n?WYH8~0*ax({LAM1t5~#slKkT_A}txzd<gsNJs5BTN}le%
zzVKxS(Ne|Hq+$V6TDH$VQo$_tS2%|Up9-BFpa$TJ+Y?_~I*=blht<+o38A}_vS4su
zF%$j_4MbbTNh1CJN&qBGkAHEppf$o(-^sDzoKXOEY5O*^L1Wl31X0A~w7SBJq913S
z5Q@~s7-Gf}yoq8%h+GX<A{n$Dl`?zgXvw|lb-!F3+n)CjT%{-|8fNx>CRt@GNq8gV
zM!3YTBin*H_0iSWg-0e?W1SM}n1T+p(i{Goiro_{7^%z9`W$VdA3M&^QaH)<yuu%}
zSiExOA%-fa``EVG)BkLZTZzkUi^o-|zW<)E_WO=_2Kk}dXD#mh9geBz-?o=)a31|)
zCe!;bjuikDoo#=*BAVrp9r|jSmv-wFHa}^x@$~S2VGODU3vE?{s^+1hBEI4L6>X9!
z3+7Oh#CCC}Hv*=GXzzyDh&O3J+=_xM+jkuFX$3e1F|(um={`Ur+Udfg1<S|RJim|s
zoC9^+^L<!}66#1VEcF1Ks)cb{{N?TYxRKDqQBXc4zgHp!2lUAakArupXu%`FZy(_U
z?n1(ldwnq8Th4%p|DpAZPKE>?vwPUhN;DfFgLlmvTI=C7c!rpWbM3RGBImMeJa0gF
z%ZB0D_$#s(5|?YN-qbXG`oN2L77U~1^rO?+d%KS_HDRKis6owh=!9l<J@BaD7j0kv
z)JMI;q9eESD_{Mk?gEs~bQt~H;J9P`D{D>wV^f}?)w+4nEj=QWwo5gBo!`sz5$ls?
zH!Y{d3~_My#}d)41eF>hcvXCN6kk|CvcgD#lSbl6^YC~Nc!^?W<VS?By)i5IVS+n!
zq58l{*r}RMoq}>^7stINqwZ8vBDE)T-<Xn5zQ?}wgk|9~-efV#O1mK%W7dPDOog6t
zC91FY*4h?5w&|+<$7AmVcn>6Z`cr>sB-3&Y$iEP2W2z;CISHMTnN~iR*Chk_b<a8h
ze{g*!V!#5s?%73$k1@T`=fQg%@FsH%zlv7=A|uEV&@%+KFKc3Vqw`^bIFC-`_`(mr
z_wJSnVFi^*ZeWJOJpZxA9d{21lp!S`u93ZsrO(&JZT$brt{E(&z=}F^{$a1LcD=WE
zmxL)ro5IP^@eE8YVndIfc+Z!QE!~d1Q3%4ALKu?23QC*OTpY6p=tOeHSr^FSTjMB1
zPI&!7hIhP<Eo?hcAh%5D|6H(`WVO{_@-DEIcB`d$T-{u|3zrtoDv6Zo+=?7EN~TyN
z9u3oeoI`4aPk4}*cz|<9U#mwKuE^$6KhNIgI5(nC=Tm$Q>#(KW<h>KUp5!ahn=St5
z^5su8jU{yaL4Cy9F|(!aBC1K|K~D0E!r0SHOV&Zz<!<m`ZM`}z3b*!m+w?vGj-=CS
z;dPo9$FHZP6iv#;m9q*1H{wBP%ii~HACCnf;(-MFLAK21iHRKs=!hp{w5~3Y_=Ym6
z@y-+m3bib|;~NPV>9`QUIl4d?-1ZJB>w6=pg4+0(d%IDT-#h=p@o9B>fC#AJfQh83
z$k(>|({@{VKJS3SDFEz@fuA=Gf&sjtzvqX&_oEHaZA4W(|F6*b=V^6Xm#BH>xRc{-
z(ZVj362IzUw2&uf7;-ed<|sKOmD`tCY~=A;RYS7=`TmH)DZ})Xi}UrY%?}=kB^}e7
z%R08IGM&Y{rB>YO{^#&h<W_w2r-&u`HndW@H_pOA>*AUr#+uXDd$rQM_<{!)O>X!5
z7+>;2j64_b&`tMW+I=&W-{d1Nk%iqM#ESF7OxFDZ@gn9NzYoo_3-0RWPOp(<c3?+r
z_RNVu9{Qa3sh|AS!!y-R<rp2otcSs>FiX`G_eM^Bxl#~R{JAexS0t;hr;3NSE74q@
zuj+5fIsZf(FW%-g;;_!};@Hi(U|6Y5Ub~2R_VmJvOpE7vN%--&vjGdRJGAE+*z>#C
z8BJ(K5$8_T^eD*z)(UL29wq$$$q2VzU{JTKY+J@mZNQ9!)S9N`pmcG-yQe^P*B5vI
zyUn0nfbRXfw-(zl|Cu8e0MOJI=VbbWJWSon^syN_9X?*s!hC_7J0oCz_CJL`8<+nT
z`EK7?vszC9uKM80NQl2D4<*Xue)Lxre>4s;VJZkWC|csk5gSxdR7RL_zQ>`cpCr0e
z8267}RT}a(gufiG7(I!+@k@_>h|qtn=kN7A4=1g<JTR0P+=`Hg!+Sp0Rp@Qqk<}{f
zR(a!8?6R3QsyblJjNYlK&8n<y&Qsy1fa#)!k?p|<NPv?YOB)H#QOs6OGU+cicZ4>i
zXk_U-wF^M8G4YiYKRfM=VYa-3$-o?cK6pQVJn`yVk@KTiE%G6<qeH&194cIkht|US
zS<Wf&4^QiN)d*EkHiu!E)*2o;b+4kk!uufO|LncbL9a9Y(~rXpsI7|TLVWy7XIbtI
zC7{8SDh&V}n+iYWh4B*6fQmWJ7>fS&|9GC<*?Ikk*!mBrG*BJZ<SXv@;~fn$R-@ki
zq`DSh;aLtw?{;Wx@ZdXQ5iSRST4)mk!RjsnBw#z!()oNPs5rx;avFz23kO(^{hzGA
zV8V1OqQ`}vmP0tYg43#k2VXLem{1m@Z|*Fdd}Spgd$gR0lUxH+m)$@zGw~-CI~m47
zj?dn}A}gDm+5-d;eF?7wL_p5<E(*)+kwyL@>0x{OWd4TC!#fowwqa!_-Z$sPfwW0t
z#>#Y94TBeW=Pulv*fJX5!P-YotY<fKXXlO1be12H`~BIuDg&ZT?lscpKJ3mkvKa`q
zWoU2M7(bBm^R5sT13%QKCN#9hiI&s$U)HehUXx#RUT?=K{&qiJuTy7u%AlzF=|EM3
zld0#ib_qvB5|vw$Y`R9=see@Ben<1AMtnVWr!sxMOd{ENB);xW0;Pq-Wo44dK_0OO
zsKDv<;Q2Y%m^>UnpyfmLA<x_ZTy^<xrwgKLFe<?V{OO^nTPj$X=}`E?@bEy9IMPRv
z#))<8K61YzIXnPtol1XN30ifNApuwaB%SY2(4;!}<y;v+7ySOu{PWE62d*TQ9NgB=
z@GNo--cMQbt=qcU{iWoCe?jSDOWtqU4ZjR+u=4yuS*0|SaT@9P9uF-)k4(*zi_c&L
zwB)jzqPnLDJ&9h@1uhxPS!wyz6F22i6RjCu4MhTChgOq!*QYPDt?~FWa0JJVf?kEy
zyMFpr@8L;~iSk->vI<<+S9L+rhD-DCl~gx$M?v$@?XaCxe%}uukBYNrYh9M(6b?s@
z2Cr{Qk$~_2GV?u;xO9bxYH)!5UfI(~=MB@r<hGNckZ+M!;O$v1Qla`>F+e!})Q2t;
z`rv^H_uPXE#vtml7_L*%m@HLz1Rm{h#7i)h6?pnq%!L@z6&u9?vo86cNvl7pwwd3a
z*Yg0&&PwlLyuMzR(p|3q6--Y-6(lC27I`2qK0Mj4xzJ$IFL6D+?D)N*54<Ob)UakA
z1xJdfnWx1-6cA}wa6m=FL4yH2ZS<~(-8onU_aBj1|9?hp+^ZL@keu3rV6cbm!<#Fh
z`jyCPOcu(Wu_sqB0+QFtsh>sD3rX1Hwm*U_B@+rKJlm(?T{LTVaD;~F))d#i6?&5E
zXl@T9H{wUKLY!gmDB3+ta_XxBp|ko5cSM=v!0P^wbK#W>tEwu~vUdi`u!t=RcZuX(
z{<t}mgFV|%0n<721Fztik#Y0cu0yJ`kIN!16YSBKMVc!}1xW`M0w+OpZvIT<FK@4H
znHl9W;_P0h8OE-mZ(W*YPe0i!E#b2+$)D*i$f?5RRWO_8jjrSU!Mwz5yMsde1qZ%p
z4(1!UB(9HIeoToB+(^bcyX#gAD4`Bb5&&0x?P2tp(trq+t6n5H2qNz}cdmJLGV|V^
z|4Jc#`Hb-VUa>}fATW{$^#gh(rSyYsOG^iUZQ?AYN|?&y2o9xm{Q>}M$+m%ySD8FO
zVA?=uc880`ig4s1YO3vW6K~*#+2JNi6Quc*pa$-&Nu&~F)E0Y6WgOk@F)LINY3nhw
z_j+c@n7;o7R#<Z>+lxn?)b^$h9<}8>%E|xL0<;-LZ6BiH+CMAt_=Q{!-8$8w{>u#U
zrlDJrobfI!C+AY@WSA5tk_Gd|M)K{u7%pyo8Qap|ji9KHqKsyLhmanQ1V6z>TcfBq
zTQ&L|G_&{|K3*`c!CNco2=TA?CyXu-#fO7~i>|{^SEPxp%k}?g=A2YG09~BlLZvz)
z`WIAJx<LIh>MMUp#Tg}`nOOssf;I6SDPG!#IC;tsR8&ju-`SZ&$nYraJM|LHWC`iI
zHhmpWMMlvp?><vrw9@uvMl_nMy1fP^ij>!lcLw%ekU$H5!^xK?-Wnc2r(%J%^6947
znsFOc1HF0O=J)Kb1AtKfZQWMPwa@v%JSav^hieoLLuWzG&$luCXVyK|Q!z0Rz=SuY
zcNRg!$xw{7N9g~X3r<S7ppKumpPjO_NUcIBJ1>;t!R-E`(=>x<J#G-xI{Zy%i-xMj
z9CcHqRIUet8w=ULgJzQ1!b%tNGrOzzao6kkmj?RFuBJfZ!27_w(CD4LJ(JI2W?4H5
zdLf3DywRy^c9rRure*oQV>`J(NCp30mUm<wjy~ad)meI7P1<Y%85{M`0<$dYB*cBf
zbf5Kfa29opBm>oxEOjwXvOg>FkeL4-IoVSj&5;RQd=noTR`DufXgHO>be*apb<!EY
zBTxKV%IP-?{T;P1(5ydv&qh;LTo;^u0a3N5`USha$lQS9OL!z!q3A~>yMoc<&?!qQ
ztlDYRr&XBo2?BeghfMqTOam+!053ZWUY1T%d;|d^5VH-DZQlj73^G3S`P|;)iS5U8
zV;^{6u9AmO-wCDwL!>H)IwJBa6d2$SvyjdMh-5ONJmnCc$7?0S_tFSQi%4T?8nQLa
zZbE8!-63EjE*U>ME1v8!yQHk!6%Og=tQ9r{mf95;1!tC?2KksJwUnh}{GDKPPi=T2
z^>~3+?z?rObv72jv!HkSYGgbB5iVhy$?5XG;Fpf365WtOjLLd~NxtYL^uB2L+-i5d
z5wLRyTSuII!BkYFUC$>R`jny^`OqXB-dD5Vl~u}eCz2_4GTXt!1fAPiO{wW;TK}Pd
zew~Mxo)JBO2APOEgqFyyS7eZ+O+m92sYY{eF^Q(HV2{wE0#hD!@9SeFYLSW4FmfU^
zaU*QuRHuwd!wnZh%V9u!@u~67w(zvy#;ijsHVyYF30#d3oo;WH`R+NlGavz*b43;s
z^7M!484v(_D<WLW0mxO1m449^B%uDAN=?rI%#6Fnb+c%MNFwb%%!GsRr0Oom-va=Q
zr}hmfM+v)*A8f$7Bheyxlhv@8K)i%jDL!j#leO)S{l1sql+m~03Ve@0RgaRc{!(}$
zlxO0Bz%=JBe~!X{W~T0;?6hH?;eG-WKz>)8b0_LM<}mMb_Mb_RiFIdaa#TY$l>_x5
z`weSfSxF_yL#AW3d)`=;QzxjnoLmul3FkRru$TlMM`VX!#agvxXA+gFn<`t+R}Vfe
zYh)X~$}yU}y>}*8$bZ3*)ni{7n9xu_CQGfb47b4sYx&f#UXT|XUx`0mMM~hjud*t4
z@#ML_XQ6r@DdWb!qasrv-4n<-2pALncE=I)J&AN_{`@5f+LPQu{AA&3zZ(|l*cwIe
z>`#l0g$Z2ZFjd+aPx3w+rvY&M%2d$!xB-!9_A2LouE{5Z=RbWQCBcEb<ysgiCUi^S
z&xaTgj8T~z1TX;3J-A9+BqS-gRx(&1g}~0y2ivps^$%)kt<nNNXuH`3%`Pc=(JSzu
zd%v%VjD~8p=-_QcZ9$jNH<xz^9q>vPM<VoHC(e9}#sD8!<=z8{;5*oaimT?UC1N&$
z!VvRmec^i?lpwWihqr_llb?vm-MaX&d<ZTF_~o*j1GB7j!EL?Yz9o&sEt>SAsR-xp
z%BKwWnA<~oY_M|1NTUUJ(VWxn<bA42Efzd0a8<X~wPuvRQTS~0;!c}Qp$5JpaDIwS
z2uEHuR9BS;uuB&-OdzTws!%<!YjaHZvMWA+hfFA~jgyw_P|Vp^WKk=fr!hR>;OdxK
z2M&3K-W|b#KL9c%5)jzvo72!23jr}aK>+ALFMkahYz6OeYO)eo3y3S}-CqKEP2%~>
zE0@G0C|O~`w}QUrU?6ZH(=i?bn7IdMFFFX4h0|4q225o>$lzC){RQgTTC>?(+scC2
zNsDnN+skliY9c&nkze>aIBBCrkqg~H2zE8eXvrR&eL-rz+u>Y&4LK(FB!W(-zE+f=
zj9060g&Ii|Gz}%x8_Rue&G;*EZ(O=8`+N00CQWnu%@o5-86Wj^=>Q*&+F|DyM-Z(m
zSw-J*p=}~rT4&!TZaQtdcX=7deJqA+?>aO;YmJNS13VpJZLC7Cu~$m(b}3{+fs)q8
z_!L6EHiua-Q)nqE%*uky1=c{Q2WNdgTJERPUwtO`z`t&0`a$+J*8Z3)4svNEVFD?m
zI+`p6d<OaJ`vR6~>5H;J94T6&m8)Lf^CMVK|F<R$0^n(OD48&b9Raw^M2CI31tIxR
zF0X!Z73rH@U={`ir25#IVY{<pirHU%ka4PmjFFZi2L%QoG5YO{?Q7fC{_>wS%7srL
zAEBDsnp@zh8nwiCIh^U=alK^|irn`$@!afMM`Vbb(B6?eAg*T>O@u@zM`Kv}NwA{K
zhs?kgD-YeWB<NfbLFqk;>eWzBG7)it6^B^xrw-*^-~Q-UjeLJf@mE@4@!WY!trMrW
zB~%oR3QhXMrk*qu{`KvaMb>;6&naF7mS1>$b*aRU4Dn1#3PpU3hw*Bmu+P=UOEZM&
zz>+S0d39_;gJOkFlotsULr%Tz?mPBnWi1*@S}sp+d>WONM?<zUR2LPLAv4LiJ3(i+
zf0MR~&a(M_a{Tku7(O^-82EbYV^la4j`@~EXxElqtFYDZ$pO6mpW)kB?^UBb-2T6E
zCeILo<u%A4gp^m-3<Ym;d?qHqdg2cuwV<$$z5uii79ECGL{+swk9`9JJL_|CY~{ef
zpg3b6J{BYp`s8dx00=lpfteXKM5ftE;sRxDBX1(PuMsIs#Uv*suU-PzoE|*L?FcrC
z=V*Vi4l>K~3yFfFMeS?8ZEq5o5oyQ}%P&8hMK|ML<3`A_LrwRS*p6)1xxyX7bXqUx
z*n`VNjJW5`84X~IM7|@Ur)vwbgO})*ks4=w9e6()=S~EB4=FOMF(#U@fwiteO_K}q
z<Bh#v8#~R2?v4WeD8ac9)R0}{o*nSgP*|Z9Sr=9j|7q>8K4D~mOpv_yc4IMD@FpLN
z8SgwNMJRdTk`4{f*;b%o@UK3MQ;sANs3`EiExWTf(`NnHhgMAOX~`VYG~#sTo!+nN
zVhfx9jiOmuA5hPoiR9Zc+ZccYBtoD^<wgx$UBP$1-$;PS{I+q(IEcXEGY92G!{u=n
z;EW?|-3S^Bn*d0pRO6Bx9|XPrtmlYG=@-7SmyHQN!`Gtz|DdG^eKRHiI*ip5P4SH5
z;i}b>!Hn6E1?<nrq}IAsccvL?er;@dHPX(#v(oIZ&exFN9UcGE5GzM3v#G#t;U=8O
z0{8p1`Q}Us(`yl{{w{h4_Y-=j^>~fuMi6TUB{Yo)9hE<W$kY0=JoGoB_C4b?^&n0V
zQ;gr@c>&5J_hi?B$?{YZIo}rl+b!%uNiC<vBE^J@Wvxfe&yp}utwlJe#<oG@^OA{4
zWP!3^l%lCA7v)%zJa2S9p&%(ISE4()1<@->!0-jt=tPqki^~p1PCaBlL~jb(_Okd5
z@~oz$>kS4RyqU%7*lpksEPZc_cN{K?n-hPV=epUBEbp0qp)GIjnWEHu@<}24hY8Ta
z#maC}eVGSaStFP{f&+A%xsyae1j3v<#6Oh=nOPzM_Cbm&tBu>hwxFN0xmF%GD84)K
zBa;^^Wi-Z4@ufdf{m*30-#j^x07H0}e6EO0MjuIhfM_o$BGzC4_h*U?-NP4oRD%5&
zqy2&3Qb;AyYOnQs=aG?kYwG2Ks8e<*E!@Ld+)z<E<r4;jqE5Q~w7>k&ru$Q%udsJZ
zT?0qJAwP+^tG`V8?aYv3YLE}MihO}Emh!+iFXH<Xfzu}><*p((_pv*Q`koH=;gwDh
zY&Kb<g;hlBbF>IcRlU(4`+Kryc36WwfmE`+sG>sQBzdd+$^K|o>f;=w@E9F4j&qA0
zsr0X&woO#|cv15wBK4cV{Kmv<Ui1bj^G;YFK@>B?xXqoq;t5g5*rpMAFFdWdc^V|N
zsVR49nAqPXE83O$F}i75fy<+Dw=N7?#{e2`b$&UCQR;?FWxiYh>7P+hDB>Uhqi|-$
zb%6%JIP)B8coWo{yBu9#ifRId`pWmR)3>DZ(^<?G@qsog&wfy-l2H3K0M-M}4F|S%
z*q|7LX#seycUGK{O~q6x?2T3mv<oidc8^4TB78GN6r^a_BP2acO!c-oM&v?R$49z4
z8@>5LhV%k=vY++Ci;6OB)N<PkN6xL5sHEf+G}T7oFiAb}*TI5cEzitUm`-==!B?x;
z<Sy8LVP=TydMq8>QkLd&e-u@-n0Gak)jNt&18Z>%tEdcJMa3wl^n=P4O}++?uyy<`
z(IVb6)z->Q^BScK`B5aLAS9#QK21=7l%5=$UCxr{9~B*ExYKRz>z^3`q?4p!PswP)
zbK_&%B4%fLbLh_&x$yf432=x5IOe)HW%2d6Z$!%a3Ik^jV+%{y%I+N}nxm4w?4P=T
zCr2+>=T20h=h*K~@5xadfML7Ojf)sCWS)<<!^ZtbU5G<ZK!ZfY9s)R}tao*T%m1JN
zWWun9{a7{_4{Jh*@od_)KnLULAIu2=1~7gm<H4%q^twU>K)J<?3p-WnF1z*0iV2>M
z;MDHIX?7>g_>Y}%{f8dEYj{<*nnk|t43Cpz3!;-3Qbja<duDz94U-9jW!<h{+#86y
z)wCQi3`PA1X$nsp3QLLhy~SSo)J5S;I{Mb2vQpX9#+(HIoLWIai;OKBj!9O0dYx9z
z_ciLnnCL>d`25gx<9G|Lm^xHNo3L~f$0z+4N-6#UO1HT#zTEE)yb53XE@bqAF{eT>
zx}CMP7Y_yIZwa_l&9Gb~@rbgd7%!HjGaQ4hCj70*JFtF;$Pm0X{M2DbX0|RAf5A;B
zYNX+_0@Y>>pxx&qEKxCg{yr){xU0hrQ~Wvt!3uH!c^^5bF<`Os3<k-pfajLO{XdAF
zu)jh1zMsQaWK(iJ6A0jejU`B|`yfCg8Az8j_9hwjY9L^!3a4~X0G$4a3{atkPaod2
zH70@g2g3p&`CEI9$s>h=SyU8s>_)}Y=uDu4*S}N{F2i4ss6%ESe_TgLtBGakHQ&>@
z!<6t^yB>kCla~nF-CIO7hi>4v{~ms{gSV&E8ISr;!gK%w;?=O!$#QNIs<Bw^OX8%D
z7<@3U5^L}wqe2*_*Sl12Fs`IZxA%JX*!zdEztu!Za=Nx`;UsX3=cHLvH|+Vv#C2aP
z{X19v5Rj!;an|cgOe8W^UPPL8nJkF&lL_+E5i=_PvZ~L(qasGV<Yz^KG3xid9u|=$
z){=Wg2;LLszeomwwOgz@@YC;=1XEkc&3gp#E=xuuacMII)3Q42Hm1heJptEwP=T9p
zhpiLJ2yUresp<QYV8+9LG#;ecT3HX$Q0JHGD&NVepCf!v3x6hqCN6dD(SygZOE={|
z;&53$(`g-$y16jjS6G`V;9GVs@$}){6rNO9>%8N!Od8A2YC?*85O)v*=zCaq+=%i8
z2hUoVyY*4pc2k0_dWYAuzaPy;js%3^0{NkN&UBu17^rtvO~u@K1&(XHNDPq?-u{16
zIlS=v<+p}s@+xhmKk?r+8vwv5w3`S%Pu3@pGntvQqid%Ea(GUg4M1RT^CbpCl}cAU
z;c8PukphbChoKrBx5?GAiFI@>j0~`v405XMp`iW9HwI(KT9nmOM2db>Ik>>GYGvl5
zcQ-^eZlL)jM$0nz?1zi^*V2|YxoP0ubb4&7SU6cgmKB?|tENi(%4P~xTvqoPAXDXO
zNs);JOUlHG^nuMRiYlY81{Jm^wZn8Enir|y8RmOceK?U^SRnsAV!5{>o+J2O9#D_2
zNXRh$yX$TmTx4m0`|ffs6dguq>psx|q!JQrwNscD<R4-Xic7vWmML0PG!^6jhE2$}
z^&rctCbM#s$M?a+(ed#*6H<yeKc#ok>=`wD8QpI@<8?Ll>!!tB@$S!L`=BEI2mAy8
zSl8%Uk`BX~QWT)rrVHnE363qTo1h-$q0)S-5tOt%PmTXp?Px+7YLN*81j0SOI=fY0
z;eLv<YBUU8qGLN18e>BAI3?W~A)xSuu_aQn3dkA!E(pB_q{0$#BGb~trN^+_`W4<M
z`SX&xrw(T0V&vy~-n^s3&Er$}2f-^ye%f*Tup*xHll5C}U}!OS-5+{|f+s3b|Ka1N
zX%pS0zS~NmhBo)z6uAM03s{kWbFOaw;vvY(s1FlFWinxM8Yqj9|8Jo4Uhx&hBLltz
z3w(tpIXv_=3j`|f@oz!N-2VMs?KuL+fKT%Po?^}<-&XyXg8)zuUgm1FO6jg{{G;n8
zX1W6byyrV&8S9Vx<vWr9G0dx4il>DJk6F@N$0YAUZuR$ov#0i{D*+-I$|G5+5UGpi
zd@JEvdZdu3n>&Hu21c0E(^H@0VB5F{Wqu79R>ei|^Cw$}atjYikRf>V-|oE2E2QZb
zpR8WG<Xr!)defG|yn^6-PK9x?h`lrTL`asG_EV7K5HbO9Ol+@GcM15!ta1Ce=M{<)
z&CJP}@+l3e)w0K{DoSIYQP{T}&qLWb)cuB%SqAX(w1Optn%53dOsg0sb5z1z5>BQE
zxsr<Ni!3^Q-h!dx>8c;DPAdg_H^vJ)#3dz)K^AIMgqklAykm0lQt98w`Rj;Xu@ZOS
zpTeJf9A#6$z$#^4`_N2Z>w@Hs$sZ?dE-?7E;6ugc0BJq$R-ne{^()dV?chmYG7L_$
zWBUCxc#oGm^fo-PvSNNT<rg5}Gk5In&QE-dnf#%oEA=6sadc}xDg09Q?L@7_^6leQ
z#;-NW#yv^l+r%p?#33<z4r!!^h|{bil<?v26Ms3rHN8kP%e~KDw%LYBYHQTK3YgY^
zpv9=Z6Rg|^kH$EVTU^5viySBpNKB)$9uvYHpK|N~5XQO|k{&YPD6ure)0Xg6i0J4?
z4XaPUk7kRpqZcL4(b7vhiRPM07zxiI80i0iKi7Gb3g*cRfjP1Kt&`=#3DF~yD0VA@
z@C&K}@<m-d<)qYNsjhw&`sIgb&OOGiz}8ytGL4tX+|hV+K7vEZYnIICHZD8&%^F;>
zb*8iy7#sfr);7v%rtvu?Q#04(2+<8ZmH?SU+8tg68i_I1uUd7??Ft~}&W>Y1f5Myy
zTkAquYv%&|BHoXMt&8fbO5!D5zMZkxd6R!|e*#sGUZ{^!0Y|9N7HED)ufq)@(doN~
z6{ZU}i=s;@{a??_yD?BMrcI&a;pE>ZcULXw2fG`QArJn~sw3yUpCX96Q)Q_Q`dx<i
z@S$gb7SteC0QY&r_=Ctqmndr-8pW(~`coIZS@ievkynDrnKyNZiD+WaZ`wqkCq~C9
z?|(AY>Z(fM<TVY=x+9{y&$CUcB#n`p-;`a&mMK`2vvc-RY~|{woyY1BQ&wTjoPP9l
znzSjxFx9lJBk-cnoQ-qWKz;28oCwpSK~wb%lrqLHjtKE{IF%vx15ly++x>P=%h46j
zh+zpV_a(|br5t3X`s%=Yh>a!NP4R}nasiWdN`Ae__I$YHQ|#o<ce^6;M~Oh4bu@A}
zGff0-PvlHDjbRFj!00tMTJ#Drs&`yC=XZg%Gh(m=G&#Z`g^==HrpYu^ud}aobZ&3I
zB5h-T{$n!`vyo|G1~2L6+XBt|tyUNJBfmiAV2aS&O|<9Sf~@!$vip6Snt`D^>FWGP
z$2nyHXc8GyiX9Lsq#|SY1QYapY&S1hHUkYgi9iT3wUo(#1BouH^qoBLhP$~_^hyt8
zo$I9Zut2_wt6wfyzPQ9iDGdYw4u?4@hJ@BlXn>yxeyOevxJ5I!4f*po!3&C(TT;B9
zB}&C`s#Z{;H84UvhVM!z+(On#6h0Y|h}1VX(5i=h^aCub$5+)N7$xY7`^n;wl6vmA
zgJ?piyKZEItz+YVLh4SuqeD-=`(W&!38hR6{1G*GYAfIn+(SN`wdDO<p>o1t*hwCm
zN6Qs&CL<+o>D?1=6ZW%}Ry*!oHeVgxlR|`2z<M5P5B~$<Q|bLC6V%^0!Sho}FMe~q
z-Xs_lBYK5<-;m8C<rmhf8CQTSAq{^4J6I7I3(@aEThlE~z58mrfPO<U%47O?Dr^AL
zqhCG?C;wH)n^pI>MA{GL4=P8WQDWfjpzQ$~h^(MO?;w((PRCr*BzKya00(r83-DIR
z-~dnDCV5Yu4uUr-x6F1V$Qm+_RuTZneEKT@Awo9y?-vM3oc%vr$TqA$gp_(vqI4(7
zmHObmsGQ{xjU*+n#=AtY&h4y@R>jszR)VbYy_t!@hotYpPBh(legwq!#ef;d(7wIG
z`k$Oo)anw6ljSwvd0=1(-17C3e|6xHOlEeA^?a6opuwj9%>6DYrrE_oK#+Bad}w3%
z&$gH*+E~Ho+qAX4koFobVQU(=WMM>~rsI^Kr~EwY<KPLbiY&$G9lg|p2NpN6`if)b
zdl64_uG%m1dE>dk16=8!L!RQ7#e3LKSt+~N;+vsjHXMjzXXVgaaY9w+suMD%pCYv=
z+OuO&<SwAJRVU>*j4c*4SBS<WG_4OSquk_+JPZK$gq@jbIg-5tgy`SomZSvtfLeC>
z@`jKf(G&nl#s|Z7MSI=a{GtKO!TC=gZ4}@s`!?STGVr42QU=6d0UJbi8DZ-$r=7>E
z>Cuc9ij(s>1XN<o1e9Q3L;q=+eGm7JDgRMBSP<mX|KfPW)@(S}ViatnGrN);S$gUM
z9atBs)}uuhJ(mH=MQ`evoN7?|aD)8v;~Sb<cHUgs#6x;ju+ylVPqKW<)=(fl8VW{+
zjd_0VEsRSyzRZBF$^+!EP;N<SEaSc@d=dGt_tc`dYQ26=;Xnzk-u#@XzRC}*GM%ce
z2YM+oiy90|M^DgmrWnEU{vMj0E{m-VofTPBoOha&{ebD@Mo`@ai6T+Jfk6`h>qu;i
z+&SBs8;m>H=|(R5k5rcSI&wYLs*Pg$HKHjnoNDOv+yOtA#P;u5+IzY(G+0kHI2y1l
z>xV5tT8O@$B0#_nN);8pxG~Zu0#30Hy@o~4uF^IBaFBmfk5_CZVv0zB+FVNHTUfx4
zJR<^BQpr&M<%F01@dJmomLsiU+^dvmWYW}g^NivjZ6BC^v$wVDzf^Ax2+8Bxo=#HS
zkX>w+X2Z-i|FGloadOj|V?F3Ve1%_unlghqvVGPmpnPe(M%|qJZvt7IFa@sA-U>O0
zq7!Wya5QkwxN);=B`bh7!eg@a+$5eT+L)_aBs-*y^Q%7@?0W#SgTyCnZHBN7c0jG<
zjm#!^I%6ga=bF*FtJi8^oD?T;=-TJu^8XeWR3II{?w4Sj%Veju#{muQn{|IQzXYb;
zT7Jvs%2}d<=UGHt02E+t#r<gQfuBObw6QNN397AJ##x>J_sIgh&9FN^Xd0ls9~Yb^
zdQSQO6a~s8lEk?{&7`)MoBAYEJktR}_I^O+@1+Q}76q6{_i{>WcdzO?6#0yRpM`$&
zOEA!5iFG_l4(OwrjA2^<B4mR~w$Gh$C-7kY$I(!C4&2fR$f6cZQ$mRABrhLV$A2T}
z7E#b!xPpAd>rw@ocPq4OME_n+L5PyNVOf17f^AEY58S{-ceN$3C~AUY3)JtTXNNm%
z-lr?`ZjwO~uA&ME#}&t0mcZJG*vjag!%p)#ROXdQ2PD#=)0u4q=noq@<8huSW7mDG
zS69mT0rxR*QqnKlalyliLE+#Ub5#o*j}XJIKS2O`%@<eRVEVqp1m4;_iE(<8aiTv;
z7mZ|W^$zcpPL0p*Mv2+bs2GN2l8(F|<Il!_y*e|g@j;r5P`WA{Qn?ut-at0)@9gA%
z^h%yK;RX{?w3<K#jAy@{pt;$%8TL%Ef|zMdgGIG^kQ;2KZ(?@qR$-*whDe)kX4a5M
zE-(gaix{5GK`$8<wR7Ur5Lo`hn&Kcs)o{&RIG{KoHv1*A22nd;hu}pkzTIKMG}Lfx
zo;oy;EAji+_T#_a`L?i@-|cv*zB*QQY7+~xoeCGj?+oKNX}85U&`K9++fOK+3QN(t
zkutsm5xfTcy!hOw)bAOw3QpS88P&(XAFtrp13B@zvDibdL+YWw1M=spOa<Z%&Pe%x
zKJK!i0lj^Spi6l#QSOUMnA2F-G6@s_>l?*lIKFwO#~nh?mfSQj_RYRG#q34R7Bt*A
z#E5*=+~kEkZ*Ec6)7j)quF3a%jRw(hq2wj2A-N4Y|DmO-+(TRTlU=i&3%`<n(I!?+
zliQHFS^Klrr53HFmbfw5Xbm4b4MR^p^WVqbP^>q!lm3L`f$<!D&?|2Joui5BIqc!0
zht3!{pzls*P#AZ$rdup%84i~J$H;^1Kl4;x$a2B75|7t$4C2ea+#UG?qWm4d1%yJ}
zzyjPWaBX!Ul%LqCQygTexVMcU;gDG1y-56++qmY!ezcOxdX%hY4wg|=DCDs~jF4Xc
zeVDCYqgljx9{)4CFBL!=+j#Io@0Ruv5>~eOi}&$lXJjSW1*ml4qP!>e2L4IQr~RWf
znaYqu92=dzuq}~s`oCI$a~IyZ#z%|>%-F~AtMb;5D=(xx+%r{O2s}9AwIaa(?}!x=
z*y3WI;mj}TAC@m6&k(fC)=GjUho=D#DitoqZ__Uf5#IssQEzX$wcf2$Sb}0q%#O-v
zUf$owR(zoAfbmz&42$KKV`E#!15=>Ie@DaF1fzaN1>Lhe-h!a5GY89v&Ss_bZW3|O
zAxuW6QJP4|Yd^jlYMlyHLDDUJ8TmY`{G4GA1pQ<e3l9E{E&Ih0^dDwCzk}d^nE8ED
z$9{0|BzPX8YYt{CT#gh02V$$>n&re9-7fZ&55e*awm$T#WWam&R_)A@JUCJC_=GT2
zcs2;q^65~3T9EwcA}$%?{SjQmmi-aB#WtypKu#Ca0KY_~QaIm{*|kuBMsPJk$Zd!s
zb}UsX>YHEVkn`1EJcto2{LsehlT_9Z0jYZ|HXUw97ft{1K)P1iRK4Po&08h)eP=xR
zj>NxVFG;I2Hb^M(H7l4dk6Ku=l`u&~#<HyQOsnW<k{~JmPpG_bD27>Vy0B(6%UtuO
zvJAHoxo)da&x~EXv7a3iVU3tUw;tIq#-mX&!7^qNi1#Dm7X#>?11B_X)=vqe?(1zl
zT#)AZ(0wvoBl`i`hvMhS!M(7Q{K1b}G!-Yt=zxFM=PeWDOQ;!+Z;%?jh~H}}%+~Qt
zdxZ(XS20tS0qc_=i+Di(S+VR{GYZtodR&GHU`<4;#|*kIwK0!0vAqo6yHksGx$Rdp
zHkbhuC{4bt2a4^^gkRGJn08N(S0jUelWM-3z7v{~L`D#V{R96_lfvj0{mri`*PMjG
zMPqw&M&_#n4!d3G=qDk~?KkGmZ^7`8z@DPf;52H+g_2&aTk$yW#DguBA<<7Uy&0v9
zazu6xS&~d;Qs}G`nCFR54I2M$+f~vR#xS!N%CB~Pi9J4ckE>9CL}<D7(&-4I{YPo{
zqO(ZGAALUk4e{GC@vBhA#@EovloV&wpOqqt`&#sB+}X^N;^|f$+j$5HUSJw!_xAZm
zQtLEb3iJh~irJV0F<nyEj(%N)Iyw>bBPzu~=2t2l(J^@N{ax?Okbzvs>kL#tPi#Wy
z+YEyevfv@;GJIgF(4H>HwgPieJXS}q*Sd#ZEwWD-Zh<QXcpQFE{v#6qT}v`)aH3xm
ziA6A>Y?=nq=<vJM`{>xyKLb@#SqpiC!mq@?7etL~2+&)&2i~d#6EiqL2DNcT8|1I+
zbZY}k4YW*}|LFn_8enJ{mIkjpxg8xM=)OPER{F3YpQlVX*;^+nqgLTSyroagE69ZH
z-s_9Yc#jiL<=v>Q`YYr!Di8kaDqtnH_<0$%aSDl%(^Qn3iE{6SPy5hKJ$?5G=VS1(
zUgcRDJTc@rakIoF#;hB#?`%|H{45zDj*{`w)wt);B5~vhz4U{5p6+sb3Pu$X(p2i7
z%G+8(0CyYt*|?3LF^(m8zrwZKOgc{cl@Tl;{i+-d23&BFqO~IS@x9Uh&z{gv&h#k2
zjD;v{TZgX?Rh=hoi&DLSbc<CoUHUN~@G6|LMCJDTc%%oy^b>Y++;6s-g1Y`A^DnC0
zo;4bO76jYwN9CFzHFuYLI&p*jjl^rLv?U|6|LtV9mzZxrVZ#C!EC!lg;Cn~<crT|#
za}PqBtER|;Z2Q~Iy9J$@CWl#f@9bmMM^aFYHuRn)emO@OXsu8mP9KeFa}P(mX0e>u
ztpG4Mru&4#Ozw)Nawx-*xOKahv+%FVz$uSKJ*!i43LUCFKqW*EXEf%UGF1>vsp+=J
zuePFGHSEKl$?i0*M`};lecyo4^?p(fEwD!59!?U^;(06^-Q{n?M-$*w$PU^wsiS}`
z4xZK&A}Fs$1WI0<6pE{*e?kVj4K7-ro)qNdlgrxCYmW;66EH?h#-KLkjNHCNLxNYZ
zk<?N+>}1-)UFQQy#84cXE>TT;5my=oiy4QC(&9Wsi~%9|<Do4DjVO(TzTbI4<27;*
z26w}>H>xlTus*t3Lv)T;xke(x6&<|sJ1dUyk3vGS=SZ(`a>RBfj)_S)9V{G0w+&a~
z*2+yr*g?L>W4Nr-li<PpGYQOhsIU*hcLShjT>Py(s_tt}aK&DcDxZR+)g?^D08N-`
z)io_uMT`^8rfKm|^GXI>6bdYm-rQBvpJ11-v9+~*=wj8Ne7?yV^olyz9sBGaXr>(x
z<F8<v0N@hNyDjehk}CK33<DBouvt`tlGwn3%b7vro12O+#bk-E|0OEa5*&mes@xNv
zX!+3DLr?|5jN=O#$a-`N7g~EMK`5bY2`w@ZD5t&rJ6@I9@mH*y8VX6Vb;Rn<kKwRJ
z4rPL^c$H}WySHev3Z*#=b1}(+_b)Z2{4OQnj2Jb`*)FP8a6BFXK)1g-Z|2k}_9iU(
zf7p8Ku&SG`ZFu(HbayvOD&4J=5=sgR(kas28<9o<5osi(LzG5h3kpah2#CZcr9(oR
z_XpQ?KhO8R_xA_K!2x^E%z3VuH8X2wnasR1t95%uNx4Hk^nrMu)yI^5@TmpwfS^ER
z4sgl~7)8uZmwEp!u#aqs^o}!J@Zis{S-`^!B*kh3$cX5VT{}Xz3oX)yOp0HXh17j%
zXk=2YR67d3JrC!;GcxE{m}LG=s1OFMTdS=46j``g3j{yTh$)R0-3-w^7f`;~kHKQb
zYb2b<P?zOn#Ztznyl=@ZKq<(yOLc}1%P!R!pfOMcNL?j8CIQYK3BrcUIN0UUyX;mw
zI4H~aO?t9mZ3{W-`<YMKji7`K7wL_Gh8GpPbwf2n=1&;++2rjab?Nc&NbwPwF?vEZ
zpBt|=?x!BBOT6dA+U&#0Qaf$@A$f2=(&5u_dzthpIRZ7M_INAUQ|_omQ!AL4(dg-3
zF<xG{>IW@?&l9*dD*1O`$-Osj5yPA4ZTEc9!W2N{oGO4f{FW<{q6D~>Q+iXDmSgKs
z1A)f)_EUxR2e^kX8PRC1A3J}tquDtGUWze!aIkpCzT1Dxani-(P6q@^1+QC{GYaWj
z=JTE5+Y>rbtrhyj<IW)GOhwMV*C8?~*Q%(<F$Ck4>ba?guCMN#w~~N<-BmF~PjX=P
z@WGRkkm7|F``UvzXewJ?;>nDsd<-OBG4F<}|M577nDJ~`e`g3W^PKeQ$*T0SS}=2!
zAMFeL5uVyU{eF_LkQYCm4TL_ik}VTpY5Z`IMijB@r-Po3#nbyDl(cL6Zv3$zF2~4e
zFK~(DFJ4LT>D&L(nBpu|J8tLcd8&LX(8X9xcl0%`D97FC>UEV*yo08Fj7=uR++Vp!
zT9N8j<%KWAxP_ZvdvSSF4pYz4Q9~h1@iR6Pm1U3mSiNB75A$&u$c)ZI{m&QfpYHJA
zozd9*{EbbK4%kS5!G8Otir}L~Y4x?9?lM;Suc2i-1(uYBO{%Wp%`j$?d{@uIJJg*S
zq6W4>+t9pqF7kQ)VH!Zbj>`Va8{goXb?J>$wEt9U@lWL9gGmaWGN-g|A!6{8v>_B)
zXzoshBu352ca)jv&cn%IC1*c~&R1A<D!v{aDwOTB4!s!AUd6c+mPJ-CZuzW#lTpU*
zK5@}vjD<0@Jd!ij?#q+tB2@ATO8+62*~RcRcNzCPf)&sGpit~_cU;1N4pC;@_<b9$
zlk<h$I@2)%;V|!5weN1zvN}Bj{@mrFJ;W<90&SQ;hL)l%E}3+d;C?!t>8rwvDDwDW
ztmv2QFZPl})D;fAu&ZwO;Br4i04z54*nrpV1&-l=qEzG3Is@%rzfZ;D8KDw+e7oCU
zBuO8=M|u(=MMknf^yCwjpdZD^TrevzR`d+I5zYn`b-A&Q`MK32)9lpJ20E*sX)lEo
z1%2RN5}a1U2)W^>F<d58)&NE0v|Fn|aUp55fA;zq_UQHMYx9><1kdI(s+zz$rQ62{
zW%148v1b;e#UcK}ViJjsiK@r6x9BvSzgkMdP)ep+o$Te?^u_vp(gl`1zDvBX23fH$
zaPmL86)_+`F{aTR@phMwW1MhLG&tt$<fi^s)ilWs;~XXWy-9J*X8aot!tWR_Pzc?A
zOkyjAIipMBs~dB}z?H7bl(V2ovHGK6@Xv2;?#D%S<_VBa*AU##ZJv+Xw7sTRAK*|v
z#aN`(%E)gyYS`lWqMjQ5B=1A$C(A;QEOCs_#n&%xgZdD<mL0$XSa$CpYDQ+%sMR&d
z+7!T8ER?B%4}nWys|Vz3x79GzlA3Y}!JXefxwbH@AODW8`BAUH*xKJTimQ_?65B38
zohuV-xF)VFKI*+qW2gL<s$P8l+0e^Rhl9eQA~Xdf&SoHmq6l+co$LfV$4>2w-;H?f
zrnf;^2hU@EYu$VOoX;^5!~UVEGSgASVC+wJ9|r~I2Y)uTHMVuEZaS#Cj8i_|d<@;H
z(CXj&FkasHtU@txAl`r9?G^0s>)o-Pi^*@P58)=KMrzO#kA;K)S@AZ3Tc)48(5`<d
zP*^9>+8di}td^Wo+_xoxx0}v=pT2t*wbdB+Ek`?);p<08E9{iYoazVQ;Px$0Y0);p
zF|~A%a)Ce^zg&sC%|@PBfHPAw1pp=7FUe#G3urMoP9HE`@<3?N=fy%^*3XzT^37(Y
zvs(lEI9Fy};cV|MlW_20iyyX6X(lu|+t{jRAqTLmY#-AR*sHAl6osLxV3Y+hUEFrK
zZsQ(7N`C3@i}G2g$Yh^wCMbUJz^gJO_RVq&40`?+&cklCl}dd^J}vv@$X0puqZb9f
z##)R^6?FZ9N~b!L?32EC?Dn;0m6pJJ+jH!4x--xt8N=T%ZjDh9r^V$vC6HMX0xo=k
zTvARqv$yMjXj5RV^SUpq{Wf(4eu}ZUaAz%UXGVNyc>0OLQ;h*jlkq<g9~vCR&K%e{
zVB1I$kEh5;hUb3?L4Z&BgO3JcHn2burQM@tN=JYix8gi9=BDr-@Cs@Qqw6ba$zy0>
z9@H`nJ+hEFyctL=ZFbx<AHrk{Rj2SVZKd@RN1$Rp_W}oZCmV3+KH{Qko6iibK;Sc)
zkRzp{c^%90?ajfU82eU+^<r$qsSWCP=Q{iMhwugFgWR$0&73i+lauk6vvj3zl*o4o
zQGZOH5Eo$Ev3oeZUMGZsIbkQA4-)OSYOjUz>j!SWzGR@wfo3T3WvxgOi@ijknI^j_
z3~@I!v#?7mxKw9Q%c(n&Ku?k7%}yjNuF_oUa2A?VYz9TGDp~=uWy-qSl@w>ffZ?r^
zyMONmOo6X+52monxH~@Y3+0)TCs8)xLV<{Y*uw<JiXzP;BQ%_tX5$@q&Pex!{26LR
z=9+T3NpSwX=PJR!QOymq<FmU=SpiKH1O$(-#qSx3@Ok1|V-V@D298d+-hK#SFB|0Z
zSYe>cc&OJ#5WkC%lSgs-O{m9GpGj`DyiqfX84qzN+f%OZ!343SU)TeMyh^jWL=pqI
za0kKw*z|S(y|omZDjhAnHh6+Nu1sk|2ja(4IC(iTtgP^vCid%h0oz}>p1;00Ejhn9
z(`07)m8tsb-{E>tSxGUoH@~0><@p*7=|Gd3c#_Tz7H}FtR|IBYGLb*HrwO@-r3=84
zY<bKtLz!JtX7`G3)C&O^*2Ck>Ar7VuYBP&D^K-ew0!SY=Jc$){3XSN3%Otb84Nnl?
z!gz|*;mqSyPli|!?pn|N*~u~JZY-<qG*}Pipz`JK**f3|M8*w{1%zQiPwt;fY-BDD
z%J*^#A+9y5A3A>zB-6p2(9vD$jb0e3TKuv6q~oP+bhBK}i7`MKd|`h9O&XYOLDXyd
z;UGH?Pl#w#zCTo-(%#^{-aVds!vOi=TM<^KInSO<T4L;*I1;Jj{>=^L%CVtlEPZ0{
zKaWmaM`>5@-08SanI9FI^YLUJ;=o0l#UBX=W89*SBqvxK`tR;=eHFaG1`Q0u!LxNb
zYoB=RA9y*61C;hlCW^}hScyXR-rH~V?r_`2oX?KE@WhZG%8BlA=2sLubO}ZhrZ9O)
z<?22dgBXx564A(8vVj<&j{b2}*8xLWkKAh^Oq+D$emP(`GgCj!6~&*fn6-DkI+~Vc
z#=6Gx!(E+D^6JHonDaF68C2@6FXK^V3RCxru`_nwBCToVWXV72J6omQ-Aj9|3dM*E
z3Jj(1G9BlSerwhZYN#Lpn@quZYYc*+{Pb8-#ALTe#zaN&zI!BY8TU{IJ_=<OHhS#-
z;bCB$KKFxje3$|bY~dO%FwPr^X>U+kaG~VSqW)L-Qv5PW`_;Yk;h$jO9#PLCl%|4o
zm+H8&kieTpM2RRqU!u=&rB6jM?X9?tU^7#avY6B3nVPoy)S;S(_=0V1ucrl`;{9YW
z&`P5hIfoKnlF2U?U2p`pn<kU(qz3D!X_9(=$D`X6bdp4G_im*+9V~fw+DTLhnCwYS
z2U;e<(c|}K1@c>P;-HHKs;rDZd}wp<Z;i|UYLs9F3T`Yf^*<go#@$vI0ln@)Ge7(3
z-6Kuqy_4LesrtivrHDepN&bE;_;Owr9P;V`MS!zpars+e7eY_Q((^FD(hmqJ;b8lG
z-$>hC(t$WDISAYBAFKB;@zHZXCWteB3JLt-tk4=1Gnu))6vj^-q|%It4iS_v&k%M}
zL`O6vBN9G2zGc%7CKOPNYn?*;wb9X%N_+mKmPLD|?RM}^9rP|5*f*4f!wJeXmy&NV
zK2<S6dge3G-NDX@AIqAVL&aaXD7bcxtKPu1dL&nD!-I+X;A#F_Dkplt6CiJknyKf<
zldj^I3`rhl;bcLeEi2e9>hROCkYkxMW9_Mu3`_ny%SS6?N6pXX5DUy&{;~<Hs|vL6
zuhy}6OBQTc`qB3xu&6U;U%KlcJM*>pYzslok#3oX%HHL~cj2(rgvjT5C2zvDTwIjW
zkIb44T1}N45zc^BdE3gr>LR^{$A(jc0VCZ5$^?;Y(UG*uw@8$fm3IU}bDv>QevL%G
zV&2(d+Fe`CYN{G;ZV&&Netf>;W#i@N@OR%M*ru_xltSz=VU~DY>E_tkA55UYOZ#_B
zn;QXCIoD6cEQ;-FVy!Es=>iK&BRYOO-uN;~xP0c4kb$c{eACd%7HN~!==+5-v`rx5
zZMPs!>xmup(4-hPEL$3d<JvB3aJYwbC`a2r7>KD`JW0a0`Aaf{F543C9A1pWq>bt<
z^&Ne@Cv>#KS-)9{*xQ-;N>}eo$oxyM2ZpNf(a^BeB)IW|bSglpOPbVg0(wfFzJ%Ja
zpPvX2NC*y3Hcuq|-f+H0XgPstd-%uGL$*!rs|!<H7kx;g6^*uWH~L^mt*-OFVTuBH
zn(%|iSVseFN7VlvGiKe?j(dhq8eS2@1$gNdSi$Oke3iXC07xC5H0WS~m-i-Z83THm
zs&Od#^W<bll6)|+4;E1;n)NJb9pfbWY?#k`Q^xGDLZ=R&2=%Yq!&n66<0GDW3oo9R
z-vafwBowmZU^F(}4}6>VM`<DKmh#xI@tsdh2C@R5lRvx=zMp)7aVYDry<7finexy4
zeOJ%QoDw@UP&kaa0X>Z;a65BwOixLMEJi&5pgeHsoWjBZ>K7lX=EE})+PPNOxQ3{Q
z&ojg5oYB3)eAS)W1*(hgZQtA$X;oD5x;B<@ru?X{lYnpy{+r=I15|&}?h3Q8qG0^m
z=<xFz;Ck9fclvyg##oLLfT>gWkbD9_;BID1O!)yWWGmA4YX`S|vTpf@HQPhMPd?n!
zPS`5z=foWY<uI2T2(e#cXXI?*3mWnAhP9&DGGkH<B)#&>Bz({!kVh`PMl0|wYJKKC
z%4#|H#WYj~@TNlG?8=|FLSwUj`$?3^{Q>Lm@oBSMPD@-GU{NH_7ObBaQ0~%}H}1{m
zT)QoI2#_lze8e|?Q)2^*LHML-MMG#6!xzhVyh72|u=#t3@va4|C6-030V;wS*DI{?
zFVQwPQgM<3wzUM8<2o>i;k%t4=@{dDKy=gTU$g-~*z<&m)C-~ez{`tqI9lBRRsYf}
zd#-cCe^kNzb&vhudb0Oso65fJ!}y6J1=y^~bkDr^Py36L3Z1g{y!P4j7CB<mm$H--
zY5nRjq1t`3u;7H$1**xvCN%RV2V2$vJ2{4aE{IK~0<*|>soSX7R#s5_=FRG7LFI3l
z;~FsGZy~@bYd}HikTj$^v>nG&*;JQw8=mQW3Q($uX0AvEaN!x)>R|nz&!d)xm%Nd1
zGC@wMgFXY<feae9dQz)vg&|{{G#f;wPAxyXb`6fk(Jr4;L18G+9$$3hnENYm)IDs~
zgd^K*v=rzVA;H#36<g=4(Sqesn+R_I?=7~3BP{2evHnxCiZ^+6P<_iSGbnjUizi%O
z`=8*|2P1vM(_f@4&dr58G~EaO_DHZK-2%(F0&0i8{nRnED9RcvcCN1T%;5Zq#kuhM
zRvN7%l#FFa!?xP0Bu@8ybys7I39md;Ch0W(abS>szy?qVASAux{{-DI<(hBcA7Faz
z?G6kX+9S-F^^gHn8PB}0;w_k#yeFl`FyUR{maq*}Q@2M-PPR}<<epY+uwL^eb^irj
zQKWMs4wyCP7z<4S^%~E|#t9?(56O2b=#p_@3XZ}ZP!-_n`@Tf4g@8-YcOo5@HXdw(
z@RYV}24({9xJd>xWxVB3_wW}(obn2P%}hQ|feF7=n~c9EroHk{KT|$(630u;4fH1g
zeo?B)Jp%AJ+R6)#wVU@atAd?zTnrm%?3r1x^yGvF46Tl*)L#%A+y1GcT%*g(^n#;8
z@`=CSS2y^ogAL{!0wRn)YhQEv@)gUXAw80R9syq6YJWdT0#HRvqvZDtlK&1iJbQgx
zqI%22duz^x_@z(-!;5pCP~zE%y3rggZ!E)CN+9EAiT-m;ahMjd`y0B5?TBVa(21e^
z9k22(&>D{w1e^Puf4QuIS+=nBuw+3b9vy*TSm53y>1IOu#3;|lp!%RS9^}BYZbDVt
zp()se@9L7TGd24XbM0JxjC6MOoeKRa9Jqora;UA(HixpZwQ!%RqAU4K+V=bGq6OcP
zmms;BY)*jV^#JNL^v|lj@*l#v11C;GS$1fWWK+skDL`VX2xfnv-2KgN-1~4xb76ZQ
z2H4F-t_&6x!KW~_jHl++a;S=HK8GlzPYKJET{3q6%3sP$j;ps>^2fFew*8n4#uADb
zZ9T;S`pNF=b1KwC6Y|aQhFvLdEKJ~gY^!xCbnRwma$P23VEpleY=M>aT<l3pgz6)0
ztYb@LB+;gaE}v{<yOcjLo-a<PemwMx!2LYuS?hsUohpIagJyLWaH!ZxGrLsPrJ;3Q
z=n<~D58`K@>9>=5H^EmTcL-vQ@Fm{(%&oUty+rxfV6bQ84(`9F0z@}3iCmBL6DGc4
zA-6X#hL}X}=<*voay*cZtQ?R7EVPd>-|WU9(CpONe%%4-xA(uutKpC{OKAQHTwixh
zo0a+Peos^}n`!Hj0X5gX$Sp{5Noy5t{tCjh46X&D2O^4-(n`f}z>{=AEELUs6SBN&
z7?OzRU~fU<Qx6`63jK-sJO70xng+i+^~nPBpE?u~JuljRz(rwFkk01*Wzxkcy>Ik9
zR^7mr`w0NY(KbGq{ixKn-HrOB`bPbOt<#)^fyG>T1`C=P{>E=lWM_<UXTmuA51=}E
zQraHAG9&=|jU4W2h@AdAuNoLjhO{zICQ9J0gs<C24=>6Ft3aU8%KIWrIUE3v)tc~n
zC$DnmdWCyixdO@WbTOH;@p=4MGPeTy?QIT+u-*h4aGZr0Xpq4MRrE?-GK66+&+46y
z!VdfRD*}GvrDK4wJ{km?w!v;1iYClH_`BS`RwvD6N$Vl}P_Ua2NnWlbi)f$hP(^qM
z94yqm&T33ukDj(kAs<XS&!tHsYd63G*>9J}q!eBxeIJ+>^%@zc(@|wk&g;Rv(S#co
zBobl#uE6Jeh2f*%j}0nxkC`qokO*~mq4PV*JUaeA0m*$Ti#uEr(MqO}2n!mCJ0CNH
zL42`Ol}R#e)YCC?HF*OAj|5@cu&Crrg>dE^Y%(vqiW-~3biqphQphJ^9|}f+_?O@2
zhz2oGLu_dB;8({B($3e~1^$I*mdhKECb){uxlV|5o_w6xRi-a(Nz3rW^rF&VE4QMi
zW7*5uO8k9?1%JoMmo*tr;jih=#u^HR4JWz3xx?QR0EOTLD~H96;p<;fl=X)^4oa8?
z2^aas5>CF%6G2$g8ELG>EX3m9t7pY|-P}Y;ad39M0@HR>8}$+<X0VWl&K84B1dz-5
zY?#3oB_K6(2t!)ow360+fo&JB;aVddU_ZZtft2(V@8Ojy`Oin#8V&9tyOP1}p19#D
zTPya@qvVIqUMqn~wp@7h_CpSEeodzE(;wR=%rrS83{Z0^JxcF4@7T@$mOw#~A6I`h
z(Hu)~B=m-(xD)-x!LUqwuWA@mZfZMPE^2N5FV^yUak_nlo>jOcEYr)?($FH-gbE;y
zGOrogpn~Xo`pW2^?72!-7;mpI-Jdm;O|3QAU7CV+zFU~Qs9afG%<P>ym{)X^OKJ(h
zXVwp|B_yK1l-Exhz9S)@L1BZ~Y1)ieLs0Bz9BAfiKDF6B8Bi?T7Z+rFuX;<8W*vv|
zh4DO<ZOUB|DucfbjYnTY&fUpi+Zd`Jk1FlqXoXqv&k52&72b}PDgSz!?KdYCPfgFS
z8FxdHkoP~@jA^33XkxQWwWcw*g-RkF825Y)a2anNs5&J6g^l*RKh?B^H*E<isJ*hF
zPoEGw2|zGh{MfRxb>RQ}IZ89NzwY{0V^7%c&A_a-p3j-%zEX$7GI<POo$U?Ga)cq6
zq+db*-OFkii^_ok+$x$)kc3fHi4S8ZZ27ecl#0Xx?ke5V3j`?A%58dAbq4VEQ+N^9
ze4C7JvgFzWm%wE5jT7DH#8Pw(xtbaF*KZvPU0WwDq2h#KYDpf1jlmjc3vfXyo0kNt
zo%sn?`881F*g*rf8aS8EQO+AsEHKm1K}Z7nZg-b;D#PO%g3rQ6h2~hz6A)SLSnaUE
zDP8g5$DjQPZbcy_JR~HZ4fVi*z~J4F*2h!C)zfXRj;VJO^3V%zvu8qWx`*Bd-#rpo
z5AE1BiGa^_=$gPS@|NC1_Qxf67#$D_lVg_SPo>pD4eBfhnPy?vGn}DFb~!W9Q-ci*
zQg^$xHxG-_1vZdJ65X8#L3|S(qBzMCZ`}*TF(2CO92>t;0yGw9&uN-OV3q0FGQmRi
z^Nd@Hj{gyz0R5O~YTME0p?m^leuW9bR=lmLzqI5t^M@+8B`1Dy#<|LfW+7d{1vPJi
z#5qh<^1T$lk$MGScnQhnQ=-)xCsdZmm(ayfG1e+Cu7`WkVx=QAtF6@d?yzz=+H$zH
z7J-BcXYUmH$EUU?mE6gEdI>2e{|hw7V_9;_u#y8OSot6v3d>~-^rtuQmX7YbPrYKb
zbdm;#@MNjH!eZX--Jcc4uW}~%8^Gb0y-)#IWmUbu4{i2ehctxOKlWvo#Y;!TN4D{>
z8yv!ne-fIw<=ni3{ilES?%zqrR8gYfWa-=Srim^Io995>Z*8eTdTI*Wk0t-N^_OMq
zR{9X&)3M_xPHpng_q`I6pk(|IRkl+Dj+gv|uFs3#D&7U2jjBZ80HdA&go*_xUAYda
zDJ<mjs9>;#ONNYbnrFC%9edWT06>F2*Uy;)DO+iI^#e)uyLg;ll;Osn#*V+uNpOrq
zT~Z;((HUjV$O!F3-i8SbH2q2d;~W>37nsWeKR-na_Hw`!px6-{PEcm!7yeyMhT^do
z-`UpCR?2dD_7_3a&7KUMUphtwdZ)};@eAhYZcdx;;gr(vDX7IM$&y&@eFWfX@V5h#
zM|{%1ZjH)$J^3S7%ZRdXate|<|6(c=HrC*_6x~7Z@@oxbQz@kO@-zv||Mf0r+xMmi
z7TLJFoHmhcf)shR<|Noh1ak-w-2E;7%3XMbpG{6O&NiyhqIj1$u#DYg%3In*QS~Dz
zN|!x`0eKE3>_4hi&*5fKskM4=v(Q+&erCJ(0n?%c#s3Mrf;<S*=5V@qpoNlgUiBBL
z!;d^5mer{#)3x7b^mMM-d|FX{t9e*&KJHYTyUb}2!7VGe<;aqkj0x5G5H_ij&3)c7
zA#{CH6cl3r=)>aul|1Ml(wv1ro_)bkkQV1{*Gp(bVIcv2$E(UQJ6}-(1W{B{uf?5i
z$m>lB->A%hF3#2${YrbswBputGZD8(o5s0)MS%*9hpa%~J#(uN#}Bk&>}BpF!$X>C
zc0R;K>I+$o@SnatddE?L`%k=qXRuAtw07M7h72kIBVAq9a&Je)7H34FZug!8uYUWs
z7IytiE46oQbNlKNX5~kphOj!+xLiA|Qq8N@YrdVtO(@j>8!7y3)y9aB^HD(;Nr2k1
zBoQDYnU`PY-}M|HtAs#IVxdfTkayTA8I<Rk7{YHE+>pdsM1YRX2L69Q(6yz#3mRWc
z`hj{F7TZ&vDHJmZqcA_>|2^*8xKc+9yK|ial=d0^A_MYj;uDV2pG_j{MydV@^bLP^
z*FKMN7T~e?1_u!X8zQs1<pgo8e*J_<hqBqlllHR+0%)g(v&_lldA6|#O$golUoIeJ
zCv8K4g67s8mBq$4Gg+9UeeJ&+zf%W4^%^me1QK^9&*wet&C1wYd33DBfYN>K5TlR~
z9oNe9EjJwX-n)muh5FF;y>Ju)?J(xiPh&7$8xEXP)9?E(J#6VveQ50973RL)gbG$h
z5@8yqYJsO>Th!OUv^Zx!5yk#le8InyP2~?!GZr`PX6-hB3aIlsYk4L}PvJM6<cp8}
z%bFpIv}Ah+2N`7U8JZ>Vsen-#n5DXavgFM2%wIHP#Of-$_f8l*r+RDMf_)bNGWhQr
zsPK+^6IRUj;|mt*$S0eu?in>HacAc@!UMAeV&J_yPrR0*1ki304zK*m9CTU|5_!+<
z2V{E0C643nt5<L_{r>dvTg0jW@&_&^>L7D*vIPSr;f`s+s6At;s4jK2F(L@%{wB^V
z$}#eu%{C#-c2f>Ck7NI;bU&18dYnfTx=^;@W3!r^{r%R*RU65AwlA%#FZxM}sHZ^v
z54s7a;%(+v33=_SHf%nBs?+J<DcmNQ$QK{0ntx5GZrx4o2_s*X@tg`+@+)wMAbES8
zs?b&?1{R*)K)?puu#kq9VGW6)?AG(&%iLdV(8%gZ%p3yLj+yEo0%~j&88ygqgQ3+p
z6Bi{QRQ&Zmc@flLUxb~`(xyx_AA#)dOReTkdX%}-M=(K8bogPYX01mz{jfjy&{3Z0
zwQKI%kz*u|m}WV(7&}+@mPidB<p*>J<6$Om-Y;YF56e=I(?5MSALV#yoafy$5zFbU
zZJMc;-zYfx>O;%s=58}qY>jho_jCT!vjiB@NyE#ha!llQoTiH^uDU0~4d**dI)}tj
ztfA0dh;1?q>%6A+Q@>d-&L(XnJ%KnT`o?<B2lhw6;bhXSu>=IN9d@XQxx`^uas|F0
z3&~*e9_wZ3O9S?=ifvhDa{9+W0G{DPJv#>uv)(qLCId%Bxr37uxz>HR1dJ}+mm+;i
zD+KOQW5^L;E^{Gk@)6*C>s!u~cILzV?PrtrgxhQ_V$}O*O7{p8EQ%jhw$>a8Jn(xs
z{;e^N(tNe^_+6~2bk%HsPR)y-mHH}+3m=6t{;svRB-9TD$yEa+k%&=w=&+z&-0ei;
z)H(*)vRI>mtz7KqFl2B=YS+|F0<*Z!4sEbXO|RMm3<evy*Ey%3;KtclsN@Hm{Bh8W
zO#My(NtyPjI9iXB9=P#%*>+JVTfC0^>Wh=oFD?hnypC)UJx5>t>zPrmx{O<CiT0~E
zQhZngQ_97Ca(#?x4h1N(<q>7PjDY5saB=jEpiO{q=u2wy+w)?DJV!*WyQZ<&%1ZJ+
z%7rpSomLLRBL#Z;)z6ZK9Gx_jZ`8l=7SeXCh&8NycJ<45nH`q#UlU)FM76||cm@Iv
zUYa(=7z(dTB#U#MOv#Cb>{GGdwxkWe;nXf>dz4BA1T<c3(4L2@tbcp^w_H&2Y}qa!
zvGNu;;UiF1IBKe~RoREb5c<bnwPMT>hM`-_^KJ=rU>8B|D*+DW!FhtXL1IoZ6hus!
zAry{PHsBSDUQ92j_`*7oP<|!?Oq({5Ty1Lpef1KzLUgK+f%qNnO8Zk9$J~S}e^S&;
zH31W$mdS0<RfOXXX!!F8(1McKE)06)JNqip$3!oW(wq{P0EO7R47STVclxkd%A5yT
zzd^@G|E~_FOV_Ac6YtJ!Dwo}KKC_PAN%nZ?S=9~IB5#j%qb%8%#~$g+>lQGKKuj_{
z5a-_ww(6_SYwyK=?E)fMsSN6wzl0OG_WIAMk9bxMCQ=T)LLe<+o3Q0iRRo&lddelq
zgy6_{54)>6(kIr1tqi~5Z$2?Noq2Nnv*Zg5sV+s8{wdMeU8paU#+5&xxXQdq-+-nt
z2BF#x|8l`e{|XPO4(=g*6yRpWse7t2ZBn6*JXgyr2?UzIUvrm4du?Wc8k~)5>_~kU
z#S$qhZ^|=n^P%1)U!&CH(|9md<~l=OA@k19W$97ev(+WRMZ9P1Fs!g2eH7=xlfoNZ
zji1f7>7e3gF;6NPIRv-9DO;>@AbWmzR*ff8^gv+deH^ft4W<?a=?kLug6V;f)A!Aa
zS>N{^(awHMDbvYv0Nqy39kO^j2TkRSqf%kq6|4=PD<gh9<1Cc7p6v?bjUB#T@Vi9k
zH29-nF|Z@hqBkpo?!b?-53A<3gk}Z~JCa?Bh5#vbPs?-q`_cSa`VpoAJ%?p{pVdlE
zNau%NgqC9TLOaJCV}$Cv+Qk!Zpt2i=G4FYO>(&?^Ul?`3Uy7A!$ala7IhtN$pc-Sh
z=1B%k>{UsD(V>GcQZv~KuOp)Taj(By)OaH)`faM?qemE8{AbI@`lF+`r&W@`NAu#{
z>5YoM;#*>%f?V~<9JvCz0SD>$>pTRYraVhex7v86IO-f<0Yj_JdyAlQdP3-{Sl^b8
z!!aHnkobfHZieooAv<2Aow={Rxuk86$i4|$X=zx1z^^}bWUTfMrv>(jAiI**7G&U>
zO%?4(_a=f4NN$=b8U7jVQYSOVxuo;=;kvYzghs-nsNlN#yOKF)vM4#p6zX7-Y~@2*
zRUH28-Zr~z|9r-;Ssq~%r_2lF$eI}32YeJpV+i9MzH@GrulW`;RL4WOv-kFXM9$_%
z*x=puyyOAZ5Zjj+^(;wk#=?adO(!Jsa<>Moc*0}{pSsf{{IN6IF20EdMz{<aBW(Y|
zgA_}xH>%R72?FJt3~oAtuIl(^48T-5ck#3st40MYB7pFps81O5V^=E7M2U-dxD4<D
zx=PA9PuT8toS8p*fnr&$a@}cx)qUgJG^H2ie=$Ts5RgzP{TzWm27CVY^IL!w-W)fi
z@b{i{^k)~5-YwQg3p5VYwnTiSMvq}BjfciV__m{zE?wfDFAtHHHVGZcvyC%Y3f>=D
z#b=JMZTD)oi8PXo;`&er5^$`gzBW;C?{y?tXLRY_U5TeOQRv?c1GdX_1V1i>zygjb
zHS{R{fzYQ$R{IabPyf}n9){rU2W|52tb?Fk`!fvm3$gPcqfsR#-=lNkhsxentkbWI
z#%#Yg>OG-s;lq}=Hh=f?qbi}ZaT=u9Z2mccfW+n88EB02fOmtdHL2pww*qw`+!Oav
zK8}{-i1E4mcKgD=s`q|v-kJ%jaoqi_y}NDgZL-xp@F?!$m$n!<j$}XCF6T!KFyg!H
z9Pf|q{9W^N#4}&d68F%kFilr8*NgLA+7v;K!fZ9?;`bRFDI<^>Q|h39REW#b=6oi6
zfatXtSU{K<VJKMN9Pish86ACe;#UA#zhZ^Ks`RFp0lX=89@`gF!J9+&j*XD_(?d^1
z+D{H)zrXmNt3L0ALXy1H7c%&mKl&sV$v#?wO`_8DF`vW<@r`91VV@TkWIoJ8Y^HTy
zuFJ-JPOJEXck-O%0$I>8j@zch{@sHkYWnXwCO|x?n&91lZMM1J%By?~k<_|TEqBcb
z;wc!ENBE6bzT7{bp3K}ZsrbE2hztyfaB2CHgt`3o^vS4@of+5j=l9_1Ef@4qS&c+3
z-I=QeQa`X;EnA)4uy!Rp)D40O#DB~AdiUu&cNb;)(XY%IK9~+g{=SBuRgwbq7DbRh
zle89Ug8VIEDZNJYN7IQR?7neKJh0CH=pEzq26tt^FW&(_>ygI02$V_x<ozNZoq=%>
zKg)fNlR6sT%<@cd{$1F1%N`6Qt73PO{E0yT%UnKD@6tvSGKyL|6)7v%;t<GzwZa*5
zpCuUr8Qzjw0@!vWFQKy7RRrjYrNHZN-S>(d_J$GjNAjXCBMLF+$>N?NBJO3uUbYZK
zUwALk;47TU;}<nosZ;Ip64H}GPkcVfAu<j@15GG2`t}#qdF*{OLwKO;POMiq<<so%
zH6P?r`jQBRn+D_W1oY7iW!w#7vn)tiUid&Ji-oI2g~@E-3Ig!G9PCz%1BQtSLbcO(
zFPPbHJTARcLBgB3|Jo2}Q0n^&!qI%Z+)v*8XQ-;YjSujqPDV`#1UTP=lz~U)N;#e{
z=zeMY$8uN9W^dd~ZPH};rm_-Lpf9ffDg74t9;Kc*e`BZ;_Nou^1X}mJPyjMk9ei2<
zEilBk;cwl`$Due&fWE&fi9of0M`P~FtJcsI@#Z&c){k1J=4(3M<^_DDv2EB$0|jCH
z>rk;3iwl~d(N6&TkchR^SX(Bg=J`FvC`^=wc8x7NaKfZv{`#*PEZbO*`w+hm#p@4^
zI(mUMkw{o7QoCn8=jAq`=dr)i)>0<)Ra$2B)eNUoP;FVmLCw>#zUMZN91M){O44z=
zS;#nn49Np==i7e2uc;V)l@i4SPNmxIQ~DNm_Oy7mEHhoS;NYLIquB<5c6D9Qa$wiw
zk1@>fq36pTI|AE0c1zi}fSJ})<)Y|^Ab5{44J(U_Nuxk~NXG)Uomw;@g@dGr6)1nW
zI(rZXenvgQJ6Pr6uD?w<TT)EH4To=O2Tyt{C|k<})4MKe_RA07ma)(HXM9-N=rMF}
z-m%boxfVXtWx{N-sx({pw!vxo*<Sx$D?w_~ISLRzH15bee)pjMU<2HQE*T<>Lv*2K
zu-`8%B*CKBwe@P5!^vWcM!H?|Xgei2pfMofyH1U*Ajzeu-$d*ZzeN(HZ$J?z<Um*x
zPKJ+kR{Fsd1D4TWCi=VoCAj`a^x(coE5U(S?V>#Ab_w<`k*D$xYQ<p4nG|{uF6ghY
z%OR^`tKBYHk^XJ-1}sDh7oAhtnELr58M}ay(GgMucy`6ffp@fQnP8;z0s+8=+j;s>
z7XN(;G)PqaYzW@;Iiu+BSVSdAFoS8GxYd+7Fs?<YKuUC~hzt#k5O@~C8?AbFqv;VD
zK1ldXNTdivhBNd3UY$vH(1O2Ul3z@~_~i3AT3wvo5Cig56xK8iiOO>sy`G`ox#ypB
zNeMNTB6Z<t5tGT6NyjCD2GS{41NqkV4suU6zktA)jqRyvD+grg60ta{zY!?u(?g+h
z{unU+khXN$5~&yvA;8+qOKgYGATKWd0$1Q*nP7&YJUW*H{<RvgRa!wtpW?GHfB~S*
zu*vwKk#b$267i!?IzGso$42#Uo56K(j#UBkV~|YXWeeYakpV-K7_2@Q(0|ub2@`ni
zy1B+i#d!~j^Vc6D0@I+aUyBu#Nb`;eFOwq`AtDmDMf;r;kv-J}bhwas<?89-X!xS8
zLFLujI0J?<dgS;a#C+!dHE7dMY@uUY#}hC?E^n`FdnPlQE3al-9~UY7`G@p;S5&W~
zFZR}y8y0c`{PMti@Q&&y@KJT$q;E5)ls?{d9fNWN)V{A}kpNL3)GK0`hHy#QBEd@V
zy8g`Rx$iU4Bx4X67V$&O9b^#22ezvHGoQrjYDTP$vW&WtfyzQU8j}(}lqIwj4>a4T
z&WlS|H%I~!>Ed_%vCY}2IcC1zjTZ(C1%$}0r=aP*gm*X&M-*bG?Q9wwNqnn7=>a~h
zvXl(C!_F6^tMO!_vgu=ZAKm|kiJHxIGH?eB*d$DUMMAHTm%#=J72eI=MA+562B?}q
zGwiSWZ8?%+Wlrb#EA8arJI@<DAmr|ozJxX*H8;#@6bd2gT@3zI3+P97_3H32|MLFg
z!nT!?f6`10-XfQ(LjlSO@U{&c!_h7?IhyFGIoEu{)Uv__INDgduGI}NP{nsW>bx?M
zQQ$+AP27^j>)e0?=;nZrOqo;V@B#6CnvTS!Re|fEU$`^+xg6->UU5S5K^~Aad(2t?
zT~HltJAZD{TY0qW$gFYZ1PEYzI17-1PrRH2OE88N>FVY*ThhNt519N3uTuk|i!CxD
z<Re&1&p&$#AFxoteUcsI3-!V<1;s*`4DADPiOCJ-EE3*#N(fzecGj;b3g--Ktyq6a
zTu97$#(5g%b7P8x_#lSzcR!i>=->zA`pa&qMKtlih#pqy2_Bm3;=`0oOnrLfAcdbA
zA6Srz^JM^{-!F0=U1>SgDSWo+Mb)#myO|(6rxhkc`9OS2@p${Wxs2t(ni?=(=6k^_
zx|p%&f^>qA4pr=ZgF0;wF=Ipfe{I-;Dx}DH&jxKFP{}r2*J?<BKR2vW%ctD8rq!0z
zTbDHL_N9QArYfD}@98^YNl3o>SH6{P#<f+S7-~3x-7*2vwrv$F!tpyoSKfh-NDGd(
zKw7Rd?zOG6)^1|gv)TeX#@AAE3a5TP=cYDlASn*5YSH({`HJN2qy~~lH!)DtAX?qW
zRbhRs9fEs7IAO0zMgZp7%6_F5mn7NX#FbsGHY^E|jz7@+r`Dw|@axdkvKvU-_EjuK
zlhgTOwgIFeLpBB~A8`JZmUuU9^i&qO==OukjZn=H&f~A|BM94?$&6!$g&^a7^PTdy
z&g4GKET@6sHq^UZx66;2dO`bD!vVHUn+F`2gx8_-1%`g+3=?(zGqX{n(SPYieuzs+
zGdAa&xds7$iu?@$mt`G}(0wZ&sF(!m8PFd7*i+kIZ^(uxZl(%@Wi^1xz<zL+*jgIS
zyDWxmIPb<rz7KV~hJgpGeCsxZB12l9b*}nQxr&RcihI5`12ii9$f)9yBii;Ro_|7K
zLS&x#Z=Run$WPbd$nbL_0l6|Kxp3r;ee5d^m7lM{A!mKtqxp;{@*nm|f+@I&!027T
zl_~Ld^IN(BX3E-osKO%|1v9r64)QE1=GB{LJ|6qaPa&98Y$8yiJs~m#+%GJVJNWXw
zzwOF=rhx>$+3JHPd&d|47=X|fG|@Z|UvAh9wpZ;;ueb&2)GU!j_*x}aWV!aKb~g0h
zHybSbr*UcORabP?Vr)=Xh}W|v!8ms=+!G)@!oSp`C@ho~fuoM7Vk$QKKVG5<hWbQr
z7ZTF+yowbvncqMI+BT1U`z9;nkFvXsAud=tFjXx~3$a);M;b*Vz~1|@gIPf@(J@nK
zlF#$%u)#oubX&K}!!pqFUSXo<d?Vkx^9yp65NKGhi^2X!(Vwr&uhg{(I>_xce2Gb7
zf9VjUh|&3#jh{vD+mpXenNaQ};Du?;1pbc3cfDY*RYB}Q2XV2FAvY@;vu^*V2ODWx
zEiA?kh;){C55LOlJFW5{v-+oh{7YUZ*u0}J>+sm_=fq|HQ3pMcg^|NVuv20$xyu~5
zcxdjq%Z;I-MMvHj1-lOD;39S2gQaej<8%mt>^oqb9EQICIqU98`Xzg4`2~q)Vg<mE
z+iOsH1LfE(s=#miK8bSWfnpRE*gD28C^;lw3QYKqVn{0spaRXQ&0`?j#|I|RLI5)a
zq*&b};DWj2OPAk`4jyH+Si63nZE_7TAi<ph7m_l?-hBf(aLVM;$#eq)m7@mAfW+6}
zb*TX^6$jq*kKvF;P=n+Al1(fhk$<ec35LAeWd2HY>kb$k3<J85#WmY}m(dd#PN6|)
z!_hp*dk6hb&L4V5l{?lMU#KSlEqeHv81~v7_p>BoPzugV{=#sO@-Qwc9v7j}9p+EC
zXdB!2FCsPH6)GW+b>}9<B_Uuy`QueQCJs0(=jGb`PT3&~V?!F;UbETCo_(3z*9lpC
z+M}UhtrVs+r-lZEXH_VWz=IdognE{(Xs;tc;T)|WGZIJ`uXuKc=!^xIwbt0&E-J^-
z%z?TwqeBrEmoS~ya{LFBfco*_2D|V!DG>1xM3v@vx?3P)*lg&?<1p2Li2F>ZMYo*0
zU9At0AbKA!@;047>YH)!A_D39hNth@bqo+3J5ClfN{7@v>kJ2bo(B^RzjY|oM8Gsy
zepRwcu<dp2YQM(^fu_errn<#JGExU7#O3M17@o=RQ!aa{u9UgZHu~H>VXOp3KJ#AN
z2$BxHCk*K~ta76;y9)6f999WgUk}ce=nH;YUt@Q#sQDk#(2Pi;VxImUVzz&RE_$Hy
z`^;yp!YTX-Cq$bLawW%ecqq1PVMrtq53*QC&68Xn-$fY~@fhf+g9O+<p_2SXz>UU6
z){=k=Y%S8D(E~-nKk)`}U~+E#F=EBmN+%b#&7|;4k1}}s^u|A&^}>*S9Ym)0AvLfD
z&f%kd*(`C%2AVuvut7SsiS2tjl%iWH#1}ElCCetkZ%HVe5sngEeVrsPmsqQVwues4
zn~8wI+JuG(kXwpVD=rBK`2Qkc1pI$S&%E8!A)10gFHU(nXXx$HRwpJsc=k=y9ozR(
zsP@ao>oLk`2Iv}ljgg-efrOY^(4%=x<ZQo5=Z$B-uF*l!Ng8A$`e%>-A-^7Bh9z<C
z2ASd+IeKHT(N`L!I+b$Os4U8|f4ZX9-)jdTXKz;z^9s4cHi=+piN1`lCi;XeL;HFz
zHx=loUO|@E0n^lf47MHq(<s{Ag-VMIi0tueWRpG+pTRgR%6lK?^4tKPjX(+aPj8A5
zMPnm7aF9f4%5>@YF5WnR5T3hYUo<%d^K#=a(&ZY0N=myZm#p5fCdwlCCt$?}9rX0(
zPUJx3i*w`*VR?F=DEt*~v%S=JE*NUd*wo|T0PbV#L`e+#jx^5{rV559uShWbTg`J}
zm@Nb9iLx`DK11$L8rh?xAFR*U(r<RPdF2hv@Z#<+i9t@5!0ds~RIH4x%ESk#s&6hC
z81sFwu#VV%J;}oaCg3u;kH0*+WH!=p1cB@a8=m-7j3B+U+WEU}llQdc%C^@FDS}5u
zXSCX|ZDCYcafw}?)8&b8Jpx3e{!W23o-hg4^fqv+$&uq&2cFG~O7EYB9U!%K0NF3?
zGw4DI>18g!Aa}ZeAY&h9y)w1}HPXQYpNbw7JH^&mmAya7^?MJq+@P=e=-S9kMhSXg
z{hIoWxSy1`)GuRNSwMtk=;Q>Eug8oIU0+*bM7GXCVX3Z{)SzegEw9KobC)T)G#hM>
zubM~gAMa8@y~vMpTAxm&He{rXirKV3piN5b>24W0jV}U*gf1#2t7cOfeL+_SJB~mT
z6j{rjLVjU{1f93OW>e7cMhzEOhJCsyB@{5Zlt44A;|cYOIUnk#6p)CLTBW)n1gJjP
z_(<M#);jnU_!Pc&Q{{QbN7l5r5Q0TeSWt2hCa>}z1WVNcKPI36A`*d4gb+NA7jAz7
zeD*kzxK|oZiN@Vv3+&rBfGX+~#@Zh1S%Fw<w-r;<rQY+&ZzkmLV^S*kb@jkgn1rN$
zO?oq(2t6qn5wj!)$<Xxngg`+0q%qZ+udas-sKT1A)j_JrdZ1Jt?i7O$8k2~?@rTqo
z&u0z<c3QuD=R9$TANNyH0)5b#U*d(=Wl?B-sWW#$*Lyq@aRYcj8YD|>7orv$T%={M
zvEx?^RN3mMwUgY*f6zA~j0=q_yMSH$Qp{LtneBxC{P@Nd?2i8J-y;RtKi2*URWXpK
zC1}rK3^S(#&)pW<dz|Wk!B#qqiWqd<Z(X5oz|IzJfpkqAc#J~n*FeNvK5|U*#s+=x
zP-NdOLy=+P-BR7+B-$B*;R+YSRaK;J-s=}n-22k>iz6`;Eri_yH&gobX=h?5`j#GI
zVx!E8YJ~hj0C17J9633x;9EHxoP}Q2PQ{E%3OHfZo79;lzy^`!B=7N4oEO{P6s}h7
zOe>zwBs%<f9;$;6>+e3}OgvAfsG<H=g-3)e5f6&o_BbF#VIHPk2dkj#8%hU;vJ-|Y
z4#8Z=6%3}HDPy629FU_3v*x$ztThWx`um{5oMC~w?njHMC-*td_@jUsFC55W)#I2}
zVW5hEb5}VT3^l(5pUg#MV=IVXHlg5skmQrOLc3)u{`$kG2g|`~3zEoCGv@#}SB%!{
zB}mNnL@d-3CaITu>FGOtE{=w@A6V)d_<Gn*anWo6HwXHcR)BN&&FQ=41K$hs|LBwA
zyQMEar1A}i9<&{M|I*Ig{Xp!d@6$h=NT)_j@vxr+aq;o^fWWHi%i6UiJdBoH!nKcF
z>Gz`uknLvP0dPCI94_dTIuXR<JMtXt4!m6Lvl|#^0EzDBkN<F=W4L&JXbshZY}ozJ
zt=Z-$PNT4{GG|0;Gh65Qc+?9mE9G5oA@K4UT`_TA!^sV0$b_$fh1>d+sb98q#-P(K
z<3(O}s>zIIO{gjEmI`S2l0bNYN)_FB!|_q~_X2h&_rN#G90=pk_iGO*`wQ|tOCL+i
z0;j>7^@6=NK29Hvp(Tt%O;~H0!^k-8X<F{axjoe5Iehy>njx7qnhc^07aZwFjr0|Q
z<aviT?qDBGCmLNB!WNiXzZ)%wJxHG&;q=*}6TbQ_T%XS9ScvpZ;KYdY(G1^u507pd
zGZ`)5gx&CG%=Ukt3%NQ?xI&OVo2M`}n)y-__^=KYoRV<Bv~V+j3=vZ9ItJ4E=EGr=
z7e8y}a-Z~a{3&get9nO2#3ck);ip~9lTI(SUW4q~O9NAT=Oiay#RKVQ4@~Zr5ft9v
zOlA<IH-*}10V``ygp0<3zysOw0`rp(f2481X$KxOd&Azy5DN(0)`lY|IAAQTJPD0Q
zaOAAR8)zYbdfK4R1}Z7A!5(Bp9uUz9Fc&`594FX}wE2%&L%p_$FqT(A8P+IUHz;(h
zUD*UD)`CR|Pk}Lxz=b*TB=;(+AOn9&$%@mD0O7W%IuOA)G<Lf5GgFM{n;_5wpMxGG
zBVtRj1$fdfqwN2T1uvWR(&B;<UHTARgz1|^mGTOvQ_sF(0mL{yV2b<y4kV};YvZzU
zjfs5%cS#aR(T1mB(|CjT>&&1-5S-^(G5`DNd?zO73-T5#$P*|2(6~%C-6V1WsGE(L
ziG;`kGLZPcqt+=+FM_0g^$8Y*;LFZYKghs)<*LgN6H-8f{0HI~NS8%CPzNIzyqWS8
zR3`!-&i@5c)5r>*^BRD9*_eDy=-V}9fLZ0gzkLC#*Ss96OEU&)((Y;=zw~n>tKop8
zf&s1|on-_4GZ6#T_W%CV7-mL-3cg}dTpa{3_g${NAS*LXdNf4>+LCwy0T#_WBY~fw
zUVrc6<Z1N(5>f61<&Ae!f^a}EdO7q{?2`+J)|YnwBjEphU=z^fUbR3D+qTiWoZPOI
z_lc(>z%i%*F1A##mk@VQj(AlbG#_z6%ZfjxI;He*g?cwkWDf9NrP2@sDlDKS>dAjM
z4yP%(Y@8+x9Lkpbj|TjQI2tSgjZ0>>Y2aoK|37_;qo%F|SLkzLD*W-kzFv=wX1Wra
zI|GPNfvn~K5!+w*Wa(w^DP1s^+gAS#&6LW%AA-is0Fsse``hiX@U=^5CZRXrg$4hk
z?EeG(1hzSL$!ybCm;n2K4~_;VaQh0IDRPxG{ZecHFTu%xZKTs@D+F2*DW#uqwQq6h
zfr@_+!dQ$YF;Jmbz)ihF(s4izPt85(E+J(46E@%AXnmxC=cQqFaNfVc4xX2cNYTQP
zZ!UHHIzF(R5ZQQ!iw!Q$FpyeFk2n><AQ{wml?N<73<vyQe@g@;d~gM&OI~n*AJ3Ig
zVgQLY2FS+&c%)!Y31~$FEjv>1nhkI%fkASZ0ykVC2lCI+&})z9uu5Nz6OO<R0~<Li
zkK09RR~p3zlqfDghP>^Itx!U5d6RbSlicu!Z1Oh!E0v;#vE<?aHBvy)Ee!=3a@_M{
z{ZD-gEN^i>FTJve%OU;K<7;Pd!=i6n{4!kf--eo?_<zT^NDnMiHNKA|YZ5|3cy{|A
zgM}``K;wf(krRN(W=`ie;j9n$k?DMPPpjy^d#E<w{ko<xDtu@-h>Q1+St)}^E;v#L
zqU8@El4BivM=HXT=(7F6TFumI)~mD{mvW=4DYalwtmnv+`NQ&r#kjHlZLzV4pOa*M
zcd6-sk%8^^61^wZ9pddZJT*|(7UPG@Z_w5(cQaN#+@5Jr+)pJ=*Z#1kgQ&gT#Amh<
z8i$$5Qq8T*MyWho_ah7oRr^O{j>0Gjn!n36%i|0`;)r>py({iQIUR}@zalZmT>F)8
zG}>6pJyi)5@LqXxf^`AqZr<X=Aq97ob(%o@LlKO36ofGD*ITUFo9nzgQi)Sc0XLR0
za=LG(ei^Fdm`|U5^fubig}oJvo$9dxLHcF(4>be@3J07b)BB<a)urTsfS##nRY|SV
zdaO6f#`<vZk&Pq2V7gJ>ugM2?>H0&CPwZLJp;vl8Bz(SN_|BN(=QSJbr%t?W`&ty~
zL(-_54>6c`xKsd3f<?s#jinW)E|W`X!J{Zlp^;vD+-F<9UEu#XuY>*>!~pa{o&g3R
zyK5x-vr^4c-Bb+jU7&d;wvl2Rp%e`r#$S6?`=-W9(_hqPEP7?5<Eh1j-~dq+h4*Ds
zf#rUcn(YUciyzg=L6+q-BN_J+8;HMiP3N{U)cu`z&!eFi_!92mQo@i*YzbsMa1f=$
zc3ltZFG@+R2x^=x>b2Hatbf-s0VDA$KbJQ+S}}k2&QKfvllDbLHL_$|_jvf-T2~Tg
zP1K`_qF`3Xm$n5#+!=k7(G@2d(9aFW4$Oa7D$cM6S_kH>nB!0jKL1&=U{)VtSQ1K$
zm>K+^X0AOP%CwD7!`O+PD2H(>JI;|qtanyL7<M~YF>7g(nn6i3qpVX{mSW_zC~I@*
zaw<6$hDFpE%E)0tF=KM3W@s`($a&uhyX|*<e|-Dj_g>fY*Lywp;rDy)-*vy&b6>B%
zPWJcREPfgZ4x<Lu*P&$I2Xbxw26&Es1D-@<eaQI6bP}KU;@6@C_p~0VL%{4j?6Z%P
zeYcpUZHny$hCqMQ0Vlck{LZ{~p)pQ$Xa(*<$OwrGn&M$Wb7W&p$jPVaC~7J==wntI
zF!<JAqeAg&^Z7LzU6kc=C^6w2M*O_5x>BM+MuywfA}Gf?)QN=!BCR;2CqQ2xY~{A8
zwCjSYtm^`<FMj*xd<|ey8o;Q$(3J~%-IKAn3=O2RpYl~%Rd^u8c=}7y2mBPWU6Wnw
zbpOmvM2-beo$5OIPDT*OwcE^67}j|6`B?UL*cFdU1VA*Axe^*-N#@0{HA;SLAcYwn
zw^r*LR5q58d#1^j0(c@EDRGZkEmj+6Z~s*L?eW@<LU6r0Zwu+9Fp;IkIB+<Z`YEq^
z`Zu=HF7_E2DsJpc2HDIBQVIcxgmN;O6x&C%7XK$S-E5|iuS3^C&PWVV$oH2!e<kkV
zX9l`BVcz$)FC#zZ?#ytu&%l(D>0)@fMJvwAA4l)$IbpPun1%7>l=})q!@HLOU!7Wk
zuYRqdf~o+1nB#Sy@}VsAsj3pnaCsWfpzKZIl3Q3&IIvueFX1DjDsVwRTq{Ksqhy|O
z^GZ11^~EC+%a?}as?2>gKC^S0e2q`pOeL(IzD&m5oPbq%$!DC=A(oZlfh<CYLVz%k
zsl&%(*mYem?_{s{DaU-4&43S!LC-1foAqwHC?sbE%BL6wuwxrvr36PJ^ir4_b0OL6
zED+C?gS?Sdy9hGZ8f3-BZ-6KySd5Ar_1~qGwh6`<@?;mqALrqXWN?=xxvdG3xN0j-
zW|MNLOWFGULfMj$d8@wm7?6v*;2H$^FibW3>dY5Z=UMOKs7c95YFfQPS-G^517T8z
z!4+qD)b-wqS<r|oH@%l<ugj`&Z&nfa@Ez;+nhDg$4@fv=36I)C8P^4lFdA-!lgr1p
z+HVukuLn{;)xiQRzea^2?(Q2RG5@evp23iN7kDf!u@Sa(TaR<k&0SsT$KjzgwJ!gx
z<KTMCyai6+Z#3`Jor~ymCZeUxN1+j)`gd5x;@8CW-w34L&DM}-jlC&GN>&jh^Nv$t
z1o@0CJt8e15BTbW<m?4Wu2Mk$b^EJl;b>ilx<caeB^+7R(eQ(Et@#dJ(dGH<Zcpdu
z$Bb|CeyQw{5_hkd@nNJ_8@MwM>ut!yY4F&U2xWq?aWxy@zHJ`|(LgjladsHZ^@797
zTS}7(&C*RC`grKjZNt{Pz(be9<cOrBC-w)EEy;h?HPILlYMVNi@Gmq@sFu6-=KN3M
z7@m7UgYMq6wnH78W5W*Zw}xoPmQdc$HfPB#7bhKofui6ey)9dts%Z5~Z+-9CC~KWt
z<F$!k-(u4>|E3K+qH7|Gy<xqvv#0=-bVIvVFs2H%aF898(sA~<lIJ!y-<jxIZ#sik
zO`b8>d|7E39~A)2S#C-ucnVx*U)#mKS?(d!?w?M1n|SB<lI<^BOF7a3rnCWUzv~V|
z2imj|U2<kM$kX{+`aLs(Jby{KqR(>fj2?XomF&0aJ@$oSBQLsj@?zlig&>T4%K}uX
zQNsUqVO`h-Q6Lg!@@9E8bq@-?XJ2=zYI5UNvcShjsIZ|gvM0uo!IOx~S_0?jh=VFW
zHOIBRIL?a98?JFssXwJMd_=7b_3le%e&`h<q9DNV^em&qy9eJfb9F^?*HN&t$27r|
zUtfZ|<}9^+to?AZ7h%AAYrmTNM})>A+3xXNO&5H)j!uDG^9huyhXq5>2SnPR^G=|W
zE4Kpq89|uP@5Ioi0lJ<T5P6#>uHQ1{CwD!8y=@Rqf;T&k3RJ@nxG_j_^thnq-Rf8%
zmv8Q^3X!p)QQk|@JGckG!<dOEH9k~0#;P2|!41vIP<~Q%{05TCkVt$-^NFvLsdAfa
z&X`0hWmnDFh5Fn%Q2E@Rs#*EFQk=0#<XS3hTAj3FfryYncKNHIcS2%S&thtzr(d<u
zDV>KgE$O2UoK7#eT9X(^J828}pAjQ|ftYY|uaualwwLR8x73llT3eV9;m!@X!3DLi
zBby6RNu!{WZeT@(S2sqm6c<`sb=2x8hL(aMThMLb-)@H!%MIGl*pPI=<s}56rgtD%
z9dz2ig;yq}Op-=Ayn8+riLNp#sCxP=7v<O!`9t-H)(7ZC_??=#2g@}v5PKOMVgx7v
zCW_BdrU@nmU2jLco~)~~pNykL$Ie1{c8jBnZ3L}<ZXguN$`HRLzeGML27e`kTJZ~?
z!0eA#gzgn=VeW)6q$tyflb$~A{!=&SUsZT3`Jx_$VGNz-zBwlu{y6K@(Z91<Lw=Y=
z$~LWut798%*sD9@{-aT`$ee36Gk3V5hJ}KVHdsThSF9n-dvxYEOGZNTzxFc!rq2J-
e1pR+)>{%Cy!4J7H3?#H5;B(Z*&ia8B{_3Cn%H@Ut

diff --git a/16-bayes-mcmc_files/figure-html/unnamed-chunk-24-1.png b/16-bayes-mcmc_files/figure-html/unnamed-chunk-24-1.png
index bf67e24fcc92e1dce890d69697dadc38c8aa0478..8a5ea2d5083736a82298ed7eb6af0d8f42f1f982 100644
GIT binary patch
literal 10083
zcmeHtcT|(zvbPCH5ov;;A|N85(xih1B27S=iZrQ;1Sta25t4w2yh?io=>j4O3P`U~
z5~Ya{g7i>>^cF(z$xR68d(JxRp6`C^-us>JuKNdTJ!H?G{o6CMXXg3s=lKmi4aOs<
zjvP94h*3*Z?be|~RLDbzfLH0LNWXZZiS?v^I&SFPRww-n004kMAQcrAH8u6&!-r{T
zXlQ9^>FDU_>FJLgIl{oeaP;U=Mn*;^CMITPW)>C}R#w(y$BwbFv9YtWA3uJagM)*U
zlk>!h6DLoeJay_67Z=y*)2F$)xp{ec&zw2K$H#Z}>{)(({&VNf2?z)X3JMAd37tQG
z{=$U|7cX8E78Vu}5fK#?6%!K^7Z<;D>5_zmgruaTw6wI0jEuazyn=!P2n148R8&$@
z0)xR<uU=JFR#sC}Q&(5l(9qD-)YQ__x_<q-wzjs8j*hObuAZKrzP|pA8#iv=ym{-^
zEdv9C+qZ8U8X6iI85tWJo0yoGnws9ZbH~ig%-r1E!otGR($dPx%G%oc?%lih?%lJo
zvAKW$zOAh-1Ol<MvwQI1frEpCqobpflheb851pNzU0hsTU0okNdgSKj=I-w9;o;%w
z>FMR=<?Ze5<KyG&>-+feV?RGXe}8`{6bggE;Ba_AKtNz%U{Fxd)2B~^gM*(vdlnKB
z^2Z;4Jb(WD#fulAp`l@6Vd3H75fKrQk&%D?`RB`*FQcNOqNAf<y?PZB6B8R78y6QB
zA0MBPkdT;|n3R<C`t|GN<m8l;l+@JJH*emgrKP=n`!+p2{oT8F@87@A$jHdd%>3}-
z!^e*wv$C?Xv$Jz@az1_fl$)EImzS5HpI=Z=P*_-4R8&-4TwGF8Qd(MCR#sMCUS3gA
zQCV48RaI48U0qXCgFqmVNaU9<U%r0*T3cINS65eGU*FKs(Ae1c?c2Aerl#iR=9ZQg
z6bjYa+KNV_+uGWG{P@w{-rmvC(b?JA)z#JA-QCmE)7#sN!C?CP`v(RF1_uX+hK7cR
zhp|}f&!0a>M@PrT#>U6TCnhE)Cnu+-rlzN-XJ%$*XJ_Z;=H}<;aX8$<!U7(TUtC;V
zT3T9OUS3&QSzTRSTU%RSU*Fi+*xcOQ+S=OQ-rm{S+1=eG5C}vf@o=lyD^gg~yK0)c
zA3DSxLjD6pjxX9CI&|icmYVWypV#w)yEf)s&^L?PQ7VOL4>jB8aTC?LbV)eps0{J`
zLh->BI?T1%6I^*gm$3NQ`WyX=4Sx#CNOxAri!&PDQi*C80Q7L3xA<Tnmrt(?EcoIp
z{M6W80K*Y0z?HTwG}rR2yE^eC<D?&YH$}>BVsi&MrKz(1+8rHqt?+5|(W}K#ZES~b
z@BpY-`K`~~qduw@77+Lc>m?QmAkEop$g6d9C;m9`w}Y#<kc}A`5Jhu{Lqn~4MrH=c
zd|Uv`CF|K)YiPMrbl0E}WIhqU`-~*VX2128U<fhb6|x2CDw>bSr!s0YPgx0Ek(?Ph
z+t|Kxm^m&%{=sf8wTL=Wg<9kq@&%pA9|84rCaeGk*8CFy;j`9!Kn{Lub!rhc<UQ(~
zKLUd3Ou_<=v*xn^LRs_A0y)lFO8|}3{tGAL0C^_22})#yKSIIQp@qU^s5#FA-jkuq
zu;#Oqp$Y<H$xyr1khjQC6X{G|kfE{xw8>B%&stv~L)E41^Z)ha4{)lKly>hkx<adY
zZs(F+cjyFwx%VeNF=d9tD7k%kDb3x1_}*FfAtdy)UPlm*n7g(AVj1cwy^nM~y#SC&
z(bgk8V-gMuj{SL|m_<LW{xkkUG4p?DYX)(L&EFaT#UcaJHU8BsHZu|s(=Pbm?99XT
zz4{r{JB4)ekN*RPd@P6+%C#z+dm_%0!lg4P2`V00OOT~P>_y*W0jytaylrj6QMomd
z>dtt>2B+5#PWoIU+q;t=09KOUK*bdnh)%Gr{tztUL#0^Sm6jX@@smS!=4C*hL|Hd)
zL<%?`+`N`>d{w^;&Bz2Amd1yZ1OI~t(9$-=0YorV9AkBd+T{P3b2_y_QOyP5m4$p>
z{3=#dL3rnDY~!l2Yrp&?&}DC5)vT_i13OiVK9#0PwvH_+4c2j{!_wyy{}PVpnN(@#
zieqbGvLgQ|4rO)$`X2-25)dE2I>~mes6xJFwb2;n1W(<dw?B$Cf|=x3wnj`l#EW8&
zY!n0*Ir}*;y=)FQO|T7uz(->u6Gt)t0Jck6M(RjRgg^$QiE}0)<qA=2eJMlv2P?CK
zej4+vm91p*J_=5MCzIU^#Ua_JPp!JX<aDq?`<1r@F4hlsuULdefuFB#h;o6wX3^eB
zkr0e6#AVd5vPDLhYIs91mVLPH*vw0pd&zjl<!O=rdtP!h-ih<1F^YH2VKZTpWiPU4
zDk9>OP6dfh;4=1Yq<9R^iQ=4L21{kz6*@(lS`ZCJRJG%{oewO`4<OFvd>O>dMH(gL
z^=QnZJf$!z*W7KIky7L|l*U6ZUs)R;GkIcp8ncB>FobNzG`9q^iBh2t0%OiEsdclh
zVX%?bMlR1(5+sR%U^&UT>IRy6AZ9;#8CcO&olh!Z*-K#9v;X__Kfwk!h%}(YwGVB{
zVi#8$t~Pdl&G$d-Ul&AW|Ja0hF?u@f=OvHK&_dH@1|``!&BO@b^6zOknK^ya)||f9
z+A@WS6G}5Cb^bo}o8K?upwzd2|H(-x6-23)rP2B+lmgk`B&1I%sPkjdLX@=YwbaB$
zDVtX|$LJ|#HraoPQdxWTlzb_R{Y$G$1cXH7Uf;~x?-G*z4G-YgTs(kpb^O2@dHd}`
zf0OZkdoe$jO*+aZz?zX~U$F^XH9zHf_33)`$dg|PVBc-bR5vAYPma<9C{u3I?1Yc8
zH*Wf<Aib6hNe7u4K<wahj2a7H*1K}S*X(_ff$FajxRCwo$Ln^P&jKi`&;em??z&M<
ztrpcq?M>5s)>p{5&p!#6wB#C36=TicL(i^_<g~cmb=&{!{&;M_sM@|xHBw4iz@ZE!
zg*~aAgvk3+XTwvuSm7T|y_k3(w!e-EkT7X*3}|`A!^7}9d7}l?Rf1Zvgn1iqV$T*2
zoT$-k?1GV7U+NUCRpoL{wU@$&{+g=9e{yca1?B%e-u+n%61uAV_@hL27)JhwB>q6Z
zCNM=&p`vmWv3gu;CmMrEo|r3^?Y%Gi4%0E)Qkvt7J-KhKm()8dQT}Be2<hs3c+A$Q
z(XyOrDLUr^{4><q6}w5!#QVEAhW}v@jrQJ(!^xyNfg%*wZpVhl&vFHYL+m<`a=_AP
zvggjy%X8_qPD!I3TXr|CqcL4RRq2Fc-!H!e?$P`FtZqt(6eWm%k^%egG$)k}^!e~q
zH_X{rxsfX%sjMD2oG@X9;P)M0V1m!piJz0H_N}NGbuy4zSugqWr|;Xv?;$OpNu_Gz
zLyfd<my*}iRMzbQ$r46kOSQ!s_Pn1XXD#w$?UMqM;C$uNsGtTXpOtS0sZ!c0Im~Nv
zmYW|=I-g?iGUh32`(b!t#qw)CDFO7+`a48t$camE%E4}pc1-k%T=6DE<Hkt^m1=a$
zf%x)7$5wAzdiJfFEQxAH;Sb4|nT<~_yTA14thK#%#k_7t3g2vU7iSN`$v76xNfwn!
z^xEdypW0~XAqU5zO#TfKD=I`-bIUqcBs%E}#wiptcw)f5)Q*9+W<Rt@{m8KYjubw9
z)x|LUFXfQpg!*H6Hr>Rb9Rqia_G%;7EWEF>F((%+3NH4*jQsW$NvzqQ^K#{PC&gFX
z+kHq3x{KylZR`g9iV{oujVAXNq;Jr~oJ$uiI!CfO&m1B~>5V97M+C4SKN1RGSsObJ
z@B1ucld9-~(I{2)^LJfN6sYBoen<oReI^cf%tntj>lH2jI!#oq5Tf={bE7$AlyN`c
zBxqhkMYVVy#fO0pBI_i+R%_{1?fspD<Mutk`&^!9e5l60MEwbTC?cr}eOHX<_i9$D
z-i@W}LRtY<t>QtyV#adA<}^amsH@K7K+*=83ByksqO=b}zJPki{EN0B>N@v6qyp;G
zLi_Pceo>*M5!S;X_kMNaCO2_ny*t9Ey#DwBx|D#0%cFI21w$q2iCWYA`zatXpsWkA
zyHg{eirCaH;lyP<xYI}1`?9s0H9y`RxosI!Mc8W4i5$Q`8)MWrmJ)qPJ*R0mOJ~xe
zfc0!@8dh3d3K9~EE7xzL1$w<J@}w4-IE|lfQs1<;SaR;6-&96Mc$7v2)W4>JJ#=$a
z+PL1nBPZo?;kPW}JdHmWLQfOPw5QX1iT$(IRB2Tk<}#2Y8=GX33G(?lzVJHAd`?M#
z_ct&wSk!kE6^>|hb=b~3hro|IWV{a#;GTBk2831+&oo+SMOER-lvdJ)VHWejTo)5#
zW%t)K`Vx5yC%;=4c`JpdnfV@TzHss<e(Qqku!Iq9N3?&lS*!WBIxHbg%)`ubimYyN
z_Gt{mU@YflGY!%kta(M-L*F#=Mvat_;zL_t=WH;{bxItP6(7ZXU>ikjqrQIBr&Ugv
zlol*=b7384bTFiVoxX#}9JGW_M-C^RTL>K^I%@o~QOyM{S#FoR6N%?G&5uX7{~jPD
zk0Hi`1nQOL_^mbefV2V-d?%Y|%LHA#lmF!Xt=%9hZ!+jgv~i{0H|F0~?vG;Jv^H2W
z`Wuh%=n=Yhto~S8LFgG({*41`^K@FvU2yfZLrGOzBS$920Dvp)WRJKLl_T^8w3bo~
zx$+2jS_g6e(zo{Z2bhlUoFBKU*{sWBOIfq)(z(vNKP&rZ+{uhwVjf~GYfKk`u&;``
zuz7WV(^*>&-*E~hXa0=#L#8nt$%(3LdlFE;;rTU6g_GSxz&q_Qi6?N?`aMKf{;uut
zVWTFC2g+o3+?sj<%pMafK0~4r#4dO!2O7F@-eDmz4{_|D-yknmFo@Eb<S<aBfO1?w
zx<Ug=DrETy*<#HOiT9D_9<H6W>JF*=4J0n4s`i--G-ByEKJXDVy(IeQ8Y>xrOj(-x
zohz$b4cDON=X9awkYW;JQcFEYrJu%4UJB2CCx;<bWt+4g%3@&w+D9yISjP+mbCCHE
zfnc%o`PXqO3?vo=`=7X{|LMzoaIA2;NOAp@={sUc*D21{`yAVxag5^L=-lzbj}#l@
z-u?eCK25t*osH)Ge1CgUH0iUg&x6`|sl7UNDo$6xT4FT-ala+1gjDnXu5;=tCmYJ1
zc4sv3%XWlqNS+L5C%D5|lB<1g+7n{7S2qu2ss;xv;qiBzQ@#~KA?mvw^X@@ZXgNlI
zeruUM6ZZkEY)_$+i-cRsik#}qhfg?3UrC}N6?qDV`2a+vUm@X=Uj)Vs-d>fx`2EU^
zeiG-=U-Myv6ZSJw6F)H`W;Scm&njEC$kjck!PbD@q(uX1CVpC)SKm?mXcvMd*Y{hF
z<0*Dso^H963^r$8J`W*_W_#lfkCypT6NUK4<(l*Eb>$(=Vcxy@8R(nusfMcB<F$Em
zt1mqu^+Ah!eNcFY1l^!k&k?B2C6}zX0gh<X8i`9ACb7vBJ^%n%4{MX7S<YgRZ&q?C
zCB5ITF5I_W75B@tKgG_S-0giyS7?xw5%z*guPa5Ra*#uq?PEx&PR6t{bDYAZnvMQM
z*tY^5Bpb6o92ZQBdT&}<Kj(ZDC*4o7%A&3a$!g!_3hG`Vo|g<CMaf$B+RFA2r`9tX
ztV1g*?Xg^d>_?c}Kn?>YAFubD%A>)vEt;|O64{xK;=Ff0&$*HtP@_?D5NGT*YFf$G
zmWX>s!n%$snsI%S>p|V`;6jC)E|W-seIagq0b_#*m?+}zHOeq#XOu8*EGEUNS4<{-
zC=Q$ff0HO$esNrtcQB<jyusPB=)TYWLKISbUqu7)W%|3eWDn83m?CFPlGVxHA?Fj4
zDc#6$$%)Fly~JELhQ#EO((V7PIPSa$Bdc45DEW+X2j}vsh?NDCHKw=8sC^qRC1+`G
z_Rr7gDh5a|pZwSel^zzpxPBGVNgud5b?Wi+dM~y;k*&7%W1;hcqzsQczmm2fz|pTu
zkx?Di7AZ}ls0^uzGjWn5XAF%rW~g{UuQNp_@0R7gf<X!drIqTsdzIwPUO57t^84Qf
z{+WrkYj+NO@L)Y;ljMT}v7<f>L8|tTJ12Hkc;^Y15QE|w)8&1NXl@=7!8R?OjCV~?
zrR9M4JH-cRbl_OGNE6$<D~xr6={Kf|Yl7@A+b_e?BE8`8!?=&q+y<4C70*coMS{*A
zQSC}0s(Wc`XmEnd{iq(~J&O&Qd}ita-HEb&Pis_yTU_=$Z860*FgdY6MfUo?;B(m8
zE1VKQgT#@)@mQxP5yPhszpvCs{@}-+KR*ZsM(_YMl&5iQt}Ar%N3MIPlBx8YH$HDu
zpH>7$JZT>k!cMWevi~2eaMPoxc<H{c;&lf_#n@{K3jWD`6bhLX6yFr~QMB%(*i$je
zzSWj?S&ud_;GaKzp-Kw6ihOhgT8a%sDWAd02P>K3krQWFTwklCiuzc)v(U-wrJ*^V
zd}js%8ZWHStK8Tmi8&ll-C5wdw<@V;sAG7kfw^k?e8A1cMvL21Y){`Tk#xuWdi#F?
z#*t+`gpxbDmjMe>!ZWFUsa&UhYWkQ@Z-LH8cJUzC9xGu<Z1(7*bd2ddEfbWdP0Bpt
zE}FKgY0d#ObaT~xs?D2Qv3C5#nm>}vU5|<t!A7_s+`%=oQ&5gZ`~y+zo^+?XKdnRL
zvS>i}Qyzen?stp&*FJFIUlFD5E4W;hqVtgkL@Gc$yfU@6K$^A$XXIonoTkQqgo}y0
zQ!X}x7Gy5Sz3%$54~{3@ss)g)%QO_tZ3ocL-rMkqNp;_?jmV<$Pr&s14A|y=T%+oL
zu$9_(A>kpx<;{yl=~dSl8xm`-_y&F)_GkJrvld*#XfnjFgtwQ5!tK(6?m0VvoF%-C
zw*w!tkZ-P5{gHkxGt+&wY44kv{`xECvHPHoZ1s-;lf}jo<-I$DZ2-9Fy9mN9p`xwd
zh5tBZJ`9QJMLEVEjnpyPy}Ped8436=ycX+VNKaR49i1vBs25VD+wkPNId#?!tH@os
z>;UI!PFgCuo2~@=u^X(~z0ju(pn`_{v-`}yTz1}L_L+$tNn<Q4+tEYqs-mC8hbMVC
z`ug|#YWKi?7F?COm^XN*#(Q;BHO^^$(G1>oXDrM%Qhe~VwXE)0W_ijjixZ);Q#O7B
z=v)>HKP?UMNjSOTkZ&5(;Vfz;32Ew6c3-QpF1lz}Amz1kPZgOfo4{zlwsN`X;TK0C
z4V|ZeGSlWSwiBNVoF;Mu12w*)vcqH@Hy4puXk%e*Pa*?ohsiCbFSS5q`w5@5@pj9L
zEjTARj+Nl!#Ofs{w!6srZnuz%tVewsEkdfg*{{4DG0x6;U>#98{ilgW7apQbQ0p{C
ztohmdp`L(Pij5IAW~$?Ktt{DjUXi(2j`o-T_WHX+S!{O8fcqq=n+t0Bm{V#e(#-V9
z63Psy?7QzJ!?W-J1}l=R3F!Yc&c;i)JGz(&@%&RErpsA$cyMq-JKYW~;a97Z^-jID
zR|uub%#xoi1JNIS6laRSGqTcDP2F**amZa%<Q59evn4lurbITkHkjyoA{s(H=QOp^
zV?espVHjq)iww;bXy&nWfK5S_#5l~FyBTOvjnJJ1S>aAEQGI3$X<a0a21RHXO^5i5
zaGgEs^P@y(_;L5ZI{arr?;Y7Z>4~Ti)s_!cUXBTysX^dYy57NxWuj*l>Clmc3rhUA
zYavrZENpK1Q_!Xee;eydxxCu;WwHG&J(9p&q-`UDuP}Slv5t<)84;^$7G6OrJ||<!
zrl21;t1vTum-YH`#xw0KZW22bfnMYW4>V+7D(<gJa-~HxL1u!#V0Y|Rx0mRzE>L;m
zwoy{aoNPrZbyCN|m!Lz3s4tSg1?YWUPQ@c3C&N<~U&&d+_myN?+oiMaqXjoemVt<F
zquK_%Q@uTFCq)Bg9et;sJnMpB4s$*we0)_-HDqJdLrjS#-4n^|26M8~3^<Fz*)Y7k
z$)t%z!YJlb;^ubOR7l~Vjy|EECY(QHE*GuqA}!ZVJ3OF3^cN&AHqJlsGA@Dvq>x66
zxWbdUPzm3aY{611cC)!KR@cKRrna$G6)ony8@exTCuW5#d2NuN91|2A3EH0+RM2`E
zD5yf|m-z9?RNK}eZXf4cp~(`!+^>}MAPv(LUkfnnS;~1XgY{asg-jk7ex~$Uu`qS?
z8;7C|^-(12-+pEAms9!opCD|ve{A9nNdaC|MncD%7Iq9q7sr80i?As#>H%X}Ar~i5
zjX6?ih<Pv~VEYJQ{S*)<Og-R8SDwe>n#YG#a=Y2?T5FDkZ_^T&x`<vVFfY7+WEM8N
znn~CWg%#s>3+D(7@C9Ri#ZhK&wKaIxEEn~F#tQ)aUhM$(6|u*sgLw!f2bwlu7+UCk
zF;)(ar@!J|tbv3Iz>n<CHu)*75BkwWk0q}zE@uifErE%%!f*vG!n+NaS@f6%{BcjR
z*>KD`IEO>))YE|N>pZS*6AY`Iv4TwoB_Y~T#`7PTdBKXlD7N5rKgT6p15tOitv@P|
zWx2kY4_gyOyYr#~QRt`<wWIo}KRQ}?VrYbLZqtMS>u~X19vBvm`lJP`-M)V}kgOUH
zf9+Sl9zAAex|VRxcWJ|~N@{aA2j!gx)<h0I-;>yM&+{o~#g8{7=~XqFhOeKe0d04J
zVRubc4kQ*4G7rVDV4WkDH+>LJ3-!bi8TULvY_1v-`hJ{NVFF0FDuErV-`UND+D5(J
zA&m19E0^iYudupWnV!QEQ+B-i*DDcg&)12aL^ssjSs+f2<R=dL4Y&Z}vfqv(1Y-Gp
zcwt~n292uT@;?~$(TJLh=gLHw?AW}Mie7K2e!NyqJ&;WDqLZGUBS`izXghFhbvTPT
zGI8u(>cm2stnVseYqV1%YAkHuR&Jhd?a6+r_Vn_8_;+KRMsQqG|B&Lk`)Q$}h1_$#
zXe-6S*#;8ls%!aZK@IubhLr@%ivz~>al{9Sfx28K*lEUz_#j3`^QvdyJp_dlPJJfq
zfb}^2n$=fw(D%AX%!VNm7YAG1bzBqkN6h6CS7Hr@WOr)k-ZNOOCu#?^oZ$X<!uX$j
gX7i6{5_zBUXsUUQg<K?kR&+>9T~94n#rny=0V=du^8f$<

literal 9916
zcmeHtXIN9+wk`=>nkc<mC`vEVn}~>r2#5+wRS-~0=q0o$DxV;r^p1cOr57mzAxM`J
zA_~$YNazqC5DWy)3Mk)q_IaLj_kGSj_wMtY{evHqHP%?~ct@FIj+y6%*K`<;a2+8d
zBV*9LeDOLN899QC41Au39QcnHa<3Iw^c!Bip$#lSAP^V~CMPGSprD|nq@<#vqNb*%
zp`oFrr9E=w2pt_AJv}`G0|O%?BNG$T(W6J1nVF9rJ9hl|aTXR9R#sLvHa2#4b`A~>
zPEJlPE-r3vZXO;UUS3{)e*P0DPMkb>^3<tQr%#_25D*X)6ciE?5*8LdbLPz1vu8y_
zL_|eJ#l*zK#l<BgBqSvzrKF^!rKM$LWX_#CCnqPTsHmu<q@=2<s-~u<uCA`3p>h8F
zc}-2tix)3yYisN1=v=;hSyxy0%9Sg6dU{u{Ue(vvzjp1Kfq{Xcp`nqH(e>-sjg5_O
z+_-V`=FMBTZkd>vn3|g2zJ2@7ojYb`X6EMR78VwkmX=mlR@T<mckkZ4ckiB!jg76X
z?fv`r?d<IA?d=^L93DP==;-L^<mBY+?Cj#=0)aqWU0vPW+}z#WJv=--Jw3g=yu7`=
zeSCa;eSQ7>{Gd>%zrTM#KtNz%U{Fv{aBwgT27C1AQAkM0lP6D}K7IP^*|X5l&_Dk8
z<N5RFVPRq6;o&b{yoiX1h>VPkii&#q^5v^nucD))V`5@rV`Jmu;^O1u6A}^<6BCn?
zl9H2?Q&Lh=Q&ZE@($dq@Gcqz>zkZ#WnVFT9m7Sga=FOX&oSfX;+`PQJ{QUg4Z{NOq
z_pYFzps=v8sHmv8xVWUGq_niOtgNiOyu6~KqO!8`{rmS-RaMp1)ipIWwY9Z%b#?Xi
z^$iUTa5%iNu@QklG&MDS{P?lCxw)mKrM0!St*x!Sy}hHOqqDQKtE;QKySt~Sr?<Bk
ziA47G^`TIx&!0c{_xFGK^5yH-uLA=EgM)+LzI_`S8X6fH866!R8yg!RAD@_*K%>!<
zlao_ZQ$K$En4X@-U@$W?Gqba^b8~a^^YaS}3yX`3OG``3%gZY(D_AUcb#--ZZEbyh
z9f!khY;0_9ZsPIyt*xz}KYtPkgzfF^ot+&bk+{3NySKMT*((tXgf*?}<=Y-)WW1rI
zKTy>4>U}b@6J)v<HE;N)E{$#5nA-=wUOoPrR_uWu%*7|cpYEid5Br%0r<4ZIxq=S%
zBDY0feU^6l+A_E6Bff4IG9&fo&AW5nJbp&zHTbf|GCDWq%NN1hmmg86NLLH}u$y@y
zpa;KP#hRvZ*FMt8(G#PVEjzJNfLF7}1q5|;tH^#%!l&UoUtEf26(kc8JrYPw*1;@g
zc$b3y1_y|o^%COz)Bk+w$KbmYTdkGG!JUHhD(oyv2;>#v<mSeE7e1GUghKtC7T;8d
zq$z%1GR$kkKq0;izS#U~E#62Q$W8s0U3n`WCO?&d)o&x2*V@V&8>D3>mkf2Ld?WWD
zng}RG;U&a%ib5U4LkcnNf12eo*fM!$QJls!3YJ6T%SN$C4$@=BiqU+d2FIU5-#%K(
zL_%9}!kVAPGz@01iBLaU$^t?@h6SFmJ_}B#1Z|%OtWu=|Z$A1D3fMUK-i14A!CN-c
zmRGEc)s0*b{h@1$CCNy^7t{^B1sDkTz$j95oq9ZwGGNic!g(uSGQ=&QWw0!=C7X9l
zU7<g~UPC_8v3Mn@xVfxaD`KD2b3c|0o843=w~W(|9Cr;%=sT;4u-AxB=!4cj0U)!0
zs!7NOPFu?X7Fer+Z=L?{U9KN>eN;L2|Lk-Bs1Ln8MzL6F4E7EM+?EJiB6)X%#xw$0
z0ip2q|8jYS11noyY~eZ0Zcy^`n*}j&!RM+e6(_}2gg1qlAl=p@N;{)QY2y2~6Y#dg
z={6@0XHHvYDbNUWY3vzvp1*(NPj{J`p@1yuVTXbC&UMG0s(4K{(D71Uklu4Jd}wq`
zQOQjknJJ)T9WS5m!nC<rAIn*J!Wvu!_|^!~)#EWYXC-%mPpUdMgu%|ouJ197_FE(@
zM*zexXQ!%G(eCuH-eEk3(Oce99?9|hMU*V|7*OpzLvP0ki1xreOWO;(=<gi;4J!g8
z$t7%)kkS?X2B{ijxs`Z3h4Y+>cV)4Bz?Xz3MfJ%AQ=Lhpk=7t^db1yI<P0_jd5)`{
z58w8>T8p85^^Yk7zfYP7gAH?cWRs17xqWxJz?@v@XS$4Y#3ap%JE|*Nw2C;f??Z_p
zuow%vt7DHP&&f0_F2*c-C4oT4rOZ8!xvt?XxYfP!NPKdXQ0IHQU?_^u!^5s`vBlTI
z*>*G`lW;YVcfmKH+q*0SHsflUy~-1`mK6Eb>LFC+8@3ckm@kc0K09Y+HGbq1ZqGsz
z4TL-GDinA}9jolA#C?x<H7KRBL+-te!|4~!1i1^SHI%(tfYqMnb&|&e0taDng$2vr
z!_FnFAuBqMoZ0RUYx8-k!zQ7eOmaXJZvB+?`)I123Oy)ZC`RMXUHCXhk*t|;^eO{p
z^U#N#aH{bK-Q{`_KXJA;*pKcNKb~pZZ8&tD2tV-C6Ue<l$W#vha48;VjqQ!2BTz@)
z{vAH8jUAc)WJUAuD&;1<uyxinc2EH`P%vjil_d`<QLl-o<-z-@CA0Mh8K-E4@?$RL
z0xGX7JPTlyRJqUr+B4b*d&SSN-tdEz{GY6FKn)bdQyfmsw0%9?2GY(I8S2a)x_r(A
z<;OrU|LEH>()4h*>ig?$zx%tPXi@a(Y_HY6e8~-11Fxe1Uzr8ZhkkSI8M#FdTD9ef
zCQV2v*E@|U#ogE>6aV@rw~kn3pGUo2J$9$a-||8pO$|%fwTMH?xDI_Fcj$wMhd%gn
z=!3V1J}^G0vVafrF8%hwzZ4$C@jsSBI@{b|41<$?@n3r0?w5DOUMBtW-GNvp7`{}M
z+eB9#*otbRsTq|4{Pf-sq|?K`5;eX!YfWM9LQ-<)EO_^lXL@!Bvuo{rl6nw$H&0<e
z8^w+MfKHutF5<`IgDY)YG33TG?aT8kW{ox<Wjaov-Mg*=K?A1PNz4D5SFVt}6)T;v
znUo&N6|%);6VF|7xAOi=ZNzju2^1yG){wrJQRRbS{W>%CdC}ZG;h31Zp2;??dQ@Rl
z%*<f2v#$4{bljvP`qq@5zm-#}i*#^|0Co50w*IB2@$>|l5Vqf_^I?O954JU#k_~e6
zLJZUr1jsQ1o~2ETy}+jQ%Md}LW@JiFbAL#LodNVi#cH`<+iOZtf*FZ0HYs!laV(@?
zX0S*`BY@DAB0rG^kF<QSnZv#;y&w}}H)Z4XzS#y&-@581NqvMjO)u$fM4RQU(o8r?
zr=Hiz7ie4Ok*9Tdf<JXD__~9Gek{q3KkGHGYgsgRnkVeS)n{dqZ`jrC2JEPM9=auv
zG-M7QqkO?iyOn_qAEq32>?R0DY2cMA&_iu#L+JOITn*V1=rG!b_%p8!5u(`+-zU^J
zvMXdvB&AGp41Q$DR||CBx6kk_>7q~Oh2-<9FY6Hnangt2S^#k0FY|1Va5p!PGk*1@
zT>^SOkz-hGqmqra{`h3-&;3NZo6+Wun|UE>7)<wTO31x9fQ$O)Ic&4~g`FPR+arN}
z?=jnRb9%WxGDA+Axgq{No3@xgIXi<sZG;>1=J%YLvquYaLvAu8s(fy;^9UyO5MAQ3
zK(N!XV4tsNBdfPxXCyoJE4x4fs(mtUOIvDal)Iu(z(;@cKfjG2qO^5}1HVMrbzOE}
zb5sjw8lYb29(ZmW5K^nw$yokp$VR~xK5V^}=igY&$(>oVtk^P(Xrk-tiRNZY`kE~s
zA_kVI?gk2z_zlFw&VtXylAW>LsI;F?-^6p%2W_rt^ABxK&NZcpvg4R|D(K7kwGlb{
zQTcS+`AwBZ$I0JO`t-=w4!sWb>0hCJg)nmY`mb;!?KXd#X@GdrU&MXc?}Wi9NlCWf
zN<aFAsIGTwlD|O>+L|%I!~5bFOQKCFrc(=9vt5ylmwu&4+zGVpTI0pWfX_XL$tpvU
z8Xz8)?&=}+%fccSkOrSNvQ~-X7mc^VU_RO<UY>^n^q>sF*lggRdI6LR_DiFiyJzm$
zHyF+QCe?lCM0*Lnvhty#XE5X*8fkF|k&YGHB*dulJI_C1h2pjT0~tUTJ@+5jneo|4
znJ|-c6<J|R`_0}*%76)N`|%d&poIM)rH?7=i&i(+x7lz+jO173=*;|K70QP()r&xh
zMRWWnN<kysSqiEvf0ig9*)PomZFm3FK}HmNA6g4Ioc)qxF12H}kIJDXY$E8vOFjln
z@MEh9oY8&Gey@^bK?MK}R_CY#m>RELMtHn{<t&04V-or}Ge_1=$=U;9ix(R3i%M78
zOIyz>Ao;1_)AaS>qouKG=nSi)4XeFnZM9`fjmL-|qj>*(dZU-geUx1rwGgzR#ZOoo
z1T4akEV?O?@c9`mXDuk!eekJ$Q0&?TdyPM|5pmP*XPs$GMfMUn;g$1~)%DYpH?4jo
zaB^K>XG!eJsXS?YHYlw$v5%4xvo9D13#YX(wEm?GtnP*1tI2aCG{GAaUA?8i#UcrM
z53~^1l|UMScUhbtr2gzVL*O+Do=Aa-cKt@b3T<RxbV%Kvs=kEUHqSYaxOo8L6Cak>
zsB#rohdi#JgPQsGCQm%y9l=>^uAhJU7akNYlJ|$yWml7!Hi<ic>yMbxw#wie<!%@z
zZ87i6C2|-*dPC$vW|cwEQfX)bX;-;fYkq|r9A8d7%1<n)2tGRZGnlu?6>#Ej12$p&
z!S_8;TE_PLeh}Pz``<7eAQF~^7n^YH57jW#SifxoMAJL>+E{rCu{eBoZ17KvY~wZG
zrVJ&OLtJl1t?W4JHX3(_3$hf5yKN_Z{(R@_X7#IUydK`E**fzNwR8eFUrZS#WJr2y
zJ_2B-yi(*#U$9A)n}u*Q9-K<bLtwnhs^;uWjN8`>R%jAPzUqHp$q^Q~%n%$3bXD|9
zKwD++8N^@n+jG9O;@XJTClY6*4je>bes%US*kd2^|BeUwg*@Vtqy}9&Q|qj#PM01q
z^#DjP{PYR6b_{5Y*pDm6mb+KhlcpuDPZ)FlCI0s_^AXnQ)6W`c0x8jFX!}b)Ry^h4
zM{9b84V&=dcImr8G4(!<mO+k+(L{#*Epk~oGCl#8a}fm85PW7#C(5ilF2a5|!Y(jX
zwQ;~nuN>G$4`m<<D!^%8uUl5v9U<ndI>xuS<abuqK@(xoPI4eHhTzsG3@g=+2dL;$
zbr;Q9hYY20R-Oq&WP0TD=`gXA1`)f<<2Dg7oq9dHZQ&TTxu+`Jk}3A2=4{Xn=L(E6
zCPyyYWJsIn3w2(pkmNmRx`er@bPNbG=2o@68X}&IU#DijEAzGQZnk{~S{ga7_I6Uc
zwH~dA7BBGcX`dn_!Q0qBdPOC@C(Q`VDS@x2_eFi%k;R-r>vz4VC9D#XEIrONpmqB*
zIFGvG>4<;yXcU95MNdu4&IO9-N~mnV@wWg*1hcO`3}czCs1`~(mG|=dlx{?2hdK&G
z0%s@S==3gFP=Pc*^JH_44j6aRjdtW-u@#Jv^oIbwCRXSc$(~iA_*elnN(Zjtg)#Eb
zk9Lyx?{8|b^CvY%b&=Vj;rrtinZTI2BsDwZDYHhOHz7TkT~a^f!@EzIE;!xXF6d?Y
z&_k+d+kf9N{gQYNr`j!_D+zw<KUt8w(n@eh8&ZDt0wPr{JOqq?ehY!sYsV7>P0a7H
z<ym3jrK4iCcVFFB8<y!fg)Ro@JY{`$sIKpLU5V8&9<{Xk#o;dl1l1_dN-TvKFhW3y
z;c{1mc*RvoIb6|~0#Dh0qK)Xe03o=1MtwXEB(VZr!(5%N4xIsz+@t?`;h&i$Iev1+
zhW9f{BU%>#XVD{_m}mjdN~u<;36M%KwtMp>X;9rP`(G1?ErCTTd|~9r-#-%>HEqET
zrAB&Gh5oqA<3Gx+HO_c5m`~|o5REk~l8AJKw>U?#l4O*k2F;f!y2@1fJ`wBNnkiBz
z1<wKJ7q<U1pzL$1OAMsRb>-9p2+4$OUIYISU@XhA<U7nwvdy8rf1EVpzNC`ANkuYx
zc(qc2G_XZQ<~3*nY`pOP|EJ;;1d&;NT6y|!u%B|l??b&fy1Z}PA2nn9wj6b<+lPK~
z(tF=nt(K*I+n)U~vTq7KbM=Aof2%-IXlP%7>(%mo1>X~G_Z2itcOEFPi9S%!E<jR%
zIfd?1p_%?-*R>xN3G!w@H%r3-uN;!u50{*tc%WAWKa+H>q6byWdILQtf%o{p!Y`50
zO92tmT@gc459^{~%zU8gir<I)Dp6XF@64{;q!Ycz3h6{IHh4_xN;eTxXo&z0@Ho$q
zA?Mn7n9;pu1^W%~xWnd~Je8u7Oj-^>OpwPdNpzf@74$=UVTxa?Z8I9B`?ils0wcRx
zRYa{Sv5V&M;Bw9I6H*K8E==>&79XG`9tF#Pau%18&jaC;wdUqxawnV-WLy||bXTuB
zxs}3t*sEr*RrFn}`MqYWcj0&7P+vA#gIO}wQcsZLTk|eWZWQG4)F|m5{1QnHA6PUn
zz$d1N4-$xhXyC5P@s#@9f}w7pmhWF~v6}FY)}4Xk-|ye!-SwU{HiJ^6LJ*^`(p}{7
zX2pWQ;c(zIIsljquM#>tK5AFBhma<3QLqtH`>JaKQ{3H407<~j!Uj!E1bcjnpn{l8
zK?70#IwR?l=KVZnD*bq^yXGE3dfSVlkaRN!#5gd|&;cTO|6~8irJU0qZ)mO5^@5*t
zH5f_s9cO@_wuj@4Ugfve*|GT>CexUfvO7g7U@A1*-*sd7>psy`fl~EX&<ut1FZ)Zk
zG`O|Pm*iDjz+Jxzryuc*8ZGTv@K>-v`D}OxH&V$_yI0*^!+q5EaIKU$qIhsC(y?3q
z4Qjjub6zL3TiOuVlhRA5>P(<jeu15dlm*b9N%E%UxI5$Ws%~*7(N*%v8*XbmcPgCQ
za@wh`wvtzMFJMnqcu6JIl;q7kcg5><$&~Qpaq6BjO3?7y@(tQTk;M@yUAQ$(O%b`>
zgQ;LBwZFe*G90@%Ui$5Zs`W^{cQSWBue!B5=e*0?A<X+R;FR@3TpIc0gV~({XH$Gb
z`ZUGp0IXI{ZOq!+z{+>9{C*eo9dD_#W$-<|%LYt|Ow_do65lXh*-6l)fQFymFSmcd
zytsurJHl9*j=Gw0_oU-I?KfMy${$iYr3?zI?=<6Zd3M@>*kfnQiLsR3Y`vI-Wa8cB
zSj%4doi~1q%wQMCoS}qtpp~+-G+Kpu%-qD^^r%>=o4UGN-==O;oQw+&9W6Cj&|DvS
z0vlOSA$s-O*e!L7itxEE*GK<YeToSkUx^a+TOTg}Fc>g8(HJ_H>~#l;FX9wN-^Lvz
zP%ws-l}XXZa>~a_W+UA|Zo0>@U&f|5NVP@H{iA%*^Sz<%sV6Xdx1piE4GSQ_!l+Xf
zN+7}CaB3Wrj7rhlzJl$IsD85{exow|1-|w7fUL*ToPzqMZKFh?FJjX#!butvuXT6%
z6?biv$=FN}+k{U$YZq{w?A(Ze5SU&4Q+md6ILnGa5|D4oC0e3AIwCv$R~X{`BGlK2
zZ)KyB9UqShQ;YQvJLVEwHda3fqd^`z&<{hS8YDOXmt}Go@8DpDdoHFJuMV7k!*;#>
zwe-UJ(yI$)qQ;`#S)wJLyW}Ec$Fb@a_v(XW%3X~xQJ(ug-_89vp|0F4Q_^`KLp_&N
zWOw4r*ym4U&+4C~_b69*o#LFZc7f2tYquA}<MuzL(|uergKew#tSM&We!tb;k+l{R
z{inf?#30<quuhhyF7czK9fjAJR7IZyKZpq-{T3jY!S>ZpLQ+=&pX)-skc85<+}F6w
zm9*69RpsBE-3mj%1j}00Nf7ekLHm}w3%<urIjUU3uCGAh<X<o3ZPi%0l08ze?ZKnF
zEH7Y9cKh$X)n~@mdj{d!GgrW=C(wS=Ln}MRP-QcOy-R5MX|&&cV_X}~vkHbBWxTJ0
zSU&>t;RdIh(A10^#nxXyj6H+Rypl7TuX+pH+hp3EF4$Z86s#UNhAL{@Agu2Zllgn<
zEuvNH3HYFmo$y^(lc#$d^~IME>z6$c8KxR?iYt3LCPZ|PakGcI+?2O)!lgW7)5UMw
zj4HrKa;+}xL2>EGK>VZeGf;b*ozIj;l|PVsglATfSsD-EWASI8gBjyKEAc+ssEnr{
zkPe}+nYcMvc6ijeUa=QhGxNeGIwiwjIn)F1;*#x?UGM|5V{;l&%4=2m!Y#yBCmLVN
zB*raP<KPfS6BeRY!-z-rLY@oOl_k)L;5`NLTmQm;MEBOE#_z=5EHV-x^4%z7_5h7A
zc$`4S^DKLViODq7ig3J7rp9H&^hLz_Cr^aQRST>y7uE}v(mOgCjbrNp8K~o<tV~&$
zU6lZ)Wq7f3jgj)KSpG};mMG6U+r686-{p+ikGq})o_<~n;V}jUzg9xee!(rd_}yno
z+ujxA->d7Qtw{yYcG7M&>NGkxj5TivkMH$wIVN>-fhB|iXzt)$D^$s~ba-n^OILq&
zofx#I>1}ukFwHxK_7fCFTNqV3KIGZEw_g29B63{TYxRf!fyyBki<jwk;PjpII08iO
znebK5UV}VHwFq=0vu*C1PS$MJQ9OIaNgyzCXH)(0jeWqq>#Pz=vT$oup}N6;enzo(
zbQyjIinLNIo7n_ru@RQ5Xa|H!BoBpZ$)XFv`OTL(SWM`+C&B-+39`mwZ%z-gGqJI^
zNmVz}p^aGY5}tsLr@-&gDsp~UogjXega=RWCROV#ZyP|AR-B0%S1m1x9|p~j|KQ6O
u2UCf>fIT{L+WPfJ8jgIaf538kk4d4-_vDiD81Q=?GF|O!7mF`gKmIoe&@br#

diff --git a/16-bayes-mcmc_files/figure-html/unnamed-chunk-25-1.png b/16-bayes-mcmc_files/figure-html/unnamed-chunk-25-1.png
index 3ec970061755caf67401942c1f59c3cd31d1b64c..b0e7277a59fd7e0c67d52e1998712edae7f58c56 100644
GIT binary patch
literal 18202
zcmc({1yq#Z*FOqFcS%W$2-1plhaw;#BArr0cc%j)Af-|g1ByyWcMLJKh?I1X(hM<3
z_k9SzU*Gq){`bEBb?^FL*K)0Sp6A3qd+)Q)K6`&Q5s%driSV!EqoJV@DLqtpf`*0x
zM?(Y2;9>x8JbKUCfEQ?wA3ePfJffqcgFqk*3=B+6Oe`!cY;0^C92{I+Ts%BHe0+QY
z0s=xpLLwp}Vq#(v5)x8UQZh0!a&mGC3W{sju2E7_Qc+P+Q&ZE>(9qJ-UcY{wj*gC=
zo}Ph$fsv7siHV7snVE%!<;IO0tgNgzZ{B2MW4m?h7CSpT2M5RP+qXG6Ik~vFxVgD`
zczAeudHMMG`1$$o+_`i2?p*-^fqVDv2?`1d2?+@c3yX+|h>D7eiHV7ei%UpINJ>gd
zNl8gdOUuZ}$jZvf$;rvf%PS}-+`oVS!Gi~iii!^(K2%atQdU-0QBisH=#i?bs+yXb
zy1M$~$B#8MG@d+pqN%C*^yyPAEiG+rZ5<sQU0q#0Jw1JWeFFmnLqkI&BO_yDV-ph-
zQ&ZDt&z_l?nVFlLTUc0FT3T9JS%JY|Yiny88yj0&TRS^DdwY8a2M0$-M<*vIXJ=;@
z7Z+DoS2s5|cXxLW4-ZdIPcJVoZ*OlOA0J;|Uq3%Te}DggfPlcjz@VU@=g*%*AdukT
z;E<4z7cX9fhK9a;`7$glEId3sA|m3|t5=bckx@}m(b3T{F)^{Rv9Djhe)Hzd+qZAy
z;^N}t;}a4R5)%`Xl9JxNdzYM?oRX4~nwpxHmIj4F)6>&4GBV!3fB)gbhs?~(j~_pN
z`t<4Z=g(PLS=rgyIXOAGxw&7ye96np%g@g*C@3f_EG#N2DlRTADJdx}EiEf6D=#mv
zsHmu{tgNc4s;;iCsi~>0t%bp0b#-<1_4N%64ULVBa5%iFsj0cSxuvD0wY9aat*yPi
zy`!V!>({THot@vlee3G#`u_cUcXxMBPfu@eZ(m<ue}DgvA3p{L1_lQQhlYlRhlfW-
zMn*?R$HvCS$HyloCMG8*r>3T+r>AFTW)KL(&!0bk{rWXKJ3BWwH$OkWu&}VWxVW^m
zw7k5$va*6iB3D;e*Vfk7*Vi{THc%+k=H}+_-@muEwzjvocXoDmcX#*p_V)Mp4-O6v
z4-b!yj*gFyPfkuwPfyRz&VCge1ps>h&-tOA8yXrz*!h3-*!fLMG&E*3B?Z~1-f3%4
zytLuggATB$B<lTrhr9yCyozz{6Y1xfzx47)oNI5m;$}YQ$OPes-_SB><$PtT^=Vp1
z<0+<-(fv)iYgnxmD8yavE?!u17vllD&x{v%dP;S=u9MKG?C2;^%Koj=Z;R=qgNB*~
z(_3iFmBl#%XlS^^nAptiGNx!|g#V|1%tgV6?%y}qRw%svrH`UIGsikL{cFPgAzx4u
zC8gF%M4dxfdbO2{#+$6K;-mpHem%H`mNad?H%4eu?r{l6-IaV(MW^Kpa=H_>=6fTF
zawWpf@h+k28!zU=1}|KCdR%UFNd0p-%X2sgm!61J1d~)34e}8k6ov*OV0OMIgCLbL
zMaPW6#-#`0azBUnqe0}+AXua#*O{GhnTydtiWr#0=$KS82<v|aso_OcCWmZwNF{48
zrOzn^{-y?jRAfvco)F4w(@}HP@b>w%y;*<5h<HyNOk!e8swFDj`#-xM#t=xP#|nNI
zY&m#(+kyMcH&s5mk%m+k1R`)J$j?Ms-K@egZaNy3HbKI?PV{aP&yCq&K-idzWy;{n
z_)ZU84G?<ALSV6&-U8LUrEzH(^F`l#1o@vX`ZD$sq%GSo-<`vey7eF5z9xwcfTU`L
z)>s3#29oAY?#_!h$}$&YVp3UC=_&!lITEDx>WB~A_xE-fH|+6=B@o#-^*^knq%-0t
zY;n-iEx$jUv`Dic@`{^lDeG2hvq8bUyhSst=msUoYmqK(*;T4qXiaK*@kF_7J8*g}
zwcLS=5bfy)8H7E)lh#e9A21buO3+@>2^lO?$YK6hYiMu5@j|(S3-&Ixo{nJ}ME=@z
zyDv6lXObvDXYO=u?`I|<b!QT8h@UBB!OxT}RG76UfJEXApL3Sm{<bet2*r46{8HW&
z50|@-yZCK$>YGU_Z>o%iN9?~3azB;UY>_hsh%znn;bJBNqF>v%luadE<>0xi-IK{O
z9UIQvET*&3<m8@>U24n=0a`&4&IMQrU$B<?{EeHbW22mUp}T@<O0wyQ=T5gntf%A*
zxbg5(WFw*48r}kLXF)}49e%OPZDC?WV!oT%1&vx~^T)?cMh6E>M^We(@~eV#p3BeB
zv<y`Uk;Lv*j$df?GEHGJ3JN^U@e7E3cPVB+KU1vpW%3q-E+q3_kyxGC+4=nUF6jS_
z!v24v_5XhjB0o!3UD_II*23X#KpkWD@u51R`(=re`kdKEU3Js0m)*WMs2v6=sw_S}
zt1cYa7Fc}XfgmZdIUs>FdTqqD#!z`5Sy8{@hD;5IwvK*cU%%y`$9wnHHYV>q_xsc3
z8tRkXiSR<CXE_`q-MYG)T}W8#lnYLG#$%v2lv6>Fk4NGuY%E0(LP9QtL_H#hK>SyW
zVQM7Tn8$wie5V7dX8eyuQ9*?I-jF|UjfcD1=0PeV@;@~_sB2<ZW@_V)H7aD&fB)|s
zQLMfO>K6ZD-9|_$|DsffqCWSpE)aMV<p042O0?o%nmQVl?Zou27Zig)evnu6SLSn8
zgIerr6(@367WBR7vnxY+ef!TV73if^eQ8MTJ@LKiz&s=0WJ!A`4Izwqxg9}V9Cfsv
zE|BOh(T$FS>bGLclu0hQ^q9bA<|%#!Y-W0DZwV&5cXltgay@=*%#w<blIp%UFr_@-
zA19=CSR3YN{skM;JB#JpZAM432pPcsUTxo9yH8hkG9fM!htQi0gaeq^KW384HEI#7
z$KCVL@@uz$zZ@<-?!N}4kU;V^pGOv!e|h}j-NuO*@Fu`?bc48V*U{1@4HUfjeP9c&
z4JV<Fu=^dBAdF(L=W}g33>sgm<h}XeLwdD#V#WRGPwdVgKJb;MSKp^2+WU$hxKMh#
zZU|5<n#`o&by%Cg87Jm?kFdzVyuY)9{*cyG<6XLaVotzRYSfq$IB6lGjPGyD^#w#q
zY!%5NPAjC%bOy5|o0)`J`1>M3kh0r#A28GD+)YkEv2$H%Hs~4g%7>K`IbX0FF?9Z`
zwfq=vtV6|3L<v%#i)jA_*vzLtK)n0yI;7_(B325P_7iUbGbqGFiz1iTb{a$7EsNsd
z`00;K7GezVA{Ke=BYe<>@jAM+N;(VWf7wo`VD$@P>_3M`I2(%VTkz)(pLn<bHk_16
z45T8!0rwupTYfInHwt{W2@+qDLq>LI>pQl8$wRRj>h}jKxR{;!fm80%N<JX1&*9d+
zZaa=aJuO(=fzXTiK?rHVfMAVJV4N#WO_X)uLP79zM$uqkL;m7I#UP$!)+ATqgZ^wl
zL%Hp=f0g&n3+tG_p59qk{`=&@(y#f~7R6GCMgF_dPa+rxC}=@AIOj(L-yykV=D(Xs
z7bKnk@6$g!{*MpQ7rUe(`G)1nbeo~2%K@XDdmDsszbotU5~85~*IxZ!8`pbQxR-kp
zZujYGQw3UoCk6Z(z*%|zCH`t>E>cK-`j?_K#qH<+F0(&%sqwF6)+m&1E4)0n2IQop
zk#6DZe=XCO^hEJr!-7x{_^YKrMApi`?8zpd%4Gi?Lsk7{FqFJybgq(@-XN;m7er0>
z;7$5hwstB>NF?D$1@*aZx}8>@DQlWS<QFgdp9OfwEHP}gQcDff*%=Hx&TY4-@rleo
zPSn1p3HAN8^1U>xI+19+fl$(=fTGyXLUrV}<5QdyLb?fxX#85Yh`$YUh}n(WN9xvB
zR!~yc9M2z;-uK676;F~^!~S~gBj2}>{P0I6rVVV)N6I|^JR`_dOn+?M`E%ML)C>?s
z<uCp)DMYCf&SZ^8bae*c=!}9IKe64rQaDD!a;(57D*q6&^o+niHt`=qoL}*vuKhEU
zXLeQy56VN|?9aiVcDp}9fwiLb&s<Osg@AD0L2|k)&GTxwb^U&P5u4D}Z0p>&YHI|b
zh5x>&Q@XdtEvBtvf5nL@pO_^U<Skv5JAeVZ5Hb`eA|3X^x;|#~iaRhj_I?BNh=xJ3
zf2O%b79ggl>2~waAT`0(JM}3ktg=@JSPzy$n>v|fA?rVcq^Wic);+jqr1)oIqugZ$
zaEeO!trcA2D^_uZDuk>yOoc7~`zC`TT9`|me1wetYK1xrWG5Z>V47k0L`_%n@wB|&
z&(#CF$-vNcaohq2=r>D+nRg3>nLgKp^T@6al}CK;kxtbdL9N@-f%50aEO#Goad0zI
zm3rl=O-d2@RMsuy>&loyyo+^Ai?eed!kG|t9c@zwqFny%&jIbfJW)pCJQOhUJ4qBm
zv|i+ekV87QN?z7jiDYoTZCDvU*;%t%=J5~j&BXo3l`CR*B3G@vCCEOoIfUOlBpV9n
z@@6_(v<uiPlpe`aT~w1Ep77loIt{)g>D)U-cDiRSt}Qz+YU>JeVy3rIQ18v|PmY)8
z`TY8Ael2Lk<S(l%S#$4pA=~fCM-x`D_y9E^u$VzD;ej(mv(m0-vFjM?T&BA3^#GrE
zBu;Z#WHD%qDAV8?Ks!DJ=lAEH^5?OIWmXe}fqK!6e_auLOA@W$3zW4NYUI*Wf)?4G
zD5*Ao+APO~c`lBO9PCy(d4BlHkN=-u()Rsf3a_mPEs~|d88v`2@A3M$@v+}8$<P>r
zC*&%U=f~8#e~5rl&p<|;`cl_HINS1sHVs-%rfUb)^cQBIv?qQ@@ic3V=V<rS4|G_Z
z&&HWbxK{GV4j+xIn|%5#WBQ&L3k25m#=w(!w*+fl7YJu0CY5P~v&QWmV);L*_t-;q
z+H?tO@#<PRc>*4w^2bN&h=P|T_tocIKgy{i$Xhr@D15peNV??Z(@~Ci;5%vnr=->M
zRUP*%inyX6YSJlE=HR_9^YfDsxFOLSPb|uGHmj2hY;1O&U^U5e<4LrVTL*@J{)F-^
z4qhsqv)Axjp%pVW8mz-fzq(HlNl0%Zb>Zy$1b-chBOx6TR#^*Mx%I!DltTYK9oI0*
zQW}mm!B+Sg^H=we3nzjxibX#yr7?Lax(^-s{=XfT6-cPWcN~K~D$3Z+V~<m_OWiS|
z*1!MLBRh-0gHxU7JNu#6=yMN^8>sQXy1jSC{y78hKT-#cAHo8-7wh>}+I#g70@xQ0
z))Xj@qi)@wZ)93mWl-9^RV>Gb#L#;L&KLq_Q+E3$&Y!msvl16H?>=K9b?K3<0T(1=
zFJtCfd2XnzVjgd#9)aPHfKR`;Z(i&a(hWw@{E2DdwjY8&h)QmhQth)<M*j2B335cL
zG#2z?iKPt(#pXP;W^|S|r{1k=(5Cm05d%2-ikXK&NgnUZKDV(ce7wfg6bl~$^V9O2
zXv&+?;&O*qz?sI4<On#$G^nvEIxLQHCBjNDu+!T{LtA*o-c}?Rh~Ow)T!~us5wM8#
zw@G|A^>cS|12Lr{8O&;i;_A&hoo~Y)IAA}yU>h(VhMmFPo9|?|KaI2QlkLaazuMU|
z_N?FWnvm^I`qC{+e&GbUP*vw(K;L*a)xNWSF)qJP=B=}vzvG22-fLVPXOu7WXHq@H
zVuYizN{&Q?n2&H@FO@WCV-5G1mDTU3s^|M3gr6=s1coS*4lPq{h-DrSInj|ygEHW&
z*VBf?D@t*|`puqCQXtnZZjNT$&d(IZKKG2AR9YM5U`vBYo`sE4-j6li9p#NJnYrC9
zgMUHY9jg61nL)!!<AIzWqwjyy1Y!n8m&j))FMM9*qx-%SR>sR2YeH}?y<i#wf8PE@
zj{!%v!=i8I`(Gl4!b(max#_u8Zv7O$vs?FVcH}~DnHxYsYVjs~n_QWNoQk@eY^dGN
zqqE_ZN1mK!%QPQ4U%+i-OlAGC(iOIKPX<pFm^dWDXW9wXKF*CCW%zHp;gj%Q+=7@Z
zK}qzAP@xg^z(nh)=8vY&H!iL@hDjaag?7v}s|qXoZWznxo2)2?Cvg^TYYeep#A`2n
z|KMtT!>N#^vX|PEa@$NEVVkV{al-mm(!D>>E|yY_f6S{fgRPTquWrN{&v45QlJTq{
zfx!^_+zpd8#A;~uxYw@|hk?z-5&~zcDl^SD;&Ip-MgK;8MOqf9h*_+AUIO-diQdxb
z?Oohs^gQ*cZ)~=gQqqKUkk@!{8V#i8$PBpvUuQ81+Hc`H_VItKCc2=&8^5j=1B%d3
zU*uXf7DD=&8eJ+?HhhF^YQDDgi(S38GGudhfnT!T^z$V@5<x*TkeHddhC{tf^j9(Q
z-JM!Z*7@s~9Y4b<ejK_LDYEH!;_u!@t=OsCjzlF7jRfW`6z^OZb5cb}BT@z`sRqwF
zpw}=FgvuXM>}QCJ_hxi+CVCaTZ5h7sLS^Z(80GIxxrKal{DRH7&4%huti$oDz?zOO
zL76iOUxYe>Y4GdnB%$v<@b~KM@Q%j9xqgo_%{-Eb5)!eiG;DK0UW6o=UxG-Iuy;H^
z92(vqDmnJZE$iSEI1tMH_1N>ntJz_U3uYf<o`&bBv<Gie_3LrkDjbpR_paK4HNg`T
zdVW}81tb^T8pSgdfa`3mg<Q8^&k#I44E1~&_7GPhc8zhtTg$+RLoiC!5!tL^bq$GU
zXhxH|qcCk2zk+9IK?6kmc&BsN*N5d=oQ<+eovt$t?teq6C`@(~&6#CtPD&>$)PQNE
zK~<J}qN!pN)7`tMl1MH}Y_u<Z`T`%Mv;u#Ze}MHCW+;+QEwG!u0bi2_c|D5id`;?&
zCf=;RUW%R6M_&Dcs&$S0P&9M8p^&g)MF!duWD|(lz}8RknH6QgGd9|X;{e?MfcwxL
zit1mFGOJD;#L}0B`#QE*otFdi+m<ryA$I9USQEWvuarzLEzl!GSom>6*1=Efm2J(@
z*@G`%H<b+9N1bSq<s{}lMPZT_Etg_Xg150#1gv<?b}Gy^GxnKieIsJkXEWIH@|RvE
zGrH79K8O=r?T@|oP-t#I$jP)NWM=pw7ue9nGt8~XPfe9*75lOs7o<@<CB}FyXiHDw
zM7{xEzOnDpwac{P=y%YwH))Dkcp-lC%16u^Q25Z5Xz%GAvNY#R5*mPiJTT#eC&$o7
zbh%dY>Q4myb*rAVvhMOZ=7;3__0pd_RA^V8J6=-$27Y$q`rCQ=Os3e+yBRM;vzX`_
z?b?`~G0s2R2lhi^^pe-q_9%9Uw=&i`+6ico2b1h7ws<ZgUY9(2fJl)|KKyAN{)$mS
zJyhfi6NJQ4QjWff{n7xkqPP^mj@^M(p$hC_90!4z(-HMF$7q3<dx09VrXvTDpk1B{
zvBW;ct)o8T*hfFUv|qd+X|+M=$AZ}#Vvn<TAS6gdgurM$WcKLNJUqhWMaq?Pj1~%$
zpRl4_PH>NJh)d*E4L~lrNCDYkt-XgRq>K3+qbFUj_7e&16I`{BG}2H$Uzwqb(u@rg
zG9v#P%&3`K)~g|8>C5{gUjl#4V7fDS3e>s()7%eL+1Po!bIn&ucXGouuXk6_{;v8=
zADb%U+Wv3x5I9j3anvjN3Sc}*kXL1w0gA4bq5U`Upp7bOiV4p+wPm7Rh=_dkA?-Oy
zB2C|>93Tnq&kGY)P89^?R!zo2mOhxmG6kOiVZRqtklx~xsWULqn;|o(Y%l)fFnJVu
zzl${HIK_qUY)!VWQwa}nJH5TVVlE#`<;&3D21(fkHY#yV=GSp?Mk3d=sPmPrK2GV-
zkEyn`EF>@#e7;os`p&CbQb~FXMQKpBWTk%lF2Pkp`?Al{<`UQ&kTZFA-=$ImnZe~e
zv?EI%lt8}6^A?K>E~hFD?`$R1@S(C9a_WN%8QIC^{bDnJFw<h5RztDO*KBQ4JBV=;
zn+Ro0BW~Sru9+~JQCDQxiV@v=tb!PEHAe7&!6PJzbe2;Qx4?}C*br%Jgdj>f#F8I_
zR5!C9zixHviN?>fld03E8;$<Sn$TtJ_FX7Ase*#j%@~2F5b!jL!j@MuWxO$N#~7H@
zB^b+`0#o=q(Ywa;Dcd<vZyU-QC3D&A0-xrV>AH<M5u4jvK%<RqrzaVcKBdk(x*<)d
z1s0KVjmqw;Ubb>m>Vs{5H_NcU9y01u10>wR_rK9)w`kZT`IB2}?R!;@+xklPXp!){
zg@(?^`hF{biQb(36tB#*OdiJXz|8|E>4msu980l4D);0dNgMICSaciQ>JPaM)2Cky
zi`_1h<~C^8wey=-+bzbb;X#RHKKq6(zQ5@`gYj`Fb78L5VF7VmqMk)vtEI`e>a!W$
zU|<a_<N-74ONC8jp52LMn4bQyM76r>?X?33OBM0u<DfgkGbxoWw$ygiB9DHaJYU);
z^>~l`joCk@N2SdMqDyj{j@@!$donIQCCW4z=@sxDOo+=(<_EE3v~N%uH|So*GukUX
zTdi5y$%B3#ZZusD+**_5sIl2m#cPQ*U2Q?eZY&XU6&WG;L#sG59$MDlf~k2VLcttl
z|HS2P*tqdYeeTQmt+dn0_$VxE8Zk;>Z^En|N1<lU9b00;jdTtVu=S8&*c=3t@K>I(
z2U+H83FpNhyCuyS<Wh?U&@4s>XVzj7YkFs}ie*2sv`Vb#_3U$^1Qj6yRx^rrreLpZ
z7t0QY)M(S%oGxPYt=b4dCds9=&Q-s$*}0*_&wD2+CsQfN)H=_F`&kDOf(}S-hQRSA
zyn>l39;-$QKU*q*Jq0T%Cz5v9`F<-l$yIM3v!J-sz28X2MCv>``xU`>sEv|6zp6z5
zo`(WKJMr@Ns{?CesMYg%zcQ#@Lv;XD@Tk-+qbtx!`f%(t&$oh5JKgC_QhoJ{9>APE
z%FNSOWo2J5pzpLkJYhk38zUZpRgzeQi4P}}1G{i8E)2goQD<y7CB*NH%^r<cXuivl
zlkPV5*fawNw)J=%J5-!dA~ytoARl*w!fw}5lh=O`lQu71T{H@XV;npI>%$dC4EAU;
z=FaaD;kiH@=#KZPqja?LbWg%rx4sBZ#k{TEPsm_1huzPmsqtH0Vqe-Utqxf7L(&w1
z5jpB+$e@(2?lXtnw^ri1cR4Y879;LNQ*ZwC(BWC`lvY>CnfxUe@yy=UfSTMK$hD;s
z0Wo$Ux~^~@T{n|_6Dj04Cf7VO9Jkgut1W^|JE*BhJ)PY?sMsR8D`eh9wMgq>oH_#i
zlwqvQg-b`&vhyB)yE6+l{D~>Rb~rXF;PF7n(pGtk_ujSlf3Au+*+m>dIdnmHC3Bn6
zqA8Dh_VOKL8hNq;Jbszb+O;1VEc2Z;D<8<W934xTXYF_`pGHj`uYJpp%|YNf3jrE=
zvSbv|Ae<l&XKg$Ftj<p73u$%GTI?vL&GH8EuLHi3F)EO-KGi67pTMxhbXy8xWnxez
zm%_RYO~zS|_Ti)6fFK%=lS8zyr_8``!bXVR@E(DoEuQ`Zn~r`y)We4F$yA@F?p{-0
zI8G$;s4AFV5B89+8FMfa;f#7B3F*2BY!y7HNr$(+EYWgk_0h?9&oaPwGPlpB&ac@V
zPTtcL&&}gmf4K%0CCN_5^L;LdtehgkN}vI*rh+y&8S?Jn{n?R4!j;!n!FxsV_z?Er
zKv0@E>4DB9|7(-l8<EW3;cR4aVY<?)hpQil^HT*(wEL1v{8nq)c;0%5FqU+mw6S&b
zd??~}iG7RSO;Qp}J>99`wEVSSH&gw@sbE3U`Iof)jSQ1iZK0@{WGxNgXuU(W|8N(%
z8@0AN_<aBCZ}-^Nkxidc*OtTa!`^4Sf)$N8#QNTRtoXxkEU@!a5!A;{1#*bb{?CoJ
z3aY_HKgP=!Pa33@0r&A23+hNo(ks=BXS>`=Y-9`(haN%nAoL((h}#*}JdyFRE9gu>
zZ>UK73AOLNj3<_fv9moqO?5Rhx<P#N8CvX{)PR54DvKyG7H9Pco?1@s<+XmM8a6m(
zH5ItK;wJB*`vJ!5j%c1x&w$77&*(?M?T$95&3!jN^hB%iotA$#7f#QdKC=b<tcXcU
z??JVZTULRS$%1(S$cZ1rNhiK8_1%rP+V$gyReG?eKsda0Nu;;!l{ntrk8jhx_IG4?
z)gK2bf=Mi^hIc%#6*fFv#8+-HwfcQ3!M)07&BxnaenZ@fKaoiSCIm#Ead_&4DR}zc
zWY|TloRQZCzo88t$iWQULpA+SN&W@1ntp8$)vRSVhuw@DQLRv@Uy?;if2sSpw%Wfo
zdN99dluTOPw7K)GD1+v=__r}eQTO@>VsT;?cj$#NiPg->s}E(`d(yOE1qyG1|6l>E
zh4ID%bQWt?Pb$+Q%GSTl+<Gdx9~B&zR<=>+v^E@)u-nYX*B3WCNx83>`I?%V=(pgT
z-d$;8Aj}u7gb-A7n0rcc%|x^teJmr0zk4LD=C+!vm&<_N3ElRuk<RWUFSDTA{rBKr
z>-hEy6a`P2cBB>dz$Ky=^z1u!bTLMR3hUpLoHE}{ktcVcu-b;5EONpM=XiPIkh^dF
zq7HNAdiUX0yg<M+rk&7=BvVVz9tiyngkw(EX0uvOv<5hxjSH+92f5g4if%D9R`&G!
zjPQz1_51CKL0N=lAv)wcPZ_r(aZ@Nb(LpM!k0HW(RV8lRIwW@8qk3DtK~nv^X9ot^
zhH>Tl#ct#SR#DN#CV0_R)GG~N?1Mq6i(~;%pS;!T$fYLx$}o774LvBRcz{t-hpBCp
zXJM)b>eLl&GFKub(5<lcuw8#@ckAo8K<R)~_G;ppkz|P@<xv&~>ieprPQaTCODt7f
z`m}bI;wi}7hi=`-`7EFJtjli}?|Qx}ThW7kX;&Wb|CVyoF>(a<!0{C*=)U`hU$2p+
z0p&ngcz_1e*!+~u#gW09=eS_^Q7XL!5+YF<E10|7KH7(|8m7PgCN&c-XtZSAW@%?!
zDbE0eHRm$nI;%(JBP+EalO|$S?VM7Ra6+X6SkM=>j@Fj+GIk`pl$J+sLSU5NToy;M
z4^e-IE$j57r>}?3<txg8vWsH)u|>jEU0B45YUv5xO!w*JUBvgaIubqUGMa)Hrgji*
zEcUBaDB&N|^vm0_?y`)(yXafPjv|Plu#SvKH57VCy}yW&*1aKdLrc?q>(<t3y}rQi
zr>=FzdrA;#w%P|a@ZT}+R-S$`2!sNsb7wEnP2>1l?CIqQyWRB~<^0iYlEo+0G4Ld@
ziS&{+yhL`buEr9*#qKe1jS_yHpxw;f;uqGBF1z1G39sq9J0?px!hhJLp0--VkU_<u
zJt?9_(w_<Z>$aSIvTPcXJo_1z<OP)=-CLBxxBZ_l{aoFB4sVKhf*7E`&(x?i$)8Ye
zsaR<St1|08<+DWb=1QH;RIIz6xw<0tmO(|<J)S;U^B=~h+-myoVKqcA(I#y2-qKH>
zVg#7hk7roM&$n#7=aE}Z4V~9?67ycT>q2g9pVH>U7j&a$r)i`YcJf}p7Rf*JP1mpN
zQ9nUm)@@XnLd&(6M>QDGL`(3_#W4OckKJPxs@jM0*hM4NTzoB#ze6X8itV0F+NLOW
z>I{ijsR|a)^d0Pn(E6$AC8lP<wX!vihE|%VRPM2L83B0)&lm7E0_2cUVLCsDy6aty
zWIjP_@XA?`-DgoJ_IA&DPoX@OGMMTXOaMCl<uw1j`TR{)O;b1Qexh%|2~g#=#%@<8
zAXXm$3i1Y*_WWvH$e+ppf6j<8ak+Oq#LL}ZH|+z435w>A^5%Uaf~gEwiWz=qDlS}z
zji{j;vaE@~x2IQOh!?aj4kY63^*kTS0VEAgqG#~aXf8|DHj%2ac&0on4lP$Qsn_4x
zKLFp^g?R{gN`Gf;d$E={a<@RZ{1qKJjLAjeXdojn*QX0FH!vk^G9}GN)hInR2flDG
zQ})$M4#S}cF@NB)2yks2bN&0<wZQd<uZGh~>c?Ebv7`0@DRkuE-qw#f(ZiD)6Kqxb
z)!TX7A|~wDXJ&qJ`!^0xx<Ae2i=KKber)tCJ@pIxb2fO0ARk8y2<pb~7z2aWVjA0b
zdGw%SA8qxCRTVX!-0r_mDQ<0LHjrXuJHjvppN*~BEglIU?>9+r=$+jPd{aj`UB4ra
zFEI5KOK9fCO8k6Y#>-?i_XoT9fQnm_S9A1>W82j)`KXl128rq3-Cy_87a!2bjdrqw
z8oE6W>HVnOQ#t(5c5DMev9^7-`-7_LFcD&13QCO&_kRAx&cg>L>ESJXGzG|b8`n^C
z!aop<oMB6R@#N%-V^8-wL|%a*ucimjBb!!QkbZ4zH)JkruM(abd%E+cTTO@mF8I2t
z4x->dA?d}bSfe;(sSpiFZzDLU_RGM~xTUjI<l;xo_VLEfMS<a;PrC*-BdQd`q8<UB
z!Oy)_M0V1l#Dg!l5)YZuv$%?DRunRWUPNhpEii#Fle7R74dNgBaCyRNmVZ&K+5HMd
z#wGPPmY=D!3kXe2`c63rvX3iOjWD1+o25S+X|`2lkGE+Cr4oq0Sj#BQXVQq3nP&pw
zC>Fpw47~EdCoC>|uznF1mg0eXpRGp)Be=?&6QkD1zN}+m4mWEmY(HJ-Hq50_c3;_p
zc!t88G&oU774o5L?#KCS?WK<LYIKe79rvt;Z;KczW|8WW3`C5B@7&O1sEoZScDiMH
z1eiz;N)avUw<UP%9v`Ibno4QJD$$vX8EqYJt?)~4c@jMlP07)Gj?;H;4>_^Yt8|*D
zrwWT3u}0bnRi+8OXmIl*R*>8wCWPg7^E8<;kTVk?zw{LUn65o#qi6IqLOeI^y73^B
zDgey!HH5nr&u-*FVqqV#3VFSr<vqJrQ$WxmWFDCg|2*szllj;);cI}&%wD;LT31xT
zE6~WZvYphA?5>?a=Any#sKo<X+y1m6XY(6)?41nu#7Hm0q7T;YY1>pkNQZ{GDW$0O
zIIcq%<Q2us=W^e}A)fH5kCoL)LShr2CEX9HGpRM&st+y$SmMXzwe6wGNcROGqdL^C
zEeWZjePDxtMv~Z{<_r4xBO3gqPx{dl8x`yLw6$pxewK}p4zsz3rhl~r5~#zvNqvP#
z*UzY1IrMh4X+{`)N6oO=F4_k4UfgONVznGUL>H~N^@p0Q`)e(WRi6^7aj0777)P^K
zu4jHMvtfB6(D~#v;c*9$ImBZ@X^xB!-PkjcV*k7$Wpg41#9yN@g}4;S7+|euyIwE6
z58^RFIPBH%*|+1T3)S)PWg=OYVrDf^DrHNAn(?w1vLk#vY;BMnXjBYO9shKlPVf75
z`;$(Cgi0W%-KuO_W{*a;cc4|*FWZFr1PBowYx-_LCeJ!!XI8y*N*B}V?C7>CkUTI3
zrKZ+W5Pd<F&A*xrKczhMzG%E2C$cqdRX7$_<8B8a1ny9b9-mrdMo3k1NyJ$z>`#{i
z8V|xV6de`aY(RYwoQW=J6nrNP<Qr1q3mu9SlIbpaYatOmT94n503^PKcS?;_jc;Z2
z^}Woy0W5YEQwU7q@m<6o)flekvV5Nn04g{K<JPhC=e0A{R|Ug^_wnqf6$ULs_HMaF
zh*gu!rUAe~5n0UBBFhat0mmLx7CMN)jKjGNOM%)uu-Ak!&(_)*a59)du~zwZ3;hDX
z(q4j6>1s(;cXds^M-*ynl3fER@q<vu9#fbzSxys!2!3%C*W|Fi@k1g~xWaoh2FWRb
z7Iu&g&)ROiBp-QTHJ{-NAqr5VjX1{N`pIkn1!MSiO*S>fsj({Tz~^kZAmC>4I(bil
z6)EcTS7a_)9KT)LXc+Lf$Ae&PUK_3HuCMti2!<8l0WgXfbx2;+2DyNjKzq{B7r%4P
zQjQwhLcw&a%YdFpHLU0#f<QPODu`J2LggNZL&+{W#CM~X(X-?GFX2Fd-+l%=V_ar<
zk&K~Ac}fFdngnLySY*YKB1FKiD&XgFvZNQbdE`aoQG7$m_@!6Axk&Y!_!BQ4V*}&L
zBZw+13Rd6l%{#4*!PZ(U^P|=M09I9kQjy#`i$kjPviF`?>YT4*9h5}m#`&lLEG8(-
zO<cP0iiS1g4WNS!pwwOp=#y-f+N5wb?L!}chDogTH`XOGy`p6f`l5V6(w6yQy<bUn
zNzEbMDiJ?m`nScStqXt11BXC0-tk&)H?5rw-K6J^y10Pn1Cw^gfQhB{QIU(mdq8J>
zz^Qhd{Sj-eu7gPmK!+oUq;zoVq#L}NI<t_(?>YC<z&?4etfz(KD2fn*<w7-@ChMMC
zUj0*d1yzA;2jY=UE)@HwDkB<MqQk?LwB~0*jrPgapB#}@=p4Y-&_|T=Dm>v!sCvPK
zSH}xXKty1;Rc$mo*3f@&uYcF8!NyY^*FRzAd;xZlwHS8AVcu(=ndbh3FM9C-4SR{z
zFnxDN=gig7y{EK5@`u=Awu1q^Mfc%)GaaB(Jg8yKoPf2jQ+s5pPXf@+J75~S%{wOM
zc9x{)@j12Yi|9Ny5SnTk<*Pc31{SbMHQGSXz$0qlfvvGniO>T9KJRPz3Dj<0HMi}c
z%0DqJPlR(Lqk&c4RQ5ZT@dV^Fadm&yC^Is?<J>KOh#;<PR1_@t(0Xb7)&daa0%&3*
zct-7ewB`x%C?NA}bco$<`7uMWsJ#WCvM~5oM7y}aL`?Q<JQ1!*<aJ0@H+`D)Zl`bP
zdbIyp@8tC`AY;gmSbgx${v5w`k@denRx&tdGj?{pUr_l*)&~XCS=XK;7pcNeYKOYh
zgP4xJqe}R^aaTa2BKD7QBYne^+j+J(zWo){vdOh9^VX#BrZnf7)*+?6g68_(mEQbq
zeH1aMbV6B94$(nL?e>lwt{VkWMHEGgh6#&8viY;-L|Lx@Q9GZiSv_(q*9+%38CEXi
zO#^~4Dmt+ON4ski6%lxbHFV})U%~;#Uf{6mYu;Jc*i08zvIOB%f?_Z5Y0xgxGAB*a
z>h0u;+!s!?6F#A@0AXKd{Znzmt&$gN6zl|?QUN#t<0ZIPEZ}||X*G_Wmz&(Hq9cBx
zr%C`TFR-!5Zr4ne@6Ee&Rf68yIiw;PMV?ua`>1V7B7dB}b2RZw7de`$2Ae8f>#Yfu
z^=aRN6Z~ajwl`HYpOt?Ne*DT`xfEeo2Y@CELSO!@<vm8#>QPRom}b1+RuF4gA<9&0
z^7L9Z5{tlxz_t|FJ^~=O_Ko+=))t1%ZzV=1g?y9-4{V;?lMu<PY&M|ePnD0*o|N6T
zD6WrOv0??FYUlXiya7BDVBHwz1@w?0r!!h|56=`Te9WFK6gDLgc1>eZ$iSX30Pq0^
zcY)9t@?N4XLOkdEz)sMt^7uJwW#3ORTD!u>I~98J`?|IYB;~7RyS5-?rF5*;VT7W_
z`(t|b{4iTNQVgqpz(Uj84ixxeB-~LnK@8_udMD8OgB&8lA0hy4$p#-28%E5_=L4DJ
zCf%fzycF?WMp0+m1*ktHOcw;iT;i`y6COW|S06Ods=fz+1=OPU1m5(joAi%jPKzCA
z_>>%DW$!5=e;}g??EIW`K!A1xk>tNeS-n4`DXmrmIz7d$izg}0h^O3J)k<l|EWwk2
zl*DRoX$8bAEu?vQ9W?2WTR=n33YR!;VJ7xpz?>t}E8%;b1neLG#ScLBA8suK#wk15
zgS^aEH^C%O>t5ca<k~gKCOwatbiZpHVz2NZmhP8`<vWv9x&a2JbZq|(xU9Jxt%BGK
zt1z=)VbYy;NLY9&8F#oIgiL82Uz;vMmI7(~OPIRZTOL%{x8}~jAO)P*!rBmNdk2Dv
zu~WBUTJ~Gh^xZWNT=*J+ti8-7teHZ<9+Kb{HV**z?rwETuTJkA72iv#dbXOrioL%}
z0@%EIMdDO;dr1AvA~paBu{fHxui7A=s!$<3{)ALL#oi}&eF&sz%t`ktyJlkWR|164
zFng+xmJ<s*-resepXKJ?wL4PpYMaLJ4+Chu4rGV*L1(G}KrAhytp{&!NRvi<r^#k`
z5<THeJho)u5pEhb=B{Zdo3r{kE=->LsHBo)^qX-E+%xr~&4Ri(8#c$0{qO?{5MZ3n
z(Pa{kfw3mhs&uKl+{nv5>liGYPPy#!t9%I(miAX!vpUah`8~GB^G^m~XVvIxrS6i;
zZbu2HK0W&u$d{vng=xeZ%AXa<F53;IFWH{Y8xS};B|2N_-C>CQPVdNT+a9PHWQuj3
zr>hr<S;9Vt5e7IuYPc<S^lb5TZ$rD^e`|MR#R=yejYJ*9TB`-X^$auo+dT_`+K54n
z!i|j!y!bEsOxQVgZ)dcU&s<{OX+m(FPVPJt)F!GRuyg*wJEFvh3GA~d)?1{uGfMgg
z<#>*3EbIt~f<6nZZY`hSS>1fw@$!P?O((GjMqVT~GKkvvjHY)3n-7*9qKsB**nsK?
zQ$A&?g1+r~c?2fip{ld*LoWRqX>%VWg?P`qB|T1-a|pZK%pNg>A_|-jPw{Wa+dBPf
zVtTPmpsoVIj7THftHw7+%^;2Ni`md2@(2}#Ce)bC#Q9iq7~G}$Ph(b;!qE?(NiPih
zAob0O-`6il2g)`4aqE~h>^=^udHS6kCZu1aSVsu}P$m;-M%%!xbd0j$Bouy^^x_>K
zq-(~1#BoNvRcw=S1$R;^^51z`P`zOSyNlNcCX-5Rq*r^1MK<<z%5VuUcn&}{89_7P
z4EHl}2OCDas&4M|vOBQ9JCBb0%{Zci&TA7^u`r1tFnBwY9acIyGyi$s@EjmsjbG=(
z)E)>+nstIPo7ewoDO&*F3Y*f^tDn?1|LLU8W#WCZlUOriXzfKiv5?LTxnLs*bOaZ1
z-<a8+(wCNGL#0qfL7m0KOKknKS@SiE+!m$|CKPGKH;iH(g(c2&jKFGFL;5S<5V=n$
zlY1Ng_S|cdxYOK^VYTy#QC5-W#Z{C02dauBfiT3*Ld@GD5hY5^4WqaEQ1WV%C*v)l
zjpCu#Wyvr2d_5K}n82$mWbbLP*x4L1J{}VNIgNhp=JYq+t>8KAiRin2Pd2++M=kCe
zff(-GPkv^PM+r}Gco2FKD_2#Y;{mQI+EmwHB5P9^41F<z!ue-ldrk)clC=48J6OH2
zie2(WY*;yXbq^_ZifUPADEKU$<xnq_Q8IgLac!b`?DqynS}LMzXdYWZ=Cv0NOU|n~
zdEZzr`W69{S3r-xy=ucD8*Hin%%L$in$-JXWH<x8qN;i(K+;v)a)t0XJRrpIz<k-%
zPqj-lLHgQycg!^DwoX4f*c2}fQg1B*Hr4T;!3c82HHWE98bU?3h)Yi^YFg7LpeL?=
zJGm8i4X%6H_55ydS%>6RB`@9`_N=LZdALA@2WvcxV9ax`yQ*;#;gZGF(Tj|ulo=qW
zcU54yzIr=JORsmDu*#MC-9o<np^Dh-tXo+))KrFn8<9b_fsAE-lg0a|y@~-JIl8Vk
zK%608yF~8%%x~;$c8Y_A-3VG_x-g$hLbsNQv+SjE33cPd`h53qzq2{)(pk1qG(Hy8
z_`1rsE3i^xPY1?*k9rNr_GvCclk$4*UMiodn-SOWs?lQpVXEyG=%^<t2zJuF#kX?&
z@X(WFx^nm)*DAi=d_s#s6$QwoP~s0Q-8j(2r};jON?bg$i;|)rA<A((J^_t0CkIKx
zF|fP4wJH7=Idnkb3$t2%%E7yDRMugB-H$y#`2v>eqJqUFp^h|A?C7hf*0(Qv${35^
zDKECP^*D%+X>(OI5q4y7=15E`ibQWXxk&d@s~RBGdTRDYna+w1qGfN$pI2(Vyv2h$
zBllpP5zGrQ!oGTS%Xvtt;Gh^eQ@;2By6LOfF*Exu<iX2}dI!}<2z7EnzWf`T{N@E~
z2im<@v4G<RRCuMt!}E$%wHrqxv2WU{c{1)l3I|*)pgQaW9-iTr!-Q`Ils)v-RofW&
z1XO&w&S-2b0GjgMfp03TBuO5)?wnVZsXjogdFm8(6T{A6VJh7n_7HGN`-%FC$g93l
za0+{y4z{=jhOeGFs0^IT*<Zo=MD`k;vVR2bZ`fi1w-%t-OrC;lpAD88T^GkS_=kKs
zM25+D+i$P(-{{&G6>0LCq~0xKyYpMqGhBa`5)3CXI9&hz;#U`#?gAR#cMbEjz~ktM
ziQ#HwP15~2nSz~w%j&k2U^rHG?9f2J)J1KZH9iQ*gfI~NV|xF`C2)EPce*jo#fUYI
zrTzG#9_ml>nf6EB1?=LgY(tb6<t;VLKSH>uDe%eXzZ6CpASml%Pc$nU{6;b)2P7_F
zAQzr4Q0&2j;tos|VCY!%&GNZWihrx&HtXQYJ&C-c*Oj;aTm9Bm+SD&=?5Ts>hwuEa
zCWk}K^E5Kqi+38I<g>fxYGkpatXepRF8#p?d?0*dv=hTqCkT;Nc(*YH3oXEhZ24p&
zzufn)2B)ZoqCe;4JV&nsriRLA!6Xzub$|`7O(WTTR0e^|FkHB&LfVj{kyBHnpUDgU
zIDa9ZuDt61L_TRF$;UmQ!757ts#X<2Lbl+y$2g=Odga-&`(*h9hv||7iNL;NQk9Du
zJ?Ykzuhb#~qW-nuluInlpRyLL^tNo#fe8PGKOf2)aW3RB0pgp-SVeCCO0QS`58f7w
zrl{xM?dsSRUkmkdYs^hrxRhlrQVzh6o)_r+Q>xKU2b5~;{P0Kn2L}D$D%$3}UMF64
zzbe(}^ng%aG)I$k^#*Q?mjzV+>r#h5<rDvd63jn?d;m~8n#P68tSR#;_Bp&G)HM&(
z5#N5hVomkkQ%LYyjA+W`*jv}UMYVJ^?e@SFO=kChyT<iVA#U9zK*QFS?)?kl-y>{`
zKYp(?klxA5G9KOfMc4R{yOp);4-h>f)Z0ST)%DNqQ*BI_YK68d4wOAV6u@FOz1(wv
zFs?(I@5)J(9-}Qv@~-#v>OJEF2I!At1+ISPL~h;AQi3Z{)fz&P*_diwc}H^P_Umog
z*HW~4J=YGCY2&zqhANJb6Wu=y_DPTCu19h&qSdgV@TWUYC5>k4k-Zw=&$7mJZp7+%
zX^A>}q2bBhMV^Isz?-SFZ|9efmHbKzgiRQdE_+?=tp{mwFO$croMsA@AOja`JNv~a
zHbCkB*s#=J(5n#It#Le6Z863k=|n$_Y>r^Ejeyc!T~vupsCQKHFQ=+;>09jCHjrZG
z)jM^D49IG9ANcDTmafL0>F%v2Z7=wz#~=V&Wf}&PsZjl7i>=UjxQ&m8$yYMQ*|=W`
z^9}<0mV%q#T(KJ#>Zww#h6r{mUi-SrBm9plJIu(K7+usRl`dB`X{w}t>XTL~;bZL;
zSU(kE<X@WhM(6j2XV%HSHc`Ku{1qr2M=_1U>&YeesR7JE>Xs>olwNvY(iWM@cM;1m
zzEodVhJbeqFX1P4L0U!OBJAs*F1b#OT4zo_E0v_k;7(l_2G@F?-7wbcdDy+;Wi@l>
zwXhjPalDu3(NTldlU7j|b~a~y<0<MJ+1_^-t8S`kyL7q+`|ZzFTaKkJ9=Y`#O%B4y
z2>ukWhP(YzQ%026)$%EHzl0qBpzJ)*wWzW9$sUosw?BWUvs60$XFbzl{^&g%U>fkI
zO^^GH3?^Xh%LYe`y59heXpJYD^&0k!XNu%@LL5X)uM+IPcxN%j6tbV)X){adeVfVD
ztI_1V$aPT9h3$&A1)l?u0F_?haw_+d=6?Qpd&vVTztTcGR;p3@?(C<@^RxR<4e+^8
zmI0xJC#oEPhm^j*vp9GrHChWR@QxDiZZLaeX%aQps~ePL*cunxJM@<IpL^tjGeMmG
zrT*O3n(}_d;vtauQBSBUcD`=!{?xu`kEA1a>1rg22$Sbp=<#^oejjM3-%a69RqT&@
zqZ#D3SZck%0^{>%Z9p9?UBg8^lNtfPi_+5Rqxh-xcH6RNdztu`hgN!vR(Hw|5j8jX
zfwFm0n&&ha5cKdn|9d}l`RHh2(kl|r9}r>=t0*Iwr1wQu@}IZkUOeID8x88AN0Z^^
zC#CElp{M&m1s(`<`VLUAfQ4IUj-smmRsbtovlL8!rc#O3LuP$V#<V_OY!BY@u7yGE
zTfBK-|3Xnq!=rqsCS#!tsmwOXojv=O1&11I;o+5H0-do)B(2JB&NjVy9tBcbS}3-{
z=Ipr*Rjj1Nq6zd`T;SVODuWE#?npE=jGXho3vdP}zU{TZXtmn7GAnOdKS;4Za6-Q2
zbLKv508+0?ynbyzmz5&5GO*YFF(P&1=5VZuhm6@mA;V7Y^PuS-^O14Qj|}2Q^4`FD
zeu$}m2y*Pw-SEq}UWQ|SR+?l`9MK)L)q7S@&<LjUBt06KV*0+(J_@l40SSWx7V5o>
zR1){>Jo~u_czexZ4r44CcouEW!#^19>zO8I6^Z&;ij7zBC4wuwd(sOJStrJZi?Qrn
zRu-`L%GHY4MU9yDELLIISXRPgk;3}f#^B;NKcfOZZGC)h8h;$)kuPM@?}|<Y@z7L6
zXcdg+w7wF4=Vmz+ENC!{N*|{6tP$8EH!r%!fnw3x{e|Bi<DTx$W4anNE`@{J(>2_!
zIR??^ea}+txPmJYQZnc<*tB|cfU3HzE8XFVU4!ZBJT_Lts2=yT&Y2epQIff{tj~)H
zEgTM4YM#G@*XFXKDD7tkQykTl2|OV{7hPA(L%u1@^w(l?X4MTtq#}xA+jd#Ypz9!H
zRty|00F-?2pfHmpCvGwtACigOdCZI1qnB>A>aF_WoC&y-vQ+Q1#z8Z4rC{a9xb#7V
z=JtCT+b>?ieSxWJ@-+_+(LiU^ppGbY{bH;9y!260T|N)YhM2XuFLZiZma&tm(eN^g
zt`Ql8x5lQ9=j2+&S->mEdGBJ6gCj$;&=P~+f^UN$*<z8DIC2Q@JCF)}<NNj*s$#)_
zTHrz^#^X|mSE-%G1Sr$y1MI9=oIU6sBzV5KLkQ}4)y6ZbfOOg#G{6E7s`Z~R8zA%D
zdVteE(`C+sa}1syl&9DcIVP}AZ_`~|Op<V^^$^E`m@BlIjDDaT-sWtd?jvOcA=%ha
zlTXcGj_j#6$Y1}NzsH9n@ng#{Qis4@e(5Jx!P65@V@?9|4pG3;ZbQ0L_nhiCKJI?c
z$~!APay%2}*Mt0Mgj@a0VfD1Mj|~%&L(n`kk|9z3UJoA~0+bt}yE*%IO#_TRckSN|
zEyC{obH+nv@j9HERD@k5cIa7ejvMAL^#9X;uHyg3e>j#-rrJ@!h}>L61O6%9S5qjI
IGY$T~05@0pHvj+t

literal 17969
zcmd741yt4D*FT7ti<Gp0fTV;d-Ho(JNq0y}cbC^7ltw^WK}zz{%@w2tq`N!sr5ol8
z==1pg-v7JiKWk>q3~RBtalZTPv+L}$&u0_-Tu}xKgA@Y^2?<N~siX=L5;6=43G@UF
z8TiDt=dv020O`5BniTN%)~#D05C|C=84L!aprG8oeH#@O6%7py9UUD50|OHi^Uj?+
zSXfxt*w{EYIJmgDczAgD`1k|_1b6S=B_t#yA|fItCcby?9tjBvDJdx#8QJ~&_sPl0
zDJUo?DJiL_sHmx_X=rF3Ja|A$OG`&bM^8`Bz`*eE;X_77#z&7HJ%0R{iHV7snVE%!
zg_V_+jg5_+ot=Y&gOih!i;Ih!o12G+hnJU^kB^U^pI<;gKu}OnNJvOnSXe|vL{wDt
z$&)8yVq)Ur;t~=Pl9G~AQc}{=(lRnKPoF-Om6d(=?3tXLoV>ief`WpgqN0+L((~uf
zm6erMR8&+|Rn^qg)Ya8BG&D3dHMO*~w6(Q$baZreb@lZ0^!4=(3=9kn4ULS9Uc7i=
zY;0^|Vq$7)YG!6;Zf<U2Ve#_iOG`^jD=RB&Yik=D8(UjjJ3BjjdwT~52S-OoCnqOo
zXJ;1|7gtwTH#awTcXtmD4^K}|FE1}|Z*LzTA75WzKR-WzfB#pnUO^y`*RNl{dGqG&
z+qduDy$c8k2n-Ai3JMAi4u1ds{f7@9LPA19Lqo&D!otJDBO)RqBO{}tqN1atKYsie
z6B82~8ygoF_vzE8`1tsQgoMPz#H6I8<mBX(l$6xe)X$$kfBEtyEiElQJ^ky~uNfH`
znVFecSy|cH**Q5m-@bjz&CSir%gfKtFDNJ|EG#T4DuP0x#l^)XB_*Y$rDbJh<>loS
z6&010l~q+$)z#HCH8r)hwRLrM_4V})4Gl0D?EClcjg5^>O-(<3{Ag}&ZfR+0ZEbC9
zYin<B@9600?Ckvc^JiCAS9f=JPfyRUU%z^Ld;9wO`uqC_1_lNP2Zx4+hKGkoMn*<Q
zN5{s-#>dAeCMG5)C#R;Srl+T8W@ct*XXobT=I7@Z78Vv47nhcnmY0`TR#sM5SJ&3o
z*4NiJHa0dlH@CL7;BYtsf!N;O-r3pN-QC^W+uPsYKR7rzJUl!)Iyyc+J~=r#Jv}`;
zJ3Bu=zqq)#yu6$%()R^?0NvrKwlfkEdEnLWt?<QdGbAJ`Bw0x@HIJmtWOSb)$&=Qs
z*K-imwY4p;ye+S-wQ7IL=gL}eb<M21&!iN;zow$_cm+xes(9F$vr_C)GB+3OLFr)_
zL=zZ9oIPR5DE!dYGSNTk@KddAG_O5Jf9;nB__9Q_d=$fO{m;X$g-;}z$VfDpn1X<V
zUOq8GqPYXUP1X4S_6zwTG(Alfl>&MvU)iz**dQI9?x5t9hDsJaA~q_@){4e{GLOE%
zKu0C*p%Z7?mUj;+#F{%_O<}=HA<|qup<GH~fiz!3Va7=zSDC<54w_JTLa(5Dw~9ph
zuuj4VSQi=Y2o~CXBs8X1uvsJsG7<y@m+wB60}@pc5-1QEjEw{)dNSki3U=!i41{(c
z8<!6WR|^T!bPI%v1j3|p;CnKI^u*{E7#sBuk%&!@eX_C^o0HqKFNt0Ac?U!w&^Yk$
zB1rCGt!29tq%4;Y-gKPJp*u#4wn2%;^qjdUq%kp;iP(dSGH2B~U?;e|H1ez_pJ;!|
z_=-QR&pJjV?|7L<nGsG?{b$EIsc|*Rv<J)amdFs>Cq@usNXFf%Qqu|;=P}q?;C+um
zruy<h7ZVz|{1yb|JxZ*KM}O2LD12!*fAzjD_X4p9O$in5$ZasuLn5tbvkG!<m=me2
z(9GRqgz4DwS(rt^!%Ca8aDz8Db88tfUeE1t%g{`L81bdOv?^)0fY>?Z(0HwUPzw8H
z!cys@haEJzoz7tA`_i#=E~?S8v-zq!)P1v4nvr)PyGOUJt3!zC`!Bk@{fZR|k8t2S
zz8&R*q)5dPR1Wm-DtJEPtWME=x&t|-_y19#n0a2xrm<ZWHyv5b+zp$e$$kHU)8SQP
zO<f8ka64njMTBzr<)raAPU7Rr+vN~zBgmZf4U4{5La`0@3aRt`r1|5p8ZYmA*#9Y?
zyVe4+4M|VOINS|3(pl%5YEd9C9;%`z$IeFRHxkC5D(v|5(XF%aDed6ltf3ZX{}J-k
z>;Ec>B9Wx<!oghW?e?X{+N!1enk<i0sp|Uu_6Y-u)4{-Niu*OcI56yWiuGp7d_BVP
zj(v&-KR-wnH95=o{j_WFuBvvJn8M_5?P{|W83>j64>!g``|h7&yPWSPdK2m$Ml9nw
zQx#zWRx(AQB@Sq-9Fyr5nhi1t^VfsG$G9V+SDJJ~{vWOJ{}Ge^|1puKg9nIm4%BDd
z`6HJ4Y|qN@xgosPWAxmSa%C#q5G*pD++aPGuk?1>k(vAjs(YNvlF4u!yajJ}QCY4E
z7<VH1Lw2y64l%B6bV6m~N3$x)$I1kiTB?Mes*%74SP=4I?h07FsAL4Zu8~_QTl^GN
z{Uuqpcn$Pi?F;#}@gRX}K#=omSLFS_6F2@hge-OGi2^!BxsE&XUk@G+NCv<zu9+a?
z#jh)*=KvqzTzUQEKjA6>MwuOa=RetNx|fDC`0um-EvDTA`G5Ui5X%<-&zL8K`k(R9
zgIM4{0%Qoimy&`h|DT{ZFJ4hRey@=F#;93}-d=}3t(y0{fLA@Tjxg{=3559p@p2H<
zc@r;Xb|Vw&%^#+YP@PZT2c3dIm};0z>4;1-H4V(#q(d(M`l01TJ9$~dPvfZ9U*E-K
z%0Oh&)u5!bGidcI`xFRQl^Ql^QWdcRt`3E%0DgG~G95$2ghFx?7Lx|7+Q@yr%ixX>
z1JU*-;ynnSio70ixGa~q>LW4{dqV2}BoKB1u+|KcW308T63InjnaRyt1&OH97BJpZ
z8T{e?Ty^f%_NCfFDGQYju-8b(#z;+lrFVd4vJ@v&(n%#ao1%+4$KRb2n5LZ5N%QnO
zWOnar_$x+Q6?W-<;oT-zvqow&*4-tV(wQb9m)@(L_LbAyWtEtzZ>NOjMq4T&F4u7H
zWlOz=#osl!1Q|;0B~AMdVhV+@mH~0RQnC+6zaZCkN6-O04C8UYVtVd_btRI;XK9*W
zQRKt|U?!d$Qk%<1rILN8dWgELzxRXAMM>87-~n9z9!Qo}KlE%-vhTGV^U3y^9IO5!
z<IS$pk&1k^nURR^r6oPZilj#T6E<HK-By3;|Ie<ZXQhWMhZ+0&qx+Y@-R&~3!G|KF
zdY^bT)dZ=EL;-F54wD$c7Yxi=7%E_c@xa4-kv*uoSOPo8ob80TBb5K_f9Js|QGiDH
zHz6Og4&XX`oEiz#4h{ejPw(sK?Hio(5%_+uZI%+v=0P=x7qu>!LY067bc%Yl0o(0S
zd1)z_UTwKR_^jzoei@L^_9uX*ll8u_qWO~$gtAEX|K%T5lKp-5i>QB|bQI}T<Bp^*
z^1m|SjR0!I-GAS;hz8T={OfSt53EMurL<JP$F@{m&3BbPe8)X9{A)fS!~@j+PB_@W
zDWCs!!6Vcw&;BooNC52c?}=!tHUC|kBYm&%U)1}9w^aTW!&&l*+g}W~OIG|9Ap~$I
zv;Ub^Kbtyj)=K=3RNgsB%;XNd{S$I96Q~gxe|>9+1v&kOj`cg%5ZCj9S)LP+{uVw9
z6AyUlufZq*u*>HJ)m113`3|4u`+4~DF{P`@wH0Pm6o3F+CI48Hpi;8GV?c6B4+!2;
zSHU~ida6+ue#lvCi{|rfI(bq_w2RqcpC0u+cYd3t{*zK;Zbx-<ZDqZt+h&FV36(AA
zBmBi_UhfH9{Y0DO0$}9lTY5_g3~FD_-#jYCQv0Is(S%$1CgIl~3z$`xF#4v6{}01I
zIZ-rr4Ke;KXDI=s`S{<}8L*)gsQf!v$QmOrGw;doeDd98JSdlO>JQ}?v5Z9ttZ08E
zA!LJ5t0DXHk2SGP-aV3``|`(}8o{bF8PEQexvm04RQ2D9^z1Rt!lM3VnnQJlIOboK
z_$okB)&3p55D!xNYe`8jF!Ibj#DC8$sOtZk5uxzR_>X?$nI*IE_48?Ye(xg5+A;&t
zEBJKKKfL{{kbE%qcx{yY_foojq+OP$S8e_pu`bzP^9%i%7xa&na9+|S2tQGNWAtZM
z&iIwh^<a(cpg)#kAx*#>|70)a{m37IWi}}Lrbz95{P&Y3rZu%oieBGG|1cZrl$LT5
zE~%M=hW%SbGO>7*pEcb-7d(3XXGM{5!CS-odTXDJ9SHx~fULggdUV2GKh7fb#JY8z
zAUZ+l?i#pLP<d$&7t)@{D%Rfpik7F}>`B=zGa5+a_jb47&5p<6;@|n&a57mQ1ta`H
z0xO-~BNZ@3myFd(!WW;kr#g{|hZ>;<jyMeiCdzNGQ^7JA2S%ZwaBzrWAirdIN6-)w
z35a!BxGS}A-?NG=+Oq%1#<{o*)o_vduasRsN3_|h6~W)jhu%v{BQX>2^O~Eyd6~Yy
zU>UmB>Ut(&FLtwoG8KJnT<`EogS42wNg(>p0r0Y{+s|Ka)n)<LHLIy7J{+MI{`V-L
ze;l%+!KuR_ASYfL(1u$nYG|ymygQ3EWK)-Fs0{xvEO;iYuse^*wWFnTfLVeJA2H2H
zsP@4rl3MNU5Szm`Zl5Ll-zN{DNREG{-=a6uX6b!40A?8)kkTM<caB`Tuie<A2c-Ai
z!>E<aQz0+z_R9Dj6gvF1-DQ3aU+p@M`JhsfKZF#~T)xCyTUO7(AF*QL4EHAxUbVQy
zA3p2Et+d~f$$Qc29n-V4<Qp9Yf<T05(UN3m_ZbkP+3ADCt%;KEC!>aSu%@};lU>jG
z(*>NJ8r9f!YjU=qQyZl$_r^<^tw5SqI#mg6>N>z7(kwwEgWky#Nbt&9-o2j}K#$7?
zw$VwY3%MYcVz-OPm1m;2CcNi>ZB6T<p4&G}NsczgchAgb$z%(Pb=%H(6sEugnM>4C
z!6+pA5G1Y5$<5G*^1X>$vfFuvama<qpFgq`Lj1#ze0i7s&)MOOAINmpTS_E3O`M5U
zPIqRuNvR}e=xPmP*cmrtQ)^#nOegxAb%iW**8M({TqP=nWbSmQeI!zQW(W_gh#i(a
z2a$S>Legf2M9keKKl?M^Bx2HGqQMf6?*jJhH}-JOQW^qG)GWFEF#Ow!y3Bi+^(lhl
z!OpQ!u9%F7f6I|BXVfkQat&e&?j3St|5gyQy9UHo0`%T4mmw=HGq=d7vV<MB2QKt6
zEn1#AZwUUfUN#smnMtrrY4Eu|dwi@wjeYK+f20#)rO^wid>fQw%H*Iu#kLyHcAbt=
z%?UdRk@`pWeYGg&VGa*L@r&7fB#1+Fee9IfoB5()eQ}Q<$lxRSbTq8&vO}1@yH~Kt
zt#}t#5IC1UF%m09nHq~!RH_f&7MqJKlB`A3bEhcblyswDx?W{aIgIj?unK*4={OTQ
zD@e3;g!;V4Vk3$g`LboKJF@r3H};#UbZ2UUBxP=y`|2D-8?k~?d&1Zz)>;gBO)HZo
zBm7WBuTNmOmXP(HbnJ50CC0S&sPGx7nFpgD^Kgj`^LhA?Xt@pzKjw{agGynY!6)NH
zngp)VK44Z5yd{PAqP{0jnlO5ls$DziTAX5EQT*7BX<R}mBp)wQeE$GD)dpR5wCQ;S
zl0V`_b-dxf6h_Rm(sd5GEsomy0p@S9RB7cZ2sg#Xf(C6e{WGc<CW=q^l{Sy9l|TAD
zF@)7j<`e6Ciq<6(e+%fF#U(22Yrj?|I`T^kR8g(QKd0KqH30iXL(?-?P8K&tuue`I
zic0$7v&h$4qhWH`^c=WY9{B1{a4xsTE>ssh7!5^s^|gFl-f8G&E`Ks^^jk$*GZiNK
z)#KXZ5XHieo_CZ$!xQI$Z$jE9hdyY%v*bNQxt<x1$OEZ&9jm{9I)Qojo&g!>r#hk)
z`+Z^GUzi->cF0#**XtFGGsa$&&<jdp{A32_B%?cs&Fch=_iH_CFK1{pD!wrF_Co`3
z3Rio1ky~FhJ`U@vtFTE`3~G@7k$6-bs)|_M(igUPOALwkP0v_5XF|ErnmVK+Y5WIa
zQG1!Bdz+wq3rMtctWB^Q>w2wej}UKLDn6T>jYSGD#kzgs0{KG=Z<(CZTgEkEN4hx(
z-q?rQjJ?1w^l9!oD}i1q{M+w>Bc*4Qv`w}?x^CRP(Nrd^tS(6xTVoMVm}dQnCHTTM
z<d%sExZ<blbQ9|h5!D%nph^?yX~@cgAszX{d21u}8xCb5!!9%@ZJzsGl5w&EQN+A~
z*{djz32&6ah+s?WWyzK6ru9<qj^yC4dRBw5!?<DoA;KrEU%w$dF6e^V;xlJBsIYpw
z)Ysj}YckYnTzvFKK(9geb+1;mhnd$<@g^A-wWh%J&cy2%B?g^N3*lS%H{wUbIHOXY
z=9+B1@SsViS~Ojeoe{j_iPOW4GL_7ZZbHF(qsd+ZkInXh7naqb^iQk97o}zv`8V3!
zjd|0#yWO<t=PN^Rrt|`Z(^S)KN-tV4v|?eutzdCI(93B}rB|nHW{GqdX}MW8KZLj}
ztDB<u?7p%l`X#XeSR_E--qy<YlyyktNI4Pd+Agxu^`?7td!k*1gK4=oc86H^`iC-y
z%6CU7#KJUnSD(Dgv~|)U4q?{GxeC}oYPjH(FRn0yuK=WmRi#fACOV82ob?#IKY!sG
z@NS3iN=~SLIPVEo)Lo`Fim|@S3aT)<EaxdJ_kFdwO6RV4b)XVoO+*jP*^kID>~VZ4
zHS)<i)Y=ROZ;}kh8}I1t$frB(kW=V<tr@Gt^LOQtN2OH8>MN^~aUS<C;LFjpLeF++
z5W>)uhs==a$;i+<UO(w?LY)q%P%vJ=!%v=5xiOuV6{LB;O$W!&7DMl{acTpmAQnDK
zdt)YYMoQrk_-!5^7gnXhSjX90u#NNPFG3b;2@yfqH<Dw7G*~Gmf2g&;xUXygV|i3p
z_pEBRBY7_WSm}s%Wc&WjjL$&B``vjte)_Cwt*^=ZtgXMMZZ({|#wMQHd)>Oe`}5Bz
zQKbR8bj~&6h1Cok;$N(#*9Vro^$ZKVy690BUfjw1E1!PnTtoXc!2#x#onxxm=Z#dY
zdk+z7hf{=L{*(tdyvxkck*|D>iQnRi&ZcVf4`zUfiZEs9b|3{h-@aC78=9Uob#H0N
z;kT`!mv?4E5XST2g6wD$;H&5BQy=!6*Dca(kEAH-%wA?2*wlW4+OhgnJq*4~uR^*?
z!o<|T20kEHge$i<gEGz~&YL<kpS$a8g7p?RZWsj9-3i*-udrb01y8r`(;&Jx2&cD&
zugflCl3)s%fk_#Zv4vdhfvLk&Wdra9*6D=32mD5?SdeL|wWIv$WJ;=&z-gk}fJKI*
zR4FXs-FPB3;-vqzOQG_3>I!KpoX-pIi12`zKlyphO+3E~E+CV4@qASgqIoH-FVh$1
z;YII)EU2%b`4NayKxr=MHB1RFndD&{E1s*tytR@$#6H|{{Dn<m1I~S7!|KUlNZ$=5
zFxbEsLm8$~{ve2D*A>H^xH7mGvP~m{k7kojhv|dToBfLVeGhHIvW1L0**H&d;r+}T
zFWw+NB2VK4xvTDS^p5&%aed*A0+bmV``AH&IBofLAc;6vbF1C8E~PsOJ+PHo(aF{P
zry+_5+p=|NLY7gsFE(k>C);^%UB~TF7Kg55NtKiM>gHjx7z40km2KHa@=wRgLN~f7
z$F$L_?I2;c*+(C!Vw$hxAUe_>qMv``aqC*wHpEB9{ONZ{L+DTJTLd%kjz0Ke-5ckX
zt-0?O)N@vKqmx+7Gi1CtF3J93NnwsD#!y%)&-T3|#nfZx>(lsapxlkodq*2uO<*=r
zDjEuF2xB+037Ltl@+(=s+1Eyy9fY$Ea?MZ@SJRbJLJA@3H&S%NtPROdEj4PuHJfk^
z0D|hB?Z|lf8>0tmKBOKmQGjf_zSDu2pJ6zk69ZGaTanr2F@r&$?JwJDkAIT>H+QE_
z^wJC$&6=|%rc-BGRKP<8bG10j=^IK=k+%6-w$$bBoqN{`0E$tlk7_>zpFI89*&|s+
zQZ5@sgLo;T(m#A>kwB_%$w#DKtI$`3_t5%J8piDKxzXDqB8)HLGplldux<yg2S42M
z!Y99{bMljXVpCy7$BevRV8x%x3*YnE$-a2Sk0({B^wsPo!lr0K(XCUn_nVs^WY&1(
zF?25ay!v}^vVu$9z{E$_v}lgg%D<9+TVx{Le#*+$nvHA6eMrRjbT)rI$}yCe+}-@-
zp;iFw{21R_AQi{H)Q*U9`Ja3rTg3zLk;nXkxdOjdNeCa2fQS+5LoT=Mxhm(<?0g)b
zHI_S}c5vB7_gBAqnK<}kGS8CEiYCs=%nC%t^YPnFk=66+oF0w9cLLJ>`tP&e1a%Zb
zs5yEvdk$;LkMt|zs2wP#jTGXx8wUv0!(C@fRHMIfBbh?;Ju&TV*LtTPbi@pVi6OXq
zMR~v8&-~%)O#(VpxCqiuj=GpdGC=6Ov885<2n0ho8Edh?^;1f_4W)Yzyto(H)=(%~
z;&l3t8RPC`xI6px1O$u^4e}Z3J+}i&yCGIr`v!A@Mkc*e29WqW<YcM?{2JuZ+bKXD
zN4F$j)eoElrttAi93V|j`&X&4Rg&o{Ve=P{=_F>LXTH`a_bVo!jQ>hCtQUGiRT*rC
z%nJp-w>Lg!t%bMy1>7}}lM!Hk^S&x+83x_jYuSmD;t4>slMvqQ^B+Y~4|F}UaO<*Z
zWa7*Q(tzTdoQJ5hQFEW&F=2Ztah-_Bvum#d?s3kEr>r^b1r~yHNG8W)|9iS$m*lTc
z+^T!$I&7&mU~bkR4j!ne7+=Qq*0D2!3OAd)L)w<#<jbA-T!efvAx@c5`NtlM;Em%S
zp2~jrqE4r8aT-G=R@$5{b1={2(w+V;fPH=O)iI8I%3S+SQtHtT=Z6*lsfw1iJ6Gl9
zw8uq;(s<13c!wfs?sfJz2lrAT2&I}ku~)`n{&jsIG?oD&C`mD1Z6P^nIgk?%R0waP
zMINEngX-sEBk*9MDW2k2*9B{x6quK$75pNlm^Fpoi|tL<4GARBaDM*r?01W#A@f35
z1zsJq=o(rY3W?3l8UKhtEAhC_U@UU8^<GW;i|7r)BlR@zuc-N=(E=J~MZ(v0i%5eR
z6bVtgPh7JVHifL2t*sMdt49?HW6a(tQ`vJgBb8u^kL?Fnp;Bx3F;BBaA4ctdxDA{x
z_+482Qbbb2(!HtjV{fOC+Vh5F&B>Py8-!ZzQVs!eFrZ4#Y%t?a?^oZ`-P61@39dhE
zYjr($Z2D!e{Wd{DI~T{*SEC8S>TV=VwwcrnntsM*==PH=NTk!$8*L*smbu97h4=hk
zyH>7NI8hErQ`;glf4S`P`zM>RNOHF2yQ;s+txa;L*_L+0##~e1R@w$mjz8iV?zq73
z!#O)D;WdZCc7wKJ?iXL!2r#D40hOteWOP0LhDJ+u)jNtK?(g^MzM|ExeopSGhFhFl
zwg?Q@*X&6+z9%=btuc)=6}R5)aa?!tb;_!7=Uu*}HCQk_6*TemAi}Io$*x(5cpP-w
zyc8LH(BnprK-nmj7f=59(IH~g-uOMtgo#-*-eNK=B-uK8aE+g3Rb+f|d{Vpn7SwKF
z9}nv6;Pdt^s<)vLi(7{=WFM4fxFDS*#>JUV4pcSESs`$qiQ^9MulqF|u>`|K6?W)C
zrwCo+b*pimcJC~M8SJBpY=_HUz3E_Ry0InYIwpOA!rKX#o$uyjfkTbM+jV1Siz}?Y
zz5`1}q`h~l@M;KrQ+u%i1&+tSrng><?C`=R%<8*GM$JkBea~ZGIIgz>HM6nMvf_j3
z1LLMd)g`kn-s*Y}`=M%`VbO`0ve6Tj?|31b<7{hUEwR$%Ui&6eI4y81l92sF?L-JX
zm?h#wv@9Z9#@rhy=-DWN{O$BZiVZq>xT}D`jBlK5zv1?)#8vlwASoxRklESYPs_a!
zRE%59d3ZqrWV(81&qQ0oc!`!c-Bg;g4P7iYp6$}_i(PWLkFC4x+NV<tmjvFDENlyL
zC+zKy-@FB=Q5fB)kh-HFlUcBSV&Nn}?cU%FYF^ceyE~^_VyTFmLhX8Hb@$TVtis_Z
z;vxbk%Z!ILf3N^df$qtO&-)NB1{Vr*rhzj`vgwCg91l<p>*7Z8PED5J7>CC%5+N1A
zJG3r}$wqj+BqT(PfHiK)%=9%*lcxO?82XrL5)D6$b%jUudHGVAc+;~ObR{a-5-g7?
z7iveNc<K8I8l)Hc?b*)j@Uz>EOm2rB6jkdLXixrtuo!O~w)Y>Xo|s6ZoWE#D)ezpV
zGsVW$8;)D)y>keGU5pxl4g4>9Gnt+D2ia|Z%0P$91m+_jur#<T7$;Io3wySq`$nZ6
zhKDTDFB!_V)eIj)SMA$G=-X}QLT8oc$x@b+nq->Ib{QXpT-GvB9`k#{#mV9*B$LTp
zk>GI_K_2vQvBWDaF@T2kOjez~v5)sB)F@3C1vSIjQ;mEZ6K0`urab)6eu}*Odq?Xt
zm5yh}BYSqO^M=kzo`oS*dc56oCpD^hb_reKkuK|9Io5RV#yq}-^VVG1E-GYQGxoBp
zZ}ItVGUNFp=Yuu><&BtuHd4x$YQ~n!Bi5QH9rxWr(3~>M>;mO9V&dbPn3kR6?)x~c
zOw}Z<IgM7*crI|7plh=r0#q}wls%Yxxk=z)K;`6Udi|l=S(0x4HlF<mUlQwnX*CmV
zf8k@j;csUw#pQ_|BZM>^yb_c(E|RG-AywA+%TCh+sZAqc8zIRfevbAfdPe8kX(eB|
zEIpD};Om@In4nXJ6VT<+_@`Jw7excF^G22S<HZRce={5ZUFzP*mIj{rXIY1yd~r}-
zl?Sa;KTfu1B<QCJiugqRx98Md!*`YDJ^+c~d4^;<?GHwg=r*V(GLU@xr?Da$<}60L
z%1hsDoKV=OVCQ#^y(b@KDajvogAy0{DTYpfY6gOL*BMp3mMT+T7KQf`0pUE2)!|FV
z+;_%gs(Syl7>|XpJgn2*P4wuv)rt!@uT4AgO*bqI-A+QSOV-{|i0HYsWb*LW={!8Z
zu_K8+7Q2_m$uuuNRVz5c<)%)DAB+)SKK5$ZfmRmyhpl>L;2urfj?^uJxxNW1K<882
zKI!Km4HrX#oD;Z08XA69emyt)GJcTS>>3)Ltm`blF&jmFk&jhw^BuHqEnsnW+Z5Rf
z6<>dAofioZ+Zf1yq~9jS`_t(`S(Z*H(&fjWBh~xnSf|09tS>#XzR_XYlyAWAz5Eoz
zC~0wgdUa-sB&xi4zCIv8(}lfntszox(}<zlv5?f1$?aJ1vKO?@OOQ-#^tFU{Of?Ya
zd=rS(1ZEI9yV5iAjd?z@go!LO%-VgLv=5xVm>liJszDTpDpv4;(pzvr4b60IbJaq(
zb9b4`h*#tyoO-9F)n*j5p4&r9n({qDn(;J6!6mfEH&K74`Vs0Q?))FXy?43QB2eMH
z*T4_pVR|2Bw4Bblv|JqQ9I>C0_Dz}5Yq5&Dd1r}FQFpvwXnyuZtN;EmO6`6k<37vi
zS?Pn`<oYxsce?7A(axra255=8Ggeq!ybX#aW9al_cNP~Z-BS34%}qu~FhhQ1RvKix
zc6M@-2W=hVmF8tRp4U#$6`g*t?J{^*;BJ)YDBuiaTmtzj3N!f$8eiz0%E}cCisA@D
z->LBL3VH^zji@kjC&9AC1q(W38?CnuX}oCXBXn=$CLc#z8x5a3Vr`__G~AvNI+vXv
zKIQbDcFK=-_GXH50FyDFk&Y}-H_c>mOk`QHDKDGd!Up6TJqcos9|(r}YBIkhu9*~S
zI;)v(WSGCG-#U%WK?e4C2HY}#o^;`Fg;WB~`|1fUK(sI|gN++EQJjt2%v3xUO#kYK
z(}K<XXgIG`233!#?f@MU8s)~I97xPL@@2^?EiCoxbT<kRGxLa!hDYMdPx(X2P6jr7
z)|^@jhC9m$8!VFPvnB<BD?G7tz9u94I#7ZQnW9VI>33{&fWJXd0|`OpCQFE~5gpTH
z$z;XShx<|=VkS-@FnU*BlG_cDel;U4=|PG<3^RCu-z4BxdbRACc;~%zT9=QYLgWP0
zooc;_KI2bOF^Kd<_VK1t0kwYh>FLkhH{pz!!~+<k2bYtMlw(ZP3KqeXK4GDR`*m1h
zy`+3g3d(KCBXZZ9G3(qWQLjh_axY6JL~L@N*Lt(q15~1s`&_4L`h{GACB>S3R7Fl!
z1+*up6{{7eILfEXuPL!Xr&&-~YAcJ7zrVFKWyYh(23FK5(i#ygrlL{SFn;+(DwFE&
zSZv5(Nm;~7lIv=@SB`Km0e;wR$a%lkj5k4{tllcO^k$%Hw`H7j5;u)<^HwmKq8JJb
zM{$ISNaWcfm0OZ83fS&;2McZW!>-M)LPeUW9X7xvTY)%)@9eU%6_U((Aif)#MKDyo
zaWDPkBDsg9UWopTaQYrPB^f7qzPcESfw9_Wlk)=m)5?i@5OC?c-lH(Ayx39_%KTLb
z3s^D~Qz(jjYTeEG$-2F3zV%}q=0%wlr`plaGd<g?2a1pPR)<hZ2cg=}A<iUZeG3?9
z6HH9&D^smuqaQR3`QeRXMwhfG1(X+YlEcgX-2NI}7tu;ClVvC0j}FWUzL0l`@&bj5
zo$b4vB|MN125mlfjwnO<b`$iFA&G&+K>`;mwO;hS@$rYVQ?$5zoI|4lAC%Da{9|4+
z8@c8o8r!|08gG;uEMg=g*1pRG36BZu3;mkvo}r&)LA=%<nWC_>%tVyn!Cz-;ip`iC
z-%+M))fsqCkmCy@E{b|Gv*#1-?QLA1O@6AxAxw=(mnG5F*TB<t72B{GQ!C9`UcO+g
zui~bHgZ2F%%N+e2`#1t-;anfB;3aH{SNa*L+`?6d4dNwOTo$Zg&f8RQ3Cu$lo1w3)
zxwJYzEsXv^%pU8(J>L+b|4}|1^GrIgu17YY^rMz*Sxl+(S@Qw+IsECtr{Z#Tl|lVH
zyM9p~d{IiH?)`eY;iNrWph(3*AmAUVywx>yQsXzaCxAAg=Ogdpq&iPYCM4zpDRp<C
zb}%7g8_E?liOx%=tUD7u<vi#yYbmphxtM})WDP#rWJtFWHY%^^nsiKscXso$j~1Cq
z%lDGo7sayhumltA=QuigJA0AyfA=x;G-dJ}727x*JfXGjihT<cZASfJCu}@g{5a(>
z_IQxI%MdLKgA{DXV@Gdw@?x&KTar>%@DAASry0BLfrx0Knm><Y_0L?gc;|@KHG^r|
z@?0<L3SW&2qk;)TR=IOoePYVDUt^AOm&b)etKy-AP*u*;F8-zcZRw`6ADu3_cG`LC
zays$8nf?UC9y!{?c5!X6?o~?p)f1Uzw^E*q_7}py<j=Yt?z!)g^1u1g41t9{Jl8Ld
z@;-~>rCT<<RPf2PFmC$RPg1k1u3D@~hT^%q6PLnOXfxfrACsk3GkK^Vsmwc-3_ZU+
z$xAoY(|exp7pJo&nmQZ~RkhbP!O*i6en24;d}rdNT8^Q+{X^(ZQRRkW>OBxIB`f}U
zdN9p7Pp{>U0P;In=%;<%aq<ieW7CO+(9Fy6yVFh|90N9n2P=16SMXgbwJU}`gumY`
zCmXM*Jo7R!e`OmT8d~tqJYn38M9628TSJr2^8_=v*DHK*U-ur_QpBdhE*58v_WnTy
zi+}rSli18sx*x9c{Au{&GJaR2$6d6fj>PE0;6Yt?MUQ^BmCA$0l`uh4*LA}^+g-oG
zeabmYLxCSU{%6>E>SXDPNgk4WLHo7s=eHVbwm(C6&&^umJ>(dMtR|-2jq~C0ruThB
zck>H7S}1oO@;o)P*>o<Jb3vO3$=i6kN#+{oOx%mq;QIt9Z4>3?_WH9;=d@LAe`wgB
za^f7H++ooi_TO>~%4nIn=vi$g>9-r+`hHOOYq>>jypw;<(~(RtwB*}j;3cJ87{QX{
zrde>H{OxdE*~HW%<zXeIuY@5R%@<=Uc<HlSBF^54@uHWB((EX>mHeJT+19RBeNosh
z;)sJe9x8_@U8k^o&vvziy%9Ugyee-c3AFX~0-m+Onl7GY7P%mo{#T+u%=M;Z!$S>s
z##vI=qX`JN{VaEt6XW=PK+5e+A?Moc)*kL(IUFH=eK-YBY6oYP$>fzUA3r95F5hHx
zVSyEmlV}f1q?tVU;`jE0H$p}xUc1#a7<DmC#ANgh(=Ev4YvtzPBvicByj*><o={#B
z7&1YK>4Rt`JQzp}{p#yHPl{tz)cG!Wg2@pF3z8U9DDb>BbrU@t*EQE|Y>XI}4?;v)
z^|{N#yMTz!fPY$rDD`*?fBR93W9pQjYy4JY8t;$9pGpZz(Y1S%4OW^%)r9K^aVx9U
zS@Uiu6VwaeHFX4Fkmu|QFVQAyCsIz>JR}&b;5il8KxNjq?>O~-!|DFv4~oylZmY;C
zwrW{QJn9Yb;ShstlC}}N6hFU)`nlD2p@L-j+!^Py3*Ef+cK63M?uayaNsFf2iM0Ft
zm?-Hr$aylel(Bb*X1wwXFW%(CV>H~O$JVf8T8gzwCC(x1YUnCA$-y&ER+IBO+xSfW
z+1Zfb7OP{Z4`954`3(h6X3Wj8d*8Juc6SOs&%%nrL4r^c?ubAOC+{r3iVm>J=6E!r
z%5m4Dt3Aux=XI*|jBLncdf~&PuHzEdI=<L_<c?6ENnw}haXSI9otsq+L0y#AJ?XvA
z8_1;&5V`lMBHXKR%DuG=#U2+jM>Ok_M|pl<3^}?l0KJsB>1FvF%e%FYu8P`?18`#J
zT^?nz+VxQT;O=aPX&{*@h3)gg%bhpMw{H21L_u>of=_y3#1iWxlM!y0Blg`NsvVEL
zUydpnOn1G3bt(9Y?K~Q(1f?(v*WZU!bU$Re$6urT#k+>ecG*X^5!b`-=W9`6$HhnU
z9d0U_zDaNC5djOFvn36aq0pQuWQm!Urw?oGekx1^v4{i<w6o6rQxhc>Q6_0G(VaOp
zA>x|!Rr6{j(`hCx5_Aukg&vP-+^)lvn2~M+o>fbgCC}S(rm$ZdT!B5s`v=m@n4@dH
zvq9wHxse>=drlOA0M`%s^rVYS!CNUgUXFtxWYp=1Y7XitUKyM;D&$*&!0q^AZPxK_
znFy<U`cofgN%xdGAl1{cDl6gTYsy*vR{EB~VH7Wfd0=0+>nydF`3(%Wk{Y-&KgA33
zB8iTXJaz(Q0z`fTC*J=csq;SU?$UtdjzODv{p)c|;s_09GLn7z-|`8}Vm8!_9Mv2=
zYzcE(Ju70rp(N-jDV;47oGFL#QYHYFe(Yl{kR`?~2~W-kiP|MqWnQL0=aKjG$uAIT
z_8UGw+(`gb6IP^($K)|_Z%MVtQ&}vX_po!%4Z@RSACI;7d4znuQx-O#VY8Tdn&u<7
zgur$|0CuCBL6UbRhM`xRWBe9F1hvDhaD8bO0BNDgwM{;tA@KZ*Wn$70ipd9!Au6*M
zY1@<j$S5rRrCksZpGtu`aEq2^xd;?uiE@H<weu{))@koW0`Tj@Lv)^GNsD2BX|BfU
ztt};9Gqj{u6IKtF`FEBLZ>*SoVrd1&N4%v*v~NM$7A}(CHtMc(Ci({Z#fx$EPfyVm
z90SGL>4or-L0~ung<?BSPTqY~v0)S4|3TPw1=bIxK6zy@H&|q+qQdJBGCS@-Fgw@j
zb6;Cdf3=a#V8^^OrJlg-zKk)$bk9;D_WhCU{kYrMeYCoz&j6e*S={Dt&dmTY?Xc{<
zY1U`};<fwWCtZREOR`#uUuUL6gC#Xkg%ui78h5T0^H<V#BgoxS3R(QY;@^YK_o`fq
zXyQl5-H9JofG86KQiD!fC~kMk`~$>rm5N0_^C+I|omBxz7?CLxULoEk^kq0IWi=}A
znxfUdJ>-OcjB)F<|E2Y)6D-2Oh^YB%qzsPgjLg?}B@Y($S7v>D@(sA(>z;FVC!YG!
z2!8+@9Cq-WKN&L5K1NhLSIM;kP>0{HkfE#hUSOQLooxD0N&Q#!cl6U9d`QE}e$N(a
z(bnM>EbLqDA&KWDGwz7ADN3xeM?d@MUB3PoDRtC;<1b>-Yd0Z*)eO1a?^Y{%kbqf=
zub{E39%FYsN=hBv%1<fyGdbw1A^Uwfx$eNV&<e)Hom$(YapgwNLi&*cl!D8-n!+&X
zK3~-(Y0fP56>70jGXS=ngI=c(U`ETv9#g}+Q8^DXClFVc25z$q5;GqL-cGTg6c|G5
zyDMLfe%gQ^r-qzwmraH-#<{+_cRlPyIyK@6_m>I+$Yb+K_in-Ji!HIqHJ7>_zVh)G
zdCM~M%$_*yFBr~u7<j~|AJuRZU)??FXS0F3vwKwGK*{EiL*BNKn6hi+Gk-<WeNlLe
z^kVY)wypJ^*X6k#-}N;Tu@x4iP?JF51k~W(j9-_TylQ>=5sz=)<)_Unxcv2{G#SG;
zuV{`Fkbe!Ypt677mvX@tdVjmt%&V(EzWV}rlR(|y6y?GPdR$RE+#Rx6wzcxMGVn(W
zTC#v|YrmU`_nvm~5+{5f+2YO>gOhyGD-_(34<0$3$K(x$25z3WJ?|VAy?1#~x|EXY
zRJ84$s^FDz;q>BcQT4iM#DS&v$DZynbs9(%PjGgtTw-qERPbD9NMQLeabL>T*-pOB
z1m(J`#3^{89_14)uz8PEf8V#ub>FKFFQe=t?vo>nrM;#Vp|oeu9huIExO`h*5yxWP
zG_uM<CUMnX5@;Y9o)dgu0RlJO_@!G|`+`_v24(2gK2D9zoonJO3<-&Tk3}U_XGB7}
zIOz|ICMntIRo%&#%Gs}hqgo<622xUVQ6PKaki}1n85hi~$kAfgohz=>-n1z7VNg+z
z@EclG4j8R$Zn1}Lzgu0#bH?Kre<CPIZaQ9!1PLZ7$+*G4-+gFiEnu(z011K0kD1?4
zM6=;w4cv5}Vn2;9v^V9*<Nl3Zy)EY^U+?E?Ob@_kuhFhNpJMk8--let{dv#wN|519
zn5884^7q_5W88tz$*i;8u)r;?s28Tc8+CBll<F29b6nGXxefBS8B9h!eSgEgPAzPh
z^kggXvZU~41z$2S&r&r)$T!{9?5>~xm4?DX+St{r?1Qs)`&xYEl#ulbexs41s#Wd>
zCdf>Do1bDyZ9l+$^+&G*to`7(K`_f;X+99Kh;xH$@~?_0IV8}2l4(ksw!^Ip-KhzX
zrZtK4>o4$Iyf#05WQyuat}nJ8QWpQ<RvPuWZJ=p>q5r~7;5c$Rwa}IidHg#GA8x(q
zj|1o1ZeM%39nN13q1yHB<9M^VjWtq5I$|-wxE=UxM2BHvT=+-i^FPOzlkMFA)uQ}a
z=Vw^r0b%;bI)aD41pxu|5cpPH(rev_=@$rvf*QO<!u0!VPmm@>*GOKwNw`qC!-1gs
zcbXg90<LbD=65Po6n{T92WDAq8bOl1(UJrv01lA=s`n*gD!&I5g9`+I)VjI`<^2}0
z8G3C;^`VfC#y<n<zwfyRW)%OPfrt*lV7@+`Ln-&+w~nG<x?3FJ*kk`2`Rjzh!c#9-
zAOQ`-9U;GmyoaT>OeIeU&(vt<T=!T7#?m0BnfwnuWUYSB!vVf1uUNE~X|KOFyues$
z2i~SB6Wz^==!M^`(V^9f_%pO9ju$-}BKs(N4YQVz!}TcOKZ<;B(DCQ}5lxE|QFhSp
zsb9l1$@{FTqyC|h$LygB0IaXN#D39$-N-5bPy4|B=YEbhw?C;Fk8J|x%{?vgb#qTF
z=5O3-OopuUpPT3ZXoL7f-*p3Vy6J)d+8hkJy%R*$*LR(Gt23)W|Jn0&-4o`j;EZy0
zcmKO}q1_H5vwJ+`5|-c5tCj$?5&bW^!5oIUqQ~;FZF)=bvs}Mt4|EKGc`h`c4N4aL
zD$ho*)$V^ZAj!h4Lf~!<{Rf?WDgi#8%GJN>Q~=d?BmilQe!u*`cf8lGaLEDgo`?G7
z#cqoaa}l83S)dl1k;i}^JOc6!oT0%2%OUsYf5Cs^{|ETbvviEjN=0!(fvc3^dfrJH
z{s?LW8wHuHg9znv!V<>cNKK#>>y&=zH=fh^ze&Sl=Nh;<%H8$Ptt|j>yy}}{ojw2Q
z<z-oMhB3?qw2)kNVqHOQ6Js}sv?#k<hu758k4&9kOijSaGCZ%LO`t)z_-~yJAIu6%
zFYJETm=E>2kWc5rxyGaX?&te29bcE&Dt$NT3&_p^4F9TiOy3CodiW~}em&qW<r?4s
zY-#_|Ui07j5C3Po`==}D50Nh@XnKa~87mLg)Zg1==c*7_87R!ys;lEyf3bY8bJ$IA
zuByK{l_gL6{IJd<6o&K)fEFQ1-b1_d<n90Ni%&Lq9_t~EuOmwc%djFE$5GbjoEbe`
z4bb9a2v}EnR?6*-F?-u~8g*|I>G9`3EXNIH3RH~!1HYCf>l7Dx;HfwBu~vZf?N3zC
zB;uvr3GF3uuHrMaWqs7e?XAwPcP~x(eTpS#dVRiT7Ls%@1SI1zl>b~nhp6tX)#9=w
z?j9c**Bu;BOcd@1nR-guP3dxbH4C4q>@JW7Gg;gkrM_;f6xIjYD%m&kxY`pRRPVlp
z-7v|2=9L@9A3cp1!=yA{($zjh;iIlroP=T;_@~<vi=RP(Gu<6NRpodyA&2|6X5Olk
z2-PF)`V?JPi*tdC7iz^Dk?$(7+8azaES?!#1l}Ab7(T2^)y;#l76v`3P3_2vMiMSy
zr&4sCsde_Q&()?su!cU1tz)08G;hd5UEWm#_h3U#cj>M=ASs*_lVs77qBNJ^F})z7
zB+h5Dzk+X5<}+o@3Z^tXAGS8+SQp=E$A)13?4)>u;`6eN-;ti$;h8})aq%0e$)mZ4
zX2)CKJ<rTb2x?YL`I!reg^UzV5zkP1OC)B5>c1xd2L<ZoAMrPGXuJK5V%>Q1-h0K^
z*1e^(4k@pX#TzjSla=p9<bm<^?S_N765YX^HzJQf@!@+#Tx;Gf$<SGCaBO2Y#oOEA
zq_d{!%VweCUr&xH^t0CfOX4v1y!b!Q;0jy;H<UI^31gUAJtC&b%b03*jSag!d&a$O
zT0!B9aNT~OT`9tOPG0=R7`g&3&UO~JkDr=W0j-=SoRqt{G*M>uLY>}12d+5<#CENq
zo(0vsRjmsrVC|XiJN!k=h25CcC1z-;NgEGE@wKO5n8tOZJ<BFn&)u0ZY|a`lyuxOb
zoSt7{GhSFUNB%Jb;!VPzz(n=#wV%V6IUfMyUFH{;O!c?inEq`BePMAWubeC6j_W)g
zc6HnfGCg)Qc&Rt3(KDw}Qux$#l%VOTq-9zP+4Jn6N)IsE-@Ty@L=|kZ7}x52p0n9V
zpf|>5qMcw?XHL5;eA{hyGT(aQ>fux;k%bnyqtM}<@4JgKz)@m<F6X?4s9TL|B=v3-
zC~FC@Bl)52yO8t<vw2O6EmvX8f_Ngz5H@`6gCCmd-uq;S(vx%6JE@VBLb~AFNJmH@
zWZV~2Y)E*w|F6Bs%?N^CzD-6gB1eV{M&TX}D>a}AJ~r|OgL3Jp5%<dNfn(+P16(cU
z{q<qp^KA*E&tS9(p(i(gDlzre2PU<<V!eio%yfO1(QJsH7?r7M4Z|cpHPb#eJCpUU
zIA4#jH(A#$8Q!x9vpzSP7alM6Tts%a5IjGQidmZTYMB#$R0kXRHgeL;uv}EWN!h#K
z7b&x`Gu}_rr{vU`w!`Ko8N+?ZXM)oijsqF&dpc8y9n!TtVVTbhPBCZVW<>P+f3O`}
zGO3e=z+aN-ih3<lL*sM0_H$mjPj$ZFO;EKpbSrp`0G2?AwIyzE)2bC7m_h+%FmaFX
z%pV=mXlU5=w<d#+#-2x#pTn|P-Ct}x2WaU+Q+7q_EoEYN1$hwrhlL$4YT|f1EkGpf
zA2(lOmrn(59_PMZ0jOf3)JK*1eNQS7irO@`8c8S<adwXOCe0e|&npP8VLvd529=C=
zHIIj=7IKfYJ#P`wJXU@s+JSA(FtA$qIOmCk(I6MD$D*RWd8sV$Uo;x7{x85r^~n2*
zc=ZyiE;p^Qy_6Obw|MT|M<#{5j0n84r8%dT3-ifUuT<zI<LP~vXdwx)k}kHo3M2sb
z&5{an9zwUK99YutqB_%!?Om>rcN_p6PPpq4QKr>0*MKg{&ZFBbgCw@vGj^b%`umx|
zjaNSsSTvqlyFr^$Y_)yBC-%u;ifBG@o6351Tt2zcJ;BUu5Eewe|H%yIr%-Z$HakE&
zbh37j#=+*Z9$HeFP?Hve6wv&nLy9)>orP`O9K)xkDFt1O+F`FM7%Hrh3$$Df_5qJ;
zw+`%!fc@@)WI|YVOUShf5=O4X2(Ay_O!eR`E+4vI#Y+HL{t;{c3f7x;Qn>NapYHOM
z7O>C0Cr0Tfr{Xs6%DvCnque13Iv)lD#b;i!2W*z+#g1Ggs(}MxR(PNaPKJH*ktuir
z=U~#}*=GwX2Yp@X@t<f9?t#PZV(OMaTc$OFvoM?>Fe_O!AIwfqyHy?iOTNI(5&M(c
zV43Erc?3(5pNL7aq2As=Rm<XCa2W1jYjOIw(ABb#x_l}ju!fxYOjU5p9eZ+I{{WYa
zvZ)$a!x_CF+)D@?hR<7^bl7Ft{Cz_yW~WKQsEigo1V55UL{8zd;L-Qrps1mo7QFyv
zxC2hV9yT5d6{dXr8iv=H{RAhhtM13h{?5F_%+u}lw1lD(q+*PJx9B9oS{Fm)Usbme
zU~o4Ei!s*7<Bj`}mq?IXL2S79UZElXfB)hVhw=;IGvNr8Atc~OR!UK_K-}o{{{i4V
BifI4<

diff --git a/16-bayes-mcmc_files/figure-html/unnamed-chunk-26-1.png b/16-bayes-mcmc_files/figure-html/unnamed-chunk-26-1.png
index 68dfd8ae2a67b47a648e46355ba441e91643a04f..0f915478574a2cedab826ddbf9c2c313bf66dc5c 100644
GIT binary patch
literal 15940
zcmch8by$?&w=WJg(w!pG(k0R%jS7f#BaL)7I4UIwN;lHdokNU*0y2Pzbd4aLO80$d
zME#!M@0@e*d7k^6KX{mT-o4jed+ojYGZ7Eel<;wAaL~}u@Ku!MAEKe5!_d$`GGKJz
z8_$7@55N~_5AJIz0FPI$TmgYV=;-Je7#NtCm{?d?*x1-$F!<`#t2j6~xVX4@czF2u
z_yhz5goK1dL`1~I#3Upnq@<)|WMt&z<P;PXl$4ZIR8-W|)HF0Sw6wI>u3e*}qq~0n
zIz2u8jT<)@7#J8C8JU=vn3<VbSXfwDS=rdwZr;4f&d$!k!NJMN$;HLR&CSij!^6wV
z%g4vZ&(AL)ARs6xc<a_JAt9mLw{Hsz3yX+|h>D7eiHV7ei%UpINJ>gdNl8gdOUuZ}
z+_`f{R#sL{PEKB4UO_=YQBhGzNl95*Sw%(V?%lih?%lh8|GuiKs+yXby1M#<2M;tf
zG#);DsHv%`rKP2<t*xV@qpPc{r>Cc{uWw*rU}$J)WMpJ)Y;0m;Vrpt?W@h&2(Iazn
za|;U#OG`^DD=TYjYa1IITU%Q@JG;k^AKTm8J2*HvIyyQzIXOE!ySTWxy1Kf#xw*T$
zdw6(wdU|?!d3k$#`}p|y`uh6$`T6_%2LuEJ1_lNN1qBBOKY8*50)aez`t;eeXV0HM
z4+#kg4Gj$o3kweqkBEqP@!~~fWMouS)XSGIqobo^Vq#vsdiDDC>)6=XxVX6Z`1pi`
zgv7+eq@<+e<m8l;l+@JJw6wJJ^z@93jLgi;tgNgzZ{B2QXXoVP<mTq)<>lq)=NA+d
z6c!e~efzeksHnKOxTK_{w6wIWtgO7eyrQC_va+(Os;auWx~8V4wzjseuCBhmzM-L^
zv9YnKsj0cS8488Aw6wIgwzjpkwYRs!V6b=Z-o1bSzN4d~v$ONVhYue=e(dV%>hA9D
z>FN3O=~HiSFB}f<>+9?9?;jW#`26|v;NakwFJFd+hQ5CNIy^i)GBPqcIyyEsHa<Q+
zF)=YYIXN{oH9b8&Gcz+gJBvUdzJ2>PH#avwKmYyv_l1Rp#l^*?rKRQN<&~9{)z#It
zwYBy2^^J{<&CSiNt*!0t?H@mW{QUWIXJ==3cXw}ZZ-0OP;Nal!@bKv9==k{f<mBY^
z^z`iP?EL)v;^Jbi(l`*%fvYac`tE3GbYaN9D=`Z{tkBSIpsC2;(eg>($nZ5IAD%vM
ze_OW~=y)<Q(cnDGGtxGFVA_rmdC%BxS`BQL*Ud;mE`NpmfehC}h1Vl<PaTH~bPUNO
z4HR5-<BPS46^L9Z>lj6zzj`Gjn|Rch;U6b>z$)my2p5HCd_OvU-Bcs#zS(y)CB7zV
ziH`OVm&F(j?THu>)e|yw$Q5P&|LqT&0-%QB5hua8DU(SD*2boKudtNwi4(U4-@UA>
z<F=0Vb`XlQ9~~Z{xjPe*d@^-U5uwvt42*W7{!-dgP1DD`Rcp|P;apRiVI50r5C1-4
z$kqGcCr{_p5&>iw0=s_0g@{N5i%1UyiMs;oLId5q;UXY|ppr4WilK%DzJ?Cwg1~yv
zAQos479tVq8!mV^DltH8=orygFcM`DmQP^(PhceAbiCh0vcF=^xXNiS)Q0#bIAwSL
z*m5Bz8bilOq@vW5Tj0JMf$N}3X8xRLXiajc{5fn@#w@}P;;A*06W<6E?7Y=`Zq|~#
zm2z9PrL7}h(g+J<A6T~lb40^8^5F!ig0w@x)RdO<lXNVpQl#yKj2R)3p1GbGcS6I>
zd~TPz#`kn;l#$|vLP;@E6d;HsK>QJ;4oSQoATHqD*j+>WGc%cn8w2DIaNPm2FKSU>
zcZ4WK3T?WiB!3az5anNruHhI+KF}E>0<T{I-NVBz&HqV}UyE5zwl@6@hri8z!knS=
zO<c&^Mz-Mb#Xc#$#06nao{zD838En^YB}9BsYMYFeJ7(@BFr%wwDaT;I8{A#<>cBK
zL*<km=fmd0rz&^>OwQxhxM;?<L?U+=996%oVhLX@kVCxAFg3UFR`d6_^6FX=HdmXe
zKExzQa^RqBhw*bbtDj+JGMjjX^>(iF?CAKnje*|rS7Kj7zlTSUj@rkD1w!AvNxbDA
z*Rp_f92ME7xd=*pp_n5NQR35$ESdhsRaI;8aDI+uyrE6!J`UwCDg9W3(Qr6x+%-f9
znQO%fzNMAO%eHuU0{+o*Vh%YBZ2T8l)E9>PmU#_%xN2Kl_RPVHpp%{9+V!pT&I@cI
z4!ZOx;SL?huDu1M>k1K$B9bd4Ahwvs77Gm*Pj4>^rkKksc60r*!ytI05*y6L&s`b)
z$uORTa;&_`diPE7s)J*H5LW%!RE1h`dVulrb@yMx&Lt{CfpHHh_0UmQ>WxCQ%#BJh
z<YxlV|Bt)-|3v}*w~5FN5I5$}go*VQ|45^Z5i1$Fzpc|dVb0prUo1$vK2|w=-)+NM
zuxT%iyVpB2yFyfYB7U|rv2Ke!c=Me!3Df|QVHo(<6JtM(lGHvjWs3Xb3nTmC)T`xH
z_gS3B9az0fm@2ngjJYd+rs;h>=x)#+bp&&fdSws4<Eh17m&N9fJIb<-Ww<<sIS{y%
z;6ECMexBX9zXSf9lL$!Kij^8vf)1!D;sy`O6#T!L9GL#{rFk;KzO+J(PkftFERjnC
z_TpJ`|LOC;Pk$`bk|3{O9NurEx5O`P{dfHWSu@Z7i($LMR%QOn2o~^5JN=#N!ld-e
zCU)RW;g<c{ybSd7rCIbz=esnngu9p4^^Ho#FaARxQu=h^YbOj*c$Il<;bl_a2BB1E
z9%TPZbIKxQqlnD?0^jW})?i@h0OsgEZfP!Hjyf?j84C3GvFAUWK3y5b1(#xABqmen
zDFFu7F00dkE!3*5?Y2&QaJ+D!)e#u=nu<d29`d!RVljnoW5<elmABjUTaAM|W-K=<
zJ1b!ZNc}%ijHWhE5jTo&rxj2t^}zvgUV*UuwjT$wF%<Sw-vszS2<~0^;ZF-H`SVhs
zH7`rC_?Lw=Q9FI(W~3>c^{(H5KK)|Iy4{z*>M{Jgn8NMEmJjz`URv>t-*?)u?v>W>
z7&VIjG+A)US|Bi6(Y#p?`?*HcDC^)BX0M>VR6F9%X;K3{!b{~w^BAnQ=X5S{yE9<b
zrt3L9fgI>#xs)_Eo-uUJ;Z(=K#}|nO$<><C1M%t@kG8<#I(v~8eVTuz4-#hxJ;Mw5
zkH#_uei~k1Dd7Q@xMxYP(*~H#<XR);88G=jy@p6TjBrJTLw>*$acMZeycjZ8j9@6t
zYLn6eqm1jv3UuUQi5BKznlLE}i9Tz7Cr5*`Xhvy(4;VJC%EV8GWJ(C^x{!tJ1(RiV
zbCZ(kTgFBS)cVzWn_e<zLV)adTR)2d!e`~5t>ES&<OVrYF@&noF@9_&^uQzb_UL~C
zd;G^s-ZyJgH`KfCmx=wyREENSKwu>E1Hbr)^tJnuuO$<C4e+dlE95@9F04CqbN7nS
ze&2f=zFEh=msicEt@pkFh7LCJ;@#2W=N0&?Ar{Hb!hfIG1ZN-rou?=;YyS5sRPbW>
zFBK8cti#`T{}wS&y*m{a>><*lL;6>2ejJ-@APhl95V<aNz?aBygD>^y_TM2_{>$J4
z)`54aW{pu~zakYr5TW2(bX2V3!fAD>m@07L6qQh9ut7*95^gFmYx|eKP=Sl_zr3e+
z8~sx{W^FTn-{F~`Kl^J|h~T5Ce`#8zET#Q@VoKo^j%z8wbaLFzo~f=s?2GKWZC{dt
z0Bp2SoRPDmAgPhP{+@MggVXJ}$;$ESfTg;PiBFT|r>q2)=LxIL@1&&xyS&IP^;K`=
zxB8@VRc9M@VOh|JrWh?@k4cYz4thcFSiJSf-E^_1m~#|6@hbH5su|@3x>q*(TZ_dS
zxV#+FreTAx!LXxYY-tg{1@NRU8b)rN^V?)2*p(0+eSZf(R6?|IEB<|55e;)#bd>o`
zOPd;EDoBEVd7aQ|1VPdk%ruvF9Id|}9MJyS_R<ogQ)LkQWdGwap7=24AEd*hU^`L2
zS6)~5*H=W~zO=k#dtDuUCmS9Y_2+pY4BnC4Mb&e;kFl)L6gEL!n!nEpaCd@lxqk@^
zSip7ZyRiQ;0-ECNM)35dmfR(SXGgd@(>8q0w_aavs8U*&Us~$rkv8f0<qbU%Ogs9}
z6nb6h>96}UmQ|B7YHG9W%KV##XiC$j*P@~7Uq*kN4(*4fP$qZ(W;JR5JfO7og>Pbn
z%jK-VqDN9JL-i*jq~CroRGsP@I{hUk<nI{$_ovvC<%nu)^gn~#gm#{7)-jXf_o6%&
zHh8^PAMQN1UkAU3_k4}`(w{8qHYZSK7>GW4w^;d)@EhM6?ZCNeBP_Y%b6ipw1>~21
zX>Q=0qYa_*FbENOWZaF13VTt02J0uo&P|)`>+T8Bji6evgxvj_b;6I4M{Zy27szAQ
zYiW5X3&ToZpkgtU_rd?pzu46O=FCN^QB}_9iWr8;A5ZjE)Dhw`SvAA2RZ4gnbfQ9Y
z84AAHIO=F`!Vu3~ImF}F(@j}C6y9p$>XzQptdVRTA%A0h;a7t0b7G9VJoOG9fZ>+o
zau17?fHDm5c&~Y#sLU<6F#Jk1kd6-jyRuaGnX}ZVfTqw!5HQ-H)D?)F<DZI(my_Gn
zh2bJIpG&t#M4PMx<(2B5(kP;%1L-ajkn4p1@Eux;E+cB^INbGi+RGlPetf$4p%aLz
zyd$d^M-A$%pW1V|_gK3%IMtuV2?g{i6ty^VEVMW5gdKpTl=E3{V<eh}!N|?y#Em$A
z4#W*y<LP{^ebjr~5N$cfYE)-NCqbZpO=6dBphcG9&F#ulF^ucuM=V}m@6tEC6Lni+
zg)u?sfvEIAo22o@QQ7+HsK9I~k(miXrOD#V%G@i-H<-+Q=i{VdpbLCvRhaiWw@z*f
zVk5Dx@4j2yFpwzDRHwq3oq?3GVOaFRkBq`^L_&8L>d0Tr27u{Pp{LOFBgT;Dh{!H6
zj7*k9Xa`i@CJlX0z4cY#kJ@A^9t^fB2?TRu=e7zLcY^Q}Pq@v3$;ke~yQ&?WL3g^t
z$KE3MxHgh&RkW<z>5?K+zRCWI(1YlJBo7~<vq&puqU&b5krhh-84cG^?SJF6{i>hT
zdZXh*<YuIAj+oIe?}-X6(33y4RL}2STU4^Zrg-x&UQ)z;wx=)WlW4tdn}Svl6%YD9
zl8f_NLd3Z$9n-y(U|QiyqJQ<ABEjBjq~^;-=pU~3%31%qM+Rne38Puwt=`TKX4k33
ztI#9|&_|5~{T^t0*59^~Mu&R-`2Hwii3@!8O%9b@p8A^M12g<nD&j4=wy!j=nOj2E
zI9cs@WGPUJ@@1Ma%zf-6`I`8wyG&_BH)vM<BD)E!kCKN<1FR!i^1}t^j%unf(OtyI
zY(lVfm$z4L;4Mmq1|UFyz+$y<r}RE{`77%3tB<sV@PQiW<V>P!)$t*%GG;2pm_fIP
zm}~QzqzON}93&_4fo2tUhX=ceUs8Lo#@KBsq~3dlk_bTN)PHfZb>1R*b_5ELp;wyq
zn4-B)C?=4uvb4r6gN@RSJOjjFhPr@dQ=)J74EJ;5L&IAGNB(qYpN^pO_Rxw!!>8AN
zjnhX=2uQIyhk3H>V6>)znYM$!NT;{=t|X<rE?K_S%F=xWnO&=-Vzz}H#kO4K4s*Rg
zVDILVC?YPrFoUv01b<R(RMer-p`;#b4PoP>ycOxqu=8azNt-zeCY=kXOLxo~6C4s9
z;y8Xj0`fpZ(R+bezS5;c7BHpWU5L9!q=*<=NlT$s2$oN2Gj__Zu>G?SJ%nL0FL!e7
zjxx_=o;&R$jG*M@N!`i)l-*~l`yaOk(K05tP)U9)8EydM&gy%+_DAP6{f^I@N$sfy
zlN?SKrtK=W0-r8l|1)qGrzN$iy0>#Qxz}^QlH@KT!~Ni-S}=KSCfq6~jFlF&1cbXl
zM0I<I+qf{=GDvY=@x2rrAr*hx-QA%MHJtwloET7#h&6z$5zd_(Le<W(ceGv_-oc(l
z1s;vB0G8LS-R85WwsM^cI00F4O2#lx(w3j5wUq(k7hsT>%RmkI=0QbJ>LkI7JG0jB
z!S`oEO#0IY1kg~?rXBgA%Hp@_ejg|=q3uY03UJ+2=tt{ad4HXnHbOAKpHxzC^id_n
z3-IY_8-e_ZSG79*adG`if0V^z&IeUkoHq?@t!u*BY@xAQ@(7!^oFT$V%U6(!6&sHu
z@p8OEV`V=Z-s|C57!KPS@VJr6-qbC_J5T-VP8Y(~dJk>d_I!&zE?P0sqo<*gmRJv*
z0EQi*HbKK)s6>9=eJbIu6gx~#dX$npWZ#Bb1#DAXj9-{t&$qj$5o9E^`*D>JwWlI7
zA8Ql;O=DZ)iJ+%)I4$Q-JNBnvuGizve61NT|0PmI$SlXa94P3C<B8MaH$R#%Jhtt8
z|E8s^4V_hYNMfHCCW`x8I1&lIcMu?MbWcw$s=El$=$6@=1K?AYoWsED-ngjc*vWhx
zO{Hp;txSs73|y!nqkGxTl8KVkj#nof%c3{Dvc(~jCVo;~i&Q|23W9iJ&P!(aoqYDn
zDVVH!FI2R%;=MjKoVVyaL`65bA?poNUvKgKUc-MLff{{ffS>TD0snOyDYY)T!ABV9
zQW;A;t2`&-t0j+Wj91hU^xc!~*LBrHfr!^3b6+Fdz*#zfvg?eAwA|RX!9Tndc5hkJ
zwSKgDIt;6K;-SdKTJ2;q-Yi4&fOG5IEd$S0PHYWleZxLx=NcL~#;Cyv5GQ-4^?i4$
zsN$OS6`L%$ddL4QXm)^0HsD^lrPTY23}0KEsH;rdQXZ|)x6t*il_3V5ANk`iV3c;e
z-g*iiHt~BhyziDXt{U6nYZj=IGo%G&!U7bMVH2=;KF~q$e$Oj%AF{KX_g5H)rg?Uk
zmIXJrxpKmW<En7EfeL`@ruB82#{-Bu0V(G5FbYLPZw<FbGf4WJKb68iclsN+#q-SY
zC?K{C(VwQA5<RSi4D}sS7cEk<zzxfIi?f|GCR7mGHhKU1N8XBveW<jYDNf4=d1@sD
z#F*X~?lmtEu^hB#i8Pf3O%bB&R;dRBi1UkRXzK%Ex0cG7`7cQk$HAFz2bf%i$0eg-
z^jjP*DdJn#a!$U+F#wfH+Ja4&Mam8yi6mdYDg01qg4@t@Nq5D)6WuB9V+8n=e`4)z
z0G3UI7mr5N^=xDbfOb(BzVzIragk_q(h=%S3_iK_WpXq#Fkn1_a;9vFVBN_DW9C`g
z%vnp%cdy<&Fi`C<t7-8>qpy@Yog0QTS(xB(=6huaVrJeY%cUbia;|7c-y5Ltd?5{U
z$N0x>{kx#8O}1^{DlyBaNNL39!_n&%^uK8=cFbpx5wLH!XZaYB*gp$u1O?}w_zrb%
zm>Urz$*>mD!>;|%a|Z1wRMk`8MHIGYdOsE!!8!wz_=ssl(V-j)Yg5$GZa7=SjP$FM
zHX@s}M-Bn@>UVVzxZ#@*eqFm)x}OdUot2DN)ZudTr!^g8J_3UXsHxttN>ty`Jh&w-
z3nmyrQN$((lz?w5ov?qH__iFqcn99lHTIYa(ysc<+<}3)95%ZiG*c07-lBwLZPDk|
z(b&eIuK$apZ7Rn>z_XCLZa%}g_X<`ME-PWSf{m1tg(c5bc3%PZM$BGDV1)Z_`Z3mf
z66S3M`G(?Ot9>kkGbK2=`;|nUIBU%~P(fm{*W!2N=D|~4>IhTP54J7H0(B)j4K`oJ
z1H(P@MN>{^;V9U_-~pp?hR%EPL&NC-X!3H!DoS@%T_tx?UX(bAy*g2l@}Iu*)?RjL
z=$mzF&Mp!2_O+a};a@j|amVOAXECQ~Pe3`4dVx%17v?I4oTSA)y`0ubh8-~h4)f)M
zHGT;%{N9`dwVrqK&rNS)2y}&R4Ds^XceZV#9NJ&h(jGZdA6&=;?zI=s_7OYG&rIjk
z=Q9E)2|e*mB0cg<yaqzu2wU2o0LQ33LoEK;y!yxNk}tn=Zaasc=Vjsnx$79XpLf|6
z=fFid^O4lDWWJCl!?rW~WMvr*(#WV*p2NmkLrl5@QorP$J;G@F;-A|c%6U;xosola
zH30KN6Of((M9|7|lX;_3G{m3iuiU*y1=&$EY`t&#a9mBppe!c5ZRo7!NIqEW*S4NW
z^U>2mZThU>XQrBZywx!8FJ0ufHPdfEDwvO3{REK5ch;i!F;M8%o=94gKqF;>6FcMf
z&WRY&#$=CF;Bv`hOfK11uu-D3yoVv0f_<B^!0j37ub<DYd$yotfw?oaTtZHKnBv5P
zZd$AZV+0s=JQ&TDb7@9JIva4y>jk-45B&SCOdVvs9_3)%J`tn~x4z5sVo=JG=nklT
zm%bRJJ3%yti!stC2gh#9XIoW=qd)Z*SPSj$f)#*C#_BV5A2DF?V1Vok7L^w#Z&}s7
zd=x$PLl1C8FEwwfY8jNN%FhndG=(l>Ks6C-?MFg=>+Q2aCrZXkuO9RT$(gCx;Ywuq
zrq}lq%t(uPblGMsAGuEd>SY`yY-Su;U}br^F0<Xj1D^^E5&<*@IOvt_t#p!O<;)^n
zA@f=@Hpzt1u7n;84gB=<uHG7vun~C%MKFKsjNnEVxvUmG@xr65PUpJYB!TKmc49r0
zdPgD#yqX>H*N~OHYlT1Sbp+O0D%joCJv$8(|3k(#!eDOla{4&t=g(KI!qh}Q+=FJ5
zOxsCy1tr(ed0BAw5NyZ_porQ+YueFsii{B7Cg-ne?VjKkHN4({K*!RX7jNAs7+o_B
zu}eOryD)sjoQZ795i(XlU`Mcq>JuOv9mw?)m87k?pSs@<^vomo02^^vQ&tz^`bMk>
zymriX&JbZS3cms|WoVF`N{3~=Ijb!w0iS|3-g@bxaw%7it!Hp1A<Cr;0jXOL^S>}`
z<5j`wcz&*j8U=iM?0r~3Mb48qeEe$pgyt@jzLM4>e~ghpTzwA+JsS<JwJ5S)=92zN
zJoHFodLf0TD#M!&bXjrC?`<*cTz6r>3WAuj(AXx$!_-N%L-cvn5j6Rci0Oxf5Znk2
z-&u1h9_y_(WBncpu1`P3Fv`isBhxrL58vyTU_}o5k{Z+>J;*x3JTIR$VllZP#v@o;
zZ+#w=zCV=O<~Rv-&0f}jw2<%U;L^1j_HGNBZ{N)TUCf;%YEzbVKPs9AO5u~%y+1Pw
z?_}LfXpjtE+hNKQAh>gKV{zCw?Wz;I%<p^;C&(n#iRLCwvzBdrfGfs%^o0>nt`#$g
z-cT_eOf7#eC!e4(dXy#=9C2OfM|rGV2Q>m9C=|T50N?t^E5jhnH!v`He|tLNcTzaB
z%cf48KuE>~<-`n1T&8q$P@hn@Z^r58-A24!(;^KlF~T)v^&9Ji-X*{XDHXLGf!aGy
zdk-hK-lxIY796`S>!JEH406*(3_P=1jSiD)^r@xz^0NnNC7qo8Jhg(h^)vz&{+~x{
z3$~`lZT0sCNAy%YWRn-_4OKHLF7EHH*&6<sXU^&3wufd`%GM7j_e|^0U#maK{>J(d
zDG%p)Z+=tq;1grf(B68#2GUdRe7Vbhl#ie8)GdSbCLUR3dqo7j?ai0p@pI-k`s4I~
z^?)0k-uY>oR)=g$rF1RaL&9m5?+ZB<Th-486)o;5^uiiBc5vQMQv;w1yI!}XZ*I3#
z`_Xl-SXukS6PUADJIqVv>M6&d_fn`?e@K#7+jV8K*3oe%YK2-n5ak;a-vM(Sd*y0y
z|DEmDZ%HEz+cf+fLlnx>2nPI&q0OuQM{m*R0e3ta*eEcF8Kk-KRWnMlsx=)C0?je;
zRT7}}P2KHRKG3yqqV#)_Jo;Ezm2#8eUF-*Ek>CL`-ns>F(jyr}LwIzZIOhIFi&n}q
zNSH&Cq%N&s##o6|5(|Q>QQpD?zj`g3Q~KZ1Q!xTEy7cQjxx>d#mczvE4)sjRca5>I
zRJNXPWjB`jW!v1Zn@pEgbLC~ag9wP{+@gllWHHBU3C@xR+5IzZg=W1A%lVG6zagfY
zQd`%}AG-YdGbq?csc|cnrD;i5le5xoA)}*ZO5T^p%!E2<g=9754ivUUMCa&@vp(t2
zB49___^FlPni_)66byY9Z0;4Zi&Ofds8_kmP5=33qklDQEh6ZVH+YHwLI<J8;oLUN
zj`@7S3zgj>Jq%pDi(XYB7#48em0yc-{(3}N36V8<9decm*JN=7WzpP02-67;5Wd^2
z4fYC|VXHLmIjukcD$S>~(o?~Pds?oGxINA7l&_8$3crrWxxFUO2zSdE@Y$1vT-eS%
z!ty8D5hteLWq@3$?zXAZOrxRE%pkD<7G<xaVd(1w5JOMt(y<!MpqFnY;BFU#gZ(PJ
zWQUH6?<V9-RS>SS+*Cw|lHl~N6?wwv<RS-Lo*Mg`%WLJ7I1U%FN1oFuqkI=EpMFZ}
zBMu^>tHm*#2B&9~jxny^^9a_1xR<4nEoknxk5jg4AZT;w%0qJm=I~P`_08C}m7#Ko
zjOMMKyIHXL@J(D!mvYIfoKwm(^?e0rvO}Ro{!}LED8I+hQSemh38VHghSW1#JAfp&
zAH{;;?ktMr{*{0yi%$ND@*As_&oR9AeJ-*_x(UOhD;Zqa<|7gTe%GJHWaS6e@cQM#
z_K)BLYpUrTEnHu40Cw@^!zSrK!Yggc2~v;&=4RniSW{5`i=U#$4*LofWQP=sX}hch
z7BkoELQI&p38uxDB@Ge7<C%1kKYb_IPI7c{P8oNO4A9e#e^d>QEcbwdl|7R2A>yKu
zwyt^*)76$9(PZAwg}~DJut}y9o#)!{Y_WmZ_*bwk;xl>s?}|Bc8w8MswR1n^EEs9{
z=2gyZj-2dAi6@MK0!Zo(S$RjV-_xczN7pM>5I1Ug`yFFsDtmYXw3%^+zt1JDNCUJB
zVY+;?MS@$m7VA(<I5Qrn08FIF@5WSu+GWUAqub7GE7l`t*tcuo^S3?*YJHTB8Pf|?
zr#NKFkpoVjHeAmhpe&WPTwV#PU>0MCyPXVmp(vM0FSV`heK78l;GN@OVcv!|llPMP
z&}x{Lt*nq9Uds{hz&Vu|+~(^&qYvad$??W<FzpR^U0?Z;sc9>Xcyy%)WGIgf$as`A
z!A^B+98__GhS<8v`ST`lc3Ng@GCV{3GS7A#iHc8bd4U7zlgAz>YK(J_*aUNKyUS-c
zx1CY!^|USLgC=zavvP_e-+LdskC$lZt_p!hiT8$5Go>JGN<@eHIS=rm0cmbRZyRu?
zTs-o0%+PHmQ;uT0?Gs+X);#I+M~LWby}>OB%V<c(eQ*YIDnZh69&X7CZTI5O&Q{7m
zlQDt^F+MAW{uF%lnv}j}nOxfxIp?=AHZ3^@XMtVv<_rP{Bc)NlQ)a?hDVe74@REK<
z-jwG=cX!yY8dfJST_=vw|8kib&{ri7e?3SRs~8L1O`7?5bcZ@KVBJ?xV&5PFx1}}q
zC~Ax@qis)c(VlE%O$f9U6zv&%8#2IVK&69Vizt4<Yc(kyJ#sG!M;Cx|?2Cs9T!>jK
zzI18B)KUxGvxK<&_YsM7-O8Etzzrb7sNX33G*_=aF7nj(!7jQPG!G+4=^!L?Dq;z^
z%Yp`CDrvBx2q%2bZRrR#{F*6F3yvDR>w0#f5byXF*!~7Qz&~m)txJIZsQgQDt0H1b
zh$Um!5|VWTXS_%|dd8oGvjeo)M(oGTX|}|?_)NE?Y%&hBe;WFlT*TWjQSA%sHeQb5
zdz@2|!7=4;t%itwQMlE;NZwIyZq4Kq&h#f<><p0Ek-!p6jQ%B1PJ0Ns54A>ZmxEse
z%xf|v`rQG0*@Qqs)2u}??Pd>XDbutYC*8mL4x1gR59^>QAa$xi>eui6n}E?CA7;%)
zXT#zfpg>E={^uiqMT9_m)7`H`XPwmRmGz>*v))0Fx&mP#Aw8Xp#qMh_N4(IBfVR15
z`d0dbUMa-7tP)PeZ0W;1qOW?1TP^#>hQ5=}K8=fL_aNZs?(Hmh-o2Y`1R6%xy_A6Q
zJsi+R80XM?Mwc?JRtQN3-e)Z)fUalAYqR#|gpD<*%jXJJ^3e{s;<ZUXwp=_oh=xr~
zz@8Yl;ZASLMeb1yUI}tfuytu8PQQ~@H+^)%FqmyHG&$+5r}GS8txqQs^3{}h+ETwS
zJ25bPxm5YvOCb&g97qlBhIoY*hjU`0UGzIbB^>y=&w>A&ZZ@Li2eMr-=an`?(a~t*
zeJYVr`Fyo6UU(9{(;)=jc<W#ZhKD}R(pdeU{iJrTTtrpMo;WBSVOuzof=c3ULwFgR
z`AcDhBntlZ*FfU6_pPN|v=0|5qFjHlE2&e6jFNfiAdEP-kawjf;{2Kp0y~^l0VfPb
zo!TWxf9^Gw@BC<!wn3wpgyc|ldH{F9PqPOneeI@-eb6V2^E%*;q{9wKGw9f_fuOgd
z%_$h9Unuq7@=(WtSbQ&3M@WZ|A4=u8Gls|^xC}Xx)$IsszBfNh=8bG^&E#?^mwOLf
z)HX$=3h+kX*nXJf&IF9SwspN<g0s_2s7pXSd^S}F$!4Lq4${GVK`K2jLo)+V<_Tw2
z0{Enp2u9l}Q%IW7{}7N9x~eG)YR^82n3jOdGSR#N4i7ZI&@5F7J0MYwAR~Z?>vv97
z*fzXN;-~-Nj8D7P18N`&oDGF#7~S<S)`M&@i19uY!ALZEFJY#+%@~=|n>knoSOXW%
z!2`fn&axagjqL7T4WJ(6WmQY}3!(}Ix_YVV)>5L|kw1rlni>8~9HtY~aXHci)0VNP
z*&6EtRq!+{z5)DXQ?}%RA$KO*3Sytuot<1HZ_;CIu`pa<&t|-w$NVSKbW=gMcdn)u
ztEp2oW-Q~58hcOESLh*5JqgfaMvz<IAP;VDYCf)S51TKt9VZfrY3HM#^H&;Cy(7r<
z5ZHP-bi8T*X;!>NZ9(A!v&k8XeP$iY8!p#3miP!ASA%BmVe?79v6llJuz!01#MA3T
zI;v`UWJ%z@<RQJ}(=N)fYYQ>W5+40ljmz7Ex5zVAAWR1WN}rgeu0b|X$CcH9#Yfqj
zb$AL_ub!r!hngB!vsK|CYtIPT^bQ0Ql>L}w@!8B^1Nmi}4%d#hcWN)9{AHLxSK#x{
zxR;%B<HOIco+{ruufpRzZewa0fYzLv#xa4?#CG}j>x)b{sd><>BbTx*?90BWlSmx0
zO#V{Z#-!Af^s?dS-Qj=Mw64ANyd*_^Xfl!YlJ>MK_fAUYpenZ8_?lnC13)ar?K`BD
z$f^M}@#7VErVa;h%q*!7!*4RZ-0*78fz<QUAwnT%wBIX^2Vop+tT&zbEsi$}|L*=D
zh$y|&{cci!z2mfC^OS_dhi-7-J~Oh=Gc)XPb3WHCuu`%opm;yvcb5|2J7T?LrMcpY
z&6CUiXIMU5aEel#VDZ`RD5WH5t1-LSZQ?yo!sg(o?WL(-eIOnHs6{J34+#iAvApPj
z52jM%hWu_fBH=7{j4xmnM0CZuw}AfMRgBH2(Vu9$<ENx*Yp~B=Wy<p!VEn<iv=qG@
z2BDxPC>S8Rerl^v?aqTu=QqvEL_Qf*iATp1dK%}i(+{`5k^&$Zj7!WO*2B>l?Ar1D
zni)f}-E3rQvGt3Pv5#8yG+)5swi~z<3&13zT7`>1^ih8%-b(n>`;8m|g@4hiq%|ju
zVcs^#>9`eNfQziU;0`|Yeb)54X`)`;a^?;D=SYG}*x=1&LR=6h=n^OMOusiE)OaWf
z+wN1ZTo^3Dyj9$Y;p5W4@}i(EU|{7G5m`rlFXeKzof*AO_WgP%9<%zU4V$qxj#SsD
zsFVZ!I&c4%R!9Ajv#m~f03=|afQDe<nIXMZW?6l+`q~#!AS-96Hm)io*H(~zhO+q2
zPM1KRLlOb60H9n&6Md%>;JCvh<=k9~*MDV@I-kL&w;-mhVtO0gPfcDRfifoAkR`!c
zPqCfI$M$KIKCwRIbUdV(g!epmsB49eDkYh=vxD5{D?28v8S9IR-c_%{nS~7|0UW6(
z!DJuvs3U+|_`WWC(m(WBOcj9elpj)Dw{AOCmXiMwOcU@$HVjtN>E`W|r1xM~a64a!
zY5N<^-672M&R7Jq4g-KRBA&4+x4d4&$wMeP3aRi4tLgl5N=>8(R5|%kEkthFN%iq>
zvc|5yz!|@!N)Xl55%y7*JCCvjRWOLj%Gos|kywpH0PvIQnIOotkh?!J(Leh?z!HVU
z3gpZb{@`o)KiXi7TyVKy{c9fPaX$?9;y1v?9NrH;Ev*;f{jU%<9~y5nRqiiV8r^-X
zKmeY)ijkPtst}FWV1+@@mRWN5>eZIiWo1%S`H}S7ovzs+n$44StV?x5_ZcBRyeT+%
zskzvd{lq?%DSyxshcIHF+LUF$SFtcMK>>MyAf(y-#D`peXMCPWI+FPN=t{Js@E2P*
zcXM$zS%rrIqDzd=_ZnG96rql`Y5VvW!jcXP_y>fAsz0FSFQDa+`(l?xum^~gBju%q
zFQ>2SmJ04R{RWaeZyKQcjoAB*>yfJ<I>gK|O;$#>H2%VX{|yBhI>Wd<$O-@j|9kYX
zF1RIjnd+AS{~JK^XHJZu%w@fQ9k+W8GHDoK`w1KnH9IA;UQ46eSzmh#)g=)xwHAEm
znON6x|7SgoU=z1c;(~ZZAmymZf-v!tJK;Ic>@Y~!uuTL+zaUNdgr{~L0OX{0r{Qht
zHa-u6=!!r;7tcPtgP_!VYKqXi*&dO(gS&nYC}+&yXP&3hduoBOTo&zuwW0b|AP73-
zH`>UA0mufr(*CdF0mjwHzn4c=TK^MN^&fHW;y|((IDQ|`??d=^ICiTnBA6Ke*5xt&
zx8n*UIDYwnzj$vzylq$@Zxr0|6>uLNz*8u=x6Ivph}qQ3y#PxrZR$HRQ=>5B0Apoh
zNYK{*3`j0lgxfM7TKvfl*Dz8I7Y)_j|2O>OYr8urxtb%}_YF5$c((EQ`>Lg>co!*C
zC@@fMJ(P3!gAvz8V#IOMDgJ>nK)%zB<ed8EQ6=JHa0Asa{f&vl{S79$Hm!>LpJT&G
zM7ZVF2<U0JOGcbcB<c<Kte<9F+|5Y&j#V;<1l&I6i!*8%zrPtx1C$6y>jAVP^k-W5
zPLnV;5(skX2C&t7KRaA`6N$3bq`q-=kcCkOmyogV1fwsCVRqCv_&$fBV2qg@+ui-V
zr8Ve<O<C`}=xMm@BWGKDZW#cpu5klMvhR*M&r%Quy%W}~O-?l_QyV}9Z%YCn`dNeH
z_qv$KbuVF%bRyhD>775H#OaZAuvOI}0rC4PrI%oF@P8sE6x{LFA2GEPlxCwq@BqZo
zYJX@Ai_q@hXH9B$w&+0Znj1L;t1W-f8W)88I?ewC2p%R30KuE@<6ima(anJ-g+`;U
zYiVbc^)(ILa5ez{iY5N{P~W>C$Yfe7UaJDM=M}!A>S8S5@EUFj+1Z8`h#NekRo5zP
zVsBJU(K6d~=cR^RK^ug?(DJCjcrq{kUq2*2{Eg6}@w+X98>17H?4O{kbMgAQug!V#
zyF%mByI#z63MRgTMG2%74*k)2x!n<(j37hO5GsaF563|LWt-IQtLdmIM>Ziu+H&$o
z+$pa;$c_{lurhQQ)bx_5Rg_-G70UmOpz?6%lq~9&i9{+&&zX9vJ;RfAEgH^qTTWTT
zSrmfGTIOD7cy%ny?ZRX1IbIln;zMC;ot5x|m5(oe#>Dd6ZIF=I!x&_6G0{I$BC&og
zGBDj$O5y(H;de~)N7j`VZ}qYg8I#cR8Gir#f|N()U<w&y{Y;1XTJ6!RB=>d(Zn^-?
z18xN=^~RI_eajz~1~0C~CQ!!E3RsWSC|2M59d^AxQRr|7EhP*GGe6>yvZ@mg(2Tl!
z?{${(=IsYxgJOL|xfKYy=+iCs1w(PFI^})r{s6n4-<wkpL#@Y7>_l8Mr2U<4^hM1C
z@Sm}TC$zM7W5i$ZMD~~+AHpdZCP@Hv1^zJuUk9I}WJv3KlLcpIg}}?P{K9Sp-^!l{
zn=a4CoNlz<!YC(7oBC#;k>H|oy7Eho`7%SQ=+mTfE9^hv-(>IWSrry_KN=U%OSbK(
zy@dpGmHBzSXZcYmhiIujj)_N8&t&AF?`~06f9>dhlCeNukyp~KU}t(Upr9K~A*qK@
zN!g+BT5Pob$0TMoWU+qWTrWSrTmSs#VnJE=yD7nkVE*;kY+%Nv`YJavpL>8&HzIEw
zRb@XABUdO7UJY4~Ed0<JPbPZJ`$e{JtCr2AHZbe4Yn^yHOCqZgNpPsQ3HE+GQq(kE
z5hY@@pVSpz3rBo%u;1_Pgd^t_hEuwCWOAZP=BGi(lDP`yha>{X&ZkxBSiS=vZ$3?p
z@B81HUPu9|k=Ol=imToVi$3wd46^YD=<IO^DY16s&2ua8DdG;xh-IARDo)RxTr_-s
zGN#M%;@S(9Ikly&Voin3-6oT}@(6sU&d)`1ons_H@AEWK;$yzMCNubMB?Oamka1GH
z@GAfASI8X30hKVh0>KAC*(j)4R3pNO-z(*OmxTE8M<kEsGS2L>H4&$P_?dgc7}XKC
z$i|OG^d)@t7DRE~GKdRuy0d*;KdR(fOMk6DPG|65Cgddtv>Hdo4DBHv2G)%a|HBV8
z4y?zTU#!q1slX12K)M~gc$60j@?#_#<Dvr%<ORm@;4DV~Sb<vSspa}ELI{KyNEfew
z(3xZq#&VUgvY@_<4F?CpsIu8?BgW56Phi$|E-{d<*S!gw@xIb$@EPfA2amnZukSty
z?26exv=|jSC}5dXJ7eBBn&#iKR1_vzFwkmFvkmk*UqwIpJQL_#yU}bnJGB-}a8E_N
zr!{iJw9xr>Q-u@jam{%Ez#|W+{V;!hO)9B*hYmt1Vlj;qa1v7!8FXp~HH0j!XJkmv
zXH;058r77{kb?(E1DBt|f%r4}HW4AD@5=)hisRhC@K6))ldtNe#);Ik&#X7j+mp@%
z<&cb340;D$z`sD`PbQ~vTEy`Uf9{6R=SrEr(WEz$sv-<@o8<9Tzd<s;Vv*86qV&0N
zR`?YGr0?O)ZH$TeP55?(^PmYe{=pl`=&6Z~)c%3BM@~as&mv}4b<CQX1^udahSt1p
z3uG>f#yMlY)wm#E9krIWwZx_Yr4`5^y!Z2Z#;Idsp*D5c8~bht_G0ao_S2LHq>^K0
zuvv$`sB)&`D*pm|(ZSi5t%Ps-(fi+N&YvrKy<w=d!@+1vd}|@SO6!rFL3~J*C3LWv
z3}7D70okOA8>lyJJ#UDhNcF+J!INvZ_09Tnj#OIP7{ARMaVtWxahtwY_EGe+k|5B)
z`j~7zv~7>5QGWa<ra&z^K15aX#x{U36c%UNz7Npu&uqbCa;eWX0vlCN-SMc*wgrv_
z1nF%7X#zjwyB9FBPDBPO%)*V3+H(e?{xL30P?qx9&Jrj1)CQa`U^4d1E7whM9QDFf
zIcuv?xVlbg$laL?LTS}C5C&VxISTXmyd`!Km<oYevlEStpS6mMcN|-83Ow|IEJ1Vo
zPUH|C{Biq6lb?2vxLMybRO;u!PAsuDtaekn1bI_2_k4$<Z@5g(baJOWFT#Lme)uG5
zo4DhB5M6@f2asGLI%HgRAMK@eI9ZX}t<nhA6S$ZhKM{G4>p-vM(JfixBt9XU{uIWZ
zbA-LSVQu5XGnk7t2dn~MgQtG*bugL2jpb4i{e-;R{#m^l&`fN|P$%YB8(-_OZSv#x
zjpKRL*!rdQgCD%-+&kvIzS4(Z*@6YcywR<EARDPY(#PsK>a(@fphB<>wx7fPaI5^%
z3b6H}xif9?I3CbV_8WX?>{uZ5SLQcdR9v>4ab@OdZjcc*G9R*#Z=0$lf%DM+CocSd
h<nRCIgO4uWvnaT3Fe)$sKZQk8QBadFmo<C(e*iMF56A!j

literal 15940
zcmc(GbyQUEyDttyx0E2=og&gmN{dQ2NJ$OS?T82n0!p`lf=Ws^qk;$wEg?CA3=JcL
zblg3lUp>FyJ?GqY)?N4h!CJG~Z#?f)?-QRLeN$JRgpi&P3k!=x<A%yDEG%po78Xbm
z9~=0OPycBL@DHq;+ILie#|sxOfIuK@Y-}(XjDv%Ni;Ihghlh`kPe4FGNJvOTL_|zX
zOhQ6JN=iybMn+CfPC-F&@!~~FN=hm!Dr#zK8X6i}TG~sOF457^(bLmkzI>U1fq{{c
zk%@_knVFe|g@u)sm5q&!ot>S7gM*WklZ%Uso12@5hliJ!myeH+pPyeqKtNDXP)JBf
zSXfv@L_}0nR7^}vTwMIhl`9ew5?8NYm6VjccI}#!l$5lzw2X|5tgNh@oSeM8yn=#)
zqN3vU>(`Z(l$4c~Ra8_|RaMp0)YR40Z``<{p`oFvsi~!<rLC>4qobp%tE;D{ck|{=
zeSQ5~w{G3Oef!RxI|c>@hK7blMn=ZQ#wI2vrlzK5W@hH*<`xzfmX?-QR#w*5);2aa
zwzjr*c6N8~-nF;4cW`iUbaZrba&mTdzIX4Qi;Ih^tE-!vo4dQahld9Q0`c_pynp|`
zmzS5fx3`avkFT$<pP!$<zkfhLKww~CP*6~CaBxUSNN8wiSXkJD2M?f7=);E(!^6WL
zJ$e)o5%Ku(<H*R!sHmvu=;$X;p2Wn&#Ky)xefl&mE-pSkJ|Q9D*|TTQpFdAbOiW5j
zN={BrNl8gfO-)NnOHWVF$jHdd%*@Kl%FfQt$;rvh&3*CWMP6QBetv#ILBY$HFJHZS
zRajV9R8;i(_3Jlp-n@PLwz#;sq@<*@w6v_Oth~IuqN3v6yLXk9l~q+$)z#HCH8r)h
zwRLrM_4V})4GoQrjZIBW&CSg%EiJ9Bt!-^>Fc|Fp`}ZF{eE9hBV|#mhM@Pq}PoFwF
zJ3oK^+|||9-QC^O(*uXYdwYBP`uh6&`v(RF1_uX+hK7cRhrfLJGBPqUIyyQwHa0#!
z{`KqEiHV8H$;qjysp;wInVFf{*;xbv@$K8Uxw*Od`T6hPzb`B-Ad$$$#l@wirRC-2
zm6es%)z!7Nwe|J&jg5`X%}o>v_2b8nt*tFI8oj-}{qyI~ot>TC-QB&tz5V_DgM)*^
z!^5MaqvPY_larIv)6=;Mix9vD2p~61ys)qsBQbv$;un6{V_~siX{cPk<Da#j9bip8
zGJVqadgP$BHpFdxvd)9nWp?AR-0G(1Z1D&sNR+ME=_Wa~ITk^_)ASoE+NsN>rG`nb
zo*+i~?u|5hJaC%6N%fxfk$@sd11agZ(JSYb2%kz9Ulj3szKB*w_U*tMBp1ClOBNGI
zux<z@d0)W7T2Zvdx<w4eW$E~T{X@G8f6CBk%!#X^rL036qGd7dAl_xRIspfJloU_i
z;|f6IiZ}R%SY+pYcZ+K1Z@CM0uPF%>?^;h`?adH$DM<e=Xjq(%Q0tkn<qEJXO*ao}
zv8U6Vd7O4Q#abavK8_701%W!TKw2yiF-62BMQZ}EE-pR;HogE9)`bPN!-8^>U!!Az
z5VKT(LEPBjxC`JEMa10)FwqCFi}+c@<kxV?jX}`F3%?ghYJ_Eal)lXpD|#{1OCcps
zN_zn+r)VAR36;1rlb=)%8}(;>Or$@JNcWWh8{&bqw24XzwrF0p<K*xa8t-t=w+r-F
z^Ulp?T1<SjSF|Q0H?}pkW=yW*c_oM<6y-qwtSM|>6PNN#iYg=l?*QaI#u>d85OZq+
zG!~V<QDG_o*K^^$Od$vc1pyR(1i8FvR^@er7CqtIfRs5(erFfuHO2pr6XP#xO@F&6
zDyp!K<GI{C%&(XI!*ZM)f9(QDi<qe76+m1yP8k*d^tTvW^Uw(;)^-@>Bc8T4$k!Af
zdE=A?Nq(VE&wG=lA~<#PKhtNtiM|yu8QUCf3$8PGsf@V)Zj@1o)?nYcE{gSZ;-2KO
zKCyq1^;a4qEDI;{YnlsgI^T70B?$_Z5fuBv@5Ir=GhtFq^XMQ?=l7L+IHakre6(#a
zQ9cj7W1Jk8l!lL-bNNE(8C38)oQHPSWY=(0F3IDbGoTYT+;MvY0!^E?pD`5`nd*T2
z%MmEX8b2=EdUn2iN<?cvdvITr)%=_jpUSaaUZ1fk^~tl|z89f4Rd08`-1#=gx!d9D
zsa*dryr|`VLRaeRRa4I;Ug>0m{4zalyXR8dvfo#k$2vI@31ecZXy0xbkmmEy*EC0r
zeC$AIP}{~HDym1MEgreG=tSii*xN2?v5K=);Nc613Rc8*n<ZVO9WQHf{{nlP;p)~G
zid%d90Y{m?zd_17UJ9e7DVM^SjFlGzHQx}GTsS+|0Gm1sxiKT=H{ON+fujCj(f0rE
zi{vfhH`lz64D5OPGm|zxu((efh1+B5!PPMHR+4gUyyCs~{b_0O29i~Q9>1Ktaw&z0
zq*;}$ez`q$TQj0&`*1R!I9Mp!*0xkKv1!Km*=mWj`1_iaH{XA*!oRyiz&qJ>r8tJ6
zl%Ddtg)lK2T2sa~r68D)H@Tc6?F8+a99SU#uf$-y+lj|o!FT3wY-n~vCbh*d9t5!S
z>sg&gN$Gd-OGq)69sZAZii4eANhF4jKmnbEkpK2oHuUIWhv)yf`%_0a!a0<!eK6|D
z0RHXrfK&gA6NLHvtHXQ{l=m+l^Wz;PcQ3~(g+IE~Ao(`ZyVRSRID_UY=*&CKgU-y(
z6I%L1iI_IOF>iP~Wf%u+2soa9^NJ<9!yXjm#svQu%*t~b|A<S^_YX{OTke>=-%jm$
ztZdB*m|g|U6k~a22zt?BM#GqhnXaGg>hk!|<bV93Cpc8e%=m&dIEOE#{$oQ#6Km=d
z+FU416Y$eE&>Fx`YukjHpN(MROlkbcX`Ao3@^KFLGN|~%Z-*<^@bb4)%*xu+SQtmT
z%ZCg4$X`1x^1&$7NJf{>6IX$NmUM(^k9lSKRHIKySmopL2A<?Poqe6G-sT%V)1P2l
ztK{|OB|ecoss^E^t6vv9CZ~gO1!=Gv#~a=FJ&NyN=N38D@tJPK4D?qy(OQEaRkua8
zO^{zoQw@jd^OZNk=F@3km$_IX2yD}3%-Mui(zZ7qc$qdG69>N+t-xcz)*@!c#@qr@
z;rrL!Y;cyy{Tn~eylzp1Z=`J}B{Wm={e4wPJ>9<`r;$DA#kgnccg;co7&FH!#ID-G
zTU0_!4LZBZeyqai0S<o>n=R*lBQ6cL<WsvU6Mk@=S|=N6O>0U5h)AF!1@LLNv+S&a
z9i19z1M|}c?@1cPxQMaG&~#6e;+NopQ_@;#P1OOHMvZQn_&_+ohFbY%i+V{Y`YGf^
zffBr@{{ieCAP{G$4<L|O+g39`Ab~k}|9HvAupdy^#rgg--%a;ExTv-QG{6=KT<Xe!
zI5RZFcC8=(7znXLYww>YhMwT!Kjw0&Jvi~NCs=U7pC=x>9E)IHYL;KI!kiOw^^a{l
zVn@~dF{HUVxKXCYV;B%Dk&$1EDoB&5lPM+mZIjn7f{L-vf(ZubsWlfJ=1EQw!4is%
zG0(q4iT4!#g?|XR2KqhkcMuZJjxzXbR}tjwU%M)&m;Yt)c;Z&>f3_mEZf*a=gSC&r
zd4IN9>I}~L=LviK1Dvu~<+lq^EMQ0Zix2+3Y^<d~xwc&NzYc(9C<Tfap(Od`WUw}`
zfs8Vue{<o1Bm*=e-RTz-u>A)q_VC3ACruHfT+H+WE-kao{#Q&d!@*-_hS}-gaf=7F
zA=&m4X&N&XecJBB(mh+*W3_-CG_3^r!zuO@N0P(CLThVaf~z5~Yf66GnS{w*<%T$t
zPx8d#4Si%4VU>BuR0<P!W<O%b!Hx}#-0=k{ee*`%3svK-DdC}BvqKdA90rxZE>d^-
z551~%(B31h-@eh>Mh!v3{?ExFR%!@`KaU~`>=o&apx-CobpT(z-ST@XqAV8XAkXsq
z<n@B=sC$2=LhRHKyRrXp7zU4p_5Ag?DF#yT;P{8dT#n$-sT)^*Ir|G)85Db=f7TLV
zV@GAH<^SIJg)BSj>7PeL0zgM2_oH*Q9M9d669?sbZ~(3hqkq<6SH+k=2Nynb{YB{U
zeo_EFT0^b!*K#|Hf=G=JlA1Nr^J7*h^ov{hVllh_S%Jx>)e~#e`6ac>KWp%#vnWaA
z$ba2D6$)b_3Rlzj-6+5Q(xF4~<bY*uB)87pUpltd&*QkeFNAFR>pwQlR)>jjqx8?e
zlxaLw4avF_1fSP=^m|s%eWcfVN~r9B%C%q2X%%p2ovH1q&yDUg*v4<(AAQ@+>qW>v
z-9sB!CTq9TSV3_fP0<3u6bTX+*-;yUi#`n80^Co9gk{t>r8_}QY^@eiQ0~+iJEKSN
z2f)o)(OwM$f;o}JM!{Q_9nC6Q=B%c`)$uD@ms(!0+zNAjmsv(t^r+;?+45;L#_9_s
z4)|5*b&=G4J?SCoJY@v6?G5Cri%!5-$9?Cjjx4w}+WMW?uL$%kh{txG9Se#3hJ%c`
z^3q2w5+-3=C{sd1gdXOKS=;E8np=jjo)b@FthGRLmjvzPSTPExeE<6%%b8AC%)P(r
zdw7$g@_OK(;1_pG7QHFveE7qLen`4N8KH0c2Dm7=d^4){13d$S{tj<(?~|b!)4$zh
zmy2B?uNU)sjmp1dBy#*jbYbqHZth76iToE6!_wL2G~?CY?`4vP-=%?w3sPz?^i_M~
z@n09*GKqOsP;6etK3($9Oi)kxKY0Moa1jaRT<jQOs@prPfsRo(vjDSXLaB`XkMM~)
zk=Zwup%&KrKKQb1r<|h=-&%CgJ>9bcX&@+6{t7<BtF5;L!QfrL$+<llv_0QyCau5H
zhH+T*Qee&zm#6A`|Da;BxUZLwkR5i#HG|qHuAX~W5~o~i(g@ruN`SoaDG=1vO5^dh
zqOUOItG#Ki)!biAi}y1Mjo&9uet<~v_r{mCkX<09jrz)%NS_nHc{pc6u|-Fm)xPy0
z(fHy=?swI9W>_rk%6zR{zsAHzGmp{9jX}$OB8NNX;xt@T?RO?Uqtj@F&M(yYF-*$G
zf)pRSYsJ_Za(RByM=27f)$!TE@utFSwy+N+A!+_2A)?A(9uZ*zv|o`6dFtZi&>M$P
z&QMG4>M0^QeYjIiW^so^`ZHCcS<jOblRHwKG=hH7-Fm<C!vjkv#$?GT$}}(DFtNe6
zaWqgPGTLx%gDBArY({F^9JF5#6)SE$6H8nO-(B3Z+^-O4R6hWFd`Mw7{}2sd>>yiv
zl(PtRDZeaH@s`Uhvnm%l-H6I!O?K=hJ^rSAmW3V#Sd%16J{YqvchEUDcDf<JQ3)ei
zpsc24!#-P**W=l|Uc+TxapJB`omKWq7};Rhl2AC&`Ab^=tivl|De3Ut)jl##$SQMp
zXv%!&<X~B2Pr@`13n*G^G~&!m<&yQyM*rxK)A~S9zz&(z|88HhAy$8u20wR)o~AoL
zs|?q%%#@%|L*S^)()nB&)UZ=Oed3#5JWo+v_IsSLV+sk*-JK#ID*{q>)ZjLDXX*9r
zDE5!e-tIrB@0^)5#|Ws7x#L)t89ogkbDB}Q1P!|(p4BEfjMF#UJ3Afv2FZ5!Oeny>
za%-jqy5qEf1f<|2iamv8kG|QW&mphp=%r1(C@Rl4y$AK^p2g3$7eV6%){(0rPJMAK
zRkP><vUUc{9x|Mn?MqX{ymSpRRGB+(o>C2g%`;qXI=Scv_a^y4A7bf2aQ}QI2k5L_
znLrxKw=h?{6xfgI;5t%m&}xWnanoYkt%dXRSv^4`1E)1-ANhe7JAHPfF<Jz4n!Thx
zHnHQpjjq|r$Q}CDF?JIR6J$T9BAN=M+C8p-3<<)+1{$B<^s^;lJt%Y&p3Lz*Sze5;
zo+Uw^#dR7_a6$zpUpA$V9Pc4?r0Xa&5rRf*CxVWUh#+=DgTW3UfCjXbqXU%t`RK-?
zv!iJLV8hc2GV!~nf)%HtbBM&v;RYYF^RpTu7SrOFx59$c_j$69!3E(SXX(UL`VIvg
z-Kfjddqn%xjDQ_AVicriXcbcQsaNYP8`da=#iXlBGS{LhgCUUH{m<~(MFMC|Q>1e3
zvCcf&5FhFjdy^EX4{{a#>fkD+31YuGv9%O9DTcOEQ2$yWp>)sNlc`M-w@h9(lva}z
z`Fe`unEW=Q+vVN!gKFZ0RxxkxZ*DkFe-3VbtZP$AMX@Kf(rw>ZYk_s9ugg^RKCe#`
z!+RItb*mFMDWFC=UMv}04G<d?{o8Xn=|WFG2m}{Tqv3}oL)CeoJ7I<AG$ez<Dw)yZ
zCso4M@&>~xAL^w>wfT)B&Sz;txkdCy(k%RV(8__eR12L3lTbgu^LJ{{R1s!|y#QHQ
z@S~i#J|)8Dt>$pw$+)#z^tL?F!i95j5;pH&LheJgs&nDN$ti9TlzWxg)98L*{-1-h
zO<FW(OuEH`(t)!-xx#!nW7%Ak!u;_j$ZF;D;c;aw%!*}bVA1o}j`C2W&5y?^)MieF
zg=k|Nu;pN6=$`-1-D*y!rUmKR$%qZGd(iPKhE-b*WuP)PgL_BZ%Gf*a(?|3%ZYP_0
zqO{cktj|K{-abY4OPeZ-WnILN{Y>@Ni}ShoK&U$|TmEz5SeYu|0$P#eGzQSoSgxj+
z#-0=C!Kc}KlHS~?{)Kfbz1Q5BwXf<PeufC@)sonZqO7gzee*UwX?;>HRIFmFT1Iu`
zub7M;^bNPFPi7~5>r^2-C^gD@YS?nnef^5@C@$dH4HMH5(inGd-qxqRQD_Q8jzvuc
zd2H2@yqmLWs;{|stgPjihRLD5P+lAGOtL#>^%iI=RiSriwK2sW?)n;OaTHSBglzj_
z>`eMV@+i5g>Cel#MfgqkYrRSs*hKffeG>s~rmrWm@}q7X*w@dgzHG8sDj!On{<f?V
zW~BE?s2bcfP}uNn|7F7TX-IwDV=WCEaI4tl#aH|&%G?58&$_RIaj?sVtwx_|msU5L
zf2gpweQCr7N{<yb^3#!r5v|(dS|4A%6ah(wB>JZ9d8#3H4qLTn&6)i&r|0lxU#u3y
ze04YHBJZg8qB@(UWTCX8*!k+^P09PsmBsTzbx{XlPha;d*t!qmsUxx*38(tA{b&Zm
zSGq7YgWj!sRx=78<cNsUdhD+1(`^`b*Ep`FiQnrF>zK?Dj)UbEWRvmVz3J=XNGWKj
z_ZUb@T$5^T2AM`YE_Vm3H|snOp<g~;?UyZ_^<12#-P)6tj)Tbz>^A25kv2?=@q#cS
z=Lxf8B44Kc$w}%ntd8~Si%@krEHTQq8vE=m$InPn1tzy-uXN$4n%-L%{eu*s+?^{4
z1eoJ!lvy*;dM1w?-k(jKcNKpLV^-LI#L>4C?1^&{x{EGMnCz%$!yJU9vYL7G`1+UR
zdujahM_Nh*M{B!;x+I<BSqj%dMwEg#&uNbn>=f;fYc}k8ArykI&KCQ0bNxmaVK9@v
zWH$T}4rWKGGDjh@R*wcV%Z;9(w$j!Sk*hh1SC(3hEPD!CYp@ugb!bX|`a$!bcQNJQ
zPyTj#8%5q%g6PoJKV#n8fKMn$?(~s8<wubLEVy8w8+q$Yn7mh0#7`~Vdc~0G-IFBj
zsKxmyJn9q{Xc1+{5)~b0UtS?wc|#jK>5s+7_2pZ~`oL0rEidSqb6B9l90$dLVwn3-
zS+;xXRB$BO!&+YHCqcLfh*{NBH!-4}qCP$s*2dnbC3m(K-6#W9QNYtteYhj<%={%a
z1m1E{AblR`2kO1Z>HelP9<?GQhG|a+u0GzczT`(27OeR0@yiebADtz0>2dAbNlXvH
ztw~G$=SuPvB8kp>r@4=XKFQ@Nr4vALnh2dRx%t7CV(#0e^0>XebRA@+w#|Lc60SXe
zD`cA9S6y3qgJo`6V}yGk-c^5gynTXToWb?;6?GueUOw=s3FY+UoO+^K&s2APJ{wVU
zXxNwIX8!08SH3I*)lEffY}vlx8*2{tXvCSz?iapE%2C^*-rH3PyK^SYl3YRTG`p#{
zv+s?OF4Trv5ileDS{pM7OjUl>Q3!pLgyc@!aQv>me-U7F2lx6vuM|sM;hrVA4(fk-
z%?L=2z#tdLo<X?jM0(@R<+yGU5N%eisoHF8YchGYX)eWa{b4{=l>4po-53`U26`Sg
z%$r?@k}8(`l`yn((dWwq$_X$d%IOz5#DJz7%Wy9(+Yi|Ad2P!)j=?_~awk=3TKHN+
z7NY(x_&t~)m~pV2-{8{mI}&J%qV;rhkH8h5!EI?Y;xj)r#a@HE5VlpJajP}J`_{wN
zLDp*>k71T974BvBtp-M;@8Hjv*M-?oZ=Fe1hB;_IIWdbxu|Ta($qb<UZuo0BAeTZ5
z$ZWh%#0XDy7v=t?ajNT}$x$snI24u*g=LuZ$=)&>SBF2#?8Bc&j=v*}h}JBwNm=4H
zWlz1x=(32@x_~!30%vO4Y*qi_vtvb9_?_Gs=pb0pHo4#kE2o_@dQ!_Nl>O>sR1AfP
ztiHU?ZFS@@j*krzW`A3O28F5aE1L2Te@tQkX~hyT6m8X}ik1%<Q8_Xn97Gj6QH;|I
zw?kOpwivxTF%eQRh~ObQ<OnUUdAw94t(jPK(|TdvsYRRc_I*q_4^q;7h>M0iwck%@
z5SSC!_&d3BR*p#1cd0umEm`{sp~D4R>R$Q0gMTDgy3E%Ji;3<dtzrNQNi^3C7g_{#
zl^Lc*9!lBvqyRN3*2i~STSt>pLsD|D8fp8kT0gD7VjkrQ^(*vn(J`XfqsaeZlM;^U
zkaWInzP!8GQr)^4l-3_p`<N8!qO!x(caK$A)Iz9I(4>r<`!SIy3o7$_ZDc0p9=f_|
z)SDW5dfnSJZQ(3vf5*4Ufb25W8m?g8WL#fgA@_f4JV&}1=&LLw|AWMDuM9sRrDfzG
zX~~gZlilR{9rqycXb6#sH#C=pp3*@IQ!ayKbx&Sp*<$h#6^aSuHI9T8wBPcpSJ7S*
z+Z6KDuP4WDgLzPpRA^};ilX0s5HMnaln$w)+?^l5vZHSYFxxZbHC*FA{PF%7Uot6_
zUv*~=8I!NI_@tcVq6d}s_{5U3GLm+WJJc)LKVoSKJ}{B6-$c1yCeiZFGu(be)ok7~
z-^M1TxQ>3Se0E<rqzKmbuDf!jfLcnx2S}N&mQ@MN2GF}J#Q~ii8Mn|&X!ajkH$O|k
z>!{uBZzHy(!9uhy)8Fn7E-${?kGqh6JC#r{1-MTXc5;hhUMlL{?--XSO@|HqlU|Fh
zNc*jzQ<p?3H~FM$-6y*zP3wPhr}tcu>tqSY-mh~Zx_X<ozL^xN{=KcePay4NJ34>s
zn{JQ07BrXp;3>o8GvBd{vWI9QS)LVX2^9r_;SY6DC+>3=oRPTDf;%(1AU(r?qjDbh
zbY{Ns*Z?SZl>G)o<2|eLF}0F;LoK?}%`Did){ycT?ooEpL%M08??uPG?$~w3*^!J^
z`yREPdr`HYdQ>_eo46;AWQB!X9^A}#>xj=&(K8<9lC#NnE|ExVdNnHLy34L+w!)WO
z*J3?Sa`;1jkQpZ2t%O{8xDF1}S_Uet{@sj1<T;d1F6z<4hAQ~Zz4%97B(I(|D|^Rg
zD@4~|hv>O9c59SOwb-cs$Z2m(ztOVU2vVyD&2t!DSV`X1m>x^0bY1JmQymu_^J2U<
znKk46&X~roq@L@R0icUbeZmzMw?J*G+ll;2kmdc<1uHU^Co`Ktb+H_#AgD~9Wk9)o
z>#Rz&?$d3Ra8fAH*GiA0Ud1!GvQRY5ytAc`_X%faru-0MpHFmDHu0wD(o@?YsS6-K
z)t0Rs>f^z@o9493Q?)5Ud4Ko1zRC2>4!5tmNqw-tM7}4NKRXcs4UrKx;A(_jPG*Xp
z(7nA`RuIJ$3Cq?mYi3Vh)uCFNxv}`E8oxQP{*vq)rC3k0W&U<q_09|V;kQ571AdqM
zSV`S^t69%G2{_Fp|0YStw1k(eo$x)r&FBq#4t(>d`wNWSt}kCj9kR*UmMjJ!CRWu*
zO5S#jI_!$?-AT%VwOSPMCf3qO>9QTDJ3$q)X*Zf2^vf<58CoxNn{_nzsRV4;wn`oi
zaefgG_k}93NDuW9txdW%i#b!)ceju-=puTp$eO~#%zY!a2}?93&|;JgMxqK7q02b}
zD9!xYaYj&`vb8!w{UhO#<r4e$X;0a}A-J_`%4CV`gRq4u==7G$OLi1Rt-5s^7y6Dy
zCai!R|ESCk*}yuUv-24<I5bP@d)3A>Cz+_-$xma@8f!%aSqrFc+1-;r1eB}2eZv@f
zs>AU?AT{h$@h*X@D6=EW4kI$5=&8Bp4Mco&=qrgKiakh^GQx0LYw$z^3yXdRg9RYt
z-*JV`a`6#WO@(<^dkr0eYw_;SA`X+1>m#K!)ve`B5b-RK3q|B|42FoQsxPWvUlo)$
zS@NJd;PK0P+3gGqjW&8T{?vmuzL799pSs;=^3YClIVwYhxeJ%-KHIU!j;b)#9xIZ+
zCbE{f`#6(bU^WrQ7erPg1D?r0lrP^VkWE>7UjbU2pyclk8$7*hwXFnI5cecsV*uqc
z*za4iLt-{!?krR8-O2B;!Rf16GD#914s*l|d&&S-mjw%PLbK&&PKzYADD#zF@v|Hm
z2J<UGks+ub{MBYVsvcB(esC>s$DE#{;G~71N?36;X{`*{eKh15AHX@aZAD|K*Rdw`
zLh*L;TFeX&iVZpI66dOrwHxOh=~GVOU$w!wog;|4{56L$W=mp1;xJ#5u=%mnmpXIs
zi~|#+a|9scA{D%^<k$L_Rfqry2cB>(TDR(2b+;}(1x+TynU{5%lcgqxXE`}g>MCuc
z^bL=f6ah)-G`2Afaii!|b|b0xB=X<eq#WXgupZ_c5z1!n8@5DuinUQXM_96>)TWo0
zDg9*!Lu_zDlb7~HdO~7O#4+QJNYSFEr?-X0BM)Yhw(mm6E;3pxBV;L~PG}4fMp2I+
znYjB+NMT000_paPzUCkBT@|weZlWW)!4p7WIw~a+asU%Fs#!{4!jvQ=RZ~pWw|7)G
zsrE#X_eq}K_CwL5St*L|5D*;YxpxqP-8q1e2s1<ViUKf$f&|c;)4|0`;lLu;y0VWi
z`P9{i`+5{(%UvKdmBX)$AW8C{!x=MGIHAvY?u?;li`J_!LwAVahknjjmdOEdw(gJt
zy1+`<YSc~*D<r2X8W};ohm(Jkaq~Fg08oZoP;o!q>{8y9Ds1YFwG|1_80Gq#GZHAT
zBNn5TeAbTGyGLHk@S!HKYkc(mucA$!8S$|~jaK9}OF1VxH^}|52Q5v9hq+NI`3(RA
z<CY$Y^c+9pnLN?*pdQcZ4^B%_r)RQ}x7PVygP|k(U=oldM}80>N#T5bKqz$jI|8od
z)b+<y$a^GGHIZWdF|DS0H1%;#9)J|5qI8G2P<Qi#J`f%W3`PUvaR%EgyHVTo!s*|B
z)M=Sn2aU9vnj+KSvTAvC9e_S>EE;n=Q|_7PO9A9$ynefPtM<6qsipn3NL4#d?P+(g
z|JreUWcu!@icL$70K~nV4z!r2<Xz_uB@2i7PIkc^JJ#O%_g;T?V3&}T0ApTi9Y~b<
zkwtyYPfZJ0xISa4+y>`_TmN<E#Ek44v7Q$fk{vFIlz+jgRrso~#h8%o#Mm30G32LM
zPI1joM}r%%aqj5+YTc<AJzIO@_9+EJN&Yc0=E9w;fVDIPNmhau-Ft>KCcrtG-bTh}
z0S9lh5s|AEEci}rP9M{H&vA~SS<NGGsMT&M4S9V6I%<QMA9XSR-Lnrtb+F8MM|sv2
zi-drb&7R=!xkg~SDKwD@)JyN`5{DZ|kNrT)7|P5oEpf|#lV<#6?<O*uqVWo!@h${e
zMKv~!1l36#R`3E1m$nAYFBgv|=v-^*v)hWZtgK6ydwX?K49ca%I@Mefp?4f<5A^!{
zbYI&u`}*dwM@W|3=Dme~Pq4?&5`Mc)-<gaD>2;hgd|R`W@pJG|;OTdS$CR5OBxWIo
z9aVn3$cJJb6h^c2M;*B6=rJ#nk36#QmDf-1s$9B0Ur=x3M8en8910)fe{_}DGcMN%
z491%+YUuULn+zGh*P>pXI&03Ek>vq)9B^>Ee-f(c{bF=`XQwx1<fW9DdQ8oT;tGz6
zpRu79Kpm&F+^xM&rpas(u&)t2*+zWHV+-Z*;GplHpK^;aAB0*%5C@#z#78+BP03#C
z<M#1TFP%Fp9Bu$OW!hp1(g}-$IjRtDE_A}^+<l2#W#*)N?b3FwnCuG3jkBu>f>{UG
z0nK{k6#No`?>Z@-e%KdZt=6*)KWt@i%mmnPetQ65e`)h7z<w(gZ!>`XTp8rN3?Lba
zbunb<uDY>JPdfbgl-eZ|a9K@_xwPFs3!hV88`FP2dXz+PB(uw5cn!+TCM|yJ8ravV
z?X#%A=Tb+V?%kQgH~yKAV4_=qwJ`(s!=hh3%E|(HT#IH5XC`TaNT4m_JEv^*Xs(U1
z+5lICRe>~c*U_qz`-|#V7v`{3#PBa9=Fzz2GF|tz0*tXizNHFO&L*wx5_kY^eY7-M
z0TO1D<_31{d;V2$<J2~d(@|w&dZz%;J@sA6CLB<hig*3!_BKH<@1P*Sg9lEw01vY4
zdg$PI%?7~@3Szq!<Y8U_4@y-OssMk}6Xj0Ik2+RKWRFlrkX|(sJw*=NYRL_x?hDh`
zjwa%`a~`~zW<^D+s5|NC<-$@aj~$NoG!`G{fk6Dj5C-%Q)q9_cgT$>0?CgL=ujc2D
z1I~kbFr!jR5}-1rM5@wQIXRbcLm+QKcK~qZnv09-n8Sp=)Z>6Ist^J|pOjrPcRTLk
zC!_caosmRKU7$M3kYi+~N2@c^b+OzF%l{?jK(1~6o4GFcHd-Ya`c-;>Ja5Bp<22`V
zjUK67Q1@*{5D-+GkEG1l%rf@^Pw6r*)p+gYk2&p6h`~bE^k~TJ>9@%s&)GCd>jl6!
zPd3pKGL$||o5R!$>mPh7IL;MxW&9NUri|pL6ZBMWo-nefzl}q@-qyEZ#v79>(ryW$
z<(d5yR)yi6PQ}-N-6jt%hx&Bow2Pe;J7qjCc<t)l32RjFYNsT=pm+}an2K&x4nAO%
z)6#QvlKR#D-hdy&)xY><A6-IU7tKi6&~_#$X<Bfg#Z*To<8~6$g_p%@e%wGd(AWL$
ziqgt2^@lWFN%fLKN56Gb^uznrr?Y5C53tty_VU%Se}A2iJNB<`a#=Odn)iC2B$&LC
z{n}d(xVutL`t-~(X}-s8%k|D<g>BvX?0}mwvqJz_<GiHIm`Y_yY}v%TOEO}r?*E(A
zxC7-hx#P#cC6R*^{5@3AD?Pp^*V}S1e%Gux02^QhVLrC@d2RDkHB--#b?0X=GSk}w
z>lZy60kI8YW80t1;E`9B`y$VqsP#b;{Al&}eH4(VJHc`wkJg&E_q(%%k~iLU=RZX+
zocD6-F&GJQA{kwkv5nqCA1@DY?P`42r?j^9<Rc{6g?beF;LyY+MaEal#27@O(9-MX
z>(vb1$5TkJld0*_!FPjfs5_z^l?rs+4%FJy#_TK=^uMrxk_qVzwNkj^oy+*+c#NQO
za%C|qE0^vb`9#W7`DNn^P-*~BaEAPQZ)VN3LM(4xHb80o0LW|Gwf&3*X}Y89NH1S4
zA^wAA3Fme6%|B@mE+|wK{@qQ?_Elc9c?Klh-U*L~ya7lK3&as<e_5ZvWM^yju2$le
zs{ZOEO8<<+`D@Rkn0@o0;hEsHzQqSJ1FVo&#>=EuRP%(6-e|gN_pLl$cGQJXulb6>
zdm<D~&bPzu-9!?noFvi-%Bo7##JkRrS8r=D8E0%1kZ_L7ygz8!N9`=8bH6ENnhFQ9
znI6?0-2R39w@Z!#w)=DV&E==&SdA4vlZG>f9ky}SCrgG%`kI>6v%cdQsxT%L_X1G3
zn9y4nd}#RulPOvnK!t5Wm$Z?NUq-JAqrmUrj6xK&)W^gV86o3I0MwwZ!E@YIA|nbw
zQdeRCrI$-}93Dzui(GejzB-oP<px|+A&bsSv%&y`(I}XRDBY2<D~R9lBO;Lo;Gj??
zV~YTJ6|iq>wF6k*SlPJ|(?QH?Ux@{PF^_iB)$a=9uB!L@Ab`Z#`&~T*7i<UyDk)ZY
zqc;O$21auk`^Kn^0Fhx5H*x!);e66^0@ON)S1OTCJUX-(djKMa1;I~<SXiA)3_5?o
zaw*zYt$5m8u&gUqz;sWy(;L48{~wIar<bv?cxPao>VL!|j)^jY_<_>wSwAqwQ%4O^
zSKSdXh?a${`xM|Bj@)$-2gi~9gFgHuM+cR3ib3B}61=a7kR>;MzC|Nt&g6BpUj8k^
zND_!>JjfaEv0)065VqzO{NSX&uphG&dw-!wQ0eR_r$0E1f$#XTGuqyO!AlR?Z~@>n
z0Bx&O%#OPIXUMzMH!{4e7$hzLOKT5>)BY>$#as#0rsiEAeO9gQf^eWq=Cs~m{eowZ
z+r!8Hnm~v&`-9_YzloHAbEGD~$}ck7oKr8B^{KgH?-t)L&%x0Cgttxdj=^sn(1)qT
zUw{Y*d5aPduZP<Gn)jb0A_HS!>KQjMW%d6JZ20%I$Cj+99u>8TzKH*RoR&(P#4}7S
z@((~+xu2pwq?oLi4jR9ec{x&WRE_eTS)F?|#;`I=X~f_EOsV;*eIEcqPnFiBEXee0
z$7i4Lv_)Ub4MO?t*V5E?Y;{oViJUfIdE2bkmNu70H+MC{#F7VuQ89f&)hl&($oFtk
z5VY55x;^q2k3fe*02alGvJ1fBA}(E{3+xlE5I$=FDk2nr^D5s6{|GcF`b(()IfOb7
z7$^Ho<X=wo@8bb}FGh8rd+q-MtF2s4zGtQK%e((K<4p4JQ=aum{@}Q`38CfTFh&d@
zob=@*uskKm?+go`Fm!Q`@7G`q#UGA6E&-q~j8I#Ytczy=fYpBu5dSUde^BN(SAJh1
zqxr+2vmRpm+3f+Khpk<j$v;NCcIOcj#6Roo{W~s_ZTR8&oHLLL2KR^oniSu|08OC7
zpudm>e_<m3{ly#GXqhYkfVkvt%JK(gBB<$1B&#Xl6RFkS4s)h6l{~B%tX#nWED}eo
zxyJX)zhr8-S6<B^n6Wm|Irogy<}N0!=!sf4z%QNwXH(8Vj<-s@BZZ%MLWjkwjj@(&
z|0vgA2uMUVG5Mt>Wel#-LysDOI})wJVa^+l6<v{BJ*JzI!c3P-amC-0q|^Yg-hz?4
zT=jjG^t~Ko<>HA=U6^pZ$c9!L<B!kwMV?_M9~vV`Q=?))MK})=jhP@9wQTn|QZV4e
zA^B*DUto`a=LWrJsZ5p-4W2I1-&l_Pq*@Gq10EW9{NhO=%o%#tx3x947Cgh}SNun;
z#<G|o_>7akagNbM|J1>lXI_(|8?}ip{6S;H+?oST){H%+H#SHt?QQ6PVQq?k4iG<)
zt#}L&e2dl?`)s`zs&W4q*Z<v*7q&sr$;?cW7S*P&2_$a4$mf&2U=KI<(E(x|ziM2`
zoY%TqUXV<&H?XNiOEC~ORz5cL!VOXKOAo02kAJ*kLm|zI`HJA8+pZ62n2L|C?7n?b
za>7o4eqWM^n;rZk^XoHjs{I8bb)y{7APZj(MpfOci@*B`H9zh1a(UUfA=gcg)h{A*
z7nJT8L{&D4u`ly%tt%1BHkdOsv57yAW8>C~#+NnFez&IgDcVGXwmtr-M7}cOVdX(Q
zN6%wI(SlL;V>6SltG=2~j(kn_6>7(IjGo*UQz(1GbD!WQ{aH*kTZD<Jv>JJ)w{|r;
zpO8X{x>ppQi+FY1^<%D5GtWq2>!<M&3-1l@=&q(}0P~Rij<2}PpKmnyj91v$OZ}RY
z8iP0ejnLE9^oguV{bAn@faR*DlHw4ylC7P4x!ixbONiwB3k`RIO<Iy30`r@Or_UZW
z?6pfyeQUD4Yx?Bq=(any>bi1AB`}XcQo3`kh?K_mSu|>h+&{^91<>$lg4Fiha~f~4
zO>qio8a<P=64v$5GkO+-^3eZL=~gG7@RfE*&l%r5U24tKUq4o>p5w$D|7@=zM9uZN
zr&p-=&yZf)?7qvq_w||HG{0vgrOhOMnfe&EU|T9<F7)=IP_zOgh&E{?>?M^`tKEYj
zTOK*poU7VV;+9pzcfMDYFmq^E+nCaW;y;HK=cGe4j+V*mG4kbv4Rb%~juaOjx_x(Y
z=1N=1QLi_31J64DsnqU@em)+NEUVYdhdzce^Q1qMqHl*qkQ+IKl?>hbE+84rKV<mb
zdV`tu(WU!c2qpoP_m3nQa~20~^NV3X4}z!o^u|yLBd4y}v!LuJY?88Q&7F=a+hENn
z3~~FmD8}-!NnZmpntC<!RnnrE4h+9+_E%v!7@dHBdN-F5ti1n*XUkMS-K{q?IxXcn
zTjadTk9+RpUN`{4*&$=LocYdRlDB^FuM-h?wfW(RQ>(f&rxY)VtLR|%xt=d7zE@p>
zN93wUC2qQoyyvMig!7|F8S&vv^PJ~?)-qsp&QX7XVY+sTbE>oTmj3sf+T`f$w(_Sm
z4a19pya3qb7krEJ=#hryxzcK`Vhz6M%_cI;SFNqnrm#gVU6?c8y(CGkO4{M26f3vM
z;W_iw7Jl(~nUk%;B53+G)Ij2~^x?3Q<@wfKMfr+|D^y?iM@{_u61uA#7Nyy>Ri7#U
zSoxC2E*x0B{<Mk@xlATZ4ZVOx_&@!~tEz<kz;4JMOYRcBYYHjUNN)Gmkm?0FinFJp
zL3a&WoNX1B3X_Sk63@x`NEi&C+<y`a3Z-B{)v_8R4BL=>wXUvYv2Nof=4=D(4`2??
z_v4|R&wEBTlIViM0#dDFL-=LF@B+tkc6%7zR~nH*)vrCbX`MQAv$n50c$4UtslGal
zGIUsMd#A@d@oICu)<R(cPv*4$I<GT#gWtR5pzUyn-I3gA_u*VJXz!3J!l*4hWO`A|
zuazC;>R4h1-P||u9n8+DX`<d6vZzECg`%?8VpeNMnn2$ID#@}_=}emjpW)82R+wZ+
zXGfg48s)#2oE2ZzQb~3E<o9BE@^sUFw`>C-Y}?}C2T8vmlVWw`bSZ3dI6<{P=TvG1
z?L&Y3`S$$b04;r8R}kpgCH34M?^^k+*6MCWY4BAoM`-^G#*H3gg+12wtU(Cg#J2&G
z>9XacW~T2~Lu8|tTHJ98wZsZ&WfWT69ogrjEJgMWAJCJ`7RpVosmHD<Dq9bGU)&34
zT+UjH?fg7lH36+P8$HhG6JEQ&+o7xy?+u_8OvMZMqe2&QmII$GS=w!4zeZPhp{hN}
z*kuOD$FF3ATXSC9DXbiNCuCFjhviE6EvKQm20u;{23N{VC^l?9+Koo0`4h1S6*#5Z
zIQYI`t#Fnm2<8eYu<OI}>n8Pn*q_mKvkdQ`?g*SS`YyuNFNB{J5?y(+de;e$9+X*_
z2<uFUTwjZY4F``-aRzOP#lpVE-Dib#iN!l9C=Zg43uJ>g7OB%}9l)*a008kiE95{b
zcPWNB6M$(uEOVedzqCPGAHwRcBa(h}lU6e(r+1yd*x#;+3t~4j6PMdnqwsh4Ju=77
z5;6_Fet5!qx;ab?u$LQ}Jw1C^DPAJ{U5Gt`);~pNq7o0ROFQh;xlpQYW#+-g0LoM)
zY<6)Nn2B#aQbDT5ypZ3Vpa8p1M)<pM*<8GixbtI+^vd*pQ%I))4$#@J^2QHl{b^F+
zTH~fL87VKh^+~lVG^r(?y&1MMsB$=aIx(#-9eJDlnrih?R_K>G;(o3`xS}jTOC*d)
zhRI>CQ}X^;LLX!?l6J449NA|9eD>fxC%b~a<1oJMcF?ACFb^o3G`LoHI;6;BDm~y`
z=vIH*UW8SA_p~CNrFGLzmemROVCsf39`yF3Dr`Bf!<7s9T?c4Z$Xnx|6Dcoaurvwr
zi4~tfVOTFO;hSHSg&1%T)Z^mgfQ1O+igb^RhB!gXSRm{KTNV*4UfloxKTZ@{#LRNt
U-D%c=Phqh%RCQI#l&l~AFJqAX#sB~S

diff --git a/16-bayes-mcmc_files/figure-html/unnamed-chunk-27-1.png b/16-bayes-mcmc_files/figure-html/unnamed-chunk-27-1.png
index 10c5dfaa27aeb9f22376d9162b459ae1f12f53b4..f8d69a025f757b5fe8374512d350678cc218ea1f 100644
GIT binary patch
literal 44703
zcmd41WmuG7+dg{F3=G}f4bt7Eq<}QiE#2KQNQu%dDGDM=3J6FIDxHFK4U$R?-LOZ0
z&-3i(eZTDe;eYI7f8aQ7?pgPh=XI{NuC)?$v{do1sj&e7z*Bpuqz3>HH~@g}V?t1G
z{73J)Q7-_UM+VBMe?cG+7z~C$AZTc4=;-KBC=>$&0}~Sy3kwSy8yg1)2NxF?4-XF?
zAD@7LfRK=oh=_=on3#lwgp`z&jEsz&oScG!f|8Pwii(Pwnwo}&hL)C=j*gC=o}Ph$
zfsv7siHV7snVE%!g_V_+jg5_+ot=Y&<KDe{oSd9oTwL7T+&nxyyu7@8e0==;`~m_3
zf`WoVLPEmA!XhFfqN1W=Vq)Ur;t~=Pl9G~AQc}{=(lRnKva+&ra&q$W^7rrGfAHXe
zf`WpgqN0+LlCrY0ii(P=s_Mgs57pGv)Ya8BG&CMPdZekTsimc*t*x!2qob><tEZ=@
zudi=lVDR|yV?#qjBO@bYV`CE&6H`-DGcz-Db8`y|3rkB&D=RB&Yik=D8(Ul3Cr_T(
z+1c6K+dDWoI668yIXO8yJG;2JxVpN!xw*N!yL)(eczSwzd3kwzd;9qK`1<<#`T6<#
z`v(LB1O^5M1qB5M2Zw}&gocKOg@uKOhet$2L`Fu!V6dpDsHab#Mn^}-#KgqL#>U0P
z#mC1#d-g0LA>sM+=PzEoNK8yjN=iyjPEJWlNli^nOG`^nPtVB6$jr>l%F4>l&d$ln
z$<57u`SN96US58FenCM&VPRoWQBiSm@vB#_N=iyzzkdDZ&70EF(z3F$w{PE;mzTeL
z_pYL%;{E&gA3l7jtgNi6s;aK8uBoZ1t*x!AtNZxzV|{&nLqkJjW8<e!pPHJQnwy(j
zT3SAT{@mKy+Sb<A-rnBP(b3u235UbGy1Kr6`SSJa*Kgmxb$567^z`)h_V)Gl_4oG=
z3=9ko4h{_s4G#~GjEszqj*g9uegFP_e0+RjV&cb-ACr@lQ&UsZ)6+9EGqba^b8~a^
z^YaS}3yX`3OG``3%gYD^Vr6CJ=g*(3tE+2kYwPRl8yg#&o10r(Tfct&`u+R&_V)J9
z&d%=c?%v+s{{B7^i99$sI6OQ&IyyQ&K0Y})IXyi+J3Bi+Kfk!RxV*f)y1Kf)zP`D+
zxxKx;ySw{YYY~pB0W7bFCcXeb7ysuEl(K$c4*-mSn$iP<pn{!3tOd<(n%-XW*F0*W
zbI;PpjWgUT#_9%%iIr!kKa!gelYL~>JMH5m=565Ty2((daegqZRkOQ(y&1eQG`lam
zSrqDV)-klNzZAMLROq|cek3b=G$cGq_*h+?Yw&sLB@~sJry5`$wyan-P<Hyjmvlfw
z&E7^0R#y74mKCn2XJSlj%~`?2gH@YpK%R|etj>k=|Ml=PJb3x(zI;JJy7s<&QNh!c
z{pCmzNw(sJNG?hAyVl6N{<X}o<NH1Xj|ujci%Zk*6xsGewP|3&ENRjGoELtqYPobT
zlx{V(`Z<_i_H&+lvHCDevfZ_cORm4aVoqbtx@O=CUT#H`Xl8}8tHGYAfeT=)@Hkd@
zMH-p}NDYRk2GeK3@&cy=H0eOO3NZqW6(q(A(n?~5V?6}BJp>m}tHDT8(PmQ7Bt+8z
z4o0k6CM+)()QJ1U){4Z4AE*)k5~%wb66O~Y?{Gt5%(Vzso+!S3o(kln#AW)y6yLQK
zsr$lM7RkyQU*k&HsnhU-@cFQ6OY9)DD(fc(c@P<$Ukti;zbg}k%TA0Uq@GLd*5G2*
zs-cwhLnRUiwkwXI=NM&dG93Y;m3Jv=Gg=7*H2?gr29?PC`+HE_A?+>QtI8h}$<v-B
zR!qd!87NoNalP(kP3!-{L*BQSy|u9W9k-)%*=_U;ZH5)4pA|x@j?y0jS<sqIjGX2U
zKWX64OncEP7V0-t6M>WVj{|lv#R`-Ip0Iq*`Gb9xvihP=db*h^1ozaxl1;Vso+;_F
z5+?+3>9SR=tu6YJx2+m1_+Z+&{H$mzPd{X8Bb7GPr3Azc0#=uK)iGq<^TrP+GJdD!
zWI^T44(U@Ks>PNO0?Flv4sX{Kd+@Qyp>X%(cw=qZ-KLtLUdKwI@pPk4@LP2)!<LWe
zTCWa^DtNsJ)}^moEvYsn&Z>TOk?Z^5x~0=XXw(lW|A_@XM7|;n7Ma79LAx2ioph}k
zrvcMM<yc(^USeRWT_z3}&ZI!@XAHQc9q$i3zJBAW8*<LA1a?bDo6#BHjQX1|PnbBj
zdrF-=293BRbGlZOVaGR$j7-j3?Ef8=$&P{CM<+9YJ(#xA6zgs4bmr9ZyWihiPRuS6
zKk$RZ)WB}{W;@1g{%DqjM>ZMeow>gD`+`oVnc2OwJY+;#_Bj4h{R4k^Hq*Zb1M#}S
zvRNCzhft1Bu0L#`HN)4^I+OfGPLq|R8XbW{|FYot1kUtFs5G1j1y+>psfm3=`osuC
z`k%Zdv;Cg|_W$up`Tv(dzYQSlP=}N*QR!4gIwu!j<&>iw0ZHrYKrxSikANh)Ff~4F
z`Pxlry8d2?Y}&&gzG@*u$`ciE2HWLqx~R^xnWS%#c*q2ZZbLDMcGvvr^BoPXmaFuD
z!7G-l41MI%!;k7qO&pTz^wxl??3opW_FR#N%V@jl)xIq8QIQ>mRI?q$9P4asa#>eq
zL~cPk8@!9{4-RNGqFMhw{!c!t<2zRE=QIJ+n*S~Ey8oMto`J5teblo7jGRBk?Ztza
z_V@AsnY#WfA<nV?VYU&-zv)io**HPE{>T5Ys54Nu&rxoN<>jUJr>uI@|HGwkdO0nM
z|J#KW-Al{=w&^6}|E*0Twru}m+aJZ_c>lv=I!WSxQ8M+t@RDeae~7xjdF<bu-M=35
zpGWkY|2(Ej{reGBGOFc)PE(Ol_2cfYGBch;|6)sgFH96$I^%nlw!UsK%ml@|{pxMy
zB1SO&VGl9*ziJqgbxU+oGcq7yb}6C^rhWKdG%aR(^Qs}Hce*<6<gTtX@^NP)nndb<
z@vv-?h+;7Q9lnlXUIxJ@_p7)6^XJ_@K^>EnND7lyCYG!u1`QbG4}0_j@p>piSn3$t
z%^W@tY#*P_yv!C-&H^wpP=u8Z!Y?}vak+0=qvG<`t+$gVYeeOz)iAx}#vxt741@Ac
z;2g<cTxx2y>||#c9(Jsewuz^=rDy2x-O8qIvBe{EUz8sfVP7*{>V9V53MJ@myAuFB
z-tF{snq}gLm^$ponLW<gM+7OUbMX{jdvbd^AZT_qf6+WxZk6A-i3fSn+%Qe)Yv;ci
zGGc@4^*_c0di8@zz(Mw7XY33J`hh@{%|ee@Z^mC~zugWbeEh%oS}wDL#MFj77KF^e
zwyJD1c*&YBr-OU5^MUZcr#Q|ja1Bm5YwMSn7!ja5xr$11H5%#Hb6lEI0SkNo8Cr~i
zUve1o-pCZvX5t*0cq2K}+)p%5H5-f1{-61d6cXg_6l+5IoZrvN{?XeIH<%Cz->^#L
zGaWI)`cI-i@%vGtU?6+X;;&s`<Z@Nk#!cS6Lazs^B=&U?3SintjyUk;#^6O^?T;Y*
zHz<uXWu9?{xymb*7)#Q17H{x{gWR(Lj_P`JL};HRm1S^x2faiLYND>rtUetfQxalU
z{Zl>IXkN++x5xos6{WQ;g<tCX<_}o^E)CS8fa?2wO_sqUm@tMizu<otQR#gkhoiL7
zs7M%$NY3O#t$hCBds0k<PmIAaN)WIRHLm=hdI5vNxS7`54-1HZrgA@<{}QvTfO+y~
z_%%>=l=1KCMhBleC=7(GOR@kSioe>{GI{=O4WDA)@Ly~8%l)+eOH2;?8hee5BR0JL
zzlEqMcpML~fRZ`?Aqi|WBwkGB;s1E!rKV7Q_SfFL@s@wvtH{l<hmGWeBK`mBnpfxc
zLxl;C4F{nalZcJ{L>yzjnzi`|bY7W@F|rf9FJ6+vgIZ$oz?K<HU13dR9BW)CW6<Ec
z_riH4*~%&4`$XgSj!~Q%myV89SNMz;H18i{{wqrrOsoG3UEl*oqawcvs=`cw8q6uM
zm15tzWiTsTFe?C<#%;IzgBUPvDSAg<sz%%4L)a6pU7(sMr2kL7gM)$vU@wS#>i(vp
z-d+$Anreje_ZvG<8xJMih*<9j_b3$zIV+U(4XCF6_h-%mFn*5GXMaDNa_qN-LcOxa
zG_p~!BZq5X{@N)VEE&Y)+{xNZ1kT;O%j^In{6A8etbjZ6JM*ynFn+dD*x#=WGe^|5
z`hTG#eHaAOe`cfRe+F3Oh+_6Pvprh{dOR=b973O>zaz;rJMs$BC-ALDDb>8WV*!Tz
zk6b$aoajMROgF2%;*4+#l62wI(<Vnu_%LTU8!B>i@FJ8<Lu^n)p52w)L)6$@|9FNM
zO$*9h0G~SM!5sVuWuKh7BWi9u1_!-N`A2Xd``RueTRAnN;~@Jg1m&q6aJQC*a;2HG
zY$9FXXATsmvBg$Sh|r2bn<E!11M2s@ih~9%C~*-uBiZE5L}mQ!e+6<f*(>l%#scz6
z4_~?4=wl0P<DVrrEuf>chka^*W0z%Q{E=oZb)2q`iHweEBP(4qfGh|v`B1@A(3%UM
zqUzJ1lP>2UuNfWxc-_RtI@U3Q@+8TzwbyLV6(tJ}EdBm#9(ma)%9A^4#1M<F-#y~l
zKs!06Y1KhO<9HSf1%48L0peK(X;mDBY&6aboi(r2reN1dR-ZoZQ8xvtZnM1)xCo%h
zmmYFXlOhw!oc3;0>!vB|%nz2*=TB3njTY{?v$5F;JwNZbH%F7OGb{cg)%A@%W_8~Z
zau98Wx7R>lfgFuXig?Rn$yOvNf~tmaTyDvyoMc=OzPsiKM6RT95GUN)vsM%nYnFq<
zlztTXWS(MOZYT=wbGCmT(+ES8d|}49qae=l*TY=}464zQCshI0fxbNAkaN|PM~(^<
zmyBg158)67b!nJvbU}9&DR{^fLH0mhOB7Cr0s$qwQz)XJfd1tO8tl#%kWVdmJ%NMh
zjxnbT>5l;29(Wa?&x3BO7hTMfRZz*xl`R80Rr(YUvZcj=<M)?GvtT4S)a&aSX<-0B
zfhvSzfPgJ$=S7XRG<@=51Wh0wrU|*#OY;h7b9meO2o@Sr8>r2x`SlV&_TnOtio>tQ
z@e$k#{IeYJ$C$zL9Zz=w3KrI{d8?KR**!+=M(Us<?T4*FgN1aM2!>H|n2|7Su<()k
zcNF}Y75M(@Gh(+>3_}??qS{B~nk>b7n-%;j6Fk&YPw>sGngbsBneEo=6Ce{0dX#Y_
zh|=a(SG56JBSENRK?({9=LIhL6v(?#s@{V<NIYBWF@q(hE}+#19i14krXbRAc0dLM
zoE{ouy={QxqRl?xczX?Lllx<+Bg?l>(nEzYZS*JyQB}pqbcVv$&<x<{z%r{PozW?6
z2GB!+$n4BNA)THjxHOt!yFzQW{Q}hI%UciYMyJLeWc6pcz|4BgsxjakGe)b231SfV
za_}Fu3nT#kngMDL2suxw)i}lSid6#14SOs(^mZhsqW}3&54K4-&b;Jfg@=Z0LLq+_
zK-?GY(*K2lJgwEZ<p(W28P-iry$`z3$40s-?4ioo$aa*D!O<$|1<`3fYK~F?kUvUE
z3oQYDJNYe7o17gI#9>Sug;kdrOuH_>X|fR<gNEJ5K!dRAz2$#c!dMK5CpDQ9z*6dJ
zde%SZ1BhXI<zg&g8V%OVTDElznc-GYTsXw?YEc4>*ok+}=S72Pot~V|06EODPoaUQ
z5G0QRH7Z89p0E^De55o0ISYD#K0Wi}1OZ~KYWI!|&QZwj2laXdyRHLneT?uGPNAHS
zPw>1JJR<pfq37?U;Z|VyatLWS6bOqMcgZ|`s6LCr0+@m1%6pIlE`kfY`?MiZ{Kx4h
zfJ`i@UvgtOH#<Cy?A+Ud*eAG6pMC=C7vZWOv~~T-$Qu6$F)UDTE`7ZDMn9bM1~~n#
z*UeiDEUWjm9+0^~&SQM>6G6217$b3E^Y_cN@!@4H2frU1-WGaiZ$qqe7*>wz(GbP=
zb`_oLOEU{201UA)rHM#%r1vH*B_zlp%-*LPXch(2HfZ&N&qbKZ5c~k`{w?K%E#$gJ
z8eUAq(0N7$AdNDUE7kYa{yO5igK;CZlO7H>7t@7x=^eF-_6EnaQKJAR2z1Ch1I5gx
ztk>bO-#X=cH0nTd3bflY_lpbIfKJt4I44`~YMjzHz^%|90bv@Hz&v{$6+8-jSdNX@
zUinA291xHYjuZ?dtttF50>E<grC#7Y%^)^nZU>4+BgRI*@?CU-&)@kH=Yy_&{&9c|
ze65|TH;KYEu^#w#gz;cMMNm`>9=`~>Me=}x*vkeP%0&G5BXur8*Uo<sj#^ACCDd5c
zz^yATCi7V2nI=}bGq|B}-V7`+@AE4m?z;_%!Eiy8h-XjA*tVikLBz`R*wU^rI<C2|
z&JWwHV>n*@3;CByX4bVou5E|YQc~}m{rm<rD~|(m6(E_yKM}~B0T}!+g1(x7kpk-C
zNm}DRrKn<w{_`bLGvO2bPO6E@8t5eb8$FL?U_UuvPO%P<D~UxzhpjF$Jn><KO@CFL
zUDG|L0=%@<n>sT~*B_$F=dxBq3kr{n-u(d;Fm`ckXN7t(QUqwjqVdaRfDZCoEta`3
zJ=#24>EtPpVX4NexU**|Ch@`m0^^=+bL#kgj{{3}dD!YsZwLyLWvesTew!C|MG(YS
zEpZ_zlC;x4FN+wK;wl&pLTs<{1g49_F$(+lQE6P(;4X~FA6M(YCymO7#?QUp5N<4>
zK5u%G%vwJvVYQbKaX){&8#M)^1v#ebJ=TT^zTMYQqQUKDf+C{@we|xs&Ufn}h>0v`
zixV34;zueg#MUJ=t3d8#+Xz0a>&In6#IbLh!t#erV&bhj8VJHc4RW4~g>XfgVe1Av
zBT}s;9+Nu1XBA9Zcmi6|_!Uh!FWYzOS%mTj;_l3#AqVPUO4g|Px~S>~M3?m-1MTU)
zV*yB|J8_qLj{z(6$nx~_v)-UC8pO@WS`$iVXPMutkdNTU>AFE7AQ=vC*qtAmwgx`D
zkQVrQzvgqqF*DOB7|wsrJ5h;h7b(N;{*@&4hiQCIw7Uv@(^to0W`KQvj3B&&mEfEc
zkfrjk7w_j%vja_ZU)GL@z*?o>sGXxVlQL~zTiVmb!F)i?c@ZM9soz~Jo#G((pm5Ef
z7|r^(M%@|&Fsv{P0GV}=S;Gvkd)l?nl<Z~mYSIw{&ad;Cd^0_r$XE=;0-9+bs+od!
z)Z{>XRJ~=>KQI0NK)FRxKJQWALqI}@F$-m>%Fs`&#b6Pd3RcTrumnUS0*_vH?NDL>
zBk2q}!v^hok4i&z=Isy{TG@wmogfK-p>V0h<PLpG{cLh9KF>dYEp&woEHjUdWFSa>
zai;Y=t>CnD4OmzLA?2ju+7-dJSK>;K-vJC_Rhmd>kQ_M7nRY$8ryDBq6CX+A)s;lA
z%t41B<nnZze4C~H(UPWYG*-Vzxdfp7Rei6D%pMk8efwD1`PfUbtYUh@SHS}a3}_Y$
z{*9f!ScIDWwpStN4=~^YV;|l#Xfm&<0TO%earXrQkC@z+Jp*Fp_%QG_f3^$(%Bi0y
zr-FfN0`H^;wajz?YTwiILrZ=BqrCJzykmhgoqGU*!KLbtPATrUr#`MXar*^2DKc;T
zj;R<Y@eo(-k(RA|EXmp^oy0MfC!tx5&Ecsy@TP55T*Lza35ggKK~V!<G!XV7hz<wA
zVti=c?5ih|eE49r_x2S~ixW4v^3*Dbw9pOq^<A6b34Pg{+2RS&t=)2kdn0cJfZGSo
zSSd9Yc-u5`rx~P)Pr@5oehQBWgJoVA-*Z{NIhyn5WIFZaB!XR0(5+;j@hZ$o0dY~5
zWH(OCOgEDYx%?5qU|QMBn*+^Lgg(Y9IB4XIZl+WDXN%d}O<PFYJRQQ`89;SY#ow~)
zK5uOc>?!e@!G<8Hu=f`TbPRtxr*O7ieOJBcvtnWbh1Cf^ZPtdpXa&M#q$Q_$`611X
z5NpaUDqRXn_zi$)KEr~bW`rmVxM+gx;nxQfhnB){)uqM}c7~qx`1QC*9M#r3&qdE6
z2*To<KWH*t1VFVmY4j~byr}xP+Yftc1OWPXmA#!W9xbMh%ELd0f6n?z7HpM41Ls|}
zTC(~<)%k-1KHUNg#l=c49CpNBMYaIuCh{RssZ0Gpa|Z@|md@(F2<$`Z7xJ&f7`Zj#
ze3WI!p|5fMi%+04_8>CqwYG?rzwP0q3eyJc+q`|f%SV{qv!qb?E-Ag552^<du(4%V
z_F=Hv&qD#`_>tk5m!(Si8nD_*EZSf|H0?O6z6RQ$XLlb#*D2@-tQbF|(XMPMRE*nL
zew)=xcEBJp3(S`Jpb@oa_OZX2cwX{nOE3^Gx&NyDy`CVXC?j@iY*Wq+-+G2x7NPpj
zb;sgm*Cl5ZK?ZTvsOH35JTft^Lf&6*q(u#iYQwKM@Rh#bKD65hA(@PAOY1m}rN1Al
z>e?s-%uO&~UAG;%43@Rrva`0DEp^9;NP1Z_EPN~}o*mAaPEjE_M1r8m@Q7;n>D#_r
zr}Din(@wVEX0!bkQ<-!z3#^v=NtA4xjCO^R%f-e*O+K5^F|K`Cr{liq%+*HW&;3PT
zFO5e*<UcY;`jk$=HG*%x{M`X4)CzX%2Wtl}y#KH?MzfUH@Vq9^DPvH<CF2XwT=J#3
z+Klu=VHon}`S>uo*Gdp;_6$6&&4X6VF3EQ2T8oU;F0{b&yTOo;bM<bFo$p|4YAQ5Y
zX{I{ww<-&kGldNxjujxT>rXLtsi`|!pV4WiAHlSy(o}$pnA})huC-hGvKbrmr8M<M
zcJ^V!^ar_qAlZAOmIz6%>SEGn(AHUN!bkI#t^Uvf+R`g{Io2dxMgt*5{ET+dqXe41
zGuX<yI|^n0@N}|Z-o`^Jtk!~<Tz_6SpkrQ$-tJ?P62Nhcl>=G9@TT9W>b+uV!E~nG
z@nBBV++!#Ng<0alN*FvtpDIIgld_iu7<^%5*Wv`Mq|E5<o;FWr&8u_5nZ@Aw7N8C}
zx<8$ND7YC?W?I_NpE-iBbN>*uSCVux&Ha^j^DWmK-rjQIH_y_-8A-s*)nem$#E0Gu
z*SdakOhjhm{e7JgOZ}h6Jm)4}DgMq5B!i$1cv*-g-PB|FoFs3yBo(;qg<xjqEuSoU
znS1Z#QXct9Kh7Z(Grbd?leUS>xtS`Jt;FHAWO%NWoG^`xI2*b+&HFso!c@Y@X~#`h
z(3sq65kLx&nBx_^!4D8u{hbxWS!0!pXYjfR$L1Ne_WEO4k4eqm592O#-!?EZ@$qyN
zcKTNZfC#-6@5*eh%b(MV46fS^^17d|xY3e~YdK+@K05kGV6(`9eZPwX=Wm|}e4P1e
zx0@fQi+e-l^NF@S0b_Y;j9cv%@J<$<6jA^6z2_E9^k7>52x2kLPrQZ8j?Z-+#cb`H
z+@bz(G^Avz7d_{B#^fD65I~onl<e*i@0dSvRgY=uqMLZOvdc03j7)t!sf6L)?)&xy
ztK^>c-8bc^2HJ+}Mv;Nxi_%4PVatPQ?&uq`r$FaVKw^n9>!N@p84J8~3YaI+oX-Nc
zFLEKT<q0q(v7Gld=K=&_@`)dZ#ayE_2tR5{OHG|xZrky>>A9InjRGBNZDNpuVR9K8
z`NMB?%3~!m4E<@s<?C43!kX1MB}K`2NQP>FVk4=&ci?^LC?sAd_=%HTS%}?=LxcI4
zJKGyB+&@PyU00jdtL&E(7sof)DPj^T2Q8ve_{e+}(~NV_$aW&dItjw|0JH#yAU5H5
zoQ|;Jy$7hw&3gUVHK0?d76&ZVo=rZ+1wP^BoifAaYhu-%Au~dn4WoA7zVz3(AqPGd
zQBf|x{_K<~4%gZrefn)wM(=M2wMR!@{UjBgg%eNqr`f%ZvG6Rxm%r;k*wx_{jl|bl
z4%~DzGe6H|%3h8lh&%vsc2-NH+=UVLFC1{A&j9WNv4yu>hAb9hu#)0;S3F?2JqU3>
zbX*W7t~ASxq*qXP0T@g={N<RY+#NyI^8@qkglqwUa$tH@2Uvpp&(PxjXR+K0V;qAh
z^V*axFhp4_*xORHg00#X9xQ}uk8A>=N{8Bi8<XOnwWi-@`7cjRX3FI|MFqtEul}Uq
z)tOSHvh%GDq^x3DU=$OH%MDK&0y?GU>_d6tN4({J-$1V4%Hjb5^?udNYnB-Bu^~#c
zbQv5B$x6VJa%(P3vKz4U45$=K0GX@Le_5wbcYGIKTA)UYb4G~J4pyOod(IsSE)G^a
zjVVCdwn^fgi$rb(7_pcG^pAY7ow}Z_AItWS@~_)$Jp9u#U)s7{+-W}$FN>l*vgIr|
zf&cfZmax&O^K9ib>#|h`KGKCH33R<eFqdAf&7_pn<fEY3Z{FCIiCQpln*(yzkf7mJ
zUnFWOt%g`TIr!yp0O#bu3@PK<L(5Lk1QmQ=85L>|1&=YH14by;=VT{|2h3LnRnE2%
ziGgE2KLJw?S!9!K&>}t;aBX=y1_HPNQq$_P3+M<jmHinyMA)?+ePLy3&7T&s$}~-8
zUPqTygYArTK{p<<Bu=(X7(JB+j)z5+!GaN~z#rcYS{lG7pL-C)Baj{j4htrQX?_R{
zSR2D_!bM0`12m!7NwJXZ&V;lTZS(abZP#w!6g()hRTt=ZL7?S9^zXX>EAMu@G6_^G
z@p#H@sQV4~2@xrhHexU%REQ^64D4N`DTp*4#?pNUA_nNJ4<0m;^p1RZFV>ShUP?1C
zM<h8@J8kznazTU<U(FlR))OAC^+f?@^$ENHSomI$xN$NCI8HA2NHx6}fL7@%8o&5}
z+KA*RgJv2;tXZN?K5~J>n*KyHQ8tzku_Lj_`t5Rt;@xkOZ*`V^sGY?8h9qYOCA>@&
z=4%~lvg|_*e0sUUQaK!#1ejGo;k`x?Xm3!rC&Fm0_ultk(&0QKLAdj*gik7e_41E@
zT$QxK+cqkBl$CGQ^`<zl5#*s(@#}O`M~Zt5=Nv3?ba7cGnl+$>9UF!-Cu$Bj>}m)r
zsH|*v*M9DOj~%AyQvb^2f(p)Ht&W3_jM$bq@C6}hViwi`$~>y@4i!3xh-1^SS0E!?
zHr&np3t&lsj+m=7??{-#3TU760Op5~m`IG~YXyvEX(qx^hYDT>lE6v3l&Yxf?xPJJ
zD<#Oe8U<`$Jg#XE*!*cUgOIS%&4!*YA0AsytcaufYZurNfLxjX2N5<lxkgZ}vw08T
z@d&f{g_+Vr{&_P#ri;lptDCv9yw;e_WJ_n1@^Tmbx}AWPOK^Y}2+lqy_at5dGXkRd
zO$^p#0z~v+Ap^`E=fA4X_(73#_kiUJ2(GWW=uU>_{l;Ox=RoJEV<|ek5i>Y%-Mn&n
zF4aiO7v_4zaFHtoM@Mvj4+{N4XDx_@X_+@q>p9R>Q@d1kU(tjbG0Mc+P_mlkyi3^D
zJ9G#U*lxF+tfpt$WeDCPPmk2?3w(tID)K-#e$dOfgydr_<p|B@5lmv@vunSwnTlZi
zvCIIn-hD8gys$!n&@cz$US??|e%00lrhU+71oUK4F;<Kx(2x5va~|9iW(RcI7a5><
zAq?kONZ?SUg>HD|^HN_VQ&eyRox|X~Jd$t@{>KebEUJE=4i7H>UHNAdg$3_%LA7jz
zW(KhMW?*5UR{KVsHnjLdUfR$GK6_@J@=&LvGpt<a^H|tIk>2K6YU(v$o{}knneoxp
zB(MP|SZ0BS5De>VhGQe!oyvkqtvcg`i&oXD-7kNAVNID^l2VsW2Dn39Eeh5i0tg0^
zJV}XejvX(tHsk&|xj*!_@Qxt~npbb)_J{{U6XAD6<oz7f)`NyjU+P&9bCn)shDX+Y
zQDe&<v_tzz!*Mein<@Bmi4bwJK<@z#LKA#>>A`5X;SuJ`t2NnK4e&povc~_Z==Q>t
z>Xb7O)<|DCZz&aD&3e+1gGxxir8=~_Ka=lSws`I}5GD&1k4TW|x|I#lOdeiYied-O
z1<z%d1;+Q3JD<iMoX8k65_ogrS!EyfOO9#N*mI#j&A0;qSFOr!H8B*LcF=-&7E<~h
z&g!rbzwu!u*-Nh&uwdI}LISW>;`GSU-kLtZ(j91%j(47i#4S|;NIA+>66`z5Pj<X^
z_JYF$W$lHj>9rXgGha0U0uDk&Lwz*`%XQ#L`SaSdx#G^+p+(+5{`i5K%WR4-W7}(r
zW%UL)@vNI3-WHR{%6J(<gCc`Xd&g~1D^-sb%P$1*i#{PT2eL8#$fr<)889q#9;l9Y
zFpOxIjHykzeMf}|+fnsU=_AzVq{2kDHwQCrNhxSuVa(76(ZnTeq|!3H1UkNwA@nxR
zmN4BDou6F_$G!jcv{CHS85k0j1wxX)7FT^n|K_w@+vg3|mczMYX@nk5u0NHlhwzD+
zB5Zm6jW<Z(>oMnRlePQP!mz-V{13kG@w+jQ00qp963YL~AKodMq2&U!c7tzzoG;Z2
z!MX1>hI}rn)Z*^7e8o_MLKi9MHrnXj**SSYi3Ub<(F&F^ML8S8m+5-K0o7sI>IGJ5
z1Cyg2V#<IH3|Md{4KQ+yg;dspsoJ1RS9Hw#V9`hRC3{g@qpQF^fpCWtZO9|3^PX_S
z<(c3qwk6qVX^OZAaoo-!&6tHZ#kN~nPk@xQtWJ?@VRQM4?)j*qJG@KWV~?4=GdekH
zagdCGKx2zr21z$oa4aMYga7;$5L9{5VacI;>bbX%UrWi)0zYNE=@o_1MI{>0Rz5pl
zGND3vWX>%Z#OeI5ZD3`U2bQB!R`i2MqO``uN7^+Mh=2v0NFA0#Bf5e~L>jRJ@pw1X
zOG^P>Ut3v`$xmoK<rBEpJ5B6^TD>R_19yh85xl_CTT6AHH}c%gVlozcpPos>bu^Na
z>QtS9e78qtL<rT|F-gJwSWw$IaE?=NCgRF8Nb>Q7)AGU-7MU1Zxj=Au_WHW?6PGi^
z!qj}5uqPyly=||*=l#xuvjHt(_|;_ZD(GBLPQzZ;_et9|Nv2($<A-vi_Hqy|{nnG3
z&~M|D*XM~qkSDt71bO#a_!EB{i9?8&svekD*|P^*j(;zy<{GtimDi?>^eC2+_|Vl9
z5s7X#sXC;{o_%JZImk%npdkTon}xzsP>n|gVR-bIf3CPM!_&brIbEq(mrk%^^E`0d
z4;BdZ<jl<ylj{6U5JRdnsb##y{nxKAz&VS^OJ`faNC4nNzmusbu8dEZzfdq+N$Upq
zK=XAEtnLFt>cI*N=(rlyOm<L=^MKo>aIp0g&vN%gaz55S>DmM%!&L%9xijq+jqPNJ
zY<b5k%moew6T;Wr<_39ikuvJ4v*}ttUawpH^o8ayz-A{}*Kh=aTAiCdZ>#7@4{dLw
z3hF1p`wQ1h<`$D9Q+?oCNb7z+C=RvxJR(Y;Ho%sdY`Q;fR(~k~OA&%;^ZR?By$ozB
z6aET}lweq+TBv+1$Ot(Y9=TT(92($9!o&}PKQ;&Pl`sgxG{A8GBwQ<II9=UX#Z>6U
z8O0zIB<SV$#-PzQdks+$**l3uqliTHqnL<v?HSKi9CtzdwcAK0Ry(X`09Fby30%=Q
zIcG)m*rnQ`=B`Kl@yh<<Dq1sozoaNNJbV#EUL^{4^S9rB2Wfjgdh48DaHCr?lp<mL
z9T>T`Gs;FsofmdCmM4{ooPL$ox;XsilAxcnyv`WxY3CoToI5>j4^;XKR<q6StGv1A
z=uPd>Ki`jr6wIc`L-WnKbqCdwuVps!+zG*~*k^Ayn{90}hK<(`n2wp;7)0iLV9OZA
z@|8-+pnG^o-arVYi?jEQQw-FLPIkSIjk5a0hUdp;cVwqghV#Y~C{hoCXv{lWujn=1
zv0J4=3%XZx-Mzwa>?gVKWQH^A>cOoctazc~eZ9}xT^Ps06!q;rBA~t(Zk-Lu$rFYx
z^Sy<bn=~w?98mPY-iig@dM(~Irr^Mb2~SB<0_9)dZlRsq&>^zIBq}d7<>M`C3H)GY
zsZbL#geRvzAN!)Z^Nm2D@c}wA;d^(rj<-V$+JeWZ8Q?|G1``Ww&$DaAiMNw^zh@xx
z9gCvG^Fz?6O$XI<*lDDPKBp%W+~03H1d6nQB7qmv);*y!Oe0TbdqQM*idwa3PaeG4
z^=d1|t5qSod9lpaoDP6GhC#@b4KRS%);0m+`aEzE@wGaxk%wmUUm~GMVhBPUvjQ_x
z9?fu;!iQ$tQ?8%;t`SLad6@`tB|MA1UYChD<Tu9`$^?b&*84Xz4WS9pup!H#C=l2Q
z+{Jpn|F(RULV}RK_(1{*x(T_enI0sP-WvP$Ha90cnZgEXO{19)Saub6=QVCR7wpG(
zVaxNWV>>>E#p_CqzMPBG8|~)t?%TJ8iu`Pwc-!-Ju-ylmiOTiW9TzQNMAobCC6@ph
z`u$yqelL>m`%m#GvCankPl>nq)XY_~)&PSoaK>Sm;9W7dCKUwTVDEmE(`!Oi*txV7
z#RQC0Tu^i>j=3E(1Lm_JfggH)=?hp-_ExCjaW}cT>OS}~vml)Cxx<ytI-ZdGFbW!1
zl5!_ubx(!G>gVm|AfzagfEup|Xu{pYxTyc6wNG$tPjlD=9awyiZoS3aiB>}e-{zYJ
zRrd++69Y6hi!)Bp>d*L}B?B}JwWZ;T^;tD3os--Gx+#EJpnAkZj3CM4j_J1rDkXJ3
z(fD3OdIB%NVehc3!r(9467}Q<e_wM|iJk8y%R6yT%^P?}h4WTc5+E-6opOIan~00{
zriNbg+3;tZn_^v=wC}P3Qwq9ub&DW;r0uF6dQh|rDA!ZFMw;_XC7+;)v<jo#;+Q`U
zKZH}^Exg)X3>Fz4gcJuzLnT?}8>!Bhh|5LXns4LFs$5bcD{+LyW)0DWU=&mU!IUL4
zyy;EhuXv82^-k>D2g}VI)}9`FDku8%5r=^y8rwkht^yE-0WF^dS%y-%CIwrcVapt8
z)oPf4s=Lf~{cC@HW1W0Y9{|(%Dc&9Nfl6Q)ohUHD---x=D7&7YF2Z^<HdVKT<`u<*
zjpD=jX4X5iBtx%R4z73CfXq_cyQ}>J-!~4Alu-TpNv2tP_(_YX*TbIpdXl=M5fJ?|
zEm$Qz5#|$4^8m(>AUd>jk8nC!NDu?$+*<?r^{6F$lQu&(oK5_>B};PXN@^P9nqFe}
zHUFyA(*rkl2t2D25NC?<Z}m=YlVr+2Xz%VC3MipJ{6)j#z7&1lyT)9_40p(J0D?+D
zUXjDSA)*qwi^=Wj`|-$ulr)orC^@w2_b#?Sr+m-*-7%4fP7xhzt^qUw6gyZBn~KR4
zpJK3QtvcSK;9!AXHED_S^+&=ds~TxJA?;2fgP7`8{MIv63ESV!*oKqBq_zP5m0R&V
z`QCeBI$y10N4YKTj|}Dk#V$$}6UZ;exMlotfmHH<*{6k@k$E!tR9qO%yY4Olf2t7l
z$E1MveOGdfX88s&2_n#7PRt-a0IojI2U)ul5ki7eAc!McAhL9lO&ETL;4*Tf{^;x-
zGDX~Qr3Hp@n}QsgE@4msnyc{Wect5lP(SIOY2&$tfVp^)ZPz1X6J;0$C2%$=(`FbY
zY)aOE6DFU#j9f<&@GV#aqlK}uE{9LFA`XoTjK)FL6;gnTnFQp#>IKyht@Xg@^)Iw$
zu1J-t*sOFz8rsWvvEHuFNHd$`!+TMLZ&ng`r_%33;V)Vw74wNH6P~;cd<1Nc<^hpP
z6;Hx3emD6vd|Ip$Y5oZfO6YbXEdmf8=h0J+EYk(e&94}xc##4MVFc&#2Nkyx;32&}
z5!<2juRd45F_9Khj7CJnHh{c?X;5@s?&USkpsqh`)5TvcP-r^=Cqfdw504R62?Udv
zU^zu(f2@8>W<4d-Xye$$Go=KC(TQz;=AJS#o-4p;AaI^C3X()E;-(Xk^s=Jcr!)Z4
zJ*L*x{X5Wsd)pOD6Mf=kxEAkry^m_IFoFbz*l=JU@&E^dZc&Q5X=#Ck{Xm5P_!EeM
zW(v;0T!esL8+rh{{=3$FmmfDI*NXYkt!;9F;do2)`4Ts(ml2z2&F%Qe?J!qccb_M$
zw+PRCKu+8heC_mOQK{951{%*y0PE!WqV$l+BKD(+ThmHEU~IGNfbZwNh6fhEcSPqT
z3BMV`V*uh3)DDQO>#WE?wQ+cmAaRVDwN}ioA3Y$r(X-51K497<@VjR2`Kj}3Wj|9z
zub&&}6fl#F0C{P%N8QGG`3C;?wiRD<<SdEb5^LlyW*gU%m%F@7p@JXb!Ay<}(Oizy
zb8TDkiuKH(2Q<)fHToPMfoc+!@b@>1&qv-x-~Pb=P=ODt<Hi82-a*1}<d0f`TvAdS
z$=NJv7Wh|QkF;HQMk(q*W^zlkmcOV59|Q}H1{E2HQ{%!Y4udB0vb@o#8EtC6aF}6d
zw3!ei81A)mt4m)AmT4;Hs~rO8!ap+?TUcp9aEr#z7P!F1EzG`KVWQ4$=Zqh9)F9n_
zB<Lsg^3k!XXb4v{-iet&j|;v|CH?LikUw<?OADw)N4%U<oha6<tY|R}TWabek3C>r
zVr}OAf(A6Vky)!K;RKM+9jnBI4TDs)`4-7sS+d1G;MG|tT@o9KQ`rKeq5yPLW5(SE
z9WbXVH0zl7X^+syH?qW=qJ8tSA@&e4ZxEjv1x834KoIj)KxaSwIR9<3n)s3Myg%%u
zigG5a>xCO$m`H}%TDSmaB<W_?TWY~r<GU8jOD9-G9R=8}vIUTRH(fq0<hsdrNfLmT
z-7tmbwa0_y6{-FPJxgw}VEn|!q8EaYhYHvI_Hc3{4xqxI0+v5mVlDn;el6z>n}xrS
z0eu=2{Vg7h`7+*JC~9#`Y0JOXe2EK~ip#8|zv%|{;=?^ke6f0l2AqNG=Q!}3XV{<c
zvxnWH4a6|fCx(J@7kD-SO7c~O1j6V0tY|M=pr9#?CM#=aVY61CP4}$P)cdvygAgGJ
z#T78OOGxV%_^OuxsjQlcJ-I3sdSr$t1dz{t0o^J}w^=B|n!f&sqq08LDG0|8y%A(T
zu*BHzIj(~Y(I@W`A-d*7PM1ln+dYMWW)j=8*F@hD>a;+67ibS8?DPrq7z#i7uKj}M
z2O**-tky^Urq1}KZ0tjP4X58cuXopf(P6wPG^eu-nr-fd_Y(s{Mst(;HH@>TB7K~{
zNvt-&maMyyaJP4CK#%U*sorQ0BPz6Wt?<)0&@~4}P=|L#8`yf0_Uxl!&hrE`sF$v8
z*VNND4AnKU0(#T;@4jOpcPW38X~c+zzX1>*zGr9@W8M}(!a7fxhA$Ne{EP@>QJt}R
zJQ%}$SZCDy9xhOy(qsb3m0!0GlgA8y#50$uTj~#E8aiI-doepQ{~jWP{yZ{Qbu$+}
zI9>!!p#&JFJl@DKjon0&OyR#adaRs&gMrq}-|&;xdSxL1eMG-%4G8R|5_$5S^d0Ih
zz!UEn9@}Jm)hjslwaKhIFAu4JSj2PD7`@iAB^`(GUam6u;nxk8{)DIYSyYHYwNF@c
zHKA{i{jdW&X{W@cZ^lBtf`>v@5!Kx(QTTi@$L9=>;$6g0JwAM>mlY0NKvRlPLFgCI
zS!$u#<l?`nT(SgKl|O#$5UsOiB2NJ5bHg8>f3nW02-CC*&COW{wn@6eBo1A;uOSgh
z<wljdfwDM)Pvxe=stgdmSYVk!fHHNt#@6wTa6@!ube)OWPH%gW)njj$s!Tk@oek+P
z5EwDQ1%xp;)YyX0ndl4Oe7cQ|&U@FcEEwM~f3G+G!t*H>xcxn!68B?$;M@i+==n-w
z2o+QCcL8*Vo@y;var;2?sNi|cCoVq;xrHSqSp3K37HjyTv)+g0!HM}ZW%>5R@lQ+^
zoJBpNF^k4;aDcta$=$)Ae1MA+h?AO{0)mps`5~oMGo}^O^QM=V(o-4pBcP%{x7cSI
z5P$%HftuuP&~9Giz|oZtT;bmdp6YZllA3cAMbGFonSwf!_PNm5o(i5n0M5CN*T#nj
z7d{=UYg|L<T&E;>ogZS`Gs8{MTGPP@)D<^|4+v?`WMGwLSGN(&SGQ2fD~b@uawRqg
zmcD_OO6UyrK!~@wi%_K3s-`$nJ_!$|tc^6pJ=8G#N_G66>!~5v)WNGVei9((+=(ua
zm5+^;LdpcigI?IzoD(%S(HC-N`Jf1FRT6}5{MS@T6B0yMcF-Pvp@D+y({ONGxXual
zY(j|382?gmG-?x#e|@Rg$Z!_2@)P1;DGIbv|HcAjR1AROnI;42Yv?tV%b^Gq>4|GS
zMOQGNW?pjA$VBfQ`K%&<3XLE7gTvKx5~F*uxtth-@;b&Ve|R0uQM}k6)GSoo-D0rr
zdO6qC#)DyO)pmlFE%pZeHY!=P^+Ouz8S|4Q-`RcnDUTpQ@KD^kI2|wc<fTaN#A(U*
z!V@<tc;bMksH$p5J7Kd3!B{r?%ZAEcyPl9pkQX``;LxUd7ZT)MB=T#ulL%3sD4w+z
zNlgXd1j0Xt<#tl{P;5Q+2bosnInC~Utr3iOSZj+(-PuB77v`&qKY&i$wlX*o5I!Zb
zY}0-n%VgK&BiJNX5VP+U5Cx-=d27&_UH<rq2)vSAHzZP`i|Ga=^Bm*5k+9mQ2TLsw
zEFi)Fa!%Y~2=yyGzj>A1__RNM^4Yu_34Crt7iSo7Rk7&#%1?`sY+A{??^M;L`^y^-
z(Ke%8um$##)4k3|aKtdJ<q#c*Q{w<RjbQTxPj#P!_qXVA4H&Nu$rSmH8_yRW9{c$I
zK~7v>-^b>$1ksV$k*<;*vMrx4Kv5U4l)NimJ+T<C5(xDQPHexe+HmM977{3w2Fl;@
zeukQD-FCXh*M5h9WlZ+X&6Tisp2q*wS=F5Ot)zb6uJn9n2kDSFKAgMzdiWQ)3isnD
zIB+FbfeTiI9=;a`TF|?LC$1Vr4|%V<&E^2Ufhs(XE>kgX4e&yWUTb~DV-A{_M%@yA
zSc$w@n;Z~UM9;_PhMIC9D-}vtfmMG4f&jFC0`e>_xt}bpethxNdc5Jw&UJ_s#s~_q
zH(s0LFe9PYx%Qs<wxWmY{<%2R{^{`pZzF}nXFrXi^SjPcvE8r9#v<3a@rD4SZ*gg&
z^c;M#XAQ9ffo$MPZKO|N%`X`rR=-LrILpAR9CptFCd!+2r*vDOa|l?9+ZC6B?I`Ed
z{X%2;F6EZ{3N%zKG)A(wkCy96f7%RCI+Wq7QXeTteO{^u;kx^3dBuDOk!$vA#?;=T
zZpIQ-W%0%szE?37>)B-uixT7T0GodS?LE_S(x#UG<nWvQ3^61hS?UZBu&gP_(Mo~h
zT3(tEA$ZrF8sD)yRZrRsqhEo}<K5OS&zc3LUZ?KJ1<t(C2Lw{Ye49?sRV;>!x8#Sn
zll)6%4H(~X)PMV}0v1es;QDQa{k~FumQAy}Jy0(`T8-vz@c=;vL?lAOo*IpL&zbR)
z1Wu5}dzpWL4+HYY#7{?FjNbQZ%BKUO$+no6EYrnwE~=HZtMix#<AWSbh5v!hGaBz<
z5C)|o*)O0T<nwEpFp^e!e_b&GEm!p9Zo<Jicoehf^ro)KmT3925^9_~6P#uE<q|rq
zSRfgM@P!jsLU_cad4}<JJzvdI6Jx*5QU?}>zyc$lT~2_xJ3Tl*lge6f<uOK~;{{8V
z`1aQ0c}##}Wh#!uIw9yJtXHA?*vN6>?UJtF?~c@ky6e~<oVA*h4jIWgKbY-ecp->Z
zgK#KDd_koGtptOLK`G}Def1ao8Y%?K?+O!rp>Gn|ZnnJ9GkF=HYc`A^a>#jc=Mm5?
z@L)0pf}6B0C&z*=1Szn{JcB9?T;c06RbSX6vPJ>EjtQ_AI=yboyMgK31}LoNKJ3Iv
zBCNjFS@VFvuQHTasEu?)=U_>6huxbhL_`2qE(nWMzE0bd4f{5xTP87c`!?mJF7i=?
zxf!h+ao|>vTdRZ|J|O-vF{n$op&i}&9xFl6#4#zSZT(l8=8_vcM*2-ZoYz_JOR}q8
zQYOHq(%Y|l$h~3GLopB*uV-BBZNHh}*J)EUU}JFr#*dha>n*aA@8UKgMDX9Mx~VaG
zi<dct^Oa?e0A6;X!WJfV9(+b8PfQLBeN|p8lkLzovW}hkLO|^cvtp!7*4)RUkBcz%
z6ijBzXJk}f$FtYa>Vob*eYkedRny_481Z5bb90%h)wxU#VhSfTY2~==r^n~q_vbXV
z8Uk@DQSaX8d;th83$CK65Qhr{R|;97^veW>D}zMVa@2Q0YejfL8*Fk8G0FY}P%kW6
z%cG?ei_o!<Rsd8q0IIJ3-X5PqNe?v>04iVRo_}WXatlL+X(ttg1mPh#hRc_sFbtYr
zoxT=?(`yJ1ZOKeM)Y<||9;}5$cAxt(gf}1EdanUoh$VoO$9&N<v=WI9KOwaCfNrCk
zjsA#>8fytO;N|C&TSd(m$X}m2UN^~aeO{3FxOjn+<l#Y!cue@1xmif~_0~hyk0~`e
z6<H<PUCjJRLPF<Kz-zk(`_=cjxj+Ob%ah!`dg<sSOqyqgnPJ#(XAc8E3R<f`C9XP?
zw5Is8rr$mcu)9A$$HEewWGwv3=k8@1(_39WqQHrMgKi4NH5fmmKmG&As6JKmxy|pA
zhikB)uT?U%=4pvm2eQM$^iPsRrFe{DC(?Y?J~SbRl;|%Tr8IVixI|r%9YcLV@m3XT
zDm`qKHK}Rd5<W4DdKLS~#oC7C>$IrdjcBT-S{Lm;$FLRhl_kD6CN}Mj-Cg)6<K=Fj
z?Co|^1fUqX+MS?f=g8Gu|6prmn6`>h+Q(Jo=sA&wm*UmW#HpPQrl{~y93i1^#LwSS
zQC)L5^=Y+_PHfoO!6!b6ey%(dk8t}_lnj~IL}R?Gp;Ie;ly>b`oPbw@R0ffQ8#RLz
z-HFPOsbYoCm)6>vdsS2i`b%}W-RGcYLW1@$pdxV8?9=Ta*0|>4$i{~@s~C1)S3+5c
zmabUOluQYQ_`z~?BrHTX<$QuK8}0p76TY6ogY-WMZ7ZlSTkD!qB?!=4HD?)V9V()2
z@JXo}$5%O_vEOT=R<<BbxSlBAc>3_0SEFMiAz026c9lOu0#&w7qNQ^0ILW`zJ69k&
zre@V-t^~EBZH91tbq*+eKZqPM?ub6!<vCz7tf*nN3upcKoWfj=*rqtp)UJ2rVJR6C
zn~AicnsiIn@{{d8O{Z_BqpCEf{K5CQD``#*&3~M@9{eV0CQ;GQKKeDy!pqBPgyg>&
zc}DNQ*Q;L2-*!aZakfQ^L=ToZTuR5hDzZvaYPfkgtD20IBf4s5zP3?*tI9mq;+(AV
z#j>Y~ZOA=vMfqHOobzK^sx<RWW#C<yoPw!Jm4MpGBQGvt!*#N`#WBaNO^-~u&2spS
zpvD2j3w<eJ1!xj=x21dg3TJqGsPH`&;>CB-Z2hiF|J_$O7`GWU&tBEM(lEKJixN;0
zXzVHb_g#Rr8j!Uq83nc?PMe%h=HoG{b<wj8i`urFq-L$?Val1rXIuUfkPE8MzJ6kp
zCF`~RH`87zH3u<ilngHuMFUGbC%<&<ZbWGF+L&(Y?R<EOBjonN0Ai*m6x$>|R_8E;
z*Gl}R#hKupdb9ku>&cetThWY@2G47(<s^=~6^1jW6Y`xtkB|2;>p$K!Y-T8iHzgRo
zK)JwY1rNLA-6G1LvkOoZABt;fqi0^HaQ|?y9fZ?HEAurjqBweJAjQE!d2q^r@`y<y
z^%LfXsNWuA)%VAfBA^OQF%oon=$%T4g7K*7`?APUDXj~OFY9HE+4^$}Mx)lQ!KVH-
zfno#~pmVAYV-nC(s>U~Wd=o%`z|U|yVW)>6D2)p78A0Ag?A}we6JoWmkRel9p<hIy
zYSpzcP}Q`mU;~`D_kE|QwJx~hi8|_Tw{|UhA(*}<&wLC8x8XnOlyiBnIB31;x<KkJ
zr%l;aUJ(Bh#P<RH%|&6pff?*^E+lcIp_mHZRxI-}u;fF96DLi8Bz0f|y`wT(5PfI}
z&v0cwu=Jd7oEkneybl0heK_d&35NzwC%*Vf{5TUoaFGMdmjgoZWJonFe^jE=H{xv^
z9&?A)#7?(DrT0<mXvv_b{v4O<io*5W@Z#S_Xt4d>>KD$%nkznf44@6R<Cu3a(Op1$
zIHwzsv&_+(=FKdme)!9H-<<vsf=nA!8|jDy!^?9h*mHXQ1hIG^h&*;0avn}6)-P&x
zrm+Cz|03!u9IEPqZ_lN>I|Y>PZUh8WO5)Pp-6eTxX{AFzL_oT`F0FKTTtK?J;oa}O
z_j~UzID2N+p1sfPHET*VxTe~^erng*{E-mc88Rc^sA<|WqalU`G3I1)XejXU6hTgs
z<d$4^RH8Eu!(13`1YrdW2zE=1Zld?Vt4EFex$u%r3`wj!<%l3^{^T1Sa%VBMo1Y7Z
zqHy{MU(#pK@XvOFK-;=F7N|bBvy@)+?XV^F^Fqeqv-5m3&hYyir{}M>xW`0sM?lXn
zl{Fhf8${S=C!o_-PYW<xo1a=T<*aOdf0x@(b;i27V2I*Zh^5=z(d|F~`@q};zl=I0
z&%eYNpAl6rM5^HPnCC>e69l**@P#I*_28Vw+!+Wuey|?-Ieo%E5+7Z6o*QK}mc}qJ
z$Ia7IqF|}jUek<ntzy6k!ORSVV#0eGaGB(e5g^*Hpt2lNe3Z(wGluz{A%7=$8}WIX
zd`Uq$Mk?PsqqWnNV8p{qt=#lf`}78XQr9p})#rt8Fq;%OTn03#V3NuoaGoAIGq@~Y
zmwrxtz%yZQIM*(7T^(Lr_VlqLsI1^tEx($gods1uaPWZBU-e}y9qS4f6spJD>(Y2I
z#zVe+pEF7sc)Op|{za*e04kv6P-M(vD}loHaB>Ph(7sjIf2|W6Nu`g%20`Vb|8>!9
zV^vXPX-{7`1u~JW##U4TVB1YP?=F`m4;Jruk0h{NYckTt5d&lQOKjya`L~wKkWafP
zlAX@8KJynKwz>DEN@)+StkenH{^~A@GUAB_OlXR$-r`K*>L7tudvT8~7+g@e>*{Aw
z1rDZH^8K^|(IqttOy3R44vABX46<(WvyPH(&TIw1h_&F1-ylNi=$M~Uf11q@87kgX
zoiIieX7vz86l20v3raFWNp`IqS=t(Y{v>~A*-pl0{$`dU#ZvW$*z86|R9j2;A05As
z(-KjN8R!VZG&=60HI%`}eb`ba1UbunS%s)Gyt}wpV4;fvRHi;jJ%8#*EQob8XnE3W
zn1(9f8|ojAb84;kzJjd@-=3vRilUds6nCdj#2yujSE-8VgX4YNgj+to*gu&O9vkBd
zxpQw2dnChwFy$W9c=}M-vlq1cTd(S;q`JrXkw!jN$lp7<jrk3&30Zs=rkUsD1sZk{
zu$&9X)$I|WL$1_$SKe0HEv=rN@mzUD{<3eLEaJ<*QG~U5lnVFfSltX?1Z(eRofs4)
zG;{TaJA5t&Hu(u88|(gv$%;U^(I2v#o_H74eVpv9o<5+Q%8`lIYz)9ReKdG0b%E5!
zGyGe%&35jm`hbo>254>gQzH>&K<dyVF;_#(+@Y6Co)sJ^N#bn)(k;P8;i@Q<4+Qgb
zETUYB*JZN5q&+R0rp2eV6YDpKKlnigJ>)=KP?q?Lx>I&ww~%>C#scSFWgy^G0xlW`
zM+G4NKOWWc!r3E7^DFBEzpYSx)MpWH=c4O2yj~WpKP@tL6<5_oW3B`XYog70+h0Iu
zM4;okG!B3JL_O=_=EsfeGlM6DTk&JdP@wZuhbULSnXgA)1j|fGf1WrwRU?wm(Vxcm
zKKWapg)s?0&?_=_T0Ybwap2cC=K@U#h_KW#-K?Pi38vGh7{KH#xk+^k#STcn61IwZ
ze#<gj)CR__ix004#jI9+C|H<6y3yLBCWaaJmTE5pPBvmz0k_uAfO{3rsUO+tv*wh;
zvc3npuqsaaSc^pS#IQl+mxMA9))|rj#M%{-1$h(WhgiHT#xP9=+MXzD?MZNr6z%mD
zBfQveb+7<sAHUmkUtRE^)@C-`&*n(20ie-1F5vn&U|w`>Z1I=<Rq@w2Sf76Y^5)^r
zr{n3G4Log1j}=vjw$0)TZ_B3KI1^hMwuwtccYR={-zgyq)ED)4T?5{UXSo#uRPKyL
z;~5c!yL22pPM03LbA*iq!y}ylaYia$vIFpU9~ELL_+=n(`;bo$KzVR-=6IVb+wK9e
zZUQ-+sMj+6;R}+nRzOYne5!mdcuf&cXLmwy+P*K97O=PTBSRNh*75m;0M+}xnz1Za
zir>;~M<?W!#W{-v+BVu*ferk&L-mnKMeq?QsK@oQ=DXx95Fp^gfHNP)ZiL7_?uJYt
z0gX&J#3taw-q6L{dJLfL5bYKBcp~Qav&*eR$}WjwpU<0zMW2piM4_ZTQii5ZyE3sv
zwji?DI*}2-9UewHR;qVF>h+9HU()gVZ>tT2>|%oX>nMZO_GFNnmQe)7tq_k{m-Wq;
zS8h?e&A-I(YX~d}p!yV&eSk2f+JL=FZlmx`1Wo=}XYe9ieB#o;dllpR#BQXF3OQI%
znX9L+r6ju`O%J_N5p6wT6Mzea#T#Q0DkRiSyhB8BIp`~K`8N)V_|i-Pt#pG>Zm>pn
z?2k<aid}aEiYX~SkX5#JC46I;1J5pz=Q3e}{C^7;Tq|!#Ovg>;tu~weA%*7Z%r(FE
z=(35%Ic>M^leR=6doX|EXf}COq`k)lMbnzs?2(E9>@=M4Bj-PP21meTE*GAr$h-_8
zN%RZl02k64R%}#Jvk#6SFq^Nm{DT8!(khdpJN;6|fq%~zh5I04^%^CId_d?`=qVX}
zg+5=(7`JqKPIuWXCPdMR@!nuzjrE#)u{){T$(vr=DFNi7nw_4u40J4c4Y!D!KSxYN
z<8V{X8W-MW${^+UnKbMV$P->wZ5gm&*bIkK=>i=OOdbZ62`ZBTyckJ8uh>mdGc$l=
zR;u3|6ln@bC}Rk|nUnFZ6j9Gjaj6fY-|X-6&fo<fV>|k}7*^m`dI#5srj&W8uWQ84
zUMBVD*uTwGgQ}16x$o}i#j5x7y$*AI&9GS${>_tXVQ*!k(8van!MWLj;EUDIWoqGH
zNt%d*t)!?e0Ht?fiRz1b2qU920KgdntPS1|Sk6`#;R=0{8Z%b~^M6hh0N-H|0>hkV
zC1f9xA7`7Q&YjnpoDhLVdarhfVpg>9q%;at17zYC<5zUl^oq{hJ4E0Est^7_osAX3
zZXbJyiu;<j?o#&DxjeNA+UKXdR2@uE2kA3ifQg|w&X8Z`m4K1AXPoJy|8W`O>CKjH
z!P-W^)KIeE*ySh*IMBu)@+3Oc`7E!EkQP%x25=FgHshI;-6rU|1@o;IY?x40?;$Q@
zNmn}QjM4Vl{3Y%77{D=~2d6J-!U++9*sQA;C88Zu&Elp%T>jEGaa=7K=7^(>25Wbn
zDdiZ@_WvR^scd9FpiR>qSzIRe*%nbzLN<W*&cNRc*$^J#_<B+Dc4Z6P5FDRgbAEx~
zZPgdCVCBk51QJvh0#HwfBK!Gx-8Ab%PMPCd1*WVC?P-4dMl{ip04r`tIkzBz+Tbvd
zCJi~ri_tnciuY?BLvkTqe=>ctXufLCnKCS*y^QWzFWih{Hr&nTD1npL)DQWSl|ay=
z?@}g`T575L3gkDKkOw`tu0LR*_4(mfVQp)(ZeMa{6JWeI7f`*Ho+=yC4Tcs^=PbwC
zi$+4UeDiUBc=Le#l?x4b6$Zrm)S?g0q?X@Nj)A9czqv8DEc@W%5+@SppLU^PHBJ3q
z5o8|>oKc#DeeW7kAo@#EuaKe+>0~q)EdlWokFFayO{12&R`Q|N!AA{z<41Scy_3=Z
zsj{b^w7rN>ePD5F0u=FeF4JA88GVJVvX4B(Xo>n}b{XMxfYbZsP(&96R-ZgO^5w;m
zRzt6c!C!&>4kV2G4p+J1C!&EhyiE}o(FeUkeAoOhGK2OCXsY2=+}A5d_Q>7cD>PBE
z98`|JlYUB!;%hp(DgQtBb*HKHCVafWvq3n%oBru>pV|%oQG|(-9$93$<*M0LIyU*m
z*}Y`y$*cD@T;m|8GP@Y>!3~O~G*3t;&r09SZIr&O@yfmHTXQ^^OuWaZoAQm6Zkt(B
zT=W$uvC**~cad*)<}`eD`GzB^mfGic%92JBSKqUj3L^S57}v+CiTOG)c-3>4OYHv?
zTC$3hP&nkuIu2pUd;9(J3w76|Y#`9_Rg8d98{Y6XO3Ph|3^N9q@c+JS_!si6boE&W
zds^v4H;dgMb_!h~;;w7VrxoNP?U>*fvmH#?jkeAq2W2qLdA{c(mjB}jZ6&q#FSHY~
z0Q{#YWIjrjcACcS)>_Tw3dg+@{-G}nHGA#)#5iSW5S)7(2+S|$0TSleVfNoZ+<wBI
zu5d<-*Kkhym^>NOS4nDeB{T0miXYOIMeaooGax(c>*$HnY)W~byV=wkNeD}2goqtD
z^V%z=f_x+k91)sDp7e6GR*7$G0)YyKhW7V`WYCAh_-iCkhjsOtdkxM%Go<;N?>GS-
zSNrz$&OGhr)cj^`ZShjwop-^5jwolyfwe;K-|^k!5WCm+Vw@JCx!Bh?ZU6m^d|ixb
zJ}G0v$~Ds8S%8grmcGA{V)Q~;>>o`ASV_$9A;1XkKN=M(R?#^i2YBwdAKM}|K^Wkf
z%78Leda0F6Xg+>{^m&;^rsR8Rkw?SUZCpSI&QmyIN<{?y@DFmBdYf}s=gfY*pS>lo
zrH9~^E0#nOTMY2%?D$a##qYr7WA6DVv0NY!n6ykbcj)no2V(drYyz5CiD74nan+5t
z6u_B%6X2RS)6rKqYS+-$uw@!MnS}d$SrGb+Qi@mCWmZALX@YQTPK+{j(_^NKbc>u5
zJ7kMV4;8AI4CM&Tsm2w}+qVbSV*N-}?TZT1ZU>W#(E`nony9Poi4god7^tU@6wadK
zI%qus2hxLGsm8RgOXvG6)Q%#7GlHZIo?0Fd(<yY2%ZHApd@9#8a+pH6Y_VLv`+7so
zUZria7y9k&O5+5EtA1s;{|LH-irs4_qJj?@^sC`uELf{i<1hAp7Hs=Y9J_+mo!Z65
z^wk=KSjV9rVYwt+Mw6fe=b+FS6f{L1P}<Mt`-6m6iG-W`dLug_mA^NKY(=0#Wp!a$
zlmU_XhY=+W5mAVb>S7!#2rkl(y&MZBV)|lcb7=O=zA;#k(%P2}PjZI&t4GP%+S&Yl
zf(Q32Qb*^>#o2%GNW22tq!rt8vR02;1<_ZWLATUtRH{3}6hKc)*hzbBv%PphduWG{
zAe5mg!dRhZ?QER;$mZQ}BNEu^SL4tLs*{%>gp>}t#bNs??UJ!aZ8S@347Ds1cm4L<
zfcquY8yCeIECu#vq6<(4f*|?Zsa1>A5^Z(hzfIp80IzLDjz;$Spq7&SLzEH{^&+fe
z!^gy3`n220WEQCw!3K*^M^Z>612l8xuPFZ;!>e?i(_Ln#o>Aqo6Q*B@pi{&K`k;vh
z@Ndh2G&X489hOX;b5p%-)5ZIK(KIRs_ilo+%egP_v`}cuXqq05KvfTOG^gs-Ddc-w
zUL4mCGe$Nabor+<Ft^#=V-OWYoExgB%=&TWjm2&_Zd7+QxfqW5_skJ)pQb)F>;Kk<
z({VD#=s1f;!X<VRhm@o@be-o!1}iGvm^k1!YlcD$NQ^VT<zVRQ7{@A?QjJ*AP^763
zZaJ>mg}H63Tw4g6^$qB>2{FvT(Z{w<@XG1!n&I%vY+Xlb@E1aiB}w#{HP4u3rw`hc
zgwSTUR(5#?1~&_INMmwQ$_vKxwMbXHh}l&78n~<CUmi11yLK{)oSH1BD(XJ-nvph_
z=DM~tQy*snu}Y$XXY9c$Idkii_sk!?pj^-?%KTFnQ8Qx^d|Pwal{uCnH?4C243*cS
z^V$Q}!u&`(nahUzMmoYpCQ_%r)Ypi72XN&e^^)&#mOixKDTLOpTnx%>;^^V%&(DfP
zF=dP(_Y%YGjTx$^XL)@^=H3jPQoa5V{HJHSSTI>d1@^>5sV2%cuRmoYkjMr-TnH$t
zCuloHs4C#}tj#t2oV$g$l#1#aWtRhXn)A@0oJqdg7P%j~;k`<;k)3>GP#&^XB$M|T
zM|)_)_6e{t1wK9_0jd+A=jHF1aUBq2oT8yyLb6;1MB$>%I!1#Go0~SbiCfv!p27wj
z{f(I4P66789mwph6SO4~>=Ctd!*8lBkmry3a|d(a&zBrF?SLdzgk*l*VEo$?B?gG!
zVxX7<i!N$?Dz?c|qVrcv^O1_{+Mz66oN6g=DbA=rwUaqQnHVuz1Ooa&SH~ahB}WcA
zQ`bKlEQ^NGszuFvR<_8r`DaQC%L&^V+%z-TdtL{rFhOPUOv>|fhFmE$Rca}P<6t;(
zQVt>xVU){c1Zx>16xP5p_~gsiQtQ69NyMD%9#dTq?5OcgoBl4|M_Qb^x;0$r=14q!
zP(m){KhYdF1oq`)`_abNTppl#`iHhRx{K3pUq~_>be`O}8h7Ff#ZfHY-S}AfK7<LZ
zeR?k1whBnS6g|KT^pP{0YcC0;tS1Xg@`PlsKAC`q_P#q&hJhOD1eltsi=^x8e;9HS
zuC&zU?T0a>YKA+f|598o8==rPW*=*Mpkr8-sn#|#i2v$E!{>AOFpQEXJf)xE$U?o5
z%~5zgSLosORp=fxZ`N=>I^WHuDFPzFT(cXkg)j70bm{Vogy{#uMu#8EmA>FtdG4Bg
z2lDj>)`&W-%`vgZKOD(!HIdD<CGrpVv0xv%T@OWzyrshShqjP>eh5KtY@Ud$aaYF?
zAB*|cjTP;vhr|W_XzYR<L+|b>*S4=;TE)oR&$LxyTtx)Go*~`-Wd}}Ul2?-;LVR<+
zAK9F-Pdl1GMhN)UiLaQT^j`0<4}|U3O$hX}579LaTLbq*2Q0;pDya6=rzWtTW)?SR
z{R_SrYU@<;l(jJ^NPvx<!EqBpYN)q_3R*TD?jq&&Pdw1n>+wIE!q@~)<U1EVnpiiT
zK6E{azG;p$+Od1iwH>-b(%|+1m*m!sKC(3PZ1c#aQbSdXP@8Mpa^)@CW;_`Qocuc7
zLqko2EzoJ2Tsj)d&(cqVZs01)%ddnc>pHqC7^eta5bCR?JyE*k9$SKV#u*1CL4TE5
zstZGexqv?E51w(6vf$Zh6{4KnInlJ8w>F9H-vOz3z<%3W$h_iA*yE=0-~QO#t2USE
zM7^Z1yxZnHg$XmMF<+$8RHMH|)t8s%D|4euCor~yB>kQJU8{cXtwkiq$G<qz0IlWK
zG{lD}63OJ%sxMz0PfA9rUvs&X3D(SE_{s1qbTrlV9gg26KJSm@kCvU4#Gkn|WrH4N
zupo40UofFh+%h>RvXTR~pAhFXOHod*-V&DM7T7KB>b@V|7b$m*==I%KWa|Bclx<#=
zJ<57pJw*xYN^Zb1&;8`-E~j_g)3gTn3T<cx8o7hb?jYo5hmHwqrtDO|kos?~4e%nf
z9ctN%)*c$}|5top3MPjyVGzkB#lQiLRa~mlvxQMw=i4%_SmFacG_zAPaKLu_mG@?y
z9HP(b?@_a+UG;*ULZ%wAGvuTl%xU7J41_=K66QF*@uT`s7c@RmsHkyi=!spcSaC-C
zVSQ(Z<5<6Q%``jOXT*N_pH7*3?Es~}ZNx&3Xy5zQqhX55W3{W|jiq^EuQot<OeDUH
zyLS{$z)ui~FKku3q$;{v$_sA$n`<F3oE;PFfE|S&=g#f?-ajlRH|`s{V8IukWp%aT
zl#WIHn8`UA)&otykDIGL!?;-q;sh|K4X{;j!*Md&(iw*qZo*s}ds_Tf&0Oo&$M=Qb
z49j>5R9DLPS1vvRE{c&-wVYr4Gm7%fDSsJRK~_lLjKs}s&hUTLA7RwpX>eb{dN6w@
z`@thh@<tA>a83U!zh&T|Vv+&|N|q#8v_cJ9KK3)`H(ckh_`fQQGg<CSP1kX>G{=77
z5rnY9icZFi6XYs}jo7rj*MZ7Pr%%Z29Z}w;zBg@VI5KWElBak(iKs;LQx{%!*dI+x
z1vRLYZxjaGmJE_OD0)FT-)%pPSZ3+v7C+UmgqB9fM}2NdYME1@gNx;U8+;%yXk>yR
zI?mZAuln(KG=?GkKeb0CTUtTQWE|z>`wRcnZ{*djZw1|-1pfMSolskk_m2|~QFbQ`
z>KuOZKAH*8V70-`x>{@Q#J~0&T>3i^aO8wCFTp(^PK$CQ?Z5H)oJG;jE6Y8LO!1b7
zgXpt-o7>U6)CMLj-gJ86_s3u7R)YEbN}Y_S#qoST^8a5sV>9*_qH;2i-^{S()xpDZ
z`(Jby_c`OFVfS~-Re$jXMV>VKYu<O1RF|842A$&eg1=rY7r)NsVAm3Ie`Rnpu5%F}
z?tyagA@Xn05fl41IuH8sK%Gl)hk1(q_`U)=W%)Y|H;~G&bIzi=y}#Vf+pTVAPWRL(
zP$_1P^|Sg?)2E=zI=Ml34@5>@?4>rmos*=Zvrq`C232KBsTR&Mux$Ia2=fm8{DP45
z;~=2oeot`8d%O>;%FkorPEB;n3VS<ARD){hJ%<+B#P8JV9XB;oYLmIREE@351M@Ep
zG*A8#g5tJ}t4-Qk-!E~+#$OJwO#yM)b;1QiBJ=1y+i->52kEosOg=s{<1oB}H`zOA
z8Ri<v72cwIRu|7b+hAccgXZ-w5Aq6lXk4@p4VR|x1TR;2%Sz;nmZ@QLKU-^eDZ1yl
zRMd_H`4zts^3Zw`o@vnq42jIOtBRGP`V%E;LI{i7>{+nIb{`wg>lQm#7qH;UlmDZR
zj1_dM|M(EI!B<jA2xB&1XntCr@|5EVEJ$)XCfEv_f7yB(?ljnO>>9!VW{W0ArG>T>
zhT)<Y{*+S)v_oGbPSFs!O1x@w5BT`>zEjm=#KWHnJj7PyBz99BG}Qm|d_&RYO4Uh`
z@psqN9-t|^+k3&F^3wBrRuR^P_Is;dqvGtdA+;P}$-7OtczGw($&3B-^y`*_6YCP3
zG%GbCB+c<`T{i}3Sv2Ld^|AhIj^D?wMM|otWRhGZ*Xck-M3A4jqEL3_39Wnm(q6?r
zU86--qG(LU+pVXRNniBi<AZ7K<c3%#c)3SN<2I)st`>T&vBs~}QUCh8IF(!^6`-&u
zqw4)1+B$-*K6)mVnww`Rtf)tU`cNZr!~hbBBmcvt$*f*+;lC@#r>5rDaA6c($(He_
zX11-c9xn*BD}r=j;oQZ+wWl_OR-|o6fiB@)DPN>4lqqM>Z_)S4)io<NVAq2I-V0>F
z==BToKT@MaOWl>`wG_;lWgi9@)GSejqj~J2&D$%LuSaBu-PRDrqDuUW_3dm2qFIy?
zqH*c*QK8*jY*G&umvJoTu)@M}27GhqI?gE;t$bOac-28IzMn&^<Y?QXQ*5<Ipmd^s
zXURcyRV00<-x~2}e!CR(E(9vot&uvTcFBTHra(k5<sy&oTfEinZ1A|mJ>wnNet6lH
zE?CfSXyXAILdp9uj?i7opv8d*pCU>IEBdptetPEhslO3|$jy&j%zGCIMSaGHX~Fa1
zh%ujk9rVeExparV;mq4>jzoB$UOA7gZY&-%IXk8*ohJV9w4?Ojf8Vc1c$E66|2i-r
zGBB@hCJ5{QXj`xHS_lj4spTv<j`{7sBcgT(4r?YbcEJxdvaN@dHksgVg@7|AT0y#T
zYQ{}vwRXog%LsiVIFWKbs_ojnn|Q<j+Hp@B`+O7OAF9fC?hcL{qy7-RioYb->*g>P
zCI^*y5d(8(9|ZsU?bE1P&N+PdM#8{Yo$UOX3_?0#o_@G+i9TmX@+5kc@mjHH-6QD2
z(@d%W-XZs43-Pg4ljo?SsWp3YD^eKauLpc-D?&C6_NQV=W*`IJ{^r}u*aVWe;ze-(
zf!e)qNDZc8O$ZW{)~W;t5*`T#T{~hEM(xm%;gYq6D`ZsB#RT*wQ4e}+WECy~qgxa=
zP(qk}%vtBI;I`=AwBl7Q#T1V)wni7*iOol$*PHS88ZCc=j$O8c@0Ks%3f>Y&YYNHe
zOdBk(`OJh4n~v(b`G+uJI$4po)!*-@(n-lAyv!3cW|aD~iaa6ChTExWWS_Vp)CcE%
zHWWbM$JaRtBl?tJ$D*Cq?@hMV2^{1R=TAyS#SOp3n+-wNRPjV}l44$32Gm>+zXLbu
zkXCAX(?YaYE0R$#U+>Wgq6PK@9diS#Wp&{?ul`dC|5;RiDi-kiAJTlY9ge_IDdy2J
z<}i2>9<3rCLiFO(VtsbEn3AlLJDULM@$!l81&n1WH;~R{FlLRjeG^Sc9ryom0p7Di
zfHh~7dC&e7_OnHzC|j~)A=$#EqN;P8IL}Hou*}0D{B9ypA|rLxq+g@_0a@5*%Td<W
zfRi+^{x#>z{wM#u2Uc_58FNG;DuOb-JHCg;nDb7J%*Pb?!G@Y&>s(S`TaqyTPi<?!
z?@w;O68$fAi#qLweMbEwK2fMgxxP_1DH~)NAW>!p#ARNSbStPUf(81a#{gIT3ETcp
z)2y>G55F6UwQ-#r3h<lyXQWK1K^lWpMUIPN14}W0TG>{fgnCN#(ku(QJ(>M@5vZK3
z()1%gswRDX{w@CbUi2}N@StW4E$WZo=bN1es^(0D>LLVw^;8<3Nmrh+G;+}`j=iV^
z2mVr{n$(1ULZC9N@+KV!eud%qFy5fjxJ!S=Dy5N^Nu65z%$yDTA4Uy`{;S++!`A|T
zRZ=EtjL5!rdGbsH#(+0=e!M&wHu%6?Kv(gapFebx_}M+Omd59#*DFqR7&gZ6h~vc?
zg?NaDI|S|>P^AGk`dI%Z?L!<ctk%#ei0w719M$ZxxS_EJ9@UZ$yMZl~D(+DbRJHco
z8($V9C@L~;Nq=bfbUn2b)?`{=&5^Bdv~&JtP5TuW45O;GdiWP*{a8HRH-qx)(CH+n
zzk+W_VbB!s*9Ek%Hv>`B?o(ctZR_|#+xy~p*A6=LMvMH}Ww<uT_lX3q<z!0a%@kry
zOH9wTDM~Agxkx!uv(gdn-=Q|eIwcX%ScQX^QL98N2${Q4k*_p+7#-}q6v$MAx30bv
z<=oS}3|C2sV@Pf#W;C_@P6z8f9J+U*PkWNsf63f~pGS?FbIRUt>?w*lIcNvGmTj?I
zh$3IY%;tMB5cBs4Dy!-Z#&{v6a%KY#<|U??O}=o-v_0H@ba;M@Rw+cQ^$U*Hx1H-f
z^H#$-T`8G3&MfqnCu2)WeK_T&9@VDnv_oyGVuO<U5o^n1CQ6Xb$~P^J=V+=nzgbEy
z1YV?VYhM#R5Ee+zz5JLYGkLg&87yb7Nmym6_F?oE7N^gR;<g>!9%{~_$>-d#mKWM?
z?2Ayt&iB#0BNeI>cc+xVn4F(mosbFh4S@8xcKh{~<(S*w9jU%4MiGA<1Sfa5D?YV3
zrGR4nf=Ta`>yaDSd6zjFADFoJtd45O5XY3}b_PDp;EO@!Rx}KS@_&;TbcP%6$0T=9
zDO~d-%sY9)@iZ5WYZCEC;;<UNrtp$I(S-K2itT_2Yh$w!dyDVdwgSq6jXR#QDxblg
z#(Zz|ln`Mh_b@G_JGSl)j#<yjIFz~=5$KF&kB231RvfP{725%!kgn1_LzQ`z5!ey;
z@6QUrkAh*)F3Ib{WO{0xV?bk~;YS0rAt)y^whzmW;bUXpHwkPI3GsNx{w@u%H=2hJ
za~}h39=YOYhsMdo{F`O|*<-nWzZ9)0{tNL@tg44VWi!Z229Ars3I-zrF#s()2qdK#
zciIYeEcxC4?;G4PgY)XHaefxwL)VH_FkJ8q$^7{Sey?0bFrsgk(OSvE<4#ZGamz=)
zu>CL8u%8~tw?ZE&BaL=%Saej#qIns*U1i@|dC!{wvUut@?XCW-vtF+<ooFZ?{Mq7H
z-KTUfh@;xi01XbFi{WvqqR4_0=)G~pLLZO|Dyt8f2`6UHLL2p`duY&^dvfCbLA(g3
z8UCV35@oDin4;_}9gH-{dizs<3hL+fPX)TifC|-aLO8Tj<{voaELwQv$xJwyOjcv<
z#bJi6Pt7pQc=BcHeY)3}w^aG%VI_Vyg=YfFxHDC$-5fo^>lq)wL9{Y7TYya}*z~(q
zxHCo^@9cu0-f~8#tmUa_PyO=wn<K#5PNw0dq0}5`w%lZ)@BAZ>?zwUR0UfcXq#3IS
zP%$f8)lP^)Ta^g=epC5>x(^qQh#z@hXB1FYA}DJr)xo|ttJ9)KIMQm9Jh#lWGiUKq
z=Q<>z7d{WVDr(B+OKt#E6P%m&6-PEa#+76t;l%@OXrB9bRj6eB+?Ur<4Zr8bw*QML
zv362vo*-5Gj*(SB$os?g)^xk54C`4Lm2uJmEeloWxkEFEGhm>2=$$T$B+NYczX66R
zR%V!ge(tP>?VbV;?Y-ieU1Pv_c1&yX0lCS(H_`Qy6d0T(w6Bg+#W5c56{>~Hbs?!m
zT#tWi|8+}x<0nrEoNY_xhbcQbReS#Hy*>-J-^E$>S!Tl&n#sC0yWWQj`sp-Qs6kOB
zS%e^Qt+2$Mjtu|kjG6gw4(Ghw+%B?crEAtBm2%?2KGq9N_|<1K?hELdG~*|wUlEL_
zNaqX_s7-xYWhldLS3zv5vl|}Fge##SRXM`fh}!U|M}^pufBG}GxTM^>EmonShU;SI
z#(D+%w=K!y&B*o(1sT*(Vks94p3Sp6P2^aZoi5^l7THflyY`#jOqTdGio!iA$7UNJ
zAW$yhU_na?%^d3oQw~Svl=_HON*sy8_%7cvj#5jEwQS$WzuV!mi~T27zr?H=`Zel)
zGZf_CY461&$qiI+49_?BigB8)&Ayw}_+<R%SJ)se-W#wvv5RGX0o6r{&sBn5E&bL;
z_KNZb+2T}D_1;rA=+q6T%Dt4aFguGQ(h9<6|2&5I<IEeI{;~~&jg)fH8`QaBJ7l_u
zP$sFem$sfNoH%*q+D-@G%}2>1*JqbFpUlmX=^RajCfT>Nj!GrDnP$Hj0dq<-HTYFA
zi^WZ#zZY1!DEQ?UMepLmJ{L7$?8sn3Q#1t{dbZ`MVXfih4^(IIF$;}-EyvgZERg~_
zR~(@)_mpw3R0VSBmE4t#<ofWKD;$A;H1+;i9?an|(i(!eWhc38yCQ4WvFAxVl35Kz
zU|z?$f7%4=G|!(}>Yt~?ut)A;!n<;<!}ZM+Ki>3SMew7{w}oQ>2FCTMbP|mv6QyEz
z@mEy3eMO-aH4ZiUQ{@yH$eja=LS|b&pZbc^T=~##iPj%t8$PXUc$+FzHO$7fr5k=u
zTz%$)Q0rIe!XK*=T!M!=hTj6*2Q5OH3)IU$&Pkp(rsQ8`-<KTs2x^_sDI7NX;H;@B
zJ4@vqQI~g5k879MWzp&V1$iOefsQ6Y{XEs=uq6TeeuO6X+7Ie;NW@y4g1mi9(wGX>
zeW?a;0~kNj;?_gWiEUPHSHDF7F?LSzmiOnfXjL3&wsbS;B%)6hcUU7qS2=mN`*uT0
zMXf@iA8OZ)%#Lp+R)Hqlu!L2;bS3o}4{x@^zq7Lj3Y?X4e=MkBgWFp{Rua#*eDfO5
zyc#+o2VZjBX#&O!y&}uDoPUiXZl=u28Jph9&1heJ(I49fPO9}s+}t6ak3Zmcgt+(5
zmDpWO==5nTi;@V3csb)&x-j|%?`eV^x$E!(b?3$>3BfwI)3;y-<N-WwT`}o)bpwHX
zHU_6v@k{K2rwu;^3J^QJ%NIMG7>qDA>wejBYfW7H$b!%2pWe;b{w>st1+T<g*_m2b
zG?Y>L6vZoOvKvH^yZ>I1>!S|qAEE#C`9H%jl8Ef377rwX9F*+rH=$<U1ibJc_)@SS
zQ>(Bb7zKLut|zN8`x*gWv<@N^P_Nh>DzVzj{*%p}m%j)_s*>zov2V=W+Rh>f;)>6k
z2#^mSfHI;&OGTg+Vv#TTh_S2lR*+I#*@$o#sS3q+jP4L?3YOa*D{{z;79AIhf6C;-
zS^MV8Q8sEDQT_FR``&xPLo?7jIie*%wHiXtItU<Wzk9n*_w!XW28Iqm+<LjanfH?!
z<BtXsZG%*Fbh~18h-!iG&pJ%fCWmJcroT#7Ue-%1rq)e&0g_d%wY3X}%l=G%|B5it
zAvF;P{YK)9o6J0aXH3So2ykts?#upeuh3t`<`zCV-THB#`+Fehm-B;j2c`@p>FCB4
z<`&%;=BiqoGnW}Z(4uBaXGPc|c?|_c-jGJ~bn{@rip&4$W5L46J4>p#6gD5M1>p+q
zI(|}7kL!o8XD__h=ad2Q2tHITS7g`#5k!L05DlWe&=~_jx$$12)yKARD$kF;68Prn
zUvmHwyW7mp^>%%<I}j1xywH73jBhd7DShULTL)v_*`A846qX+)tjM^wYuY*@s>N$n
z!|+V_`DYwNk$pdjko#@uk?~DBZfv#s=lG{1y2!P>d&=rrlrqM#JI(NYWV*+pJZLYc
zI`gqI<0YKaY^`iP5zbQg2qe3~qFdo`3T#fiuFF~z7_DGBHJ0L$7p}mjo6`H+sP|hx
zkGZ6)gK8iX%cPpb{+i$fob(3~TDH^q-B<{88v6J#Sgo9)zWvXDKTGb@mVlr3I`Bdl
zXFPr*8QXAJ$#T`p@|cGB-DT%g$`LXV{0{ln-<0<|;yT@G0sZtQk%2-BbIgEGk_}9S
zoWu+w#T#pIOnIijx7m%AG!uztvhen7Y&F8#DA}+p2@9r@OcIrLa@MtrKkYUA8_Tei
zp~?>p()lUjMQP2lM^^wpiRmxSG1@_<9j@U&(RC_xxUhIdO;grfiU((7VKw9GT)h0d
z{G*2rlBru6lqyF==e%Ra-)J%c6J9sX*e<A_bxRcg9n>W0jIE^@3z=?pq$Pl%zf<vG
zo$Xd9tjmiXdz~8sm+iQlcIS=BzUqox+ffPxjFj#`{}t!V<FsBcFb0zP3XVbuDbjGE
zX7QgD3E9NKg*I-Ns}*Yg4VRx1Xl6Th?eP)rNq`6MQL%Ad)?^_uXCF05vc<5?KOuh{
zvRHu+UIBXQqAGU;Z}lXe`2rG@tEc(#a)_6AesJo~F!XEy{nB?nC@#)1S;iH=l7%)!
zu}u8RzDzY+!p-_e4<9!LFe#BCNm9&*iUHoqh)EO#p8+2fozYJSF#X%T@i7)40SsK<
z@_vL&yf@BJxRDIO$r#!M0*_4Z4mLFF=U4xdRsO;_MW#<WG9a4|%KdZqAO(LV`JG;I
zCKmtsucy`*P+=0bxy2|3o1G*76Hm`r7%>b}1j-xQKO88$m!uoEwlgG+tL|rcB}_)y
z`m(+JazZvIx@9*=;SeBz9mce>=kR+-@aj*Mwjo{Yj(YU5@}_R%gaWXfJz0MnhAmM^
z%`(>8b%(k^;!c5hYwX43K~jKDhyu#`&mv>S=5m(&`gjv2iC}N+uR4XSuvJPE&DPjO
z`-;s&%T?u7&(abq*a(`7at`J{{8(E(>0&0W7tb6KX0_W17dyn*AU4ly-#oO9Tukpq
zS^{9X3Tcy(Q=8pbIK}z4=Ae!n(V?C@bBp~)@Cr6>d>aNnbn<;DT?y$1-mT`s!-&OE
z`#8vl9M^nB+OhR`!3knoByue2F^q9_OL5w5^F#idpPR=ksNDXybu5^oURT-oWcz4f
zv&{F=y^8OThEFs$cm9uDYRus4bli8?S%$+Xv0kU^y4~}KV2SIkbX5q5{m?txQri%&
zsSQ~}@c490glwXzh<0jq(gFrYmpMvTCX0$Icjc3T+oD;W41Ws=L!+)avSl3k$Y$Z&
zi+-I^!dc+qglXp6yW41id@`X4@CWyVfR}1agEjn4N0Tr7UEIDqHZk2Z))XUJmiKvh
z2(gHjF?g`}q2B5uco?y7d2`I_=Rvmy`{zv|ef-A(F2k%>w_H^y9<dZ*`+p~bmvkMq
z6R@mndlhLRkX&VbW5_tX{_G%AQR{PQfctT7J&A!F+Q8;1aGQ#&e@1Fu7jEI|<{ehc
z^2DNMVJ>^UwHtIEj_pFa06yW$#!m=ieYP%f>GmvQ2GYK$OvnGCaUP=;jOm@Qt|RPv
z#R<XP-h<Kh^@^{l?SJsTVpM&4qpZL0dwPKby$zl2I_aZNSQVcKZ0q;m(?2@T2VkL4
z7!nVWdM#4HWG!GHe}h)NT;+95l79cQ>uKWJR1HTnu`wNY!yAop!C6xCH5GT4$TTdt
z+Cx0fsf%ey47$D8ZM92kjWgR2DvVew9e6>te~X>+<cujUBI@F9ITuLl%?DX(52J%5
z!_PN-!ucj=NFV9j*U+s6=5tX$h2?#;s`Q)dRwN*`IGY^GuV&OLuX&ic0T?j5Kg^`6
z#=98`c|UdiE(FA-5e%vf*~L3e&|@Y7i8$o+&oF35lX@lP>3DA<g&{8qMbC+d_MQ=K
zA&YzME)A6k{!#Dq`A)pA*C>ch*1-WZa*YRUx4q5kM+}9X!GiUh<HQx69;?IS^`=}0
zjy=MBJq5YFnHFS@Le`JPUqM(G%x&n#vG6^le^>Tw<CUTtCiM@cgupH<obyM;m640f
zradioQ1Sq@U_+amj#iNxcNjr6K3?P#sb^;bVyBVPmpfjRR{*q}S*}$O5Fg*bd?mrs
z7vN-H^Wpfafguo^j5)B^`oKM!E7&1{w>FCw8duDn6QoQ3^sRMJZbfubdJ?-z%=BKP
z7rW{-xlH^m=l{0-8{k9fTR_!U`$0;+E9PDSqdIE;=LrE$6_rk@&&w{bjPb02+L?yY
z*+?o>^%4k{=qfqy^YKcH$X$qgaQu!Vh03=|n?R|J^|c|=1yM<XJfgYyvafAw+U8n%
zn--HCI!^TlTE=$pdUIswQlYk#pEl^vrpwKzrN%p%mZIFmXfx)%US}n2?ICj$mwRGi
zDU}b4S~w_ygO?7R({jkSb{GKH+Z{|%1yFjI_@9j~Ap+Rb!gO0_yDlPX`q*rvnxK_W
zH_9y)D@ne%olN-A2<R5WR$cOKqJP~%OEV~fF2LGuhc|;6h@c74uieq9z00#>g_doW
zE4&g$`_2<sXcDW%XVW?2e)IRMx#vNo@o$N4X~OJ%`BjzPv^Fxal<zt}eM5M-6LKTg
zSzul(ag`M0r|#Us-{7Mnx3SPfpi7w8oA>I@9`{qt5j;FGV{nL9)j&&FG9?$y6pz_k
zW<;}4V9#^}umoXf@}ifamn8T)6B!e?v5m^!nGU;WWz7V^Pu#P6X0S_$RksSy#54QW
zZkK4WGel7e;HJD@UmSiXcueVVh7$>D1C{>WZhATzp|I<I`(!S8#PRd!m0>AXAS%R=
z0VPl4cX~^ZCE~KOR}t#Rsp*M|gKZSDz<h+$*QOzn+1Ix}EIZyPb<8lwd&RiCauFY6
zl!)k-I3>ln`#g=cc;W)FoW%9a?WxCs3QPHbKUcQDN8rsR3F_$C8q`f1A=y4psX|0=
zPE2Uc>K^-ym?I#wv=!&J&iW6Sn&H6S%u!FRp0Mx^c8E2lLu?{`i1dTh$+5YS;(uMm
zLvX10<wbFt!2Px3JA#XtR&S*WW^<_7wtS)aQ9k$G12tn+9MSYpZHE3${V3{=v|;c2
zb{Ry$io-~N>tD}mGa7fM7t^!Xrxdx_R65c_g+E{f7@{sEjmeI0WjJtQ>x?fzr>W-T
z?8ZqPqh_#$g8MJYmrQQsPcDd^#IZ$@0ypvXoV%gW4G|!ZgbqC81WsSxm&F*HqbF4Y
zK4J*V>DEa!vZ_#fB%n0w8^_jPA-OUA-<-p`m#B9{N?Ho#8mkAd+|Y&xqDv~iKM=Zg
zZ(mS!Do*T`igjX#C}$%2Bdnr0C0f-kUF$IntDnHNlLH5T(nPrGZWEeNh(eaWL;!d0
zWYBQ>%xz(plOyFLei$tSl$g1`_NSML+~aoLSFvKaCi6L2EcIOX?F|w#3}dGTrDpZ+
zm2X3I{4?796ZuF-u$p0{9F-`_>0R<qOB;F)J!f4xBKd4AlgCuqdAvImG4SOx_tbNk
zcgqq8mJtZl(srW`j-#(b*X=1`G;ZyCN>OkP?sZ1+`y*@}rkMAkw5OV0jafFj^s@~u
z0GYzMaud^q4#tAG;3pj%l+E<v?F&v+9g7O;v)c3UEGozX7nNZdju9szoAyhWv$mlK
zE^I1b9E@cmd5pqF*Y{(*-$s}6we?aSYI=tWSwCjM?c-=*#Ke7CnIr{0q-zS$6ZVo%
z+t8``bSq7Ri~Xo(R|I-N;xny!9;6vD?2o=(v8D4?Q)>SR324Saj3GWSd0f)jD`l&*
zAhbd@Iplb!5cIl|9C&2S%mqW3s*8CM{HcyB;{e)zoAxdh;t^0_`k^tlR@o9F)Gde}
z&^eo^U{>B9k2vet1eTMmNiz8zvwLeg;7LI{A1Wiz2l9yqZkfbW43|HozVR<8(YQ>V
z4`#U2@ZnvoCaOVo_*F%PCF{P2bWLLnCCQSC)dX?=Z1`4%XTTr=wO-FInZj_g|AKn^
zN=Aj~>ZO2+d|3No43b7hw3sgGMn)$zBq`NM*BES*Xv#lgqQINrwPHbL9hALdEsMfE
z&*6F&jOgF&f(##I6cLv$hj?qjN(bI~k~{Jwj~#&v6K9$2gAss%-k<n&%U5sLqCOfM
zo{0;ZuEPBBKqS=d+&499h669>0tuh1P!Z&YzM`67#?M$DdHu+?Z4d+=Sq$sXT!1FT
zUJ6BszIruTA2=gL07blVG<n7g=*YAt(c5SWe2sYfxvg9Tz~oAu@nLOkO5#uq+ee3w
zQiDywec(zY-m~z&g2s3zrLp5i1WNyKz`Vr%-sFuc!t#Nh4)bPhf^-wftnT4WH%Kd>
z8aJHXNCxwGX_#r$!~(`+iFK^VjSRf_BLVSog0WUH>bYs3T2U$cGjz#*sL>&iefOmE
zBS2Wj5i6)LqrQ0O!kH=K2$9Mj6e?*PY1v(@RI*g~1jhoN$u|rXhwOvkjwGUKwwy7>
zVGQW}JRA!A|3N_+6q|$$6-KQCNz&9Dx<C>vMvnWVYJQ92eaB%*+7B7}pnwKWUJ%ez
zHyzpzVU`i?WW8~CtJA^qji<TY$Nx+n-x@xL%K(h|b03*1V#&xz0~xx!)`8g?@gE}W
zK7Idu-22U6L(+gI@`GI?1Jq<W$?L@OvG6tT)p9;yKqntWpc-_9ifKs&%i>9{{d$ho
z$hF>B+A)8a>Q!@2NwO);&3l?&lY|HB;vIMyUv6ypZa-!i3)8N>-#w;_n0S1tdD?(q
z4N9MyeE)ztWY|ObL69O1v4j`vsawgp23=C6X-%FVP@IFdCE=WYI2!opvELdKBEvp9
zkOa)pj-q}x-yG|+xKJkh!R4ELbpxCtBf#K;2<6eZe|=|CK!kez97F_6=9$&u!#4)8
zL-f>8Vg1M!Qja+;{x)&Nh13u^|1-bj!Ph|pO%+*`ux1;~8c+@+u$>Xn$^fOG96)w)
zr-|mKaW^>rIqn{U+sM8ni&HhvyZJ0|J(tp%u?W0)QL`tzUUYFZr3Z*e=r$L!LKR#>
zTZz_=sgy3nUxy=#zbkBtM}T^aRk1&q{JXg->-2U;Y%7{qfM+Q?JjH>?@xH!%@-egT
z3oYrMC{qi%-BH(}9~{Cka%&>TSLQMOy1roNEhzG!a<0;mSU_zq^m#T5KU7a4vDGvF
zQ?Wdhz-8+>A-wV#(Rmq{)=Z51G{(g#f7wq;>Rekk($`n^?TVeVf3=f_QfBSH(*p5X
zp0-MrhqsZamhMuc@0u!|I^_VWdZ<uhx*mD4>9cm{<yTE6BlTI#=fjMyzJbFH6n=^d
znvGbs+qzbphm$a*=blaLPt$pfnE7{0RFgr4sw;!a$YjL4h)Cyl0>Qj|H?3xZ2rtCb
z&+})FtvM;IhAcrxm_hV3sNIp{;e8Qq$}=L|SxY%TA5mxq18~IU7-m28^*Np`AwvTl
z3xylJDV>KK>6PSUYv{@P$3{pQ*?hDo$4(p}-g+HoNx!U3@QJWrk?PK_IJ_I-oAkWL
z5>&|?95tO<xQKuK&k@C%`Zl&b)&tou?{GmRw%s2iSD=o)gG>$mKJ%o^(&4Ssg00E>
zH-e}g%6BmKGK@}Ds*wW$+VcT733M?6J)Po6i+bJKV=G;FJ81R8hfalpqE`ryL5Elo
z{dzGgsfD_3y+I^mtf5?maF>}&dz$u`4t0C}bLLQlXPutoS<AI*p&T}C5Ow20YenW#
zN+S5wW2nYkEAX%}e<Jr*U;N5e{j}r_wTPWd?x3geo<%s8X_bUYbzqaYSS<~$%uw^?
zf&U7XAB?9{e-svkC#FrY8F+*(@8pC_<%0UWz<$hpgB94i|M4&2^z$(SJHX^PKu}yR
z)WdwpF7yBzcPM3Lma-T?&zEv5WK?)1#250CW&3UXWmHk`c<lq`{DJIldV&`sRME67
z;#rK^tU3~{IT05`I$&0&kufq_0u`Oh`#1k{09^!6($ZOMjnYU&&E=*>+b;*jdqcij
zVmWj-pR8vxy8QLO>2yB*if$*;8y%G?Uvjpd5hLu+Bdcc0YMP0jdX8p1Wz+w393m|p
zWBMpI(>C$$b&>>pwd$|QM7{Bb+dD0C&Ah*7^UMbB^!_8s-Koy!{j{60|BI0%P>_VC
zN2z|ud*I#SSi+&CV06%fs{OHM$(w(x>=7#tnFD6mIV1@K4S^UF$_Ka2F>4EU=xc=6
z*|!#?@sI`WwE#v+aPK_zPhPPzvYbv0(I)?41>$s`Kr8m`SP!cz_X6enn}Y>1hW^O$
zM$^~xDb-_(p$}V%w_A4nL|MwY$M?kJag7m7WQlHW?6WwviRTWOJ&%jSM#qpR%1WDV
zS@JGt3(Z$pZCP9@Mh@r3<Ud8Y5pNA`Sav)4a17~MFU}SYY;?lkEMjoCKY23!N)F}-
zaVW)WCNYW%^8}lw(c<YCQcUPbh~EqO`!F;e=g4P!=Chi=?<)urRX!koe07r|nL2m)
z!kq@#N%o7FAsrHlpi1HsfY0Bh#?4<oZ3op>GgP=y4+D}8OmdB1RxU>vp3g{AKJaw?
z<k-Zrnt@GyPTvf1=ibiIHKM;ze-cM%#M@|5EKdFESDT(TZoRZRg%HBf#q*Ervml}m
zT~H}WAJZG6?RgIJG7b~RR%9U@!GF@(yqudFPU$`5M^nZZ`K<rL1t=Mybm)1*0kMDR
z$?``&_A($okp6tVjC?!O-us!j6lSvQ@`OQ#SRV{`U-$OF5jPvlSOa&PCbm6S$v>~2
z+ksYc|5t+1bT4|v9enSA+7FX{<LXcX>{a)KDn@9LzB4_vYE<i+XNAkPM~8H<BQa~j
zkkqo3Ter^gb^Icb%LY^+071Bz9PtaytWVG0^L)zN;znO+|53iC8nAEZ#XMz|f>72*
z9dq!_f4v{`a##&;AwFKla>6$&V`{t=&yDV5F0{7)x_G-Cu-hTSqq>TL3RA<gc*Rdo
z4CN{XXvF{-9Bpbfmx2JeZ@yu|B$_gNBU%{7P#`ac1yoJSR6~Y#7A_~&%kG{YAk=wX
zxdE(bT*DAj=Odu1_JX)6PZ)Ri?<&<l<lAXyoPj)7zPU<B=SBGGKM8Tvmn!c+4w&CS
z>Bk*2+<N{Q&v%<2t8ZTusYANzlR?W#t#@s5ED!b}1?`Eu21OT-+XOIalGLu9(0+>r
z>#ko%NE8p-((mJ|(Q6tXk%v-IJcmcY>*Dfj<i@T^Y@Od9Oquu3+N=~0vA)7yg78l`
zsUWKKzY_ku^CO^qTD!+-`NmCwTTD-)7#|XkQ-HuuO^{pu0S78XyN;8`hYY2^_qviu
zm&~w#-DUi1pUNxc*;HYd^KB3sywb|Af(6r-0AvB$11E)?P(^ctfwQlP&V4FexeQR`
z^oj#=VwmpZB@F>g7XfCvtAc{*%)J83r@%j6Au%J)!sukv2TuD=k&ZBn)LaZ&HLUcT
z7%Vr@<GikgVlM(!ZVs)n^KEUD^dit;Y{*cUH0H`-ei{1d1cmHPkYi=IQ5o;yOV_1y
z4h{M9=#hO5O3x5%&puN*7^(Obs4CoRR~gUuUZ~8Mzo7N#tCm66f>ChXn$IDk@gSwv
z*=~-wXxheL1j1=wTfMZrCeG<!94M19%AGNZR68L^vaUpjU4%kd`tOl3U}F2;uY$yB
zA?`roXrbri(?JRjwCtDdIsFP^v-`29VWT@VXlYf>pALbY<U37t=OC8Jw?3yJ3`199
zuU(T^S+dU+lqNbq^j9N_U{5klIcp;zuVs^*t)k6KuOsZ!%q(2SA7Z*(gQhzG?pI*0
zZ|_LT;3|YtpzysheEK`-%}y*4v~HKRhg+!7`+wE;ol#LO&7yl|$Wag^CjpThBu6DD
z;Rp;tvPhDg2T4OVkd!F#C`mwql7_qiB?<@#NFD(}Bn?5baJR>E&+&Wj$9?Or_2yqs
zS9gW3>gwLzU9}0F*i<PvV}MXLVDkl_ODJwb-|Uu$G~469hVn;d7_|n(lZ)it$rZ&v
zC@)~fb^#)$zETM4tgl1T_k(Uqw%<c7knI}6aIrgCSEejyr!&a#z8k>NavWfW_VGtO
z|KkTIlPpDil4`Uz4H_OUr9iXN@|(`yE&IcQaB@!><?$KH{xW5<Wp%a9wHRQ!L}Fe5
zE#Uyeac-MDgPmr=Jp(Zk;x}$<)OOUmRV+5BtnIMqd~n_8gj=v`;Hr(#j=Q5l?Yn*m
zQEnS4Ws3W*JVKwpL(3y29BrfyfTim28o30JbyBvjskuP0a~mdtQh=f`Kk>LZxltLP
zFJ>{@*Gt-j?*g5s<hq!o+e@N*x1wa0<@U{xcv;s__pZ_>hkC{<=p5J3zM84Y;G5p`
zqQ)Yqu$`aV#-#6`d>|Jgl8t}dH7(3Db+*5S0o?n|v!J9DrIl1b-dt}&NKiiWswon$
zWP(%Glp2Y)uc(;xi}qPE^sX4#lq=H~xpo(y0M}oD<uof9mX9$WYp|J>Gds(+Kgd4K
zB**3dlz~7UqMRI1VZDcIy;GAQs}I9aS;t$4FCKHUzA%&qp?%Wen+8v8JuElgE<ZV%
zj(I(Yw92O`qDS1H7mq1%$tphq`?LdxI2YMSTWmo^&8#UH0HfA}*W#$0v+K91JLyf2
zU2k*1GiIC8Vc-QT%8RqzhwO3Wlq{WA$YFuJ3(pB$b&tKQ-@W(ZVQSA&K-O6@*@zIi
zh39&b{=nDp@RyDI=06Nn&7Loug>}rIvB*7?Jrq93pE(K0#5?4}qxWf-T3e4A6tFgE
z&a0}G1^RE+3IF+VaQW~w+b6JjmMLnln}_EUU=s0lCcrhcU7@*UrI7>lcemKQ#lO#?
zT{SF#I3y<)$s(JGvMaV_A68iLaYUL$?lWUGUK<d3;=jKrtK(;u)34vXx^c`Tn*WKc
z=Eid=k|qH@P=3@@4nWZyF#6%8)4=<1l)Iv8XpO$|7y0-iKla#Swk`Uq{lAOG^*=Vq
zO*4QF(8>T58C#UZ4JGC0KRor#@aR>@>A}2x<oVl9BAI|p*QRbAG^IK^G5q{2HeX7P
zC58#BpM9+R=Ep|5EHJ8eu?_`unANw2;glLNcjosm){P&lTG`WL<2wZR0t4^6F1XEH
z>UKAq_d=K1pjz??y&@UXZ^J?VV=_9>ZlO{Pad;Eo#-dG(UJl9-Jp%+_f$FBO<d}5E
zx4PkxOVxZ}2Yj~xdSZ9lUu8>pZ1tu;qNTQ~dh-In*4wI!Qk;&^d9)jw$b#N_!i1b-
zIG0l{Z-^O||5DdZN#AwU_66IoOn|!!6A?hfe~A>4l(%rmLy|Ku^GY)889CUE2v<Oi
zl3|UjiH_)60kl3I8WvM7A~Hjg{N@@h+T~Lo9$G2j+cmRSFb>et5~(&HTh9fyye?dD
zpfDk8e_m@~2F2Ro!C%mH#J6G4qGi*#PXZ^i;vFkss{YW4QHs_P<cR3g%WIFla3`$x
z^<T=ppHzxicC=`_mk@IvG<~(}O;@l)D0dfwE-hfvQNTJ>s<?7awGuW&AjRa~5qC#_
zp~@}-)6KCiguV$JT75oRZ!C}wJ2ah9m&0laauSQQ42ev+YKHc~JDv7zcML}i9{{Zp
ze(Zo}qBJ_^kFtA7w6P1pR6Gf<zJzSRQPLbMbe5woq^(X|Far9z8$)&EUA}@T!IBai
z`f#w1dX7fUX4x``4v%9`VTuE`8EM5ugu`qYI`UO#-s>2j0trkv5ebT<O%6@$p1zIb
zp5&BZqAY<|i7_A|RBwcX%jb4Igf*+e=m+K<|NMMD_jt)I@x|=(YtYx5gL!)Q7$Wtf
zlNMl_qiWr_!S(|Dl!ij?3bi-)*%%Evu;?{wRjFiVx0^H^J^TUp5FyfJS)-gc-n0;=
zfvMevKK09%knWaVYM5e_4+iq^AJf|8FvTugeof3aNBZf~hY#T8f4iR?ac)_5_Ar)I
z_74E31>kInaI5IqE$JU`Ms1~m$j^IB#fpaO=x5Ss(QNl(yhBDj`T)CU*fXuM=#nBB
z*ZGYGfWRc7@#eJa!zP*mH?+$QMn$zmnSJpRV^31>R0`P)@HlahVe4tpUc43fnJk+v
z^c{iGdP_xbQ+K|L&rwC_?D-gN+XboRet3Y8r%QjGwRjlKWD4`AvL}3fW@7O@8Nd{l
zS$XQewH+*`0M`q4`!uL+8lT78a+yZg*G`JZO?xjAF$$rT?tYpfe<FZe>-M^)X>Kq$
zsaP5Vpswgg=$Ra3!kfru-SR}TPdL4E*kLgr1=M`G$Sfy5@AVU4(n7?5=8lpfBk?Zu
z++zEqgbUbNYOD(jTEPsz>@oXyJIz2Z)tHZ7=m6{6C)VQOgO`^yC_LRLLA#dxK$3@k
z=biPAfCq*97CiRxkC-_tX6rp;xd#gXJqsIo_b2IExwWk+zJ4T+m*I<+n?Pg0T)eZ2
z1u$);x~_fhb<Ke$h|!JAuz;y|{~N}dmq<B!!q5U4gw`;Ug`>BHh@Vr5gsnf|$+70h
z1V;5UnPS3xJzbM*eS5=Uq%?|X4NY{n&$XV)0Ab(3(q2Xq`9ULj5=&sd{i2cgBru_d
zUY%0-bUb)#XA%Q5maNO*yX18diVm%a(SXlT3XamN9~7MMj9sWRCbVgY-Hz50e#co*
zh>V!dV!rA?X0X_Tqyq9gRmkZ;38%}V=yw#CiUAOHG23(#E*a9YWw1FFcdx_q5%j*N
z5vdeAH-rp)iYK>E0JP%A-yXtN*_Cb(?tgxOzrg&!2i@#Oh9zu}!*D{9QB(>5a;=|{
zmG*2g504tgcqLS7jK9C@;qVd%lP9^v4?c8iCc5<p657z#Y?Q=JN5xm^V-VHbL^lp~
zF9P(`FWgtUCF!oyxvf-B&MPtiXYEHt&UdKjT27c9?_5w{{DT<hM*)Um_{~)^?lCHl
z%ckf8>CBMYm;%97gjnLaQg%>yd*)42@`{~20WeYU@GYS3VP|m2DP8Spl3I>+R&31i
zbR)=l-x`a%n|AfnwfHz?>-0$?94|M<Z`x&$u<tyO^a@vIbU?W?@hP}ZlLW^Vm1*?Q
zf0lM8C6Ksqf_A0=u>C|w-tRgrQF)qG%$M%JbC-;(Adb2IXXTkK{bQf+Y8-G3|JrqW
z@Bj{C%SB<mo@EjxiSF+_CKD320l5kaNl*iF8?i&7P$r!VHLX)eMRZZ+(l8J^Ph5SQ
z!ZS>{V$q6ZF8-wpiQ!G2K>jLN>w_y3LNMI(jHR(*uHpBzIg~(t_Xtc948X)(Dy@6u
z<UlXjwp{RzUbp8*(Z*C9aPR<eXjQX2(1jOTY~b5gXSZ_2XtjRZSiDE!v0+LWxA)Te
zd-4G?tl7EzOb)!_<)wg}uI%ZgcS~aKw_YiexVLVGd2y><7a3s5In417-MX8f1bZK2
zmCp@cd_q=V*K)?A)ReZ^=c5p}S)eZcT!jS04p*xPb|+|6y0O90shjQ-mL?BbJkmbz
zM?&G5z8cDa2v>BA@x#mxBg2~Ovfa?@vQ>a=cfXMp&GXoyf(WLbSQzPwo-cnf_smEX
zeMB;qOyp_kb(rc+o9=%hi!)N!+lh_VCN8Owl90Dr^nAjZ8C#$I^f7=<zZycZb$hf>
zs!Zq9!(?dxx$(#Jpp6gYshWj;G_aTwAEd#iaKHAUVMm3BsZy56=-5Wgu%MuIe+^`2
zY(N0aGy1f~yM4E{Soqt~v*2&=!^eoS!Fp>jcbf+RDot<`HsOv9HxJFb74ZNbi{{4&
z`qLo(Tg15Q$`vc_>T{IidxLvlhZx|&TzB6XizCH-)XHr|2^G@7Z`|^?BsVH&elk>F
z`^xXCUVyd9DZ|7mNwr>()^Va2zm55wpha`M!%*L(U{;tSODOYlUkcxpG<|xiLw?A&
zUM|G^@k5|BA!y@|<Mv3*H_Ek0nqb?@q$0YJ<~%Qtl{=+f{5D&;RRQZk*$d;!Y)^sF
z`7rIZ<cVEof90SQaehhA_uY@&nkIe+Rw%_ln-e?)%TgZ^#Lo3}37EeM6T*d7U80mT
zFcR?2w_aBm<Ak<FtZ26Mff*`z+>Q36NaR@`B89C8$|FWv?L73>`N93L)w@CJFD!s_
zroONXEL?Q9@U|ivDLKFFj|`wIcF@6p$gaeNLJn)*#%5x@`pqqVC+Y_Yx-5C;2|h|+
zsgs^2z&Avv^Qs%s^U@@keEe(DGezQ4T*%lwepP9*EBfMt6=p>+lzg^-T$Lv@gUE(R
zWeK2EMh9>aU}{%B%<J3Zd4_1JRerS&cD5M92Sgf~^%NU!Sj)VZ);zK;bw%en_`Fu9
zE#mj^t5e{-HPHn-y^uk>wxN#B^3o9FK2aU++KTg^9`*;6?x&x#zrJx9{g{=0qwz53
z@I`Pu5khxD)R^u*C53ogmiIG~tEvFq8!C2;C<1&&&w8?Zc#a20>u)14H{j@tndC}K
z4cyG(VZw)h_@F)11I`sBRE;WM{6H2{bL46kmmgGV03re*5_?eCh1NGk-{3u{a2bt{
zqUw%R&%V+U9x9Wq6$V-8Jh7b7vjwW$Qes0oCto^UmI`uHwI=UTIBfL^eUh;<q=n0$
zy=N24afPZW=Q^ELve5$}aL>L|dOrE-W?PJxm0p;B58&Wa+?(`}H=ORzN`eRCa@+Ty
zh}@Z~&nW;s`q}SK=xjVGqmYagI6?UbUu#_!1TLNMm<k`qBu%A?=!!k#0=vB?2+j51
zGA&O7;MZ%fFi(bOt-1G?s$sdb2jgEJIdZ-G0y)OS2rA}mbIBKvuP$9Qs=Js34BZS2
zlD)!$K1n9Q^}9TIL`9H6Arg0DLHI~WJPu!x$FVpG*1R2I%tQ~y{zy7T)SbfIv&gV9
z1q-B?C0OqXy#66La~u&AB?CXWF0$i|CcG%dzCAIaXH$L$VB7sRyLK_BLK8mIgTN(k
z#{@zb-GF7eRin4C)~i&YCGC*rBmXk!{FG@%njPaTMu_5P%$%N7_ejA*<vw_SU~(Jq
zV2(~LVR1vB2NVsvrrx)Su-zMQ6B@Vxn~4_w9NB+5@GRIHNIchP*Zl5X$_BP*t)<HY
zCR(O_xw*x`%d%B^WGynF%2OEg1Vae_Ox{ycC^JHG7i;ICE`!+mOLETut1h5>I=u1^
zdCnVbZJ&cB-jb#TP4DY51<NzJ>hrUAyZY#*18wwd+h1>~a7T%yeoNB6MUbH<NmlFS
z>SQZ^GIy}_ibfgDhJI_n@Rh(Cy3<aEeOmd5?%_<@ORvyDN-wyz4UmTjmQ#O!#en6>
ziq=?OYGIhBm=-z<1%%3;HBCjRaB_KF=^uRwEkm0bR<++<8{sGL9P$`$`zF4EJ<c;y
zNyS5j27@UYf|gyCOS0?>09~yp8TZ?b=ObHqn@A#<)$Ql(`8kXpNzQ*R`nrjspN)?j
zFT`~{#AbhtD5wDx2~kdB=tYfWwm7qNtWhEtSd5yv=13*>!RK2wQtvct%cnJ>=n~=K
zf@<Wa%}0&}Ci>e!OTOLTIrXTq=P0}Sg}wgS5{(ZAgirgP5YSWW&)seNgevk<XAJuh
zM{^l1=ffi_>C^LmIOtDZ1PhO8nY<t|=)0KCy}#=<7jhnq4<Oc<v7cK3whMXfd^aI4
zc{VvyE&hX<Ln?46+S>C5Gz64)g7|WuSP`ScWHA=-b3P$roX+-KZsCXa0__^uTxjd?
zN4?S=;ZB?x?!A>Qp0vslf^6DO@+Lu5saIcaZ7;rA;hF(4j%B;$#Ru3n!>f@L0JB{^
zHk+GV9r7NLF=J={B0zotsK1#)xWt$4SYUqb9lN7rN3?ee-l6@|k*?fuiifhG2x*$j
z3vM^^#cqpqHi)qXgQ<eba50p;o)ndVyLTgyc(8d1%l}E6*x;KNdP|qUlO!l14-d>k
zCAMMhSizg4#@zS2Ihn9BULKYkJ8`~i4D&j!I`C%pxrqWA*%YIAY0xE$fsSYeJRn=l
zT;g$j9s*#QLd+)wzw$j(1RaYMN%72iS#ENJ0T;dP7IrDP6gO#jxq-8(px}g@fGI)U
zv;#|s6X+O<>A=|ICnHK@700jVgOo`?FVVY`YgIi6GMDy+MvDILj3Q3+z$c!N-U}9+
zT5Q;A!#sRAnuC*$5Tz=HZsGkpLIVV+OiLgnhEl(e3_Er^;^DX-0%ygUKdx>1^d4Xe
zD8S0Yic<Uh#=C7$s<a*vO!`4Wb$qpVN{q@acVJGmMgQ}M6`khTc|P6wW#c<Hpyw>D
zr$td&57068M4tV7t)85pPp){OHMm6tyjW7a)!ymDv6$4^qtNNGL<5EN<Q@W))r|Q?
z&03=~?p`jK)z=`Ueszs?e6-I8RC3A7XfKM!udTN9^(zIc>E7@vzkKI?kpbJeQ&*LN
zQ+O~>qjnJ`?x<u4-mVt~3J@OV-UWgTKNRx9JZJhgOCKrf=^k7#k)^Vp1iA^|zMaz+
zKJ3?^#s+w2c79X!HvIssyx6`nW+NKo!1;B;$M8jIHXuL4d~bOEBx_Ukt=hHsEpXhQ
zQSFO=QauCYIST1oX<Y9T%tI%ULv+~4)HvKTVn7iJXx8P;M)%-E431%mnOiZY#wyhN
z4}3OQ8F7CcNgz!ZFI94B`(3`JVEi>7rHDadkbl74^W^sY>y`ySTvw8ig98xY`^8?*
z`hZ9GPm)GLg`yEcX=GB;hxyA>5eWIrq{El`AKSN&u0a|tos()QL<DF#Mng5_SZr;{
z(10#AHk;oD?K6T(zUSkUKftQ3WMC=lU2Fl!cP{Z;cIv+Ck!%S%*xWcE0o4xTXQ9>L
zFatchgVdTtQY_MvP#Qcqfd^~8cyjMHHBemrrXQboVw3^qgI+zJ3BbRFFU=D#)o{#6
zz@!u3>&D(@Bm}U!3$=GQ<k}K#<Whkw>tr1bBl#F&=VPFF)gN}kuAs#YQg%63U%Ued
z->g)v!~<7ft$W{SZ16puzyo1dOM#Vw?JE<>F`|(p{p6f0ovfKm=%8o8Ep`uA%?jR2
zfl87u1H+~Oechn3OF)bhD+O*LV&^e+Uck%uquH+k?EO_iGOv(g*r0)ob8~O)j7UW>
zQx>C{EIyZs5FjLeG)|nCk!xCyH}U7)ro=iJ$ad0V&PZ!!J|g2rhLCQ7OCh<I?DsM6
z-blWV4ft^{L~~rpsjSv5h}0m^ZHf)(JnR!6Jeht|=U`H3G<E7I^2L4L$*grNH8f~8
zj?qJ)$-1h%(96HqJVunSDe>fP`lUX6kmV-pZr#kD^e{JuKu*xQ){#^1JOD9)i1O9l
z{rIrF!aTjpnG8$$rgDfbKv4JmZqLqc;Ql-<^5~Od#62ipiE(wF?JH(S`XR@I)g?vo
z1^`nWa&gzx=LLc1%1j!Hi(`JNPSv`66xfRH$`S<a$nDN*W46`&^C(!cS-P8IHE;Lu
zbRgV8k3wo#fU}QVvm;KiwGug_T~yxo4){M0dwb{d47QHnlhh#Zx(%%(lG2W2#RsR~
z7Hyb~T`iP6XV7hU=SsA3Vny+0^zI1C&g?8oifr-uc$W;r+0DYIOoHAp98ofp`ASx@
zffd3!)zksLGH-G9Cg*s;VZkcz+j!F(L~*9i9-{oI_n%1K2~17y%2IA)RDW)}ryFwT
zQs02A$UJ>O)T(H=ItjLSZpM$7_6oW`Q1bHkCj_6$vc>0UrhH=BY2FK!GysQ8yL?qC
zZkrzivrgjQ1McZoR&{ZySHP__&g5(S(z<}j2f(oyYGd>b4^YajUNHLdmhOJlkteYr
ztHT}1m1t7e;u%etu96Oh7D*jVjMdJK*9<9uB^U}ZdhmTGMEkIjlJyaTE<qWVvA&Wg
zw%0WrM=6l~^jKeNeur1Sut8H>IOdHb9XFdZdNJj3SC%%ET1Oz~zQI+gZIFp)zwy-&
z*<7XsSZJ~gn2Z6AWy}yQnPk9<Jj*3MRR$TM0j|?4h*V;9S)l)|ohJm+!kOeivF^R6
zsU7%Lo8b*&gqR`%>9_aex2NR^L-UAfgwTGJ?^9*KTKT4{N1BLix{$Yko}Eg9YXNM~
z3Zg}{q)y>Gl(hHMi5nr^e;QK)xn1F2t_X#nNKEEF=NTCM#B!Vi%YL3^-3Z+bcDwe_
zX6U*-94lT38+<Hz4IQ#S&s#1IPU;|DK?qd2Z<2G`DsPSGAhKJ;!AIg?uWy=d99hxQ
z1z51x6`8wYNd%vuE|nTV5mw@&B~RF^AR8g_s{9Z*)*rgam+`a{f$O{1NSHPP$W8G8
zn|J6(BY9=y0Vq7MH+%eb#5#n8%}*+>_^*H4o_eXrL6rDI{3NnU4EA{S-7UcZr_>x1
z_y%u&VVc_M?L$`tp@3X{&GV1N_8jC@oaCONG&6pbl;AEAahI9euU=E}(JM>7j!tTf
zHX4vfUJ3xG5nDoGg(<)K;Ufix_GiCTra+X904o|$hI>b1hV0`Sfr4N7`{3Zv1!`R|
z=3jWOrTy#L<Nv@H6ER%U?ESZUE{*`a|0eTi_v~C58+qbjf;f0|%M3uta&W?V&Hf>d
zjfX4a73V*#<x`V_ey`@a=xxo;?Sf`;M(^go0dT6yo0J82e=%!x3CO+P6B(ZJ>KCc|
zENHu?5F_~n*!qD(NliZ=Nx{slT3W3{#?t^(xkz4AwfzQB(NZ~6ByYfhdjC#{a9z5&
z$GFl9@d<^amSE^A$>`1~y|VgIYe`}1^`-;&MZw<wMBUa(Pt<*EH&LMYbe569D(j~w
zo8dZzufpE8jv`J<3zCw&APQE%ggs_yrKx~ru$9kQuV>+R2{}Za(*7o)U|EhV#BU<d
z=f3OmJCD6lc#)C+h&xYJQ{+$Q{}zC!ow7f{Z_>@Xt|^ir@sId|nK{~}$DRIxR-yjG
z@*mM-V7zR){|=Nzq7hlqzoJxpWJK|M@u?B{Z^a6oebV2Gig8AizZauE{zq32*8c-h
z>__6?fGciNko;a$qxikn-AjAT=3niy=_k3uokn2#)4E20bmd>1Y9cI1eu)H9dfvY^
zRU9n*+YJ8)zzzyr!fFUftfKYP9aCy&G%eH>K7}U*qlb1hv(1|N=(=rYXI+7Gnl7Ww
zPqVp!$cAP$0}etIM&{#-wgai|qW`n*GXbh_lH47+CRcaPx%#?b1mK4Zii92u&ATZi
zQsA}s-8ip|2KL{pa^l6*0f3ZPOZA4)-(Z~s+=tR84BlL_X-2<r!w}h!Z!q{wfJr|h
zR}>N@<WV5toBW)Ax&L^2t7XKwsg^q<V$#~)M}*s>`cVWgD%swRv}kD{jKFX+AwKQP
z$y{#BG~}SpAB<QWECu;5!?DUf<cgtp_Vw?R)r^je^rd=N<=9_|hvI=L?Q9i%AA<bz
z0Qn(iW<7b2UK8;uRqhLb)n(jjCGpfhiIKRvD%wmi9a1y!=4>aZ(wFZ|9iAcXQvVc<
zZm)T{TRY8Zy}cXx-}wC{qxM6D301!8LMwqawp*ENR@pUYkjJ{RYaVQICqetZKuHai
z4&(r3NIpII=>rWuOWuhWNncnD_AwZ-(<+1+Lt{%L0zzvSURrmwW5z|CS5(8*f2A1A
zB25@CRoFp}lUhcK=*`cR?A8@@Kh^#@ZLeSbYo%w%Hv^2eFL!d+pfY4wtpqG^Dl@Oe
z3x0K7BkT)#IcV6iwF>VmRAjS$h@kSTWrR0P8UG$aMy$opA(VmkXm733zdvg?uf`Bm
z*f~R@`H2FD!b0vWg+vaNN&2a!pPmf%JTYR#uKz{5-j)A_dj3D_aUTG_URXefUKXTy
z5To#}J(ZcX2s``kos6bRvL<(LGLJ)L&bwZpK2_y*%YzTVtHQ$6266EAZ7Rr~boe_)
z02gO9Q$`!Xaw5ph71G)|qibs}3y@c(8Albu#z!7i{a?VgFjV)RVzf#g<yA>`p8I8<
zIjH5*vh&CCF5f?e`F{)AB00Ovh&*#$l88p|{OGk-&F>2|(FE$;#Rpi6WCR?8%$BTF
zZo;a{D$ay=UIoZyI^+#HoH%|~6E8h2aJCw+&!4pCZhoMix?7p_&-TJ@*OE$!c0hz1
ztY|!L>#3CVlVqFI)Hxe%{b3@08F^L_(k&G*>_+))LV1PpAz$qE&%jS(+${df80V2%
zFQsC9MgMMHy_{M9ncgQxMwSnmdsmT?Qm!%6<%PkQ3s;$eArZFDZrkMu`3bJqMFCrL
zEiJ$&q{%s5;5cHr><}}u&XUSE@38Rc+1$NWx>Ua6lA+C(id`j*U>eOy2q&fik8LFe
zW%1DxSpQ2XYjbnfi;IcN$Xg{l8akpkZmnzYPQNsgQMhUEv%l7qx!l1{LP-d=y!XI)
zJ(Q6ilGski`wJn*aP{&_CuCsv-Q5_|n$m3L`K%nJ{F)_NdeSdep8?E;A_d*Nj32ue
z7sIsBmVV#Es3<}vdrxJxk)c)A8oTbAcj$8Xia)wU8{%HaHYlJ+I$@p~J_to2sgtFK
zD`!uErf=WQFO%`*PQu7lhD&y!2hcv)ba8E@o=KRWSY(eaBLx)7FktLPV8z2WuD`@i
z6h61+c1b9PXShVHaN62+-b)Q4%;2@e)*#D7|4b~bm3aLU&9X%P&Obyw32+PjAkMas
zgET6AFEJ1-au2Cjksg#;G`Z1QaUue8zArqKp3hEM$vo77|0S_t7ZqV2P<vaR8acOf
zwsP7lw%hdpl2?n9f6mdEgZ9aKCp1Vr*hM0rTe&U^m7DvrRR%u?$avUq86|fsl&iG4
zU0`W)Jv}2py`XQkr2Z%xH4^|k>uM$$+46vv$mXqO3;e`?>tFg9%z>+vc}<A%xN|w~
zCn7(w`i{=VeTX>UzZb{}Ljh{1+%W%%v2(Rh3Gnn@;)Qt4)I7<Ps>zK($D3p)k5p4f
zM{X`ZUOyME6pcQTuP6C;sT?ic@s{A13L&)!4Zq#?c&B{ZzR~5Svwurb<RtzI=E+O%
z+s7ZoE1MScmRagEexxn6?j-@bQhMutW_b)SxBc{slR(oNG+KGL9UFlJ<<v<_$M*7B
z5|3OiNzS;Rxam48o9}&G3u<j`*lnJnW4#1fam^cOGWy?{_<zg?kz;vLr*iXtAEsgp
zVHD7I-CF5Xd>%PGMfa;H*BgFR@o2pDD!FIXqjKxSqeBVscDN>@IiLVGPNK?)ZOLDP
zvKto4O(SV^Zm*}Ym`!N9<RuYEo4R$%;Qa6;C=%s^)pNe_Sl2&lvkn(1+SMx&5j;yC
z>+L9?_{g5hc<v+7Y@z?4=j9$Xhs)uKb^TTLu`ETc)#c~iNQD}t>JC@Bpqqzg0)LmI
z{>2^D&Hf31uuK}=CLN(9jA5I11oZDzQo~O>N=`ra^*bq}HB45vD6K6Mm7ZbhLsN~9
z?-!$IV2R`5OAElGpH1CZ`S;jriC2#gNmcGq?EuXc*VR*$PS0dq7M09$#Myatk34S?
zDt{Y|0@8VAF9|i3UFMaJv)G*)VItLxFAw@6<!oPYR&d$1ZZ7iUu>^WV0&MRi#a%b}
z%#E@LkuM>kH)f~3?V)|BHQrL!@v%FH=!lmCTv9{@3a{Pws(x}770kS(KisEho^W~K
z;KKx*;kpYtGK~;BmfBN(8edP8@QN?Vd;ZE9*P3nmV)8ll&TVCMB7b-~Kd}t0prsD+
zvx}d$KJ-~Vo?I$<8On}uBKQ2T1IRzSAd?`QR`UYD$tdQ%5}G%=(X$W*9K1_6e}LgR
z)$Q56af(+yQc71cr;q`wA&_<o1XxwS^##acpO7rw%hNaJxkGr23<c(S+rm92>eUg<
z20#4hiJgFQ7{zT?^fZ;)m#aW<@eog$&!W97up93!0X`p-S01Ez2~kkB1I*~+XRH&T
zw|0*T@Qe`5A+u_l1M?BQw-L;*<N<5FLEpAKd{EyB%~>?WpX6N-B4Q7*ol1yp>fsoX
zdQ*Vszd3-1Z8-hD!H9KoUg=$WCAyy{+&uyi@>7DYhbPowouwRUPeuq)0<AM<)(B?7
zSuH+6VbLCrF#KB>NZn+BiI`Z^YX&QBs5hfr_(~&Wz?@+Yebe(O4pjs*Z-{YtoXvT9
zdKpPWDg>>C)_59mG+$BZYX&PpjeU%^WAai;*n<_NQ+p5gTMxEo+$<IcsYV8`6b$R_
zqewYC@c(H3s~ugBO^npqBZ-3W!_Z^ke2d+-Dby6f%sqR9B&}nyF26@~+Q<4A=<3&e
z{)rrX{j<h$tpUd318RztRfMk8+<k2c<$D9Ch6rYk*DznDvvpc!_x_M0My%%QgmA!@
zCX(CFpZ;##*gw9eYAB5KO4N)zC6K0Rj6l_dwYn70lK;={{6B2{fAanRpMlrTVTK%o
UgCWJPngI0EQqxndP_~KqFU_G6jQ{`u

literal 44698
zcmdS9WmuHm*EhUphEAm$krF|=Q|U&L?p7K^VCX?gq`N^xK&2a`M!HK%a!^7AaRh~-
z-|@Qs_w(HE@!TKZ<2jD^10Oigb*{bQw|{HxwI}wzt{N#30}%j#R6|`^9{?B<0O(x;
z81{+p<W)cR0l2Sipo0Au0)arGP#6q`gM)*Mi;IVchmVg>KtMo9NJvCPL`+OfLPA1H
zN=imXMovzC?b@~L*RNAhP*74*Qc+P+Q&ZE>(9qJ-($Ue;)6+9BFfcMQGBGhRGc&WW
zu&}bS-nemtjg9T*&717(>>L~%oSd9oTwL7T+_!Gs;^E=Befu^qFE1Y--<>;m`1$z-
z1Ox;H1%-r!goTAgL_|bIMa9I##KpxWBqSsyC8eaKq@|^0WMpJzW##1L<mKfR6cp~>
zy?gK8Jw-)DB_$<gWn~o=6;)MLH8nMLb#)C54NXl=EiEl=ZEYPL9bH{rJw3hq_wPS=
z@IYT*|KY=j1_lO(hK5E)M#jd*CMG7Prlw|QX6EMR78VwkmX=mlR@T<mHa0f4wzhV5
zcJ}u64h{~Ej*d=FPLCcvdi?mYv$L~{i;JtPtDBpfySuxGhli)9r<a$Px3~9`Cr^BQ
ze4aji>g((4=jZ3|?+=H=0|Ej90|SGCf`WsCLqbAALqo&D!Vm~VczF1;XU`%cA|fLr
zqoSgsqoZSDVxB*L9vd6`;>C-&xVZTE_=JRn#Kgp;q@?8J<dl?@)YR0pw6yf}^o)#*
z%*@QJtgP(p?3|pO+}zx}yuAGU{DOjl!otF$qN0~CUltb^zk2nmq@?8a>(`~FrDbJh
z<>lpX-n@DH_U*fO?<y)PDl03is;a81t7~d%YHMrj>gww2>l+#x8XFs%nwpxMn_F61
zT3cJ&+S=Z~f8XBT-qF#~+1c6E)z#hIjYJ|peE87Q)6?7A+t=6E-`_tlFfcecI5adg
zJUl!yGBP?kIyN>oK0ZD%F)=wgIW;vkJv}`$GxPD|$JyE0xw*Od`T2!~g~i3iPoF+5
zEiElCFR!eutgf!Et*x!EucJ_?jg5`ZpFe;3@?~>#b8BmBdwY9lXXoqJuiw6X`~Lm=
z?(Xj1-rkQNKlb<cfByW5Mx%fI`gL${aCmt5`}glZfBqaD9UUJZV=$PLlatfa)3dX)
z^YinIi;K(4%d4xa&$X5z*gZh#rEcm20CUXWKS<KnFGm1w01f4P2Jrm70>TxYe#XH;
z-E>Z8J;v!*qR%<ZZ+~0lbk$V3-CoaG?-X0l@v1V#lEdt=^=!AwZTF?=rjM@Qw$66~
zw#PpGlpB?639{Zk-sn8e7kM<%c06%BHYPGDt7O2#^OlJ>b)PXgHI+z!ur}R*EEC5>
z^ESQ~l#eqbf)kRNI`SbEXlZQOXdp^oKdI$F>g!9HP}pEU<NLoXo*@KSmm7BOx;&BZ
z+EI5Amal&_Sv<RW{<F~j?J%ZDF6dH6XCmgPjLX`0y-<#4<V3;Xd)IM1L{f%kIV(3m
zX;44%(}u}j(d=fhXl6lfJb9m;+QYO%g?9b4n)fm^S`Q|!iuI2>wGh2pP`6YZ$rKKx
zhC0+u9h%Rgfk;cnSxm-}luHHNY=pIJgkFIh5M4zI8zl<VZ>&Tl2Qr!i`7H%Ul28Ld
zs)6{(PUuCE3K&yCxf%tEg99Rt6;Jx#3LT}9F=hU4)BfS5Q+Mu*UTg>u6V_^A)$!m<
zKI_?iOM)*sZ}?#8wr2N5SqLQua)T2hp6-Guxz$lP;)OWl4YB`TWSRb}EF>i3CwXz=
z*{ITAlNay5gCxE}2i|UgTpBs!Ce{2%V7zqS+k1)AtNQQz7m&o)|GY0Oj_+Lv9nlO~
zPl^8V%<(2fTm$NMs~@Q8!{stQfn($OFS|D%ifl@h>eXs<Ai1z!mC~WqiE7G)8Bt^-
zT)~z#zvqubUXfUz-;cvrz1ozZ{Eq{8lBrk+#_(FB4v4!v4Bme|GrIj^W|LpVxuYld
zXlO$6MZeuIGR}(ht^D^v6XE{(iROtnv~`kug7`RDd$eNhsZ}6e_Kx^n9!ApN&&qhJ
z<j;-eFdZ{r@wJxNg?Qe+p@Jaq5qbKV$vb>bws`<nbi<zeE*xK+(|cm#A4Hqaj4&)(
zPrMy5o=0j>tx-1N?Uw~7)a+)kCE8o(u(%O&o*}u0<!g(nj>Fk`u;JX3M}HZS!Mr!f
z*_3Qna{lZoXz_o<N;aY<iMOJ^I?Sm0Giyl0%(yidg-seZ=lb$wO!qea7&yT?G@Qe-
zJNI0d_FgXfc2phkbmhF({*@Jrv&cb+T@1#g@sC*Lk&)`7ikLGVz21R$88+{OG@Q?(
z-m&B^*EUVIp8iMa4`OB&taLQ&+}OO-vB*i2JAZ0g+jQ*GY0@Blo|*A>mBOE#n13}J
zz?lA(t=!LDEY=vL?+`uwdDl>Ze)hde%}#JlHv7M+-8u|s|BIfd2zD%#u{N>Mr?40Q
zi!2RjzWV=!(*Mt4|NnP|U;7}K^$YJt0h(p%pC(V1XqVG+<BtdI+Oa&+wb(W5F!p#c
zFE{^qPg|$!%zwbnJE`=m5Ut*zV%T$suH(LpzA#esbf<S?420>YN$QIqx;kC$5B)}i
z?1J>uepQ>id$QIkB6hOx2nOYjtYJ*2N*srstlDtBdO~KadP4V9H%0E5*pHEky#MYs
z3RGk&h{^h!POqE)mt*^Q=deZ-(X06{K_--EG{5G0MCb+AK<HpOat3IwYv@c+pwOux
z4@-Sm1OJN-Nl^X|(;9FjJN{u6N-XyNL!Q%jHP^NN=WqTuIO&i-%m3RtSepJX!u=Ha
zKkzXV{{Qgschvv+@Ijd7x~|T|GzfkiI7kyhD@nXMUk1ByV*AhT0c!WL#JB+l-{00W
zSNYbm-h4Hr^a4jR{x1oTdQn9!JftV{3M=*0zG#ROUTY4^y$K=X#P0n6?9c9aJdAlM
zwQN_jga@X@@#+!jvL-_5|HZkJ_Xjpy6&y<gj=PS2SZ9Ekzl8oTVZVTMu+@+8y?~-F
zM*@Q_9o(ZyjK#YI{$1fe++wY{{(?^=?iFZ_s3){u;O*22vVK_hh63e<twWkZ$P_9o
zJ-O6r&~iWY!$N{4t^H2X9cfm-6SBnpSoR~CAMyTt8vTByT-JW0Jfi{rF2ZZnT9y-U
zhy5ofOwRM!;bXWw9e;sbm|jg3v>zUM%!-Rl$Y932H~seJ=ji??c-;2`W~NvT4VBWc
ztL2_N=2O-72k-gz+JY$7!l*#YM^?MlKOsKpC+~1fZa2LEs2fodvPerFONOJBS=frs
z2?N$>rM_xb#y=x|7i6C`kHU3QL$DQbANIE>{Shp!QO(Rzc>vB^dzhQDZm0EC`%@_?
zU`CYxJ6(Ujr)eF_#ohjj9zPV%n$_4S&XVz`T9a|^YS@)PmvG8oG>WGmc%ulgrA+$a
z`<uD>d)Tn*XrA@%Ia+^o!gp75pp?%4n^h7o4s2GB8|1wla!jdW1)h33K5Y(cEjkc&
zt2A>KmBDi7UtKQNUvK@@rGNV5uN)yVtjW;q$hquk4<Yk-djBbe&`Yv_4}`5ww40PO
zp#G7)f2#jic03qmNhu@xO)?_OiO}mn+7xtUY)&3%rX8FhuVd5IRPxjVE62h9Boe#H
zU*8ATYC}J*oOPc+XL7$)kdd3u`A=iOLxnUXI2qv`Cxm!uX+A3Zul+%U>*c7p6j&<7
z20V1X{hr<AgOFi+JdI*dtvuw5)kA>D)0Y+#llr&i(Z7YL;e1_hoJ$Essiw<gdF+3w
zU$~~&Ig@`4NW6ZU>i<sgU()$&kB+)ZjVLz`lJr(V_J5P7CTLsY9gl^gMcxjZLYb_v
zbxjcx%*kdP@Yk5Ggc-vB{v4Oh82<NpwccB~{in+RTf44$z(3ksutbCH$!_%mezhS-
ztl?CMBF+CW1*4gichK_&I%L*Aj&`a1vz*dfD8@%oHBz#QkBv5{GzD-QI6?APFOCFm
zPMTsh8G>%bAtg3lUZtU^fAkHOa_x*kF?SSml?xmGHd4S!V<>`6w>>?XxZ|(u>%pmD
zY~)_&kWVIdJ5eKfd*+a|lurtc83?@;9hs26|0<3GF>k{KZf?3^4JSg{KTPu(fLO>J
zU_%+LkBdI=0$E&}eE&El@j#kOOzII0a_^p%$bSjcdjada>OB7l`0|7ZV+ELyV*h2c
z04Y8?ATs^8rjH`9$eT%dJ1q4DRI>z3U>nD`P&gr*WyeFQr~GW3|5~j6t#)=Z-y0am
z$v(o?duwpT6wD{jf!w*KKl>Lf=s)RXSkBAAML%H`N`+!0JVcIEpyJSaw%wkcjIX#d
z@hLL@d&YJoXcaDn>Hl?QkDi^Uc7_1G!Be5Qm2ZT|x;`WExQh)UAD7$_o7F8R1mrb&
zeiY<Zijdh%(rJWes3O!-E!HE4rK&)~Hk$q3p(Pa=N<Z>i4X&&n4uV_|(O$;2I|ha4
zB>gMhe(fMIJO3l87i#JAmKP}c04~?q$76w4JR~^GrG7#ZQ4=e>BMM!KP4=~YqRWe{
zlv?3^+(q0A2|m@(hh0qok1Xu!#Hi9pgXez*g9X@JdpEJBn)<XpnhK_kgthkl6+{A$
zl)q-8c-beV62Q97kSkW71adWEdlc+5efE1%zyZSgfsuZi?_a++`LKR_1u4$|>&vB^
zO~k0y2H+n7YoAlLl>wX`hS{EVxUh!hhX^Jf4cfJnPjGM6!wD64@49(z2E01zw+o^u
zZ%s?)w0*M1DsydH)bmGd|4~qBj*fcy%abl7AoG?f7e;iW)f+@;;)3=T!_47HF_`CX
zB<PW1y@mO$rzbTK`Ffc|zkWLz|H+%1sc^y`-e%rY*?aI&!nyop+~WGgDCtr2c)Upg
zdoH$PoN<3?w&R%Af}12LJ=vK)yI<Y$R80>*e=~WW^gJ*MVsU40_>6Mxp_5o~7KY#n
zjfu0HVhe|ZBR2j^kz^l>3I<(%p!|JW0SqIPq|B{I8ORKi3?%0?_+Rl5tOz&8#$y%W
zvow(4qxIsg531B&Q9@8+BsPJvoTgZG#}T&OI>^6(p09=@=$9dv?|WU$UvNe8k!SPV
zt^(c4L!I{y3mBn+h8G(S5I8l+&n*)D0{EE&CgdS3+7oNceD*R(F^%iXNwd$B9X}jw
zYDysgdZv2|(T$Jj4A#NI$QX)L(@({wDL8af>I|n<n#~h`bdk)<S+Ty7$AjC!xmLDq
zh6o+3D2@Hl5Bs5vDX<~~thyTWSn%IeoJf#;R0NFK9h7S0e*jvp&6i>-Ar!$*EVr<z
zz1*|BclZYfL*7!bUdRUoAc%(|f3PM_WGuR{D9S^sb48JDgmu0po?H$O{w_r7Wz)TC
z8o84770$i@Utb5vJh^)L>mZaK=#KZW)8h*gHVo)0p5Vhn{w}w3v+sy~Re5OiuW!6a
zA3|qJ(REr%Sbjm=<`bcw(2#l{OAg2yPV-uXiBVaf$WI^5Vx7die+_Yx=w8TuCC)}1
zu{E6GfiTA%Ux1-X_A0zjl=A?p=ATsy3BgrHYNZG{YQ1FQ6F1T@6&EAf9+1a;W}!%f
zGn&R*1t!@xC$Ip#Mue)-r9l+oBEmGWfx$-$JlHw6hIV|+eMQG6fcvi~*Jk3&lL>8E
zzu?Fau`6l+FmJ4RWZ>xZ<wm+G5O7fVI{9E{j}kQ!*+I=`&xjmc9A6$Wd!V?DkNsW<
zsgQ6YtX+1a2uP6U*5D&!J{f;`vN{Pt>9Cj<5dZ}U2+RL#t4$37Y4{c;N^=;RKkgA4
zBZdoudthTr0kbHiB4=FrM2>7AaApu)r(Xl@7@=UldT>qvm};M%7;HoK#1uqpFL@DH
zzPVHMNS|le*bt!fT6WktDpfvmBNHOF&w}j`<ZZBX9&nzaHqQ-3k^^LM$7i_=tOsbJ
zT2>QzNZAA6)*xD_I<#jK@Pre<iFmg6T+fUY5%#?3FCFSb0j#)V*xkiU$uwctviJG<
z@8Mynmkf|JFduuEgI{Z$cDpth8hnS%6K`}85B*?=jt&Mls+Dc3{0v(mBMNZ=U8#&U
zRyHT=RD#y5p#7YHwTKT$;30~1{)XP`F-%AH!Wgi6Cx#0ltVuGtjiHF;n@HQyr#BHv
zl=>`I5Y)?Edu=Tew8B0JQw}~~!+PNhP_uk`*`}boFLdfkrJz9%!OKM~XyOvuR3@2`
zO#3$y042@?!_cv9p?tXryW|IaVj=SCOoy^}S@SsnYYh}AWTtjt!J_vMSQeBaweHvQ
zAhPTxg%;nS7-2XrBBDBJ17<^OeR;f50<U`Q2L<QkrD8m!Oy56dVF7xwho2DKy4MZG
zB>P~$JP-bxHn73~xs=hacCkY=5b_Y$h^V9g2$gZ+VGuINpM@O2v!s!_G@RS?On)14
zZu<|sT@=>{GV4ffYG)T$wq9Hlmcv>=2s|=05u?c6L%ypD*!G#59padBP+MzChMN$%
z{$QzG|HaLAEax09lAnhD)zYe1FZS2}#<E-5LyiAr5;KtwIj_M(Kcxr`iRX!htnHAY
zqtkFuNj%Kn>{VI+IPd@J?@lr1G>2V%v2a2_&Mp3Rf&#L3{by|q1ti#jn2-qITe8|e
zxrJEhRXbT>-=zeorgZt`=SCVDRJ&s^c*v|&ITY26^mBpH@rgCly(K}t?K%`4rbN}I
zmC1oF9INu{?y@`xF7CvVrYy&A6mK@}e72w6zyWaiQsRBx=}6QVOwvYX-drBiC5+U6
zxva~6$OKru-lqs5$}!^~D;q(ua@;p!U`->qu7TYMIIrPfI3O5j&Gq2OnW2KndsuR_
zplD^@t)c>BnrlkcAg@P@Nh%)b9fcHeaRLk(4%|H3!cY$(RQA(Hd6xwL8Vs|K6LZ`O
zJ3rNcdWC69B$SBSj46ku0%FVSz{cp{y(GYUxQl-}PtF+^R#K=q#Bn9zatngIqmPHS
zr9@`ojQQKyA^a)TYAL3qF}I2W973RAC0wj%=K+I@OKX-eIqJUF5_hT|LV(=sn{Vss
z^S>FUQ^KueI~4`8(^m-$nzTsTo?U~#>3P%vS^Khbv7`Y->->)X;`vZooH_uTU!PQb
z47nP*aTcr$Lj<4xcKK#LNg+%FLI*qxoj5_O6qHH4OaCT<{JQ?Z2cbdKgw!zP_fxE0
zZLp4L47y|V1hBqr%s8bi4X1VgQ$k6Ks-AdEArIl#8<OreHjQC>X#uL=dg`z_q?-}E
z+qk9)yc$rz|MtN%_%1{+{3ifiH`Fh0L-TLZo$G!kpv5x_c>+3eX4K}`Gv1mztiH|&
z0B2=GhO}wVjx4ciX;Cw--k0zHy1MelQ-{kz638z7YbuW&4ad`+VvE=B;6(%|zOfTd
z`0wOH=q7nCinMn0#>l%g7-}wE*Jv)f@yTr(<5)bD(np!*ukA>)ukwt+o^dhUCxxvq
zz+^M@{A~40#F~b4>OvQ<7+C)T;4RE3?%8Lt7mS~8>m$gsx4GajSyqTV_ub2vzhWVY
zSOO%n@G4Q8|7YfyXY%ltWz{?KhYSI&5@I(I_LkJ8Rj+=FB+Yfdyrr}6S}A&L7OXo&
zisC0*;L<D5d$R0*f`h5I0qZ{iophks?<$Q*Y$3q6aw13vD6Z!@yPBNdjqBXVE1QW!
z)U?tBQL@4i4AmX$O7e_42VhUk5_K&7&(rOBELk=Dj8w32`}go{_fDWi)a`uW!5>uG
z$b~%Y^?S(5z}ZgIuEl4JGB%i6LuS|<#f{>(f#8U)e!v32`j&)V1D0eYNstRKj$?uR
zbE(NF8SYYS7z1zr!pB8e3R;lSp7*iu{&O1FsWeKjFYmpnsf6o~zE1_hDjJ^+E*uSz
zlCqv@5UGkP%<BK5-=T*Ts_;=Y9bE?ekdC?5KI$wCj_-(#-TC1f+2{{Qm^;6p&MPaP
zzQYx?^a_A7DV$!UO#b*#euN8!Pe?F-T2H;VpXqFY1zjtwgAWF<b62j}U*9E@HYhb?
zl#}x;9{^b;`k0!F^o-LkHLA7qucaM30%^~I{3!js6>g-X1&~k8%m19230ax{3-aoP
z7CK}IKp}S{!?C_%V<}x0gj#x(yXK0ALFLU)Z^E(<lcDGq<qpQPk&6LYu%>IlegDod
zKw#&bgiQtd&llU+#YCQ!fY$DM7His^TsGG32*)$HYyYFWffCsEp=ix|guZ*08=3VU
zjJ=NvSefaH-VJL57CP&J(aD8xL2EHSvW?mLt}LP=je@3^0zZ3Ff}e)_*!;^=-@!oG
zo^Uu3=@YANuYaZ%T(g7e@{qRGyAjbuze{d`{g-I<)ZL7=X7(5^Uh=j|%^fZrjLK0&
zm(yOmh$QlnWJ&o6K1S@UE5aLCMeK%s0+-8B^vB4jhKKe#%oc^<ispFMB2mCbfNkOV
zu@mBu!Y-Djoz-VMjZ-&**|YhH$seKtdN!7{)|%RJYMZBh*iJEpYAw=CohO>hy&#oZ
z73+D_yy4=Tr*(U*b8qW}hmL-*t}?0x+0|JFeTcD%txJ+CUVok?pQw4Qe!NSdd3iF2
z)|7wz4=r2U>I^^E>rZQ)?Hie-7f6q;`Yy2-$!ABxh@WU5_x7W{qdu9ZyqTxt-g=Js
zE#}APszZjKFRHUsd{`9cb3R^#OZo8n+b((wwIK_Gtn_F8ywj^KQ(ZA%YCo3A+&WI|
zZ2F~d)s3Do;oGkxaMQj=^iwFw+;`!=JA;yf<~q4g*Ql~f8g{f(*HcuTN4@)U2PyvI
zp!N=`C&lDuT_StHr$x81+x?N>3T~?o2|)Cw4VIobS6dSdZ)N(iu_Z9c*tWU6^Cabd
ztrQPg;ageln;SRzu8<Uz&wM`jen(BQ+DmH~Sh=rV()(<S1+y(act+Jylom(j-TD0s
z2?|Golo;aaw^#K%i|CX%Z>!$M>^8Bd-<!OSEPWCq>4cd!+X??)_r!RJ5X;}lg@RC>
zm7JuE=H)|yu#TxkH|3#RL{alCHU-?P#Jl%^6_V(^JvRuh!tPxU%274-k6!+iaUuJc
z)AY!4E<m^RtC`zG7GdgnLVYNXZ}OoV*H+D2f6R|)cXZpQbMIvz@O*-csn-UQ>KoYe
z@OUEwS$v1?x95U8<t!)<GwZO<&y8t%hN`B`Nsk7vvj=&K<0xCW@44y@o&A~_oOX`v
zb8Ej$xcIW8Js%l5*G10o+QZpcNDzD3Xjca-E11HmdhMCO{@O(xhS616Xx{bO8?r7>
zjrEA}V2}JWANInZtMX_4;CokYSH-yFGyA&U&vG!#9iR7>Wn51oix1+dTDHh0%+yx+
zWu9TDyqVddxb1oNKXKtaHP+d_2A^5tUdA!JM!#ce`TkBW{a`CI?A7lLL3)nFK1y6<
z60dCHoW4rY`TMI7_rfFm7n4$SkNcjfK<4Sra`U(sPaj?{A5GF@pO4Da@(H&btV-^9
zL5%qnZxRMUp3cM3VI#D5Rc=$$CPzWud+$lFp<2JB)vDKx8`LjTl}xzxjkueWxYQ7$
z&TGFI=<fx&QW1j1CN1&Xb#2Ur?z%rGQ{it_2q$`%^{TG(ruO8iTi;(IUuWt}Tq_Lr
z^HUKa8`~L^tFEMJ_q%lm{=vG2<aZV!RQ4mpYl<!-ssNU#0o^+S77193z-3xC?3Ds8
zV*sJ^x1A+FA;fw@!<cwQ_$Woa;j;9~rQfa%KcBu^p7bQ>((I6c<&V<H+L9miF{?}!
zcL)xrNz`l+60z%4Qx>aY+eCk3gcf>a&ECmyWbp4bMr>oG{+-q<=Ej-u&BF#P?j(4A
zd^mfAo!^~)Jb^PtsWCEDVJvd*;$sZ$t+Az>_Q^op0ioA78z6B2;7k~WEz-J0x#y5`
z@!iDt%<uGu9PyC6gmv#<K#<M4ANg@GF-`YxxdZd)Siih@6G^xc!J6r|2<%IqA#DO0
zZ}HoOX*`Br?Vq9Qnli)N!I^Jr&YQb<Dqo0K&FKrOSoX_ND+`9awz2!4G}Y!?B!%^y
zH$>o(j7S|q!tX>6k)|pb91N5e!8R1Ve81o2^(dZTBc*SV3TV9vm~Q7wz~D;xoKR2B
z*ZFwp4_oHjlr!T`=s@#$$13gvJAW!Jl(z$3_99-F|6l9aaYE|(4s$-3&o(_2S#d<3
zO;O1a%zE~XA8(41)v9q^8JsDqJ%Cew<JWAmu+mmBC{PEN6;(6!dLpHix?dlc@i~Ph
z9T}!)9*gVeo{=8LgUMRxN*@WDxWpoJ@X4+2!v{ZEQN};cfQ^LK&TT~xZ1tOquj3dt
zSsb#*g)4<Ed>`e(LkBk4H))Wd4wfkRTT-7z{E*vn`<&ex1Hy(Vmq6%TX5u$LUFy9d
zVo$E@G=$z^yNAEn7%nQDqXt0}Z=$=NcJo}H#VPUM0{yaN?)US3)`!wJPDtEeILV`{
zW0`ZG=6}!<xgwk0%jh&~`YJShoK*wsSXvwep~vR#nhP=4UJaNB6u)+;S+ZHeo*fS@
z!I~=!K3?L3&VU5x!ED|*J<Z|<7zVINbz{?ptsC-@9_iU{dJ|2f$iD#O3$D!%*A4^#
zgu9Wfk-HUy7^29yfFqdQZj|%U8{zrI;9ihS3(ThWV;jx)Vt&vfVm7!BXt%Qt9y^o<
z%X{HONrZQa8i~o6<h(z6k*aTlIF{JoF1QuB0z9;bzRCt*C<CRm$<Rg~wNqp_9K!2H
zxsf~lkVVm;ofg^LjZBEeI<<|xPO(t!hZzYGiu3S>;7gAR6Ir^|z5@}|(SQ9G0ECK=
zAtgkZn#8h+?30>OwggGoKaDU$&*izt@X(s((q*Ul@7&zCZrTyPMVreaioEbmf~TyX
z+iw{f?%EH&ogv*qKW7qfNlbrf6zD1~jf)3QI6KdO9R(CTaa*sy-#(DG;bHT8G*>jZ
zNTYavTKG@|;ptL=h3kqW$ScEh;}EnqQbBln|AT?DYtVegn7l|6#<Qn}%*srMQA~EA
z+k+)dI2ICIis!{rDZpq(21AOA1jo;Ps=a38ItMUABgZdWI#%)%VPlqJrtS6+Z)iux
z53d9dS)51yCUf&&*LngbQfi>*L@YmMnaI3o^T#K}5@)C%?EEA~?UgfFy?q<(p)dPA
z2({n^$(|FqLqH$?leTw{rQ#hL_wD#Imo0qIUF2*HM;x1+#j+A3tl2i5@R9e&zkU(>
z{`jR-OKH>@BXnhTu!oZb<?9Wq2M@O5-Y`2ZRP5l@j*}#T?ud^9PZ8}D&|hwuOEW^=
z#yUrf-zV2)9d!>hWD%_Xkk*9hPG5XU*6tp>K}g&aab(Q9>7`T2fp$A}NyQTgCVRee
zW>)HZ(A{q!@6?e5RThb^YgLK-U0d)(oaYrwuHfLFGsF^k$7_u}_-vh=>;xCXG};hB
z48*68t}plLqIn<RY=(5qLD6Zth_DX<&8xwYI*&k20t~GqjT9U(T$W$NF<8!ZCO|I-
zE1Lm}tmG8FExU_(mb`#`m_^dnP|RE?KM+V9T)yFt=)Z9T?(E^4pB`gv+i4yhjkWiu
zqOxW&pUC+2VRB^UpFrCfyvUW0Cu^?iDJl=N>z)un3O#U5?hW3s`o+)jS|%uyPpBWF
z5Ye<9W3n86KeQW$2>SJdY*V!x6#ak-##iRMyM;c6wpQ~Zqf@?Xyp*Dq4$h`XM~+O2
zY-6XQVuvs5vBco|G9x^te3<kx*}@;q6cK{>k_)Hijtf%qV$y$nh)Ley+ZrviSh!4(
zT<v>1LIILQtY!I;No`<dsaEeonaQ=NA~$7hn~W>H$u#|UJ}FvVQztr_e>w^04AKhW
zX(z{XT&Umv8nOA{CuI)-xZHfoi)<;$*oFjOQ6$23Z=5uLkI_ng_FZn?YUw(T$+YXP
zFozxm2pc@giFg={6o1s4mr=RzgGKfj_38Yrmr_~%1AP>=`-8pmH+6*&3)#SG<|Ey*
zlkX!H9E@J3)Y}pLocO!J8?O~uIjFVEByiwzEirk3MOC}3Ym_j+8^6#Z>*1dyUS!;7
z!)zFwcKOwObjS#^O_j<{bLPY)u{!LBdm<hN9%A-M^xZK<Y4PVs?qpVkKrd=wGUeXh
z&*Hd4w-*?iR)A7N9S8R>`pcYBuf+V^`G6Df{GoHD6`Wz1{72*UGK(9MCoQ0MyU#3o
z^MDv_s*4yzUSzxZ{mcW^TYygU+tz6&wsq%(z%XozFGwWlhgbWs-5%&*PjdbQi(YF2
zXpgVSltf|iY8^E7^^HzjGrHWbnJPC4kGv?6N`!sMa|VgYw35i+RDP*9M_ZnJ_Wple
zy08Quj*I1?4m*G9)HJ2aAbT{!SN5-B@UhpZVOnpcB8OSi#S`+x^Ur}rHs3>A1IWce
z(LCaX8Sv6$M3PR8jnqiQB*z&3frucZJ|(R@OCHp2b6`WZeEA&yQj1V(pA}`;Y&oSP
z8nX$X#6c%igDi=nAlzA@GXC%6bnmwB@ioPE($XO-NDvZ~ozjZG>6+HV?gO2>#=S)w
zr}ysI%(qp(AD{Omg@mRZ58xt6K5{>IN`YApfVD*Kh0g{*$nm@sXc0d}Nvy8zBEo#Q
zE-J7t5dT*C5;Qpt^yZk0+syRei!JeDAgxr;a~XmAp1(NvZf22r!Fnmt;!K$xwCj_K
zw^#*RT8WZZ?mLW^XxS{D;DPNQ9zb%L8~I2Y!V-?VXZ>f~NXzTYvN<XukXI1eNQ$jQ
zBn=q#H;GjsED0zea`?630wgG7lJ2od!Zt&NzNv~nz%oI6m4SZrCiH$>3+&F-=%4Ff
znQK>mGImIKK;gAEE)V;i)JCrfr@?~RC*ef^9)Hqh#pONkxxY(R`&ED)d357KO%%Z#
z9&fN$`5dz*#ekA=n_ekjMCe+Dg0AbVy@N#CF{)Fv6HQt-uY(m>*cA7lkU!oA5o((7
zlw^E&u)~QXOU#@5<OB!eBzAled`6*|c#6}SPLC?8h-eCjbr{OUP$5mKP)maKGk=8q
z5yj~Y#P&&3K>Sm7^9^Ky&_^l4pHYyG8GxZ|Fc+<0<E5<MBDBH?#3&3X8|~wtxblSF
z&&z0J_Sp4Z%Wi?hhreiV=(-&=HNN762TwYEd=0_6QDQ*TBzii!uB1y7)EpQ*XU>2`
zh0AZxOwcTUJ>$khDm-xc6SWTA%~!dt0M_+Jo|Sx^GePd)T46?tNDyZS%#^y{3|ZC0
z;DQr5CKZ0-{y8Q~_P%@t26T5ez@ZBxkpE1H5*Y*JcNc7}#yg4mmD*MIsa{+Vk!*Ot
zFhRt~y_EyO+wP*FbDo4e@7@&%e$4sDsUCol6U}k91I9vtpFE6Zfm|F}(2b65?qQFE
zSkPqJM>Hcq_Li{v3hq-ael7<*d<<Ubz6YeXuKDA93@ty$-%NdhqM22Ff`~UAgknCp
zXyzI<jIgo(4%W6~mC8C2=Z4AV%<NPK<=#KFf1QPf2?hmb1>s418vnT8Vtwy>K+$zY
z;N^dQPb@6q;5DeQ+9f~~4MHA^AO*M2lx!7=a-M{jpE^AbbTwkdL$l$bf0gp@%O85w
z(N)nQ1AYUNPOn)<y0uA}S|l`=LZY^+f*^nlRPPgcnTB9pAOl_F>S!Klu!58HT%f>}
zNthuXTDZjdS7yL)X71y_OAi8pbGi(t<j`dDMb9ltcOkOP!mt}04yMlmVG;!;GO3Pg
zvN!1QoJHT-RTQ`3&d&Z*XfXG%j2H{Da}`3<BnEZ!b?mmrhf}j&e58>&>(q?;S=;am
zl7I2xh>6(CFT`wk`yTdeS-wIUlP;iFX(ajd#Fs_?E0Iwq`!SvpJ#alL*HYg%pSm?-
znjfuo;3o_T&Mu9*{bA*Gm?fyhm2wmtRUkp_XjBU@8}r%L(^tsKN3KQvh*C8tLocHa
zW1<LH$c>@FqpKiV-%NF#p$uYSz3zi{BUuP!b#k#3#}vg4i5`l>;BO&lv+<LIMRtD=
z`wMy;xS+tfBZ@rmJ=0h5&Fqnj8)5t-<Q@;_Oaop7caVCxl7^)dAU@gK+|UwYUK2rB
z3sk`@bY|C*&=RUKWi0-eUY3_lS;WXY*N#e({N&z*?%-hLnNb-XGL<JfNxfW)GSW!@
zR8=V&lth5Kf5YWFZP&9L!yWT3ATRyy_sQ+V_~hi#9|6Q*Kbm&xeGP4jg@k=p{PITm
z=Wty~!jN6x78X>T_xPhL;_N2@q>$>G_IZ)N_pE289e53EJ$T;tMuq1oZ4OpS?J8&)
z++#jpxFMuoe6>F3`jCS|fc`^<=oWbdfVJNkR8ZN^9<0aZm@TJSf8x3nL1fYWy*Bqn
zpYqp^bS3Ya1|521_j9<y#Rdvqi|4H{#HzQ2f7SQxMg%=FE52%y)RqfjihVEk($EJI
zaiS;t_O`;YCpS{!S1~sp`t;9{bxjPr-_+%oaiKeBOp}T`ALjGKAsx>Y;>606QY6>N
z3#hv+U!)J!VXtT3_V%qs=qCyF*&N8^KoTT98!JNApVXwYIf6>ke9Py|C<ePAqC{?Q
zgsGI-K<~vQr`b8zRN2n_htz3nC6(86_#~(c5>!_tpAf%ck_C~6Jg&O)$1QII)twsd
zKpO(Yq042z=QUAAIbvXp?Ey-u3OHKs6$QP)tF={dN>A;$xcc!tC$TXYBfyu&at+l#
zu+qn~^n9k`ezU6$uQ3P~I=@%);fB{JDl}Q)@!{`KHS6ZV%bB}vDouby$6@a!wt*-p
zhXph1byxzoNMHL-`quB#kGB`^n6CZ+<P(N%Y?nqNwl7&A&DH6_HDu@~^1(HvQVDnM
zxA2%zsjp3RaWU+gp8k;$nE08~2i27M4~T>P`&k>sT;1PEe(2(-nQup997MHOlX|0J
z-gJ29{I+*8gU!f-Oxk|1fqBITbzQ4gWN5oD`YHiYE|6>iy&4&wW92;Z7OE=za?9TW
z(RhIYie<=m7UGDo_ns(PKoumzzQAb;_xNPbAZ}r8=TWi$fp)hqe$u@GN{@^qLEyW1
zUPm!bP5l0G1B-)5g|Cbf1oPY59tP{QoA_ZW4UJ@ii~Kl92|F({W>nKo{Zcl*Uri6C
zC#*f48nx6hZ-pPHtF2%4-3h;yDfHo9OD?3>;+pq@K~@9_BJ04@I*2R&uK8dfB~r!b
z7IJX;!TZIwGb>F>@_RUGtO)LGZr>|0TaT5<;UUo4c4|`H?iK3czL)xIe^1%1!v}Fz
zLV1H%4X4M1D2w?yixn%ROQlF{60D7b&{>Lew5Z4WuxEXF&8F&fT)!Xrp70aIIusiR
zM+-H<!f^-C`i>gq6j|#ek(h$L6sIgS#~X8d^p&ls3%RnrnjpYM>r4%4#^G}S0-|8Q
z;>4r|55Ix2*6?85z5{LVV3f~&L5DG%`%qwha2R2J9Tgtt@)CJJozUhcaH`5}stl6<
z;lE-S%X;z<xQS9Xng!i?`jC4mw6hvBBrESzOJM2GBmUD`3Bh~b_388$@STNN^;~#1
z4r~(y-&cARmBd@1t2<r{YRU|}O^H$sKQvLs1d?$m0P{@jg~>9tLOKb8@vS3L(3d|D
z*OD8AGE|u|p}OubqweBpTT}4M0D{)fo=`-S#$6Ir0U1Ki7n~OwWsZUM7Y!D5E<>gL
z-%Pcp_GJh`*94JZ`&_#y!gT#7v1|*;#LsDb=v7SH_>Y$%>J68V_IWD^&`=m^EjMQ2
zwwG!}qF=Zt-T|XL`6a546c4>(y7-D~eQOIptjC*^872J=X44pjrA3*z4`YzB^C4DU
zge&^dtXVdXbJK_7u*X_^mBiiDssc$A1VRjW?#<gc!oX^5oPH@8>9r@)X$x&tpTNNs
znOg=TUlyj5`7g8ThC{6Egcww?XM5Z8dyMsku2%UL_elz9;H^qsc;_J)gD@(HQ_zs8
zT|B1u>6sY;dR+PjEd_-=P_WB|N7Q8pUcf#+@I~ynIBNI{?<P`)N+CUw2oY6(D5a_C
zo6yYqvNCELBe^DfwGVvr+8~(6X!kwu{G;sU{Psq2=lMr)DJ2qARy>+zRc!Z&7oh({
z*1EdS19@6(1QNmnC3z3r`7cjlP}#eH#C{lxV&Fo2Qt}?p*}L#;jjQoj42>AXdsD_j
zn?848^BPXNr_GghUgNrcgpY}7#_<mb**mrCXfgK2B>lczB1FqcBLxmiW1BuCY+xd~
zJ)oHMd_uHjIbrVZu}kBDaf;;t45g<fd|hAP_?HFEvinnEU}VyngKx<g5QQWi3C1-e
z^J0D4i;u9i(GSMS-ZRElOWR(s_5mV(!mlXU^b4R*>?Yl45xahphqqTwPQ72LJT+7D
zLT%I2B1|*pqnYni+3}4}8u|*lD!t;)S`)vd^2u8*G^wSD`JUTCk31wrnEo-M&^*+n
zvukH5($B+3GvbwNuw*^CF+r&s@}|?{-lJJa*yB7gLww|9PY}S6Fr!>CDb~2i*V(nb
zwY>iK5HV&eH)Rmq8nig@chfF5*0M#+P;iU`>u{WjrAI}{t~EM2GDYL5f2~BHL~^PZ
zGZCUCGMqDhu9C}XeJi-mN7TCoL96w>y&kJa$OABJRc={$h&BZq6`|>;B)ex9Fi`UW
zZ$Fr%K$na1;z2^;h>g+(QG8u4jEYxnSrW;$EJx>zZw5jioIO5mh&2?Wjh#G`UVJ53
z*jIs*r2=OF)o!@i*S)!Ah<pPn6)y)Ns(SpZ)UNE?;uXwwHVLOxk0j{rz@!+!+tD&{
zx5Wh(qy^?(^Pe4256K=`k<K!e<TS<}pq34ak}shsMnf2iu@ZETu*^(ezG$!BRtuem
z%xBrz7~=`li=n5rCkXuSa-Tz(@bg^ysO*01q<$@dlnt;D)HbccTdWGLMyidUf7M&G
z&HiXciNt+!6Zxs?%OsQRO7;xxtM|$H$Zm26yZ+TFR7UJ10WS=-JV*S})IvX<n@n&Q
zo^sEo-J&G)L2`0{2{)w)sZOL7{^(W{)TT;ncx1YW1SP+?O(Pv;GgjGwhZL~{^5#^v
zzJl&fynBuX_XWqb5g*1i$q=@}{F?;Dn9HaeBDCztz?~z?u^;`-{J?WIq{=AY?}rr*
zJP`tWO8aSI(SgqP?FeMvX}9M*M4vRe3XcvDyKjT5YWKC5H<8_lxJG)Lqx<;<MV(=+
z{=bayT?c;G!^T(=k8dDhO}2a8h0!NncK}NF`D%lzas+WL>^Bj9D`Wi2PdBF+!K9kP
zomVGJWWCp$7SN9_s-H#`^lKEpDWAHe%~g6!u@G@=uUV7StsEqXyALbZ1cF7*3wPy;
zHkUr#s=1{XoJ$N(S?Ps1S>AyN9tk&zvxv6ezJZkZRw6-yQjkO*CTLMpfZgrZou+yN
zvppM-MS<&fNAh&z+>C(TSE!M-0IVa}tc29EKT1GB7lB;c?{HL-6w>`KPjCmmY{&bW
z(+3LOK`<@!9CyW!7G#HX=rAD9dGIgJAYbk6rn%y8R>4SY%gFl*$9=!kFr#>qFG6Ic
z2S3UR{NS1T1Wfqw^>eO0zU(IQ!Jn@9E}eR2j?*eJ`<cmRW5rKC@nO{_@bAAaYCl8u
zCL%P?zBY!>E|E;-49ReA`pNwcAB(VfXyBh6(sV8p1BYv+<jPMu72N!Eb*T6%#z2wU
z6S=C3_r?T8BwzLX5F=j=<2kX}6gd30V-^NwXuUS`MT?GX&P|a7wN8iZyaRd#ox=!*
zU&GGx{lr$^&1F1_3M6H^m@>}sRV;SApaR-ez<6i$SwY2tRLC*1l@H&9m0)`{h1To~
zF-DBVx=wXfp2iL*r1He<Epmqf2c8KLywR=JKcGtc2HzBBpC<k(DM+sS<>x8007j&>
z9o4r`D9YtF2*&5C!SbFJOF>EDWmH7&TR9c+v!>-YeeEYT&j_KN?<Vf@8u|c?JP!QD
zMtmSYThx>wuG2tFK1k}EYMq?M80S$-Q?Yu~+(brv`f0(p`Caszp@ilcw`4~-e`%uQ
z-5wtGEe6xUWgXx}Tjf?Xx{GsP3L=<ZcZ7%ZX@t|?aIbYQwer~ge)!WwYoC@3-O<ad
zY{39pSpiAw1w`)aBTK-~Rc{ST480afBwdV>79z7#vUJ781a`f<wn;F6YKNeRr_;0~
z1bmif#*8pfQdc9XJI>lK9<d|MaN3_hQM3@$y7349H44ap+7ir#I^V8MBhLltNwLk*
z3eKhQF%s?#?}&iJ;HEethJo91!aA1$#PyTa^xx=q5M#Pm^DgyqsU0bK;8q(HRsXX}
zDXmW`a#M+$7hvu&LBo1}yN0=TO*T6rijZqK@L5|42i-*M_^PzYpHmY}R4Sp?FqNZ+
z2&2DW;o$L#h4I=Ci(6?Vh~3^A1(P#x0ZM-P`vXf*WJPg4K$fULrer^6y5w0S6qJ~|
zk|5_Iuxv5>oi)9Qg{7ryNv#5X1~RnIXkp*enjZeq{nfkjFcwA<s@OWV{p49mp1mcg
z9#He1cyN;6MJC8U3Yq8j3_-Mww^CZav>o4fk?e08GTE~Dm3Q34XcZ3{JWdQ_H1E~P
zlWjMyK7RV>jyaxS$pP>9bc%og4*&Q?<68}`JG<gJb!*1JLIoE=@A>gcqLE}>?At~6
zH57Mz+Q_dgE<B`NUGIyIyl#e5+OLMb(q?aSorm|4H9|>Fo3eSyD__w>33>SPF*s#^
z{MV;hZs6c|!1I)em4-aF_&nS(v|YWQRFwG9*}BZ{3uZV8A2M|(=_?Q?z(-k;?C@(L
zVtJb{LVaoZ0R}>fYB;f=pDO))93ec<fX8aY?$MwFShuD4L}8WmzVze@PY^p6Nj}ix
zW<0f0H8BLSEv5G>Vq}!cQO&cWYthNps#C|#B*0>t$jh+n*6^pwGgx$E`%83&D*x{`
z%!>N&V6EWrXfY)RxJv%lzS}#WPwIGs<lpBDA|451N5`j(JKy$IOJrcJ!UI-Unm!FE
zJ$kx{A2CX^G?wvVPuH!RbWFZNAcd6@{1Y51Exk(Kwcaz1Ocetz<aKJTq?QQ}-qSab
zyV$?yoYI_|g$hpJ)|Mw|O^2-zLRj3+7njnx^=ZR?8Hppbyj^>{ZSLe{P(lhdon?~H
zbqbbh70UYbr%<}321f=cvi$qT=U15!bk(&&&DpT5=j-Gniv>(54JH(~3gO4fkXKZp
z-jxijA`enxSS67S6(2YjtlDyaAwDPd+$7tvdHhjsg_rWR{|nZeuU`&@inY=Q!-9$v
z6<C*&g3H3`QM<Kv4jF^fKB)Yo`xySah(8dk{e7_lwZzD1F&Nwg_xH?Zh6v7uJh_fO
zb}B%RG_Si<#ZR_;+`wM&lH07vmr@jQ@(|CgC!_!z_V=<-gWfl@;U*8|dG0TmizA9X
zCWsRSrI9!LrVS%&k0fBN4Mw9hYP^P&cj&+PQFPtQqm+Q~Z-DhA-N#9c3e8U6xmrFC
zDTKXt2IOp;Bx}slMA2=uJ~xnImU_HlKQbO$DEsj#fUvwciz6Bd)j_J%<sG6R2rf8$
zEYk8g$#YiaJofUUqtHZkKOBm)<&c@Zs(vF{=yAg0In%XiRYy7xb2{YS<|o7AYF@k!
zC^E?lk#qo+wvzc{;VnUKOPlCQ-Aa#&u~M80_@zOx@7ILAxPlf8?`qAE5v}8X9X3?s
z2%t^Ub`2Mps&#r!(hl8W@<P}erC=L^8`dWxpOZp%EIUS%K*PIBE%)@ymUtVNd)97M
zCVg4_(~xLWa{{!N&npQC>fV4j$Y<sIAdP$_;?F73*!|4Ob@2g7l>rn*Lxb=RCJuz5
zTzmu;Uyp$}NX_Y2*-=(!Ia8&YZB@|9(IU=-tJOzbF2!}*spIs4^rFagc;M;#4;sVP
z>IJXN4c$e*sY~5aRGk$DU8ZwtUbUmcM15-?U#h9(_I};C?S{^U$iHF96XVqSZ9lwI
z>2yuYM;M{71t#sl)rIlBmvne<PQ4JtPgcj(Uf*>>_{5!Cg{uc(l%ra=cb*x`!&=A0
ziIIkI<t_c&MQ0yrfIKT?zGMU4-%}<wNc-5tpWESa08{bjDpwPo&lbNQ+W<X#a21JR
zUzG_++6c|Cdvq3O<)r#S<m7?-lUj|H;x8%IeMi?pYi#KtapJ3E6A-OU^h`p>zfO(M
z4=#lb4z%w}wuexozSw+f(>1y4*=qH1>x(zZyas7)dkL6cy@1p#U$ePX-z9`pGD48$
zKCZp&Nex9dP7^RpoMvM*`FdJuy0~}F?llAK2sc&0peH7x=!x|&c4r7u)?#*i=?kN;
zJk9g*;1V8u4E?i>#ow7I$m8-qR2L=(QaG>QG#A^o@t71Bu*5OjH4VpACMqA7((!+r
zW|W`&)cU)JzeFY^u{950N9CN_gp;nonC;znsVPXLM0v0)UaRnNue{P$;h;P6b}$Qc
zEa=Nm^7S@IlI-AF{8a7!0N{^XbzOVk6~JxgNqYnNPzu>3`riM5r8mTe7!M&vNm=BN
zlbDd?k`}=%4Av?j=dCB`U4Xu8&nPT~%cS!Zp<?SW$$;?XJSULkFrcb3NN9ARL9~4&
z66Hc7!z8pBL}IR}fy7-z8+|1{pWH)oX~pstlal&YO#DM12JCR)EKW9{lDhP|_TC{a
zEyL=HB(mbhJ7u`zqyFg+CT<c%#rhIjtj|=eavdCh1pwt?m&sE@Q$cWUVKjd&#Yv53
z_qsSm)2)+h+6@_=b|J`0KCh!N=7P$iib~Ry^GsY{*Op4YdHeBaJr`<5nRKdB<|{hf
z53gI3q`>6EJADWk;Dtjy!{iT-2knCT2^S8$yf_;|dwk5C<a%zqJ~gR^#MqZ7&)00E
z#LE75z0=%4%sX4zb0*5WOB!10WNP}Tp;^&rK78?>pVn6QK!dFd^NAes?s$7cwQYZ4
zauTNRPEV($YTa1i-vE9uCjaC%KC(n<PjV!}8j9QXv1>rw_2BN#$$Vu@U*S60IfAv0
zPohy^!ew-y(XaMrz&)ePY{h9=-nc8AgipmSCtXsersziZIR4#`1Xgnr_}m~qXBdub
z{vT3TF9*`vHEi!DX_YhlBCwc^oH;3oo$F4C!7nlUD@sjQ_rsKpjmS3am53yHXAh(h
z8vdzgSDi1swJ2=dgzyW#!|KMez0lezN6WxIhl3SKur$|z-Kn0XpK!&f2P2#KI`oT9
zze^<{eO$PTvm@UCJb2s8N^-bGh_3xBg*|Drz7XBZ6*~uCC%z;98fjuXq29T|3`GJ=
z2V~z<dR^!Sp3mmct-~to@$w$pq~e7e&NTMai=QOtACi(a?9Q4*s|}#38$E4gemX{-
zLKWu+?JD2{8TLX)0p<On%QSIKT(6lb2H&~hx1fXE7~6>wOm+i2m0UT=opR*jtEg$^
zek=UihqY3wYWW{-WZ75o1x@T&pXvz7eP8i6)bNx%wCg(h%7n%Z5Ivetr8_IMe%{~s
zNBxsp;xF;*XPtKEwkl<Vc2lj+v8p{R11+3m?h;>AFcLF7%_+%@_92!2R|d9<X8m^r
zdH!g7-IjT<G57oIj<?fw>AAE&86lh5@Pk@bL<L%v1i!V@`TFiHT%E^OP9FL?peU-k
zE}v+=I_=kC=YRbcU>E2}qKoo*o^?H&u>CQp+H+0i5+8hIkbFSrg8EaZREd~)Q`sir
z8S|~Pa?qE_Z2;+{tz6T4eJE+Fpd{`s$(@+eU2mOtS(wBPI@X+cA&KIext<R{P@#HN
zHYPhvZWYgnh56zP7+v*+qZTBvg<_1f4V1JwlfQY_D&KcLcfx<llh@Q(gJGE*hzL7O
z3aSYt6?af*xl2$|AJVv!rqtQ;!awdM;W)qbY%oUb@%6m#4t`2mlhJh@9(I#SOZs23
z4jAB4@MzLYF7g~hobj<2nU!sV+e%(H+$qMizOLL`dR6hx8R#>*aS?@*TqZn?sV($1
z)H*8^%YV@tIXdcqJ?zWLXI-5&x!@>igo4miy?JGX#3rS89y<*fgB8p%r!Qk5lcKO-
z##Y=K<-A{GKRkDJ9_uF?`i?xb+ZCZRZ#%-vzeNJNsbyvpVDVpy9KAbQ@%hGmXjD_r
z%_G<DA{K1-7VxzlAJbO`C&b;H*pZ3;#kRxdgy41g@HljfYfk3F?J<a|&k{L_&6;lE
zSl*PFVPiU}6nTP-S=^yC4xA+@kZ-&iJ70OmKf{1r9Q#QCdY^JL3tWpIIhyYAk(@ny
z^vgpYtW5dANoA>YtY$0mSn8+;4u10nmqabJ(=q-ev6+{o>&vAKAu7rFM2#Z<!wW^a
z*M74*xI@LyO?us?P|5!}k;xz#dbUT380Eh3ailB8c|c7teG?xQv6Ir#c?A$j0s{9Y
zmj557zJe<bCfIUtcL?qT*We6+1Pux9&fxCugFA%a9zt+;hrxrp+W^6Ba3{-mcHcYu
z3%crdSDmWvTenaBWl%{0i#*FUps$|^@D!s{SP@B-b9}1y!a|RFSW!B0o=L|9+KNPu
ztEisF0rwAIdoA>GiPKEE4-j7ZC<k#bGlUXP`N$!-+NShtJEC2t%qT7Ad#MEO;Ogn%
zX{*lVRFJv|ImXV7tRv7>pk1+Y@#2q}p}ui;raOR3oJjJMJ=D@50vp#=CnV5Ap#9%y
z^G^icna&V#voko)=pqlWv2w9z2V(;qOjHnT#IL{xsCBga%f+3#sB!Y3nziIb>v+N$
z!mmc9*W1<Gt6h5R55lXU49@o}F~UPe(G3zW^d5Jg4s{3xTv4?3B&e)n!|KBf-q{a>
z9>^r{qfMBH)98o(fk5Fz2{sj8yDnHKIUVu<V%`_9CFq`+N-=s36HxsLjG`A{v5?iF
z;2HOdi7d;euOzj9q=4w<^WOPc!cf01NQ>LiaVXNMgx8r)504G<&3J1~2};!np;U4L
z`|>k+zv4Cr^5>0w(Y#{tq=BT{H%@Brc0|p$9>e}H{D+3FZSiI?f<6)@4FD=<tE1uf
z;Q5D-YOSX)ga$IcWI1=SEg1UNjjzoeGhH}f11grVC)q96k=dTAVH(F;_v%>;+L;j)
zddR$B6i&L|qbyby4nN=9GKSB>fy8RC)D!{GR3ROyhBYytlF`-`;W+1-jQuIBfLPr!
zYbgxA-DC<R*nWytT+YUDXsYUp&@gVwt3HdTXvKgWy@r_lnUC{Z9r`6q`YJsD@&gsN
z@3+IafM5xD>tOwr>t)s}+bnmBp_1<JufKR(WhB1k8E(abZ7*v>{S9P8<Dv(T8%w<3
zErgFD*>vx?w{_u<9Zam+g#Cta-QhaaEAxe|f~ppZdTn`Gbw8;u)Youpnm4LQ-y*g-
zb5LdciDA2rZ7lA$|F^Af5e%qDS;&4`j{9`Aed*{-Otm&qUerF1et>hc+_#s2Gx=<&
z>qi-Z3ASv7TN`N_r=lnb2)Ur3geEk6*WY)gUr<dC<O6@~J!1<sUBG_3m&rqhQJk$g
zqv+;0h85PwkQ46!uJ-M;?s+HxN@4@%hp^cBT7jOTBWhcP!-~J0hn@m6lS%EHh7a*W
zsZS?)C~0st(#57;Y~$Z`Q2>VVSE!ZUytZ@ezn<Lv&^dNT%a2g5?4srw|F+r=2sH7v
zTjU{kNUZv;igZr;BP*J*nIVxxB*Iz?r8h{&;KDy^ubWl{3lQaIW$O6;fb(MvkMB71
zN9HacuyGna;q7RVnABFk3Q@PusUAUv*%|-inDog@sSE6lvEP5+j-X$H^c4F@gEI+$
z6U%BTEyjk$0AQz0eG2xg6^Z;uo?e<q36Y+SM5Ysc-pZIDkFS9qYMw;>JjbgIX5(J5
zgxK77@^BGaZ^IH3w=;kZ$2RS^QKmhQlPZ|$TmynAi!5A6?2Js1&74vxS8)R3NJ+~S
zWt8NAb3n5Iwtfr(X54O2s(OWrmeSqv5^81G<szx9U~u9*fJ847M^czI%>Jqy({_pU
zpk41%#izkW)DC**kDEGvLpx*;#QiZ>gF>C))X14NS%7{RoYVMF#l&t`uF};_3;=YI
z+#<P)Yy(Jd6talA{p7t+)S-#f5TD@=#;8?#DqWsMT+}=yC5BQDv}ZT~1e@&Z1)h_-
zLxi<a?l~~M0xP^Z&p6r;vzea8Eyr`=tt_gP`ehdY;a<UHka{7=IONAbXOQ}J4xUZ)
zTer1fur<`8;y>mCB)~W<w=?UyU6(37z8t=|VJ-~mP<yY84bYy3tIBU80O$=p>EyUY
zi4oS}b)<8Zn^3%vT;}Mi*lpW|MBL&+iOO!p<MzDdu7^h~5lFJjovQNos6_t&w2hn(
zzuMDnjOq>}g1>F46<q=`L7G3y3naNSAjjmW5L(V(2>NKn2ReXv;idMh0o(0R365pU
z1_0*mdaf>*J;LVzFzJM_wDH?Q={wUu@L)VPx9y|RwW%;4h2-4Ed#LZcC+E}5RY8Uh
zU`^fYH#|f)dxN;uQ<K2WbzZ{ZgxjEk6n?C8f;ybKAe<6f?1~R5U_Ql%C3lz2e$9LD
zs*nfxcX5j>IjqlB1%UcSN=obC0ezYm0xPirTtHOR>*n9^mZJV5PclwQtZK_bkF<ZE
zDzITXuzgw*auEc2X<Yy^y}jYpig&DBBs$7;o_b}puRJEI>9ckfTf^`tvK0kLi5oJ!
zzBR%zRXw(+`&lK=>pQ<77|Avrhx)k3I9!0rH3Nj;+g#bI`g1=j_c1qN6DDbpcS!^b
z>UZgn0l4Qy2oNm02*-N+dUBE%ru2{-8IksL48AwLuy`X(1lfeT4JmqLr`3Tnr@vEh
z;gPLm5H2S$`3`%;#u3hJfY@_W08oD1m-PG%0(zw`jLK%-e9#1(mzue*yX{nxh4Km=
zX_}Q0AP|(Z|D%6*n@tSTTunqij|;L(xM}<y9G`9YTO5vF8%+%@kQ%;3p;gAtMEZtC
z9ntdV<QcV0^RXKR;jlfD7wx?TNM3Y&!1X6w4VC$}i~Wc10yA?YkH0NJg>s9$U>)A;
zM=jVNuPy*-=n$^+x^9VIC|d7zGN{lHpzlC1Jv8#FS{pV^l}qT+PS6TXLj-s(MtM(Q
zLG|obD~5R=xeh81-fqBNS~IIzq_}l4Ey~3z1g8PE6Df2dMWTNak)V2AvO_oF2Mp_M
z<%lfv5PXlHVQA2ZP=v-eYH~CLN?Kjuscd89GN>n;DYAh3_Fl;Mn!Nc~$mra94-Y=~
zcm)^N1=s1@=~p(rH)DB}7qgK=syDm-oUG(RP`+^tdF8`(RrY2W;^ekv-sFRkJgx&;
z5SvRg7bi?pSVJ2ig{XW_hX4c+ov8JQtt%QIP>cSp&wrbPw58V49?BJ4`}9JX@X<YI
zIH{s_L^d4&@`xgqQu?)~HK?E-fF<q>1hnwi9+PITsL#|=n*l_L08NW6g3C=Ycpogt
zwUgYy#I-Up7K~ucF*+;0yMUnib#D%)BY@<eW6#DMcRc9o7S18k71@>cg!S~Q*T3(!
zm^;Pje@VbhYAZsun|0-4<x=GiXn`8yd>aD+IlKR!B((0!-z&p|)1u=1P`Ted%d(e3
zNY5#1d2*Mou{pzk5(~2L<A56e@`74+1PfL*f>+fObN~Tp=o@NfBKfl{ezJjEKct;>
zjuxr?{HTRc3{h9PF4wX0m3%~+eAw!4ox|oUCJcgP2c1pIpoolB7VuD8JZOdp#3=J@
zTlcyD^k<o)=|3>;Zeu$OW}ftV7x8%!0CHm=VY}L-lVmyOA>r_>$CNd#HNt1pj7m1%
zX~7My<QBkF85sj;eFBsJjM1DlqPX*jK!c)$-9_9a5lt=nOgQ;#glS^WXHkB(9)*SU
z1OPahG|9de0-)Fv)HU2}%-?P*D+J)78>#A3uJZ~)0=%fSj?K+>zlz;0v-bj6wZ8@O
zvabwyCpfGhTdbZu8=;0y^H?2O!h03?{G=h#MuPC(I}E%rsb%&xVsvsw^%hNA#~TmS
zH3VY>)fyEgq_a(dlGhBJya@WpozJUp)#@oVZm^UG0WIt42LP3`d2KTVV6ovXZ$D`-
zbe7~lb>ez#sVVrEq#19;<j$g%I@%u|yltxOcx#q^*jY*%6!mk$o0khS@USONPjP`j
zY`FWp!Q}W0#XS!T$*)7+OfmH00OHVt{8CLn@G;kP^Q#4bt0x1clfCrkei;z7!uT7n
zt}Dl!fU;AhAY`jlS?$>-TUo!+6D?y;awf4?FB&}l2XJpn<n!mWr=8e$eYSx-Sj-yj
zq)F2&F2!`goPBKciKHu-X1MbcN=s>(!bWMNt6knA+-t8JG4U3Y1BSi(@eo{`3^)dO
z8`9>0cTOl%FXsK_0D08|4|#mk&J)jH3CA{g8P}s7<th8dRp(X^Z?<e4VANx2Z#ox$
zrWa-4nYl;xu|#XtB5s&#A#L0VE5E@z5}EJdYNSEc>L0WCn-&5kn%dh1KIcA;_8)kd
z5`ddb-A7mT^Q1;~U<I;GQFDdHJH^4n(2sp|P|N2SI42(aINzAPAo5=BEe>f2y>agA
zEB${SLfVOC{fcY^%>aJ$WD->r<$2t?4LYvJO4*Eb6jS^Zpy{NR6W3XT?1a1)EP$Q8
zGB~#Zxwr8TApK)Dl^Z~RylFf)-P9Bl^s$Kg&_+S|L(IqEw)#nLTM96(MW+BTM8AvZ
zWwLg2a~#cyjR0bBois+4BMC5+wQfA`%@qxs$?BWnh9Nll|A@^PD<*-opT^%J!gaaV
zUbxmF|208e`jCU|?{>3o+vv#CX?)6O($VqM9nvig9;rmWkPd9p`Xa^qi2csAu^$~4
zf$F5!7`r!If^^%8_4s>WT0P+X@jpXrl1SM_{PXL}O{rbUegL;^k0LQ3ZQk<oLrxr#
zv8`1(Fwd195Be+qg$4oY!;Q7F(ciMxG~NYs2BMV)M0XX(a12bI5V)PNQIxpxzcWT2
z2c+%WW4u-_xUA{fCb7J=&vB*)!+yf>(9^~rXw`}9K{Hge{uD2OXVzA_Oq#i<et^ft
z4j4_UnZ52rCwA6_z23Qo{V*x1;+CVxU9+BJf#c3NIk)%I?n1yVL50K=a&)b33m6T0
zr*R<P95h^q3gh(?R4nEQ2N>0U#c9Iynb|lD{Z$R;$6sY2S{c&RePUT8_5Ol@9zq#u
zK#vQZrFi;`6UektyK#qr4L&u&dU0+?ZSDHrwI84to!2lBW*<jH3DS0ZQx1t&IxCS{
zGe^FM+fuffF7a!runRxdX8KAk-#dDrK;8*2H<VJ%lP@cGIv4tjk43M)y(SvY?`kyl
zV~q2czLlyF)zapN|NL020GO`3P-9OMhQlr8J*_A?sfZi1FhW)Nv;zN8Nx7Kl8~l3E
zhmpGbiEyCYU#6rwxfVFM{ax3d-{W=$=N<n%Z3gILiaWC{EKke_E%A?$MQ_j{Ky^<x
zwK)9|+jCdObr*cp4NL;7UP~$!bSK-|Yf*Vo0gh81hJy;{w;EZYE82j7zV`!vI)ve%
zY0Q;b$=|Nnk{9lI=Q5(;s0R73RewAU`bJpk_2K8@YcAL3a&}X<EH48LLjhV%Cn0se
zwMgUr3>$sKK+_$>Pie*Ahvqt8ZlUtWk~SdPp*{RP<q3at@bcNgR^j;9!NUgWS{|c-
zR`ar#KV<P8<fXb_a?o2mE?j>_>EEiU{h~WqS&DXVG5}CT18hB*Y!{P7e-HHs%gU0H
zdpP*m;LvJdfzRvA9}1c(Bn*yq-ssF*^r#nITTMl0iu)DroGjt{VJzCDRpMo8nSa?=
z#gfHv%g~iDYn~RVj<7qEDUY{2IBuWj^{kfbt!j9QVs+l(lE0bXIa?Qn1~l<KO<r6R
z&sX~x+cNt*4Gr3yAi340*7=2eXvPoQS;St<PB=}^0v&%+w|&CULox;Uc57Tg$)f?#
zw<usy`FtJFb#;o*(CpgchZ=r2wL&8xo_d=#0Bmt_pavvte)}UGo+hKGc8q*|qx0Mt
ze^Vhtr^w9y#K0bq#_7`j-r!Uou+IklKpKZRnuGudh@1)7D8;wYGWe<$Cd|VVC#_HP
z<^v%L==N#70a9l<y)&Wi9yvTzob_^0mYS+?4d_7GPbc=_R~&XJDYs#;Z&?L+Aym97
zd^OD0x-y$%<X_<3gV(npnf}~YPSEZKuI_D&$s^L-`(Uxz+hO(1rq^a!R1u8w_JB&x
zkZ}6b@TH(<`OJIw*ToZimgmZza1>L<I8r|m^qV1F)ZE65ukgV8fhkc2>7YLavwF!3
zaw78MHeZj!YWd8k>=?Ab%k@u;wzOW$NT6>zK1%^^yOgO46?$ibEM-Vqtp>9Sr|N1m
z<@*PYn<ERGzp&)468IE~Ic)#4?qo**dXpap$9(q@AS#V=Z|cg*_Xr>0oz1xu!Y!77
z>|{7(tRTcY{hx7YPaIj}_R~%M#44jcn)F?W?CsNzwi`@s^~96!qs)+&P6zY;<Xix#
z?WGQ9{?;GLAM(;Fj?#eJR|9|*EIJ5{Ral_aWXJF3#^VtuvT4mVb|$`6Sc*~1Ei<_S
zxnCo7@%i-v@AkSE+%gvf4f;}AVjD5UD_CofJ>0{*fImJ9NekKNKi1M)bKHto6F`S>
zL6!Nrlg=dSa`ogwaZqdpA<(j2Fvc1Q{$|EFnI+))`vxv8@%UG-!gkTON5l`*Q0e5R
zc$!Z)I4OrWw3x;LfL!jLhKt2e)ZXFtKWH^h?bW=Sb-wu+kax>N9K$nvsR|8${`PKb
zfm%P>ixTnC(6<fo&K}EF9wJ|*iDX@?%zUdfxX)k2Yj@DPSuGi*$*w>+xjsX|&^Fw6
zS_NR%6pjnCK!q^NijhuTNDmeY-8J&Wtd`4RNJ(x{S&ce7WYLGD&1<&(ddwerdv;L|
zp*+DuU6cB_(Y#6S4RD2t{TXBwBLy=!kazX-i-Y(9i6dw=>GztJx#o8g5!LkyIj6u>
zKG$!^tgv7$IQ)_7OdY}KT@iA~%6N`7ABz1tgsBX-os7hL(FCA`A8(o_()mQIZY0J8
zdFh1f(y%QS(u0McZ7byz|EoQzbR-?Jn3O!7U&wH5xo~&**Iih^^DeNb0Wr;#l2i#y
zJ?RD3+2_w>?4{PQihBu~M8|~TM+zmu3&XvurMJ<0Qo%s^(SKyN$urP(lvef3`C{$4
ztCE)4WN~o5Kvi4BrgS2*=@UnrLdil_^5|Ed8QT>l0-+ZAA#sy4X<d051J~3LA@h02
z_`B7lP#$%WUCX}#UY*oV!PP-$>AY8;<KZ^6vcbymg0+W9#uf@&3gDXIDOobcuVu@+
zvuKNy&qc4v6F7c9P#}4!W^})QS8ccc38Ovds?7#q42|e&f{DH0PM9|p7t!)X6_cDq
zOD#<6wp{D>YR)NI&X{+W+_Ns6N_uRjqk6Bd@b|ANP!)b`qQMGKYb*jhc=yN>Bs~Ma
zRdxAc_wq)t`yS(}_H=MOUPm3d;9cx@$f4oawz<1^z%kNVO`GvS#8u0r8l+~^l97$j
zmk~hC!VB~g8%8GPLnnRWCd=)auI%l^LKfu9-@^Q5MP<7_t=M@S_$db79no6Oyp`CZ
z6b>MOcoG1F%C-&fL>>FRxJP-(nPl#dd$q>Pz2DEMg$PxJ$~$%jnVakdAxRUl8(}4;
zFNkrb#W@qK_qDuaP)p_<Dh<PN7A(=mO$6vB$_Km;Q*V4KW5@sCh!Jk(x773K{)<%8
z_eF1-LqnN^mprNX=v>oWI0-og4SMvTO|0$}V7Tq35L`otbmn?;76-6>Rh6cQ_>8a?
zH~a2pC((&fw;d^M>3Jhmfk(wu|Nd9@N6yc5pROFM_2@FNV9sHhkA1)l@o`kVBqk?n
zAq^LJ_SAsxX%WJUA6nhRJUaU}2?T!D;mf%q152HK8GI!U8aMgd$2#I3!GIP}fUN3)
z39%de+9-wW?D~73_}Rx(<v_z-r8=@b#hy=1;Zg~M%Nc?UV!b-11PsB_EGAUV>ga0b
zRRU^c1|310%vcm6eK=V;z=)QNpCj?hdpY>Vz^=igt&1ONyE8lnw;!6-<%Z|#zs(h2
zm+Gl^dK(e}9Cl%r3rLAv7#ru>$fXgapgTD^SAs?_{+oTxdbx$#Q+R>Cu7hc1e4}6E
z{(I?&QK+p(>xceYq*$>2-)WlM<wS*>D5CE(M{cIB<2fEV|1_PfioE6)-C?~jPLN^?
z!#SKi+WvSj1OVwKyhy(W+nW5?CvLoF-1v1gesz3_`y;I}h}u=r`%}=z&Z(CsMHcpX
zi|B&G*Do2KoyQI_$WCA;OTF8@daTOslFpU09*sg7Cebyy^<DZTLWFGjt&bbuXEqDk
zi59|hDPYN80P*e`$xYESlcLc#4yMwFOY+u2la^G=Jna}}7>h<pV|SMu&dBz$`cWOU
z&n3YF*zQ-H>w<;Rw>3Kd8lR$Ss?_oo<M(W}cp+y&!g~+gEjUnw$)p6G+r1{?=BTbv
z>RE|lrUBq$xWA<lV3S&!G_L$N`nguJ3~A$|)>m@m?4&nw!IzXSo|{R|5iirl4U3Tj
z1Do5raIEgwz}&Z~FTJ+cOtW12?Qn(on&FIM^i;V^{#bJ)HVkn0`>NpNV)^%B_QthT
z0w}%iIC1=&Lz0#>)Nbj1YXK0GT-*?7fILR=F`H&9a^;zwj@A2IRSYN9Fi)uuK|=a-
z@iX8=jsn2{Ct=y;n0SdQ@YRC3@5Ou@@m_J!XqFr*1hxFZs-3OC$+w|edDY(25fCId
zbuA|S;#7VL^qRt~@pW5%a3<Te!1@~^qDC?Mg+minrBt(~*hK}660{hWMWuXq|L3<F
zqB&pH57v%<b3YePqc0uasfy4nk^U!`u_c!MLTDOk{kxZ5C!qX3dpv`(UP&TSEos6v
zbgY&|aTb{5*696We8o;NUA^Ws<)2TF8mg7<ujeC3N9^P1rlL`xa)u52cCoJG_Hl5Z
zOHt)ifbZLP0v;-Nq6<xD|6$?9My2)=Uw^Y7s^BEn4h&|0hMD%|K!i#%I~hsB|J6k{
zJ6Jc`I!=gIVrwUW3i&U$`mfy`N1Hy6Fc`dmvJp&P(wgi4xluMFaj^lkslSdUIE>_s
z1M;>+<E(=ctiL2&pu$g<dDQ0}^YR*IQ`ASt;hu9k0iS=iP1-jrupkL#^!iq)K;KoT
z$skgq$HPOh(T&VqJ7-p%?Q%c#8!;~P_~v^Ia>;tb+ZkO5RJu^(w5{;$sw7#ZcI8xx
z=x0i6;wpdkv|s@P>ggQr^IEwQ8-At^<$S=}=JCtbIOT!U$NG2cmFgg(U?jdr*W8Cw
z`n2w`nmnU2W`=$)^0mP5b~kOL;a2>!V&of<V}$FzO~skL>RwuHmu)mY5T}1S00rWl
zIW<d(Cump4^)o}$VE1TAdke0)vmd;j^MXPFh)MM@zO_HpF!@aJ_sj_j=6O8DhOtvC
z9CWRvTpl8LfGM5fKL>;_#vI(bYk`5(OK0>+6Bi?6L8#f^LkKr^nns)R-Rv(tK^k0N
zEp@AZq;UHa0t7WZpvM*i0vE@y3Yii3<0|^vmO2EAArk%pp3OzZr)^q`*6N@f2@T+o
z{dr0}sahr?;d#FaHaN^k6YtjG5JQcyUs_>x&_pJpj}l5M7OU;=1zN1-@wJOJoM>3F
ziH&m8<x&0AU3MH><7?XOvq|Jl<B<4p*Uizfyine)VBHtF$3yb?V`kMuo3ozD_>GtE
zP%Pfxrgz)z**vq&ueX8-SnY{54ZB4YQh}<A!!grq@N4aiYg}YbnUwt?htQqum7cWE
zl=n-ZI@!MEfxT){F&n6uX!3kZdgZ^I)kfrapq$!)-)a30qWL}pd(o3gxA&%o4IH+r
zmpwon-{LW^ctx#kPaL@=>yp2#s@aY7Htvnb>$x;b1tgP=paitDH^BS!h@DPxKEY>N
zq}-?&J`YDIpt=<dd#Qb`Ca=!eN=;1%i1{^qe@S-=6N-C_4CN3yA-IRC0@*mn(ax0G
zpH*+Uk!r^u$|7s8xa6&G%IpR2m$(OuTHsuyy6TgCA*pTaXXBMkiY%lmf>AZ=`~AbZ
zeuNUhJ~sZC2}nI=2>2=u_x~idc}S-`+?1)F1wJt@V}lJ+pOD8rumUXKq?tzF3%!$9
zdMuevw=uNppw6FUdV64a&+-NE?9g-s(Du?MGT&Ult-Vr5+~G8Bgb5)CnAIkD)rCCR
z*vzVvLmu>ZsNT$yvwn?sFs%wv{HQKo&qJ9r%S_Y->!Q3N!Cw@gx2gnSU0DqVc?)$8
zrOXf7yJ_Z%kufBlq_5ZPG|iyTvU7n6IH<OVT#l3q@{F&S;yx<Q^N~8r9082l$2d56
zeu4nNz1^4%KEQM6k~f*A<EeUMNbZ^FcF}-Xsu{`HB5zL7%CAd8C#0o%psLK%$&$9S
zd&f!PV^jnT^j`++CEl=PHb{&H`p`_phBE^;em6vf5#90Idnj)<_nyNNzY~D=D}?9R
zs~H>Gxwsco#71Jo<8r@zNx@k+Nz_vxRUeZ*BnYmGBc9NNzUE>;eqrgy`s+y!O#DK1
zBIkubba9LRhYL_eUu71x`0I<jq*Ev9hjp<~4J)T#zH#MsE0#lF^r}>d9!q4`-a=dz
z+(*1Y)l-c9Fw_63>GX?VzAsxr{e0g`a<Q@L{0Q~ZCQ1rdg7syZtj8Et>_k^dYEh|}
zMAA?S^*L=I*~h5wWwk1X!!{XqjY>y=*+vIzw7d#{3xc+KJ^;pDTQ4N`L$|bS8pQ*}
ztJL2>g&1o&lxy~M{H`~IVBJ<fWp;(Y3f?s*{<*|w?*Aq_zb$*E!!Gr;snH-i(?r1k
z3Z9?L2&%o~r66aK5Hpz00SuAaT<ib{wE~_!U#!Q>B+3TxCsR$!%<H=17Ye?pG>6@P
zGRZxES}rpM=-_JW51@|in~pXh!s~BwHAdx5Af}c+8f99!Da;c1Vb5K8fR!k>e=LbH
zEtg)d^kE=;Xi*G5+Q)!K{9-O}CTfE<6<sT{I#_xCWEq?uL)qZ3l}w7~Z9zc`Kv_-x
z3+E9UvCDe6!<0~N+mS>VzXNA6o6b_9JcvK3Wm#rl=bOL-2;fq5F$TZELb4BlCRSKR
z8AcahDfX<FUHs(cw~rvpxi*io^UchKjxv9AtbU1UgFkU#=RHI}tvCI`pjG|?xG`#8
z7btuOoeFEFF9fPNi(fJOF9#lB>Spv9?hKNiyXSH2tBEn@r(I*ZN0VK#)AE}{(GPOZ
z3p}vJP4CZXMH7!IV8Z$TM9}}-EQf|ukI!0#ppCJQmi#v-7`e=tKtQ9tFSZzRiQKQI
z<QKxldxDc-qSV5`SjEX|VNmLS+X_WI5{bS^&pd5fEsB^m5Llm2n5kAvJlFmzc(0Vx
z0~m6?S7HrU_{GhgIdO}wH_HDeV8NUER)VY9&qC(45uhB-KQNdu5uisR8FXW3hpe21
z1}IOPg+h?#5}8h0vJFy;F?rpyzCJhRys+nBN+^C6Aa6It0Ey$vt$sTBmW`yZ{>Qzy
z#W9h*Y>q0X^j?Z5K;YA>kSk};fV*^iu@q+9tfCo1HgQ<d&QfUwe(cBH6g|qfwS_35
zBS+kx5wbi<2Yq%L7@XcA4oC!@`hfJC&g1QlHD2%Pt}37W+=yX2fi<hWD&-D`6u4O1
zptNV5Mx-Wo;W>`RCnoMgo74J9gh~0u{UOji9uOjJ@E-^_?+V5C+fZg|b+5z!c~J=A
zzcnD5mOzMLwHJy^<XRuqllm{^l1$DUZQnz>jdzq!6y=7dJDo1&rgg0!3&~wr%gdg!
zSyW8P0E<(=X1YH{a<}%iMZO>Te&!6oRh5zzu_Z0rg-&m4FMn?tHkqLg#|pVAv+@0s
z>VmX1OQNDe-jIdTnLFsfk{j8K*eFek?i4!%RQ-O6lZ=e^`9}~S2Hq;q-K{*D4N*ii
zIrxvP)!-%7Xl29pL0s10QRfn3_1aOoOzuz?oQF}hoEuDh=50~E9;u)3fOP(KM;U~b
zM$sAt_jwZ04;kY!B;Q!8oD)tQbONFfo~z1M;PrC(V_W|4JAG1`eX`}x`sa7QU|?yD
zqefYapr4!%C|9s5><<{pZ}>h_3irbj7t-r2`N_g_2?UVDRlRF$_hX%Qp~<YG%D(rq
zjo)ZFtayxmZ|DUsts?QZu!>*oQ`Jz>xND*P6*37}u<6n5MNj>PXek!ruUByxfI%`9
z?YITd(un&^`3BKnnTx8GfJN5t2Mc_#T#7C%C!a|;*h3YdU6){-r-RK}ATV}sM}*%z
z6ulN9+UMTQ{hQ}<Y(ai<7pZ@hKl3^XjW;qpI}NY+n6-oc1EmScqqZ@mqMido{}FXu
zd)R%fUzt3f3gG8@Liz?-^ps<TphyoUjIwH2diBPX610w<swZEK2-cbH6eF49RQG;O
zQSS?yF%uv!MCe%OQ;r?!0~Y2*+FLDQ${x=@mD%tT-fqoa+NTtj^G|m~%b1V*y97|)
z7U$?4sj&8ATL}%ErsvBed0`bQ$>bhhv^A`30S<X_LB>ZsN)kCfWEF7XJeZv?2^_31
z;ohUNRx$(ge{D)AWgXY9Dim<Pb$Ae~8Sq?=+59QK8LnL{EqgMPo9+<=RIx1p$LQfH
zl1(<>*|u#WvsD;WTs~K;V7aEe9sVP=3x|l#*_?#-<YP3@fhS_qs<^sFmf(>pb`8f=
zyhn&84%d~4=A!jSSQ^cnWQ25q20s^q&;@?@?{>>AxrA3mX{L1dAcTRQ37WSGtG49*
zc#?8nIlO63mkG&=`pWk8UsLpHFRhrk{M(N302^deW`Sk4=Bx%?W@@&v$#gi61VK8H
z`#iH4ixY+5p$?NlF2I=0>ncv2$h9gKlug4_eE|`c#0a*}Ir9^(+5Oiq@Dm8n^_W<;
zqB}I|Wybcz7Z?z@fRsJPCZ{=X_sSInw>DqzUr|<vtWmIU*KV)jUj(21L(vJs>BI=n
zZX@S<Qf_?*k+=;4Ecgj<u7NfVg=3mWS9Yo>#Uwq=cb2BrDy##1<@sK>2fuX5wa0BY
zqGgnbi;+O=^^;;=gW#M#8V^n|O$a6V>k;UD`|Ka|y^b~TQ?go|4rD$y)8*Q7dwsKP
zJUTf+&_kSfRXH!dzwC|x1u`lOnJCOmZ?dnm%Y>^cdJyLVk8JWsPN}hu>D$>EAw54Q
zrjExV-J^Y7l+D>MNDiQ2@P9j<M=Y)<JBO6{{fy`mD}Nk*>x@*c5O<kw=l}zd^|{JY
z!@ve}p9he><eJ>`MO(*-zjt65%WBXOveX7tm6b(N3J4wnHtEKe9tKD)YaL_w*WW0`
zyL@0S4XozqaIPcz9%KlQZ!u6j@5g=NQaIJ=7Y5IQvecR)gPtH4erkvMDq5b+>#jD7
zac#gl(o?T+3EF@5K6lvPA;qv~kz&ADzYa-d?_T&c*wEk{V<InggrWo1lp9fKKNf~h
zmp$EK9rM&h<DeK=>6kUg>+yWRSo^cs^@Dj}ofW$?HZbu!BTv`FK~9<8Gp8Nu!fa-u
zWi6Z1%yA%cjVYSI+i<ZX5$AWLP*PNYg~Ht+YkSD3hGAcx`x@%+S><{ErNIM^zv(aU
zuC9pYuOA7HHdI(KvHXk5n(%(5P=I8=Q%`wNf=&ps;JQB7Td;$Hf>{jiA1iyu1{jO(
z8LYOLJC}a2n!;@;T3<!huSr<f%&Wgeo66XDt$bcLF<cW8Ek_kupTy86A9cPo=ZY}N
zrhyshq1&AWx3$39x857EGai5g)lO$KP=T>Dv#p(|$?Vv=ihh6oy;&W$6aUq<cs-T>
zwVWMGU?vB9&yDnG>(8z|&UQ)7D?ye0po_1zzpA;Pf032>^jFr3K!B&3_3#uRO7Rk3
z*-~isg}6rJgH^?LJ>#3Y@MoYvjHeKOC&aBJe1TLKwY$RB>M)8`-p}SSxEAFJmspql
z^T!!|zMXQU!fxrT9k;KRh$6HhfOmL8QallINWCSzJBC)gWhf2fY~JR|s`J*vg}SjC
zIMgExy%YLo8G<kw|AZ24SEwPrM<cbmGgAJ=o3b+Ln`@fA3<F0AzSpVZnv6FM0ieSy
z)C?2krFqt@Gy{rqRs<;WuWgRG&%OwN`poVOvU2q!Z;kqE#lMQ_Q_A*c@Y?6wXLWkx
z=rHN1ZoF0HL$T6cQ>ZHnq+A$M#TN0e5+QbD$pUO^V-OzZBwnQ^#pnt);=6_7WSIhT
zr4FyHT<sJOae@^8d;;5Eq&-7tm@8O#&qD+XuwlS8)wi8aINqy;$D9EA2_5{3+j|i&
zBj4ug!jJ$HRNkpv{yWV<X_=E+1n<uzGA<3kM@BO@f%CLgq&}m~v&|4nN%sV!qes$2
zFCC8K;ML}XBT=JIWmRfTP)=Yv%rf6SjDtoe#P`SKKP<1?wUY~MO+1O#aAf7IX9;Gw
z?2WlZ7l-mp*05Fc6XhMtOXKj^PTkb8VQY7GYtP}u8^bF(FdCV_GmYJAMEMJObaB-q
zIdB}WJjw0iSd!PWb(eIY1<5+^|J2gQyLXm`riXEcLd2IgW55t)yno>{=>wF8`=hk)
zV<nnT)BrM5<En->3$dFM!DoL$;mVR&#`I50FsXJly3X_$HJkTOf65O`hEKN%-B=vf
zk8z8b-^K&bV<{Ep$(4fHO6+1kx__J_JruZrA#5m$jp4P38dSWF8=vIipv>zPiTz{q
z&zry5P|J;q8tZS|(^oh>UR#WJWgdV=@J!;*%x|xA8CZEf!P*yeZV+FOIgU_=8fnbF
z@;=Py9H}y+&o4!o8YH2y51NcqZ8x&2dgG=WJPhFe#=WO~{r0(77ic4!nDD5(u}ZRf
zY7m9-+<fvJY<o_>l=i|=$a6)w0flGM7k;MJ17XsrAayCbkxm33($-2!{~;ihkmb}u
zPPkxDnaugX-U2_H8-C%#dFL5p+>yKtglDJ`{d6(85o~FgoECg255-P?7ZpK$IdXtE
zhdow;nx=_zaEA*YSSa?7`~|FGIWHy8Bj@2Ht*xcz3zb+z-v1huNB@^KCc@i2zuw!C
z4A@^?Z%J=bd@8$5I14jjtXW41NNvmf2`_&$(ACamAGA7>S6HUrC9oR#=3j>Hwaez@
zG3!)msxmatw7rw<V&hi9F>=W1)E`kcsk@kuvGJNN*Nefpjz<<majx2)ml-1rsi(pI
z8zDsytn-pIKZ=p(ncL$d8w>IL`?iC_?cvj$D8r@ijmQVXNI5F-zXV-ytp{rb?IvEk
zIa*y$pKfnO=!f~tVaJu6yd%j}@Dpv`Xg<2?p2pE3_@;n>#~U289NAPhi5X2m!nSLg
z0r|+W@iG&p8bGpW0>A1mZ8yO|RGco8R&yN3>eL(h8Mh!v%SjgMQD>s6z={K}6q3s;
zh3W^~q^NR@h7=G5nVOImLYCx=5+~362@yV|4WT{n(O<3(s0f=>o5gnjW`#-9C7s%p
zEd5JnzJDUhzE>}CGuW}D;GJyL03*eF4DZ7C__mP`&XRU9mB1T~A*GV?SzHgw<c-li
z<56S#M6S@64}D%=dJ){e2pnn|WYJRbfGlXte(L`X%t13>E)wHCa9<=GykZ`J71f^9
zO@BjOOPS&KiqQC2-L2&im2#CZz_Q232n#%onOaU}lqW%S*d|)TYw5ghYRj&4M_fH%
z5w!)Fq?dtZ9X}Q8`r0J4#@9x83l7BVY3#u}$^?~P>C+@^XBQDA501?T##my1<$JN>
zZbTaFPFeBk)Ry1Q!?JA7BS6*fCfQfe&%gFFR_l5#uNsGE=YAREo{0Mf$m4Lu>>Fdi
zCPV!WsY^mGE-{SNfU51zJ0b%&Fero4F0CtoCF|9`@p>R_>-Y#9%6bX^0DdXsIW^0~
zp}z;8f@0<@k$Ho-?2#*7REv1fV=IqpxgaAnN;`~Akg?#6TYc1%ga_^98o6(p#tlr=
zxf~){rloyi6jx~6LDK&WWW7tPXwSir^I(xkpfsFY*?U4S*OL|v^XMx3o|0r;QH35;
zq4Il_vQT$gN7o>NN<ERNm35Fdz6SK0RQh$c#!0u){3m=owuR8`jwr>inZ6yj)WFRf
z(u+}K<j8hRdM*3w?#W5IIyNx8>A*2QVavMe#Tu_ZC<ED#MX8p|8Br{ujjjq=S^jxY
zXb<J);a?BQ6weXETsj41OZL3SO-4^eiq~=o_m<`CLK^6(1J{d@R#4mTR$uVz;&u*w
z2E>Eeo#;MUDf0rwvMJci*~LB7mHCB#dTj9Vq>tzIFPk28bhpqakTlkdf$$<_n9gu0
z=NCYNPDI(4@kYD-rACJ#UaY<`9djH19+Zpy$|FB^MPJBsShcZL+HJO&Dnj-Ai=I1x
zw()XR_3v~R0tQW`DuSHUGZrx6vvfhy%vBh%JtvITW&7(-B*VafiB#r}b?y(;JQ#0M
z6QMncAufi&_Ddcg(07)}YgaXH#r((ml02p@E^z|kUOs*N3mV4|^V0dpbn~!jj9JoE
z_iF!HJ|6o7!L9O+CmC(che{Fdw>~H~o@D6A-=9MaK!?ORYLh6q#I~&|9VFe}d{;<Q
zBY!%ESQK@+!vE7`o=_Glk+E7`)Zv<Z?CF|DEiz9SG$a><(G(N3pT<6I+CV*!`>XOE
zgL)#VUrf>1;%w3x{vWl{%X_v38w_ux##vuuTO*=-)kGP~38Rzl`7#Jy*&}?k@uJpc
zxNBIZth_(@(_DO?a*NmpFIR}67$NVDd0z`NlN#9uU+mR_`EwN*4fB*)6s=h*XmzyU
zxm9fGl(il;uoqYoUAoTUhUXr4w;|1QL0<66bt#NmdXLcdJ)uw$(c(Y1<qt}$?wp~`
zF{nI718Ar;PP|`hjmQzZmbVdT)>#JJ9m?xv&f+|10{Baq14i_Y>*>-Ynq)JU$G<_t
z)!lLf4cK=Qfy1BIbw7NX;iv%`KhE}Z)I=s%ENNkjcAG^}(bK{NV!sMUTP6i$OAy|0
z5$-XTg>=QVLh4K^k6vQvop|q+DxVD*w>Jz8Ctsf=;}FZoP<EDe$hs^nshXWd{3yq~
z0!2CXr?qZBBI-d8Zo+w(QJ$8k*0}e~UYp(=k0>BaHy9E5^RQR{2x=}r5qk}t>|2*S
zUD}H|%-#6P8XZJP(9LsH{bTZT#4I)R!TIY<(+A}XwM@(UeD*SkmKv^?7_!lnPBcV*
zFW_CCA3f4W9{>2$QVhRSkhiTrxw(6Iv0cz)p$Ru@xQzOBr!c$`OM5cTcSh>d$MrRM
z$&@boM%Rp3Vt6h~^2?1Aodx+<bs`Or$5y2z$ORYC^jRjKS9kp93)SR0Q!z#PH}U6$
z6@H47Uuu1S^kN>~bIB^bXL?tj2$CP$DTZDXqPCxn$;M&ciwFrPV))gMk>c6~9a3BF
zGw=@Ogiixw=HDSOa9`I!N?1Jn8+z$m3tB1`R_2HoaT4JJN05r<&KAvA)yyk(4z{*F
zat*C~`y%Onr(^}8Qp!2BF^3YQuxm5dh$#B;>l3bYgiDIkaF^e1Esc3;o*z42U|<t?
zeSDYH==`7e>Jib8mL1D_QhJhYn!*p1h#vDKh{ej&ZvO_~wbXZHQ!%#}#e2LsI0S70
zS>cs%4PZ4hRMib*jZ$fjMA41wbQP!+ll5{5znrk2sl%Q{ZIDmzS|J|}Bibgm$Yeow
z)E!kF3o;CF8Q<uQ&YZ$rG%YK`mi-t`K1sB}d^}C>MI8G&k(WHZV<=kCTOij+eO9&k
zzkxbtB>S%^15fabPQ14`iva7e&Od1}Xook(!3xAS#c=-!nP$>d5FN%nE+Sk**A#eM
z>36iBbgZ6Vkg%4JqmRsWMk|lW_T#jp@c3_BZnQh=1kpRdTE3PAkflc8>tqNkM?({4
zi<P)U6+S%mpZKjVkE%N|W^TvZcPa!$tRuS$69p#R(P~Bi#yeDS8oyLL@CFV1RM4{C
z`8@Miq-t|~9T80S89pd5l#9QSv%SB%Nh+-x*9(Zg_T4gxN?HlBiHfL6Ot06Q<QUE7
zm@mR@Ii0s_a8M^=Q+cVYmnaG1e5a1JyO#HzK)`4Uh7kSP$l$>m_wjctgb8X5)m*mU
z8=}i?HJfw~ch*BoV>Mi4FSsG0ctZojndZ|g1*21bx`Hc3=-usa4>H)KGSnq`MMq_|
z&fxiwf>yGG1Gaf7Tmq~LdOSyqhU`w5qWlrrkK>Zm=?{BFeVolu=nBy_SPjz`VT6Sx
zn@mo5={8K%MKjNdWI*m&B5GJK6+FrgUh)fHgu4$MP=YMI(8jt@tC{<xQQ5FJSKvL9
zb`%r}Q-=}5lhkP&i9tEc7G$t{FOMd2M;((YdktFdg}>w%U-qU6>%1$9xhXJnOh05z
zqUeJXTRoN>M^GGG5G^)gL6N%N)?6}GhCNDJlEOI9aepW)BX_l@Bp2;~Ts~H<Wo4e7
zJ&FP7aj}(D_S2xUU#0Q&_z?vzO%Ux2Al|8i#30;#SOO2lTa*HtpaR6Y-%m#iF{nOS
z(f4hN>CUGAOBG$Jh1S9~WDKe?;g|~~XI(sU*{?^gr{gx{cmv{#;RoD%_mcokq+Cem
z2s6yNrAq+akEH&F5vajOe2-TbNJLWWfB-*w6ZHx7UZ%a{<U@oh$s{lyvhHn>kEXWp
zmE*WAK=96r@LE~H1u>Ta@s&u>T>K1i6u(wggGJ`s2_*BGMU?c^>WvYhZ4a~BAg2w1
zi2X%aGP#KG)d-3yS(npB%}>;a)?dia8zY_DMiSrCNGKBS{xc$-$xFD+L&I@awUT^1
zFb-W~tbXu(@PY0IBZW^lNR3v+^2WwgRBVuP2jy?#qfoa%v?7%u1fM|KkvbHBJSE28
zm)UW1RB1h}!qE^(mLkLVG*0T6WIc)P*e5`hR?W0-Fn{_Y&`~OAAk5J1u1?pWUGP4A
zh(a47$YfGE7V=J1D~qyOa5cliHW*?wMJ+&q4c{ErI;aZ0DkX{CRyTVhLrT<AFt4Qy
zs@}^8dk4&?9!3nnQ?iYR&_$aQIgA9XMWwzkIzoFB1BY3^VL*bOxRu4ieg$|+Deg*W
zfzUl2JGJY5hkVH;Y#F-YRvy%&esN;2D9}C+!>AZEp+?u~K)jN*+rzRvFoU{|q_3r!
zpaWiCp&0TJWHROW6>hhhrUk<h%bon6;3aD~&6Fg9^me36lSqa{@q%L*RrH`-j8|bt
zrw#^~r&^hHeavqVFxPxRLh#QyhIAzn&W{@XAsdT^c-2H<yi9kP(ZD>Bl$GAWo03nx
zyZ8E2*M}tbgVLipHm$u3DQUL*rw}ti{wb*J*I1Ra7?J-nMY{Btea*c@Eap+q0T)FF
z9I<Y%sC3IOH1PVn?7uGYHTzDPtA{0^^DT$FCxQ*^cx)PRCZn!QZ*g&TbV(ww+D?fS
zKJjt+yfZm3S*j@ai7;qX`mlQYjJnhKXDo&vx#BMSZ_&O}(dRdEYgW#TqGnegogcmk
z>3w!xk}~x5!_Ys-8nu0^#ZXGPvwM-scY)DKfmk`6E<q4{pOtt#l}k`HVINK0Cs4Ww
zheS!L!q&0=>U!Ihv_c9tp|)|3kwAR03Xc^c%85@+bt0V;(n56CzT<w}Gic{d+VqOD
z<R_P3tikzbXiO-Cpb&zk4EiFh6ptKqDwTo)hK!cb!(7!lapz)1;04)fp0@Vq_B3qe
z`A<>V_lWDS>g6vzW8W$y3{mo@V7s7V)FAJb$SUTV^urzedrUDRWA?q4HR-JpG@dmV
z@jR!XfghywTC3Drw*od8&-++!hoK$EVW{VX?O=Gsu?*c0@!Tm5wmXq;o2w+Z1&{Nc
zrAm^9`=!3s_<9l~N5AN+M>1%y)8$!>(ZRIc<AK|PK32G+aq#6Y++KZ&r9eVWOpkg~
z3hJ*HFZlBO-u4q#H_KZH6}O3o)va<v5IwLTjOn)WOlrKr7cCS43fl|CR5{vkt>Jza
zRf$N8HK1OJmZ&2;3QI&S6BNK-w$RC<$Mg8^o*)G~GK$d1cdT>mTnzRqNLX%9&(Nc*
zK!0)TcR;AO+!cbUJteZXT=hL%d2Kud_)4kg7jGh|s5UN0(cKa)-L$s?Yc1-^tr@nj
zC1S|N99s@gZHU@@5YJ&>*coA!Cp**$khyru#4iR^TJL#X18H>WVMd?FC$pED$d?Vx
znRlrdd-5VD5w<iAdL#3ni)_f#8|{@;x)Ab4{5jVK6f2VG5;gA+B|z1tjH<m=oAOaU
z?-=kOewP}@j@eX`uaFO>kaym7upPGFH9>$JO4{^?IwXh#*x$dg+cKA{T*bOkruOv8
zB`ph72>bRVhIS17tw1`jjYzCA!%5}`qSQtx?Q}k-4xrGpLK^pEZCT2{>v@oos0K9#
zovUx$Q@jhyV~XF~&BS;x6`nPgmcYQcsbLoYRsi+hQO&t8J^a2i{h*4A^eWzjq-Gj7
zp!iLVU1vu;>mwVJk^D-6`T*s8I8p=$#!zB+GdnGeSiT*Y`ufMvD_>q1bFJ$5aegg7
zxdL3Qm)co(xf4hgchSGOQ#AQKc#(tjk4oH-dewjo)9NsQaW&*!J9e76-oloK)pBY;
zOhVQUk=LA2s@lW(P*X~dbKrHTqQeID(w)O|y^g5zna*3pCI3KM`|^N&?E883iJ~Qj
zat4c|94Q__u;yQS)bN)2nj7=2m=}kwB0*nz`ha4yo5ksVikwHbsG@7NR(=81335={
zCj+b<&lQ;h9D<oJEjVK7f|Fk#lI%6&4fIidn8xJN{QcyW)_LNm<n?zfxIORV$`xiP
zfnem>7M@g|Oh&4v+k4t#nVbce_a1F6lP=hmTD9CAtL0)`%k3x1gkfgHmG7pR-`SY(
zwe{6CNf%&U^wpyKR}5qOXf_B4xu4Nk`38<}3`__C3r_BHh*G}!8A5VX2*PqJzn1Mx
zO!nl+@~E9&uB-6e?Gt-Ykxz)fEY3?No1m@54AdlzlTMd(77Y+gR~kW1iW>b^rOzE3
zrB2YE<U8syY#}EF5d!9o(ab<ebu)6RzAMiWK-#2NB|PuoWTZ?^<^*3;<w?@(v5e+W
z<R$BMju1h`H#Z$})b8vE>Z*xwlgh(cgH3qOo$)PfMLA(%GW$3kIq#;h&xS)|hFt((
z>W)7c;l}clUTM-Fu?ij><vg>mU*K#Ic!R{j^hi6NgCJ%zEkki%44|fQWD^CS<G&>|
z7IRP(O}kbkG&+{f*uY$S7X@sXhRmmnlJ`Qq(E%sn`Sw*>i%Ox&@8=(Z*^<Zne)I_`
zVyS5-|G3itABgvbw=tLG>ahIi7|Q-ev6I*AJ=rVOwb94BP>vyr*)a{&M6O@w(7)Cv
zj{HuaD&W{B71fFp+|Q50=;=xKweYST`*yR+tuGPjnxM)a<XfBWO;T@PZ4H-1Ps?pq
zu$GC|0q#~0+0JRx=Lz~BE`SNT{mE5Eg=Nputxkt)jac3s%f7i^4{)zW`YD#|0=a(D
z90!^y2BV`6YR8ZrXj94k=qVQ9hEr?s@Ku{sP=6Et=o1I@4M5}5orUz2WJ^5~{a*;S
z=NE{)UKZi!QP%>ou4!pd$+r%|mcbuY7X4aE@_Mb{1Q$A9>8T9T#_kktZ|Pl7dw@ss
zs8-gYc-``6WQrX$KYuG6y(OSuX<$m+Q`>Gpts?OkB-^6l>>&(ToL^Arj@R(JA@t{v
zGzgZC<lU`x#n%@l+l4c+7JM9s@ETN=#iX~B_GJJ>xbtB1o3Tp>zE6vJZ&+^&Rg{8}
z6g9jM|6W4_YWmch+(#YHd|=`@j%av*?}Zc3SX*}1xetaYX|f_16AWE`ClhY<Bu&uw
z{|Ux8iT!1YzY)PTK<%htbYQtp?}SiGIhE77#H9&U>ju6lAt}7!@j5}=mxJR>{@l00
zQO^-74SH-b<_i!m@I2`l=AJOr{4t4H0=ZKmYysk2d8et~E<TG#V>m+x+*v0bIg!qW
zP<p(5r_!0Lb*8L<K{DOGs6R~Yf=@WejhAMADmJ^CFU2o=?g@5`y3=+!i)Tmsh;l8P
z01+7a{~7gi5ywNA@MgZ5{rVU+|9b94-S;IuE<znY<h^W97`6&hDqPg07843ICgBo1
zMJfT}dGeU|PZ!H@<?3<#?ZWw#@@hIcm6LFRD*=tLqL7C!nj(gy`JK4<p0^6|4GPCA
zu{;=7TwlIN*9RZA<!T_kTQ+Nx3(6ya$LV$YK;jj$hg>}$dALytsxmRf^E5|*Zw>`6
zbvu!TK^kd4toE+6;N*gFDXmEq49B5Z?T@4~vkBj>GvM%BVsJ_5thPQpkpL0B?BhzG
ze9MowST*q${((e?XX*PkTNT_rATWQ}HNOJm=V0|usCLHU*No{tk6f`nd`0c2ER{mR
z^_>*u-50GmqC?Q1m$IJw^{y=OcFge2e`U$dq;}!+PEh?^eE<BbJIKgkhUKRWfQF}`
zXmn|Ta6#s>D;WGv!wwYUHhm7S!K&46h!(zeTK}iH?~H14>Hd|FNbg9I-bEnv9+2LP
zg3<&D(gh(DsiA}_ML<C5AkvP2^b!Oi2uKG3Y0^RwL<58(gnr}kJ?HhVb?;jLweFXj
zFUd1AyZ!dw&+I&9ru7|__s0Cr$-+o9?s4$7UC@mhCYWB$As?NEL+xC_hY}=h@ONl#
zENmyx17R2L5lEIFaU?_&q?}HMiRy%m%J!~rW#^#9ca)i8mmQRDWaN>}o0D3z<TPJ;
zy&yCCeES?dI0ZxOEbr!W(PnDJN=7o4zCYf+i1AuaO$~9V0bURT+pTFb58w7m1&l71
zzeZLWtm0R3x^ukT*o10}3tZ>5Zcg<^utcCf6}2*_&uD~oeVNPIqppl#F_P@SU()8j
z<=PhVCYUM>!MRUMJWl}ncz3+auVz9!(ZM-$AmBzhU{E?J$5a!2E}p?A6FRq=7u1Au
zB(_wdjWGuqtX~zkB)bDuY+P3aiG36X8r67E(cug=(RHv#E>TUCo-y`wxvPOHKla!X
zYUePF^hn_;bb%>|L8P4>7Pd8<Nn0byMWobD7mE!txrZ(|e%e{Wi4~)fb0q|*|BS2-
z8e;{q-<%GjN=EsPDh;8V#}@)p?9MhW^>A?nbWGnLeTr6$cqqI-Nk5mlGk5<Ut%uR-
zo5EK{C%cV)LYoOrV%|Pe4uKeU0gy5^NSXEGesp8A@jQSKHimSlJcqThd0w{bqs`7-
ze1J|F-`%27W&SlA?eA@38Y+bO-%N999Pr`+u(pwOOF;LPIo{97^AEl8aVX_$CaO9W
z5Zv;P2*+ymV*i!@j>GYjmy7;;IZdeKp34CPXf3OYAvM$9ZS{j|^-pl1coYTB<&3m)
znqyZiuN%~!L53mENgxirkIu4?Rj}0Pv{7%twArdUTyG`kXIUC7(!y`H7yX^+!p=&x
z*JoF(Wh(e7aNVT6AV^Wm*WzkjFfAuywk24Qn0(?#Vd1G<byV&5Wuil~7d>{z>6wpE
z--`=u%6xsu?WB=D{Dbz>3``19O{f<zvU859MT{@gy)Bza)ZtIY=}WrZ!<{+Z&s?D`
zvR}tKOp8(Gg{Vxjc3o}f??Zb%&(Ae%D|l+G9ZtFRw%n6MM--Arn}c&wlmj;Km%T{E
z6puojTl$H5?5e^6b1hzh!FrYI2XfCMNzR(+Go@oHx_ru(HC~^B*E-AVfRB-=WV=HA
zzyC~u19@qOCo)T{In6&jIujpG)KHNY-GROUYeMr~E2W7cx!TVZP{yQm16ZnkW=+J-
z3w_O4U1%kP>^US)gronjcUziQK-;`P_8uX-XJSdWYdER|69B7s3Hh;%({I97RHJfb
zHQ8Hx^$}EBh-AcNXz!PtJjyPSn!PULn$@igIlx5pYEw7q5ct+K3F6YH#cg{o^mTP3
z_<2X&P3%<~hLhLx5QPh)MaEtu%Uj+Eq2+_1*jN~XnQ<c$<#$4!fht-FbL7V=#-rpy
z9mV`JQH(7hk3o==^H>f{L^@LvM36dK0Z0H(Z<?Spu%cb%Y)ru<Lc?8?R;(q8Q7G*?
zD!S*RGT5|-r^}Ms=+iZ|7wRh=+TpWL>g{rm)W2`#OC^5v(ZBE_pbL0S%@l*s!Y_YY
z!47V0E(Z|o(^D}g-gSC1UR`B=*=>M!8l~B@g<k;_5CbB~?-xEqUc2sScnwi8y8}dY
z_zp<kNhBylZU1~EcDQUK2s4J5c!|df5kV)d1ceV9r}#iCSxNCIZ~l8{8snz7p-=VR
zaCc>blF=vh;E`D7yLf9KL{tB$c$ReeA`A79;*Hg(^TbtiXK3%`xk&m*T?||*ly5{b
zQy5qkuB()9Za-p6Gk4j2Q}%~v58ku2U&M`aS8w}ep!kX59Z%tY=fyLjID=fRxfQLA
z*xS%|MHzP1buyD?)zUJF&OTUI0$QA&9{~ZPwTA@b+hC5Ul%t<Zn1MNHFE)b_M0DPE
zV~t8p=*X)k&=fls>)9sNylr8=0>dY#d7Rj3qd@*yIw$2Cb5&|6W{gr0CYs2tis*GO
zfT)w`Rdb(M{g?#|K0%2)*-aZ{04dAn0OkoaIPo%d>v3}6$K{vawSeRjFEIq85WxQE
zC~Ian?2>P<LA!RTkcTld6sJYcs2*i>{Ja)0r_PVL+tSWKdM0J2_>pcLbonSA1(~pr
zWx5v}J~{`xf~YN8_R)s=y>p2Xe|yv&maJ0$m9z^f(L*X|o9D_za-p0>2SJR+bdoX&
zU5{L7g(-Pwm);Y%1G4kWRa6NG;?9G98bd5*5aB2R1S+-=RTk+NX170R6uV4J&Q`W0
z|250OI5Uopscd2^9bk(O{zt;A^R{owzm>ls>JOvNKv56eY4u7F@w(N++=P<mq$v2X
zY7~}T%$y4lyEH7ZKsa}AF#|)5X_#VHMm~7Np5Fi9`=B$fO~|VR4vDrU=Hfj8m3P;^
z=a+Rsf%2bq%{W8PL9wT|{c6t`hkkbKmD@(6Hd+Tj8yc1C@MU8BLO-wEnLwpKtV;Fn
z9{ZCuX&P~>M)F|=3~~y^(m)^MKVm&n%!-E|1y?*$Lo-826l^B#X)7h9J}*MWlC{@z
zIYBX>g*6Wz{PnJ?XFWp%MiFdJ9T07r-#^1l2=5aRQh0@GEkR8y0+Rh7Uj?l4{<PPz
z+)d0i=>;u@SPox~19J`gM6J=;;(+9&x5K77m8Yb(`2?@!1Hu@A))-lA6>#6Q`Agaj
z%J46{V_4RTGC#-ZR<i-vq;8wgm`A@|z{YI@mKEBW?MS<hgQaFWty<@rWV=-SepnB#
zO$&jb<zj0}PLo?i$BBAG)J&tPV$y*1+s$3iz`~4rh|Rsyejkhq!`p33V-7zW2qIP&
z(Yxo7aw`T5Ybqb1q(|lzYS6y}w&w`^wKfBo)<SG%D+L_nRjk!}kyRcj88Ad5Nq`S1
zJm+~p+s~9H<equFJH;;t8S*v><xR1gkaj_zoVx$^q4naL%Gkq{#AhP$)J%lz<t@M}
zExJx1gOEio0I;eHgJ&gJ>_6#tgfuX_cKgQD;U17zgc1}L2<`Kdc=l1Kpyz5$e^^%f
z88oe;l$m90-3j+uWlZUDT<nzFR{-{?jf^Cek5Fpp*W)Z6-yzAw(vF}G$go(pbAFc+
zS0`P79J!ea^_YdYrfNwHHI7-)ggJwR#mN&U>$FOPI>^%E6yw!c7%7;uzLMdhSTKBX
z_>r@wPVR(W{At2ih>@6hc)izO^th|q=#GWxxBTxOeOI4S(BKc(2aDNo&;U4Cd9cWt
z*RN3f0Unlp{rk+yL;&-$I4HorL0s?*6k!?g*MQkmsjyd##E-8UITfe}NazEn40yW4
zryTi7;d<iWg~QVQn7YgPB$BxNuKKBTpcgclREI!gViLdkE^%%HLXTo71SLhp2@6(h
ze^3i$%om^~j|Js(#f%O6Z0>Me+Dn1$uDmo12X1$RIStUKs?euXOtDT#SEKJMqkC0s
z-<U*e=)ObA;Ke**A#1tGw4~%|Q|9so2dUqLj5x)*{naX7lFi4aumq&5VYtmFHO&o)
z0&muU+N1cLGpKR54fei^sj{w<GDe`7YJI~+unSe?(F11UIn0j+Jr!KVPv|7Z8`N{|
zpgp_*OUdVONpZx|8~U#5rI##;6`zB<*k-tnAToC@a&Dy5^X#}d0dDyZFK?$TzrGIf
zz3$_o4hsujEV78BpZDw9@d_qYzAw6;f8Sz$QjEQCk_cn~HTxN)LJDCYhShJ0<SSGn
zW9SBacs_6QcICcNk4I6D)vGTgD+#<WJ*OwGwBGhx@OhE>wUIu9<L(v2a`pgd+l`EM
zPVkP6?<D<jhHHl-0O<>i@eAH!LI}inS>vn6rzMf@h+BUZ=V#r`M8ScN6-ehTAm>(D
zd-`AEd!7@dv3}9N#^kK?45c?<Fm(CkI-`YU$}3`EzV;?rz*P3lB1;==hYg!zjMcIN
zd!zcaIp;MfajBB0f;qpVjMZ$3+mNI_#cJg(x{iZNyM_@Mv!`c-05*~ue5EkO_O5Wi
zn9qLn!j`TD7}23nQzxlF>GrXJ9B^rt!-^UI!O^nfDRL@EEw`*!?FCSWdxCC)+Otwv
z0yFPmSSos(4!(IZlfro&>dZ2QDSe{X#WEvl^gsRJ<Gv%aRh|3Gpt+H1uNyM{OC~GK
z=9dI-jarw!?uZLPEmsQG?CAq|uj;fv1r$))Qu;XR#mgM@+LdI@+>%z(?P>PlMtGv{
zD5i&=Q74^r;W(o$GHXc5<8FEy;XZuvp;#@-4KWA@wnx~g%+Q?F&}Uw}C}vSv*2?cP
zs?pZ1e@JFvo_IEOUV2TE(wGygl^vdJ|Klw+J8>Lu3#wTMLH2OTBGOpV<muD-ZzA@x
z&s0g_zKJN~zOG;Oyl!Vo<h{^A!gSHn55UFf*#>8*q=r(5Zw(I*O$N&N!v)8OEB(><
zf>ykz$LPf?{OiY+EA3Cm8NI42s1#9>#@;qxO*|zbPG732kv9cd%n+CFKiMyg&3V@|
zWphFS#&+9kZN?FDaN`ps&`iFcL5={jfh9P!N4B{eHY^Lo0b^XBTBAd6V3WE5vdMwp
zm&zn*k>KLj<~UJ~>*g)Qz>g_iaRSq=MrM+fCheK=B3Q|E50Bd(<$EUDiW0Of%0}qe
z!uz!&9`r~ODqf$6&snHj*WWhVFA=~uZ4{L{tlw(@7?p70KYDjiL|-eS&7mjdcGhk3
z7n}YRMV2HcI=%Q>naq-c`a`BF+CYK*6nlJ0as^?$ON)E}K+sVRhkWqSms5d@cHj0v
zcy_@ql_t9@A4H!|zM>^CE}pn~xXGjfWjZP6;K8QB9Spy6_XQy2wLdFpX^TsWGk~!@
za&I%Xy`;Or{zTCTlwTR7l4azP=4xQ`^wq@4@cBqR!y282eP=%0xX1U|3F<VIkpPUO
z+Grr!4E^m%j5zo9>{}5Ed%uXc^?Xo3eba<SUw680x~1)NCxLt|L12{;{SOs%RDa;D
zXV;iM5Za~3kiiuTG!E7qy(ATnjKv-tuyIKXp{4pAbP&m_g;X-eJ^n+@R}kV(79>1H
ztsT0#iJuv8Ey#?j+uS^Yl%BoBXH32k>afA9<TTOn2t5fBK`bV{tjvC#21zAIwA6N0
zki!bGSzfBOiu9GIx~?_q?xP~W4Fz~2|C$Zr$J*U03ypl%jCz|&VqK6Prd#|`U5id}
z@x!n=gnNkK17TrW1{wZFVSC)IcBhENoBUYuYAQqNr=lPaY98zXE-8>6$xQrmWCdc4
z)~6+z*Nsaop7DHwWDSg@BQo&F?gK5xvkkx9GV;_Qg=h~A9r24GWXXF!d?v3)rF^Xl
zF)&?e<^RZhi+@o8|JWAxE1Z}^ZQXNlEidHw>=LL_o?-ePxUHK#nm~gz`S2dI85(~I
zd1rR_eG7^(m4ww_{NzE@Ej}V(3_2}FoNZNdE!{4b0lffXCSzTqf|t-TlfVy~_Q%9*
zYledTe6225EgggP2;`k|>Rae>UvbRI;-Ae+Z<oGictiaV;s3yc$rzbGi{HAA`(_-;
zD-@O5!#Cu}ep9p?9-K2nhr+(caf!R6p-%}PB2?jEtc{$#3{9g(zQBvHn*+EMyvJC2
zh|dhpoEsy;wRBbt#&wMz;T3uISu-v%F`8ulJUFphVK~rxq5AuzCMiTV^ut~9jaG$@
zR3DYUK&7_Q;@R)`eT`)kDHj)lXQ@>X8R<fOy!9N!7@BGC5D5gc+oPNNi65W#%fCSs
zH*hL520$ect8Qr=$$IklKGWdRGqFP(#Bpr*;qBvbmwQh}%s+jGA@1{+cl+Jr=CKdV
zF5oFsDCF?cLD<ISwO;GJj43UE9IKd!4*@1<wmZ>3H<+GH3`{*&#$TS{p?Y0(aI_>5
zv@t9};may!7HCC_*o>*p3z!tI{c#Zp07U+<Mt9}|IH(IcZyKsVi7z2MPA(@PHIHWt
z4DFqocn{ddZB=P#edGpS{6tHTtW?OEpaXZu^7+tE6T!~_&-PDZ=}**m#1pMIh)7?f
zm{&x^gp+6GvQRhe)1}?g2rWz8Oa-wEyXKmm1iQ?Y;~K>C*yHl;zWyxs1;WgEd&k%2
z57B^Ty>z0ro8}J1In1w9>m`>}<heniFN!htqv7gmV6nW%Oy)CrBSJY$8@%L)LIilz
zB85r}Pv(UMl6$>`xE%RQg|IU-lK=>!B%QN_oiRX10%3Zv5x@mZs(Ac<ZA>XVp|s?A
zY(Gta*}T4aD>u&q<i|1HoDO26w>LXITwV7ojLaI|V%G-dn194%n)*1G;xru99KdeV
zZ2y<T2|(8H{k1Flf^3E*_^lFK_;&q>Xp>>_>BNYRXCa|4^ZdV1@DJ)c`+i`MQ-1lX
zOh{sUxLR5$Z2f1p3XPlxrR5<jgCBwmTYkH{_Z-T^=%L^KWF5l-5U|hyGKF_3ZIZ!_
zzQ4Uf%_QONg^1;&lam42WQU`)Tht{}<f|$(C<l{9b<p~BB=Be1=aSnY{nN<eBvf@^
zQn{o0<2u_uP!hwJx6mm-{>%OI4J|aAY49~lxVQ&kzct$R07)7z(kGTR9J`baqoIM`
zjsX-DEE0Ke!WVeKSS*lDKR41O>91c%vCQWEXtX|uNqftaG@*P}9$eQhPv_N*q7Oi9
z91F7~&5!{$&}hSt@L8Jo%Lc2tz}hiBUID=U=1eEq30C@?g2%72Y!R{0Ux3+W1_@m;
z;|4mnppO1y5HVs^%P;K44o$Z&hgS*RZa0HmD9!tCt(%4ydn%-7p+-Iyc{-H)_tDm@
zZT&N|^DYt2ADwBVwK6hui6JwrE@K@hncEuOniE6sxnUhqX}t)$EMf@Qk-i<sAQ5T2
z21v%z<Dh&za3?%;0;Izh+cW@iZXWzS-Os1<+@M7t?Xr1RTwm%)!Wn}cBax>GJ7bHy
z0<)Gb1K^O85l}6q*tO|q(<-C)xpR&XEGtQe%o7szIY~6gRJn$v;wc?xl$+$`>g;&f
z1Y*PysGaaFZliH$%&y!(BWH&ULOPXDU3*0^Ha#M8x_D7WvKS@b>x}U6#`0eRyTa2A
zD#(lM2OdibFanGvk1-vp>+p^KH#V()kL%5W;OpW+{wwC}eYer{cW|^8edIBONED9}
zNSXJJp|nU8YBLez&G|J*K^ls$$usH`!7_H&<OGO3y}vMDotvD95tskj#GFQkAj(PD
z_46~3j`ADUK{xqbUAEq30;^@Wqg-QbPzpK|Z|HG_ICnq%GaxTph@yY9U9eWzbyI_K
z5Lck&<pa_r2gGLL&J=7PeUhg^#`gCrGUSLyrXMrlGJ+<xOGU7_2JOeH(9Gd3(1W|T
zr|h4#s5|DbClVlj6w8WrAKxL%<@?2Dzt(PEv$??4I(;w@m1<3F?Xp)wrz~r~{*Dhv
z5#JnKLjk-P`Z1Y$+u^aUbNw9y^noAp4TP(@ZcgN7W84B9eeb(^-G$oyw~yyWxvxzZ
zpI`mPWSjNT@6u{&3KO3mKYWR+U)o|39^7waWqRJ<$uhjn;trW`p*jC7cGcnNa7{>V
zO%3hGPldJi_y`=s3-z<gZ?Ka#+A$B=P*wX<(z1zjlC3nfm*mIx&Sq>>1Tz$Oh&vn~
zOPr393bBW9gtE2uSTfYN+MOv+ij_JZZLf(-wNN8lLueB=he>i!Y(A~MJS4~9=HLT|
zw+sQItZlQTkPl*u-<+8^9FIIMLG6=kT<QtKnV1xk&`2d3s-{SkZNPggT0e{X6SMB*
zTs9JOyE21caxp)3pLh(p$I56<9S$vCaeWhJWZ!C7Ih%hk&ZO>`BkYjH?w-r&w0BLr
zEkBW`)L^yPl9*Pr)QG_ku|$EuGEFd$-Dz)Rs%1SYy#pc8^D>JQC(Bqo8z;@B>>5NG
zAn=kyMZnA{CG$m;(x*I&+O;*bVW(%zSRU2Yjr{pU!u@Ek*evmd+CtiZhwj_&(t)~Y
z330OLL9akbeE0w(J(QAWx~NReCPQ>cR!8L54YMhHG<Z2+fS%g?>NuDw1-f`eo|QS9
z6v+#d7lHZxVy`Jc9S}6)SC*P)@$F)rr$<VnuiJfuY<*6(rI0ogoS=`%jo>7)=5pdP
zp=v;Rt*xL9d#+H=NG|Ed^~Isw-eu)o^_2=<jY>)?Rk|l)k2oj3r+XirNRo=WW^Z<9
zPlb$C8%())HxY!GNUj)+n#aGrDb}ui`d|V#RtW;jXOUT3lI4oX%LSP3(h|1`I=7mV
z+}27Ol*kHABPdZr-xRO75%>x;Lbf9jma`5W5+dlNwXj9}qO&1mg|<Hgl*yY*n?~@}
zv;%x37?D~-H}%+E5c?lI4bb$P+x41cxk{?ENUk?^N<Yo4&}Ud#I=L>+NMDi^RECp8
zq!OWCG)ida&`9vjFClZ8D0zdQW%z3MzT<y7U0>tddkoM~$Km=(406E@C{io5!&z^&
zj-CjN<P$*11(D{JWEUW;W@`fAg$){b_c??!%&3y(a%8i_Xm9VO(h4JfUdt>3`OOFb
zJzSZJOg&#vbkYPkUJC)ebV0Bkk|5@ov|>yb#`)v`qP)egMTro<=@qVj3_G9ofdNBB
zRuh@<iuip3Y{Y_NmTQA_bV-s>`@}7s8A@&lwg%L3VYS_m@%wu+Ky7qO)Pe4MO9nh^
ze7oIOly->1-3T5mtD482<ERysNCXy&-SMYNLe+b}lW2(eElV#FB^p?@XRg+OqQt{=
z`#Laxvt_`RWDKB)M^oLE8W*T`Qnn~e_HTdaOX@HuYuDe>rHCPsDXs9?`hU3@(@FIG
zs~o;NFQql)kM0TLT>;ns*ol{IO2WK8nBVx1?9hmm;6F#Fk+h8d$?SOwO*_Vt41@O>
zRf#HkH!AVqAJcA70FrsuxNHwAf5R1`jsE6A#ciL7dLa7psg2ol(bEMR2h02Cv-^~8
zWw#U;-7OW*X7{HkE-?N_R|ogt!v)tA-}j&Br?2@Nhk}>p06PtBOqzYLQAX)(AZD?Z
z$kXdV`&x(YAbqY7b~Os}a!dE7;0Y<_RW*u_`-f=ljx|C_5X|e4C>lKV7)rEob%g5F
zh-D^iMUvQZ?GGD0)cUmk)pDE@yBXzv_=gb%dVOYpcK1i&iX#6-SAb?E+2yw{@dnHC
zI+Wq1{p%{A+8IgZe?2Gnza1i#ci#yAOFGTtc>af5p6CRI{ev7{r#r>3Z~ehn$7HtP
zpLmlZw$8z9jQ-bW%87S>I{zoW@H%m*m7mr-i_}Ci$lXf6ilE{?{qxZ6xz)uV4$NNT
z{6j2^!u#o;Vk=Fv|D+WeEJ(Mof`agI{0We8<+lhIg#LfBt^viT@;frD(6q1q4g!40
z<1_VFAV)GFfxi<&mCPFOJC><^5&i~S_vElWE9s+rs$`XTBCHLXm9<Tf1c{mwDV-h0
zFEpjQYc26B_n|}`ppl-=oYZi#f9jkfQ%DB6NDV%dIz16SUf5FJnZFB^H;?cIN|-<h
zSP<*m%Y5Yu8BkY4_3kgtYa&Mh8~Bs2ZVf2kNja%Ga_CsMKtG<rLX`}IuKzW&o6@@f
zt`C|_zUmxUoBS4C02=AXi-wZTcb!psGUlA8DHGA}U|;psUHlAs>)_tpgsZndG@r$c
z<~LhvUSI;!93SsHkLFzNI(I0_SiIbD3qIV&PrtvT8~-3S7zma!U)|bpDkvT$P!xH_
zZ@vbVA2gs`Txo;%ho!x|#RzDYoL=4Xt~Vdwz(dXNzovr5cZb|VZy%CW*p6>tWoNcK
z@l)fspH}@h>hSVc@ijqzfs$uq({cT3n21}&t>DG8<Xfi;Bjum+tG~V?&HGgV{H;MU
zuSgge*gkp2j<=;NO*w;pA+F?=Vl69tyL;eoL;W98NM@k7tbNqRH&S1N%<x{s3B!yi
z7t`_g&Ro~!+Sr`H1}Fc(v($_5f&C4CnDYbB&jbAQ=b`^^@4w&~oaQ9cPxNv62tzEy
zE;o=H<0`+MuE?N(FE#Z5U~IOBQyBr<@S<E`QIWjBc8k>CCwjKuwJ3go*up=Pm~bfJ
zw2$kn->NC&7o5d2hycSVe-rp>B^A3M>a~CIH;VZ`1m7zVnBaFZ{y)TWAw4ncGY$Aj
zywpk>MKJfpmi+h8B!6vgc^&n%rz>myboI{8yn#kpx6w;G>`Vyh>xSzt2<dF<BG8nJ
zS|Wo*5sA1J^cjEm|EN)wCZPB8gJ?qA16L0fqkv}ETja*gilazK{i6eEwyM@JP#AR*
zw`l9pmk&7tXI&goMSXz+od5jsNUv`c_8wI+CC2<sv#+W+fteh0--kemwV#*eIGz6Z
zl9f>%MYMdw`-0>tDsmXXa9)utudoE#eeIGh2=aR|d8!&Z`*{rRKqwMgVwmIUq7#2+
zpX0ZKc|FH2QyrV&b|aqkN(2{HuU<0Oc)HPT-@iQ}HKm!z+CvXv<M=Z)#n~yx{~?=C
z_OXgZJfss&<KgEkVIEu3N5IL1ul6~5=Tz6vAQ`;krmcQkU$p*F^wqJ#VwF^IOi7=A
zwuOJGe<lb1r)wSNf<@thB*xVCkd6F51%3DrO`B*`Rb9}nKFOn_GX~Ev@oG~2=eKMk
zNaA1H4DEE8tQ}+yeZk23NeqKRF%TxekyuC9T#wTwB%_-}u%@DNkqYo(e;o_E7TBw?
zU+RL4&y6S;N*cT&5=!x1twn$dz^GOqyW27=Fv_>9Mfx|AQ0x<;{W-B~$AybqLxm~b
zV2Gwse@=87wkfA&A@)6yI@)hNsLjUXaKJ~m?Kz-z;QKizv4&GaPf{V4K2XPeqV3r2
zdMS6=;77_*4taw}E|~vm<rmTa3b=x`Gp>ix%##>D7hJgT&hZoaaBnSL!#qqZ*nrB1
zg8^sut@+ibD(3u+)Bv8}Zd#l@OL0{m5;e0ElyHtezC9D4n`_>hQ#u={Q5Ir>wp*ud
z%X2g8`>#Svl{<EaHiFx+ulw#RpQJ8XK3b$r<u2<TbSh(BaxZO|Ru;xe`LHfhN;0mu
z@Gkl`_99;V4YQ_fru~-2ap8@<p!yW&dRuo&65Zz$MP<h?kPH5xi;*_c@g2|YPhR)i
z6ISBrXYQHp5CyK^Ym8uy=hkkGCp@R|vC@gpWoH|D3#)G$Ru&f9@>Di4;E5cnJqACO
zHG%qDzUNrWXkV8N21@a2hJM8KST!pa$a;M-zY(9y(3~w=f7#JKMZd7o!p8}y-gNT2
z{%H(@U(X2t+SuQ5Ak_2Nu>WT*h|J~rW)-4zcJoWq`q!C{6q?T`R_G?$mbcx8T0Lch
zMMusT3Az6{x4^A1q5?C3)<L|h3M!>-0I`LAm`cgZq><=mL+S*YnAd%-yOU_JIYpS;
zXmj(E+_cj?B2DNstu)a8Je&V7Bg2kkQv~*Ht_~|FGR)W(qr%{On(Ya%XBo!;rwQ}{
z#PVWGnq!y*7sNT=I}f(?)oFkmT$QZN|I>fIb-=J%^LO<&nq$SS3w1m{Xk8CU8c-K?
z6muw9Y6o)V{l2Wm@_cv~)!9#P2bw$|y%xdUok~3h_xDZL-+aPm-^0S#xVVePA!=s(
z(;Z3`;ddWH^2Ub=lPlmZ^(kw*U>o;p<D>-iduQUnn@6Y9fK&Zklr3B54r5~XbQX$k
zrYH)~;d^tbVZ|i-Hkzn>Z<?sOylW?M>Hd;ln(_`~O6&5JLQ2Dd-H{bEY?fhy?A+f_
zO+jgQJp6hTggANdxt${whrFBNnPNq1&{~vkfv~&2jt!nwUGaWI$x*rhDE7JAlC`%!
z!JDQh)o=krtpl#FcC|k}ia&BFV{FJUqP2h}gMs15xqH271F!7lO{$JdqjnrJrkyfR
zkc3<v{PdZL;ju^c`l4W&@z|8nv$LDom>F++d3&mmi44*Ph8fR?zz6pb^PRQ<8PMT!
zR>A#{75_RRjCb4qRW46{dK~g$05Ubjw6@HRgsUEY%ASp}?7nWS`6T3xiW}mbUke=<
z)fe=+wwM`&82uUGeD8RGRE>d;9;X0DMnCoW;2G8St=IfhdOCaYBlEDEpRAxAaWqJd
zHzKyK38U#e=D}_CH)(`EU84E0)W;<7mi9?2ba>sa*orC0iVPzw4rt~lb-Y0g>0Zsd
z7r5c50N9WAZbTs*DY-Uq4Q?Ve9o@w98rL%MkH%^Tw3(8YgD7tuQJb6_hxxq)9cC~g
zsyPi8s-3y8rdH^bT}OG7n1)^FCA>4@XmIS_PZOUBgJWI<OVR|$guycy<g77}tX!pm
zi~X-mf3qGc@7TwYnMsQ$rnjStc9%!qS)pO#QFkpKxTmPgoYeZ@N67)EqjFs6OnxXK
z>nC_Y9iPa7i>5D_TJ5L^pEAhZ6&BKtsl6N=xD!498g$4gh<#rS-9B==VPmB-U@INi
z`Rytsi6;cH=@^xJ`h8mRe%*RHgPaD&UE?~&p}CoEuZpS_A9na;XwMzO8a)<|@s(`P
zZ%1`-@Saq0i~(ZfT6H^VkqdtG@=n3q{?)%kh=riP#Wxm2Dzfaj-+ljOQ%U2B;|XAb
z!avs}I_0|L1)hr746!}D^IoejuVtwNujSe9&nh^hD&36+)OEwBV*ocPE-f-v;s?RP
zSM#jZ(-^GKDlon{yGwY64jtv6V=Bj@@K9;dJ^cTwG4qnh2)uU2c$c2w3jU+3X{1r5
H295Y%*q$~F

diff --git a/404.html b/404.html
index 3ad921c..ca51981 100644
--- a/404.html
+++ b/404.html
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git a/89-transformaciones.md b/89-transformaciones.md
index 9a3d778..2b7f713 100644
--- a/89-transformaciones.md
+++ b/89-transformaciones.md
@@ -106,23 +106,23 @@ animals_tbl |>
 ```
 
 ```{=html}
-<div id="ufmafyhycy" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
-<style>#ufmafyhycy table {
+<div id="ociqvanemt" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
+<style>#ociqvanemt table {
   font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji';
   -webkit-font-smoothing: antialiased;
   -moz-osx-font-smoothing: grayscale;
 }
 
-#ufmafyhycy thead, #ufmafyhycy tbody, #ufmafyhycy tfoot, #ufmafyhycy tr, #ufmafyhycy td, #ufmafyhycy th {
+#ociqvanemt thead, #ociqvanemt tbody, #ociqvanemt tfoot, #ociqvanemt tr, #ociqvanemt td, #ociqvanemt th {
   border-style: none;
 }
 
-#ufmafyhycy p {
+#ociqvanemt p {
   margin: 0;
   padding: 0;
 }
 
-#ufmafyhycy .gt_table {
+#ociqvanemt .gt_table {
   display: table;
   border-collapse: collapse;
   line-height: normal;
@@ -148,12 +148,12 @@ animals_tbl |>
   border-left-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_caption {
+#ociqvanemt .gt_caption {
   padding-top: 4px;
   padding-bottom: 4px;
 }
 
-#ufmafyhycy .gt_title {
+#ociqvanemt .gt_title {
   color: #333333;
   font-size: 125%;
   font-weight: initial;
@@ -165,7 +165,7 @@ animals_tbl |>
   border-bottom-width: 0;
 }
 
-#ufmafyhycy .gt_subtitle {
+#ociqvanemt .gt_subtitle {
   color: #333333;
   font-size: 85%;
   font-weight: initial;
@@ -177,7 +177,7 @@ animals_tbl |>
   border-top-width: 0;
 }
 
-#ufmafyhycy .gt_heading {
+#ociqvanemt .gt_heading {
   background-color: #FFFFFF;
   text-align: center;
   border-bottom-color: #FFFFFF;
@@ -189,13 +189,13 @@ animals_tbl |>
   border-right-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_bottom_border {
+#ociqvanemt .gt_bottom_border {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_col_headings {
+#ociqvanemt .gt_col_headings {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -210,7 +210,7 @@ animals_tbl |>
   border-right-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_col_heading {
+#ociqvanemt .gt_col_heading {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -230,7 +230,7 @@ animals_tbl |>
   overflow-x: hidden;
 }
 
-#ufmafyhycy .gt_column_spanner_outer {
+#ociqvanemt .gt_column_spanner_outer {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -242,15 +242,15 @@ animals_tbl |>
   padding-right: 4px;
 }
 
-#ufmafyhycy .gt_column_spanner_outer:first-child {
+#ociqvanemt .gt_column_spanner_outer:first-child {
   padding-left: 0;
 }
 
-#ufmafyhycy .gt_column_spanner_outer:last-child {
+#ociqvanemt .gt_column_spanner_outer:last-child {
   padding-right: 0;
 }
 
-#ufmafyhycy .gt_column_spanner {
+#ociqvanemt .gt_column_spanner {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
@@ -262,11 +262,11 @@ animals_tbl |>
   width: 100%;
 }
 
-#ufmafyhycy .gt_spanner_row {
+#ociqvanemt .gt_spanner_row {
   border-bottom-style: hidden;
 }
 
-#ufmafyhycy .gt_group_heading {
+#ociqvanemt .gt_group_heading {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -292,7 +292,7 @@ animals_tbl |>
   text-align: left;
 }
 
-#ufmafyhycy .gt_empty_group_heading {
+#ociqvanemt .gt_empty_group_heading {
   padding: 0.5px;
   color: #333333;
   background-color: #FFFFFF;
@@ -307,15 +307,15 @@ animals_tbl |>
   vertical-align: middle;
 }
 
-#ufmafyhycy .gt_from_md > :first-child {
+#ociqvanemt .gt_from_md > :first-child {
   margin-top: 0;
 }
 
-#ufmafyhycy .gt_from_md > :last-child {
+#ociqvanemt .gt_from_md > :last-child {
   margin-bottom: 0;
 }
 
-#ufmafyhycy .gt_row {
+#ociqvanemt .gt_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -334,7 +334,7 @@ animals_tbl |>
   overflow-x: hidden;
 }
 
-#ufmafyhycy .gt_stub {
+#ociqvanemt .gt_stub {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -347,7 +347,7 @@ animals_tbl |>
   padding-right: 5px;
 }
 
-#ufmafyhycy .gt_stub_row_group {
+#ociqvanemt .gt_stub_row_group {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -361,15 +361,15 @@ animals_tbl |>
   vertical-align: top;
 }
 
-#ufmafyhycy .gt_row_group_first td {
+#ociqvanemt .gt_row_group_first td {
   border-top-width: 2px;
 }
 
-#ufmafyhycy .gt_row_group_first th {
+#ociqvanemt .gt_row_group_first th {
   border-top-width: 2px;
 }
 
-#ufmafyhycy .gt_summary_row {
+#ociqvanemt .gt_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -379,16 +379,16 @@ animals_tbl |>
   padding-right: 5px;
 }
 
-#ufmafyhycy .gt_first_summary_row {
+#ociqvanemt .gt_first_summary_row {
   border-top-style: solid;
   border-top-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_first_summary_row.thick {
+#ociqvanemt .gt_first_summary_row.thick {
   border-top-width: 2px;
 }
 
-#ufmafyhycy .gt_last_summary_row {
+#ociqvanemt .gt_last_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -398,7 +398,7 @@ animals_tbl |>
   border-bottom-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_grand_summary_row {
+#ociqvanemt .gt_grand_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -408,7 +408,7 @@ animals_tbl |>
   padding-right: 5px;
 }
 
-#ufmafyhycy .gt_first_grand_summary_row {
+#ociqvanemt .gt_first_grand_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -418,7 +418,7 @@ animals_tbl |>
   border-top-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_last_grand_summary_row_top {
+#ociqvanemt .gt_last_grand_summary_row_top {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -428,11 +428,11 @@ animals_tbl |>
   border-bottom-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_striped {
+#ociqvanemt .gt_striped {
   background-color: rgba(128, 128, 128, 0.05);
 }
 
-#ufmafyhycy .gt_table_body {
+#ociqvanemt .gt_table_body {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -441,7 +441,7 @@ animals_tbl |>
   border-bottom-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_footnotes {
+#ociqvanemt .gt_footnotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -455,7 +455,7 @@ animals_tbl |>
   border-right-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_footnote {
+#ociqvanemt .gt_footnote {
   margin: 0px;
   font-size: 90%;
   padding-top: 4px;
@@ -464,7 +464,7 @@ animals_tbl |>
   padding-right: 5px;
 }
 
-#ufmafyhycy .gt_sourcenotes {
+#ociqvanemt .gt_sourcenotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -478,7 +478,7 @@ animals_tbl |>
   border-right-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_sourcenote {
+#ociqvanemt .gt_sourcenote {
   font-size: 90%;
   padding-top: 4px;
   padding-bottom: 4px;
@@ -486,72 +486,72 @@ animals_tbl |>
   padding-right: 5px;
 }
 
-#ufmafyhycy .gt_left {
+#ociqvanemt .gt_left {
   text-align: left;
 }
 
-#ufmafyhycy .gt_center {
+#ociqvanemt .gt_center {
   text-align: center;
 }
 
-#ufmafyhycy .gt_right {
+#ociqvanemt .gt_right {
   text-align: right;
   font-variant-numeric: tabular-nums;
 }
 
-#ufmafyhycy .gt_font_normal {
+#ociqvanemt .gt_font_normal {
   font-weight: normal;
 }
 
-#ufmafyhycy .gt_font_bold {
+#ociqvanemt .gt_font_bold {
   font-weight: bold;
 }
 
-#ufmafyhycy .gt_font_italic {
+#ociqvanemt .gt_font_italic {
   font-style: italic;
 }
 
-#ufmafyhycy .gt_super {
+#ociqvanemt .gt_super {
   font-size: 65%;
 }
 
-#ufmafyhycy .gt_footnote_marks {
+#ociqvanemt .gt_footnote_marks {
   font-size: 75%;
   vertical-align: 0.4em;
   position: initial;
 }
 
-#ufmafyhycy .gt_asterisk {
+#ociqvanemt .gt_asterisk {
   font-size: 100%;
   vertical-align: 0;
 }
 
-#ufmafyhycy .gt_indent_1 {
+#ociqvanemt .gt_indent_1 {
   text-indent: 5px;
 }
 
-#ufmafyhycy .gt_indent_2 {
+#ociqvanemt .gt_indent_2 {
   text-indent: 10px;
 }
 
-#ufmafyhycy .gt_indent_3 {
+#ociqvanemt .gt_indent_3 {
   text-indent: 15px;
 }
 
-#ufmafyhycy .gt_indent_4 {
+#ociqvanemt .gt_indent_4 {
   text-indent: 20px;
 }
 
-#ufmafyhycy .gt_indent_5 {
+#ociqvanemt .gt_indent_5 {
   text-indent: 25px;
 }
 
-#ufmafyhycy .katex-display {
+#ociqvanemt .katex-display {
   display: inline-flex !important;
   margin-bottom: 0.75em !important;
 }
 
-#ufmafyhycy div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
+#ociqvanemt div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
   height: 0px !important;
 }
 </style>
diff --git a/89-transformaciones_files/figure-html/unnamed-chunk-3-1.png b/89-transformaciones_files/figure-html/unnamed-chunk-3-1.png
index 15220eef7ff937f0a7c1a10125aeee8697ed159b..66b87155d7a5787cdf4dd4c74034c097b1daa11d 100644
GIT binary patch
literal 36380
zcmeFZcT^Kv6EB(&0Tlropi~d2s0gSu0U=R9Y!pR6sYe0nAT<z3IX0Srh=5X~NRy^?
z2#{l;3q-1v5b4!~k`NM7Za}e|^WF9Cdhgz~-h1o&!?h&Yd-mQlvuDrzW+q|AhWh+l
zMYjR~0RHpmbguva+z<c&^!H{E=acspGP9gN0LGWD>T&)A0)bpyTp$o=)22<_+}u1o
zJiNTTn>TOXvSrKGty}r{__l4^#?Q|$ARr(pD7bz5_8mKR?A*C?*REZ=ckdPw5)u{`
z77-EIvuDrVy?aGPMa9I##KpxWBqSsyC8eaK_U+sE#~*)4OH0eh$n4+0|G<F*2M->U
zm6bho=+NQAhmRaNA}1$z^ypD}d3gl|1w}>0W5<piKYskgi4#gnO3KR0Dk>_fs;X*g
zY9~*gR99Em(9qD-)I4?Sl$Ms3wzf7H42D1;r%#_gbLNbWj*hObuAZLW*|TT$_4Uu4
zJ9qy4`3n~=T)cSk(xpoV1_p+PhDJt4#>U2%FJHcL<%)@k$<?b@O-)VB%*@Qq&97a%
zcK!Nw3kwTNOG_&&D{E_O8yg#2TiZYX{PV_*8#iy>w6n9bx3_n2aJY5rmZPJilarIP
zv$Knfi>s@vo12@vySs;nho`6K?c2A#yu7@<y?uOqe0_cW{QUg={h?53KtMoXU?2<z
z3knJf4i1LH;Rpl*iA17MsK5UD>&~4!ckkZ4ckkZ)`}ZF_cn}g25*ivB78VvB9v%@9
z@$lip$jHd3sHo`Z=tqwp#l*xse*8E#HulMrCvkCc@$vBq2?<Z1K7ID=Sz=;hQc_ZK
za&k&aN@{9qT3Q+!jeh?8`HL4X($mvlzI>UHk&&61nU$55ot>SNlk@7;tK8h&yu7^p
z{QQD~g2KYW*RNl{dGn^IsHnKOxTK_{w6wIWtgO7eyrQC_va+(OstSX_R99Em)YR10
z)?%^PzyJRG?c2BS-o1PO{{4p!AL{Dr>g($p8X6iK8=IP%nwy(jT3SAS{Mg#s+Sb<A
z-rkPG;XZx()X~w=+1c6E)z#hI-P6<4+uQs3^XI<4zW)CHfq{X+!9hG8KQuH%AP~NM
z`7%5_JTfveIyyQwHa0#!J~1&dIXO8sH8njwJu@>iJ3Bi!H~01HS0a%}B9Z3j=NA?h
z78e(nmX^q5GKE4}US6hBsVgfhG#ZUgr?0NAG8hadlgVPS*lhO5Yil^?4s5x7?z#^E
zAQ7_m4;VeZ-~a&Z1Dw}AeKjC?4i81`9%rsBv|o<$MxUFzqk{UpOQmeZz{JFNhu-d^
zp)Z{BKG>w!KRB?wM{H+MgU2xhGbj6av)YHyohsokuITT|IH&oPaBArzv5mOI*=*~e
z)i9Iw9^dyYDar9^P^F)lm9(_Wm5vTEh;Ek|$XNQtK_0ZOuE91)SZLxN)Xp6$#{mbV
zU2Xy{=yrVoJPFMX10LFOdDs8(f!AgM<Wn(Np{%l4Klzs6!`}MS@~+K+F|PTlV4De2
zesJ#P^v3SPZXgG7Ui*h}S7~v_%MZQqT{d50NdofCTBHh}1d2a8Zq`wRvU#)AmUhq*
zUQI9=?*%%QTCf5#K8~UuzIhzaO}-XGjm_5{BIPX3_6wlpCO^{3$nkDb=1Www+V0Je
zfm>@tu8f=4555QmD0_HCnHRf7_38qh1|EWn`pkExH!}SmH_BdTq#nT*l!B_<J~fAf
z<K8axpHyqR5~sFvokEP*jq6Q}Is*<3TFmOK|Mi;1BXY#dexy;HZEra6>mTh$Prl0P
z^7I;$b@9G2d^j!(S!$RNsgE??Lyc%)IHgE}sDU@{yQX31n^S_QpC>*uwB`$u)O7wT
z&o^w>aj^y3mLxUuXU082@>SVqQ`%MTS(f2gswZzjZz<s!-;l^NlflHI$fJbBk>|ob
zGTEjdfY;5MI^JG5gI!Fqo_*OMFltRrDyBm%sSWK_9&_?V9~B9<8Jf~g;fyX7sE(JK
z$}m;;>*IHYa=E4;I?jKjcf8PJfnZ0=q1SutB3C5vC!lRr`zJ8$sdUe5%tzZ*SJzDu
zX;-TFjmSM}(Au^%GV2eI#mw-ASyKr{pC!9;E<nDoaWVg}c;RuiI~Mxlk22yE+yO9W
zL9NwfBh?;r4_`aanB#=cwho+KTu<n<_EFmnHTjB_^-dT=S_kHf;tU+Vv?7E2ZDR}u
zN3ac-rCr6Ly$+c3<uK%+ogUqmvU2YYQ?KAnYue=3kh>o6DRQ@-0`;I{p*+8=*8JGG
zSm-H<p?k7=Gy2i&&M8<-L$|MNC3=*epmj=)9hA0x+Gi*|!Q+5+OH}XUHO?^>L^+&`
z<8bauUk49b!eK=Tsf8`>%9qvG0#Coxn6BKnEWCHAg=$WVYNY<H+&jo$hrhZ+$|<@?
zE>v>qMm0xI=E|)`1Y5081Iq7iTPcTDj_8uu{4RrHO%tpu1h=Di-Ik1=7#d~M(zZSF
zNUI{Aj1lP?YoK`RI`N6BC8#jjiZwl!Wyb9HGy78z%cWJP*+X%bO@Av^=Wqv|>szoW
ziW#?z1>20eZz_#Hx$oi;%YMLBh06&dL6+WktGtsNE#H)92P0^#6aKpA$1%?<l3bWt
z1`ILxZj^s;d)hKF8Em^kJ-H7%)hy5OQk))`<YC&$ILNgyjAq8KYO(cVe7Zee^&0V{
zsV$D+gc;|-bMy3?My&#R$kcoRI3G!RPh+bfubKN-$Gh3ndNzeve2hf*?BuntSj^yu
ztggOhUbNRl%E#(TM%Czzb6Az|M&`m1!Wy4uNzNk#r(7OisFs$2{mo&5opqngwK@O8
z@C7$dBsi{}eYbJ)EkARtp&%%y(LG-Nqy>!I)4~#`6Xn+p)}0@qkR%Sk?<Q<oKBQ@L
zQMD-8^^T7G#93q4FwhAr8b@CGt4k9XAvoU{UOUgrG{z#+VJ`G>^O|aR(}NOM)cRim
zB{H%If)a&G5xmp9b-8lb07}^(FP3AEBOkgwvL)3PWm1u^cw!_J0^QH|2{Uy4+iaPs
zbJMSoq3<a#SvBCtHFx@C6C?@`h+QvJq5Z!4<ha@;!sKenHe93pktEMKMLYc`S8K(>
zF*!Yi&JWm6X8co?Fpc19!yT1M*WA2b^9^a;81SHv7v&#aqQ)3YR43Vf{)qg;BN~j@
zHFT%;#8IRlxUb8Uz5Ok#8w&-gpLhy4T~`1HjskSu^jKGb_i>8WdvQq&xpCB4kIQ|*
zcF4V&FJ3O<jRy@%S*ZtEk!5Eg<*ibMZwod{6_Uapwk!7HZNG#^NmB^msq%KIoVBwz
z%p+pPhgvU6id7a=)w0-p?x$_{reZa&_CUMA9}d8-u!n|dY<OjF{C_%mDVJ!ap2Je)
zzqqXdsIeLg8|kv>FtWZD-CO5(pZ)&h61!pH1(ltU#a!)cc*V5;2sZYA!+0m?$<Am@
zv=#PTIq&sGVoblltC|HtNPg5lFcTm@kjPj!X{8^4dkcc6by4x2|M92-5Ry>k>thSg
zrm{5h8gem+;lM}Ol~L_WRjr>o^=Yd)!7N&lAxq|3WTad*a?l9dVWJj(>5#|UPaSeA
zNBhfjY3*Sg;f6Fmc<X}L>sKA`X>dEcZg8`UycgtwoUO(V^tk0!ZEWzrcq{;!^nX-)
z->6XXpm%;%iB#cxy`e5O(hi28%SEk9Z_PF3M!vXh`))cXbyFW?`BN<Wi1wBdKS!1n
zp~jG%S~S`ngqlwVtJm0dEm+j5y#3d4|C2Ur0Q{nN_g%-HpPe<)>!GBTm2yN3ee&1L
z*9xN0PKMyCT;xkdZ5w;E1FKmggWBQ(b2gmO_Qogsrpy#lj>g~|`Qim1l`@#Ldgx0m
zjO*oG0FyLR6i26CC-a4A*tC=%M2-{&h@H;QG+jkmmEFFv(mlDDu?uKX`yX$%9FUwV
zuQ!80(D_BgJ1k%3B072y)m7=bwLXjR-meF&d0uf#iu5i1g{<{uW_Ik8IM}8(fkEcS
zw#l$oK9(k}dPe)D2~T^E;E}cOF%4M974nfGSXJt60=xmbg{6Phzq*PV<`%pVYwj+m
z8OF#O`E<Efg(isjuqLAXe%`TeX_@yy>24ei5nat<=w&W4O>1Q?$PIk6($4(^Cplmj
zbMYme!S`nnONVfEIKW!Gnt+~t|DM5pmfQ2x$<Z4m?A=xxk?pblsMiZ@nbPbXLh(}X
z0n6L&vTz?y=3fJgqun)_T=43>S-tqHh_E+qu3INPLMwN4Sy*HUHhM!^;tiLN1k+fD
z(>&)2izF_aO?zyv*=oB)a_%5FxdOZVz}j$}FkX$(SIB;heg1A?rYf4@@|DhiwtALl
z6cHR<Ypp6tiNcILwZR(}eT1@o>U(l23|1rbF2zuNET|^Pt%guME;i5dVv33KB+uVZ
zJrQM2CEbOoF}~PPPfgF5X=^l>Wd<z2e!4EY6_heS-dbZXdVICvDQ8a|5|?e7JT?O@
zR!Ppy1gBYGNxp9h;zDA!U)NFIvBIbe$dpz~KBhI7Q+4WRm)vKpxO#qjT|Q5~%RW+%
zFg8zE8cFHGY|JKH0Q{(T26Eg;+Nf;4Y`e-5kaj*nb<ui@JPW!7eSE-jMiGu2vTdK<
zr<s>st9)6hH<@=V1#fsMD6ZA?BAY}1W9^V?fI$J996z-C#e~*AF!LD2RMlLb$(irv
zTD{~b)?HXU*_wPe%0f^45kkIN?0>&^Y5_t=m5vvzX21>k;iD|s1r640)y18KAFfv^
z(=>!J#NIIT#kf_2MR_m$c$0aN(|vfid@8fT@=Fz;t0-ZyPDQ@<gCE<nWzn{3$<>0r
z73yHn(z{Z=6gR7o>&#q6vcg-Tcs&*7-$nf_7dej5)E2J+aj$^cdr|e9Au=FbNs-`Y
zNF+cx32>DgtpPCDgGvNCJp?G902s)vq5xf+A;nuDrhr!<+`a!!`d?v0DG*{Su3>2#
zcua(RKhYw1HT$`v*5_=;xJ$c1#zLqWfbtE1OAik~OHKp;v)*Go`O=Drvm(yXJZQRy
z0H(m~0tt}DXq!nCcC0k}0HDjM8{qq?!b%cX>@B6RY&mU^9IM&l<tUke>5)&{%#A(*
z=q|N|c+SbcvNg=D-i#8>--@ceO$SzTp=Ep`_>@?2Vg*Exw!NpEZ2ELht}lS+g8jA0
zosOt;;E*>3ke77?)a{J{Qyr&6gd88XFo&?GG^tlNM%QrBEwWCbPvni90l1VL0ld=K
z4<Hv0PS=UGH`R>UYQK;yOSp7z{cQbmXnhgJlQt<c3yxBmsXz$L*7T_@I!O3<gUd~O
zk5|aii<O<t2j2m+w?h2B3gPl2P|ct?VJ_ozQ+`~QcjX`=WYYU?8u{RZ32lmJa<XQD
z+ZoExK0H#?J^wadsAq4SMP*T0ovSr6KLX=n&{CKk$SjR2&#bd_qbf$6`q;04o(%!i
zyy;lm>MJ13(ql!e1%Dqpw#+P_6=-Xvcm^w~y>dIF;t+)Ib#@1p-i_C9bH|_rwgF+#
zjQeZs0Qu4hlp}@B)-zQhLPX!F!%4=#L(fXhAt55nyr8V_loLr3e&D6CK@97|#mWfI
zi5A4O!zVdsqNw8>!%@V`UVi{oEKGqZNdul4L8m<)gOC3KjkVO+A+V-Hu7xNK{W0bo
z(`;P?LLlEyChd9TEJ904-Vi<TF4*fdv9BoXHsBSPu{DO<w=GNp4(7e|s(Kqoyf;w=
zr<AEk7TKUO$r|e7ac}a;YAZ`eINsp95aiKOO0+yULiHviIdn~fgg{;JebA|a0?f+L
zY2IAyy-sg))vAkaZPuiCqkCLyTG@Xb-3?loLcOL><z0RLW{Ek%juby9BDynVEb5u(
zrElZDY3~DoAHEzBG><d!<R3~RcSp)l#Z)|E9A4}}y#ie2MQ;Ha7pnrC?)=xL{MWet
z_m%no#><%vn0ivYJg5S#BOi$XIoSg&j(>mu4?95GzaD)xMAcUmK*s=`>gGUP#xj7`
zOkbdu<sA*MaOFwVSst`d@pomv!tLt~mo)CP_wO13i0(z1ebn9uP~pP4T@b=4WqcmJ
z{15V`5aeDi^1JLhAjj*-e+q{HTsEQmo*Y4A?gN+x|E2-gLD@)_{&P)o_;z&c3t?Vd
zBj8E4J7=~OJnPFOG3MK1Z@7ER0vd8TZkD{i$N`%MX%f+FB<|?7p{KzTEmqWl4|3P1
z6bx$*6kw!j7vc*_3ajcaQsHPSbSuMZlJuPjhamx*JwNEIxLu&do-?8>F&Rxyrv9<*
zWyj%3?ZCHBTWs^oe&6Fsu~^Ql2S_{yPyylg068p9G+PD&vv;6w?fPLil>m`D*0|~H
zAhjJO3UpFke=7guBol5(L+HJsh>{fLt9zyXXOoNp@_Z1>qyNE|<8rzV=)c*=y9s?3
zkZ2GFWRCez2zyX%iN_0XORV{Z;Tzwn05lB7{I!@*QwR1zv1LBI)3vSLZHt9%dvISh
zH6vTtA7W*omYC-R$<fIblc76G0;IzfrXcKMI}b-3B{#pH(&Pk3k#daKf#iokdh;}k
z(WkRCX*%Q{KdKloQ1+HuXEZMY_{Rjz=2|L?_7N7tKZfE>)yj$O!sbrc9x58H93`G2
zVWb^TvhIjA-7>!veEU(O28&wvo9gDAKxf%lXB!=WP_)_z<Jh{Dt1-OY;}43O^k)`3
z4ID)D6)QS@{l+0N=3NJVQ{p2aiz|2SxP9%|tg!)Urf{}gVT9GR^~b?KnGuvbG7j(t
z+t+tVnM)5F1N^r9b}Q^r0#w<_F!8V0t?aPI&yOi{vAn*#T1-_`{T7Jat?yzz5F|jh
z*OK=R6P$Xu9leyP9fbe038Zm}up;o=xQtCx4jHIgM3<R0#kv=OizVLyh|2xECS6BS
zHlHuQT<Td(rQ>!4Lhf<{mCb(M-JJ8Ns0bdiS4=C3b_RcY8;TBeve*!7DG-*Fs;w!C
zyK#k`n(DRtpI`n3R?k8zqrvf{nSn{T6od&-&fmEEqJWfveU>X}A5R&)0BW6BfpmOK
z*X!{T2XN!MHbM#)b+JwxY0V+Zq6GGlz$cXUz7BT0d1^)QF6a5blLgR#M2ij07`@;s
zi8hHa(L5*Ux6o(tc3`Hk%O|MuI;?gwGgm<*bTC^Xx-+6Xh4F++&^p71P*@DDA|u`~
zsb;Tw>MZ-{u0zQa*0m!OiqWUc03|LQ54#p<?c|=rJ2f^kM$LGOWiZCvt5=`!_aZYU
zRU30)wT<H~-Q!j<t_(3r=5>MsTo4W)em4s9(>hi1B}AH<bCHB%_pp+_%6lht6|KCB
zi{mkSUPpY<uiD_A=ge#|>nW*#^1_`Bz_^O)%({e1;?3EIdNB=KQD*#*)_yg#3NnA8
zE8&)f$M{2X%))l<4bB~`6|wn$Nu=c}h(h=?PlNj>tD!Fae<0TEuIV>N9*B<4A5Ghm
z$GVfa9J-YgAaZ!R=8O*TlBelko<w+i7+^jCjV}T$RMta^R*(i1GGvsKX4R?v+sE|0
z-I_I-a4vNG28kqIa>sUbP)~^wpljzEx7HluB~<EOF5ia4<t-=+uC@4-V=3JwadJ8+
zqV+#1x@^W>bm^A`xNL#s@~p={fshm_Va6Qgw882oh&+%JGI4_0sxrZ;F%=M4Gpf)z
z;p8u);gA&eKsU9R^j_ub4xu$8`)#~GAbYer;8U2RmNF(&VuY7t7Qdg)8qnXh77Bk`
z4(Rj_(6tR>w`uL#|F}$R?2$kn+y-d{D4$(>_A=wXzhs-%_W0jP=)V6l;{Sgd5x?BT
znw*x4@RU*^=kbc=8tmS11K021Lx45xj3H<y`ALgJVoQn^7$*JXS0*@YnsToM;C#Ft
z)rlWRy?f6Z-P?FQ*5|tcn5$?V6*Rk<-<*1Ea+{>GqKkC$Zxpzu)MU<{$^uFxtkxh~
z`YN+t9DKL&!mZn^gD9KCBt{hvIeG9ROFlYVe<n`$_YuM<!;knm9#qsv1Bxf?Fw~+;
z_vGp?;8zO(_e>|+`Z2Gs6VZfS)>JwA;^0-q!7tRg@*cB_0{B^Zc`uQy*8=6oY4OUL
zZv>(jIsRnwJB?aDybuSqbIahMc9Lr@UM~ZsntEadq$W7^WzdRUPxVgcP!jPVL19~y
zAWv-8_qzmw84(fXfT<hzSD)E&z6V^$Ke29Q`+!U*f=PEs1Q*1pt!9=$ze0SGy`?)y
zafRtqtKyl$H|OIrWeL48;&*g0>fr~uXV~DtMZcp<QO`O1`GQCazE~YH&)Jau#)hQT
zL(Xs10$lB#Lm4&;t(LFB70B6pg9aWS*4PlRqbXD*rf3}nboGbGMlXxSh@F6uvxpUg
z{H%PDu~a{qe{kGPcGnikcSfy|^HH0qNz9O=4-+>#{_4-}1<j<!V~Rdmd;1gVQ(?xx
zy1@x5=wg9Z319l7!Q)HmvS)siWe=*#7Jt$kdwEpYH0jWWGqwI)B*$oTkTo9kaW0Mv
z`NL5;xW+gE$SxlAEr6&DAaWa|vswFFII<3WtOFYV>yt!bUQ=^l`bP%z`L$->*1PPg
z`-yRIkiYh3Zj^@XxOd2ew_|&B>Nj|9-Sp>mgQ}dY6a=T#D~rXU6JGS(ZF1kSmv5L#
zGY~#Z26+cU*lU2@jdj3|qlQ1XpD&IZ5L6#~bQ*{r^~O>a+uO8+RKDR!>)`BlIo_*&
z=K!1q5*RJ|T{4vBUBka@?hlCmL?%5!aojK>aWn6aCn=8~s8KTS@VY_IAp6!3dbOFQ
zvYfhI1x{jykE_TJO-VSaFj1J2%sTTb4&J8dCDL0|0li(2{mlhh=}3ek?F=Fm3oBM-
z1eE~!qNv2uKOnr56Z8bq_*o9k%qdde5No!i?63aQEi#i_oVT`%vnK4?Fk<3q_7)-3
z<Gt&TX8(4y?Bozd(YD!ox5{%miLa_K&(@Cgm3e@(7x-vVqOk2mRfG#2NFCt(ceo(W
zndXVVUwX6_r+LuDg8jH(HLK2_eW&sLwzXIKmiuNLpLZ;yWlLoe#~dt3KMog(@vRPh
z!il|Hbx=_)_%F6LkAhTPq{V+kUd5`Lrn&|<M<3kK*x#7+yT1N>?E82AuB|`AHP?&3
z>FT<t@=X=^0g3y5%JqMTTDtw;(dGY6`hQ{x{>#1pa_@h`gYqvmGCK?bBM>*=>=yX+
z_<OKD5xQv&qrU<mQw;W79pQC8^xZj8mj<o{I%6n9k~APu8KB~#fo=`f<ywoLnt&3A
zi`Z#dHMWOR(n~f_HSw5r7sDIE+x*2>;N#Zsk8A6WEo(2r0Ulq6no~&~G0yGZ-^i$Q
zYlceUIGRs)=+bm!$;K<LzkkhcChWmwXpO}`iMkZ@{RQ{=zIYl);s8kGrH_68hC3*;
zb1l%jC@{tNIwsrMp{|_&kF|Np&q7q=qA<}K)M~VERL2Htv_TRmG#p6|<0L6ELH!@U
zuP=_1bja{|B5djAu9va5w4}@W@jH#d`Z3F*n|h0W%2mV;j^i_;aB&UTJHY|=EPHoM
zhB{At+lwpH(ayrd-GMW)Yf3fqwE)y%QU3a~cN*2~to>4Li&uMM>J_fvVDfrQr+?z+
z1a~tQVDvI@Vf#>M$o&6RFk`(FLaG>z@?^5(pG>@Ywc*r$=K@!D=*MCAO8}Ryo2Xmf
z-+1;5#+S>5wnfm!FOXgzmTR>$-GBnW9l9an94Eh`WTP9-0Se2)xipaTO$OS><m|sY
zSl?N50*#C({)f-G(2>2{`j0sM$$MQU?ymBNH~vF0%o#;v2iomA)h|T+Q$W6V8tY}K
z!s?5k)5~Hdqw_9J9+Ka1!F~v&CWEb3fWe+3RmnvsD0Qlz)A<(}l+`o`xS{1!33W;S
z$%*yHMcaN?H+3}sVg;3O{=_8%D@FaR+oAFQIGi}Hl=+_G(`RGA0Qc>`uT!%B$Bp<Z
zi?f~R2~c3imBM3Aw-VHhXD7N=galO(Np~la?zUBbvd`QJ55yn?4ZyFZGw^Kpu{S$;
zaqO7f8g{{y1-1(lU|E&v_3%M4J8&JF_0d(xz0C!O>KQl=+Fl<WDMb(<`4PPwlBI|w
zoqWQTb%H8A)vK=f<gKIDqJeoIfW9SUDiMqsXjU3Rl4AKyPU>DP`ay4TyG@-=fc0X?
zyke*u;_j@ZqGh|^q}$=3OpYr5<Gu{Xo_nuHuOMP1G6v$U^-?d5@8kbn9}>9GE%hE<
zy_h|-$T_>Kq9IP_f4_{=$(QWI=<RID<=FJJb1>h;-v#2Oy_TT3E(a*tyN><r1Zhy?
zEpKJ`{_E~;bFz~SRL+p;Ooe~L=)Wrs=O6=(pZ|k7;U!=0`;z4QFY|D?I@H&$A+08%
z`Cpbs9lcAIRb~=`7X*_Y|3@<rSJd$zbhE_d`m}h)Mrz`_#SD;vbD)~Mj5Lt@&`(!*
zy}qHybQ^A5N{ih;_#q@y<%jmH+taH6_^k%RnI?Mbs#-&m{=cj+7JRgNsBbv*K#q9&
z@ISxxwT+2-Ux0*-7=#3}dkKnhcFG&^>W@3-DbzQt1E&x)!!Mmf+>8I$9g=FUQd;VD
zERYM5)zWVHheilYiJr(j=@p(AJ2$2L&+nW>Dzi>{VSUg0|5MBdczu;B^G+U*C*AkF
z!ZAF)kR;l-cjYli<E+2OM_b!BN#cp*m81W#{p#rA-p8YNM*Zr-I-I>1WlEHP=hl9d
z>dZlE3~A_8AI*DFo~fsBVab^CkBxCT(E>$G2S~~PwuNHGhS`HJ0bPgWA?nY)^8m_4
zw}9k>uFjIVM}AvKqm6ehb>ceR{!Y#Frvv_9wC%sJ<68dezZ&s9k@tUs5e~p9*<Q&?
z_9qde&RzdAtZy3t?sr>3&E=d24({?!{cwV&a)M?Mf~>pYu6~1%5D={7W6Yvl`NG2K
zyR2!0lppm>k;$7kp0@Qcx>ayWrc#L!GCt~^P({Rgzwg%isD78Xl>=aG_@Nu)#w#Zo
zEy(dG<8anc+DT6J(mxP9R|G^+dqjf0m_E3fa*}xfIkJOtu4eHcFdxzV5_Md`8y{1d
zvsTyB$-&;(w0ERj_BB^>l<aD<sKpmVN`S-=c97#&(%Oq0ntF$sVXa2hkJ(jxk3GX?
zRS()WmCMcLdMvD?mmdM}Aa*j;Xbm<*x_A>SIAGoVfz_1Mp@Z8qwy%58JJ1)2{c2P+
zRQ;q{ZyG{T?TE`6Zgkg?^{2Q9NG_G}){%WB$0<c}j0#IM)(Xxf;4kLolZ_1rPe7eH
zKp#c2@F3x_klf?K^iP^+cNN+@Z=IcI^_U;DJ)iC;aaiKf`E_id2tZDHQ#mL^XN@;i
zLYXZIUAtba!=N7`&Yk4|BkT9ang`V;@l2Ct`<C(xdh*>}_%>n*R5kT^kkYcaYt&n9
zC4qzfMgEY632eRQE!@YOXDw<dpaJ107*B&aB}y>S+23J6lPZO6PI<9PfVEJCBj<%m
zWWnaeOv4uIz9|KDO=yjn`i}4MWu~tg@qHaa4*yiv5wI2qB}(Ax;Y;j@_Q>UfjuMHQ
zQ{+D??HKHQ^VALXF%f-cX;mRf8r!xW?G%GFG|joPS2yaEQm+TT_#O}$I8CG9hwoZ2
zYy+$XKwf9*Yrc>e7rEGKDpqvk<L|HykZwlo&^OvcpOpif?fN9^cum=WI*xRQ&sGsE
z9Iw9ea^@gmet25xQFbXNJChQG>M3EIp(a@fv+LJUsu3X-d7{>BYpYJdqlNC&AU50L
z&y>rohjz|unWi;lN;|G)d|C5Uf*9ka<=3%oCS=Ema<;}-ceLu@Vq@jTsG)c<R=qcL
zSfUcnw&o0T7sP4q&+Zj&s1aUrOx;1tzOMEvh5K6>)}|>kcGlD;qgL?``q;KkO4Hsp
zMl>aTbqH8lo!L$=%=@}~tK49};o$48fn>}5UAVqaNxPD*M!aI2y?Sg%1NX;QdAMf<
zy-wTu$4wuXL8eM7{KZJvL19b5+~AG<Oa$rU+aE%`ACxUgclb%aZ-7~!xYZjaJ;2a(
zp)n%PQ6nxz6(ekPI~Js`tT|--TpDWSgl3LFolH<el9vqo&FpV?Ev{wF`qyd!x8d}R
zw2C{hi#Eiu3eKQdl!lb2!s4|LzYd8DnP9bJy-kP{YeSMy8ttKTCSKFml|+HcM*yYV
z=(D+d&_qzB4#6K^M-|Yx#ys#rLtRIVyQ|Ma136;*4Cc!w;?*=qOyNFv2en4$Q?Gz;
zf3S2p!O6%=oeQz-ifdQse*>ld6+*vgw4G-6x-2DwR4uB?`=gp7$#^uSa`0_B&$z%c
zDYTwL+>)ETEk?F`9A3k@ArkErPmf0fW!-ANsb6*jm^Yz0nA5F@jOf(s&~UZmSx}bE
z+<@O=lEtqiyDYI<0aH2UJf6gdJ{*Q_ox)3t1c^|?d_@th{4z6o9%gE4jc1Gcgm$G#
z94Ts>JB3z49`ue&TzxAl9TZ(dspzibT{<?sT!}5>(`8w-OaPoBHmqu?Da3Z?-cfv*
z_`Vv3=8IjuPxu`*wi@?2SHX5nAB!>Pf;f`oy>r+)=^5`&`q85Q>Y(vA1YR3cU@yDd
zYo>xsH?o*d)X}keXEoXkq(!EubIB<md8WTqP6IX0+gJ1W5)smqt1Xvu64EIh?cF)9
z#l4R}-CapT?Q6KiHy}R!BajJ>kr8*iM}GP|%W*mWc@&o~uS)5=lo(!e>#($X%|{_=
zlRHv`3uVhJ%}Zu!<JQTCC2H5M|Mv{5Jb60;+NX+q-&T~K4v@$^XjYLn3B;S|!Y)xn
z)X(;`DZ){{{*1op01@UmGC(X=YGdXz5G<N4eI<>Na<M_(xW{)lS~e$aJG$d-&}L4{
zXQR!qQD=)U$z#@K#$!%SS`C=}>&(o?T~WoSci@IQIxnMA^A4Ut^;nP2?O&{@q68g4
zA->>GT<v)_nS9h*o7Wh;QB8e<*VmFANyfDJ)5QFjq*{|uHZ%CTxkI2XpOm}fK@PJu
zmSgIUI!z1b_y5QsUkA~XS%)S!PcJ;Hz0s4%8ui_UYZ@B+BrO6H$vRJBOh!($Jxf=p
zb$T=^8$QVYvvm6e4>?v|w@WtCqsudkT6K2pMAlBURN(=B(n@<$RPRM2l6*Q-?yf`n
zrw-SH-^-CX`HyB{oLZSkpON)5yirepnKiA&qJzrEnc61B=p?ib*uZF$w-aj$*{VT>
zntD}Dn+*@{+_Yw4I0Z|WRZ)`OrM;g!5ra9m&p40_9hjkF?A!JtF_zuF+>q|sbCM~$
zW?yzbY9Unypf*7vDeYBZjhz6|UkC_`Jpup0bV#A_=S~M@#R1?M>ExvGle<(o3!S*>
zF~77LN`A(RKD~jGIBaohN64fdZi6jlBk&k^nC|?J^b07BbUnnZb+GcSmQ<~>z02Oi
z4W2@jT|dj$ooJeZ+R7R9o3unOC;9RAbG9{aq+<cf-(_3^8|8h4c`%>{Z#jEDpBLxu
zaFG|X^j+(7{VnFdRK5TF6pV8Lz-_QPyMI)3xo8;&xM#iRTR^<<1JD;_W6R<dNa}z@
z!EgF*kuSB|R)U~8WK<#mX?P{CebySZ;kB1skb)lK>X#pa7MDswLb3&$Ku+7g>1xS^
z!utd?4Us41EP+BSj2XO52;|yGW?5u3-GG3@C^GpV+|3^cG;_YC^QoOO%G``P8@~;w
zjtug$AddiC_|{7}oU22XOEyn=Eg-lcG8eg~c=hv%Oiee0U03X<A7cP?YWE_dVD)r$
zniCI%fMh)H<=d63fc7kqVwlW<vlf4l%c+iGJu4+kxf9MGNAB~3*s{D*=*(UoD#q+O
z#E~L*YR}2spQERPpDf*7DkJv5F-xO0cg9`{PQk6YR312=Hop$>JgS}4!R!o=0A<Us
z)e62UO({OEr(C*nqDP3upL-Nti4&71=2x6tdmv;0`4$)e`Q8iBS-LFU$^nR1;8aNo
ziSQa*u9vX5if{_L-z+??x)H}sUVW~0nf6g(<9J`WaNRm?gXtcd0V&$<{?!z7+Q748
z-Yr8EI21fL{HF#NE?Td^EJ1oQMBxO8jKZY`RE+#MF=uvZbDuHF5bQ1!;}N8_i*lDH
z&GxtH%Y3~^0*??nU_c4m!5uY@W>(@5_ci#)EZ-VX<+|CJUA-cSj#caFlxPck!^*~{
z@XqR05!EMN>{@o-J{^!yl|$jufHu5~s;K(OGR|^x;Rxav!s~a|i}-g2<z}|X4}>;5
z$`z_BQ^;Q`rTcW%sjs?R8lL?n?cEN1z?n+}7X+(_W1)4xF5&gcmgZ>8b)<OFv<tqx
z-_wodDknDfaGnwU73yvwjC*$9JVpaN$sy0Q`_!fFlaA;qTqLwTq+}4I-gQaM+Uv{q
zM4Xf7&Q@B2f@WV1NE+$N3!ux^OJU6X2eqobQd3F`y2?wqAt^|Y9JQ)pL#>*3csX)5
zOVPB~mgedBu<`BlCM{i-*<M^_U2H(f9<+^Z@U?N|-VJ$VG~oRt^yreb0TaQ`AABZ7
z*_oV@N2R53DdjFtm@g~hw0T9+eL%h~bSVXaz<tWHg>}{#8Nnu%TB6Mv1J61wVGoSl
zmA%Fq^;OU&(L_Sc>Mi|Evb(?Tr)oky@w6+Bx9(3u(!?tV`uUm5d!!b&P;Mr-6>NaZ
zCQV*cH`(dH^s1siY5BPqnM*S+d(pJXm1Y-bNQ=d=hqO?xErt*kB!0W=FcvdhAZyDC
zVJ~!sBQ=e}ucclh1K`!=HrE?`n3ks?NL$$}TX3ENP3%Sdt-8;_Kdbb1YZGsI?6W5O
zgz{NYV>D<PRTex>_bnvQVE(X5v%9)o)0#;Lw&VLKai6!Pqq5a2)aVQ5zL|~b6Ej%v
zv}J_Hii+C9Y>~?A_>S{!Qzbo>&}tJPu0>b%6m&3U5^jF;tHE!J6$tOmFU&C?hQ`4B
zNJqEgjJ870wg8wV%`)BGeCDc3=-%D;t%9PIW|>%n;He(Db}iN;e`GRjqZR<}PpBff
z#Hk47SlXq#BbA*#oM(wGro>3IkD0SgM}<3P&-=NLBo;k~VI5?rt0UM~Y9w%x4!dk8
zjMU<+Z@Ov!Ec)Fl1qHoCysFIv)WDb&l8-VvnpkR<fZ+7M`7A<|8}Wr-2TxadjvB!I
z>rq5kqH-I%=}kvF%?=zhgt3X$FCkBZNxY8YwTWF+@RYntpkbo|JB5&DU1CQff&}3z
z*vl|_wwh}{UT*L{d}}VV=6Db>B-S=yEQ*kZ><HzQw|R`J%;Y_;&5mc74qv>GIqpgL
zpjklNS$U2S9M@<X#Z^LdkgY~u0HpL~wb#KGV&1g%3H~f+KGT5Ll;-6z*CBx@04@1|
zJk7UTKrjy@_|hknyFSLL1s+sDYf?oG(to_~{!;gB7F%?z%$?%vSX%21vEA||S5uEo
zUz`rezin>Vj65~{bS46+?AN?&w6Ny(#~rIo`_<Ok&L;L7SlZla>9?-u4xpmqQ<ccA
zN<l5g`oJokV-n1>bT~ly`bMqfR0T$5#TD8Nz+a-<(il4GF_ei1z^{i^hgK(*W^0nh
zQM@+lrJO?9*+URaPg9vAK6nl$4Zg}}b8VSyd)IA$k~O=0k|!chX~0W6KnEJc+vl9B
zR{5ZD$j(#C!!L%Aow3web7%T#PPDh7W{2hZkgue$;^1jQo2(~xam<=dULMj}F(*f+
zMN(F!q;P4?)FnRB99F3$cH5`eIADqzrEiv}VYMea95n--_IN|3IZV<wtrsE}ANCzg
zQGASS3D|-zYVq}}!^qOk?lFc(FF}XkMLu0V=&bNlQ&S`hsRE)4%T=FAZ+YOrsTc-x
z%5*8x9e;5rRcj9#_n5Jw;ssbHK=z(PDTG6YNOe0tEeR#~i5PWH1NyO{MIpAo!uHC7
zu?O$r5Va@FB{lIwt{Pi$U+U<b=BS{xT3U(E`%%J#49rB_0VTk0NDHSteX9htA?%ct
zl{~P-P?P$!5p)Bhw%qPZE^EA1EQBtM$OgLz;<V{08lJtmtGATYu3^d3Sf1(maP|>~
zlJw{j7DhCBS*$VJ@C2rAL0Gg(sftfybubB4jO3FU{u-D`y=D2z;RbCqJOU+<kQ&84
zQk|MNe<WBxrp=NRG^3-xAE?Y$z%x~!FfhJ%f7w(h!}E6OCY%H%PEC&}2+|nD`lU7w
zM{$vN!MQF7;sBQ>x96-%D}RPBqKIfhJ=uRV?)ckjWZ`B*o{`cxH|u0pM+(WPL0$#o
zSa=|Xz;A%;IuNiK3undS?JcuJ(a{lZh-j>0)Omo@uys}29AT1>A+`w(un<PwGMTE-
z%zCG@a)aJHb^>y@Ay0ZHf2>8M{SY7#{MR5w`YqM|K^kkRHwysI+B}sub0=$ibWG`z
zr4Pzi!7vC#@ASGzy4UO&A-+3N+$><n;T1?{o(fYl{S!(cR)I9#r>`P`8^QS25c!6F
zT@zztnY{?C396<(!h=Xx;g)(3p&d6Mge_)^_d;P(U9%ZRY?a3I<0w{X$hI!Y`{1_<
zC-6_BEMX{Ckaf(`R>B)??EoP}@GxRog<i`sI^cL9j95)lTVKue?QaZ5#xgE8X`?%J
z5nt{NowHR!w}(IUN#<)wWi26JZ^BIz;<^bP;KiDq$H2!jN8?A#VW)ad+xKE;`6aH)
z&eBi{1S^iqj3+0QzAS{fN%Bv{&?UCEt<^(^y-_!VM36d6xk(*<;%p@v^Xw`qiJ9TM
zkalREb%25?WLe7wS@sBie^iCFiS%>u`E4(`$O19x$(eve<Yr_#{f_kLG#G>n#6GTC
zy(6{vC_%38G;O*c9QI@gDviIac!fPmu!1q7g94$SNb^9yG_r@7*Qiv~+|FbVWl<cj
zE}gL0yeq*|ntuqLONc20ioy~+Dpud6HDRxO07s@!ubp8mUBTO}?g66uLF@k2N`+?O
z?HN_>DI?0jBNtA|?w@}8;lVT8V(A@O>}`<t7)I=-D4#@f8`vsU)6vkYAJml=W%<hX
z2z9WBhWQ&#@*C<#RrUJzsO9@(u$SV@hxK{Wc=o{4s01lfyto+pIdu9#l@anFJ6nO}
zxX4!UPOS2)m}1k^v1N#zLXiA*=Se<z5MK1aw~L@V0i>fF&xYsq1Ha~Co7`PQ`p#%E
zsn#`CtZA$_L~3RqSU$OSG4<Wji_V~2N5*2OZIx0`NE$HDtcuXAPOZJcG@GsZq{u>R
z)4|Ica1D%F^h|HfK&JjCXkRX%YHB)-oI_HYV+TH2p7B1{-TG#oM{m?YWXHvXnpfah
zsp?*TlWBR&S8@XrX~FXR8}8j$1*Y6^M$wERf1Bj7S(3X)AX_}sY~if}OU+)?h#d!{
zBWvv6k6`eV?us>%Cg!(Dv)EQ`Fmrs7`Q8&#Xrall7{y_HBk;R}4<h?^GM_QhU$T7I
zCRvl`LXbghk(omceLwWW3^o`D^|Ak|)6T{(%!u75Yt?cXa%3P4x*7)8_4hu?S&&P^
zhj03rhc~VvO6iTE27e_ZHG+2&RvTP%zssAqM2l-p_-}S}yk}ox6&;`}(WnupGKb_|
zza3y{?vpG@eJBX|`pt8ltH3ZdX_Ko}Hrbvz61b3{X_dKAtcACIUnepHC2aWV`4yv7
z!_WJ!A?_ipY_%l{{Wnjkmxat@R!JjNLj4#=xWzEPW|W<MXmN}<>7?v=wOOdxhKC;~
z1yw3o(MuyzW?AG824<TpuWx-&x*D;=HH+ccNK%B;3%8=%^zF*#`s|W!C{C)KGm>J#
z3@{jpwIdXyacrgb<0aWs5DG_s^lK<?nw1^u*HVb~ynQWIh;;-AU{P9v?YRcDfG<pd
zW#AZ)GFibo4v|3Q(Oa9P_@qrq`|JMLWbnBxBS+tOxxCqp7y@K$DhAjXO)N?Qj{`YM
z06sr9te??)YEnu%`qYsVeLt4rRKb^WL)ztNcN~}s_%-Hkn=3l-)^Ra~V*hcP1)}ng
zKQXaaTBQ%{MK4L7oRJL@rvw;aKJWZqe%>t%kgvViJL|!jEq1qkOwn>@@IgY*e(Zbp
zO8W)D1Z)2SU!K30xu?KLblMrVGb5UAv3^lSQLp;B&~?yhjW($}(CIc}7C06=SNcVX
zHl(v^h}TUinLe9Sm=H2I>uHWcR9|Q5HGa+WR1cL{U$}#ZT!8mqmG(Xl21Mq)+Y+}#
z>d?9s`wrtv`jplLE_0oK7CiZ0K-zluev@j4s$h@n5uI%St(+Crs0f$-mho?w)EzR)
zjgB#;Ux?;da?B=<sZ)?+UaE!7UYT!gg4ght3iS|D0Vx>&9mcbhW_BBe^`$!?ccIe-
z#jBO<BZF(of=TC7#HIalq8@u9FTf&Uv40oZGlX#7YFH9l{k{DDRT-C4hS~70BUn)$
za!sPY&h=lK7WSb@A&far@jl-8Tkx|E!mDo1Rzh!{Qc(GNfAMD#f2kg%P=Sk{mAHH+
z@D~!&XdgrzwV3|wmCOcvIaRyDK)+j&3s_2N1FbZ99=9PhPU(&6mh92s+r9Ti@C%m>
zFn@ph<_Zt_&AX+`5_HD=$`^e2eUorum-7%(d2;$@PFKTj<F<Nx9&}%q8|sy40mYN!
z82xCqKMaQ;kN1!m>GW`4#CiN?PK(S&i<k)=C0tt*d`n?Pz?aix^P_J+afk8bd)Rdy
zYpz2-TMoLCQ5u$^mwXb1a7J@k(Z_%(Dt0!v`60kA(RI0VKDXBEy8ce+hcmwI7GrY8
z8U~dLn96-SM)vWy#txur(#W!zVS_f;`f}gSY2WX=FFHo-^(XGbmsj?CX=~a+%1Lv|
z3SifnFpyZ4r~ilC``nJ=Lnbq;Qsnr=kDNAk-i~2mzwzP-|2h12w~w1*z&vuazsM<*
zpFud`7VU3kxGrXp-nj1W`@B=mZ~G(<u9Rat)K-e9CXA-Zp3OF;QaPF1AQv4)TpsxZ
z%?w1R_8Zx%TenvhmZ#1Ic)b4NJ}@Juk!vdjpw|(vK<{FY!_SB*WtGpRn=krX-?lh;
z@s}32)-6*K(hb=yJbgbL=dLV0?Rnin(kshoOY1SUp|O(EjM*J((O@V%MpJy3F>+ap
zI<dBRR1v8Lsa-r|nJYg3cd@MBV>-0^;p(aE=nBe%9YYV3D8F=}6)&OM7|$9_4PPG*
zvW2r3-qwpB#FLA;L^eT@?!cdLtcn;qt=9kST{kUjYxnJab=VVE$M{J%OMC;aY4OrI
z=xbk6?ddUu3RMbkZ`-ixYO7iMUp&QpDx0V$k(2nKQ~K8Gremi`H`@Ty6gJ0oUdMB1
zxS@+k=3ml<n^0{-n#igfo_6&)*pxr*B!w$muyC=yS0hfqlDdt8q{O92vR?>m0Q=_j
zj7?KJ{BN(!u9gM5-pk3EF6vpy11S?L#qE1wMRr`TZmlXxPjhs@-i}R<_Ka|wunn{x
zyFeN_7!yH`IA=bs`Ha8gG@I_ITki^)BqoFm4LnV!5Tb^j_PI9xHM6Xyr{H^p8!aCX
z|K8v!fc7b@Ifs<FxdlzJBa8T|Hb%kTs4-U)%|nB?S8v6}xqUr3$s?F~w;5<!>|Xdz
zt?@S7o8$Ph;*50bHCWJ$b{wj0PF{z}4aI>M{3UT}ArWI__o4brTUh>#NUsM?GsV>r
zN*V7pg+lm#TUR7v%E(<99Om0JRVy7tmu+-@bOZK=;{rA`_S4FbOYBS0GQ&i;L5V&z
z(o1>Xt~B^GU%h#c>bct?S9P)1lyyK=7`G&(SZ7=9;C)-Rp(IYLyD2D>ULGhmx(TVt
zs?T1>ogJvn@40^}`UP&BT2=U|8CyPVsIr|%OjcJMkGd)01rVP*wzSwOG5INyxta^E
z9z=fdtlny?fid1}%#;H`Rs*04Ac)`-6(+Y)2cIw?PJk1BF_{AoBcD|!iVvSG47IWi
zto}vModFt9*)5WW3AJ0<B~I#mo-cb*h{6v?DQ7=DgGy9$eq|{WwxcmvYU(3K;5zmg
z#`pk4?GTrS73_h?QGf+Ba@2isbVk+3?m@H0QgUoYOG6y*wqN+igPsh}Q@dy@5A53I
zey{2%$oF+{`{MbRO}%WZmy=z`(4$*P`{^A_M}+_L3n-$(z&H@!8l>zAB3~gX+|u8R
zF6$fElSTHN^oW^E=3i=$vG~OrTma&Fs{#bO^F3dBJ~~Z*u?g4tk~#vwRr*YvY<qH8
zYNoF@d~gi3*0?ypd$o1hR=1;AN3I;CaRwHg2}~)?WS_)u@-0&HDnUHM4?GQgONhcA
zk6O~ZD-OOqMhx8n`g#306*0^H(jlM0VO`eC#v%@=A(=%Kk1xz~@7I&?@hsT#?wA%h
zL~&krl1$}jARI!OI}-<}xjc50e@aC<K|Kz@43KvYXdpaYpnK|jK1?S*@ldva?Nzk6
z&-IJc1*WElKr7KlfK|m+hd{nV%SH9<Bb-FbEVh_weue!4ym~c8q+0V1f6;yRJrsRq
zG8a}Rn4M~!&8aM3Z4!;$jYeL-NJk7`<?5o{ZTu5y2h+-A`QM3Ls4{t(gdje@@{8od
zfbg8+E-kC!;(55ewcll)>@^oqBDhXVL~uJ>n{Zb;MG(tk5!jdP9Vz=@iZK!-7%j^7
z9Iq6|(pBhhEnYI^x{A-qlhR@d>ih4gEVX}oWevZNo{rq>Jsz9DBQ{F8u4Y8cO0WM_
zBX;ASf%MLrYCLOHke(q+jH5JmEl!g53}cOaFImq^uc*$L-kwY&Y&oddr6?WO+Gfsq
zvx?eav~CYY>(|?G<{B5kqypkS{tK%0u2M1DafTnpcz;p2-MCm?RYu89v||MFtOLl{
zv8``ldG1Z?>x8@GHlkI2qz8Xtt2B*g-;b8NY<I4J-I_mAt+uN9yzg`z&~NAzV(EcL
zlH~CWx3?!CDW9aV6a{V3Vu$wB0b$j8FAt%^G#m$~ExWb>ubhFf2QO=6FZpJ&O6yH*
zw_?L%zxIko3q9M->2v=%>;we6XWyc$UxqbYkx^FUz(Lu6Hd8kNw9bxUSkp=ayU^|W
zDWOe*Kaz&Uhj*ac{hvhbV!D3i!9|0gm6ZAhD#E+~?r%1FYxkqtKI<;p27U4XNaWf)
z0%tb=O8Y!O>x}*1B6JrFmv3X<TCe_({N4e8@0k?eW~6d9p7S2Ys+ls5ygc*EwhVZE
zVNx+lOS?ELU;FhTlE4j1TZ%}|a{bzrZW0TGR~XJjc&e>6)&J<I$;{-Uf1)3d>ff~f
zX>D20R2AX#WEbuX6fxo#fIQi<vv6$Ip(VD^E@xKZT=iob@x_?sNLA!U<B^RCP(r#$
zXKL`=TEcf{X-nc}PLk8whR<NFcRSIH{bm9WVivCT>fgp1Rtu19J)B447><2KW1Px{
z4->LBO})fB9FlmDLG9zZ^>T|uo+)0}Fsq%l!0$h56gmls3=mLG;(W0$M(UoC$2Rn*
zU%E97xX5NPM9;WEuHTwC4S;0)khFQ+(C@vSOZkC8$S`=+5!uC2&kLRwZ}JvL;^PC)
zbQLkUpk4;ER6zdk?I<1ns<JSurAwZ>&k-Vv`<J$0IahX!rk+_+3@hHarTmn)!bZRF
zvKs%<MdFLAFrBwSHM$;0v2VQ28+cy0Hd965EQ7?~SIH+}cwZ?Hx7J1)YPc^i@j2#N
zY714?7X?R>{-pH>ZMN-P{!F-c`8Y|ESkC#yr^WsuO_K$A1%j@m+_ywM0oB70pQ=s^
zpPN3}X-ErX7ZCOa{0*m%&bI`OVNWWU?bYJdusP#SegbW2kT$oBpHw%NX#Z<H<681E
zSJ#X^nS30-q`)6+jqi5>N}MwDBc9gM?V)eyB*MO@m`b@c>SCvaCpX2=L5h`2uVPhq
zI|p<+*P$jVlgrW@hliMx&YjxjM+JXmEVC;yVT_JI$V)buIo&4`8jQ3N19Uy>IIo&t
zaq35c^0|W0v{q}M$yg*q{=KfGCm%`!^m0@rE}r*CBJJ}SPmHF7BWT*PVtPkRanjbQ
zET2~jITt~=6x|`OTf!c+mM?x7H*W?nX@#u__%RP>CHVZ0q4EBrTf=41QDD|v08XE3
zk$VyQ+>=L18{^b4gy)sCSb6Y#p8PoJW!h2$VhG=7T^M%PX)CJj>jD0ya1^T`oxSt3
z1r^|Q=zG%l`I24_4Tl-59J6me!p(7QgA`;}%Bwl$BRwf2Pdm5beC{MHl(W%v;zb~C
zN!S1ORus&m06y~7QSgZZK=gZ}JdxyQlmyO^Q-cwi^F2mqicdk*Vp-<eY-2<Fn#v)5
zsk}dmsGq<2u2ILo0{cr1%d}=E?0T1auEMDYxXH010@M4v;Xqs<$RYOd(D=i&&8P=l
z-+qyVr<Ik;ajE*NK}<}dX6oX~$3Ihfv*%7&TZ_H2{Zmw4qCTeyx$Q~YL<w(pxdKI!
zf&6-C(Wx7sP%k#^FjdWkhK4_+iX?aJfarYBklt-06RlX96^NxM(F_G}RUguM08~O0
zLblmS3kDqPZutaj;`o)NMb6w#Qp~-X<+BbAuP}<-kRZYr+B07A_Up0%Gk!(@{`J}w
zB|vwLm7<@ZkZjiAF<)!jOutm0U<inZu)C7rRj~cp&Z8A^Ztj=lAc_K+f-zO$o{G&0
z<k)j|o`}m>q&C2}_uUeYEZ2ravI!sMb<f7*m+L>{YeN2}?1Do=rUwqLq9kQpCq*<M
z8e$xwINq6_)w3xqTkYUf3`pLub<k2jR59%RV?z`%nWACYiN~Hy<FO!Zbk>ZRGCC;k
z)auEaU<)AK@bv>(U1Bb0Hji&47up>7W9^x`wvOSZFLXX6V|4J{!*;CPB4R3%EQ%w=
zp=hURzI$+vdiACQaAy!5-ltV@W7`iQ!8pICJoc3FL0#HY#8E<VF&Q}Zr(e19)jr>q
zKNykHitCeLs5~)-;urHDdNV_Djfr#|ll9n9S##1gw2`n4ViY^Hbcg0M`SEzB5ITXU
zlY{12xKlIurw&L%S3B8ro6yg{_d-T3S?N#DU%D}aS17u1x|14oF)fe>O&udRO<&>^
z4@7c=N9bdbyA+D_ILe^PDK1c8c1^$3CPb)pgna*<IE^&-Bi>h>U?q*2?bEp=H%0Fh
zJAj5~&@|;~VscGkqg;In`z-za6SUNvw71bZnYORevnd)ZM*i5UG?Et={4LXO?D*sy
zdp=0TkpX%6J{sXOO)_LuQn#-hYuMEF?3!%K11m&(9Y_}slut85`Nw^2$1p!=OKX}E
z;5nd~!TwgH`YetRaS60R54S}U@jqI7&u}=qsBM@rI?+jp9=#=cH-hNBC2E=)C3*?Q
zB+-NDT_yyBAkn)!YNEtwAq3G!4@Mbfd{;>B_kO;A&wIRn%;A{X*Sc0YueH}c*PcC4
zz6_-^<Q0thXcKFx;#Vh}HJu4I>448fXP&xWszK;Eo8$1{Li5?+RlK$4)6$vg+Lli;
zJy0g4KiR0Z7f<%lqlA&(v~=&|xOKnPmVucSlowwHuuUoUd!<9XYj{icdDO3J(7%M%
z`us*KH;6Gm|1sX*4FxVKglL0=GJ6XKf!ZJs-KGx0>hPrz0`wLP?|jZSqwei#FYX_$
zEyr+!-|4N{7E88D5};@kyv_vbqE3hY{kRMF%>RYFq+!i1NRF+>ikK}FI`716=e2Kx
z^XKFN7dk?3#AnH>lDG;V9NG1ZEce4U(Rcjtkvd`<ys6LU1k}JWa`zx<it>Ty`EDPB
zANibKa3%D(sCS%oD6wA-9Qzd(X?(@NQ83$y5_C~`tZ~3zuOC_1BKI>3CC1WOV82--
zuv=;G_~hZY@<6jp#ATQG8d6M_z;9SM`c|t2Zt!W3()&Paq=5%cj@*rjna0v5{&Y5k
z0F6R|%HRK_6x`F~JV{Y-$v$$x)9AkDc>BKE5wbnS6v%m3%x<JZJ13b3c3OU!(@EBk
zl;SYnlfy&Dw1z>CC4?NTUNK#e8~nJk6>py!_y+Ve>S*@Ovceg*c8a8%@?~&@uykK;
z0z_91aM~M6CrWCpnp5j?I{_EOos>B_&6IWaGz1g2{wDsTTqZ8EElkp1;c~nu>P_pZ
zHjVS|!xF2VQ*g*xlOur9V_gM)XXl$az#-~tlM%B6g4&*%TDy=u5`+Lz;8n#GG31Ar
zTvh63xVT(`hCjXO77b|8s&@@2RaGg$6lYr|5Iq^%gm)TO`odG?_Kg|1`p3r}6pnZw
zzRbNAKD$Sl1Ds(0#5Q+ztEAl5fLWLbZEXspfkr^eBE?__re~`Sq5QY;z|J_Jnyxnx
z!qVrJrDvF$g95#YCvSe>!$RQvPi4y?LQ!HcUxTyNZE^&MI0R0>W?*?rj3zcUi2I<_
z#$U3yJ;~#z4397h;;JVG)!g_B4s7E;nj|!#Vq`PGqeKW{!HuSL$?j8cY#aG8{3nZU
z;vB_TrX>}4YG($d3Guwb5<>SD7bJlLo1E8&GfeqE+6|{35ocs9emHT!*|kN4u8P*e
zEf;usp~)2!{8<%y;%r<PV<^ORXNtmm*3?tu+fm(%y=S8g?LOMB#;o`1e(vjUCSjhU
zxrd)gyaPSnR8tRpGOq*oG6e<8z@w2kmlvON4`YQlOy70JJ<HCXo!gzE3>gyMELQL8
z951WVMs<8|`V~=F+gho2k`@EHMqvU1rPPb#A6whG|FV<4nlr^piNGZqr7!)Y+8VIX
zV0V&}GuvR=JLQeHk(lcuP^1$$c5kLR3y)tL0;f(ZgnUo?$VHtUkxWH2iUW3z=@?|3
z@+jZPm!p13RDPc5I-D@m7lyX3jk04icp~ft+pW-&g3xW`4}vbnYfpgrtv<9i?RuRg
z^8`zsWQqh^fM6D04WwwRFI9(i0r$IMq0Y(_7vbc31@h^(7^68))$@A}DQnokkH$^e
zhrkoq=Rv`0T*%Cew_y-=17=Fc$C34rF_o23zeXCcGdY+9wiXKERO!S&4%32PtB^(I
z4~9a@{xcS1#{a@-1QdfspR+M?4JgCELEw~#D@5X=sy1-eLcRmqFUBfg|Lz5tZD|gO
zolT}h7_%7=nGk{*uW^VNgqRjC4xYr(TXhur1)+G@!E3Zcw4gCu*qv}*qGY9;O9xNN
z`yRZIOrag6p*xOpk|i4DqvQpb;DKVvhA0t%;)U%p5}i_819t>rkQ@-~He8ks`I*uc
zwtKo$>>?PMY_$OAnxk`!>UwfUi2#EcIVcf;+i}^6l2_W8%wa9plopw+`QJdsm41}p
zxs(4amA21PvA<Qc4EqbODN1nxbEZmKFt$zUO6Q;$c%)}~{Mh1?h-+@*cANbFkVg#2
zWBx+m)qAy=`IQcpqhZ{z8+6Laqr*0_vtX^SAHPIjgveVDL<KIvfn{ETHi{fjGfg3#
zw)9N5MUM*Qop-EHE0{a{+ELz;zWLt5P#8^?VP*s5-f@x1YLZis=l+)%K}B~8m=*-i
z4uUblAOlAa4^YkH`u+?m^t^-a!3OnOIzmQehG2{MwevIuJcVzelO##Q_9oDxn+hQe
z8Kl)L5~tg0wt>oUK358hR6<ZnC`1Jek^rR1Oh_b5vzzb7DY6m(dPVpRv5xgsLP(#C
z2O2r6uIbmjgfh?wQPpkhh|(cmhs2FW-Y8%IXWpOD?<EEEdqk0VPrX-X$=O8p5u06R
z2N!VQeT<j3cn8b&v*548!4VavETVd;<`HV}+gq7_3*=?Lm{r*i-e!6}8g9WdCM=pA
zU5SRc*YZ&S@-#`%1`d+RQU!6sqTjFLTs~piu#oYjU>AkCT;|^6Y37L(hlhCEO&glK
z1=i`{5;Y+t>!C|_mTn%U`@T();+id-A!rpJ=365}4M^l8K_A6x-Sj0xP?KFM_#WJN
zNa6X|413q2rG;R1mB-b?96lBPgAMAROT5Fs#?Is49S^CCH7H^L%l0m7P$D`Dc#*eS
z?>Dc!^r-T>OlxMKmL8@Ia+2?p!%SZwDoznkNxU}~;(;=&x$KvwD&K%yY`~LH-cEsL
zP|ju%olRD5P5V?b;Q4wmzJ-xBU2I)88G@zyQX@fdqZ(aC-rJd5$4N$o6>{AX=J186
z8ES`b+8}G~QSTh-wP_W=ygXme-XU2SeI&@bPch@PR73$>!jlU=Zs5MoO8OsZ$I1LR
zPbzEjb-?_uN2^!f4}wY^wPA|N=t}85#-XJ4cRU4KX^aU3F}FAFnD@HsQt*OO+U(^F
zX%Vkm4V+aO$zYk(`k-K2nwS(4yb>bv%P^3_;tcSYr(IkrykODqX2hS)m_;n~O%Zv^
z4CT-FYvk31eaRU$a1Ik`(WK2|J1Cexs7RuN>CvRcFvWoAhPBInc?J@h=*ttMZ6~FF
z&>E;YhC;|8@Y=gX-p%~NGPS;;QPK{Je(s5I!N=7sqR`SV$~_xF{6@J1{%xAVDY4?~
zCkC4kt?3@Hg|zB1+4NlRDC)_Bz)FQEkkA+SlP+_~o<v>)_A@?i1FlQQSR|K6^^E`5
z0%5~0I~Gg1KGfj#TI{dzQ#XlVKf|&J+KsB>E9+)QbE*=B!YoOAw)a{!NBuEvGr@(5
zc9cRg?R%tA<&0I2*Y>?>hSF|>l;4%voIf54PUa;s(U^6?2%7?ysml{5yB89FuE8~Z
zNpP9u!yxf?fgPe4v07hunQf9CwHsPj5%Qk0=Cv!a9-F~(qb^EYv3S_h4hwkER*Kgu
zyH6|&uMig*o!i#xl?Pp*)sp<}>l~7IrpHZ}&?5zA{=vWRV?|tDH~HoZy|M%0<?LO;
zG$QYA75Caqf4@B447u|4Aiplq@3rOJAuXBHW|hl89EPd?BxPbBFZc`_zN#zNk}p_Z
zUut8KlgqkuisW|+o`}|G`K;rP;ZTaA*8AAC9d0XP<zjzYW3WZoIB%nSc~^8m4P=|*
z|KMy^wO2nucqyE>Oyv>v(v){_yq&33VP4r@M77?&4Yl}cZrdlM_}Y8U778Uieu~1$
z8H@sJY20(FRiP*WSV+mP!&+SEnlkC75y#d?n-ru{eF~<m!Bg@-CoRTJc`^}G%Ci(`
zw9;s-HlL7BtTsEbw^Nn9ZIoESJ_>gaO^M+0;4;`(K|0}@zPNJf^JzyNp-8gro|ZR|
zrPkqc(M=8LW0jD@z?q-r6M>&kMV5pBAw{C$UrJS>V)_&?3u;?Xu+%-`k4mzuxbqB`
zLUJJPY_j$fKVs1xUPn-xFV!A=)s{ymyj<>6by?lqhQ>;GzaT&flLo=4T{C(*?I8`f
zK;m806(4C(v8w9iS$+-%?STV~S>`hMF*6C}l_+CND6ukJ-Qx2#>y~N#K%&t^X&T=1
z{RyU$Z^t7Kd$v8&{|a!VPx!KZd?RnUYmM&8Oqj?wabd0uNnwo2c%+Hw`9Wjx*minA
zukCLU{xrOI-bsOAcL?NQGdkX%>svnH4(+Yc=@HV%wD{;9a!1W{7jQT1F}IvYZMR$E
z{h2H8^B~Ps?cdM@#v(E;-s8r0dwX}-o&@w%7m;k7>1QbbHOSxjdILO}`5C<w*A&TO
zC`6!`J^gWPVbXj=TzFN9v;qXGApnIze1O{GlODY8hzeE2m;Aobjm%Tt;G3$g+9Sig
z3WwYOr$m?_t`<tbq|Du4t?|yFBru_8N*flB)ei4wCmLn^Qv>|Agg<ut!Qy@HA(Cw)
z1n2M@dcjGfvHH>AxIz!0$o%gesW8{F+~|i`o6lai?`p<BVmc)kCV^S}$4+0>XZ`xc
zN3vs0J$aCEbephPjaySlsY_|cyzaXWro!68V4nA$>j~}~5s=N$KQ++@8q}t0(^>XN
zIJo44c`gIEu_4U-nN*9bF>^G;&-Juv)%RPycNp-cozvq5Nf~DtL<pFdFZ0<1G5Shg
z#aRxm`lxoUf_bfI3gSmgzQZfjYvts}dv^G0@P*>ab5+}Za|+==A3x}em`ofPP|W?Q
z4*>jFtJ-F8Q%$7kMX_pBt<2~byi|Xus1d<^!)WqZ>e!qHoQ&uHcaaAhGKZTuTJhf5
zkW&cR(14NHAVm$|g9N|zS!zUzosTp<;`0tU<}5RiDXmxT$deiDObm7=xP0>T;C|Y(
zt|OigAKM*VS?)M_BEadnOto*uRUd)!DqSPefj{INo+HOFKk;`*xwcvn{ts;t@OBZj
zB?YfYrBt@=D6EXH{0`mNw3^iZcAG`!l-N5D%<eeZuh#17@^e=SQO4=I<7PG_)=1cR
zvsB+H%<p(fRvVV-P5)F4-kY(xY1q^p0)_(MYfFDa`CGkMGG9<jnqx%BdazAc&HDI}
zYpp%=5bwyV)V@aT#O~K3y@m(PRpM7+(HjAe9zWUtnP)6@wY$HL65-^4D}^7JdG<q|
zR^xyY0qxgEWv@g~65=hf8_)pDm<{kT562mzi#3U&p-iLxg{^Dqpl9Bq&0tG<mKIvn
zGFQ*U-Sg8tUu+cjuOQ8xyCo)88$ad9!b9MavD#RC=b3x|=)%g`*U)bluy{a)Y1^-L
zh6!FJR)zX6dqYz0<6JCi9`YbhUg?YRl!Ezf{SKtuOGVsnn|}M4p*x0Pp|MgZ5xMgG
zFqpG3MHkPXz`4@w)H7%MO#fg>79-nP0hztEE%>t#a;{ZPItb|q(|qf|zZoN>a>Of7
z0RuoY2?!^Q#1>3V*m5DuXH0e@lnw+e*GNhCqU<$F9cTuYG3}EzxwH&}iaIqD;^MIA
z4<h3dJE=u^;`In67)yoSg$-YC@V^A4f9xj>8uAxV<X^$R@OtoT$1Jx6$`r^mty0;*
zi0eZ{;%WS!-6`qpjpF!Ys;oP>4H8=1Up=uk-M@>D+Fy?EymPWD$cCNeyTvNe1tt=A
zz8W3WSjm4q4c~-Z7ATu;u_FgwxYsw1IgWS;=QGE$JcIeN9h=m76!JeT(kOiyGdwW-
zBPxTVE(YYPE|iip>HjP|b?dvdR7AUg0oB1$<1T+poMjpGyh?hcxe6lcdNeq^%!ipL
z?srVN*Lu(P<J3Qlv%op&#@5?Q17{}XuZYoJiBPF4+{hOXiN&q2sKWFTwO-tLck==7
zuCuAQQG95LGBAU(5r8dZEdn*tj;o!%8TRPGuMg2rcUQE;N+sWrx~a>VS*p3`U<mh#
zG&b`ta(1b7{k6Y4l`+S_+wu6Rq3J?m+gH$Y$QymOhoEUI^+}024;xxHLcXqYFyp2!
zR&c0C2QEUiN?%3S7rFc^T$qp&L3%Y>URlUseO$@~$QYfWke-0^dXaQHW*;Ee4)GfJ
zatey6+U(&P^5`bYslmP-5LYvp(qeORRg;ICN7DD>>u(q&j9A(-1(tpPnTM4ny)lAb
zTuWluT*ly;Qz}S%eY!c<`s>yM6(y~-$-yvNqS>PpTWW=!CVxHE7vH1RJ0-<~)9?)8
z1zq|y%;$U`D4#>l9`f$8)j%x4mysG`K<UknS)_1s-&1FngW1yr=gk5?g`-&DrnZ}z
zmzi8?o-%Kc8AJ+wQMoMXE7g{YSQVB0ar%A8jl{%7RkqUJIYM@(x9xT_-S+lM7xuAM
zgK1#5XfieF1iI8ddH3321xI3_Yz=(fbd?FY*%p(e(-<dI_l}#}`pPe|QG7DBklRAY
zb7BOeH&7twi6EUy+O|MR@AKnwAabO2rQMsQSd%dR>6RaL55o^I#`^A@i9dSIyd~D!
z&5E?PZF8CD8X;Y)*ri9?U(wS)FaRov#FZQS=|i>`W7ZC7Cai|&MstK*c{0m;Lx>8n
zyvw(usPT7LH1i&G{|*q-)#zqpA>`a;Q2{~YXXYQTGqSd+$uTM%fHrTV6P&1B;}eEU
zBMbe_g%SqD<2BZ5JC)SY;!Pjjd4*DI;$A`2TCK!Szdh+(*HX&C2g7O;u-pB+<T;)S
z5-S<I0##ez1CkLebo;t_x*LK|wmGOn)Xbvw3COIkuo*DiOqDU*x7`@{gU@47u)+Q~
zvrD4S=tmkysXVB(^q|)^sYg($_{Qf1EU|ZrV`1JM@2)$wYaMGQ=Jh_T8pZRmA#+dj
zGF|ND?W9Y(Z^mlO&Nf$tvLR=N&vPOW;fb|O1w`lNY=M^P;)T9zL3#y4s;Tt?P>Ckg
zHO<v9APyI5G&HZZr9YSmhX6QLwT0uELZ|tDLjnpq3z}fH^b1}RZ+eNfC0q$8cQNEQ
zT+4}HqG+ok8Cq0J%t$3(_grZazo{D-%5K#l5n{L3G&ovBqN*Ji={g#txGy!|5`B|+
zRNnzt8BX)=TyAsh@Wl<<4<D$>e_{q2x&IOKt70g2gF9emylDn!bv6T%{yaNDxAzGJ
z7z~6^unVQMO`tOrx|MOV%OIguF1m)H)GhlZNzd|E(6$E8JfGR0hg$aYZ}bw(Of7hY
zd)s)3IcGR`N<RyHH-f*2lKv+ia^_8czw8&_7bXdiojbv@!GKW+Z2_~{wy(Wp!F8on
z9m}IMUd+S}ZDv=Nw@fI+`X7Tuj}oQ3MvxjfAxMHU7?$g{c(#rm5?n_-;ztfE@r9N5
zM#Z1FJYY=`xHnT}Sm0?sdq};G5i@Qbol>kf#GncAl6hhTwn~yLidNJe6!J^n?Ng-|
zjq0`xzk{ab-XyL@Gy=2N8%WjaooQk1sF{^@%nx2qd#Rz4F_VJibFKO=TOsnh1>2rD
zutwP1&wj<wT@d2MBPMgG(ffNq6&cwI!RRMT4LEO$_Q|fLlXxeH+*qg1fF9~<-&K6F
z@#bl8idWvn9eRukhpEks1hI#*DasvER6=8qy}FIF?!?o$|5zVdh#!$~DQNA(1+Ni=
zbIh_NLk<@|V8cv#iV2@Szerse=6U_c4}nmX{Z8mgpq43y#HbCD{#<QUIAMkOy5LV5
zlxQB;a4lzsxJmYx!qNyMy3L66ul@wcfx4KG4K>#(4eK}P-*}U{NgGh#`-V?X;C`v-
z)Aq6w4+3@rypL=Kx#ymrHlBB+jo#;w&=|NwxZ)p0lnm{8UlpnkFSuv2#4Kq%s_FE>
zRwr4nVbbcW|4iPZ_fDD}OF`kV<-Bm?-zVCVu6R5vJMmf=d2<sPExq>a#z>*=_L33=
z9{5lN{!06ac%dgF%<bvZkhEw>z{ppS4Qw}ynzvzHtS{H)_AWjz^f`G*#Ys2n+Tb<i
zPN$8^<oyM`B;iNim&&I$d<3>qZ~i*9-Q7$)C2y-M6Q4Su{Y_fcbeX_?BtX{NKTn0P
z!7_d6e-S6!8e2}Lzb++<9ETy|+_Jxju?z~eL!~`h%eV1wTc_a>jpDzu!&XTd(Dd74
zZVs!_O^dL)m-I`D1u=5pXEWeTdkfh<>`Ht3srKt9jI}ICVqFkU%?ouz`_KtgqgI)4
z?!$JNMU~;knk>1+MdHy(k+r|02ny$TW8(8}+tpLc{zo1)YEgdnbRyJwNFN{I(1ev>
zx#g=d#i^=Kn6qcOZWxHYpQNy7?8$F;h@CESnS*B&3k<HCGetKmb9EeF8EngOesgU<
zXs<?vamc0|0<284iAMDynZ({N`CDaa$~>NicnBg>ha<I}*$l?#)IF_i<FbyDLRxai
zO+Hx+&0$JsDNKC6a~^ZUoy9ADUmrv?Y75CELQg$5)u`bL?D>T0s(>gVuXNr3!5Y8b
za06}gUGG+?uyp6$l`Fa5GCYT=*xO8Ouof~nBXq<g07+1@g7z6+bAwFd*}B~3W_<w-
z_>mpzb>c<U!N+5ih^e5C{fA2XUEi6tYnaH?Y>G*~b?XWIBh@`f%43=@!*w-cgPf9R
zRKtpB)E&ZzA1J6ymzb<p-3usE4CxwFx~9~F3v1NHl~T{vTm1G!>T2|R7LPjE2#GXS
zY52;<f`HV?LEEPgeI_789_$8fxK6Z=&>E%HnLZo1wLX)X3fxUB>f~;AYgwC}C@sd!
z^n+{&O_WIReq1kcjVG3`Ul_f4t_7(<6@dV9_OFEK$v3MnBF$u_+ob7G#Xfs=neW3P
z(}hZP>e(Uvv14vGX=SXL%W=n^>p1D#ZMQl+|88pwqeR5jQzB+V#Ys%OaAna{4jn-~
z#lR`-fcjVbbh5X8hHDDYqv<l9>FFoRT9I4kJe6#to~oO-{rZN&87i@&kQ>A9;JNki
z>TdndotGi9EA5Hu(j|+}d*s&<_=~M^)aNo43rcIDI=oCO)p~C_U&b&n-!S4j^cMzK
ztH8g6p@?6S;;yiwFa*h`x8Y+n<!k5DqLCcDNSAWe?5za?O@Ts(s8?d|OrS%Y6Cmt|
z<>I0l4EyJtPfMZfY<^m}8FoZjNo%3a!(wrcEsDgE=I^1yoV>`88P?OGVUjjb*;N7<
z=>7|Kwlfm6ln(?uAZAk!ignafrj5D=i<>`)9*J+bHXLmWrNt~gQ)no8TJ~^)kw_^3
z%QTE7n`;~S{#p!!t((0)xbXhz17=U|^PN-LK+Ts;pO9jh{pSJS3`sTc7sQbKC360`
ztOY^WrWxh&YQTdKyvK}nYT4y;%fF*<^^O{mgyq0~@`n>3BOf!pz61Mufd5!>FrGVb
z;1fm}o(h5E8vWD4t5@W~RKxVjj%Ax1*br>`9OxVXT>z%n-0mC!5f%u`_PzLk9>+Q`
zs0Iuof309uPRW_e+T?w9uj$h}WVEt7IpWHz3$k!3RGi9j$Ka!#*jvtZWvukjLyxk<
zle{nFlS&?1ZTc_4jkk|OvV;I;<fDke;N7k7d%0CK4GiAv16skAmi8807&$!l%MJ+M
zu(v!?g_MQmv>n&Om&_0*osWfxS(mXszDJ8HkJW2c>Iwb5u2m>H@l@#cSz@VVj@4Yu
zZrk=E1YN8g$u;)sPn6<U9Gu*)Cy2K;vLQDzUVTyEr!+m+YkI+p+%ft70H6i{AFvE&
z1f0oH@E@tQO`g8|O`$qI1`X>0ET;{f0~m%iooQ^N8+ubCcQJad{qL9$G>*BVozJ!4
zr(YkR{UZffMd?_7Qwd-TaDZ0A23R%$<v#z)0v<%81xztAl#X9QW}A7D`8;Pl-<}>z
z1MJS<?f}4eXsU|Dh%v%uZdYgN9+Ej(wVq1g)r!In(k9b7DT>S1p_Z|)?i>p8(HrxO
zkc2qxB?sAJ=}^;yQ#Xt-^tV%ax}`KL0<~e4U$P1`2!s@{E78f5@6(RHSg65!ERFO0
z9|S$TQ`)~fa2#Dc&Rf2MnL@3FD!SwSCGH!V@ny{e+Ej#)iTuf*z%}y5iT1;fV<9ok
zk0xii;|lnIPQmlFg%lk)W>>CX<v={qzFX5W=w5WpHMFy~lN!m1c#<H}c$0Tdtv0}m
z8Iw}n6BahFv&LpuDFEXpK(D|cTx>{67$nwe2BdJdXj||^rUp_(IW4G@O2c&+xYhN;
zByYU;C~^LyNpbyM4TT)U%;NJdV@B#&hcAB`4ZbJ46%ftQ)!4Qb;%7E$(=$J$MKg>J
zcP9nANN33!Gs`n-oC@HJnQGAx(rxa{fm)XO=crE^q~WIAfjW0L)K`a%VNR;gK*xJP
z(~HtE5(2*?><$auAwi?g4>WHCD%a0Hxoz6%-tAs;2YR2bOdhXHeV^jv>PLdnRNW-L
zf=y*zd%DQo3A9N2*uo7-u{(@)_j2r7Zs!swyUDaOCInKurs7+m;hqw5$<$qz=M|ZT
zy$<qiS^@5D_|u^daC!*b{*qG%Q6ZwQ&)xP?93pd?a^dh*7b30x;9JI1uzyxepLN5y
zooBM2A1N{K%`haP=3>t@RMjlp_g$XGG4H3YGMqKCB3Trc>Yd73y(?$z6q$T{29`AU
z#`kg1o(c<QHy~v*5W8#yC=-kCsrg#I(WZ;jo+1$|W^f-SnN_Aj4KxDcaXvB`7Z1Nh
z^A70*r^?9FAnMl-bI+n7b%z}D@5HMjeS5Fj=nsTJpBdR5ASL3d^_Bmp1waS0Kkb}P
z!N~<!568Nq=xqd$!wO6*_DLMb^4~WA7qjvvjNbLOqDFA{nE~1Zzh~I8VyOl{Bmq^h
zDa-{$I`bAcMX}OavIoCtxHTQ-rUSHT{<;WCuvQS~$@O_}x>&Y?8ddvgJT0l*)D9Lm
zY*$E)@RkxLHA%9Q$R+l^VQD~|OdV!<B>XV(AhU=>{bk=bE%DG5DH<W5zZm+L`FhEQ
zoK&3J&(mtwhmRyh8fE%1?tQ}n_%5Fe3=N=QG~fZ0O$vl2_fd<%cx8>0Gq{&P>C4;9
zf>U0t6QFmA|Cf{E0#;1<Ov&G;P<LhokTt5hX;C8l3bY7_#J$R<fPBban0EgS=k{D5
z`xtf8s$^7fD~pFe%@=B-QL>A~s!aqMC?Mf^*Dx~Hx$J%quE6K>raI&)95nX$%hpIN
zXqsK{3VWuoPPb9vZmXf@$D`qqVlu|oC`Z2k5MzK6UpW(-1Qd8&lkfw1B$Cb2W-e4v
z95|st&8;PW%vwncQ14Sp)KuxpaA^$WHRS`^)c+EJfIBXhWs-Xjd>S(`Pvz}CbisBl
z)xz4t9_mUiSi#o>*DH%~mO)Fl)NSJ?nQEppRNBLu8}=d}(s^?{Z>2{{P$I5fG+$CF
z!M)`;#$LY{RI!fNm|~r%hw}9O_S~ueF09Bt1*!v>m+ixM`5q6sCzlBSs!N>wV?V=A
zUy>aIN>)nHQ{BF3*8Yk9Z7SY#@90q+8>hF@+0$A9HxppHM64=lq~|kKuOX&D+5uYk
zGEs~<%NU88C6%Sr&oX{&CB_%z>L03-SQCDF@F}*pNb2ebV=UrFa$~W1E1Qn_4^y5R
zy}@|@5XY2iMHumDT0x`|Y*!^Jmy$c;j;;B}+cCMVQfds8Q$TGDv?1{>yZ?9L-#Ysp
zVjZYmVG!4UG1Zv`u^jf12Rj2h6)yQouCL%@*r*Eaf35|(RSH4g+Fn|Za*o&ywEulL
zcWxjv-Ya9-CI3rVMoGbZ>Dt=C|6wtscwo5}QnuK1p8tkF7rgdg1wk-THiM`M284{r
z^3oaorf6h#f8R5KNxRo~bJF|-C+#R3UZ))fRrgVPDE9SDE!$(=mzU@oGtR{VjHmmt
zZ2<u>Kl;y0o13Ba8Nj}|W)<ZP`X<n3R?7a9>OQtg5rMw3&t6|`D=U4~n#ed-Dk}q3
zo5-jfOwGP!xh*Tn2&^+n?Llv@OG;IV^!74t*<+!Dy@6u`rw;aZXL6C#0Hb-4kw`D5
zP+Xw58w$w>NM2$v<D=gx5$Oo*7XZP!ul@#<c;>~ZTNfUU<-V9#Um3oB!^$+BPO`T*
z4roBlw7aK&2hfCJch8!9IeoG3^z=S9LS}jT^przt?@gLg-@pd6f6#W$9{X%>uMRMD
zc5m;+D(BTE5I#=mRl?t|Y#)MITXAqm5;awoO<)jQ95xL$0~{P8Y9d-3oaZ;p@GWYe
z8hm}3vcl=xoIgmhA-eeGZ2nYBZ#ntx4J<>uszHx$K3lM~2w}b@*RAIQMF)e|3+AT7
zoK1SFnJ`e9;HO{n_uBk~pxr~l##X(vP8w5``JkmgD-Gk9adi?|3j{t(vce9eUSp=E
zd{O6NPpSGpTJuF}!PNnoA7OGuk}v_$FN*R#!=2Tn7?HX9uPT)dZnO2tN(*&V**B2C
zR<>`x+l(}X>CJQh%58sO^^!pM<gLKr#<f9A4YORx<^#hs>x|E8!J4NcFJtb<ZE7j^
zK5kvaQa}DUPb#IinFi9May%+EWEz<pJcsmbKu@<Y0rXPnW&uo0bMF@rITU+EWioDp
z7A0jRQg}YU5m!JZGc)zWf9H8iXdeT1b1)?%t@Hf$1kwCNx!Jevc(g2y6(E{HZxh<%
zvp@@LG>FYi_hmoF{Lqr*TmQuDB}CS`F=0-PAjQa$I$1t6hU5RZ?jf6C$(pdB(9t5I
z*EUB$R7cKyg8A1SW$&fx<DUzf1<kb=3vG-P>8sy^3!CI~7g~RY%Ja1N%Z8tfmfbwM
zC3P6A37;FBeV{Q#odwG23>MUzjYAInOv3Yx8Bwt+@9}!hv{Ljta%wkwe5)$s?DR=%
z@9u<Y7kyKwN05#oqMR6ZXIqyTBpW8Orq<1_E_%t;YW{IS>IAS>a*LNfOJ`ylS`c2b
z_m@9k-ZhW^9I*JkBRuiQmWbig^{Ij0%+Vc<4_(nBR<aUmaAC9jt0U!0wg=Qw^Lg4B
z_j#Lx&;OBkg{Nh_?FAG4myWHqcH*|lb(y&JQ-`=dn<A0FP<a2oqTb<!Fkt&|F(1gM
zg0Udom+w4-!p=+o!yry$0`$Sx4hK3|xDqsi-BsI#L_AA?s_3L%IYAEfeOxYTu*-HS
zFxdC&-X2H!h_dxx5InR~__q2O`BzOh(|qa?;U%R4iXMf=zMRuRrQG0u@`=PxF>+)V
zAv$YT0_Ai1_UjWqmC2guPY@~4U;k=9ke2z^?v~!;S6SDBfWW8e=~5xI$563O<5-=x
z){$G!+5|8n3lk{+N=V0>wvN}eT$1>*(F*AYA9^UoIJ`Sq9Xx-$ryeMo-eAx$ramW~
z<9_|9$^3QkkGP*Pbhh>0&e0I-;yb1~O-BxaVQ0PJO6Mv$-5+f3Rzju)+H@~N-R(+M
z^rRk>Aah?6>Q)~j{foMMPtEnWwJq|v*QaWpCwup|Q2whP(-<Od_1DVzNNvt=e_MTm
z+hSYSA@_AZWY^6q;y;*-nfGR_yT#FSvh-jpY&He6&m<fzc2aNH_}gT2nddUKPf}{0
z3lP1%1W?Zuh4=;_7C?~5oaejrqjYOdKviZHHP41}l?WD6C{|Bp5&#j3MfmXjP0UB_
z@i&~F%|0ci`!T@{BApqSAAJ|1X_T3#YI+&DXZuuZv>t*l+zCMQx8Sykh!d^sfYh`8
zMtXu>p`eL9KgQwO&O(U7qXvh;XJ7ReZGfY<!@X>6U+?(OEFN#|`uEPDCQjOpf9w2P
za6#{JZl9T=7To}U6x9!z-w6Hl9PvqM90i}b%J5v_T{63R0a8HUxT|!i)mE!`59@{q
z4zc%x7`>SuA(gO#uIuB^4AjdH87_ctt_gTUJ7r(rO-BQO6d<<((qR)jneIC`h&}<w
zt>3Dr8F!5lSV}VlGh&ZDt`7TlA|MJ5%lq5%h@WYnufLBR0cWku>Z8A2tNKZp^cPe)
zy@7fGU`7`JRpxpF0>}^M`HF=d=MJ)MAvzV=J<EIC_rfd6ffV!lH>}vr@JVoRZc%`N
zS7i_kgadgEf#cv%-k{{g!2uKhFW+WVUZ%<MORr|s^{C;+VPoq6R&>gg6>NnNNI2R5
z^5477^h)18SdICTrkL`miw_`>-krF1m=;3k!m|e`+*>y1n$<=&IcftOHWrkJY-}n7
zpjfus%^`*_RAseUs{*BrEp##jgmalVg$gSk>;$$l`lR{FYb3Ll-F|U|k9$p&(h*EG
zUvp)8cXT+xoVw<bz&EObE7rS{NRzx8mw=l?bR9O(V@WN2J3pNlh}#+U3rWdO#RIzt
zc1S>?kJ*q!9GwCR?JTl6y&RLMoosilF>IG)P^RWM$@(F;QPPX{`7rgcm$#d-IVc5>
zo^@0$FFjMI0MX~yFbHQPWd(Ub#KA0=p+NkKL|(vfs*Gx%e6E}7o0u4e3c7{;Lw@m4
zp)K}68Cfm@5Aw_ViOw?kDez*l9f(XZrWU7dB<K%4UR=|h?l!#LrA4f!b~_3gMy>Hd
zPRc)Cgpyq&5`Rv#)2Nag3`4zM5q8dZX-wGKKUJM6=OBFd-A9#`IO$1bjJM&ZC7gfF
z5gBY%N5hJQ_e~^y3JG#^RAZL9MEPv`_>uW6gI8>~sc_nf{thnnT&ape8ME6w<ID0y
zq&lz;QQI{%qo-zHtXZL7y^U_WQw#sOi*xDc2h&?YRb1`N#+G&?1Ktruf}fuCnDEca
zx9C=%J!I~U_164khx<n|coh^-%W8`L;Ba76Cp%b?;uzUu9J7@o1?BBoA6ylGAkfqR
zVmtje;F1k&UzxRlEk1(}^Dx7y4XQEnxi;;CP1Qm1yGN~J5-fKla#}x-LD>FW@qTt!
zgZ@L2rQy6y{E8J#i7)}m!L)csCVpk=hO=+F&>#nwO&U!~(a?~9Xru$>Yj!-?9ZLEN
zbdSUAbuT(i=EQO86sC_sm|x;~e&2W5@=Eek2VM1@1{VF#3?~*hok5Vh%9m(gHvWcw
z?BpG;=g=iF*cX#ApVgCFd1d%vgQ-GF&mjf1M4VZGa`vBC^%qIzGPkT3eyfMpRIW`u
zh!qpfd?j5V*meeq9i41=rZD4gts~C4%Y~nYe*vOL*(;G(fLad6Um4;Y3VD=G@2y=O
zgA+;ten&tgUi*ESnUyf~Dh_Tj(D{83a<sZmpA7SGhr9|66~}>f-+U;NL_8#nk0VN?
z0T4H2oSjeF*_zR~5Anh8fKoD?Z>srp$}tTWN96bC|2Mhd|4Bmm-w63~j(3wzU#j80
SoCpr^r>Ul=T7LWf^ZyS{Q%j2g

literal 36406
zcmeFZc{tSn`#<^`LI`aLp$L^w$-c}KQi-t(p?8ukWF7lV^=^@ngtAV_zLhoWB*_*g
z+hiG&-Pp!5*4fSwE$`3gdtJYC&hK2;Ip=!+;SckAJ-7RLuaEn_Um^N>+8k{BYybdo
zT)d!Z2mnkF0AT#Elac;Q?R(KF`agjFWn(S+zaS8Zfq{XMk#Wb49ZXD2%*@OzEG#>B
z?qp?UWn*JwXJ_BFYZnIx2PY>d7Z=y=-Mjbf*|T@=-hKP_?ccwjo12@5hliJ!_rQSz
ze0+TT{QLp}0tXKs6ciK`5)u*?7Cv<7(BZ>}MMOkIMMaMsIU*(|CN3^6At7<}=+R@x
zjvYUKTvAf<#EBD9Qc}{=(lRnKCr_R{b?VgV)2C%+WzU>BBPS;(FE6j4prELzsHCK%
ztgNh}qN1v*s-~u<uC5LSgCP*e*|TTQoja$Yp`od%simcL{``4uZS4yeE?m5L@zSMB
zIyyR+FJIQx)z#C}gF>PD`uYY228M=)Mn*=)#>OTlCZ?vQW@ctru3RxUH@C2`u(Y(i
zdiAQ6m6f%%wT+F<wQJY@`s=Uj*RR{!+TOTv!_Lm`=FOY-_Vx}A4vvnFPEJnF&d#@P
z-EwhpadmZdb91|W`?kBgyN8E|r>CcvmzTG<w~vnx3<mS{_4V`f^Y`};2nYxa3=9ei
z3Jwm2!{G=7;?A8rckkZ4ckkZcfB${|{{06J9z1;b@X@13At51;A3qKa4Sn+DNmy7|
zczAe3M8wmlPa`8EpFMjP6&3aT`Sa-L=$M$8*x1+?FJ8QS`7$mpE<Qd!At50#F)=AA
zDLFYAiA277_3HKO*C{C}Z{EB~O-)TpOG{5r&&bHg%*@Qn%F52p&dJHi&CSir%gfKt
zfBW|ByLayj3JMAf3yX@1ii?X&N=iygOW(hLUshIDUS3{NQBhf0Syfe4U0sbrp+0>0
zP*YR$@#DwZ+S*T_KGoILeg6EpzP`Spp`o#{v8k!4xw-kvmoF_XEv>DsZEbC6H2Uk;
zukG#a9UUE=ot<4>UEST?7!2myw{JZ?J-xlXeSLlX{ry-hc3@xthr@mU{(W$8aA;^~
zczAeZWMp)7bZl&Fe0+RjVgirHPfkuwO-)TtPtVNE%+Agd2!y%0x%v6|g@uL1#YG~K
zxU{siyu7@!va-6my0*5qzP?T(k;r5+g+ifHsWcjG=&e-{T?SZhUoiIs0KtbFf1rrb
zc{>0&1YFcSYwVjajfElhk77T};|#(*G9Fod;D=0y@LOIsiOLAQqOY&{{Q7mV<(Q0j
ze}x#FlGn5)RPC_cYpe6R^6>fnI{US>njBv5cVt;?m2I1zomOvaZ5{a1!d;C6e@jd(
zEAty0Q=fKu$sPOXQQY$)BJ=J&#a+PjM;RfYqk9bYG3txF7H39kYU=KSgglBnfY`f7
z?i3&<;&dIjq}loD|M3^Ql>s%g5#hSb8M~n7)5)RzhW^dH9<C!6*)0*VYeAAYr_Q4u
zc5HnRzIqU!t(Zx*-506HM-wSsl{gkicGp6^q@57UjVIAXCy=o-1<l%xrosf0@tO}{
zksAZHKDMZM$TsJV9f<O>L1p`q3ky2hb1cxxUq`+^dvl2$NEcboh69qB3sSbO2ATkW
zsbYP$xqQ9zYoN&GGpJ`;GQ?nns%$V)d!iZ|?S|<ES(P~0)#W#R_jO4Su+MA~VBMUd
zA?$6=4eF4%tHCF714YiDJEOAcaYg1b!DzIv_&Ow@!HE5QzW9=*5am_e)hoUiMjL~=
zrwSUs!PtkXi>Gjw`Sl+tLY9NQ<1&?jvDRuO6mzQ3YAlLsO_;S5TJ=~vALv2m;ZQ44
zq*|$UT@<h2a>3v~53Nw@4-0tBz&_uFX^!nLO}i!9T$<*aQP;oIiQ63#V^ZMt_8Gb~
z(6&?=|8CeNkJR;&Qjtreq+2O0^O%pt67t&x_na{ao`|b>^*%msF6gz-#lgPlt}IX$
zJYarmL7{YY@k;RH>UVvj{{77lE~23VE{j<B(pQB}tD$_#sNB6es5JILy+1x)#byTT
z4}LUNnK?A?uf+mimWLlzvDT3<2y|wvcm5KF?B2t2!!cDy0Aby)(WXwS)g?x=x=u@-
z9upvC-_qV8HRTcKJo{kCHH$*hoP9<vmCQF`EPuJHk}3=($P{E+lKr0Wm@j_olj*T4
z?;;Hg$+XiRYYX$?Y(|dvk^KgO6^UsisHMh;p_|8YdHe05stB82%aO)8aaCzU@Nx+*
zBA3_CSQqweQQ6hEkuiA`MX(hKY7Lqg%1w3dnYSp29JP!BTid$sD1N4RNawg^FHn@S
zD`3(MHGRTcpkcz^M)jSZspal~&?jr7r0hWLkqTJ(j}5PpjC;h^`}m5TAfu$ku4qOr
z^`m6ff$2kx)M2YU#TpXoY*>J8&qtfdY~34AU8t<>_)~JUT2T9pC?@g9LP8uFX&uhy
za(0-fpoB(iZ?3wbvm|UPfhVTObaAYc^7<{9gC}*Rz8N)zW1fYbz{Qd0`Q1b_ia&wO
z4SU*|ksgudmFT`_&Rx;ATGpXeTBD22E3xlHFCE8m)M1UMRvd>24%y5_!<N({*f_c8
zW83g|@@vUbVtv&cw`Av~*s$lhQj6f6SR*1C9GZ~gV^kG<a<xB__d&_`mGw5xeB7gc
ze3r20Lj{T>B}lk|h9O&9ZBcCBL8ns+cP(@dJ{@d|l?P@BwaX<9%Tx>N{eFj*d7|57
za5kUAk7foc?1vfU;?hG<DY^V;H6qa~vhjX${d$4@+i#l3f);Msz`@aNw8RF98V<?`
zDKFq%LraX*0a3Mkytz^nhrym*Tq3Kp>!k2!)ip*80WpI4VglyXay74bJU8jUmlx>B
zh(1NFT-&?*jK)ayH%nu6R3MyB<@K8dtbV_4ad5IYHN3<LqK(JH&eczmht96M&QCT$
zZOYdU8IT%Zfc5;xF%_ts849%PP{FX4jLfmVyTT39YuUKxRackpDH=j22_&^1l@~#V
zn@6{!GfU@nI<v$m9;%I&aRK~YgKLRF)e%iVSpVySwa?VkKAIOtH^j<`qOMC8bm>AQ
zrKH!^PZ>vA-mBwbsrcS9TsoDOEZx_88V+Xds;O8jYoSOCEmsa^&+!nWU#(@cG~CQv
zKf8lu2FDF}c%39YE*LJpA=+Fb4&M}TDR;;NNt^Q1NX(j96X<_fH4Nbz3%0&zwMX3=
zx3^3RbrlA`)2#fs6F!BuQ41*Qg?=U{ek5td1SrV(`Hrklu0MdYx}3G)i$N)C%6oNj
z4T-2_qTKSyaGP1T=1unNizEL=cD%*^w6Y+=`uM^XcPBx7x^zzc)KK|H+Q~&SO_4fH
zp)I_DgGu&IV7^7JNGmDaTvZ5|+nlsG!~s4kU6K;mCKrxzYmT_rBcUyr{|`zh?nEz0
z9glaLma)}-Zd@%u*RkD%j!(F+-Jl68k#SyebsmcT!zEvDg%hfr3)=90?F*$!{{BU%
z<fd_&VRti%Df^xK{gqnE1iPBt*^$U0o+pt>H346%t+DXSmD9dzSLNAE7dtYBivykS
zYDkTp*LMzKJbiVoJp_=#y<VWpe)G3IV$p;RO|?WiGY>2B29!(h5v3{vmUc$svUEuE
z%9lq>f)<l%;QE&M#^^p0DKn62d!Mwo)oJqV(}9BIq`->RLys`IYr@q5#Z+~>`s@{!
zkFwN}C*cP5&THK}9$IXu%e^c&-Y=x?16fr6SAm`aI@K(0@KQu_WuSY^aqw^5!Ijk=
z2FzKp!{xCT=Y);TXcue(6B;1R<Y<+P`KT;>s#0bE&fBxyI}X(?t96bbbwwT0J-;%J
z!#tx-f}R!(FTtG~x+ljACnWZISH`&MQa{<u%ip!o&0YqT-5UHjSJ*bHBf(G<7}_)*
zOZX@GO&KP5Ekimft3}RjDwVMp%qYlQim$hJNRD?q){G!!#hk??#N9JF00+ZcZnhs1
z6mL@)tAJ-mwTim>PL^7(%0zf4@5g(56okL8tEfkrtrCw9_?HX8SFrWotW?qCD%RKv
zE%~y!dC8KfL@lc2{KQeZCKLc~3Zdx?<HDuL$iXUu5-$-*7{!Qs#AuCr9JePmdS>IM
z)<|b&U`5fuWN`5HAS&zIDr%}**6m7ZPHQ*x(w$dShLpR!LJ3sF3gMb(H7R|dX0ODm
zvr3Bes(MABc>xn@=8fBID(<#raAi5nYsp?19@g-z33@8VMR%~DV^afW6G7&j4n6IS
zN%v)N;L*h}ib;uR({U`vEBirbs=DsFpo<iue3m7hqql%`9=yILyr^KTXl=Dp6V#nc
zP#e-!IIxluusl=Hn4!?mT~D0$d2Lrs!gZ9I2Nk=81<GC_bk?GJvg_qpRIRE(goAyk
zSATt&Po6wB?CVWa$}p=fC3v(&=edriq}+0TY|j;1wx`p{;y%{}Ms#~>6rGFUPv}bN
zICjry^z?(>$YOYM8acc~2-0`SmZ;ILnVWJ`>Cg&aB2_uQlxMwa2ijMyWU0H5EE!g1
zr^)WdZr~~Ww8H8s{c#EpZR2s!A&)Zn=!!^0>0HTfxkb?0U$OEF=B!fG09NFwKC?-g
zVEX}F>+B(wyo~BI2C|q0mXTbn-i3hZ784zsHG{*K>V9coy*%2Jn1TJT$J7pi{biO+
z<jtfg^zrV_l}p+JnEX=AQo_A(3$24sgQY43{`dQ%7L3U7;?X=(S&$w_&?r@5UYR;x
zuCq7)lX>}>HD#WP+1L=Xh5ICv1u1vzXrozy!~LKxsU*sK%kSmv&iuHAIytH8Pu?`M
z=7r>nMP~~d8_Z6(8M9ikcz;Sd+mW&?46hrjn0*~rea+kjRm=@9qw@OY01jP1=Prot
z4rC<A!J>y{2l6}+rwe2;=nEpmcR^Z!Gv@(*K15n{5Cm}Af$Z6VJPve$96tU#=zoO}
zy#Nv}blTFUOoo@38E2vD-uYCP;M`ewN&GP=gAL*h#GM4>oEZ^GFWG@A<y{$`*QFIy
zV^y51E!Fq1K}>)w=XykI<<iKow7-%QD<Eef2Uwjs7Ry5X(jO2?G4<bPKi=xqZ(ctZ
zJQV}|n?au&G1)7F+~*%Mncg8O!+<=W&46r6^tqh}Of>WleSA{tS=)(iQM$*{A{oDv
zW#a_JmXNk6WXnM2`&|hBTsDNBDtWG!g$W@o=%wlY`AxqYxw+5-`ds&K#?8qkkutC*
z#^V~2(sV|lPo7oF2m!YynpVzF_b*aM?@hKXKMT3N!9G?2OkS63pm?%%>V2To!Gw`m
z++W8f2p$=ag(Jn!kEN|Cs3)rw-x4{9$O4Qp)*x)!&WwIb$HVkfb^$>mvCV?FZ%8Yw
zy1QM!Fe^&?e5}m2S=`u-qn%RvwVMrN^;e=nX0)WUWo7pgru_X>cM<t<9L8qK*V&@7
z&WBdeP@6NuQ2y6WG;1^ugMNw$2Rh!fmRj9!<*VaLqrIQ~-hY+$xzr=2N!-{7h7Nk!
zEw8WQ;5X7=!ang~Hxg!Su@e$zDhIJA8|5!N?yt)W&>Ykoe_z1Ljd*5GKiQS9ZRHBb
zr8bDbk6atjUy+42RpFtk$rVWYi6ZV~`R<~hNtiJxYv)7~H}A#{xY1>0jHcz|qt-+T
zp+>9m7`>k~Z8h)%2UMB5^U=><2Urn58A(#`r5+NQvBB0;(1h54@@+D?f)YK?Y2(^J
zGN#td^c|jpm~pmIlEFb%v;LUG7=b~h(O~cIXW5&@KW;ME43B>OOOLT&Fu!|W!Xq*I
zMQ}5ARw*cuy)jmS7-Vgy+sDkZW&y02{b-(hv&6X{gASl4K^b4Om>q_5%liLHVIJjN
zBs>b1SA|VDvW$kmbi4e6)qpd90Z!cy0B<ZN*!$a4BUd)-cw*h(kt|Q*iW*KE?Al;p
zM)cpH408eC%-Mf+%D>9>Un}$f4a;c^cul?xn2D*lbw0WdbmlaW`(n%fKR=Be_u?_h
z4|vVXf^Gnct}lZ?8G8^njmi;8VrLVD!>=BMh%%xdZ+jA%49cnMTp72NRh-;`uw4+z
zo5H&h{2&MURJI8|!|AUX|Cs(b63hYe5AGb=@H$Mts2>947|^#aaig&s2&%|`&Otp4
zkqNW-=bHZJJJGx;JR3WFp5gNRIFgWWGUpl5@ME!W49=33!(Hm>4|pm<_vYJ1!eEl~
zmv~ONU9Jwd%OiOYMW68EmB8)ASVabk>6rDJVTV#l%IE-Prr^AlwyjC%1R=A!eM+9|
zdA^_$nKUs@TDzC|HI&ie24HcDgaA5sLJE(~tbA+>B3HV{RvWwgd3<+yak{bI)S%4B
zQw;Qtp2?kNWY8A{TEqbpPzEPtA2aeMz%RN%w2r1NI&50_|2im*4H3SNwTj0vz%k>r
z`u{ixu;4+IT-my_eH)gS1<A^&FAFfC3xJ}!0xr7Q1J1<nWn3SWn8Y1G1j~C`4=N+h
zGb6bRx1KoxUgZs%YLvH0eTQI`l{zf=>Y3=)1sOeIv`cQH#3*WLKC&Xe5AD|0H@0dt
za91|ss_rC7cs(wOW5dU++d5_14Mb95oH}Lv#Y;wFbO(8@@Zm^Xb|uL^9QD1Wk13g#
z_6P<1vG6lcufZ~Eip#z;lo)f55VZbD&ybcD2d{28KYqD<so(8U!t5%$>xp<9`L*h^
zB2#P0yj*{**n~*(uzRH*4Roz2!V_uxhmAC6jba5F(+Jtk5%N)1f>qyJWjmuaG_Cbx
z|6sqYT@DvH!nx79WKvXqDA5v`S1ICA20N9#@_=`eNxSJmIO-Y;PIvrRWM8M)KlZHr
zB{@Oj^k+B^RU>j<Y^q0wYa-l#-5Uto_ZxdJADFo4Z!sbwjDC3-GOio2K3VgO#VezF
zp5xE=SkXF{KtXoIZmT6L41nIZbQj>4{3V>7ClE<EgEx!vq+}18fgj`^6X=ZTFLunl
zh|qh?Of-m;nk5@!Z|_2^|K{=+gZvAV)Ft`RuMBB1NvHTBlwTZ~)(fb$B`0)%`O$=C
z1>($NH4`!zh%?=meGWvc%Tp-|>}JyOuLvQ*%EOM`2O^gavOqv*inp)ZqV;o8X@g!V
zZ7C-C!b1lLfg2AWW{<Ec&phLMtrgG*bfvbWCeuAtjtlKrmnoR~uzb}w@w?qp>{OZB
z$YnL0XNO5FyQX+Xv+dG#Zwa+QN`<{c4OCOeICemx1dhAZRqiWk32%w-sBAXtCyz$f
zvms15AT7NLNcLQ*B_Po-fgE#YH8we+C>NGoFc>F8>}OWeYU3^QDl1Qr^oWMnYJeyE
zQVDZrJkqKq_83tO_G$@_(sCl_YW@taXFvpYQ7TmCs=4DeZi{K@EEmuI&vD(TTXwpk
z0WfrIH<2l<+1VoS<&<<u>y`am!BtKWqs3x9KQtgXX4JIxf-oQaWBPV9^=bTL_D>NE
z+q1zsj&@rL@>hVsh&}+KtCgSh{#|bWw+)&Bp(FB@hykQT(s^B#Ph508P~^1f$g2Rv
z!6mZnovviF4k3dLZBM7^Mr=YtsGAEXco?*1I>ssjPOOk@=FO184>Hc#KsvT$<kTzi
z$Ir<z*VDU-pg`x|?I{)!)(089^5d3S5f%*Stj#mK2%%-qA!dy`)Is_>06pZPtHZn&
zMt{GZ=pg5R%xJ>~@;A^W?SM#u=z**;6S7>MD`PS7Zn7FK*D)E$5Cd}gH{*+CP-214
zcl-K2r(QA6-_{b=AE)^&Ql?1|-2hD{dholoo**!(2(SMBw=5yqs)KW$k*=7xW|9V6
zoHjyXlg)_ezbfPZOB(#&LH~E`-@nxRe*yJ$NOK=Nm!j+TX>%woR)2Qdw{7e<Z%0}b
z5-XJ4fl!th^QiSKD;(2?{=MyRY>Ms^ph}4s@b{U$T%YKDYqs?3zE{5=N?HIyQ*wCE
zOq}UI>CNTY`u*(UYw?fUj^?JS5=R8fUH>G@kG@!jTP;n94iIq?NeKE4xw2!Js4F{`
z?*M6QRU_aQkn1{C3z~{shKF@St2HyA8_TA!06MbFk*eD^t``=!)Bbmr%m*ktH?+l!
zfCQhB8c;5mX1`^yc6nmf)hNZq+<DH)20A+Zb;ojQT`pIPg*oQbrl`_=%k;P5`BpHk
zh3FHOjyB$*o6w6K7#SwP6^v<NR8~~LpyqS?Gw3s|sjMc(d=%fE%Dbtz$cTZL74=ep
zzvf`@*!v7R{$QBB8CW~}KyWzP{-v?k06NOUZM{@Ix#E!1xutH;U|Nz}i{&^2DS>~z
z*Xo6@U}&YN==`NI8Jna)!RAWaqxAi}x1syD1nBCBq<KMm^&}71rb72=x?uQeyA}Xq
z^M#uf5nw&5MeEk~z?ALTH#p5Wbn%F2l4jH9edhIm5>I!#Y0e_w*@8VJM)Q{3)MQqQ
z1WxT~{oP$HP=0w~PJ&$2zmss`HcEL8dEa1TreChEv@HZ{<Qc0_uG89NwNGu&AI@$8
zGqG^6OarQN6!wKY5u*RQ>l!PEoaNnjUg)G#`|E%3TDWYx#Q)>yrG*9^+*5g9;yaXa
z*nR+W#pU&ZpB{CgJTURH4A39{bk)^Y8rHqB_g@Hzg&Hzcd~^Ykl&I3-ZZDV(ZhE_*
zdGwdxpL-BJuBW0%2m1ct^<OT5^XN~cPOpjb8Ko+@qS>jb`1=g{U4tN`{dm%fjIc30
z(=6!+-nR+e=5T<NPb;*b0@Yzoe~fb3i9URv>z<mt(>b{*3oP$)#JiQhGw_}SvlID_
z^XCbQ*?9K;R65Sw?(J+598&Q<rmpIqr<U%NO-VsoU2JzHNr3ME_|;*B{a$Ps<j#$i
zHkdn5QsFp#P7`?RGN(;iF-Iq-g^&{OhQn}^zNRSKifp}>ZU*rD%QmK@c5EdLz7{xp
zmo5^tq>iA!C<g-E2r`-Uy$Hm16|J7eSQ&TlzPG^?t#0Dn@EPeGyZQSArwlsQV>?Tf
zq)JG<nnE`>%(<Mi1|lgo;43Qd2+WFp+$2rI`6JK5B_j9!Sp8y|Go^Xa1FF3Ra&B*7
z15W!ovI(c9Y{6+i@LCN-dHR^Nn^$Z?bcMLT$o^lKUb17tzSi?$l8^rEi#59X+E8oC
zfbIcAT(KAgD<l3y^;+i}t*(Edzv4T8B4vDt&pS8b=c3;g<Bu6Y4o?7jB(5v@ySCbd
z*Zv)}g){z3h<^#OfxB!P^8Y`GMu7<%nky;s>WTXrMMt*eUP%PB5$Eg&g6xb~W1G(M
zZ`;f83mfWRS_vszz7rwN0SSxabdVK}+;ogQAli=)u0F&Q_Fo!5Dz%NlX&>28U9ktz
zcTOuNG8UfOGOnJQ3>(R*5RjF*3^d*AM(;NCBf57(QR(cOV2Cs`Qa(Mi<$Ja4Vcm%P
z#F)JtxynwZigqLg{cyPs{#bmW9-`g&feC-`<?@!@AA7W8Bc1e&Q5hbvRzvjS)OG%I
z?M6`0B(>$uuuw{?lETr%l%j3?acfEsh?ei+pdp|$1CNkj+*+TdNZTBLFtv`y($c{3
zyp)u8?3=forKi?na;`)M3WnuXoSDBA^S=3+TgX)2<?3gn-QJ&+uE(wY*oy_N=S7UA
zr8#||TzMGXf3tcSsAHP;mXknxK2mF*)TG@0O-wQe{5?jU--@0B&B$9~bQ0bT|0pmo
z5h(x5GXK|LJ$47$(zR^VnTC7Tu4-!X=UFcT+>uI^Pgo&g7Q4`wf8644@ZT*HfVkKO
zw!sU^+ok&DWc^7(-@V7^Ntn-37d9|NQv=92mwIL+%gy_LzGcWv<Z7?_FTVpt^I2yt
zSvaN-Ij4C%eXwnHwl11;;&F0}>gZabS9tq2(7VMP^b69w5fw>#e2`N3wNEF4#VpP9
zInO^Bg>$01ohFG<(@*loM%Op;;Savq_UM0c07QC)T6Hmalv3ea7~avYc)s=@Y{gP^
zR%nSDWbhy>#QDZ$wucTE|D<0d$O;3#3XT-){A6>ZV?XEDXFm^LlnjBtv>YxYn{_Xr
zWtM{vKyu@ahP72YR~<Q%XvIMjcPWgl$H!)<qjEttkDvR{Fp>3_4nk0pX8yFi2@6zb
zg|Ah4n){Ong*5+7nA=lRI{P+55KK$gzTd`tc&Ip-K4mCobq`l@aGXYNvf(uv<nCiR
zx+CO_O<)_g;`ht>1oFxuuFQff`WlS}P_X9n#}2+0>w#BoBE(7}Yx}J0lIO}t;2*&f
z1$D)F0SUKcF^PrI)6uO47TfsumT2FHiE!-050|_7Ja@J9=Jr#KzIW#NpR&e=kQP9>
z+Yqi-(n=x>95<Hwk#^b=(p1mWkz$j{V8Qu&ry|7Elv{q%YEvQ=g*OruIov?+lTDTm
z0VM_uhb5=}7bW0eE8%zazX&^curzuefA9tOe=w6qJ1N^M{Bppb*LO@c<jkw#h=EI^
z4Mp35)=$OM7zOXjtQFEu3NX-D)^^+m*0*%v2;$PF!2fxMXN<j>IseJAl~2_-(v|HI
zcjVoEdE%S(4L!%1Cl*$cgZx~FG&<t`B<OL3da;qj(lv?`i+p{k_Mb}`0z^3Om@2Mk
zFheZ!V9It_v&GzofEPS6TqEDlW5HY7mo_E*wCNQ8c5L)hTPuW2uO?NmCnstOT@3y^
z=09{Zp<%sn0b`#lLlRe0Z~hMni#wgC=w6T$J3VpcpU)h4%L*0UBfKv9c*Xt61!62-
zo@H+oPbIb8cQXW>I=vwATWXN5lJ$HXzu4HktJnqg4~+Sb8|j&9cvmxDKC}aWMr>`5
z#XqRE7b4(owBDeQmpl}xKKhv|!u!TAWBnhCgNn$Np;>}H7Z<OX{P`%A<EVtOD}6`L
zd5{VEN~Qq6c_HARSbQeJWE(R5VE_cj88^LZV!&X$!mp_3e?4FFFJ838$^R1KUqbw!
z%Rd$X#9EwC(DxVY=~E&96V}HFvEff>4pTm=FySgUnDC;RqiwV`PWT__PMk1WhO5bn
zdcSXV0(q9UoYvZ!h<{%g-a_dPOW65GCZvUH;$B$?1>7t#D(!>aC}dLmvQfws`VV-H
z{$6UCJ-NBmQ~Q_}d>Hy)KtMY1#w)b+0KOLN`uDmD#ScDob}80>H01V{VcN{E#A2IY
z%(hy6FBkRfAcu^5rk~f%ppr6iytLx(2|?(xl3dXT%ISIK%H;m`Pm*hd0=q`cAmwoj
zwDK9PjD5q$loz*Y#7ItOyUuT-n_qw+%7+9P)Qeuu1b6xY8*YMq19Y%&_ogGh2dU$V
z1H+PGN_5O5Id~H@VM2Bu-^?@=fQX|h9vb4=lG^)c?ZZRk^wn-ZzzOQ)<P!Du`cK0g
zhrCZLQLS~W#UxFS*l4G~o`_pFw(svAq~zt(z1{Ou%7)5^qBiUI>_8J{!=7c$L`w4p
zbQs%~&7vs@{kxFskwoMZx<?tmx#3+vqRMGle_zegqKC|SY{W?3f1&m3<k39ynB*7F
zE4optGAmwmeC0=+65k$FhJ<z*Cq&5t8_@*)UWNxzjp?K3GakOH{sRVdxn%Yb6z1*0
zyrIbR)<9~~D+KPd1D(GrlSC!X{V){0&s)D#m40(e6<8eEwHetIGWNcQQ&Fq^SKJo1
zk-{f59Q``CutNFN$g;ZoQ}cSQz($lJr-Cf>6hoBTTt49UA>(2w@k`S-Ozm<`hegmJ
z;Sub|PPQ<_-rY#y@pFXE?Vo?R=5*8w<G~B|B0o*Gq27smq^?x^8eiT<v!8OtP<oY9
z48wHjNLc1nYrfr@3|31n7sptXHp*LW(^ar)wuw^l=4Uv%6Sy#Sbk}pC*9Wyi!!S%e
zS<pVH!?=+MGZ$bWhFq0~eTXdU|6Y%ElCXb~g|n=zjdgTmEFO%}Ze`kjYpnk*N(RO|
z@T`vp?$T>TF9TItcl}-crkhWp#*Rg<#Cxaonr|alt*hzQqui5=dtXCaPiiD53_hV*
z;K~VtPn(_n^Jz`JdavGo%$@nfp;OgTW^j(<rC^m(O{16)_e&vrid>TR^~|YmpK=yb
z=57)B^v_tE;U0<ONe<WFExm1G*~o|$%I|QfR_LUB@s+T5P=ShANu9n`iP0MNnL1i9
z1AiUU1`hFxXn3wTHGc7kAXG$*j?wlWLV9|#<UH~7cp%R-@#)$^-Hax)w$w)IK11ra
zf;dUHQN70j^p{vO7HfD!xU^6F{)&?73Y;1}<$<YJQ_|?$(dl>T2I{d6T)uD*B!xlW
z5^ySd44NE>7oe6;Nmo0g|C&@d+}+66dlcC!-@c|nYbrWLses|g8L`9`4v+dTqGnO+
zcSNm?qWmEC!IsMX8F_($HF{O$TnA@mD32OAYI=Pls~4TV9Ryt7{jeV{=fX<IJRXMe
zKUevJ66K#7+2z8G?$8o4i-oPULz`i$^{BG5t^;~F^xB?uU?M%>0=C#`nLUX3r97%C
zM5$_F<+@FNluc)W63}~ni`m~;le+%W&?`IU>uyh%qaqUsVRsq2;=SdYj<?x)f(9}k
zRpn)8<zrf8;>pC_kxMtnL5WMEk&>PLFwFiQ7+JSI@38s{ddmTn9Gxh!P--F10<k>t
zLp}>La~akv8x#q8BLE(c+ml#Xa_da90iuas0ugmV`%&d?KsmrBGL$02FDWRV*$n4+
z-U^2$uxPrjxZzAo#!t}$_^krrmyF6<H{J|imGHQeRDBNDc<U)=QoImS=5c=-J$yhk
z$WZLkx|<JF5oNt6d#B|_cgrnhtrXLxccV6tj@cdcgwUWPN9*|xtKFQ$n^)a<lmcK4
zbaQXpkAZ@`UtG|W@qF3P`MtR`#-$Seu!%p}kJtXfL+Az!vRS(lX24EiMLoO~i~5VV
zr_lhxX|hkUAG&~;6%$cQTMIoXIyDxnennY@I?YlQN?owB{yef(^%7<<h8VzuLpmkk
z<L&O8k;Q^c6Cq9_J2JfHsDFdeHa7@!6?nS`uu^(!tHLr<B)!dnYpXCP<qJqnW8_pl
z-fD-}i=45e5UAn4!;q;w4<8;sW9)j%UI=c<l}}`<#Z&K@G(8h7{EJ%UiL`U;YBxw^
zjtUaJSOH-@jjfwL%Gha}cyBb`ZmLRSgwI~IabEAp&k}5N#`Pz`lH)t^^Y5w$3yUeJ
z4qZsPZ<MWO>x#5qwDsDGKue;mwpC#7p?Y1OGRCdKVmcIIw`<~|<>L$(zb;xPeWm=;
z$lmn5NZ}k&4!6~a_<^+8q^~-k*UgxlKij5`pZHbS{agwO$DJJkyzP9z=8KNrqY+aV
zaL~j{2xZhF6t3XLg2yM<IxbVG&C^w`+<0+O#g`X$Zy&G+@L!F#9)?na(?7CCSE)oV
ziHehFC!|FbGVyGPtno-yWeLp1^n21(iyuLI`$697v*IF928Z7!u(+j9{hJu;)vb_|
zM@TefAl?seC&<?Oy@B069l6KzUOo=`rcD*#_{l*!5VI+Th0LKGGwqeB0hhv*yQK$A
z7&`eR7bFQo6nD$qZri?BaH+sG-PhF&zsMs6AEx(LnA)7MF#>W^$$a>11?v$lpk<3a
zatB-(yR2$wQq=8I-gGlg_!LAlBG<N3{w6ANtHb5Bt1oQOPC$?;>fHXH^EzQIeO=RQ
zJ?E=mdjsn=4`p0Mtp+@RGg~}RWPy`bSE+WZ%y$q(gqv9gJ}GD3N)vz_qSYt#cNAGx
zw%v4ALkPVYIKqWIyQ#bx(Z{O}2HzP|c~||0J}Z70!z4V$`fFU;-yq7#?${zuNa)Y<
zF^g1l4lS_4lNB*~By}1xP9ARSNa1DK(1-K_2A$!HZw_lgBEnX<>5u6QD_Hel#q-<s
zbDj6T0#59k6(EkldRy0Ga30N-zFTGp&W}hnW80T4jm&pFKt9TuVC3fx6mQl|6dPM#
zpmi#*$CeO6PoWNZLul(1<s}MSpr2qtZ+p$L8H~6iD=EW?EZIOlvSV5;m2hgXcvoyG
z&92h^X#x>cRiT6YtS>cjoPFZ&p_F4-aMpW8MDIqTFxR3_Nx(?6<1y6MY7F<bi6M@?
z>d;W$<50#86G1N=DQ1U6$Yi~%hQDqz;X%)6Agm4De{ReNM6c-<Xv-EbHHV<Md-p}R
zO*3rmqFWlU68-c!aq<Cx=IgMpw}cZ3t)il(lO{M=gs4yhPM;6WrI}|OD}vf(b%SN1
z=9St!q^=Is$+J<!*wxE~oO)jP3P^OH{49c94n97~zNkf+t$ge97Z&HQkL&aYRpq`H
z^>6v@Bo%;H>24&|Z(a?Oe(QC*N`jy*We-Yar7P46>)g*&Prw{%pZB4abhA(NC$gyw
zH8uTyN*T%ZvJ`kN8>~@*O#y2hM2UvOgz{p`lc%6>18843+ea^Y=blC`ra@_%Rwet9
zt;F0V<dojyD+eJio25Is<oBfM2lwPHt4Q4}Puk$O?CvE9CDHCs@^qUSl(Xkf7>?VM
z1x(*7?AM)<RhV?99H-8H>DQlhJ$4zZFQJmzJP8#kK7ee^vEj;zat%=DDgJp!J?<+X
z#Ql7_%dX<uc>r%}RrJ8b1Ir~M8;pRqS%T-3Qy-&LhBZ0ab!U$)aHWeNJwwOm9F!nc
z>OuKUaK3GpORdLt32LniQL>FV2Rc2+kJHlV_^Rwwfty07o4h!!1}WL!qJi+4sc3<*
z2rA81X2WUU*l9_Qq=ic;6;J<0r$Jc(T_L4~!}Tm)V=t)K3b$b6+v;ESJR&^n<r8Im
zZkqV+BFc!uY)W6@l1jBI&#LWDFU;OXn2kO9)vy;$<B@qMt^{}R8`5&hEtnmsXR8Y=
z_|XNX@8d6}PGSV!E=D}0PJfwKz|y9j36rJ4rt4V0tDlyy4DrrJIZs7<)9rl?BT>Oy
zBpOImJ40)M(N46EG5#uloO4&mF7^5vaaj5`mSt)rebR{!$?fEgpq|tmi(BHB<HnaY
zD!^}lnDCU2yImh8qf08frDp#~^@cfx7y8(X->0isw36E<-*w_hI?Q4`!Ezu2m{=1L
zNR-8Qd!i1%^ASgwZevYs!Z5SPPr5Ljg>w~`s+5&3>_-ao%sz<$ts6vxO?XdS?fHVR
zHaX&fRcif$vGJWQ$!=4@Kc4;T<#sJVLqDt(UtzH$9V>s}eV_V6L0Bty>00{Gfny6Y
z&6DFU$DGyk@-ACj${Wg*P9I5xEhT4_GL}aM4hMO!eZ5)WoY2DZt0Z`E*cH(U?ZE6u
znz_Co#D+5yUsYl7?QRX(PI}gln9GJot&}gUxKw3EK!wpkGl>#wHcHm(N(#h%_~Jej
zKTTm7KXM!C<4f&4=_LwRkgjf$V0$yY`Kz9Hx-*;fVNhmN;Sw|=xHt`4_0j5E`rBUI
z^R`6FGA_And|gLRK=wvkV)Z*y36)f^zRs-C>XT>qwLXO<@w~R=qEBV#M%c7L7As~F
zl>wK4#I6e1TMp6T!~)@8h0FKDD-<+2S@RxPvrK~gv(i#JMD-4!J?{p1*rqX!BUN>=
zv!FNycNW@oQz83!HYS25*U-|3$G18KjUuXde=vg;P?twL<}C*E0#DJF-`HkJsoGB3
zB>8^N@(c$DqOhgYST)qCzHIY*ns>*aUINq%P1$E5ZIfae2bm_qYrKf;_!W9`YYREk
zeKkCrM$Ql5L&odSnv<*OC~{5zNu;11t%7>{d=;$lU?71#-Rs%9f*_LvM!9G9%M`ri
zP6Mp-?s%*k6>~_Sc??4yepy5JRD2R(w7mFlXMK{!b@PrD5LV;Etu#h>o$zF@wrk2a
zCvy`zB9=|t|9Yr0d_s5~`*Nwf8H$w~{J2QX6t4si+MpHSssn@yb!0#T=UVKknIY8a
zqNs!W1HHAjYFbZ$&t#oOuG^Jc&XStjqk<TYB4U`3J^PhsgO5FfxA?Lm3#N8?*HuWY
zt(-)z3ytSjso2?P*F&It%G$mr7pq^!b_E7VP<DJdU*Om{3?JZ~Rn<@n@wxz#$X}6v
z<{gc)O}$6c|A1n?G>+m;2Pc=MCWXILf7y_Lu@00OScndWweBk`A9)}e1R*ul_(8Zf
z96pD+8$nEvbRm?tSrCgnAF?hMpBLn}>a$rAeJ1`~J@%bQ8?%`XklWF!cEZE$0KE)Y
zR<KICVC@#ugrvwMte3GUA~>^kYH6~yCucywI>T(gDxI)`)^o}C;g?2A$fZdnMtn}|
z`l({;&l8qOnMRm}j^qk>R&!g3kwto8fawj|Gl6OXV{gUIiIe6n^U3!yQQS%Q2}j%g
z;wjKFTE-2;m`z&1hx5<iFY`DdprR8%OB&i8_M{>CZnhX=_Dz1}aS!ML^k3B6Ma^(8
zx^PeK)AvU(UsHalW3_5`mNW{&8laxwmv$qE?{?6vc@lPga(3=XQ>2us;F7~FEfB5N
z+R7-txL}kHy;3LK4+4vwNpM1_A1~-iq~<A#??K*=#VZ%WqyAjq=RSDw;iRNd6Tm@r
z;5Nt?v#IvI2||-uNAmWts!Pa{SCdidOcMfH5V#Gm>I2^=Iq|r-d#Rr8y^NjoWt}99
z)pbVD=n?eCs0%eUYdyUlNFxsPt9CKP;{J&v2p*fXDR0=^X)8elg9AN)-QaY7msd?J
z#PGgKaF5!Y-q%u-9mafnn2BN6#3Dj6FlVbymygV&n3digM5C6T-Vr8YJ`%3+zWnQQ
zy_6h;R#@Foc0b|@O3Igqs!SOY%g-iSjXOY4G8AG2e4z9hKrh|RJMQ~CjaRrD!Lief
z%2p$NdK34Fa*XRm4M!H|!8CRqAaS0<d{~YmhC2aTw6axp#Rt8quNBnvp*QTdMtT?A
z;hN}|dV}ehOateM26f@a=%vw8OH|>x!q1O#c<nu>Hx8K1W}P;r*zuW27(hc&lC<>p
zI-1@<gdvJL3HC0?FB9K(uIAh^+LQOR4{;+)A3gDjV{o8CS6L?+AJdzdUysa#8PR^E
z3s`JT?5-HMm}w;0YlfhCX+}2EHIp}!MwO9PpCzeP1x%E#B7*1N!}=bvqEyOtZkK%B
zsKe&--LAv74R;6kvN=rHl@HF&rqW>ZxM+CUBe;+XA%b-tdf7XDG-wCl^rJL*aCo-=
zV_DyRB)C&XBTEn+dap}RY9UhAcoNqhl_ClFmy8G>xt(B?YDOiDmU)itPgjFb@b6RC
zUU{_LibApUuSk#Ua%2SP+6Q$^?xWacsj|bp=58Y1!2)Q(`Rij>Kpa(<6*{rhiUkR}
z^WQ{EZ^)eN{_8bu#WUD7NapyqW4QXBbXymn0S@`XNK}su<2Em5>@|Zw>noBjNcATZ
z1pLd$!|+kW0Pbw?6~&;@#QXJH_C!Oj)zNe?%zlXt3|z@I21k3&Vl_y)1Kz{QWWB0I
zc%{G91aF>7kQZteE3b0=eZ4i4{`g?kdglt?I&VE+Ut#lLmqPs}rRv6ZJxz+BE2(?1
zQ5KmJ>}CH()5#{E(;o_gx9{%R@NAxPCQ5(4^g~t3;Z@PMK9JF!#gAdICGg=z2MTlQ
z&@kK%LA8PQWUM-^_^DiId)O7-Ln6r(wn5?5<PUSrv)SE3;BH0YY3y}O$kp$wP!T+A
zG$LjlN?u7!k`km2QxrN4!i#f+x1BuV`^DMOhVokY+SG8T1RFjekWLm8A4S*@Ymlao
z54oPU9;-_C<R@~ZQfuEFJ1eZ%I--pCLZ`y46%Va0ui>RqV^okcd(r4-uh6ikLX*3`
z3L)36)>a;==2%OiN!~QE-77uNdX_Ce+SI3C_o^ffVS`y*k}76ghd`?yP{`>zO~N!u
zrlQ;7oYDSs#K?h~1yLeyiEjLyA`nR~q<z6$G;cDs`2`s^3cs4)%6I$8!I4`h>XlkU
z+FO<fKW}@4qxIM?&B5*I0hD!cr6NtFwilGR*8NSloY=2K6JH1j)?Zk);f5rx)Vma0
zM7TVp%AwBsuQ!-E^<roCY#f0?j>RuWCY0v^U35;$^PlNi_9*jw5!meR^SsJs>SsVi
zvJK-#cJaDVW!p=^5Dl3Q6FhyOG(#3$5URbB)(j;4zOE&DtvlQtC;RE>Fm44>jQS>)
zz!~CctLWgDvqI~Yscia`JTkAAknaWmR!^$f!gmc3qgy|hp}!sv5c{7+T<MnS5In5b
z8GlV}gMCy48s2x1!Tk!)qM$J?xl?ZTh}H@ik!$%N#X+~?+uqHP12FCq2c)VmW2QJM
zr?IT-BW6O#nu>)g)Ci26M|(>=+2b9{mJZyqK~PD3>xYTLpW)g0^dh*;>iR6*O&vE(
zmiB!)I~`mw*1?7*(+7iRRJ`|jx2Lfh+Q_U}VsLo?<=P@S>S<q}wfk=H=HOJ0iJ1z5
zHMQqM3~;94l7w+9>hj79U3Qs{st&jr)uny}7fq<doF(=wrujE}t$iwqTc(34b+PVO
z^*gMke<%${Mt&I}EN4jDf4@yypBP=$dS1j=EOhC_-fh#HouZgH+rR@Dyk^U_ATDl}
zVZz%dHU&7W(<%mp#w)Gf<Ga36%3-w62KOZ<puBwJHhdPU`8wt^tY}Pa{*e;gU^w6Q
zu$w?G`gq{E(b$iOYLgDvqd!90P1smq2Ql$!a^g))kwDXCg!!XQfEnb?ydQ)uPOH)v
z+M0!4Yg4jwB6`D!byy$$tuMlIhgYk*AJ{e_OZoT~ed##2%O3slLU(F<_42Bnl!R*0
zMnnBpPk}DOq#|J;n95XIY11M7cJoPb5h&>W9K!gaI!;O2xA^*C!s|7Q{pf%%p8M%o
zco;)d&kK4@WPBW#zDcbcMF6xPVzgcF<HQVspB^B@rjQ<c1zTfxB7r!=tqg{dEMT!p
zTre;DZiN10h#p-?Chc^+S$}`IkYV!HvC~*bQwX>y3H*^^!d`Zq2?OlkRwjNRtgdFF
z_*Kw4dSJ>6uR#Bv2<e%Y+3XK*o2j`)F`$&Ag5%qYJN$Zi!>5l;*6oETOvT9x{d@iM
z3(|sbNF`><t@Y_#6D8fPQ459YO77QIX5)pI4F9U3d-38wy!bJ?OuOK3r<5<uGvTZy
zQKupv_3zFZ>1&52*>d~_Tl#^lY`XP%g^=mY=>fi<P(e-SYtpNdr=jYFH|KIF7v$^!
z3uXG!359XhL0^i3*P9&uTd~eljp5@v*as>Tt6vqaFN2WDvxmJgSv@jDR+ldc)&<Gd
zjZ@w=j)Px*>w=5pMq6rz8^m%Hlo3(n9+IbsKt^e3WuMLs)VzXV#e)W!%0~RuB}_^l
zY9@h0%hmldq-#1Edg6<8=b>%?yVO)^bs!x&ePqckiw_o{FYLV$xA9@7m=rc(+d$l*
zKVVz6Bu|X)7q>5a+I7LlZ(HA9!R-CjuV=IfTL0RduFxPa+u6k&X+@FL(4@{<J&ESz
z%BY9X<n)yTIxvs>Q^u&-@zaDC2>(KYNyHuILUy?w^oF+?Y=Ekzm6FzoI^lXNV7{0?
zc?#X=+<d-}cvpplqwbF6Wq{Zd;=$OrrKh4g33_i&aXspv#Z~9tvT`i?iAl%lwyd99
zFCV1MyWM(P&b#1A@J`5F@^~G*7ynXusVp)jwc{Xs8Ly$|aT!w8<4KW-Bu<~>C<uX5
z>oHpLEBTv^bMA>Sj3GS}aKAdL=a3+=l<p<Vb4>IzE}#PhWY5!PC&FG<r5$|j3O5-E
zOh5PA+KwRQ8GaW#R^-w!QNl~ULC^o48HE%V8xnXxl^s=lYglo72mCTiG~;ATb^G#(
zz<awEWyV7e;%(OSZ&VJ$P1gfCCUUG#f3mr)P2tWT-$0jF>?^8U`d(z#5W-QXTJ_C`
zWeG#O@N?%=zcM(4Zf77<cTChfGqqLi$Ax%({Z=g!ux8)j`1F<kyJPCe`UZb`up@XP
zQO&d>^cK8Tc8x1lip98PVBW0lK(cw?p&CKMJ(L??FLO)?6<AU2E?2C2@K#d<{2(ia
z_Jo&XI3K!SOvBDAoNoz6S4_wR7$21S?aq}Et)zDiAZqbk)CE}>x8FG{$EQX_DSmX7
z&VUTK)?KLg6AHDSs6z>O_qlK`wUGTj(1kk<6h(|IAcRDmN$g1(yeEL;g7g55i|8gR
zf3thi<N0vqArIZU!QW5FIFTyiOm0S*E}S4oy&gSL`P7<`j$#*fvTNC6gTK+g#nd}_
z@zA0-vT#<>|K7*xbJ4)eCiKl?{T6V*nzXixL}c}S3EPIeg^dqN|8|Vy2x@M8gYLQ$
zD>BJJw9}3cdCyL}StT;<>M%CU%EjUU`r?||-k{l5nJ?1DtY|tqu8OQ^_6g@ehgMa*
z5VS-1=XkyiES!dob4J=Hge^ow_A;P(n$@@ZgBj61xV=he-dod|lbP?03_WT#>9(M^
zYQS0#*3%hNO^sA4f>}ch<fOMPfR=7fm3mC}yo&Av(akT90|ac@M^nM>@OrFp>O8n5
zIExs^OP*a77*<S{Z*Y&&n$1wX_A}Abl6D3VlxQave)m6y#M(i<4VWhqMYtRx_QYrJ
z!aK}@Gwh+<Wt}WURk(*9Tv*?&gxR=l<4mj~sILfXF^>=<Ry0?4<O2nASefnyv%7aD
zwujYUf(~EIzWy7t9|P7eYvbp%NtQ|#&C-*vccJ}M8sgxFh}Hod+>WGfNVA789}{Ww
zgY6Kwp@XwsT5gLa_rT{tU4`r>3|>B9{89k}@fUKxk8j%TGTpV@5N+$y$6g%~YQ{T%
zDb{03NMRRok6ZUwKCVPAb!qV-H(kQH4VMr?CIzm9c&RVM1q3gOMp|_We+>wo_9EH1
zt%Z2aflLY`)~wupsxKk3=xD#K#0J`5BU;D}n6!zi<tW^2D29zIe4ZM1P3ni9-^KUG
zni^WnffgXNl~yNSr3vIMzQQq(b><^HapHaX_lt8bU4`FUQ5ZJyHVjlGx~0{OZe5?X
z`mc3K#7JLZo?sVDO@^7RL=K1v3>PO~NM@6`zQFh!TkM*c8NgPdOqB&X%!iT<dH}CR
zT5GR=u}~i;#rbXjqDBe6>*VcZ6^7yg71t@|$On@4Ol=>*;F0*jp0m*)?*XOY(2WSd
z$*JeA9YWbwD|pnZ66`wFr9tNng^=f!jzy}!F~%E2byZ0NQ_K9j&mBozYN1i{_52aM
zey8Yd>)UC<mbAYSt$jM+lReeyyum6x$36ah=D*TqY6d@ADhWvrQ`{3J{F?}!hOoPx
zbpj)T=LSRjF2t6K&^z#dNx+T_fA(kUtVSe|SEc_s>Whgw&~N?Rpmp|04r0a0K3*hC
z5Kr+*M9H;nS79%t%knAl;iqCz2snsbeo6TLA6x-K$|~$z^SfO|0r(|;JH`%#kof8E
zQQtS4!+$od$_b)X$fbvv{^(aViUI`%B+g}vZG1^UH(MrGu6)*PqksaPnCvS!Ak$hv
zFQD+~Ss0nRGFi^PH~&U;#nH8x_pY||b}93Xb}5W8WPHZT!}Z8UUwcTeoh}ox;Jeg?
zvzuQg>VoI`tw3n@$wnIN{WAOVYe|h#y7VFpY0H>E!GZj5WcuXoR!;+UDg%)8uO|6!
zd^O2>A1EmI?8pQ3neCR_h&%lA2OQ+(IM6*RBe=W!|NK&gE+cW|qvg}Pe>7mw+o{z%
z3??tBzZQg~{MNEDX1Tab-Ag*ymC&LUE?MPRSI`t{X)(rX(7eFjU$<2Wfb!w(E?7NB
zc#+{q(Q)oen83PazAI_Te817QBKf0j1&7GH^%3)AZ_mP1XX)P-)uWM~AbL-(Q+G9}
z80By+62_Z4B*`|r7r$`5$0bb1$=~3kA%>C_GfEqm&T*s+sOfvV2h9ED)$X_`oD8<a
zE<`q%#Ed!zGBw8zB+tpYL{PL2a4*gd23vLTp2(jmmgTk?S}+-T$FP~nU*>R_iMUf;
z7~bF)E=w5^ULllsf4E0jhEhrb-3RXt4?X{Ci!8mE5tcmXx&|4!)V!d|oHcyvM@~1U
zq_m;cbL`1f(nH7Zl6sQNUUR-(N)MNJn~$q)6i@%i+H~!j;HFwW4d=}SkG!_bS9BMR
zd|+mI3h5`EPB`N6X823II%Qw;!q29L2l4HPJgPkNFVNtmj>5KmYu%OITp6;(QMXLp
z{eEUIHJIGa_Ueof*5V&LR_$d5yqPc5LnY}21>1cz8Ae*&aE!aZ4(_8a*W-jPV+c6T
z$mOCHKVtU8!UNgt+s(o(`8d0d;enu9T0ZbIt#`f6yW8!og;}?B|1#CR9?J`CJH%br
zK5wJX_n`;abY<N0qk>yGD4PB=J|?>~jQ}rKsN)N>9Yl;E#r{Kg**=KI);F}+Tl%lw
z$SE%fZIIC!TSX-CS56qh(f#zgfE(Dx@KZC4{yuYOsGI~-Y|t^lKd&KN52WAL;thS_
z+YP}pAYW}IBkw;5C9WiYO8Q>5k{GXkqSXN5Dov4rFH;j5m_4g32ds>4;I@3$gayan
zfZ)=!tO}W9NMUuDFH58|8=_U<Xna=T%U1Foz~Sg-=Mu<4o#1LY4qnHM2@c`9`tc*$
z_IPtP{T}ZJ<;t4inlq-#mkpz6{6`bR9GNm2PA>72OJRHcGJ0oqhVK(k->?RuEwx-1
zc@ui~{5<u&PRplk@b@;TDAUa5l{~XH7fHyJ_1w7X36FdUqpBxgb<~a0^P=@gh~;@)
z?MS7yjEYGgodCU47Hw^+$cN4THKO?epr1QLN6>ZH<0zwi!GdaNhku&`gpt{XyR4qp
z5J8p*z9rT797;l1AGj<m?aggtv;R3U@xpLEtuUAp2u!%%f5@Q+p!I(wS@p5esmXUo
zR#xT%eBq23=rTw!%PoG0(Vy<5*-Tu+Usa!BTsJ-$6xJYV;$y=vW4pi(U?H3wuJ8B%
zJRK#CxO*+p?klERbU{}bZFH9@2(lvWx9Wbrhp0-&Lu>ej_LWsM&FToC<=7F{pr5yx
z)uZxwIzHKV_JP0{&sg9BM<nc=%o_|MxNyf8SaX{FeNd3a4Z~<`tUMC;Ozp0~=bseF
zm^?$_IfCsG$>+5flvsEInz%-Ab$s8Be+C|uc%*7$w?dEKYYKwrz#r5BezqSq43kUg
zt%K0B<pwvdyAPZdp`D~VP5v$HV{M7D>4Cj0#Gz~J4v`UA)73?wstY5W@O4}waAyCH
zG`}&5$?>xLLW^|4*DvHxDpmA!JR@3VF~7*`Cee^!eLQg)G^5g&oYbvVl2bxzXC~g0
zdy}jj-zgEWcy9gn7Om^~BH`?Y=qr&eNng!y-_DLW)dOAwob8K$$teZEBYB(Bm)P+U
z`gf}<vKtb+d~A9V%KNnB6-P^KT%~Y(*CY^QTTMR7E0|Hb(mu1;Cn8wA5*i8*`WR7#
zx)X8xG9<ClyB&|zAynRm%q{;Pt-WVBoZI(5%osg-ixR!poM_QUi_YkENYjagL3AS|
zNJR7?7`+ocdP4Lz!ypI}y^GFh{~;&mobSuub3I-R7x!L!^}W_-?dx_D>&SvAt8A~Q
z(p%xE^cbB^l7<iJOrP0Ty}Z2lSo$A&EmdXR^|{3TH)q%`-}oM73{+yT$qvih_30)!
z_t$xj(G4se{qo$-XgXjOXYI^0!+N^5aZG9|h(_U`m%1p>p42`Ytf$kPa%>lFuAM$|
zxvcjgh7X?3*IJZz)hb`6Tt*_Q$yWO7I-d;ZE)SY^eGAY5irvf3t^0Gyptcuh?L1-X
zmqo&kkd4*I79zOQM&1>)ee>=wn~MY=#vXanzgeCrB3-#WAWsIn`n%ph6tXWLn?9kf
z|ALM1zgD$n2evULDdw*oPhv*5dO3Ua<xQd|sJE42zI<4Jb6gvV<J-{fJE})ND~)x&
zQMk^7&R{j5LC{a(>A_q#g0KiNFP9U9Yq8w)`?134nqhY9H!!4%g`PuN9@4*G=kz=>
zVX<+>m{8SOfD&Jxd|4x>emtPg?Ob}6v0N_)AA_V2tyH1sAC8rJ8KiVy-vcnAM~uke
zc@N*~x@j{IERRl`E=~iA88LyDnW<gerW3?viOX@o5JuenO!N=6*`33bLuJS7>kGvM
zD_miQJRK*W<`!Qo)IaaqI!I;ND>rM*mXfEth-}02_9X;yINJvYPS*X(wmDUv0_^<0
zOuz+N7N;id6?QXA;&VKa!P#$<3uLkP7Q=^k%(BxH!DDJfLE8jrygLmknAhh1-VC@h
zkojUac5Oa&R;2amdDUNDYNA-^Xat@c*7fsxMCU7NW7Qm*f3|4HU!)lA^0s}pI{k3t
ztsT~W1p?383=gi4A^z%B;C}soX|Es@eZU7h=RtnUw^&tU4u5(lkqAyCaPEMx{|ash
zVRc{0cgJHex!+T!Wf5Nzs;WrQOhc{MF-oPuqo^MPhp$W*21h7U5b|rjv{incfi8rH
zo-|1JL%k26dTYrKj^%<7g_!r*A1)@7jvBiL6vlZ;^LQ)5x7Ub1zy!dSl-h6zGzgvg
zqKE*8oEIOs7Onu3&mr!O8C*71PV%quaHj({@Qj*<6tiAGT(=fd?g6HGfCX3#2Ho1>
zLFhEV{ijdP?m&BNJao4@wK%7PvXux>i9jj`Z^*BM8%69lKyW7V0*(uyZ(f=tcR6pB
z_9J+*@XNGLUfOAjm3-qGfR)7!*ktT|z^xqOmIsQ)q1XiRH~`PDZZZf6T>+c<R3Q&{
zC$9~Gq@cHUv)jP;Bzbq;B@Vbq%uG=!HLEiZ!|+txeL#Q(4`TUWF)l@7TDAt|RW-L7
zINt3|S@tltd3+dVrsv{mf$Ff*3jhy=&eHLx##|96C$aa@IL5&snn7Jypy@QDfAV0O
z$0_c$gca-B3#qwPN&dIKouv8eoTfa18xFT`j>gki-w<>|4v;3$u!6&?4B;;KUBYrc
zD@WpAFC&CnJ;&E&4r(=wrT(Im?x8I`=~w@*V&Ja|BB0ZdlrVTlbrG*1(qn%D3+4`j
zzg5wO596+{EoMmN%WgDI<Rf8rWUxt(GUENm6I6Azjjkykb+5gCE0TiR;eN@m+>YC=
z6<zWIx#cDhGP9DNKulT%0vA~UV$_hv3ie|K6QF=VhFe*};DB){9l$gppavfh2GRk+
zk!)*#b1NRin+Y}7k&9G^H?laeNgS9GoSPM~LhcERbcN#OzO4F$z3d6{Z*?jAa~A-P
z4fyj=0IVgcY+_ILP;Za@(k4>rfBqC%Zp(kswD>tv;9K<~Ajrn^zkkMoMK<Q29kgkA
znC0s%YKg?PTH83KkpoD`38(?$*sy!A%m|ET84V{wh?AA3EWFqm%wdTqmx43^jD^~P
zfJGPxhxIWm@}Y+QqdbJ}<y_ne5dHH9X^ntnqhrir?`rc#!>Z1`jTznjLf$|fF+KBh
zJS4Y--1*OG^giEy<)`Calu|2(3JojbHVov$iohaJhG&_{7MGPE$ohN4V2KM3+iB9g
zo9WE@oh9WpUoGHV=TWO%ut81s5Rfq|Sb!BwN`Q)aPF}#v%wM8fvMxJPx1R!H;dYYs
z`O?NmSNH;@HB%zzG;=P*H*EnoKyza@7W*rAY~keMmHo^AZfpq8`qT<f$w9>X-4GCm
zNo2hyty+cUp{!Hl>vFg)b63AGc3)9o@x<((S9>{ACQDub05Wis^O6CbI+jJdXq%wj
zjW=sSsymGpB^i08XyYQ#>-F)}(k7MOY~SHl;t1NF*OmfiMAe@HUrg9TF?6^C94wg$
zjIaSDR0Q!DcRguQyW`jPyru5wgn+<^x|QB!wSn@SXiLbS&OH3Q@Gbhzs$Gys465!I
zL=0x_YJ*w(Q;=q*D|%Mc$qvN9^9}pjh3^tTk18i#-<W3J_H`VagZPOct%tQuOx0B2
zOun$hZL=e&M2}KaVwoY;n_Uqw#8(4Osyw9(C$i549ew}Ou$xz1!9~mX;&*1Lpjq(_
z_;!C?L(&yF=C<8dCL5RP+iF`5pJ8YeMCjeCDJo!kN4Nu6E9HK_r3RN#gJf!n5JKOa
z!>+<nAC6y+3aM}gCw%_R+&4n#vJI-cq)g)Wq*ZGF_d=J50+8vn_pF^y=P7LqK!0s<
zy0cjtOgrae4IA9wqn&x5-W*mZLIx~)DnSJOAac`9`l}l$)9=Kj3zziasB-+v(lU^1
zS+7BTf-MV>roLY!yO+m~Fjh$tR%kQ15f}EH#?#2~tvDeRCJ0UptPz$2I<dz9EljlN
zqQ;p>i<sa1j%NwLbir$FbFka5iyYLe@eMi-LBy--pUsyV`vqdF%8p#US~+Z$<bDQ_
z8vOhzF<4RdUo#@a)nvIzACS4PKsubUAO+FT`DVq@7Zktwy9W;%$6dmz^?DNq;)=E0
zPu|gZ8YJik6Xsyx8hXR+3rVrkqvLb?*;gJL#!KT;(R5x+tphh)TpV7i1eNF0_F6s{
z|IIAAmW{TU2H%(h@7n1p7-4TwxJywku5QjSe$y<|8y~8WWgd@l4gu5Y+N0hYBLm4s
zF(lP@!-wDAfzgr~zto?!OiKyi(joKaHiz3LtWuuD4eW3d$LA_!YmvcH)Hz{+58t#O
z{G0Ue!Kpl-@qMhp5&9}s`qA_?r)#FcRHka`ndeqv*H57{f~-k0VWy-)Yx9;i^o{9{
zzP?mN+SLvjyn`GK_rGW1i)0svx4ffNXLsEjezoV8`PdvTcy18jOMp^mC#>ddCyy4v
z{f{`p!yUHRmZx{OhnQ*uw<p`uJ)4V*G-=PjXG${xeHZM+%V#u4S`>Q?3|gQbdTBW{
z9toG?ICc~j@@p;9HaeFdcWGxnFayM6)d4h?3;=iIf6m4ViWQi)n^~TJ@7Pw!3M8iF
zrqjE_?$w2zTUtEO!%cAhUTE$tx}3Dj3obJ4tKZo%J%gaNS^dshs+e)4`d8~c-=NyL
z_NhCf7DE&)aLyCgQ*RF8IAI9^1@n?xoPc5vj%%Tm2&*$@GWZ(*J^3J_dG^iEqzuf@
z;oco6-1G8nS+6*6`~#u0x4GF<b}K>+b8y=WY>AwsxpL)2HltaY)_I&+V&J~Io0byc
zK3_A)jpQHathHgM$>XO>QLcG5+sS8|ZZse6h>zJ}SVvWPD$kJulBrtWR<ne9Km>i(
zmuS-~x)`-0(YH@tAVkQ8h=yKhPf;4l-GqUpSix!=tzonNbx^s9Q@PofBse04-i|`E
zlD`<1yx?k*G>f(BV{_6qs^#XX`!C#H@*qklcWgM(uWT`<OAMtfxL{jB!Hd2CfxXx;
z={XeRNP^1gfaR=i)*F$iVnN2?ZJXjCRU7l8Ncn!tV@uweAM<YG(d8gIOK(<eOm$%f
zfTU%Gee;A>W9<<Croo92%xeC=w6(VG&2rJIGFf%_Ww51uv#91)Gt~4OV0v70plru(
z{Ysb!F7UHJTP_Caofw82zw4wp&=mPjKz)#LMC{C+l4r;9am`r@dPnY#y{QVJ2b6^+
zE9E33PM#@W<gAQgAXkmsut=Mz*10p^AWJj5`oHe~4k#X2Emx%j{hf${1U4(26WYM8
zkWAMAlFT)zkQL+8WvQEVGwB{y@RgW!d_WYbE}*&70ItSBIe;uDs1}W(Y!^UNA4bZ*
zD^Ftk_sIVMoJ?O~d<`Uya9Yc_i#y6Vf(%mdS>|dm-LF?oW!h@_PkDfO7Uqy-9J$_H
zx^5e?J-9{gr@GyoS>9%iJk`ULAb-k&ADoD{Dv8;iu9}(EdI7K54)Gg2RznAu{e&!<
zM?!N=k{ZbisK|MNPJnARH&yl%Y-3?6X#rY4p#}Zc$Z<r_WqrKg_WzIsubKp;+Qx3*
ziAeNIYTJylzPJ#bc1CtK6OFH_*$0C*VIVdz6}bcSDdh@@ie;&uqqIKNHsgNyjnyyK
z$hxxo2a(MGRyAKV;FA0hT1)U`><WdeBJE0Vla}ZRgICj@c-Pv;6Ux=k7O#a<{zp%~
zqDWeW<?7giqQjTglq)YyW@LU9TYDFf>cFSQnjh5%U@}5k3tN0~?Lbh^u1=y2rU?F1
z08sG2E~2`Jj+U~XH%fo|&<$*mjVhQMc$6QtVUS!9(%wy!0^%|7yo@*&PQlp{q65_6
z0%`!))!{SB6(N<36Pd!H5kB+HwXKCI$9QP;?xFus&STA>zFFNn*n~zex3i?)Z7Z<-
z&yF+q_9Ik7V#!RKf{d87Q1@09H}cW(VzZ40EE=fJT4?}Xf&WLHp{OVJ8(9-SPxS~;
zv;-)k-=~5G91<=+C&9u1NehsJ9+eS?Q@hfVAB=E)>RD<?p!fh<tox-VIk09sqFP1q
zzY77#dwXNkNd*jz4lmKv)z#d_=N3s11FUC=Y@c)?mJ;|OcM+$_ODRp)WteRTp9b7V
zQ<zNJG~<vqPyzFQer#D94grgn#Gg38n<%ftCtydH+b7EKzzTV`E9R~|d-GoPm!<>S
z@;2$Ex}>@?Hik#Z$U6r@U|ic9Z)2HUht~U)DGK`2X{e;f_X92`y!MC72Ua(MPNNo|
zjbY5Tt&%=p7RX@e4s@ulZy-^XD3}6JgN5lLG~g2|EGZI!AbFIx>hP^IqGJC<iroD)
zIUdQV6`$I(+xpo(#I(AOx?&pCeVSc{HN(SLoezFY!BQF>BwDiSC<<(yu}yIsVhg%n
zDT=p;fX*KpuMIuvQD;Z@?Pn_*1MvT3lP#6)R0-WNLZjxb_%be?7o0uuwQvUKZ~f>L
zRCx1|ql(Vq$3>PDWQ_~MS5a{BXa-$!qk~E7<-Q0)QV^pXegsd1*F3#L=ttiD8=F<H
z#~$1u5>bNFV)}!-gw;IARvpGYC}1<Qq+vj2|NHo4BKNQo1yaoRKszfr%gm~ECVO`-
zqOp)U^(7zpr7hh+Ag=Gn^lRp&U#zGRh$;QDO<#ZA>HJs02Rcl<fT2l;^vjD<_5^HZ
zm#k^#$iEUJ5KAgzGBTmFXLy+@gXDM!NRV<kz7VZBC`j@$Q=npWp0)s;v>JBGX&vAI
z8ALy^vy_$*cX25izaS)t=MV-qJm{v0ntHRSW=Azd1UqE2aGSiVOjltTZ|pN;s`JXO
z`^=cuLCvXj6E!$>r`o@(A>V_y5|A*#V_scR_t?0!!9jSS`iamxrM`}iBupV<f4fx2
zE;kWpA;1)km+q!T(mXC%XP{ayn-PO<-wB|c9?a{Mz^??cfFoZdsmH(sc8MTj2HU^S
z2?~6G5KD12MBA_b&iZ?*V9-EDanNG(gFK=&`*g#IbcvjLEm{X0%4(YC0`63Q*@Zg?
znYIhK{8jA;RZ+kC{n97Jtz&>#ki7=oM{PJG18*buG*55aW>(I(M3t6Ab<v26j=UlD
zRocpAE|n^kvH@%$gU=^+y<uY1#U2ShEBNhQ@)h7V%MpEX$)mS%pi_!w*BR+CKV9Wf
z@oh>^QO<g6T8DrP(5Zp(>g+24b;)l@-i${}a!&}2?7C74C5`Wx88u`78n<8#BQkK(
zNpFQvduOOYPv6wH=Q6hY<&t<Cyxg_cL`VFB3_URxX+G7gCwm}z)GdPScO`ucOH7i#
zXLMl?E_GoAw?0*6E}?TM{#!AAuqOwQ($R1BF`oUHWQ3Jfvs7y9U6hZiQh^v6yfW}!
zGuY5Db^BS47=zyT-YX6lwr=ZmSNLnp7*?62p39it<D9nEY_!GOy3kwoM2MA8tfqPH
zo4=;@=RBL*9F(5wjyQX6uAx9ff6|$m=vsH44qu$Gt!Qasm})AHEKZcHRSe58@qSQ9
z5gz8yP}{(jHa;3pbH3)rK^U2Ob~t`v@}g+^Q-9sVl${epmTfK0n)T>x@j@BPw7mTw
z?}s_-fQ0;x_57nJWwOfVy@}`5wC{i^In`bXfxl42Yjttota^M<uZ3K>eZM80SLnxk
znGMckYbJheEC-M*Pcebu!k=k;iml-s|L~BKeY(D@H~a8b6tg`{bKIp^DDI}^n|K@6
z=}})(A%l*6!I)tejSo8eHybNj)y-vRB&0E(Tj%REes8>Y{4&sjSRK%8VSfIp9Ot5M
z#XUx<RoXcN0_vFxkYcvf+Q*X_Ku?9|WW+4)#$o^VmE$`+>ld$^DUHP62ZN*@iXf&u
zgQnSN&AXU*V5FQDG@pIa{VgNsofBHtj~2O*xRviUQ;<IE--?c$^p#h~0?2cFB+(rl
z;cFk0BaEcrPH&8&_kG4naXDd!MkZw0lNOWd@%-nfcaZDDIrZ*CW$qhw(8EyZ%>#1A
zs^WBHyuJ%^2@4P<>z^#Jdht4d`5!(M8(O!hUt^qB0Tw86)=(>FeYZCq*-C+7NLF4U
zigy$3N33RL^TJ`*UsYQx$FqJ&UXmB$6TCyDzegugl0Fk!L@qSR)0&ljiS&Ba)X`0d
z(sRbvfUgN04x$;AVl~aIqVOW;@aAPcKT2mTEakOU)C)gf;n#1nV1~y#dtkxtF_(3}
zf(DgGDvpNGV^Yf@0FraMm%^CSRRQWl^Q%6oeHQMHyY#hZer_tsnP>88;U`0ixa0R|
zhWtrcAN_MV4Th7%eK($y8Oh^86~N~V5XM99yZua96d(cEX5cM>)7+ok$lV^@^z0Xn
zGazGD&y1S$h<E>SpI=VcqFO2##6f1dK-kb~8U}azas${);Rj=OT{>U4womZvub_4~
znXOjJn2qlCd2`nmTwiG$kYZf??KpAStUZH%En#MfZGhia>{fxCng7X4JVz3AQ&5{|
z9;T*u^wDguZlK+4>{ro*bJbw-{xL8`$27&uynphU_-=j7v;HbbQ6h~|w$+&d8fKpP
zhSBy3Mj;JYrVEVwt<M!vUtpdB@`B<en4<cj2zW<%Q>zM1+g{uy%^(*Dl%A!F)j7y+
zJ>_9<XE^0~nC(8}HevCZ2eA`h*i0lKkZ>7vt5BO`j7n&1*}Z?hQ0);<x8zpTO3Uer
zQtjAxj5~N@mvsk`XuxIM%}(3)SP<SnfH3#3Fi?lkuHIvJk>c%=N0<mv05l`%VtZiC
zld_gTbO5Ay3?L5^X0yoY-|Wlr_u)=uST-bSzsVj8+RiGz<Ug$tsd38}3%FB<(JUlb
zV3#0G!4Ri3SZA`!X`K16G!!H?rwGqF?)aWee3)*W7J=fH>nn+Uf7cSo-|_%+#Pc9_
z1&i2P#RPT^Ih)_{zHlc=RtE$m8{AoNt4`0Y7V|Zp=|KI>{gh{nZ*m8px&C)lV;`$y
z#aY6|=H!9pAr;VxTOQbZx6$`ekpsB{E9CnJaR7~wCGm4y{xZF+c;=BL*~t(TF?ffZ
zks5WkyMPm6pz(EeqcUE6LD#hGk+v%-E12#_-}mpBtJNgG&NB8kV|SxIRW;w4x*)i!
zd_BJ!z)45|o;T+qOB>g9Lk6fL;UFPOqkeLE9TLuPQ-SgR^aMocJUWh!0o}qXR&(uW
zHcO37u$0GJ=qZ+IPvrU8m)A8DiLf{CUCDsG9;8J3CN}=Lek??j?zDucqMBe`_*{BL
zl1b@sQM4OtT>W>o$J%fz4Of{-wbc%d&TrQtZ52bd&M<sbb(c$^3`q!ehq(p%gKJju
z0hoe%IUN)|w$p(UkKh3lyUeom|G?w!_}IDS;Xm`^gGQfTvZ;rvb#dH-IIN-8>?awH
zQT}@e?Q2NuQV;=3C4>N_2eKf9<_Nd!hSLI{21W=X#x$d~q*=k^6l%n6Sc;!6>Gpg@
zFHyM8qgX>WmjUpy9%iJZqH(q-E(T(xREA#whLzy6c`Hh5+*v-@A8Y```w(R{?K|WJ
z>$0rclyV*IVfotY_3W*g#A6NE>6Z+%!pW%zfN3TpaSKyc42bd<a7~0&2w1eO7&&R~
z+1?db0GMtJnqQ1>>8qv%c&56Awru3Lh&4LbKf^IySZuiO%Itu9j6U{R`dB8J&)L#9
z!-_%vD3#!WII(mqFiwcNfm#+F;BKo5O@Zvi$n8A~0+cH)LDj<Qd0T2HiB-&j+?E7M
zHGuRYx)O)cZ_#P?!u0O;Di*LejvR2n_^t`Gw||kdwf2jel(Q1#$0S}J2AW>N{)yv*
zZa7Jd5ZJA0xKDDd#aVIIp*tG4ws@wNlD0gh@J<UC1JkC5k{8SdFW{yW9ocsRv$C)T
zMf7ofACo9l^8nZ6ecDKU(6HMxpI`#+FXgRX6yYZ9%1_SVybY^tt-mrRAse_AyN4L{
z6gx#BY-QqlwG#<&O+u;-+owVGW?9L58PL*doE`s4d%_e00S{4eML@B<`<jvZ>AKY7
z&Idx7xHx<EcU_LrJZB?P^f9vR`ds;PW9D6gtlpl+$2;*6EC<-hP=9HptOq6W;0Hq1
z7~%RZ59xwiAv(gUdbdB2N?+a|k}O!re^HvQZhV<ht4OU1vm>_Mg(lrLGv9fd*bfxW
z4OEE+6CaQCwqBk`Whi~8|3Lf{yWW_vTTs}AbRDO%%$^&0R**>jmHaegpO<J~cbn|O
z%r?vhW)c+)!hRY|tlkPZS7v#4`Ab$~|Dll<(L?p`1kg7h4buZzwGB<qE~@(76yEoY
zhB=cwalU;vcQX6&&|dvAO*b^Hhb-ufB9NIUhj{&?=m~O9>zFHM3$J}>;b`PqNbzHP
zxH#1NSJp8vG7ekTxLp)%`Rj(5dL9wu;;Lq5_m$=&bm6c!X+K!@@q1xO>TfanhUST|
zR;gCJTZzCLB)e+uNMdo@n0%y-&FsH_PV&i;w6erCCS<yX3U0omwPE*Zv~V!vjvR@w
z!^8{*Q6P7Cw?XcndvoXPdvjNLCkbuX*bQG-u;*u-Ycv(xkqt5Ie==ZaU^*7;kB5Z%
zlzRIk<MYy3q#w@kXaWMv>01UcU0UKXU4ebmudjSV5vO7H+Uup70iW#ce8oNA!){}0
z3RZAeaP<Zmbm)8IE7>Z`^Z+aRnXbDaQSrc$>Q{mY*Yz2~kC@2CG_^xWY^7~NxzGo1
zoB!(hurDd^eft`od<3%Sn!farZRPjC)Vsn%>VSdtFYmuM66u$2zme(2;GUy6B^W#$
z(`U|09lFfidO>$)XYCrGbR{}NS}>GyMKDWkAz8I$e8;Aat)cG|4$z4h$iP>kM{v~)
zXTOL8WeoeaqX><@r###`#=BBx6!jPn=G4KADlkp`5V=l*{XPa>1LMy7<zN9V4r|=3
zX;&`VFQ^#r`hr72uK$4vm^>egWrNDuCI|CmUP_c3t+v5Z--GtzkF?y^#C>i#nLhCv
z$uf}gBx)8EZjr4)Bn8n}oSQd3E<$Pc?jpX@Vs4es-;(^AnyYPV;s2|wD1NN6T=sD(
zH{yjWpZS*G2BKObIl+STonbi!no7~YcVa^QlK6wk57i4HOe4}`Fx(0iD<&kv6K(m2
zOKwlFA}HN54sR`#iCs+*hNB=oK#PLFc3G&JsI6T{Wk*tOlUcI0OKFKZgwGe(H;033
z;6bXS7+4{>LC=DBAV*k}dn$J20)Fui`hT_4E)R<_nre&~N`xF8eoU{gIXmIuJ`#T6
zV<jkXOK((Di>pFj8|APY5@smOp3)tgkN6sI5x4Aiv}hB?53bAIrdIc}2Fcl&OG<|g
zdqs8%E(U%OFnrYTYv-Vki))+@3AFI%Kx)@|9{H%$4ZW9Xl14TU&!xp@%F?{otmY_C
zz<}deu$P3GGwKO3>a{8S^m0_eyhTvRRZ)EGqxib>u#K6j^ALzpKTgiY7tU8>KSHhP
zdfCiY#h-KI>qR*9;20~FUai_n%x<E<>@gtM`LH5MJs@1703cKkX-pZ&e|$1_*_RlP
zF63>ww;$WoLIZ)>!zDm)li$Gk7Xl--x6n*;6)~L&iMa4-{bGPnna#GoZ2<G=qd00!
zG~&HEO#?Xn2}i@n&;kDve|NWT_q=XkilC1n<x1o3@xC_P*4rW4wk{6|<~o?ta4B1m
zUF%4c&^lH6&o*p?VjiS^fD|obQvRt((ZX_v9om-H+=ybGWaCVjyJ((!sp4wjVw|6H
zq7*w-LDM=}dbv34$(9x;uOl<hWRcx-dWi7()T^SZWfI(Ye;}X+`wtLAcn{v2q%N9%
z8NHiR!BI0NQJd|ok`3zfF7cC26@wk#wlxXk=Gz2C#3!lZ)4&)iT{!N*4j&LJJZ~8V
z`c`rN;_syzd|+g$HEqZY)}B0T5OdU(gww~}A}^r1PCe?shCt0?h%%sR1>Gp6C*b^f
zw=GSAv#~M!k#_Pbu;_bKF+lvWx#45LtpQpnAouYqZA;^aoT{INZSt#?H4G%w9V8`G
z0R$+#KQI#K4cOSxM45;&RRAF1(?5d~sudMOy9=OGcV$sGkiXHC2az*TLbJwSd#L#W
zWWLNJWt4Blq`a!21jkG<|F+))f@%&(=71MmK(!oXN&#t7xd1zweZ(--v$f;{qt#4$
zPg8*FH<m2KVm$OPtFGf7=MOE;e*n0c+rTTE@Oh{?_m;OB!wxg6A-!m}FGQ$!T#b6m
zxDBcnOhx{1h4ec$+flumi_z(Cjq)hgBWRW3wg2H^-<-Xse%Qq}l^?Ob%5Lj<Vth?u
z{T0dr?&qw(xS91y#k%WTlE%rdwuJa#K10V<_Wq;an*brld&{MdM7Kw?I{RNy72dIA
z$_X?!5`^%MJrIQv8)eukA5wp|fX{jCKG#_^H<~A@=bT=!JESCzxWh}A7V!-SX7PJQ
z70L$o!hN53y5aF+I3n@m%zI|m6ioZHLzA^6;$kM(OlX`~QYUf~wpR6_FY%lpzF936
zRWBc5;Hz!3>p*NE0_sqvSJ|fpI$=hGxYw8z<y%lkmq7yEkI|)`=?NiI87lgm!(TD2
z+UPn%FksY{PXtZ1@{$O(QNUYPQqIg@r#(<ltpCED0LA=|xdma7sjtL}<B`xGCZ6DW
zK9Z5O%9d2?mP|#97xLS$9|lJ4Uwk1%CFQ<RfJL6DkZHupJrXIk$Ten}P0I3!*u-G(
zfIkuf1;KuEc4qVRDvWGk3_kMrBvpmV!MVXG6m%;XRC;}8sbOv<Z1W4L{0IAtQK4L!
zIcF5fhmM7_{0B<a;Dim{7g=~!A@v``k`oD>RAvswMA|=)E9N<7bHg0cCBpykgPEKK
zfd1nQ<h(cl<M1Gug=jwfN3>az;?h&aFU1wYFWMqQLU1tO`&UCk@*hH87VJ4He(8F$
zbg8MRSdn_7C-NLIuyI)jzu%I25@Zhk<tDjjV+{Vt6FNO-3dgCe42<@qf9z#zThHJF
z+fz3gtG%NA*u-SpEQxKcDKX(75<_IL7u`8<<~FqU!|ssF+??Zk|7BW93Egym&^?o}
zAmwkE#sV|`#<WjZS8|b1UIV9G*QLaVYY(cp7C1`uYO)}fcrDm7yemiOq`Xp&*q|BM
z>?X^=`$%Hq@vjXxh@6Ru$+{YD+{q^<hL9|MT6o>o2EKTO+rQS<Mk%=$REv30LDgmb
z$6vORd$Z<dpL%$(LWHoeh)Oh66paSXX88nf<VM{;BcjH_nqwlsd@2C}xLCpDFc3Bt
zs|qU^3yUD;InBzB&kGN3AqAS6`OP3F@KI)TpwN&AozdH;Tsy7gY)aC#jJT^J&rjrQ
zjVXCPxjS9<Ag!Ak!BOb<UL2Pq%*?T(MJxE@yY?H~@*TNnI3$kvb4Ktv2gRGL1V@*Y
zx$~~Qw4q<*Ld*KTwT+;Q#N)qSD&-I?ek-F^;X#~Nl*S%^PBTp9j5mAClKhDN;bhnu
z9bwxo<*Kms*-!Bb^U*OO?1=7Vmiy{56#0Mv(0fkb$XCSrYhs)E%xzR%viT;So9KjB
zst4X(wuA4r=7QJ^Lfu_ViQKIgiGRkBzfLgw^vtPMh+GVIuZ1D%uUjW)-rfTjM;Ylh
zEOfhi^|7t}uS59%Y>)vszqiu2Z<8`zBg^0A2YJtH^HGGHC^u*?bbjng8-FR?tu0PC
zSe6Tj4l^nF+7@PbIv$!H!VRV?LXwXNFW7Jj+Q-c|hvAfkeW8Qk^8!5CE<Y<x#vD_x
z03IaDHrqTG<?l)rZW7XKrNGmyBY|x18&X~p$FRpgYP2xd3OiC3G<o>5t#Dv{wl&Dq
z7%ps-&6Wr04$6Z2KbO`$?k>H3EG2p5stS)8QuzGNi&GvK9sCO>8z84C`68P1yRhC7
zK|S*j+k=pd*khY;>P)7ofnMo%a^k3mT(rVz7D4|~-0PhEwxF%>Am8!<nAhdRjKX1I
zkL-5Ntq48X5~0k(Dt_B>0ke1V?Sc7|a`2~zj|b)SHWoGiR<ESMI;`weNeIL1aG@i>
zvw~sqy_eAC{173>(5LuI-=99we{@Qj0T`f&-$vIFv}gN#bDhRrH9gf)B;Wd^MP51n
zkG<#w*0p{{yt}d%)F}ltU3`x#57Had`O)Pko5T$`;|bmJ_W&iTe|h}7d%5e2FSlbL
zc~XD;T4`2K&o?kl1bRAS^hc#;=0{#KOzstw#7Vk3R0Wo!1s&B|Cwi+Q1lhIqahHXs
zmEL|B$B@6oQogoP+6)V<LSJS{4JYiwfs5TNT`!_RbYfjgS>sGZB@+C0)7mzGzn`Gl
zEFw%RKaUWV>$Kll?HrUhJG+u{UL$ZKLSeT3ya>dR5fsd}sqRk%z2SFz^FcIQp6Y72
z!Vr0ASKp1|+A15?YkDd7G4vFfscS>6F`krbI09VL6kmHd*`Y8(ykRP_d$PSM@wTST
z0C&y$<koK=GJ23L3*n<yT@JSOX(RCP{$bm8%l^%kWA5w%pSg?^8H|Y^L03971+ApT
zb>L5ovgPwQGA>0}+>GtyDeg7KT;7{adfIj*{4!ZtT?S(s2y1^#zG0ZBzSNAHH}TL}
z;XrHs;Tzd1t(%YdBQC?#6F4xAX<{xyOx(tMZR~@iTAZlo;%kpEaoQKk#KkRte>^;9
zC`hYCy9~jha0$YAk=V5tZIQd{smEnB9vzGGcUfjPS}2Huj;+%9DAOeqFexCEVD(#>
z@z~P?Ywo<OymDoBNWr}T?Z$VP$ro)&yZuxf_G2}-!EQe}(u3gF>O_fJKc9<V{(9af
zij?|J@UwV?pLs~2E!dTLXeSqPdR|`7c|pgnqTc0ISQ$uOS)*9^(46a8h|HC#!komu
zRui9y(Bf&emInz^Es*AZC&Dql3M9lkFLqsAUuZDGPl@b-8MNU*YINZA6PJyJ1uwxl
zo;M9TVDO6N98JpJ3P@4OCBb|fKSmpag+90pJ5FMq)x<%03Os~L+icUJ>5LG*)<|wd
z_qZO$2UI=~p*%Jwnq=@qj)kQ01t@NOSvuKPK6mdNIPb~2;#QTn%PYUq&N&*Sd=*Z;
zf79%OE?jEo&oGTkF3+$d+55Uj6EI0;_bNieO<@2m(081x&u|bN3yYkcoEHl#lmJMC
zg~bN=KmPh&411O#Hg~Cpp_LF37C|=w2HB<fDiAI}{_X+J|M5rK^4ue9qn)$2d)3CT
z)U2)QdpkJ1o%`ITO?b{lP8t`VpOYcoJw0h_-uOByw!H|x!AKw$slJ2uagD84rtf$1
z%&UBlu5Nnn{-N@HMQq1<VG2rXV#hD5&5VYDQ<$LpBMg)sPF+s2lF}-|3`r{Bu&j8=
zD-?uO*c%BqsaRxFStbcutssq$>D+pcXP7X*`!?EO_~IK4BPahqi=kwGD^SkY4ryH;
zYY&S~2&aDOb3@omXOp#sT9Rn>qGKsFchICr%iXt02Cr_?{&y)IAK7MWG00UmJ&01S
z*d<xD`s@6v#(r|s^Ce0t>&p^FoQXNotmQ|B58okJS=;ddQS7QMpVoEwc;u7|JGXo1
zE{!uA<<c9VaScA{Oy%h^tNygJaeN1_R+4BrW&h#O5;b8Te)#+e+-ovVVAXC&WWklc
z(lQyMyx$^L|1Q*lj^+dL-{cLNBG)!7RknQ@3|L%8ZlO?B2@aVJ&#fZH=ZziWyQ(_I
zeSyIQA%7HIG<{OSgI13mZ<)+vDq5aUtyZ@z%Q-td%EPmloiAA1MhfxXl>~uak^mW0
z&8lQ(U6P?)DC;SxxaTgD-tNp_3B{6n<y5ZnBf;+35QC?>!GBX$lr~Lkjw|y0ChC`K
zp0<4CB!<EO*OJAlvxf5r<2O>OBc`pe$$kEIDL!k<4~ff%IbiQ-dp7-;%0<Vl6J|Y|
z)n$lzi-w<n5WYBzE~XK>Pak~$4=;_2NWuHjwV4ZP%tJl2Mp~o@y`lUi9=UWk0dQZl
z%gM9hIZt|hE(XR%_7^I6*z^+Q4%82JG{@z3dOs1XJHBg*EST%I?YoylRchDwB4UuG
zYGZXUe#RhbJR?_UCXX9Sg&O$y8wjqS6JE)ztM6;S;-zH07vJ!nFUV&@YDU$(<k*5`
zbLV_hW#Eyq^}8Q`%!?YV%C%{)o~$cFm;}f)`7HJd-D5^3+O(E>S)6}QV|Hs2=1ieI
zj}7qvT+@Fx=#sVl2!e;z#L7wnf<Mz0Xy6`G#rlj77{&rJ%=&L^vfu~b#lm(30-_!T
zq8E!zjOfp2g70Eu_h5>|zuX1lG%AE~vG@tpF$Lt?m^pFB+wy3v9RMH}GopX=*3InD
twWJR#m=4H^|Nki){Qr_t{x3v!FRRRTV#)cTR9KiV4OLy0GNt<={|^iZNk0Gp

diff --git a/89-transformaciones_files/figure-html/unnamed-chunk-4-1.png b/89-transformaciones_files/figure-html/unnamed-chunk-4-1.png
index 70642ac77cd2135c115ea3cd22b32918434e1489..da0485d28f3e76c0b79637bed2a190b9d6805854 100644
GIT binary patch
literal 116640
zcmcG0c{G*p`|fTQiUwmLkwg)h=crJUDN}|bM2ZY$rc5CkBvUek%n~wWNXR@-QIcd%
z=2@I;tG?%Ve(S7t&iUiKYkfZL#kTkRJkLE`_jO<Q>!YM7yMvUTlt3Wtke559LLh9B
zA`mvcC)tRfIBq@_fPc1I%4t~>2n^oD|291ApEM#64iMx|om8`X{<GUr=fUur<iz!z
z=dGp8{0%uWq|c3-dw$Y9HEF9@pPdm<BjO#Te&3_K%SgFQgw^}G*1=;|_fM-0MY+oi
zhM5_ME&jgcEHNE^p+!N}EIzSg`k}MJ{O=c|E<N*E`90RU^R_%ctb{k<HgLIx`6rHV
z{_~m;<;4E&@6X8zq;`LQy2))L<DZxB2~lp_{=7U+`2YUaOFi3%<j$P&(&6gp-?e4q
zhKCg_9zO;Ke9H}V8R?TqBO3;-@<KyHmlQX1RMpl#sJdF+KlR`Nx%1%0psQwP9FOIu
zrl)5kNYu85*WLW|nwdE^H&=+8`$EsKl+kO=qpGSc&!6x2ITbUm6UA#Ot4;PTptUi`
zQMM~VnPu#f429UAb-Hbk+CcH=<z<4>`%Qmd(h~Mg{CP##O4#y${Z=>T;F{S8cdoru
zRKpAR%>}~ijQ2>F8DuE1wM;fepMBkFPVIM5M<-?cRMe9P{{D32TbnA12}&37+K4yw
z)WN1Z2bJg)*aAy?DW%VzJv+fn&%V|Din;lNaJtfqg5z@5R#r!k9UDIO+~5DajZN0T
z!pzLf#bk9)Z*LQ0<N65Ej?bUpeVGdvxN(+bakfKWTudy9<A&{wqil8GBnflz%2dq-
z$=|V=?xda{R#sN<m-hDd!#|_iUVRZ&;R;T6zWy^wKIx-jhUQGtFnM5=+J5axdUCSB
zs;kFoXBrr350<aZE>P~EUlruj@m<>%s=y|3YSU@SQ(tOn)AH?Q>8YvX+3)BVI#BJ|
z(~@B*Yg@A0r>wlZzsGegPSV)KBzSo|=%8D<!SFYxoG?q-q+r*N*Xw%%eh{y#)P8N5
zCJko`3eCplu9s0Y-Tm@Khv6YVCCTQQ<)nM}HfCt1)X2zO?44fJ;Sv-Tz0~6)bqL16
z6RMD>6F|{q85$OLmi5WXij(u<jX@>GnfnCH<w$n<m1BiVirYBc%MI|3fx$sTa?iU-
z;df%aDVWq(dDz+IWMqGhea<nRl`tqHp`f7P<V?1**k@&JJxumkF4o=g0q#^=Q!}Av
z^R-@eH8nMvo4s>CJNpI)O$-cnGcoD=<OKh-sbtQc(sl2WN}V5WdG$HRVvqDza@xxV
z2GS(Lwj)oy76dK6Hqw0k`nAwuHpZ}zKUu$c)sKc-ufUGSWo02rjxK8BUkj8vc57*7
z_{o#q(v@XCd%Vf%q)Ft@o^_ZRdYo^UsCHJL#8dhhMakvk9@5cD>?w^eYX}!N{x#Fq
z-C4!D2PnxI#Xfa*hTbIKvE%p3Gb7&4iib>+#St9LkL6Ase0;ZFR!>Pus`zRmGx?}N
z%pa$B>pCDTyjbGC&EwIPQm;w7f(@Itwx;VxQd$kv(zQ(;J#avZeJ_`O(eu?$F{jC*
zm2$6E?Orm>G_LcI?rlrf*v^cltjvA3sH>|RrrBVS5i<9F+t#f~1C)FAWZF%B|JD62
zJUsmKr%&Yj1b)u<xM~Rgw!3=u$CE?n&IIf?y?*_=kx|>bI~E!m9j{*>Z(C}wu0G4Y
z7ki`3y8D=3!KaTOKQ=cvfBEv|>sOawU-B$^ic%XrIgTFvQs|fy7`W%gEB#^@G8!(Y
z#VI3OzNx`TR~Y%%LdQMQiZ8@E@wqtI`19dipFTaH<@w$iEfeOqI5A-<D44U8K_nzR
zeEwIrBM#R0730~&0KYseEm~=4a8Q<{s;VkFayfuOl*1-ENnJOzZ0CAlFw1FdRMJj(
z$S590`JvZxD+Qy|)Q_sHtSra*U$Y}^$-8fC-LeJuty%M{GeY!Entrh!&(Y%C+=bQO
z%k$$sJ@4*}Pxe)$rlwLoR!Ov3U0FP=_*}`wQHOYa$-dt4b8PIwbD=9&uc{^ej)^%)
zLlb=X!iz?xa)Vpu#Or6G>MPt+dya+vjVDf%-#xLj_{hDN4A!rE?(Q^PI=KerzGh~u
z2M)aN=+KxnG&VNAx_((&PCRCjt*xz9>v?0^5`XN{TXvV&j>CL>GYgY_)rXTS{1P4i
zM*k>*sJ_ACh3|eYzn7Yh)z{bKE?7N0yu6-qd~;oytXP~L9PcT1?XM2<cvMz)s@$QQ
zw5h4-QqIjAtqG4`y)sEwPleS+_YPrsQP(TOLqq);MBknZ=GwAl3(@i&X3Y!?zK8G_
zt}YHT85$bWuPl!jx7=&_@WBAv+1}nxzO}C|T+sFRtcH`*3O4fXjnDD4JELCXe#*+q
z+EXt%IyiK7b{d(OFbP}z_}Q7c+vkTjlgs^p0IoaHZ?HzJ)Y!;qm$cW8eZ<{DTEG%E
zZ`(Hb<A-^7{zb0f7BZD=f7Zi?Ek3=zOKD(YB4{(DL`LTK>O$lTVQaQy$Hsa~y?_1s
zmGn^9!^5M%<#+y#&u<wS8E39iFiAN5`XZdDCSp?md}!!edG-CaJ3N9ny0UKdeZ;1k
zcfN@`>Ap21Bjc2*?U}-Vd1foOfWRE~&EwJ9a`BpKYeqmoz{a~#QBmJ&Lr)YH6&<BM
zx2xguam}O8VrUIBJF6WN-Q3*ZJlnQxX*$irPw7{l<(P-q4TsKo^Cspqmh*Fbycpk!
zI9lpW@$TKbmND4as}hZLy|~PNRorZk%ko&bpv6r%jkYx#8yhTrdZ_8;OcTN3p7lVj
zFv1W10{bs&GZZC!!PRvwCFLkvVBVc^JsKLC{9COF%825>5sM5B3~po5H^1Zx+~`uf
z!7{_VQA0z+)z!6&PU7~@Pg}^S6+(Frad2=PJa|w{>`O_>$wl^b#JPcR3m7HU;pMTM
zp86N!%lIf$Q&Wqt1<G0a&NP9m4vlNg`m}Yy8rvqw|KwX*v2v?hFVFa^nDYq;tp0j8
zPQpw_M`vT7NBa0|9Kt($j>BP({X&**0f*;jXD_Oaw|?*>3+FSHCh@s{Kk8DM!7X`t
z`RRc=liHAI9n0O)n>TJ?ka!dllHjnHEpT{n@D;mI;t|#OMw$`Eof>kAiV}7c2bjyc
zyLFZAR0jqIR##Wax5^xR+~|;7y)W7^$oq|;S?jg&uZ8<}dQ%eJMrD$Gt6|?q|2W0l
zA3V0toR*f}=p3Eym&iSI_Ya10m#w&bNmJ8i{Hw10J3l|NZd=53xW7h1U7GH@uC6Y#
zJ@$PccSzsQ&dz@M(lAk1fz7(qYsci|q;JrO(0)Z^`22VY=cU-28o^w;YRjJxShK4h
zQCiK8T+Gl^AX)77W>TBKJ^r=3y?w4T)8x+h*VKSB#`Vt=mkv1iQ6Kvlz_4%s{*`Yf
z?%%%~l(TI0cyzNn-&9DeJZ1Aze#p|jzo?J=bKv0%*Mg5k6e)JTzq@JCsYs5Ml#Fb8
zdiv$>LhUzMS*^TJ7z<TjDaO5sd9|0UD@dLzxY%_~;>eLBTwK2f>cX#FxiWIFW`Aee
zZ<bEJs*nEra$xzjZnwv}#2x1prfp_|ZpK;<)*F4CA8KL|64K-fKFr1jKUU8uv>k0v
zD(QcbQqL7%^bdNL#?wkNQ4%jDV+BRUdbt1aeTQSFolKlTX9EKRvmd@PFf^pV$s=Yj
zEZEtO)&@1_-ZC8ftZ|w~Zh4~D-QcKHDC=%{Jw3g|jQ+@;`<|YsLm&HE$nW00`zhH1
z^)c;h*NXRc>nk{){0sfIDsTlyN;qqWS2i{_l2T1nv*hzXl4G?Q78-jue`l<#zpqaw
zm|c*UcjQyX%j|4hAS^_+Tf@z9o3`zIx|qiMZa#Z{vQPFH%c0%!*#a>NJZVL0y{W$+
z+;YhKFB|=-``sN?Rn_gxii(PXZ0A*WJgU0-;>C-DkB7dO`<*_0T8AsgY|pnzySgy`
zbLY;f>e1-B-@kt!$zZp1SN8S6{(k<;?+N`Dq^qmT<4NZ)lr%R_A%`F+7FqTC{@MqG
zVAh@zI&!<*<`yDnP^GLsjjH}%MB%nU^+DvFJ9jiv2>R7H=7tyImz0z&VM=uuVl@J=
zC~fU7fRXAi%bzWZBSdVI5MlY+3@ZY5Ny8T9<nE71-Rug~VHi^F@$vDo$}6Z{H3?NX
zbJ)A3aanix%nLF5=1h|dZ3_rXEi|dx+0CmuduMEUX5-=|U;Y8uQ2}Zku!y6~vC1#c
z%+1Y>d^q!f_N0Cz-Ei{$h^19n@Y>4sg+yJeycOhUq!Hg2y)!d5(l|d)vyvmhHB90T
zY7BI{rGNhXNx}^9H`oxV(H$Nf+)?DbjNG24l^HKWkB^~nZj{n;aB#4+JkA^|<GmBn
z^`Oc0+NA?ax&P%fF521I85^@Pmvwe($rMjUopiSoV`TnO9mKh~xOnl!$LGS<LspYD
z&yJmwmskFpjl7QfLb}pz!=@o>5p}U6M~0Efs~;O=e5$GOYk%I>WGiMj@eQ86Tbi0n
z$8NO!Side(d+!I&nX~zMc^73_040$5E~fRKIB`N;T-?vkZ#y%mdP>3<`Qe`iNiNG#
zG_=xxw#w@)kBiHX%E$B{?8IImHfU;U-numzcP{wj13J}gQ$F)gmQ_y<!N7TVcwmCh
zg>J<}L`2B3<~uF&>Q>d(*WW~_;j?RuI*G7O<5!L_s-1OBd3kOu^UWI|gw=(<hdCBK
zq?_;I0vz%7^b{5nnpj*(I{T#Latff-P-ApwXJ;QylgHpx&du)Hy1G5Pckg79SQ%;K
zl)nG!)vJ+_p^MEV!8PJGLya&Tubm8LrlyNC!z}<tBG;v0V8E>^d4Wld<!XZ2MaC$H
zC{<u`1GS+D4gA*{qiVeuZ8!aet+cV<zkdgQqm-r>y!j<rgS)PT<;;VqY@2J>t_9Id
zcu-!raABu+rdeCkg-D69PZ=`y8dwb?Kyt`U=Wp-to-nTcRP5>sWR#+nx%UeB*3y=i
zmXeaYxApP<ZoGJL`Qk9VNKUS@z<ydqMMZl3diqepe(EX{QzUQ_J9}kI%QUh*61&=0
z38#e#mEdG)4jY~eFT}KbWNb!SeTpU~CJuO{=oLBuu>~nEoQVEgFT2~olSCN5mX~{0
z9UUD0HeR$GP}5IMMb({qi$zp4JisnpyCweoXlLdiECl$K_&o@fu8aM_J<jvp*tT;!
z#(T0%Ti}~m)jk>;S(fFQ;lAPFPLxu_4IhNBlaY~eX1y+Wwlp`kkCCwmQ8ig3o$b(}
zs(!2M*GGY5Ys2_!+S}hQ(GTOBh;YfG#|7XlD~pfup^2ek-^fUkeOZw}(z1bwh)9}t
zHplVf6EI^yLw|byrCJZkZSM7ynxjTnuMRCJpeA>&WfK(CI7&S_)u?oF!AT;xChp;+
z8#C{T6NbE@R(WD9EHXWLsKnN*uMs=VT{doY#x2WzILv|qG3QsCf=W>So<EPuHA|;~
z&W!%c<MxeSF0xAi=)_Wj@<~XDrsu;=w2!RD<v3^j2@Zb<il{S>&c-P#E5nL7IXKLk
zVovuIIeSV+7X@xiIQcK6t?Cvz*<*Fc^S;N<oH>)0mWJXA=)1+xfWW`)+2cReyf_&t
zX_%P(8>?AGlE-qaCr_Sakqs!zR3q&DOMx`L0<xpxR8I(hGg1|J7#K`MaeI`p^xy2T
zbgE;rozfB>fjTszf78VTtxRLYFI;g0z24O&C;o5+Q5SaY+7-6R*VmU(%<kK_Z&Be+
zQKZ7_*RZHx*Cz3QZ)1{lIVw1SZ5OHB9OWhR?`vY~Sf_W;^EIAiBk=G4&#k2wEuFO{
zxR<aCZjIvlhas2vu6)>Bx50U7hDT7a<2)~cpYGou9F~=p1!l8#9BoU^ary0_JFrQ~
z5N3M(IKCG7w7sB}NL{&I=`3`d2j0J!lXI=<J6@m)?r=U!;HP}{*qvhn=84FO4HiCf
zM4olGmX?+?Omz>N0xjRQM|1w4K7X$G_|c4nonPGabf0ZXY%Hsg&;mXMXcrN&y1Kfv
z(;lAQRofJEQbC~>R$Xo|Nvvyr{#4r};j)tJygav0@FsBgJ~}!rO-&(T;k<Y6P#+{-
zKLLD<_oG-=ke4_8l6&jj?XfT2-P_1%Cxq>ZB%9j(jEqUp9f0UAs1g87b8>S(efd)0
zy0!{H);DRlRKGJS{&Ho%mGfW}8Bn8$gz8->skPLqy1LZvh&aHhFhL8w+qd72q@ms&
zM>vrgPB;);1JJGoeunu>-huRM1#_Tobhu11G!eQQ2CehHpq%g(coG^aeC2}&B4cpZ
zSHu#eNUSSuw@=fTtvmPA?%5+lawOAD(srcPmx@zu4>hUh-Me>@`8969T_a#zyM1C?
z=bB#10!xNh@%=n00`F5`Q{@Uaf!B%BB(MS__?&tY8JP~Ai1(BEZGgh<%;n|(jm!A)
zBVjd;;I4G%m)!M?vptWccPRg2Fh!)#ve+R$F`>mM;WYH?m*N^Dwsr>{uSdgRMOhgN
zAB1jw_zm(B?yLW8AY3>3*5Sjio*lcSZeK9xFgp_T{JEIpoVkU?cy;!Hz$(<-*JghW
z4GCHOkYizai4ce-$Hm8^{uDTP@IzTy?CxWZj*cxOg$jUzCrs*DJuLIS$Op37BF>?f
z=jG)^X^B*F^XAQtG~LYCua#VCEQ)`hxzJKwZC3sC$g^k9;KJJ|7!?#0$hU9Lc>7i_
z-^TRN)H-eHHUfH5qGoA!)O7Y&fB$UXL&;`;q=pcD5yB-Ptu2;Pyt<V9C6pK>&gmc3
zFVYD&dCozfey*gXgo07*fCsxP*>Q)NYmSb~dDesRuU;)9!y6oY1_S)r{Z6C%E3DF}
z=GiXkF#hY*yLV6Z2kSSF0r5s4uZD+TQde)sFLpADCD@(Q(b<2#p}k#zkFT}0_36#|
zi3wU{3Lxl}-@jj-4{y)C)#n>ebT9&;t>kq@#XT1sTz8RkI3+J`;Pk9wTe`k0cEP4n
zT}eqP?|jp}=OvdlwX{sFhkQSd?mm_Ykgt)d9arGIvM@PQGKFkJLrtxoPzT$ypB_M<
zx9w5?*;BlRUwBaJ>+9nqk-CrS6@*Y)t*+RY(=hS9iH-Fr?FA^1RxaElovXw3+vOx%
zpl*R(hV;7s5Ts^LZm^KEva(v|k91%Sp&MYrRHU9LTj6v~Cnd5XLn(QUYiZjG&xZ^A
zUK&l`&b+p|Y=Im8KrDW=6;)K4>>_;p{9eBNQsKW3rmJQ)9xES6!VEYI{)5WwHjV|{
zz<6ioD{pVv&<(x4y&jZ|4^el==HJ1o&y)kI4f_=o6mV*!QA&eBIl7nhT>1Q;(o{BJ
z|B>kods_e=mc68@DJi(yvtzzKKBjF+K}zgcoPoT&&)f}SJRrDjB(;+PtdWk4P&|MB
zVt4q{r>B-(25UoGKvXgO2z>EEQ6IJvUrP{K(q;<`5_LattP<QywwMiq*8qHR<}~oX
z+WZ{qY_$*pBvK-7gDZ@|N6QU@=mMWTJ40d$TMKQe2l8xcI4B|#b~Dd;S-^QNx|Um?
zgc&ewKi@Sa@0|=;4c6@`S{{!8t#C-37Hc!Zfv_SM7Z+I6+}P(=zxLvurOq8XbO^-3
zBv>d*%b||+p4LRwwWU$+TIFp@G*NDX$8A2m<t4LYVq(hB{DPgu=?}Ao?qWH!UlQf|
zWFKnOB*%U$;W0;x?tFgtlEyE4k5yjEHnT;PAv)h!SN3T?>f5((qy7iPTU)V<uwR@y
z<g7fPHjQK-V?D+Bp?Y>{_PWSkS45b&_$bacUglHXwa;*4d3kvrkAQdcPiJgoszYq5
zsHl)VbB68k;p?Fx97r)D4z6%E-<!8HG{uanpMrWo{9qJ!c-k<ik?)N49q@>akKegZ
zptiKM55xwzp8GyNd6vB&5Hj~OF!c5H%|<W~)VTLfZ1s9(nY3laeH(>hq{MlW$*HN`
zB%AZ)KX!FFya(FeN`5)BCS(T>E9(!Gf*y~MnSwuKsihx2fH;)z2n4=OOiaX9(!moE
zbI*$N>A^Pb*H*=u%PK47y3Ht2QL(Iz9p?@{bug6w`ZGCJSW~muu5TLp&ZonVO%MCX
zK4he#qS__BH~IGwMlTvtPeh}iKxu+KPA3e??sMtp!+wkr?vV()t|mMvx!h<v(NpMn
z2QHU?tK{AmU={Ex_?U7hVqYZ^ITK9R9Bwx0ogT6%oIm9`snWY>9BkIM)s|Nm;?BsY
zYGry+PPJ+1`!h-ENrwJTFJ4=9gyDHmqC(JaHANQQ2ELZ0s=i)8P!O@^Q-)y$_8G;_
z$9}7Rt3?z(>S;P!!*;qH?Cb|TCKjD+-FFUkaPqBLp@=yxD@*42!Oe|O_xShH>`U$;
zi+gW~RDhW!pSU!M?xWC(zEs_Jw^2z{8@Zy=)_48ghMby^z<%S2nt;oSok-bb05Fe7
zsh4t4-#sbq-7jX>`0-<WOe%P8sUsjvmC0YK28D&?+fN$<Xb?oaHhz8I$Y__Rqv+_6
zh<w+iT~kr<s#Z3N3b|BH)Fx>5Y-M$|Hn=g5?R$WfHg7C}_tof5>on~XLm?r0=a>1X
z*ClOjNS}#dtaegRyrTF7qAjG}PTyyN*LJx1v?jHPmlH5DxbFj$==Ag#O0KPB4p>Kv
zuO+o~9I+JIAEPFBoyZO_h~B1@1_V}2HrU}s!KC;qR}%M(5DhmE1{wY2i4{^53SE??
z54?^PsYmr3*h@_f=9V877NErA$B(yf-;Qwo^2Ar?g^4ThYbu7y_x)Dz<g+9`zP?kF
zlbt~Lb;Y%GMCv~(M)A2&wD32$vPz<wU=_jb*A)i4O)`$(wll*<XR<Yq>gC*IV7>}4
z+5b!)Apu>2jsYo@0J?|*o>qZp<B*lXFT}h||D0u-iK-JoCwTcIVBkok>ngJcazzlm
zzr~FkAWzf*^rS0c1gHlt`Pk|53J9=yTw5rWFQ$vWLXTb=n6^u7f#*@qqZI7g?=ae)
z3XsgLUsM3P(0`xc(>t@nEr%r~B}GKECrR=Rv`_BY6Ns$HZp`yKGjn==KIir81!Th~
zPo8YtxRFETSxI-dGqwZdzRU7!tNydL(?5u&;JFtB_3~3kb#-+!vm`HjYr^{|qNV$~
z22edLK^Bl?W-h6$yoQ+MaL(^hLPA2M%aTu8l(D(F*!A|5L+747y%evq@vc<2*qp~B
zi=xGNfKOs$1n$-Uy@u~b7hhXjJqa^nY-LT&Xmi{-gzHbl23!1vhn)1>^E&rG?wYdE
zKx2h3C(~xQ8SzLxRlCW;B<sq}n>gTEgbpSV8@b7qOYU2D08N5?Rw%AX&@YzgPD}U^
z%&ng$vaJ{NF(`-zu=rrGFJ`$}G{+*t)u1NW&J5@qbpg#@!u!D%<IRzCR|uEyE4|-G
z+s|kxEe9l7j5S_Gw(3erd~4RmArfa#^`xdBCyb8@s=5lUur`DTneM2y-qP9YpWlw5
zX2!On86(gVa($1a>*@+`j<=WB05Hx-+jacsBoSHD4{T@_X$Bf|JF`KC5DQD6A9s<`
zj;QP7{HD#PWLN;VnVFeAD2XiX`*#~7KA~GPSkJH|+L(Gtuw44r{+$&y#ETPr0~c2s
zb^@s_%6(4KX+i2K^;cD~j>{Gn>0xd#^oR)EKNQea-q3+T-m~^_AUOWe^@zGxBm{yI
z4OoXqrM*wz={z6!Oinm{js4pW1#Bij8=~p8$+tjIKYyBoU@!Nh8Ra|#_=w=LOS%hY
z48RDN(^zD_cA-dzi`zPXx#3~Dv<(Blw49Suf!T)0h=}vcANd=G*dEIbk?m*@Wq<Ke
z((1?k5n3X#Lu8sK4YjqkU0jN$Zc)k5qM)oXk7UyaEuy3pHI&1xar0(ca`GrDcUp4v
zlO@5<P4(A+Bu?o6OtjBUY=Nd07WyM6^Hj`l+^C5V*+w9zVaqrn<{S5oNNemDk|NQ1
zbJgb=ze>;rgZT<3*Txa}doOB`6rI^3G?jo=EIz+IuS<xyk&R58bs{fjGQSm@ymU_@
zPQMzp)Q}&?S{MsI|16pkAWS%{Tb~<z*oRcP6ay$Z<C#x2Jj|c6Mlfmqrv=bRe=*ng
zPK_gi$M9oKFgLMpYZ6W4a|#iFN|;ClHD0ZME*!RLpgtlwT!A3P4SQo;qIJ^N4$H~O
z(L4(5JBzfG<&+*785wumxHd%j=5>%|Hi=IV6VOqK)Rve`8$s=X?#l%D+|rtJ_Nyzx
z41#GxCTZ6SxQHd4O~?}Ln`%$_k;cQN!utba9&rVxrjE(-@72Gg^X|6)&FEL5FJ4Gs
zM-ix*L~p04T?am}F@#Av&W@ZG1Y`<#5og-FH#STPtt%j55@xEfEtZqt%YpdB7{`&s
zP?W@+YZ?EXbAcxc%OXewy1TpQR}i){BO1Mweb{a!L5MPhJyDF^T|;XSshxxwr93_L
zv1E1FH<&2CZO51bk+`$G^%o~ofp400forNyeB33iqN+Mr7k)VC?0E%+XGgVE^6VI1
zj(#mhYiVuuclBqMg9ovTG9<UGtyO=ub4`b-sR1E{^Itzh(wT3YJT*^`^+>KvY#LKU
zT_o&R9`m`=e2zkLMT$iD&aW@n$*1Afm6aia7B4i98fSKPNrq*h1@!IP<*n`+j`KG~
zMMbB2MjyR+;R4R>Y$OfV_w)@BI)2Uxc4+JdZm*~?0axXF$Z$iH`iBz}Ufy%>TMW|C
z!35-KUoE^S>eG$x%6XD28JbAH@o}c10MPTqLZ<`qKaRrULKxqX0|(r0?&rN)`T9y}
zeWau-ipLk?j*6kY{WaXhN7>nlkrI5Tvz;SitgoaRm48PDc4^DDZIaGQHXunrK5}t!
znF}l<?xVK>0vErIOZv{6E`OSbmRw)I2R&=>`d?8qwX|enh3R^Q`B(?ehlCjbai641
z!PF(=h8JKRhC#iiYi4{$AVvf*>&hCQ9&7+@wgM7dy7Eh&H5JPkKiQ5Qv$L~@JW_P>
ztl&#FeF@4h3knNyKQo}1iG4t;!7bRWBsL^X7%w1KfzzT2TVOVzI(k<+TpB(08>F&j
zHdxs0-P3d;Q~_96+!&0J90{5a(VQ;HEd}=0r#Gb6Gr6wtAb>u4^axcM@*2lDFF(I-
zv5Rve>1+6DS+6C)IGaMq1fl|hQj_q;1h(^GD2Bl`=86P}g6jkSv#_{mWNiE_pAAdi
zNdj);?7h-nys1U3w6rwn)jKo*h%9PeB)u&x3zZ}UDafNoiakqv6(U61dV1Dyj)m5P
zilHst;GU%VErp(=2aP}qy#m2a@d`;{ukz6?lDEE}x^(H1xUn?Z0!K-&CAx9e=&7L>
zjhr=UCw5n=5hNi<rqicRC8+lij}lD_IAh)C&&#9f#R5D$&_M_Y2%yQ6Z#zoOpr>~p
zZ?Hj{L$?#1mG)1cauJPj-f;X+AjBV5aGq!w?MPo<S|av~pr$G9MbrD?KEYootdWV@
zjd!kr5Js<D>nJt)G%~ge@4-n9Pcsz%p1TzE`~c30Af@|JDFQ8Bcthi4DCy>n+PwV(
z11|NMbs(?Da`>;elWe{R!dXLApGzr7a~q@CUf@`^Kv4I?BhHuaBWxeqz1^8_EBQ_&
zZogQ71~fH>MHv#vsr`)1%)$Z#-Rb0asLggizE}U><T?d+k&hS{7z!_!Y*ZU3-VK3Z
zyN%RtQ>J+*?_;@E6nY|_LVteVx+|otp3pd{2n>v-IJzUG%w-)NG<*Elt3^WVk)gYS
z1n_{zw6k7kWONoei!xtDnnGJ<Ns(B)N0Foc!7nLEK{>uA9h7URtBbp_b?a93i{bgF
z0y_#UF9d-||31BlAeAr1PEg}S8RM4Lcb=rzDi5auWm1cU1o3W0o=dvw@n~ryG-8`)
z!YNw!!*)@RxlVk$TeN4x@yoQ*76j^}nvL*1?xKYlfM(=;XjbsYQmM!sN7iWR>ARw3
zwmBc##9xgAhI9ux_dBtGB6EwD-e5skmjsy4L=e46N*FCeo8`G1>#Yi6MHU!nJ<@uF
zn>)S1f{1mqP88$+kv;STx4R!G9f&uXK##b?b>s-MhXM-g%16u|k3i^^m%2r|qtL0S
z<ax+VxOVWd<@~r{zN>zZy&BCQd6W|W(1QJ4spEDNmrYDe@_G39)Y9cm53Wm0pz6AF
zFi4rGGU0;CWXVAK#)05{k>c08UR@|Os@a^Z5vaYCeE&}qGD*tHiZ>z>(Eo}UqXnuy
zo80Bl6DE9nZK2=UuweuE9^g?hM?)>I8j$i1vap=<^eH#kk=cl<lnbdF&62lo-vWQa
z-EQ03?jR#mO|<}R0Bs83$mxSkXtCWeH-~%*1k2di7*YUY@@LWrt1DM}flVTwKffjl
z2fCDN*^6A9{PHo13RRwrH6%wk@C}GlvP@O??F%Os-~f`=Jg(CN=l{EW1pVwIC*Fu~
z4M5f6A<Z1&BYgO<EVuQP%$TvwT2a+R?`(Ismc>^}=}!+C=^`c5$77SWyjX7+yKzv`
z)1M`qMUvfvhFd+lwrD$qK6?g#EkwT|u?S7|>bkmPoSf$}`XRSLbQVF#!)6FreAW4N
zGEQZ9WpNssk>~;*I6hjl=yrgjMg#x@$1$7se^8Os&tsp!(BbP;&C%h*wSDJKJ!}U`
z>ZHWSC?ml(l|Kag0kDV*d4fpNSbyk4U1ASv#FuG!cKePU;vjdYszxRaHmcE&_MclF
zsQYNr@T{wTtv~pCiMvm=MN0Hv%KM^QK;+tYKW>B|*&HEJ(MZ&ol)LHaleFKvt}NVw
zN21z>9ksO)yKiEQ+!7M2Acj#$5vy}p!u?$=d`vR{46nAl^7JhKQ3d3ZN=N<?h2ttH
ztA&LHVm<l!?G2Q;abL4x_WuXXh;e33o_u|6?S5e^*54ynV2+@v`T6;sr+&zzK0}QR
zaEH1ZO~T}Y$0#y(>=U3Q_9KbiuPaw35i&tb#>T~oJ1sOqN-%rZlh|wc_z}9JT<9_a
z@$(5Jj#!mpv=?Juz6@0`w$-mCLR|3N=rw|XHbtohLuEd7OUJGHN|9p)1<ALq+XCNs
zMZVT69@x5#=6rbkbU_l$AHsh@2PHs5q~u|bm7cZVkugiQHa2@nHbY)R<d$Y<jWRT=
zE%MNC12!gMMsQ{Ve{Zl1#f`)KFI`PdI~2fHWD96U?2*<6d4x9Bo;`cYZ&Oe_1VuZC
zvI%KHJyjS#MCoilGo)Ppd=840|3x)Yqq|h!(<8~r=~s0XzSMy(*v<X?X013G<hrtu
zkPx(@0`>`}VWAg4%PA}GVlHcIL%xi>?e!2vI`|XScm<TxzW47#gm=f%Qc2=0f*Jx<
zz<yx^b903pCnOjM1SOC<QKjI3_l{}wl*eX^{N_e)WHy|jq+I)e{A+t^f%`Te%@nRP
zbLBKsT|$G0isp$y->o4|bJZC#7GHk|$xv=HFtf6-xN~v8`U%+MK`9Xa8^RP03qe7_
zNPvFuEQlR^Oln7|gP%U-v-s+S%C36LVdhf@C>!q*;2TL6D3eg##i_@J^BR#lNB`=z
z^n^hG((RT;eM#k+bI=~Ln62(m++7pb>Gy9etd{BA$8VMeQO!<o8In%KlSI3%?vu#q
z&qO-6?_q1C0ogF)S;WOA<$0Wx%wF_%$lOrLavz0l+q>rt+|MN?M^}G;Ut4|A!S+XE
z;?{6l_U8XiV!}CY)*QQ?`4mSGqlnG5uC<-k7sm#U?eQ-42?$WnnFsqca+^q{WIybN
z&NAG6w{)6bAqO@3(E!5kBVTcdpS!vwaN>B?)>H`s?;_JFi>8=z><q9wlbD?n`Xoc<
zhdrF8>jc3sAT(VJ(0HC=;BwMgSAlRtp7qO1*+mzZ+WGOekF<1DdHKH2xPtgiV@hjn
zNdI)gCbU2*NsDR^&)8O@b5c1ty`32igzsx(T-LX4MZW`<=S7zoNEF%#pEip7$AWdZ
z&;~}Lbb*8o#>E81Dc76Y$75ZRPtIR>`$u+l^h6XOo(E-JOOr<aV<LM(*Yi@ALi#=m
z&$bI+-`=>?_RZ>ox(Ur5seR9Dg3m9{rOvlppijCpd8fzd^r;`q%N^fGBBHjfYoa9r
z4OX0_)~`?=wDMpZu`DA;&}@ZZ3jy;mE9)s)*_L1WJH1<g8lju?r#lgkf4cUI|7l}E
zmE-^L;j4%DgOhQ8T^@-te?%%u?5&ijcBiJM&iWr)TRC&|=+Wv2U>;(*i82+x+qU|Y
zPvz%vlP6v#uovazG!;718?bi2^K)6%YR-IB|E8H`6Z<z+6&18$<`6-Fz8kZ-4l9H-
zLg4JY4y*~M{->-(sej`O%J1L4St9X)>PKU5alCj<@xleLr)XQ^LQ#NdX=7v4`egv{
z)5*zcXm>xls_@UOw{LHvdkwK8h)^4J+j)D+FCWhm&3l6`g>sFY&ncBue|FUe$=nz1
zu1gv|*W+Sy6ZtvVuqYw%e6)kVOzy^w)7RVYuiatv?@EoQnLze{2Lo_i$?QY~1(~M{
z_TJ8J0V(UojT=N_n20v-064d{s!87u3~T`buC@s-hwc?*4tA3U2qN>YgPsGHLV#_!
zJRYHmTUl8N4FMV&i!gCuacAfqapP?aA_or|pfK`ygh&WTv4i%w)M6(#VD+7^Vg}#2
zb2Vt^q1Yn^_IwN0BS#W}zLP#vwUldJx}<TCiD=RnYoG32ZBHF(_kMZxnD%|@V;ol2
zEVky#?)19vO^l47Nq89>i?C)3o(J0on?<xOwi$j>T)c)*8o$!8-WD2qVX#sFjh?}l
z1r=Dm9l$?o6XbIdQPCEERB0ZMP;{Z8-h2#b2G^d$iektb5VuWFO?@aW#hs@AYODxg
zVDUhC@#qm{+ubjRMl60y?r|Ow<>6803J&gEI~rIu^uv<gp_sF~5+%9AGk!I+3@T<$
zLy(yzNwbBxAfpE<M_213zVx`7?4?QedmJ9>Gf;aZOjuR#uwMzicwrg-@YF$NbYFFJ
zb%&!4a0QEkt0RgQmS&DfNa*!!Cq~v*Q<cOS5N95b&iK=Je*T<Hrh0Tx(Ii&G{AY7q
z4a68~eN>#G!1<;^kmQIkmwosko&D`;O+T!ud1OBaG^GKQ=&6F3ux~5G*`RK=KtYL)
zThhzNAZF#_npO5e%z?18of+&EB;i!T;cGn`n<~mTM-44m8h5{&zWq~4qp-%l;KSnd
ziI*{?gAM87ZK8?2Tb$0GB~=c!bFaGEY<mQy*sI7a+%v(AQd;i6%6}}T|5y2s4G1fS
z?ac5Bo0E1%S=5I-Y)0G9mo^P+PR7Q@&aa8`fXSVknu_5!v9)19i!m2I<?+aqjOJ^;
z?K%5kH4fceOXfm0;HkDIG}@+6C<k3%sI67FaG?PdymTc%6~`SFk#`|UJrYv7hl8x#
z?<qLWP%bPSx3a1ZxRU7?uTbUd%b{7Q{Wa?D&FN$<vl&W<e35JB<|$K^2O<AEJk~lP
zMQkJyhTIHRRw^H}@Oz+kl$EV`{rWZJ7f>1@6nRndqB{ehxL^LAS;h<SJ_Emyjy?`u
zD=exbUH`Dh%t)Ib_2LLQJ>NBywFVh>NRy~x%+TL~`Js7_>Ut6c^Ia))EGHmfL{E3y
zwr#eVB^n#uOEi(BA=d&rhha*hR{~ZlG&~%EaX;EI0IKWKowl~7<k($Y!HyulLA{bQ
zh_phzhSVD!9i3&8Y|`*T#)~`=La1*BZ#vDT<xZX~0scog=)l47G7Wf(oZ=FcaD7qi
zB7Uy)liAjnpE|X&tc>JG+eK<q29lb#IN@#^B1Gqb=^>XTN@YAhJbn7~{rmS&vWkkn
zQ;ZN97ClKk`ut?5u(5WJM=<62k+yTz6Tp!}+TV${;&#Wz2C`-*fF7>LavL`hub@6r
z+mXGrv<dcGXYX#>RuAlieFE&NmYJj@_d-N+ci%Kt^g_Zp-`%|w^#Z43jM32ltJVl%
zR&D8ljU^A&opCT*9*?kfz;(sV+ufrLWlwK!Pd$<Es3D^eq0OXtjz8Z`P(VNrTMx*1
zlc+)%IA1=8URmu6@prYI5?{euK}el^Q~#|wU1URgk=n(kblUCAVvH1lV|47`%`YEM
znjmv*<XKAJEOE;H<5gb&2M;E;)3cZBn+b&bQ#466Q9N!KthvhDR!D4r^z~Lepo8`|
zCNB}<y*k|2>(>+5Kb6F%iaI&B`p!VMk3?NwR>peh5DGIyR3K$kF!o@6ss$a(*PPj&
z)%0_$iJ+tU8QKthC@PqxiC!5rA`mDg0aw$DSHmc$8YI`ixuEF{8r`?Z?`>7DrJCzu
zs)o<OA|4_L^eFJ)WTOznAyNLYvIlksr%NQH(juz>v(;+41Y<%9dzzXiaTdrkMV7tN
zm=}6jalNPL_ftrI&xfmzSXR@#Ze>XPE^N>E@ReL*KKB4T$l1w>NN2uyv06@}4}PRi
zf9s9bB2kqy3gpYd`XOg~G!viXSa9oRpA~T63#Lr-5|UBMz7tC>l4s-(<lKg0fr^?s
zEG$gGtd$*%J2!VjS=m(_kGR8(l0RyiNr0C0e3Y4^{j0jfh7loQT?}{UmyWa?&}4sn
z;-<#pvQuvy__yz5Gjcw>z@)BTR^hGiTrPy)p-8|vu}ORav?}Bd*YPPGDOxW>KK#Hr
zAV_zC8N|NuNm3JO09;GKq4Ed}b7QdyayK4?(rM*CJ*p{{>-BY5&)>{ugX>Tu^SY34
zYI7`_td+I3<hXY3{0c2?V$P>epBnE39Sk}P$4=(|tg^T7DJdx_>uw(|iX9g{Yx=E(
zo&z}&hizJ;PJ&(s$79`Eca4-^4}@T>N}?QVGjW*dD7D{Cn!SGI7#rFfSY?jV#>GX#
zcxT5uWPzoFD2|sM&;6BzRkI)ETOAXG4vB8$!pZtUnzad;$$ML-mS*LZAIBf;ArObR
zUSE+JyOf-_#R;X@dPM_rn>#xheOHj?P)=88wx~Ku6yqS0QrAUn6c4{EXJC+KrS62%
zj7(XD2H7_mJ{_w(9k#&UJ$yD+dHb1S?#5VG_8P~tDIM))3p$&<CPW+;qn%(#R4{h>
z?(q#LKRj$^_~~bi&z{rgo>dO*HuSO!(@uUukug9$A;$O+lWFjiEr*9grx#av2Q&T-
zk%{w?QqZL6)dl!8QL@~!tel@bawK_;sx>ih&n9GUc<yd#R9~uQ8dtBfGRLAea*!h<
z9-P!5lAo~!@k+Nh6(<y)@O@w|boG8!2oGSYPoeW%e~s|r3o#Dq31Oc%T_B+YGaDJ%
zgUBqrUlE}xR3ToCqY$lj*guA}bWg6QiKupn@Q8iyJJf`**LsEDmlx?G#gm!i%RZjT
zvN6r@^Q(W8WCX+L0U3nS8B%C?G|^pKTW0{Lz>*)Oz7FmeIk@r8J>rdmf@3&w$g|j%
zQ&1=*L2NbU!1X3+Fr<&e_m!|eX;6AeW+z<<hN5V3ARM0n**BC6O%Q5fe1WJTdO5o8
z0|Y63+R)II8=rGsSF0%f+R9;zq%>Sd!+2!sdxv;)v=f0DJt#rGfYcloUEd=DVS(bz
z6=u{@s{st;v!k=s_60HH#WZ!1+=4t!MmMuVHvAjmQ&N?S`@$l-5H1*8Eb*l|?hV^{
zn-Hxr?|)N|Suy=fmRlqwRBDgw@Ax+YC?)(!S^1vr=!0Kh+bCJE3zzHY<<s><YLe3a
zsm2Lr83u-n;R08jbHn~KB>41hXjt|9e8rL_KiJ7Xo3z06-;q|B??C-AAg!C`DK%e4
zc>mq$1d}FH8rwfXAi<;Z>c=ZrR<f^aRaKc8hSYxVEhTGD3(7W6EUo<}`R`6!ZO`=f
z-s8XiiCgd5{gNe1hkpx;`eh(#cR(|zekjtFuT1WpEvIJYnW;KKks8u!kH3qL;%YW?
z&V761bj>p?gZ|$TnN)X7+mdBnxSz|?_R-Qd_J6aWBcx|;3#@hM=ee$wuyd%;Oh?Z4
zii!8@q;XlhTm93*CTuD*_*&w=QWs9w*=c4d^!!7j>YvXK+x}PWE0r2QyJQ)Yef`{P
z37s8Y|8Aze`L=Ve?Px56$k0*Fg#G`vC7W`&WYgV;50$URdeNp(m}vfI=^QD!LaQJB
zKMvOK%G1$Rs0!Biea6jw`!*Z77w)Lx?!OOyb~!$NO7O;aM$vdv=UnPzr_Kda`!hY9
z?Iqv&?~vSIVZ}T4`3}^TE?Ih!{k>5)<~@FSuD{FYXN`=nCiwSS`m!ldrs^D7{!YeV
zy!9gazqeiCIMT}~S>Kc86Z#}Zz^j~tUQ#d1JT%4L8gFa+x2<lyJO~LZD<_s0t#|P3
z-sZj0!Xhy>wf@aLGta-b`u-8^>C=;ih5o;~TPRtEn`CCD8=f$V+dQKh`p?8}soQMt
zh|s^?;b%B;%)i)(Y`${W!Js6+$NbmXv<q^teLwZ#Otgk*ul1j7kWljG&1MOS{&WgK
zSC^*w9z*Bct{wkQ!=IHpb8X2l?<T`|pKLY{jF9V}kt>z18=;ehg}Ta<E?Txsy9LXp
z2FYMe&7tEb8V~>HCP)USj7pYfQ$lEW@BnF&78IQA%5Kq8SK9M$t|MXBCkiG~C#MI$
zz91f=7$bVr_*Yn?+0K6-<M7efpK4*rsy?!VQnmT*jl03YpQ_|dv`(IP|GPg@7Oy4t
zU!!1>Tz4%u#^a=<#Hc|*a{0$#@{S~p3q{wIsoT9bNt66LSlvG-*z@}C;&5OW?ao)%
zK5Smhx5b23^@$U<A^Tf({#mz;XBq|68Js-IHy$)ESvpDRdsm%2|IeMIXdZkh-EzI-
zO{6$Z=9VS9<~#MJHVxcA`M>XS8!300k^9zXxOZ1H^O>%=b0uN?yB!@PjJ9mu_HVCk
zuas9z3z{F7xD!y>KVopUD{EJ(rt;693bf?916%*+``QeRuTHG|w%@k%nl@9KscO@c
zLz+xGvpjls|5<rbY~7Nj)!nGdZrv=)=mI@);$Q#gM5&q_oQ0ZvsotFNw+W$Z-TH5d
z65(^BrKwt~2BwH1QFH(F#1mS?wSE2i({-(l(tWn~+w^~&nRTwO0&%cfiz&@|5MhSZ
zd5P$pssDUXjdpbQdQrNC@MvlF&12emyGKh;Ui^>sY#@?yzq$*=9q<1o<vjhUsp>=S
z&h|!;_6BcXUyBqcRyOm`<)rLccmHa7x#GVslu0tlKYFzEm_FAYQd=y8`QI{H1FIg$
zNH6RYc>N@>?c?##a>I}J;rOPVD&C<dBmUV+lzCSg#r$+ba{z<wGpbgt^}m*S^(HeD
zwWP(D+{a;I7>tInhRDpIbR8AtCb+GYxU6X62yK?%deNE=)^ADHIJY#{21~B^vnSno
zK-;KmA#2@x383HMN98W4U%+rq*@=k{Y$Zr#d@blr*VkWNveeTnFZDKb7HT}SZsAcZ
zD3O4Pt(27?kECUW&GnA`(*j`jne|Cj)PZi3Erh7|wY5LSy2`sVy`!%<3!TXPAZ$%)
zQg?8F0GIWq1AkUw0^MYF%GX3RmCzj^%6sw^wiDb8e)Lb?{`F(5vy{@OwY+k)gZsH?
zj%E8%;mGuwFND1zONzuDAMrqa1BSvh0F!^YH=k@EEUdBazNz%cwBf}{>ZLaIaqt(a
z@mIUw9oEhwX%ijVwJyt-q9Kms-mFEh*Y?qN!h6@senj3$XD7|=+szDVA7FU0X{7Ct
zy1GaWnH0mH6_XK%X?AZSC>;`VU8|d)ydvos&wOTOEGLHldhFY`@E)lXf8Ti?U2eg6
z^nM|WCyWTCP;C1fhKHoV73AuqdC6n$2e7wAOM;-ZE#pbx;yyw8NFn9=B}+~XYEe;a
zbi&=^K1A|pXk%mpBtrRK1jdJgh9Aix+b7tPPM3N+oC^+{`)TU5lAAxG+a4>A_JmfB
z1<kwMC<pJb?d!y36bl+aXwRJtwMNsq7KIrIV4Zob4N_wtkqK8954(OAIHUI>pEgfQ
z>BldwrAb-)-%I(mJECG5HxXqUA|D2)FGPs&TMsB;q_Teuj0VN_?R3Icu|>N#xpm1O
z9r!)zZQ7Bxk%=k0P1IXe$lAF5r7R5XCGLUZ&pn9haLotnyJ!{I-2#Vo_N=eefrXLe
zbdlRXE8kzcNJx-|%a|Qm-?>Pmwxo-ovP7AGp!!8%ZJ`tfZ;FZ3yHby$dcEG<y!PpJ
zG$!YYG6`;7RqSFquWvFe538<cAb4~vO5>xPZXExzrP-$2Hd*g0Du&vR?cixnFXqzB
z+&Wr%_wK)W2rvK8=e0d={T&yMKdyq2X}>+H`8LJwi%{)xivJ`|S`N%sS#)sydLd4p
zp{e_zwl>AG{&|d6=8ttNPO$px%I_zRpE~%f$hkj-qGFK7-tIEakbgb4(7whrKPpuy
zo?Wl`Kqn=K`7E|a+X%XWa=Ee(k<JZV7NP_`Rwh)oT_on(TZYKBN`<d?Vt&-^s}q<>
z_PrPa2k=b(1+g4;_wrQ#x$-ldoSe~0m`5*13r$#56jGghz)D=`YM_YVi+}yv65<x9
z6f$1FzSfb}4FL;OkjY%$U>BiOd$3=)8RK=;)k&asp<e-I4QL#9kjk6d=M>!WEtmn!
z+n_a=lM@L>6?9he_a_9xHP-d^r<sNR_^ZfP^>Oku8Gh&5BSaThR};QQKTG5kp8#!$
z$|P9!D(~8KKZ##CLQHdibaWC#j$7UDK1`>6?|m(omV9eqRZ4GU)ORK)78aJeL=J}}
zsw?JmuB(*6ixtut8rl81xw*pD13!M>X-$x0bpT{X^Amzw&^Q<k3I6m+Gd~#|4#Z?&
z#9-@?z2sT3U{(z}8Sq}5ZB~C~(D8?R(2WZVI(3i^Q(!hWe7}{CQe*A|*9#F_4p!D!
zOp5u+QTxRSP13J2-xHaUQCs1-SCVLSC-!H!V4mb!t;5tQnFm>}YX>dHcZ-S|-xy9V
zV=zrsKE!08yEvGOj@%Ocx5JbqGgdkY91zehDQ=J;b1%U<8-@1od}Z7qd==dv{nliS
z3)9^DQazu#9`^I~9U_y6OkdI~*Qe}%HXL~Z+7&425W^8c<nhdaJgf4K4vd7@;n+ZT
zV}1|XpKxgTeSE6We1lRS%~H@4;0$_Fv`(B}M%OL6%@Ne7n3!16%G9~7?z@<n5CMo2
z)X-r3?8+X+ypnoZiwtp;89gx+M!K&rp9F~kfdg?A^?TWNpHsYOilJ}bW{^JMdOwQp
zrm@)u_ch8$aTZSXN3V~c=jqOW?7y!wAYj%e=qXpP@yEuCZ49QmN4}xNwskti_)s=d
zuC}2eSdLXhLPTvl2DU)<E$cI`Z_N6DNFPd9ObB5#3f|D;b4^7@M?rt{)~y`D6-mu4
zE!&wfk%cDQP2w{PxKXa}-AA>T;U}~VzJ)eYTJrK$Z(!-iE}aoXJMy(=3aT)ur&Lss
zN?ewP59cRCbjcfP4&#T^5QI61e&0E;;_@UI$tc3)`}edDO-;9&W988<bVVQz;?x)b
z?{fc@w+jd;@Hyy3sIL{DkOmimRv@G3ZS>p6&_Y6M-bNIf4<3VzTr^G!$hPz2*&Jrb
zT6%gf)HvGia4FGce+zhuz|x)Hw$R92H(E+{Oj}LNp7#o`Pjzve*Ume!UTJtt31(wJ
z0B~+8AU;!t(xJQ;J%JHJkIR%p={qEfv<XZxCzBV~_EoPsUO=8UKGqW>#8U3ZyQaym
zjef(C?2hW{5y<=0z6v=lSZcqA_dB2luYO0Al3SmXlT$KBLqsJp3ppNvBOMYwE3x1j
zMDTIQXM*~uLKFtTA1yc)O{Jl_ZmC~Er3~(+$=^4_f`3hBbDiZ0!(8&EfUFOrrB(f9
zjrGd=a#>ROlYYpga1tS))D2W%Jx}Qq!3tR~#b7vb<aX85rF5bF0pEphUqsGQ$w>x9
ziz1sn8>O<&yRr?0YVX-zLhCxHQsO5L@eet1wSXAqYptbf)-=*HI(;X{>{x0@`Pof*
zI;-FTKvE+Hfcx@FbZyK6vyikeTX=KpS)_e5^~~`clXW$v3#O1$*01Lg5ty%18n}cf
zpFq8Mo}?3|0u^&IcMb$>v!fk7P)%cm2W?s{aIoDk!oo)S`iKKfXnteppZ$g4UHT29
zr5moTSC&!giAu*9_U&_o2|<DLv7&<0xV1%L_T8q0a9{`?=p;Zga9Cg-4HwADxo8>h
ze4F7ncwz-^fAo-esDglLb1bCjOD_$mw`0fvLM?b>P}k0-!-%7z6>I%8&%F4L(5<76
zq>G1AwTF+NaHHS6zR3(Aj=7)~TatV6_^~o#8gv1WQxjWs#Jv@=jH<meeElL#f-)Y=
zAz@U#7lit%sOdetEirKzL^edB_(G^xCLnu6YZenXJylim^K{t02kPS>t1(BQp1juX
z2>*y~BlgqdFGSunHBDUpA{C@;m;1iLJN1^ckdM1e$w-@u=)1LwuMrh(A-^dIYhU6%
zTgk-jZZ>1z;l-Nu*jb+WoU%gG@s~w!t{$m*`cv!8gQm*+3M%ebIWHYdB$3(Ge2}GI
zy=#1Fxv%z1PlV$;jc|d;`O)Qy7W;bfhP+AshC%!KMJH`@b&~9pwC`CwD5Y<-($on%
z;c#&B8;IK8^z?Cv?>uHF2;V>!>i=p^D13+66`!w1_EsY|5arL1hjc}!r2Ne3arvBK
zxb0!Z^u)w{Utf9w^N<O1$lZ~PDocXHuim~bAPmmm8J5cW8SPNw!%DaXH>h36s#S>7
zefPuA;Oa~--9C+_#f%F7>ls?it;Saf4=JTdvWrB{#o>V;BQ?5@S^Z6p(_+H=N%D7|
zZk{LcNlDc$ErF-Tq4k7h<%#njQy|sV?MCkb!KWS4I<!zeWbP$BEhE!+u0d6EUQt(9
zH%oWn&c2uP>av)FFl}J;3_NVq=H1esHZ_+PDoAm>X{$7#z}f{V5@ru={gWr}YLY#y
z@Cyi7wbEccJq)D;>&rWLr6!Qzx$l^^rj21{7JMS%+m9b=CNo?=F!F}64T9YH^UlAQ
zY+2IE2d9yI=mjsdK2wGiRLU(WFHc%CBx^`FDqcO$#kuPDGJjecZ6OP*$C}Cdcb-Rv
z00T#W*u$%|_x=ks1vD^C7W^*j?%fT9AN~EHi+$f_cA7(4v-^FYqaCYPMIRKr8>Fz~
z(0NIGCO%q)4DztomrDR7JpIiDF3AwDKD6^<BH&wT(a}g&6a<{%+wAPN_I6lk%C$&D
z0EBl;hCzHWG(7B2%k!bC>WF}V`pozUG)DY<d}yyryZro|lkVm9@uB2uE@X>4cI?27
zG%F4qm^+q~oJ_$W!UrdWdyb5ZXkWU7AwS5}3r-dvV%-#%be2a*v#wSF|JSy?bN7i#
zb}$Vme|A^Rx%Zc|CdhZ0^;vGmlnHaIS%O*;p1^?T+TdAVg@wOyj<&Y8#K%NYQQ7ua
zQ*i`U2N%7`@*zI1f}XzGI1`Zp4O}ZsD<i8T`anU028UPGRdl1Msi^@zkiMLn3ab+s
z^6(rAzw(U)fUTvY?t)P2@bljoy7=M42aLrWd7JK0)9+w!A0cdgCTp^P9qAO@9t(q1
z2+bvkHRv6VDPuYmVT{VT{SlN+p`mn)jK9Xl3K7!d$pcmsDRK<A%@$MaSlcvO$~+_d
z$GuWf7Dk}yhj~{+j5}za|A{8{BQb0J4+sO8mXwx}fjW?yTkoj(D<yU`b0F2kcWiM(
z!};sSk8B(qBV^|{WpiWb@yC1QBQo9ZI1-$tx`qaZd>T7CFsi_Ax`s!1$<L&{e2Hml
zgUgqtqLom&Ld{I9_s|(f9|Qy3qy&5+?1PeuD%)mQg#e7Iq^kN-3DDPQajG8*t<Kk1
zHm$t#JEeK?Zn*&r;!%<n8oO`byivJ$5$*e1=!0r$bs|Y`*;s=9K~ZDQuWxeBx$mXA
zis?3wN{y5*E=`e>Yg-zVWJ~^Ce<JvH+yM+yUt0PD(hLVD=LWZ`tE(8EKs?vya@*#H
zl=+bE1RJKdu<;n+X>4tUx?yHwVhGb91bpeKGiR_~Jom{pPdIAsbZ+m5#>NRUT?2qN
z2q1x~4MN9Y=o_T;^(7@FiV(Ly%%Q=G=ceFUVxW?r5MD?}H(ukT_dg?2DMZa6$<CgD
z!N;?4ULGD$;|B3zn#>u4<7hDBTroKkf`|=uTXAK_uiPIhDo-J#;5u~r3{yb1ZNJwJ
z)Kk$*?+J_;rZ9T~h=&xFCm~)npeLV{l=MOBHy{rtIZ+i{IZjJB5L&{C=0(>+bF(v0
zg}f*~o_&Mataf`0)Z=>nBKX(WLQ%GT%k*g7-`#)zbjK-7h@oFMvJzEP<l<XPsd&Gp
zAEn1db7GCtHwk@x@4x^Xec>iTU={Qhxa)chD|g&UIOpQB3d7k+z(a^W3`I1yx9?n^
zXXo`Uh1#zBjnTQ6=~LO!;;v~~)`KP^rF3)`Q?)NH_w)!$%WWn9wo95MTHq9(Cvxsw
za<<7`rS72_*P9H<$;m-MLB(Z<pvi@7zKkBKIV8teXCiL!p@Cw>B(sxnK*9NNH9WAa
z><3!4K$YhU8a2L0E?M^W9uVI(AUWGRQ17|M<tar{<+VW!&h9~(L<1MJQc_c64J}jD
z(z1v6&cGk@*7?rMw|{d-Cnqm=O5k}MC_lKl$6*-&V7~tTscQ?i)%EpT3Fpt(!Mxi#
zF)TEMuM7!T%}hA@?0cH7Z^c9Rr{vRw_v08LD)${(oZfJvmZjD5k42J1JM22sF+4o{
zB)K19rMtVkoF2%0-`@8%m3fbXf^2a*@OJ=+HVDLBYa1M<2X+u}G$(JHIq&pV)zb?&
zO*e8XF-@IhyK+~~!>QrGEQyVW;=(-1l2_*?PY%`}6BhIk*}cA<%pOXGc!=1&{a*^~
z16z+Qq<Gyt9~>N<p7|T1O^8)vzrK6^TnBxa9pvN?*%6<Qfw`ak7@at&u}gK0d?yN<
zBQzr+xjLSzuSPC1)mM1;q+TMcCy(AqvdIqVn}k`78yz_`VAwf0kP21c3^@&342_J8
z;C0uRY{$Rk$?O)Vr=we3S_+`&Kg(B-$3LCfQvs#q`3o1I@Pjx@XOH(%{-Y(!OK%(X
zil23DL;Pw@Nk|pnu>J+D5{<(%@TQFfSP==yg1M`tWHDSTUVk1MM#B5c$01IHnCQUV
zvloZORkg(*f44*(fs?D>iMgbTKK|d*3_dz$W;AH6|N6xnb0;8FtJ9OLU-qHFOn{Pg
z$+{y+leA&XZr_oomMA3w^5p&OHXJ}OJa{lhi)(51w<DnuAfUB%ujnC463nZADw~9#
zk`VB`380cECKz1u_xI1v$`UaD)L2GlOFm=LAkq-pV${r<w>%#=+Y8zGtCm+aT)N*w
z!gGFnp&Xew?j}n#L+RPLm>4T?I2=Kw1UU0;OG^xn?HR>M7hyUJGC#r)ir*G)v&)wW
z1gK6t*Sd}Bj5D2`o#hEVPdkWM7wKdR;eYV;CSW<PZTt6?F>{g-Nt#TFC=^MQ22q()
z3W*}A&_K$VDKrozNi&ktAoDVoxg;tYjEO`OqWXRpYrWh1|F-Y{-P^l8ujg4--S>T+
z=P~U2e(VPvTx#e>M5rYt3<68&Ps(ccv@h7&<{h*BMh6^iu8mCYVYleg;30|i#afxk
zE@qFqMW?j;vgPl)=w|QY;)^tyDkkWmqne6JHzMJ!T|MDQt#HlaX`*RFqfI&9S-={g
zsj11~neXg<OxBMb8~V_Z_Q01YPRavfAc_&rX-vA&|D$RcF#_MnjMP-8X@|W!?~v}>
zcRX54Jn|S-Dk&-H(NB{h16?(>56EH}r4t0)4Rde(vg=`LvO9lE+s@ZoOoOhf8Bn@3
zJe{RA`S+lM!Mmjt3)cO2qjsidW!2=x++`L;981=Y@MU_C-eC8J%4esa^GC{LyCF(r
zysw}$xq~u!?AWoxhpz%yxv`=1j@K0xOO`F0S{jHu!*?Dz(J%1A_M{{yT&?hhLwt|Y
zj6YZ6hiOxF{=>DYC+Ex^$`t?Wmgdk}cPHO=rl!yK_L;9?JM6mLMoHf#|Glja+-s5T
z7At(43-svQppvyC7*!i5`j_cm!WoZYSV5o2!YG8%{)nXry_enDhqjXW=XbEiW3{x@
z)zxvwZ2Df)Lm)p>AtOiCyqDv5_|T!bq%>q6=<Cs2wQB<;G=;cLHebDTY5uLBS!QCU
z$P5^))L=Ya^k0%~iKjV7jwT?yzvXy-QEh%=_7JkdkRb<MToWhJvc^~?wb@e6kxHJ_
zr@)weLIgN@@??hER=I666HX^4(iXb7WY>FoCaTBet>iCo)fcp0j&!b<PqH!C3I|Sg
z-Ou;Mn>%wmw6ov&Q=?mraml5}r9o6_&;7=3YAY7l9Q&q#m#;3(@bkO);AP7Iv-|hD
z^}n#KL&l3UQU>iheEoATBo$9w@rq7g7O-i?zI}QVm10iKJ?)$@(}hBP`?SA>y5jaH
z^8>@99>2^q*BjRP$0jFVuPZ)tlWiC2F6&^CDy3zR+~MCawf(A9!>Jw#r>ry94J{0q
znyCIaf0A)hy2o9#R#cugbo>#kd8)|=^es~z^G*)a(3lc7QVgXB$Nu{~73;Ms>fSv*
zwq#A|4Bytp&I&qL?i6KnMXw&7M@Uqx@Auz>ciccorQ62URW7hFVU_~Vnr}S4F;+_O
zw=rnr(d={aLDSj+{Ui4H3&&I4_CD;aIA_-1e_MW@m)8$E+Nz1q;>`N+@G<!7H^vbe
zQu+?H4g$gf(yeeXJp5Eb!t>f*`gS=S6EKkk@w{t)A+-Y%fD+LVu8i|mJdqHu+1&gt
zIAnQ!MflOVYhGOMr+wzO;|?ijslN#~#S-N@RTXd`#i0Cv0eI#mojjRe9AZ#iN_|VO
z*z)>zm(&e<QGTy)?WSI(B(oY4I`3%7e08O%dC%uF57f`ri?-(H4<OO)#pD0};Qfzt
zbDe0aZzyEjtZ{R5Gc~<OVd6cp6kLnyFeWBu^v>z#m{6|Ct+XA&6aeNq*s!Qih)Wsp
zu6ognj4t~J9(waFwxQzt=Y<)YydS>I){n^`avB8s|H<G5--?Qhb;gf3&csp{!5&sE
zKL9$w#yWTE<n%K;y8E{K@1QWOHbfyzv=_=WDUv<kN){9te}XuYa`yZ$$7IM+!81hh
zG$%(=K!)qdH7q>d)A=rl)4kX}%1Q$#dJU42w06|9Q#W0ov2^XZ@S{ihPqCd!Tu}Y)
zwyAk?W9^Mf`REPp{+zeU(}jFWw7o%Z5D#d*$>A2IrAfzRq)fN4c(UTxxuVb99N8le
z=I?nNc3`(<#yR^A?TnMZ%#1616h1dN*eZMTq<Nmx6JPyJPEEzg)ELB35YRqbTUsh^
zXymIT#xZ8{gmCcenLNGkfdh}BJ5=uGp+J1{<N|%k+_}$CiP9=jK=qc9x%%tNCA3Ia
zz3bxjLXu9M3dd<^=T0M{4;j>O*4+8?Ddty#nUVkanHiaxbhxq+CTOX4iZ3I;n8-IF
z|1i?9Z<cI0=+i}p9adZ#I&7Hc(Fd|e9?v%(sW@f%+Z@tt!EY|(cx~-CMv4_C?t1pB
zr*p!;2@hpoPl34+QL|Y>!+?4ogR8W(G`<pENX2}qftVhm(9e1l?d<FvKQqz?MN)08
zyI_C52&ox^dFeE2ys^(t1%-u=BF1T`s4xJ9MT#f#_Zzcml^Dc%fj$U1)!y6uSc=1m
z-nFqA8RF3Ej#v1(MaRc?5a6Uj2V8^A($eZJFxj^y`$@!el<{8qPh&grt!G_#xT;;B
zo<4GR!Unf>p_Z>A$E{mQ^|0L1(&o&U{T3Txdc=U%4&O@k30`>opwP>41C?g9P{My^
z!sBBTo-)KXGa(@^4uF;?kI;{KNK;kiE?v9gRYren!#r<XoXXzYIyCWM*8w*f6UwE?
zt*pQ3I{C%wM#g1K4qjV!x4W8}TE$S=z3<A)Ev&5C2>*ZGk>GC>V}5+%PwXqvPNpO!
z^*pV!x$*PD+Sj(2X*$b}kUg#U>MBkt8k0g&I;#&ibNaR6=aIe(BliVXe25x;_rQm9
zMe14_+n<DO{aY}fpeB7QriQovci;T{^{cJ_o;~--DmyH%z}PT@PwTs9@7^qz5(Gim
zf=&xSX|FYITr1<DbLP%9vDIOcP%cpRkBQQeLx-@JnVj2cZMF>q93Yl$_3iD;x3=#x
z?J=eXheAUwk_s4-OlmDmg1KxuX}A~Q?=OTs3T-bHQCt7W$XC02sgYP<NJl&6E9bf4
zBG5$;A4P6SKzPupQ&lT|F=B7{c~(0uxrSG|aVG}t_`3F>rNY_iJy7Pk@9onzutSty
z#@l)+13Sr)(SJ&zA;X7<Y|GELa;H%>?Cb-S&?Uz6MnM6-Y0rd%VPT$(sHH|H&Zz=I
z)>&VrTwGaIW!jyPrjqjLL!HQ6q*a2#G?1-NuE3j;vcgkX>*?9YsAhX+<^~H3sT~Ft
zYsaJEj19UR)p7US`l6yDhi(2u#4}~>okOvCda5_y@a3aNXP2u!O|>*NJt84->izr9
zOP1uH^xSgF`CU_k%J&+NS1-?9)HRUr^fz6$so1%;VhS2!f%m`nN>??$uH1PZ`Q`=G
z9V%t1bgyQY#zs9BWtQIK)~r~8&RL~UQBUvqL&vt)83I$DyoP3W>C!d=-Baw2nF$cU
z1I2)IuVU@?NuuAOaKb7Hq&ZCcXoqigbxNJ;s%CstbLOtf$;lCaAG*KFG_kdfbg(+O
z(=?9B-)pzuKOLP~v}gK$iu=%}f`DH&veMfvI{ppc+x?Y~we=s_dic;_lQ&|F3Jq2M
zHKy-A9=4!AbLRdRL(}%#c0Dw#;qcAJk86N^2HmC8-PGE0WkJW<{KcT_7arMlUD!iI
z&M>4;169aL@4VbGZAh2I4YP=@&U<K$SD8F^xylK^<nFA>ho{lJk()b5`}%^tMd)PI
zf3ZE&Bfz5Lza3#YK)~A*$fMu-xBH6eT)v*cOcU9Frv(M8pTt2*CC#Y<n8i8c<WiKN
zOXYJ^kPV+uY@gKkE??0BMFN?Q=a;SmJ>+0@A`oeH(DLR@Ozi<yAcyZi)<ZcS{V=ls
zzM2lZ7F<G2&p3Qkl-=nEr`p=vT#R*I{w()Zr0NyD0UPsUgKyocjhXIkb=*U7yH1B0
zO9R9><AtDo!l3pQ66NQY=jZPHP+V-C+PL%Z3SbmUO?Njp(F=j(AU-dg7HFlH1W>>E
zbXTz%#htu!K*5hB3qs*ib~Vx<`d|Z<V=Yj5&~MQXGDrIty(iHyHGNZJ3zqGo@!gRl
zI7B96DA&II4yFSNM>-Dv{GS$}v~=;*sWBM|g(LHx=5tKlO-wYK)&`)DKDTAeCNHtD
zDcJLG+-J7By}*xWl)%^;{U)qWZQQkQbWTdI!o5E7v@Tt^AYA8GA(Ev|Kv{@hcqrh7
z4mS)dtR8Y>sT0n&+{8s0pbYjCZ8&aI`_4ouj9W9ndyIzBj~AHvBG&u>=I9VG*?9mI
zDt6k1#|?iOZTxBFRHFT}OCSBFn6Zna^yZ$P|GZ_~c_o9yWj9soDLJ1!ABiUd5lt9{
zqs$T=A6hKIi!rnH>oeX?hvdt(bX~df#jVo}Bkvzj|LHliw%r6s0ylIOOFE_{>0v-v
zd{WShdW^AKtNC4*RjZICOO(?gir!ua`YWF<>;(@Cex`b$Yj8*ic2wVsF0XTuK=|BR
z+pb`-MVHQ@t=}il9X^~;?3AZ_ZN=j~<>kR8S)+ZXSFYOjClKSKpP0BH+g)XX>bEgV
z9371w4V=8$rgk<X0Kh)NpzMn5p}4<HEwrZjt^<!z4Ro*+_TH8r-er1dSXlr5{Z~0T
zxx6?xCEa%M;>}!N0llJG;bVU4zJZghi{or<Y;fJ|VReDe<4Mie&8MPlCC(^zS!*Ie
zeE9%p&wfF|%iK-XrcF<<7V6e55PMsJnj;NfA~REa;X-|W6^2@O@j#~t<TgQDgAN-9
zUv0X0@nRD?d>EZ7FkOBauEJSao|nAb^DHe*&5jc7Ztfg<Oh+X{@7NwwG1}$kYR#M}
zHZv(wf1=%!=U4{LhXv=B38S!F$y@%I|5moCuAA_=^va?$^K5*4nn?5#V$!<NrsknI
ztc?7+jXz%POY|ufy$B$}*TzO5<II(QBf~ybSHmR=_EgVHV}fkzfe@HS9W?n~>B}|C
zmi5Bq^6Gj+!yTtjuVT7#jPLo2eAh)UUNFhyuzdO4Kd&or#MIw_z%MU(rlil}yFz~6
zidI}&Jvb~poJvOI@4>sF<Egf;LG5!iD6*(og9k9>w~U4xPaBXfh9UWx^wDawj)SqO
z`dvs=v2=Ep`CijH!*_i99nQ065xSnt@BaM{jUo>$lXY*J=`6SGmrPUuUqa`TT7|x&
z?8O^3ZL&ea({Za`l)rvWO{<CVH>~@Sqem^13aI9`2Lyz?T#u-mSnyEy83RSU^;Zn9
zQNvJ}BO#FKtTJj8)G=jB95_H`R+a)0jBJLunVLJ-bhFs9zpkTswVX5O%-OT5mxS2l
zs6KHiDSqEx8+v+v$8vq(fj&}F2PqW**eZ7Zc4Vl2`am7f9UyKB>b#HCS?P>bYDdWT
z?TZ({tlSCMqpT#L0qT=YfgGTH+xrvG;PDul7&2rCj)8nhRG?R(Yxwd(O@B3iFQac_
za(c%$<Om%*1|X^yj!^7U22&DH!C;bb|K2_3mDQyrIVI0O+hw;<kz#e8VteWE@#Bka
zY#zLGZy*3RG>8Mdm=kJGUnjNc>Za9a^5e}lH6OG+WUag0^jW<5ktN`;|Mnwnx454e
zq>UjIAvDd$rU2(C(OVtFi_ZihAS<gn@fn=~|CqO>F?vzn>J2peNS)*3<H47(SY{lS
zKk&NoNTxGqOod$r?gI=BZ-cv|XK3ROL7koL?R)m@sZ+pexPyB#=_&xIOk2FTl=!Ur
z^H)#O(6P_OwGNlI48g{(2Iv5rb3hOra@#q1^6NHiIF73%M!nz;`1jwQ*NwTTyAK|K
z)XZ3?2Ni@lbz1@R{NTX0+ky2dYALfcY%UCvyN<j2|0>Vcoc#YN&qf6Mo_P1_viH-w
zcd1yXr;kScfrf*_&SsCjG%5@ilnwJ(wb;E=rw85oSN7{i)>yrISaWl>^y36-F}44i
zQ^+L_1#fGBy!`TMnGE~LP4Zc1Y-~)+086Gn2yE;U$QwXSibkX6IaD5i&m#B)fP$qN
zqYV-e7sC`4E11Wodj>#jvxO4;1b>sB7lVo*?a;_b`A+4yuUob@ZCIz{3ldEIw&;qF
zR@i-WW8=Q#*4TJ?jbCW10Z@_{T{e2yusvtwi5;?3bda#XPWT`m`coMuaWZ!uf9IY}
zJKlR^_}90&y)<2@d-FB`n=nyD34=Y&)E7shqGmb2$47Klvuo$EE-uw*4Y)`i<Bx%*
zGFv^iexsHxNC3lE#-=v_nlT}nw`kF&v@}gTdX<$%j2zkc?OXKIkFj1;bX7cCujW6V
zd_OM!Uggcv<5G$*1TAwcTzVcR(>L_-Jn!nh1iTkF%0hD;I0tR6)7InFX|`BSs}Eou
zCoU~RI%7KFY0j{@gQPb7{*jvTVXK_U+_{HtR?zdMH%_xy_T=L_=@mqR!(m}r_Xg}A
zIO(F}LpcF&R76B>QzxUUcnC;~u}mj{VjmUUuyEDu6mHM${E}N&0z3Vf=Y4xmZ}4Fp
zS0PRq2bKPm>v*1ln#W61v4Zzz$Rqy+$%NrOhn#~@V0#<bJ>hC^KS;peZuW@sFc~>H
zY+jrNFZ|6SXDwlhuK@@c79Prfn^6ocEuh%$FIlbb1UZg1zys%p8>qSB(SW@3=+w{&
zFg!}J$g&0ot7!DrY+0zKVLk6~eq(j*=+T-@W0B@vR;m`h?oYq+>$7~)#fv`8({Czo
z5Fx3mb7Ec;MvWfrG_Ah7fL9QNfQscB<SM8vb1wRfX589P6O&}mr3xbYDd2};YE&bF
z)WFJZg{Aj%B=lKAkk-l@$sJGFm($%q!rDkOiZfhAA0V7aN*e0YZAVa0Pa%Dij*oUk
zbw$NLIOx552g0f@3+WO$({ufLpnBQ}hEg?ae`lB%c-Dl@KVy-RdAR45x{dXTlAUif
zt(8=CTKk9B+^!Q3MikuC*kEF(7KkA&fFwNJ+M3cyBf~B^J`CTtZy%-a(~ssCcMq1h
zOd?J1PQn)3fQTU++uTR^iSXc{1|H<*GFVS2Iw1J}V-F~`j*gdq;iTD@+#>+SgL?mp
zrMKbTYHny5cWH9VkC+oD1Oe%dtE%s-A1E_$=MjA(X!z#OnM13R#gdgkE9+C~u2DB!
zubJE*)kA1A_t`LwlyJkn?H)$<l~ybWD)0Y?Ygj)7jQLbzA~P4t4b}XP2<hu>{lmkp
z=gcV}eB~bCJpzt7*!sr!ZdjoZXw+5+uC@zJPj3a)CFP2C4+xkJAC?tp!%2{5Kxi2H
zqCM`@IAYkaQy-KUITl^EFgBJHM)mC)+@QY`<W0P;@b_sVdR6m0dC`g`OW=lh>8^e;
zO0>Lu){YW}76?va>9taR*s9-VpR>TpPhtNM9sR9HzVquBt9P&|X30&N9s>@_T^ZAt
zF~(FYt5abkZ&xzj-~Z|#5Z)p{Twjw`6A6XBB|hG<*Il+5!7_+BoH?wYn?yZV0!u59
zOd9oAX>ZSv<3Wup?h6(#9{92FgvrcskpnjoNU<TIp>#cQ%<9Va>7!wt4#>N^T_c4T
z#1n`<ZThPE-w}VjoYI;^rq%o2C@eZifM%I@vG4Wq*|VlM<qUF~6fCYNCq?s`{PE5O
ziCoO*g@y5PaabWs>Xu#ev-3K3?AU6E&L$=A_k$0sTF4HZRZ;I0S6z}+Q)*H@HRjoa
zqqE<=x!EUIbC;BX=6?@!f#`z{K;sekzM{g792P7+V$7I2X6jbgTafC7uh_R+#Vqa`
za8ohwOf>Rhqx+A=m)iPQoe1l6u)rT*iq3Zw?!lKxOXm!=8Zc6<z=!><z>Np~koDlH
zyFWoobSDtwxp54y$?i2wc#4GzhY+yyb5+$n`%zvK7lS=eg&05PXoqa$Xzbd5@Yt~t
z6J<AMY5Yz-E^!%Xdrxw5ZEK5~lFx@1FG5Q%Sf+Yv7}ZEdH{JY8Aj#hY{}BDID0lJh
zz)$r)W$T-AnyYNQ?u&LUR`Q)(w<2g#DYPHCIBa77o;|0S-i7^$)r&#XO^YCEJ%bbI
zBXhQ#le$jAlJIB67)s_eKX31B)S8D67jl8B8qC9Up|o?SL@sR7+xtIgj43q5+O_9!
z>kkQ;&EdeKIQQ#imlEyZ7qg~M4;bNc0V2L~(A#d8!^7ovO|%$fER^xx=y*u&Iyu?v
z$?ejsc;xP@`tnk7>b*x%8hiKLsQMxqz4OMuA8^m*%fBfE@YOO;OB$z(|DqVA&qnbx
z6rvuxJ{d$Hh&}b2_M6~7ZB|@D0(P&x1q!owh%%z`{%{zGkQCcj7*Z;Qw5>QVj=x5P
zhqv+1%KA<SGSv0Rwt>e=Yv)TH-mS^VT=~pdW%Y4aS3}A;Jlx9P;tfoj#msqd8kyXi
z@IE;3tzi>^eEX~=xN<O>SUW&@%$W1(>7mB}@x=k~U6(JW^s0V$A6Rl&FSUCnpC9yj
z?OSL5GcN{vZMPO(6S&~N1&88xIU_T(Ph<uMbsb@vp&<$Sd3N8$aypSS7rWuFF@qk0
zEJmvZ5yS1uMCQ=10QpO7+?ifeUqg^j!f+e0p(r}l)6*L<Vl_%tC?sJN(L(xdTVaPm
zr><RNEr$Fpq-h?%QL}A8I>UiEFknJ0cTgy=sK`rMZr8ob{LdkueY+VB+)AK)e$?R+
zYqBoS@1r+dQDS1mL)pxO&uf@EKZX;ufS+nbQfR1*<|L<2ACiEaS{o{2U1n64Hn)Dd
zX*neD$9^!%Be%MXgkxe)<UHsinz>-L%h*Ez>xiA!&mxsV1H2<&z&;?oDq6h()4~M{
zj(inO=(cU$I$*$nstE-I|1jzk)Op<lJkVsMF8>JuJf6_H(tfqGvz)wq^~aA30l9F%
zl9yMXXc#RryjXHpd4GJLkEqC<@HAES=Qzovz!fW3qJvQOKKze8d0Y+C3N^KW^aa+|
znnd5kvf^U7mRG`Ltd_t_X#ubiW`O{nY290FC#NJPjz<@n8Z@+b?>zzChkhy3I4!n|
zhZ0r<1_bC&iH<JoqPYFmztJXIRk0>FYSKn+o(@%gliyn8797Qem<Ek2+k_ng0zM+@
zS9*sstoB;+bSpl6+6c@;RA{Uzvokjj%yp3K-(U2%8!-Zz?W(uU=XolpX#nZzjvhI}
z(ghP#1UN#WtrZ2Gd<Tb$^Rsrny0~7cqq0ov8Ey*|{NhU+%<E^1@KRIA>>K|O3<A3K
z?=ZNeJh#7>hlhtSDCy&ll5!Klb@_6fD_Z=fSt+c^k&27dU0x7y+HJkM$;`8VxCX_7
z5nsFgQ8HL8TU|=*dH+y2c=)h3VH|@EykS8}xKgg*;U_17Ud442l{p(O`xW9gp?>Km
z>N#=5y{W4qnF19uudEh@pYFzQy9!ISN<LBeP212-O!2rx2!qsPu1YQRA7uSQT;YRw
zpF~1~+GdOYXLmKrE9jIUGexpx;X;#W3jm3(C;ONG-1PIsg19KZ_;>G2t_=L!G2IEA
z`0=rfcnJS}_u;cCF%XVWqF8SntiA$IFZz%EqaeG1_C>=iM3GV5h6W8mVw3i(oXg3e
zMw)Jyy%&SZCI|=5)N1M-2@6yEY;>?&^?>dlC<j^Bgd-o_y0V_K)7J|&Ybb5u0RnFJ
zY*cHg(8<1Ze^;rA?*-5`U=>L%-M?Cif{BP+w3xu~^=$@&jW}|ozPiMdn|b5&YxzZV
zp$wJd|0gN5*<#O)5Pb(sBePSaawp$)JkRPKl}ibuj{c_Q%h_`Es-A#;@YJI@TiL<%
zHgVk8u`=469*Pio0jXF{(@#{SRzqY2G%~7YUus@mnm$mLVzII1N9x{fQ~O}Wd45H{
z>K!eu<^;WCyG?wrX&Xpt{mYCxcc8h5#KVrqoBPRBpoEzUt>jteJ3q=WPaajyM5r*>
z(wqb6#-s2gou45cy&J0^o|r^ASB)+tw6Ge7K}-b&f$9dFFn^_^V+VobxU|!6cKPf=
z$t9H(H=P**X@Ov@bW!5C)S7K-62IGR(O>?0T6Sw;V4sFrz8CI?rRohZsNDN+mM|Ar
zM?u3p&tCbP_t(99ce~t6D`DB+yg5(M2(CtF?ZSx)&M2~s3_WxRT5A?V4im)_%+40t
z?%Pf1t$6O+2LS6^9<dh3<0D+JO8E0ryWGFllar@>Umpo~&LeNdwB%+|HiecjDBR)r
znUj7$%BT1^4-;nOKDt8*9G&T3dua8OCgg8j{AR0$Bqb(F3RICx&ljnCI~CXfBFM<d
zEEq9w6%0WzS@Do*Z^?Fcxt&Y9|L^wa;D}9nqR5_BNwWzRyN{Pw`ioF&TZUaupC116
zZUBO^+WQ&~>nDk4mNuV?WR<hCRq>|F(_`}Jhd>*)cNs1Thl*!5p2DEO$a^p&c}h!#
z`_}B)=XcL%qFq)t6}Wl8xHZR+SRl<IbJ*A%8x?pNUwV7{6O|oPY_h?3+6b8Rl2`!j
z=L~D$urg?`?D-0$Ne4&0yaRl!zWw@r`?*QvD;__3B>VoTU&JA;oEuvQ3>VG`GAi-I
z>@`oLR}EZo<gRXhcj5ow=)0C*lllE2t?hNcktN&>U7>4v$qp)(kQwEg1ZRSA=JNMX
zjev3=MCy(kKD>)CMCKkuBX)I6tBGfM>k<$_YG<#F8_f#6R_RuCEjKeYMgAuqB#|0@
z1gawQ#mp5Q*dwgPjfy9~0($~_Z~mANr95lL_U)^A_MaEFJ&J9vMdYEMT)ZQP4|fp6
zEV+BP1Uz8x!Gl9H*JzCx`A9<W?<9oX*1Kb0K&6?nLH(Ly)~n<;HO9KaWhIxI@NxdG
z3d`S$-V-o5aPrg*TyX%fg9D<)w$6DyU|&=<Rdd}twfsypsI~~ZZr)Ua9Er{7^$CgC
zI_?;6)zHX@Tby-Ix&M~OA1+<Ih{?R!an5_lrAy+vhMfKNMjmU|ig>}aX}2iWIcfKp
zBd`Ihapuowl9NUI*J?*kvJX8}_fDNmjEzCDFpA-DXVIF~*MA00h*U_;$WW2EoSpp>
z<=XFNr|fKDlytc#qvVg<b^`d5a-Qqb+->54y9l7~BCixqV7~~-93Q{aAll2@yV+0c
zk<ayUUx~8bD>eqWmM@YIxH*B;K<mf&1r=hp&DrC}2bx%dmOOIMRGEA-Az}W)g&*I%
znX0)ZcXQiT@^SENA0jcFa_h;m_PJ|m59sjkeThj<jy!(6RI|=$HlHjd1x!3~pB~*0
zowd0AaQX5$8jZNP<-OXLJ1YO|pr@nbLtU$>x769$hj1!;)Yc=|%-DWS@if1tIfxnq
z&ELhFzjVEZvO`P0G-!k1HNCjh*VeTh1u4a3LQ+yTSc7|*EYP*Iv}n+Z=taBo^k5#H
zi3q5C`_^+~O*j=<P>>tH$!_z{+jsANz*WRGt|<C_>>ni`1eM+9!7e=`XO@9^Aw1$h
z#x{6ksfD}x%KHZur?T|VWL{?GdpeNNR~1|;>XGWoS;tzwDvlYXKXd7H{~=$>|2foO
zdu_|B-l$lT<IFpL1-gbpj?G(14ON(*KRA#r27uB8dUI3LaGmwepB_n}!i$9dnSW-%
zKX1MXA|j0_0QvLTGiTU;wHfIb>#^M2QV$&H5jj)MB*Eg6w*7T~uGzhrOVwsnH1=jL
zDrsNkm6_x<v75jogu`_8$r-t1JwpjPj?$5ByoyiHHw6N=yD&)o=0xGE4~pAMnT3zK
zBR$++yj5HNx>(l0TwxT5G;=}v$~+(l8l9KcY=Kc!o{Bt!E9^aS@&64mfplesSHFo~
zYu}9IUrk`YvHO#*bw+`PaaJ6YONjrR<f*N0-ShOFTNR?DFTMA+=$?=DSc!FfQ`4Mw
zz`|>`939)fp?M?aOz)YIZFZZkH`t_}JWL@<PEGAZ*2n&i2zuhopKk$s))UyiI^&Ej
zI!mll2$;#Ehf=(zv6>jHd1Tn|KgGRMYNfnTx$YvC+GJFu#1Rim0gYl(s5?=4@scTC
z+btpG761Wt>%0|-Jo?96<%%dvCDuN)pj=L#)WwbK*Sb0ieOvnV8FB$u3zPZ|IHR-a
z!e7n&b%%Kyy$`*-H2=-(t$DfziRMmETa~Z){*o2=`Jk!Ntc^`eTRAp-%a(tDIBu2Z
zb~$wTaPXXyOT|?oBR&^tqWzsf-Mebl<42a0@>Y)RykoG2Mp<#MDdt=KcBzk;IPuts
zxD6g2V%Jjmkez7o034w!Zf`mS+_XeXKFKU5FDFWFhe6MQ<E#x0+XoscIUNg6L>Wzn
z_@S}s{F}A#Q`0RiQzx|q+$<~e!x)Kwf^6GnWFG+)8F2@l$3?^~&p%|iOn(-d{F&2~
z9PwnoX<QqBg5m_hpG3;Ojs<t7BjnVi>XyA)bE=Q_qLCAAHrF|tS_T#9b?vZ%12cK@
zmYqf=zuE+D@r;!hxUgLxHU<n?d6UM_WyK0lFbu53b$enjiFdzIqyok|(N4CeW@d}L
zY9$rx*H=AHQ4r8LhRDs-#Cd4}@aUpNzo_eO7)j@1sAV!p>ek)6vyN)UQRl!CGHXb2
zOA3{ipVGG%;>!i;kzK`g2pkGH)}ZTnX0h^S!3!ROJ05Dl7BrywPoA7hPY2qPi4fz3
zaGm`cgQSj8qAvJn4&xek?ub4*EHZqQvH^aRV@msFMIp3sVG;iS04emG!Oc5nCnqNU
zV1H%Kmr;K<Tmu<2F`b!*=)|?<vtJ>3e%fTp-fXfLfHQ?49{QrmFjiKZxVY()Lw7c&
z%tnhtzsRTgv1#)*y%(0>e*QLcm1G^LOZCUS!$|=fU#qGP8gwG#!;md^du`v*Y2Etv
zf>rDExXw1V?I%n~o`gG9uo!WusFau5vCEiqjIKrWPr_H54a!W=s8613SF0on$t;Es
z9-bgvzkNH-O-Zut4^;og9XhPj5fPmmkJ8%y00GxIA4jG@7bF?uW(n&gBR{?Gb+f(4
zNaN%O!^!LlW@V<D`Z>mdl_MS=K30V`bP08IksZqU`d`{({qYG(ooO3)YN*T}XzaHd
z*P@#(o3U7|I`6k(eSTh^c-J61t$+3ZeV0GKd;7NP7?60#?UvxIt3SN{IWZ;8z*G3=
z$mp&0^@YcW7)bgqjyT&Eb<a9>e4jqezq&ZDVGjoxk_7&ch8kQI3qI7CZ!NgQE{AW2
zG>8mC!cz&hxrI!8koXEroX@#xabr=bsB3(;)n#&|RY1>v{j%|679=Gl7o;vVR5C*3
ze})IkvR8JVII-RN<41868E%R8sF`RXJRw7)U9Gjcf?Mz0xxEGYz?v^#Mrun28o}=Z
zU#)d_M;m($i6{38fko~h?ra3-o`0Rtx(NghDV1Bc4c3{ze|l7mOcwW+l$0!DsD9dQ
zaJX>I<$Fk_{2~-s>}csIfK)ObBfHo~#7GO1+X|B=U4c@l$Lq-;nw7pye0!XEq5PBh
z8m>K;N`@3DT3b$Tt`{fm*v5`X&p36Yx;n<7BWe|7{%mp|@zZQ0SS<pZADdbLQK-jr
z?S7&0=-zjfzw+cL9vX93l&Yx5L?J9hgRp=l6(9jSwvA2q^R-BtlV)@9?wchNRU96X
zqWE%kudBQN%Kx+gMlm=LK4z=mKmSA{FnFjpuHO)uJzKXLxVWfCJ?`T&%iO%-+c(9J
zdnCCF^vvEY;wpI%@5Wrp%ch2ck}M>O5l7v$UEeQSh!E7e@Bfqnjr-H5Y;lh0Ue4x7
zcI+_@G<e>|_Z%p$+_U_A9kzW<3izKI-=swu4as&-_3BVsEp!WvZ&vR<Vf25hTf<Te
zA#(Ng1wmW{4hI|9cgn0;DX8=W_OHM;GR<<cVl8}$P|g(QM347f1(DVzEDP5i&zl%L
z1w4k{wR1nn^wMs<s%DmtY}aT6A^uQZtv)e>!CPoPfljLLB%iTP6B(O8(U6n+2bw$h
zHF+5sA2yl$BmzxP$p<&Ekm53{2z}9xdgbY#PD^8YTU%FmBNY#+H+c5HVybCRKepCC
zokg`HppeqnRxL=1jdh@>qOir$i=M8sFuA`Iv1O9`TggMF-8yyZnS4=PyaHy5u^;>E
zzy(6e9F&@#cZ{4l-_laTpYn|YyvgrejOmOSFPWBso(Z@q;ZI`c;@#VoM73``OR>YY
z(TtBU?tld1?DE<+`76irFgYu!N2MsDp(xne@&(Z~2eJ6kRnOv2&rKI)A`87vmm%@L
zyNq=0$Q_7ydI`#P&w?7ccA3BHB^0+`HMr5LNtu3n=o^b6p?rtCdk6GVO-D<b^mq_8
z92iO;m!5axkRw8to=iZVI<q_PTseU6N;a@mLf_VQlCtvl?Zf$qYJqRt&xeYl@wl}(
zMDyqj?&v`)%YCMC=wzpU{?+-@S+;ST{o21tua!;QVoC}LVHZGeqZ%Lv(WgMb3VE;5
zn;SK3@L{(s8$RDO#oS3vO9N_TFykS(qey=O+Tkp{(0YZpHxT+`yGt8fsPNxD>L?)I
zBeO8&aqZ{({?Op$fWm(Bv-*;Np!;-nbi&aX231BjeN&>yd{%6&r)p$f|A^5M%~9u+
zN2hM6D%dZ7MQOXv@yxV4Nur}1498|s1hrAraYIhD^>7-z-6CS}rmz>cCRQ`<8MNsG
zjfQ#ms-N{DyRp>Fdq-R{U3Xlh^dhwxzrWt>5v;ozzg!xi2W-fLUPVO2(IH5&_1LK{
z$)iDcG(PG5%I5DxRY2R+OW54<a|Phkl8u`-P4s@2$<F!QS@7<_UK4$r#&V$Cnk{GZ
z{dMu^CjNr4au3VO>{i<@=gVh9*|gfnOrDP#G-wdKyryO?;n6{T>Ys5Q?BVyp$-~^Q
zco@Fi#ijXBSszO#izlwt|MuK&=-j#aO_DmEXB+F!v{ZDQa{tSjX_C?J^W)2|ha9S-
zknVjc^5TYSxv|MZx*+A&h!(|LCys06#$oUtx>{yl>M6AfGc&~{UP}G?-ygV%Dem=|
z%nHf~?b=L<ta-*98Ke%y0_Axhb!(aa{fE2!)_ET1*Z2!$)~)QSuP_Qp1Ix*FA0Mny
z&Ad>Y`i?;pejlgizR;4`PaRo5IiE^^Q*aUK3-+%x4*WnE>+QZ&gi!S~u#S%JdD?tJ
zm6xLya<ky=nb;)VuvI#D<k!L_sYR&!MAb`mk&>3ric!uswq{?-Qmiw(#&~YG`1Onr
z%)}wRCsJn$9}ir@ktyRSON|2K#7qyYzH@aaTB4>N71Ccw2?YtQ3@v$2Jic|SIP5pA
zB==FHMnd%CHdPNEb{FW--VS1wA~*K!_~FAp@c`ia098({c%+!SqFYo0=n!IIRwtm1
z24&tAsdB9qV38}3)0drn$a@e%CPg>**IA?-fug6!<lh!i$GM8=E7;%po9X?tHrbTM
z9&8RD8P-JrJYi^-7dcU^5pZf|>BhCMw@L&(J3Z}rxn)uFo=;8pvq;LA9Y4qA#dTC{
z=a|aeobgc@C`_C*iRbML3<qM&RHp3S+i$*sV<Wa3cG-Lyf6qvAzT5K;b6vO>MepH9
zEmDQ6yN7>Wv~R8Ll~ERxInhM>B2zx~bJp*+ZMzNLm9tP!+XKRQ+T+;Hvztb0Xh5z~
z@m1mo3pb&1{(0~f$A&2zEXJizc3zLA@KL@C&hNqz^8n5XgfO?cUxMiY;i#!dB3Z7d
zy!%uCZ`%`X?QO1|m66i-Yn)-E(0F9@+)vLv;u5xSfE(kIJANhNa`wbphXHp|I-EFa
z>Paxwxer9P**X%1C=zVq7`hAc{cznwww#>_(1H4UADl%FU3<~9W@ORG85%3Z(I9#`
zjJkUXAU}VHt@g-RDlr<KHzK5`WIHmlZ$nk@9R?HlW+Gh|@)eTlJ9d*pg)={DbpK+7
z40%9Os7zi&KWgj1UAw$k!#ANp^y59C;L5f$RIq${YV1#=POAZwJn$E?Z%ymx%j||Y
zTacQ*TAa?X=(6&Y#6%%6De2kK*&X@s{nPFL*$N(VzUZ?kikC0WKc=&O&wv?6MnxOO
z7kY7CWX*f3g|hsxyRi&|i%x?5-=XW8i4mvU)bXG+Y%jJG2>uFSE0MlA@lE&*Y|*#)
z@X1#%YF24s@a^v1st**OKqx^Z<2A{zWumI;d$7*f*ft05VywYyA+Tv+;6t%{of%nt
zBK)RH=jw)4ao6>MsTbm~*6sA~-_B~Y?sTn_jNYGMu9iw?T=0byaQf7#HrILBCR^)u
z@S{F@Ij9hQ{OY=t93aaXnf><qw6^#n5fKmt;;|SkD^80A?JFg|09ZxniHpiG_wOd4
zY&0}`xwqf=wSxp~EfAoX9#_}Uz#$dzK(ve`y<|RMG8W+s(MDT)19{{|?G0cCMksiO
z)$G%pHMv>b;hcNznM=Pzx0?Wn>#M%}K5IMtd5T81@1+g*WcHQzG?176Bb5seNlU8*
z6CWzm`M_NQ)#Au#R|kjKwNB5d>`Ge9USVAWi=C%&V`(a$BF?K;N!T;86dN=m_A$%2
zMvRbY9v|_rk?lAVn`W2Z{P6yL3jrR(vFX!45c@Q(zOfOol)*TGg*hw=Quo=zB=g3N
z=NF6+j_7)>5kJrY(>T(cYf;)ezC$7w&CB2eM}-rVr<$7XBp78kacK3^!tm4#i($ii
zZ`(FEpsQuq?j5u*J<@BVdcOGnX=U}~;`?2SjSPK=aMH#9Nb~;<s_%0$n}7d?T_VrY
z+5jd`Uh{H1x=m6N>RQ%9F4^w`T)+TMVWf*MP}{NbZmi^3hLnvEi0kVFt}|Cm;yMTF
zLVJ7UkX*af+8aUjIBq-`yf|{19dLh}!?$;2h9lw;bO!jH1O}Q<^;BXa9z$8=;=>MI
z#Mgx}2CAW4k#QhE$6E|xD2wS6(r+ZNDKq5&P3ajK_*;sM8-q0hDOgH2%5|k*!G1=B
zfw(Y_L<|J4#4$i}0p1~awYE$$u_>c?z;o*1(J>#)o>wvpyKnby6La%sfG(<HG{$IT
zd9mMQ96Y^_-t~$hzrO|GO*f~Xx9q#O&c}|ODm+th|4g(#ed+EwNunPDmCKIm*p?f7
zc~>r9b{Iz~@#zneZ&8Obm=+E(;6pK>!?m$lB2}!l{)K6+m-}?X6ae=A8xnH>-CddV
z>?-5OFZl81|2|#*wp%0OFIOa$y?Z}9F#!eu^Ha7^KKAtV3@cP0HR{RDn{5RYkjUX-
z`B{ifYsAa&1EV6+Qc{DO{@K|$y?Q@&{m?1%t*v2~pV4tCtE$!lcjr8Q+=B+Wdv{fp
zN_5Uj$?KNJ@bkTx4)U&@mOh&OfxJeK{>D@UZy7GG8W5m$f|^v0-Gvhb>-)JuIzx6?
zty(cZWAh~X0q+%iF{zOKbCs@2BHH5_GseL@Z5?MB&)EC#Xu=ymKE2BhP)_oV<ZPEH
zU|aAc2yGPD-}$sW9T>CcxN^9g{Nzs<6r)$DCBupd)!CT@0HaVZ4)|~dIiB>s-1tTI
z#rO!&`l12Zg=h|ygCyib2T_~GhFl*0(_b+gOyRjgPmnrnNk~+m6dx!dwvHS!WF=G@
zjSZ50*Tbi-P$rPtR8pGh_?gM1@S@qnBm(>jXq*J}lj}@Pi0BU?iOERCEgx%UwYsKj
z>|C~NcD9>RmS1c9tR9Nni+h^(FKZ~U`5#GHuuO``sF)R{>?j`pub!-~*Z-fM?A|l#
z!aw@49Ck5SSven?kiKf2jxE5JKmbA!zG$P4@hRS=h`F;Gbhwf<B_#u+4<Cx$(Y`xt
zMqr$q84}V9ryh$Tgi&#E>CZ)qbLP#Xo4ba?{WdJ2$Mg*g+-H>A^;I_M>8#!3)HLT?
z4pe?_ZleW}t_4_lQMx7U$s2j6j{csp)vs36827I?Z5#4Jq{$_HpAiILPCc0BL%yn^
zp;58ObKrrng9p*dY??atCJFz}ty^QqwmEa^RBe5|a^6N`<6YOTU0Z-Sj8b7?>eAba
zetw<x)g!9aCi#5$UNv`fx&FszrvvwvO86T@Q%dpLg@{D4V88tz=Mly2f&gSym}1w8
za?olkJAwn#pBEI|x_uid0+i%$8b3~iL*4Hc|H}i!fsKJH6s!~;-7NDPEC|XH4kq;#
zr<3^1cEQJNbT@8Ls=5i_G`OGBSD4r4|7IVXCok!A5rgJWWJGknH`J9;CD2x)<G3Nc
zPV-L1H!nh+@FZ!rxOg+Sd+4cQnEZ8)xFTUc`ubA;i9veE%v>EGT3>Ps`F{z+sgOrb
zSWUIYcoR^AL^+2Y<tjxFg++ToJHov5I#uH3ylV=Av8gHhTn@n<A)f;d0i8fmv~=0B
z2rD&d#qCx9J&$OeQGNH}aD(QzeTNQw^Ss2t_L~*Lb)X{SP8Vs#UX+y3Rmt`3yUX`C
zY0u&LS@8Xx3j)VMqyO4a70}~`wuj)6r#HBv0wRs7((t>$(5$F*;<EyBFJ!TLWL=Lg
zT^41#FRyc)^4dKccplI1o-Qm`q@>nBBaG>T8#pIf#_{rC1IgWg$<I?FORPR(ggAPY
z(P$&(Y)%eP-TjGMhA>D$UY@c0w$gpY6%2oiMff}3xVEOI^yXnGv<58Ly=EZ{wYF>g
z6>p^mpVs6aM3Cvj#0mKSi$?k}|NEXm8n+lC4q;=XmsjPdPn5Ok3pe-X(GK5Tig^qQ
z3<fTS$e6-Tz)X=x@O<DmmjYdA-Opgfz~5rQ;uX8*L=jox@%9ETyRcgUm1SH7gYjS3
ziivv~Zy0`tZ51n5udgT0c`R9?IAPS-vC17Al6zz=Y3q4is8zy{=ll1LP|1fV8xLSV
zkgf}rqO1EjVq6MVXQ*q5N=v<K*`K}4(NPpdAJtA7%Jk{~)j(q@m!rvjZI8dk4bF7W
zw(RWW@dFT3>^ZGk#o!Q;d3WQ_uQjz^kl0Nz(`_9eRKBhpvahs@#E|vP2Z9cc@~SYX
z7)1T?XT}_v(USU^N5aDyagq~AYxwS~Xir5ENmKSdu(W8k_JO;f@?ru+euxW=ix+|v
z0oSRBZ+l=&I+XoU-b+RsX#&oWQtIkXCnxtl{m>|()g#Kp_KLO2u+CLs7Z%U>cPL9^
z!8!ffGt9}MIH!GpJ6_O{?g1H<k&R`Cf4lD$COf>Ta8CYb4qy)OpgjDv=W@;5DNYWt
z`)^Ixkr*II4H@dTdiD6jhug??6L72$m#F2f!3RV_KqRq5b(3Y+4l%TLv~+y)OB2?u
zU*G?iOO@FmDRd#MZN0)t=3O8tvuonB{ME0+rp@Gr?<m_mrm1dL?XqFlh2464OikBx
z=`&A#k!odu$qOb17OoIdVw8TiD_8g{5p`9zdd>GguBA&3x%~fduh<lQdJxpte;?)H
zv{=2DIesOtg_nlE7{od~zfMc*ww`+Smfz<M%Z-C}iL*=_Ku|iEUp@|=-&aaXzv>Di
zx`vvvG6dTkSLFX!+~3`U^SpY?u&06Xe<l<K4sEvl&h`LmdeUP1P_OTa{`!j77y4h5
zxQO7&O4r5u*~RsjFBWEn9XiA{nRO<Q9|t%GT3)=IGRHXUebYcquH*_j8^Inh_|qp{
zeDU+{wi~zRx57rxf%9j#Oz!@By=zuU{G5x08_d!gddA5FsT*|DvNFxQGf2td{=i;C
zi@M#h3XNYqb;rFI!Rb<Cl#-R_<i|IUd{w4Ulk)45N0zMmy-z>C-mUg6{lNbnyZGqS
z<*aMPC%Q{U*9>@M@OLi$>Yn=MU-v>(hhEG!`4(uHAaO<z1lj8ygu^=w7p`7?dv~wF
zF`mMpF%_MZt12r6eF_9p#MhBjA`dJdFG@N)x0#&~f_}nNyb@P0U#{{bXIA*To(73?
z|C^W(HyNgG?v{7*?1l~LA0F?>8riY3n;=XPbnot4(oKHY*<&D<Po7v*tV6LcB%V1#
z;E*W=doX>!Sz&^^!<bnohQeY9`bZWoS$tq}<`+$tc4AwDO5K3JgQ8N-nn^EhZbxml
zN<DWld$X<0KoytjvUw@-R!PD+ybxe}^4~<-QMV){d?3L58PaL9sQUBgKMFx&i<;TJ
zZxc5*vW~Q)NU;&VtY06K)7PmIv;Mhr2Pa47I?(6^2?I+idKx5mzWeW0R6P+Inximj
z*sl^-8P(Y~$9?8L3x9WiRoF<axTmwa%cY}XRSoH<py2B4%&2W8(9P+jByoE#m_Kk7
z8|&H%q<MMyNI<)~u&FV5nQLBEeg14?ZQXRU$K>(j$J1OhGlRZ@rK<VwMAdup*yWM_
zEGT6n4dbUrV7yeM=vPUlFl>zPY-(=G00l>TLxPz$Z9AEAaM|Z3lD^fySEH$rk`m)A
z{kN)ef$>R6MNG!B#SH6@7K=vkM2EE(FwLh_zjjrQ8=AfB?%VzHF>~haxqox6MdHHD
zGmbyqzl5zXXKNrXF5(yx*kwen157A1!p7g1C^N7lKsju9VtK^5v^4fH*fFQeE)8^j
zyu2d#M3_yY#Jbi}ASZT5gncmY8Ae8TaKZBa$SexH5X3@mZYOaEq&*1`dLuKbw?XBl
ze+PsuI^J<S|JS*nnw@ndRJTY1i)#Za7AeeJJ@DHn9Dn-V#v<g-9ac9eVMWTb%yno)
zVQpvUhfs||o*>fRA08HrC1BJ0EQj1+GCpWQ70H|#v=8ptkP>MapL%*efAVDI$JvJ*
z_>AJ_q4iaBz}e_k_|$B(`SSj~Hc*GK*2AMqmo7YKrz>+Sdm5BYFx_@~)rRhj*0^kn
z=Ok1AqMoc<5&xsRW91@M9{=#+i=Yondpj8-$vr~bb|HDrunfLRsu#-d;q%VmTh%M)
zL~^q2+_@QRUg<E`G4HICN8+hdtWqBtwutHP()CU&ND5R^!okqc*yQ9_rO%w&keDE2
zhF(ohnGLbfgxC<KrHsm)b3Jd$lxV7Wx{?;yMg}!C080<HMl_qDE?{OtuwZb&(lW2~
z*~b@N>R&0QFd_XN!Ox-Nq-V&r>%8qBulmZMJ$pK*PwR4-C%PxieZ9rO2m9r%Qtb9V
zoE}~9@QCFV?=jP-C-j>>CAnjivF?_dirwiRtG@1@^P}IQHmi}L@1ARC*NW<bkIPp?
z{jx^XfV_XWs>W2Y;Hca9W1`}eS*nW{!^1@87I_kzA3x@+KwEH&e9=EvFEA;%%4ECY
z&e!Dnh{=AN8OPwA^HXWo1$|3P(c$dv)Vl6!zO8zni&dh62P^hi_42|g^+hjk?$oN0
zA|=W75S<hRf9AxY)Z0k|!LX_nfFUw0Y<ltcmhabEyP=$(G;G%s&n3J?Mt&p&oyaw-
zR-wy3$El}sV#FD^Ni;R)CL$JZVSioD>KBsVs3SQsef#$ZkP=5DreVHoKPr6w3EW$2
zG5=wxOBmnOG&f@07`0hzN)>4aw0es@pdctI#fWh5%XD!ogjpEpc-!rz4AaW5csC{S
z(5*Xnf-A-E06`X5{rfTz6JQ?Q1%#0RfX74~kUE8!?lT$R7(yGY3`<;&M?@?FMfkkC
z2+4eI$96d#I&S;XJFiM>?5X+5ry|GoNW7prnrR;S-;WF>wQv>q_3IZjr+{AEaO3=r
z(K9~*roNquLQJ1ZWvra2QKktNz7m=d`ev+ewwSK3Dqp^Kyu$_dZ828RwS&^AQ5NIt
zAZVe}G`T1Mf*`$oS7FOngr(RNx%^n0ed5Hr(RmxMgj>fQm}O#Mki56cpJwIG*HwKK
z3r11k5o#p#`En_$HqQT0GM6@T(V~)(0gkB)aF8g=y&N0ayWeTl`%_w8PJ8`O2g2Zm
z(X%FJCh-P*>z19^{@L^AKo)KN36KDwDl0@254sUb?0Mkkn9Na_Lyv|>gtu(`cJhMl
zh1y!J;qwoxsuqr1+;yE#m*-61StyCkle0|zskQ<qZSa_v*Zi=+6t`2N8s^_#Oi9l~
zdCoB)H68>S?r^?ywDh3fc<TTfs%Uz!1x;VypHVUp8>aDKvtZ{~`L3Ud-(>#1b=AdM
z6$VK3^)ovUhzr_b#`R}45v{Yha~zTY9^RQa>*Qp9WYm`DrE|%T=p7+V5U`*>z~Zg>
z;*ZgBpy2#|%`s!1($41RQ#*>NWX|53x)f;*=!$c1&zV1e+?X+o=pI*f3NfrpI(wGO
z?l=k>vHk*+1g-IRUj7;A=`wtcefyy=pFEK&8!@&6=7P+^$aV6DzQmMJ>{R;G1wD9T
z>bE=aF*<lB$UD&_lC_m+eVM+QgJ_if*{P`r!zOm9OBV4w&@JIaN{WW}_S+m(vLDm`
zpv+&ct}dD|{cdFE&Qm2iXTO{<*z@4wv$pApV^!ju@5`?7_;3@bw7y|&*NFy+md0J5
zp!4RqS0oeARK^VKB~nF4kAmN`?6yB<HhzyFxpgK?;EL=I3x;51CWm2+ii%VOBP|br
zYEc-FKre?PDB^{5gX;lr;QO#&hJ6JJP};B$$-#4NZJQY?1jhpULK^8#`^GPFm`yTg
zxREPQg9jt4W-$}Vz@<yNtvaI_vf@SS-(P$60@}{6(XfKFcRyk1Yn4KBMf5c%=K^IW
zatFn=-`W>Edj@o^!Jwc)H2%r{epCK2Dvd`hLr6Lt8j4sIrFExJKmyEcl?4)p_Wr^(
zU19h{9s~%1S!ZxOnWG5+XF!d8<{J!eR?$|PEVOIa?l;R{FDsx$qe<pP4?NYtVA|%8
z6YrByhuyo!B<IW98i5~R{iJIdj&NVajK^U70Bf?!V#!?gTGv<mm$b<qDs%4q=@%DL
z&$@&y!o@;a?B}9)Hq)?i-~|0UWt}-XjD0rxOSBaYq8J{UH*W)CRFG+gp*SpP2d%Dk
z4IVLcsHi<0eLf?j0Tho@LMuOTzyKTU$k=8{d*t*~H8`6oED-SYbO09&gR(_C@SfRi
z2mGf6aCavX(CIgGq(DhX2Gl0uo<{16en4~JAk00xa>W-(5=nun;Oh>W*w8aq5fl~W
z3Vw<6FVz8GmV)kveGeU3;wj1xxd7$8rI|2z`?eiX?qO+ZX->{AqZk#&26pbmgNC&A
z9ViZ0g7PP1Fl-5lXDRr)W&eP8BG0Z!T|f->fOHR3ZQ6L8dM#>l)Qq{+)cec^mB!yG
zQ^gL8otac!W!WAHfA%yVZk6Kbs&uRQozmf8J5DA5j8I-(4H;z%qYks@NhuU`DJ$P@
z>6qHK`WQtc<r6U8yK(1^5eBII4SMGl3|OPG!^wn>vsn81=Z)!q&B07_>@iGf?V~;d
z?*eu8ZbD=J%56VNm4^w(gvu`0KPml_b}q3f)7<IzkEU98`_!vfkx^~G`ln@26(v9n
zO#)yL=-OQ8tno25j;ZVFvJ7b~rm9(w&c(&OW2OU+6E_m-Gq1Kj&CSgyoiMb33Ssur
zL>g`~16G%veDJ|cu>bz?=76#vh$8irL^;`oY5mvooWeCw&V*4!1}irRWobJ+#t&$b
z6_>f=WVTuy_aRG<b18mXQ87ikAHD)+X5=b%6YvED^tsM_D<wr6e}r1hb$;C3RRG}w
zpg0C$2^R^sgfrUI)|v}X;p|!8#_Us&=QrH_&Joer7DxdKJT_%phfed4JaoL2o&74?
zqocqqS2crN*D_49oTwrZy<>OtlVp$o^&PwJyPxdr+D~(B`E{ekxiv2_xJ5$NS5~%z
z0Hwl<bV&z51V^yEFEw?v?d!xh^P0?78*LT_g-ii#0=b`OZG92-B9~blVBWjej66e?
z2WbJ^V_jRTC?JF2tm$G$^`*}Jd%lWL*?!a~vk!v?std#YjoF?y-s{}UJj<8oL=j_k
z8?j>%M9QR}0dxSRHJk<T4Kh6@4tg6}PF29~n5)uG`&)5U*=6skdT^b5b=?J7b?aMu
zB_v7L?Chm(VGCj}w~&4{PGiG~nEH<3=vdY;QS2DydSCWzax(R@bBj!AClP{v^(xZu
zsW__|z}#1@G_2aR_~W%-zm^-;AURaZ*%-qAjw`9;fN;||nwZWuo2a8|r@Sryz<)!u
zg95tEb*Nn)&yOta2JN6b3xq-KYf*<B+&kiB#=zZEqb(Q?_WArmocRvE;dnkII2hqn
zEA1059z;qezc7E?U2;XMMLHrp%5N1}mn8%mhUVtgWd65Ww>Z!Jy%!!7s3ayCU|z5R
zT9s$Y&cyi!^)r3HpS>Th<?346&VP$Acxv4gyUjaZ+_{tN-bTC>Z8G^{<y6r=#mB3j
zZE(fW@W#W-lmTzsx;CaAInq}(q?14hIKNhrm?ak5;9ACgmI{NCpRU&aMUU0S{1(N*
zL%p!VAun0!dLlu`wc9AtxLZWkr%&Fy9gMyKxJb^~xpv(;gvUe+lNmE)a_m>G98&SR
z%juL9yA0>vPW_`sFWdQ5|5JP^9i29NEt%>jnn8*!PG{Jww`iqvxsPmO5L3#$z#sr4
zIhGv6sha+AXlTeLuvuk59q63Ar|8WRbB5OB$B=)w6t47gP&P)s#JbMsBj&VB8pAGV
z%Iw+SMQ6g*#mas9O3rYU6$HrxD^uprK8?Ws(9JCmBd%RN&Ua+pR{V|x0eF3T#5zLI
zyZ7(oqoe<E&_twhj(xQ)M?w-Pv~TR56McJ&$*Q#?ihJw9_GvoSb*dT1F<wEWBtWS^
z6MSSwk!-De@Q083A}{`8dn+$>`x~k0Uk^zLnogTgP?#jQ1HP1K+d;5X8g?M3;m)*y
zTVGuE9^)D=>idFo4%*Fj?X~IS8jHn?A6?U&;62^m(XsIKd=0llQ>3Y<7%})xnI=lo
zKofxs(cSgUKbnp68R94v6+`<l@v`1YX*U{*D(EMPMZ*6+AL;#`2maREE+07XGHLd8
zhc^CSbvm|Y>Aeh5)7snT>DT1W;QKevwI8V1P6T|ALk|v$V7mZAZVWFC=+&zZ{2g>|
zOIMMOexPW)uj$*H=^eHyD)z@tEHROp0P2EnojY68U_cCtZvV4qi-uqDceF(%MMdA{
zWv<#ePyJZb_66Yz+2c`Wz1D>6!Dm=fqr2MgFZ$wf9CpZg*)k6h13Em^PLbo?3@)e!
zUIT?<%s4i-9zut45t;~z>y$iWG&O}^kFyT`#j9CzZAj}#`MTc5W<W?Td)b%a9_E+%
z?-!!@<@-W2v%Q<Vm_+v;v%|W}UX{A#lgvzA)HSjd^ao#WO&&5x-S*RKt1FJ?&0)WO
zapbhms-mO0b5qu9yG2_fj(43|856$arz$0kr_*$P+jF7admSJc8cj>ujhL8al<lIS
z@*VnVv<3wQp%4BkT%By~HCyY#44v$p8{48HE>7&f!vK6#*Sq#DB154Z$pwM;cfWAA
zh{|<MEV>j1;i+&;^5bL;C8aTZ<EnFOHXaTS_aU!Bdhh=H!Jx9T(uQQP?h<Q=7ee5)
z7XJN|+Z`@%ZphfAsN%fD=Inzgzf(sdJb%_y3Df)p!K=N+7!SvfG1CuiY%Hj|5~=7~
zCek_);$ttsz-DJqk3hfBKFl&V05^aJbs5{=)Cika^9ZB(aegd{z~Y7nO<+f-kq#Dt
zFi(1;8K7IC&yFlfN<8i;UFPol`i_^MK6}=DuPxm9-Vst16qxL=4-`9C?vL%_8P6`4
zj3vjw_L9TizI7lv=KF0%k%?Z@Zec~uU22!y{Fihd(#82y;mmml1BD3Sw+s-js@f9n
zvp3@Va0!8%*}Le|@o`zNpSS8M1;zI2_0IG2!I?AK37Lsql_a)G=m|Dy=Oa$V+`RS?
z@qzi3Jnnq4^xwFsp^?Ho{%fU*JA)qXq{Vq{9f6E=0z+zbe#52sB)o#+5;_3{Q3`+o
zQKAa^Aa!dG2thkSRtDJBb#|Qu@*8_xOY26=Y<#+dCR7=AcV81skrUDbO{+EQgoW6I
zgs6$$!C}!!S-VkTD_>KDRw7ouUBB(?sRrXViv#MYm2)R}kLd$yqWwnS{L1ZzzYYoh
z#QD7sTDj%5Q$AdaR%-IGN?AN~q;<iw;_zuTgWWoGPoL}_OC2rNU(FRaJC??S8nbC#
zdyzu{2x?`7XfBp$nR9+43G%!d2G@c=_Agr+?yw=CGZ+OqHTd0AiR)lcPoAvheBC(e
zeSbRLkmPFVfme89SYC}d_~P=7MLmvMUWUv6df$KUW;6kCoNfH6b27`PV5FmOq*YUr
zr7?MOlrmYcCo2aiklOg;UX$;)m@&{B+dHW*D4F^A<;!C;a<OSwujrA?GGD4);l$aq
zrI%$t$Cti--)t&e7cE-`be5HoVU=v#>*ywP%oGX+=QOPy_U2A5roAI)UAgEJcg^qP
zm5pp|`SJ6ogg*O<j{Y-t@Zi@}dq`OV_wUz+g86wndE!>+J$Vu7b>2!qajelL_4HIS
z68r-K?vm_&j7t@#Ck}{)I3$n3yZ`v<ppr;CHhwbO+U)d8FsGI7iXl*!uD2R+I2o8<
z6nmotr#)Nx8wq><V|S~I4w{LhVu?$9{wnL`+LJR(cI)VW>eR5s_0I}QZcTlE5_2q3
zg!8j!R7*rGAcTX<Ickgj3X>N*=x*A)`6}dXHaqIhx>R+QoHpCl?Yy4~BuwF1+xrSP
znY6x;ogI`kKzq&07yIrhEeQ>lZtv>ayH~HG6N99xI{q_zg^O+YL5okX+gMo%_N!KH
zWUCEi99kAXzgG0p44$=OB>-;q=9MY`KwGCHWmn+NWtfZyo;-ER+SXS6^1g*{EQf@O
zgg~TD#+ftXq&D(vieDOVF2Cq!V%*0K6t>i$5PB4h0}hC<6?2{^PY(WL|MtU&>4t_u
z{occ+8xQC%upI^m9UYw}x+#7TIN_80e9B_UPMxwhH|sE3c%<(clO0(Sa)FuaKh2|c
z002<f{No>9^f9J!@v*TSf)M`|{BrfHD;qJr>nkHukK~*02?Fhayly5JY6b8ts3ldn
zgN)D^<=TN;jaN8y7_=#>1kCI?+ZONwH!Ea%_eTE!wGnXZCe`~52@fT$2HA1H<{QkL
zr#))aSB4S*7pRY*UjORr<b{25T71e=MSVK=3=Nknnw{{r$H3T4Eh9$um_M_-v$No@
zwRz(AYj%TBhA>N3*U*4R4qC5%*i_Kmyl0Kk&Ymo^I(D-S1=*iHA9CU<48CZ3Q83{F
zV;pDHb!Jb~ExstEv1Ay(pS8<dj#q;+)0^X89j3{kHsqxIM6ZLlm%{dTu2)x}U#Dqk
zO)tTo6+zd~kt1iCnhsV!IuRC7#r)yJk(awISh%p6jvCl;ndVV?d|sCs#4=?xjgOh^
zX46vi`AgX~pw(BgCAXB+)W-39xU?d;s&@WmjA~F=SP!h~iMV(S5iY30cBm~CT_4@t
z#6>gD2P7tCUGTmp%})ko7S~4CzU>6M^6{ytCDdwY)iVd_X`Q8X&+-1w?K!`ith#i$
z{e?UWlG<=_z2PLEUH3K7B33NX1s!PTPqCZzqyJP+r(KUrPpDmYoT=$1ygt0wHFor}
zJ0MwptW4wT&YLqw+)_{0qC|nc{Zvz<anAW-SQrL7&C6X~Cj;RDx$q71u2Bi`2axv8
z56YiWWzocmtFzF9Tu#k^E@YS94=f+Iugd3g-{DVGpP&NbZ-Y)H<&fF9MWWvQ&`D3B
z@7R@6*$DtE*P0vqo2<@vd3|*BvJO8-5eD8jd@(``jh8&dUtN9uSWV5rQ$kVn@<vf1
z(+M)<ItH>fMdA39f<-U)+3lvSo5YvK%D9g-=TxBx)Tj!&gRPL-Ti(64oSo-dHgEK}
znl#72?Uj)eyxpB_Y>0qY5-XdTNC9}ICm8Wj>YoGm-vuwM%zfEWto*e?JM>ZbG5~9n
zk{>hBxa5?l;uK5gnK^N+^H4$yFu8-5TGd)Ge^w)*!-|cIYk&B@QSw_J9$=8*kA?~I
zJj?hfr4GjKt5oguUG{xnbuc30GzI}sVFF3TWx{ZW6)KUT%GZ_!)cb85D{0Rf16RJq
z7wsVr4m_Ac7R|sFy2YIPwBEy`tv_`u<)DPWh{*3T#mM7&{oury>uUKk1t54Go*Iiq
zn0uZSgky}>)v=%90;`VsG=9$gb#Zmg#Z4L`7hjeXI0I{vaEqB{9G7taSCi=<jM~TN
z*{8y1504euJ3-M=mw>3$HZ&|hUnEYU6egEYnj)mg6@CCgW-d-q5YJiLKOV*E<%9}{
zSFXpdc8`{whN1Q5*3Cub_MOH&c>ew4Y--BnkX~bb-tXt@#%K8|eSyd&egE>VjPI>?
z-@e8%6-r1SeD{7W=?4Z|=iBo8sA*hq`<sHkk&$S-Ef<5rE-}&A$VfE396ZLt)a>1X
z0YQ`8wpz_sJ1cc|k7>i6-WjwrF0Z$HaE72uK+WOEsA)(3@mTmoHpKf^$F+1K?~-S{
z@Q+8f^VP3k*J_p@UAulg+m1KR?y^MlDCGqeCF&y{G*wR&3p6f0P}*hDABQS&V-x*4
z19HN;no?iFJWaeAoPo*C)<p_mo+z4nXJ0V5;O?Gbc6d&TO>T4A{MnFvqCWoUm;~=j
z>?v~Y^Y=IR11~SlbXhd}yAP!slcLAM!tR=N>e8jIv9UCDSpbl+Sz(Wi-PGh<7YY6T
z8n<3pGG!G>F*Nxd2LZO>s@QVf4_iNWtix1;x8JK{WMy`#Z#6o-e%%>l?tPxFqmD`Z
ze+YZ;Kra72?E9ppVK%fBB@OL?$Vfs<OGQO0?L;9YqtehI8Y)RDt0WCkNLr{QQ50Ee
zDWioD&+Dw~x}W=gp8I~D?;pSGclFEH_j{h7^F5B^cpvYse_dtf`~Z2J%~lcCv){dI
zgR{)6x4IK2zQ!6BWJ)>Z_U$(;Bg(X8HPoHm>-WDAH&}yt80)1Xiz9mw!*I-Gc!^x$
zym>QeYi&vIo99u4bvCM*#G`I@^Jet)>EEwdPaCph)$+NYtFqGE2FM$Ja^L+VBfiJd
z8|S;;ovNdKyS-wSLDeO;fzV=r_Lq+ZA>|a%wsZ~(_~O7AF!D4sfYg$|bICX08j8a$
zo?Ca{V8SA9TqXij{mdkX%U}1kvo()*R)6x~!h}jSiID^IbIQ7|8|*#%f|AvfkpoTa
z!WN6lt65+M15~zk@a6!OmX>N<l3mX-c29zgWR8wN-vu;X&DUESv^f3>PiFgNYAr*D
z1)Psuw7JpA(%ihz1W6SU;om*$PlvK$^z0QYR`khh8y`cHheq2pBMzAd;F-pyY|Q~I
z<w=|Ly7kZ1JF`Qld$qo3s_%0hZ(WFa36b^jd1C^cxyvms-Nr=Z0Wx)+L2P-h3S;z2
zq&n?>;c#YD|HO+I7wEW%%*5@)@t)2ue2<WR|M}^aJ@&60-P3MP*Su>8)n412p`A^T
zUjKbo&Wji8N6%g+#GYQ7_4wEzU*8EVA`H(hJi9{5qcg5q*>dN-T3f)e`LSCSa~D+^
zE?E+sW+U!G8gr5jM>C%dVTDV)`X22ue@!gn{fVv<PNWwVB}=d0xp?8q+v#2|UAw-@
z4_554HBzItXu{i!sPSX2u77fT^%d-^Kh8%mNACl)hQ?cS<<)sU+P#(<^Q_Hj>Z$D%
z#Y(by^8ypQkV2s{?JNi74S)?OwmEaAB)9j6pBbjR-zJ5R9hV~EsSs*+*6-O<{nnVE
z<C~IBsr;;ZebcVbNsUmOs}>|W)~Atk#xR7w_3H_|4Zyi|Ude^k?G_~iVOt$J+JvK^
z<HtMMpPLuW&8AvNe|SmzGysT3!DJ_NxpwnU)=iyZwy@yV{Kju9zn|}8pynMKbsF(x
zTF}U8D}Vakai>b-livFbyI_6WX{fOA&5g`w&-{$PxXwKYi_>d4E-wYe&XbjTZWX<r
z^ES2b(o;@eUPKOd{G>C0-4)f%RrAz;9wRZq$PBU_je~VbYfHYyv1v@O64`;Y-8uQG
zUacTA%WgYObM;8-G5FZ=h>D5}lbnZ*+BUdOqNr`p?0laWOd?ZllYo%}-`sjxt}>As
zSzRRm0!=+OW*ciN$9LFmsIsz@j;}CjWK^3kt~IgrsLpDb>g@CLGcz?6J8g;$!WFSY
z50`6gvyzv~G0UX#CooeNNZD7tO{%ATEsRh@Z&1W_56)84#|ux_Ropt!f%+0sYrvA<
zcl!Oh-@(nxi#T8EV6-|;@<dVO?99i<+>RgLVs)DuF5Y;8T;+C*^?>3f1ja#L<mU$+
zzRnFxOzass*-o*_^Bc4W@#f6YqvjY+@+5NND0!?1^C^|N<zjEm;TZ?7MSTb<reR-<
z>itlTy9sSmoVe=M^Sk*lXW&)>K82^<oV`@H#G-HKTs(>4h&eZX8#r(JjPA)P3ycp1
zN_SFFOmSRbe(~nUp&sAc_86J>{M93@_8*=1JlX!*7q2yLI3_6WO52Amgg5Dq^MF+)
zOS_Kg&xXE(UA%m`imQO2mM%mWK81SE?}YwKfNqE*JT#*j9>=q2$&yeOZxUAxe^DaV
zR8)*#`&=1L4yI?+lAFWf81&m5$vir`v&ZDgX+&AJ4e=JdDE|>t)#IiVkPbA&h-n-!
zGD>CucpP+{o!L>OxG2H+j3=Wh(PjKnpU!Tzr4p|nFH`TRnA$6R!n?e-Mn*>Xzc4W)
zIzR#8>r-3-en4xxk#~)c*H($33`ZkgoB|bz9SBQ^2>wYy$E6faR|@-v+%GddbGeZ|
z6LA+D_r`$x*v8t<K66xm`-yqhBgGm%wk|UC;-Z+ly*|jVKf{3$ghR%Cuyf7m!a3t!
z;9lA6iv%981`=|!FPV%N?}uEGKak9w-4`Jc^<Hwm7X=ZW11}~jI(6>cb)wN)KCY){
ztJluX^IGUGaF63w=axpJ<4sCRqO+@=hgUi;q$i*wanMQsW&E0R(*X{6nHU=r`a8v}
z%RiWF-!DKhHTsm(`g_bG*5%ID?<gBcQw99mg-ylq5c%dceRlj)9xa|BvT*e^%c$nB
z>3Dnrnr*Hv+f^}%r)R?s*49|74KVa<+8Uia>8e{)p0-XBl7q_B$XoUD>n*wH2tl=l
zr)PARRLoQSr*d30e?Hi6Zcfe^4Gp!=q9btGB`_L3er)W>-G&J4*ld;c_%Ujt97-(c
zMz{BL&ou>C+lW4xV7pkG^Tfat_d@9x9K2!8ntI%5cC-`_r0B@{(KSI~vVe)6i#0J0
zOiO75ivm5s=Zg;psiJ+yu1=Jc(irvW)hlQ14bXP_JGCRXSe}6t5?#pumXQ13l5noO
zHGlmKHYs{&9~g1wlsDxP%GU$I<vtHrT((ZpR#6G->h$eDmUbbtR(?oqY;=peP*bv~
z;P$b)@8Ld^hIc+N%-;UygKxK`RL&eeEVpA-1;ra?){u09=!r@F+5Of^P<@gI28BQ<
zUX7?!1p^ZkcA@P_tr00(v@I}EAl?}VY);PnfOF0C?|XOeMoD<Yi2(+>cyuNJkXDHf
z%YiwY_hvkMrZHiHw2aJ$S)=UNt-FP-E?64QYxKN45WYIgWcFTXF`xjiDbuA(UuJHn
zh3dG7JL%BsBMh#<;CYETBQ<P^_peAs==wlC{eR}OO8vJ4J2hshRhMf{n?HM&fvv5o
zmty<nL?&EKP!2A*zM<jJf27kPEH>IW+P|VKCUDwd)gfPtrp=fhczExs^^4<IbuP?^
zdTlcKlg^D*PkL+)-Xo)g{mE3`CSb|yjW_-M{SAan+p!b-n(?5ZtsOFYi}mEm%Smwb
zT|jcl52!KkxYx@Fo|vgv|1@)TbwzXo<~o{&SJ|6AeS`*ii@wi@>f=39-F_Gw3SIzN
zKYu#S(<A$^+TyC`pKbrlX8Fli#l`(ZmvRY_GJWwPQ2#p57KSxgJN`BqHi6SHtKy;k
zMYWJqaU87nvH=WD|9hOCYetUQ1-cz{#@u*Mb8*Vp{3|mu?)Msfb{k8?PmT+ot7m9>
z8mU<!<zRNUf6U;)HsZ#p4zfK=FXVpa0V?m*j5&_MV?vYYd+GuzCK=3<MSuU<3l|=*
zZLS1t={tVyVS@oReT=`y)uOT&n7r$J<;Wy`JS6(}4`qUzs2w2O;h*B%H7(t!xC0JT
zw#}cN`r634Ib!{b=#E&tZx}7x`l->c^WWA+%2@;apMiocI?1(dZrkS7*SxQ<J|^9{
zt0vwkd18CBd`gwS{>}#~SEN)n6OP&)cdgiRiyH#BM=8%Pn#DNtacYV4YH{a}A78j@
zx8toMpj$93!XU~z|KZcq{h#*1o^CX59?4D6heka>!IqU7WxgwRV~xqI>2j~H!wAUs
z9IdRPvO_cM5B)?NNlDTZlPysF42)u5kHE6`xEo1nx%M|7Ep2Y^krgcgkhcCn#qYX*
zn$56h#yHHvWizT$J~ryoC77^;fbhAGA6LlsKd$Q*IN(mc?xLyTESp=bsTmuy@yMkE
zati)`2MnN3@(T)FIbAc=9BG_ItN*~j@@%!R*|X60QNb{YLT9XcFa$GM4r$;4g=mz+
zvmR&rx2^r3I1Ae~YmW3M+GlRBL1)a^(IDxsPTFq^3JOrgJ$q)B*x!BN?A8r8GRn_p
zX-$YNc<=jPEr2Ib>wx@^@7AnXL1@_MQZhrJ>o3a4w;8d!zSzuWQkO~#6j5}pZ<2aM
zoy-VTDSjYx$S}-w+O&>O>knx6?bYk#&CmuHxFqC*qBC|do?Rq^S5bq_)VlihkNjHH
zQ9;}K#pgJup&Ls^UWpmLZR4i&!OsWm*FNm`-RN=Cevpx>VXLDb;(#JfPmy7H){wU~
zwubb<y2G-<tDce*DTHe0GceZFS;>^$jGGhc$uPJZY8{(CNQWhmN5bDd=sxJ+(qfSW
z{P^+LC+aWNC0{BlO>sJq9Hl>WGPGimJotlf_U6swLCaC3&~Et?BKPe}q~)N9V~7kz
zD=!0syW9YA7V%(i2ch_}+kotS^}5R2J&dfcu0Hze!u+nKU%$c-TFsfGJn}xu4GO6x
zwpqy$>t7FkQ8UQfK0INT0>;eWo8~TBlrCSAFXQp#laux#<4bVWHT%!54#XF)<b-Qi
zosoQBnZ3CuI%YoC1ZWI=eF1rUO=kQE#HdQpjsVFN63w5H5%ke=uxV4$<=3wp`mGo2
zi)RLmzTfrO<C|BPSHF2U=&gy)yG7Esd>cmRZgxI8c%JoJDqB?_3zF{O{(#_ZZ`;9f
zmg!dlrNk^rkP$FY=ZVo<ey^?N9lf!1>B8OjTikT&s^2!zrUUjJ+P+2kUE$PzrxZPC
z>;f()Uzx0SB=gCYOP7L~YCdvgVD&!n20dsucIIEKEmNpP62?_Drec~pm9?NbX=!Ej
z4kFF{=rgl2Qu&AXD)zynw$sM$ZtYH4EnaChtLL%lh(|ub{kx8mX&`GW>3g1jxZ<kC
zaAX@7+lQKFZa8J!=qKr#+12!h*05vK?|7V;siwDW>zoUp{p+iCY4_y<Lbku(@=a~)
zQl6K-G97AcI!ryEgN3S}rYMD-aeO!p0v5G=Xi>XG86tAB#1sl--@;2*zud1X%2XS-
z$hmmgT&NI*y9rO`k80ee^xpSIfW##GkkQ<M9roQdf=4v#c7+ZQurc9niJgkBs><>c
zyw3UgJ>!eFv^3cVEVBgDOWpE*|G9JJ*p2p{G=7x#nz1Ks)oU6>lG*cnZzQ4ETKahX
znPmM8eq9iknZDlb^u6XqIP)l-GIKH5M=0R#-UQ7<jzQ^DbjQOTbR<j5qa{e@-kP7^
zpDbDd6SQ)q=lmHPUPSwj$NPcuoLRNQ;cZ>H4KDE2&}i}btwz#%0UATS(|awXL`5hH
zDU3YR78e~NF?{>DX_K5j9nN@4?Gme@VO8~2)u`rcz`!SHRZ^XNM~tr-k3^kYBqfj@
z&YZDc`TfXhYm{>ovYj_Ff<s8^ZVI?4>5Y)&BfaXq-=u=jQ>WfsJAXSvo(wS6uS7;#
zdXVG8AMNc63koQvc|^^%uJ7L6?)RLzb0_PK|Mk<YIMr*^VT^4cd5ZRRUGKSWmiC0r
z48jIM#rEv^2{W|v@@wl~UWBe^yC`gr#+C0S%K!W!f2KHW-Kut7F8#rS!&^u6i+(!u
z;nV|1BWE4)^H)sJiS%mS&wj^q=PV*3E?Y11dF~~GltYJh6tDuY?W4Km0tKUx{J-kg
zrH*^qrlHi!oVku{tSBWS{Krq8>?ly8w7wZvwsOrHlSl2?=p8zIX!;L;R3>!=+6op%
zb)43h&AS4FVF3$rf474+kWIFUkB(#!gyxHGef#wr`TX;v>}>ed^V1?<<-MSQRxq70
zV>aSLd}YAHt*xywdzCv;6L-GJ`ZmZjH!kzZ6U?I55^3>DC)<#8naK0JlN=hl(Nt<*
z?}^j$ATyzhFP|Uyy!Lg+u!Y?e6b3v$l6>b*3y(A18;_Pad@#+TuqT9=Hm)g0GBK!q
zg`)?<{=`;fkK*KO<F5^@lL3DrrP7a~(B|~h1ViP)X3$iz*a@1EIWEW;K2PxQ+pBGC
zSS`J?$E0QdKm3VxHKyvUQXO<Zl2s;Hi=3Dlf#8%vIxI|OSDIL=?B9Re@uQ5u#_7YR
ziM=9chtDGsBU%R}U$=7Q^%W00{dLcYfqaGwI50i@Reru^OWvu)i;tL9xf`#07f@Ce
z&Ivh`ZO#^g?M>TJr!DEX*+&(ofKu`Op__(Eez0N3rqfK3XsWCf^Je}<m2q{smZHA%
z6n^yqH)E*8+BfOmJP8p*|Bhyo>i$7N!Hz|-<Ky&BU)vI&)Z~(+cly7`@kGBJHD|RS
zcl}HTJAdw6@29LIudMlNBGB>x%<<~$ucR2<yF1lfqbLRgmQrO`-=G=CG4e&p6wr<J
zNPMS&Y$P>>$!^zc^McX!01j}ile-}r;xh;AWvoH^bHDxy3S)sA1RKIi+C-MFqDA<z
zMvv_l80h)5KITI)GJwhN?@&`PArE})0>ZMu{Slh|UZi7R`Pr+XHjT;Xie0&Ak4OVD
zg2?_b`#$aH_HV)BwY=PB9oq9*vo3!NU#=*QbW3@d5QtdzAo@9lUcDBm<7B$=r&(nF
z?Zl&Ms(CR>e|sQ7_83TSWpb!@JOXIg%N?ImCWB1lbJL5F!rQlTYo;f1ad2Q$_%k91
zJ?>XDE{vNG9C$x97DK=dGiN>(lR?r!hS%{5<S#YnxOZNB&2}2HF8ULq_b6UCWMD(Z
zspIVdTRPd+VUmNejHJ9*Mv3ElkvXC(SM*t^!oL(0SnQBX;2Hd1DVLKksaNP!T;DeI
zxe^^d;d9=dB?Xcti?ts&nW;+LyzAn`|LbVed3zy?4a#W8fIm^u5!2j%ehQ;VynI;~
z>A;m<Rln_S-ksv%F?Y`M)>oHTZn^bcuorsXqm#Y;Irmb6-Mh8xrw{Q>vk^pv7Hh>b
zL#-_=K2=pIVev3;-gkOu3cUDF9d8WXxTEG%t?!MlYtsk6VE%*XDvb@4S)|v7aA)Na
zg_8d761f*)HTj56VCdSd?t-H!FVTvKV?a+UmPMFLY`tjH`FzQp%W)fma3GXnj_9Re
zB5ZBel+As7as7|gX74^E_IR}7swZpFiaSl?317Bo6?nSlKO2pY36^hM{uS+?beumv
z<C7}B7&2I0U58zKAXn@pg27C|+?YEAQNjmxc5u)FnFr~9lXPd<?&E>y;v5It@A)k)
zC<>JlenL#UIc)`@ZlC;uB?;Gz{aa488KkOu_!E9_tMP`USY2l6$U)oJ;cu(($^EA|
zlbJGq$WfFg2s2MPxqKWtYLqyaA=(J^^$QLz<iEIfmUkIGuUD^5NiMmqI{L@;yeHh1
z_EzvNkyu#yd4ZeTm`JlfjrTD)J1k4^QSQ1>#IZ-T`)eWLb4Ld;Jh?@68@lc%Sy^mG
z`9W{d(BzI=oO9Nx3B=j(IYX$fQK4>3@1!X^K|d9d+pr!zoaSC%@}qr3lS|d&o?UYX
z0Ds4&#!NLa&Jg4Eh=22MpOU)D$$?6C5K6b)A_hZLvxP4GUzy}d@7Z#ixle+<;0;**
z_3OsWhlDdl*05X0_wk=+Ike`>r4<{8_vxm($=T)INUNuLFC|Jn+<w2`kpJf9PT=o3
zj{l~UJ(i)|t-W^FXw@pPA$1aPc4*?(s>M8P$g_z7-y%W54y4&YMGJSKF6i?4=D>)L
zLnGlTN=i!DfJ=y<F=GQA2_{?or}FF{h~C=oO2dc$!066#f2Ho&J1O)y%6DOfIT+#7
z?9lgq{T56Zf3zgQf8LT{34`C?uhh<19eZ$Q*8c7jU1}bsdd;4P(~$YHWwZ4ZyLA(R
zU!1PUH<V-#9@rAZC7pW=RaRze1(?&0<^r~|qUPFSXUlbraubE?QekH5Y!e#(UR=D&
z7$rGr-MY_pvF4rCna%a}&CJMP`y>YlSS=3$K=Kn5w|ZVJ!&UC{xJAr7`2rp8cX3q?
z9k_~$7wUTC{JYP80qoDm!<Z15d{o2>Z`@#b3|tr|^S}rHPaSzT(X8XYI&%CT7wx6k
zK>!NgRDCQu`hB{8@=vWhlVl~nHoV&FzWqy&E6=EpX`W7L*e(8#$bHK!3Ts}#dH)f)
z`+V`{d5+ikTgf;Z@)I0Ka&@yWv>~kz0{p54mySOyb71h&6)>Jj0_$@`c#>v9LINQo
zQ3)hZHE8LA1<ruqp`MM=V_urhn#Go=RJufFWZ?%uJ*SNtIuvr)c-^|80=*&QwIHIC
z(1B=aB_=1cz3<P`kuGpl7oK5!jMyg?$i~i2G^+FZEsCs{i)_Ro^ZU~mZ&MP4qSl6h
z*7iR-)VjpqP+OXKx0lx8=ANyqjJK5o=z?a5!)TZ+rKF?;1jrvUo?&K23Ak$PN#buc
zItczeQ9Fj&a!xA>f0d8Nf`io&UX_Kn5sOswg54rkz2-<jz;Zn5<Wl}hw~@5bStR9M
ze26jeqxAH>`rA4Pc-^-&yJ7!8a@ni{0H9&fKwH<x@{qT;I1<Ek9cl$k+<tQYFQ9ma
z$(R?N2d%;il$8KjVBv3|G9Xath$Kp>t+)rOKL8Dcvk?&@{-rP^^k*hpwKDFG+L0hE
zo<4oilEScl{r*S@rNv$efx?|WhuRDENls<8-3=uh&{G~hR*dV^^lhmm3hI{M^l;#b
zWV~PX^#GZ-xrKr_MHh2>u)=Oby8d(XT5HPo8=wEy${r3W58dY3?ZmHtT_meVK5ypP
z-n(~i;$<b%e`_QXF9R<F8FWoJ=5^6+Gtl%sxqBO|KH_%nv3+~n<7deS#H+qSjKx*M
zvzbV0-j<Yv8iuiRt9VW`;srPk;V+8z0gMid2Ep4Qf17al@l*4qH%#3s%|Ek6=I;K}
zIzr7cV->+|(YGU!g9T{CZ_iTbY8W?{@TLm<-q;%)P(oBNeQF)qPJhLNeWb`XcPS_s
zm3Ux1L41IZ{Y6Tm=DU3>Z`W(=2s1o9>5AUO==~-41ILl;_h_mI3p?~Gm#zCRMumNf
z%7~Xlp?gU^d-V#iQczGZ?R3aEc<ic90<92B)W2;B)WO~sas0z&lEmZ#d-pyrxC#t*
zKWOvU`2xPv2HPIx*H<eg*+>Ye2tKAybQI3w{?4$7t=Y_(6%Ja04MF7V*Qb4IDJN<w
zD~Byz1F}UbySDOtET&~Dd7b`Ff#yZmt}yI~_Oz(|gmnv~&#Z=IfXy)4G(h*!pkONK
zetIzR-Gk;6(X_@49^63?&7Z0*F6d8c&8w!;Yb#>8B$96m(^}nFlxm&5efolzSO~b3
z?Y!IQ@J_;35K2deSXX_v#6)t6wvqsEfo^3=WoDmxGrf$=Otxw~&d3<PH1RO#ofD-(
zS*P58%3bfoOZ&z6A#qyVATcLW)V|xCAsNR{p0yZX$62JVU9XW|@V~or!Ss4{?>=aG
zI&ce4RM6eW#37$L5^caq^1<$|0PWHOtl-f}yZ?urwyi+GW$1&ff~rc~@*qHZLhMT{
zYs{WsS!p{v^tir*_q@f!u8nsS9Ebff^YHk3SmhEXncgCaaaUB9-bH|=sTovXy{Bz0
zB@<qtLfqP%b3AriZWVET!1%v>`Xm*^e85Sp6V0Sa)9V!rBt^=7ewnP!L;0R@Z_6?s
zKMstWKX!A_J#xrBYqSG6u*eml1Y5M!6AY^C0Zm2wOD472Pq|Pp4(K6XovAasG5bkI
z#>`NSSsxmm-{+j~e$%e2VhW4mG#w)zK0d}UUQH?zf5?E0A#k!?8@9Z{hyeNH#n@QS
zmgb29c33;U5(;O2sA9~O{$A0oEkJ>Ac(jhPcu?N_=b&@}o|^i7`x9K(TU+Y&Gu@b(
zovJag&0T752K$DZ*c`#RfeB{btD!4TX|VPu@1KiULIB2z;Vnui(=KD<PAfa+DE8<P
z*uAZ3tDE|iDOvqJ=kQ?ex7T%(0OY@x+Hz6LV_Q*-rEqt})iVefPM?0d`Z|r&G#8Jt
zqm+WPP+##H0HBiJ4HQ@Iw;W?a78XOcce{2rCBsZwtQ-pOjucmGYIkS0ki<5_V(2VX
zDiudRC>?-QtazO#y7W*cZ0UN)STwakRT%UHa3-0pVkn^5FifZv<x9!=NPmLgd@X_+
z6R?e;uNvT<!ywy{)mvZ0%)*x*Qi!gA0R@I|6-JM8ZML{~zY|#e`Y#KQ`*!?2ciFO3
zG7Q#cY=NsVI(X>NG;y>GWd!Zz6Zc*@K({R_jiP^B5R1}4S!!sJn>zp!F>hvjMa`;{
z7-`xq@BFt^m^yv>da{|qwiEbTO|6KllBqzaM@9=e9LDiJbLNb~SRSP~q%VS~07f>k
zY*`5&z3gPR)Y|4x(<?1)w_Y0JM>w09Xi?$QD%w^wHc}WD-F8;@x_$8A(+ks%2+H5z
z#P3`6L@;P)bMeF!hsIx*v^;JP9~-F15J2;Ggx{@#>&~C&u;LxlOUN#1mJvW8le3wo
zp(h>J`U<9Ktn)pt;-F9z4Jvut{+*)N^uFSLKjB#D>C;7U-%ND8312v3=acuCW8Q!A
z<iK$Ge2S#9No2c4u}cFcDKfS5opL4Ge3-xjlqc)~xjd)CFR&yky0N-C_6iNg7}`Fj
z01|94D1>P(D|xW`Iun*06X7oQM>ZzG-|&LTYN}tu?AqOa1`V@jdR(zS*3uF#+y7P8
z>5hh3O|};9jAvQS0_HfH;QRC3>}k{7SmGnhJgpVrWrC3x$LPm1-O-~DCwvAy#S>UK
z8y1F{%ZrA8Mb2<hKZSy(H!?%!xS?B_M=`gsz{_AGg4H!^m~?wpykLVK1R69Ogl9m1
zvQsiG9+aHual-w^Oq<G)Wuf;1?VBI@DT&^>jK7RI`4kB_KCN5_dfBDStblv4L>ote
z^_G^s1hhU_I}W!_PW-ljlAc0@O+f3{t&8Tl!M&DF8>>X(I!vEI<2eSHzU>KVBfS5y
z=GOhaRkKf9$jp&8P<tI(U$WvBBd}L2<~HTdUOI7N$B(2TAX{vh0|Hb$B&1$Q&h8Nv
z#J_)=BzwrX5O4SE*T?<lc@Wp0ux$LFl5}!gepQc&>-vi~UgO`5pP0#1Ax;KbS~py#
zE;hEbJbB<i=U%C^!sej8O=6l|U^y{L!6~N}MPP9Teo4@5#l;FK9hV3=(qDfU^*M?k
zg!@Kjm1?iK<NWB&6%W}EmCSi-@of07lC!DRbM@F&Q>MK4*d7(CCvn%@`-vc&?P&9?
zaIe`FXNRcb#H)A07L5iP;Vo@@PonN_Nn_8S_#y)d41=gq-^5*E4xA#X^8Nd%YQ;Yo
zSh6E3f3L4kh>y3cbT)`;2Lg6was0(;Nt9w#_>(IlNF+xlD$#eNqbZU6HAO03?{E1V
zW^YAqr6M4`bQD2W3MVMHO#y+24jx3*!;8fu)JyX$EHZ4ziaj_TKG}qpEpSKzDOTj)
z{9Fh1q0<8n-R|K*P02G)Od|hbsd$5pIyA3sguMJGK6}-_zMYZv<XzQ-rWrzwl!1gW
zEWSFu|D{)KPWbt>9hJAL_s9OiY`j$h7IjQ+SJ8I3<u#+plgkiDN(fL(>gt7z!;l=A
z>|(8kT8N%uRD25j1faY5u<nB+4N7D?oosV6SAFEjP4C7nfO9=^q>Z=u^^3Pf-AB<1
zA~mSZgt;acE?%4-K98-4G*<5U?Is-I*=3rA=b~o+jY#?uMl~7o>?eR%ut{A9#SZPy
zd2H$E+XaXx+HtvpVIHk_?$uS@@M@7a)$bH1DF`j?pI?uQdUGe_ZPCc~SsJV&y*8&A
zNS>Pe^&^wk2$&oeuizMZWOx0Xrs=b1`yV*a)wCJkI!G{yyL18-sgBx`cNz502p(RB
zF!H%%Dy#tklXHdRUc8IdVtKuQtxh7!yl#7cahO5Qup|})4m9FOuN-cYxcLF0-%XeB
zJ3XpikJ}xh6TZ7#u6(#4bitKXs^d^iO~sRu@t3jxCO8Rla&lO^b_q}~Q^~9QH)Y&b
zfrS&Mg8imVBm3^fl6{~0ER4E;eqp2S#?Pk)_$==H&!xrVJitd;X{o_E7u>lKA3Oj2
zWHn5E#^yB1_e1ADm|Ej23>AdY8~$uts57-jbdY5j*h*+%(v0Tnx%Eqfs0HYw|F*oR
zw6)vg{<S!!9edbQneGN>14%l2mKEr+<PNd{8)aUxf;{>>W;&QR;oxnw>0HiQx-Qrn
zbT*%#&)Ds(i~{6;F`M{o4$OKB2Kw64B7#2`nwqmPX<nAeh>^#C)CcLf_bAv{cKN+v
z?%j9QDor;``k}5K6&K&BQqRPi;}`OI_ifuG?}AG*<M`(4SgH6J(i3Ze+V9FK?*3c9
zWU+^EmJ>xjpFaH}nC4-rWrdQV2r>VOxhQmiZ&765x_x_7L7aG=d4l|4oS;+5xo?X^
zKGi2Zl`^3-3wdBzAou7f#*Z7<^myxaMr+Uu1V+cnt)C%zAcdhSL_P6sOlsc0zhPHs
zzDG7l$9cLfeg_{{tvV*OELvFC3<UBYQE1_-d57ub5c~+GJSYvi-r6Hkhj!YH_ijgl
zX;V%+$TB*~VZ!K>Ghayd@7Z%m^2)`F-_WxY2dE5K|BHfo_GzuTOO~jAA3~w<fA9`C
zeVw`jLj^T8<3znlR2L+FYAkqg?i{)2C-Qqv7W&gs{*jS`ccs<-`qhqVodFC$#PU&*
zXV2Qu5|d}qF7d+G1BkwB>`NY@UR~N-F8U4)1lGX2vF<REcgMt(bN}1_$$S(=a&k6=
zEQYbLIKlR$va)yo{-eG(D(CGy^;$Cf{T6^?ntIa~n6l45Daon8Nk##K{prehAtR&6
z#L4(M!=X+yrbvZdQJJ)4HgG&Q*4EaRM}O0%*j>Bs0|hQHH$Sz=)6_Jhe)I&H##j56
zehrykbz(~KiJ4k%3O6?$3ykZe6k`3Qbk6zkcFPtX3+D8K6~BqvQ_imjg{ExiEQsoD
zs=m5ddBbJzxioA+U0-qG8C}Ec6nD+0Cw&q?#X*pG@!2>o9Vcb-WW31C6eUjoF9>D-
zB%^;wOS{VZtFKe3^y=5|St!wz7X!!}a2SIOLGXWKK7zWfE-oKK8<>t3UsxpG-y=O>
z(Ywx4QVm$*xal81dX&`@ygJb9`}owRPGx$4S&LyKM!XOIpS4NVq5^Y*QIy-Mu6Nl!
zyAlwO-zO-K8fDM=Mzp3YR<J^YjS)G3Nr9KCby@8(emPL2xSI?4FCB&QffB~6Xd5F3
z(TV>Ik)OCkR=_6Kj=~N8jBgHEqs^-F!hGql5Aq2G?&<TFjdJ`Pc<+qv-2}bU%ERKH
z72jF^r*WIKboy=Wlz@TEk)kGOFVnYRPxoWCW%qVN7h-AbPbry^A^HZp=SzjHo^EDV
z9KV5u<o^S<x$5&2VHfF4V2*hKoFL|g^d%Gzxa?JY_`ujM%Tq0kjrZ7MhbsC5gvI=S
z`2*~jav>rD5#6#xy}@P|BgfAAuNL66a!wMQ`u|Bjl~Hh9l$f)M`tPPsO24542euJh
zYqbBrRZy)fYo5JZQQVTgBlKj%y4gdH)&&l7iXX5)Lwastc!q-Wpt(n5N6gTCea$w>
z$9zDb!}T81rbR7Tnh@o4d)f`trkGUyXZ6qQ^~Pn5&z%|Ew5?@x|J){z4}bK|*V`3e
zsh{}qS10z~j84}MfxMx<@i`n2ki&f=BoLCE*%Ao7&!f~`ePLWgNBo|g85Kcq2&;o4
zmFF5}91;ljn;|l7wlJ|IjwueK24{)8{tgn`+3t;dD-IC8M&|(g0n-?cx>9l6Xu#Ef
zX}&)}1i)3^kas~cgw)a3-#-=g7ps?of)?2;Ef9pC<`hwA3C3z_M&_)gIAv?Opd9r~
zy{`CShHSr#>Nis#XMfDyGuC;1?oGQJ@&jxADWP0j*8873dp3@W1yMaFbR=bjTyaV6
zZ4ElGh6~>4H2x2ZwpQ#i3@e?x7HcfglQuAz`79179+Z(WF<&w9V%q6$ArK+{oxVmZ
zjEve|`fv_6eGnzs7V0g)FY2%>SFU7x9%t%|hXp~P<IfJWoeBEbeSp`=ZfN35ViZBU
zz+lm4!GEAs^D?$|-?0uci^LMYvkLe%e89uTLIzgKD=Tlq|8PzTGnBJ9Nxe#{trQd$
zgV0J%%Fsur0vk5M@>&JGFl{?6n3j?;77Ud64)ze+mL@Q|KrY)I9e-f<At5p8)1-Ut
zvZ?4(SyY^m-~x0*kwv0vI2utJapg)X5DYLkeiMDrYFb`b!$T)Srj20Ggq4qjT_JOI
z)}8Ba%kN5Spii4)zi#c?PqDIw;q!t|ooaj9xubT>*|T?_J(Dr1A*OI6KOM@^8LFA!
zf8<E{z1`c-xx6_s203NPD617yYjj_;!Ovl5HDQ+%b1NySjX?+4LB<*4`>+rIdjFek
z0|C+Tg4(YiKkoagSj2y#VZqSCUvBZ$746iwrIW<CBT{W8p&_kz7@p5DoHAu%BNO^8
zOycRs55@hbpB&OX7Uuv6vdrbJzXRwlR9(4@<D_2W$BaqKG2tR`{U#RgYkI0Tb(y8*
z!~p{&(|_}rKJ3Hk-z_&Fz?5mXzV`LszMa&x?{~qeUC8uAq4*!-E#ilp>zWws=jd-k
z+sHFB*RKuM2c?CQ+9B3fR#|tf4j<LndEt?~>O&bd(|i3k0kLzGX&>Df^unxw>Ws@v
z!Uki%_gEtjaXXI%wqQ(C%k<HBT3p-$Hjpec#Aamwy7`(DCj4${dUwzW-hkp7M^Crb
z7Twk}D}kwi$;{`@#eLhx&``2z<m4?%EUA6xwY9p>@Gkv}3JMz1U9a(MFUL;R+S*U9
zx0XXG=VA=gO4Zdz9aM``<^y$p{unwaH`gsb?91nup}I|splphf9XuR2Bu1u`eL!U}
z)wrFcBRCyqEhe>ByLBLI`{iD|d>I0?m_Bkq0OEw(YzTc7%OG&4-8<z#rpon>@=QQb
z8?sgOz{((B-;Tu({bs5$O@tV93qxQ<qT!6fg;uFBP=<j%fTAq4dX$2hLia1H)3XZx
zuz|3s^H<9wRfJcnC3(Y!q3sVDAMo`pTiaJo&Vi{x)Y_d276{q~i9>%j^a@UNm=wjW
z5J{GX&q?%dK*s8OXLXqO#^H;@{t%6>b+z?u4Z$gMVMl+$wg{gzHQ`kz`j+5>qNUW)
zFbc^cEiIim;~7PlB6MogmMmV3%>55bl9~tZHe_8YujGG2F9>KeA`6Le^i3k>wZlMb
zd&B4^(`(+h{IEPLL4wP{L<U&BWd)<ibmaz8Ss^+>2FkUcr(LOK5ep4=S<^iVeFn}E
zZo9Zd`WVE-9-p|`@I!rx{O)^~q~}}S%6xLk?d7ls8q4DoYQKGBVVY!!8PgKYMA^2z
z?zcS~DL-k46Y?^{wa9kuTAW#?*~8@>m^%2t4YwOeS}@T&)wG;_%L|iq884z>!d!!z
zh=P*nq)<Dsv_ls9F!Di>d#3HR9Xlp76Hd0(+REO;h^#iGR(8E&NrMLvRWO?3CfZDn
zAXf<`H3yONEDp=n-%GVWtEsR=3GwWwiCAYeRuyXcJm=V0CMfTWyipd~+oZOe&R5q#
zU(4+22hiN5r7g#4n!x16r{_u_o`bO<nA7@WFG@)TlEfND7lvYcAKHoq1_WmLH;Lu?
zT1hO7;S-f>p9LhM22Ub@!qY{YUnGrKTle$LtJ9R)ZD@JhY2CJM+J>SMgbWsAwezEP
z;Nin3COWQ$I44E(^(LAHoH(&Bbl1;E(?0aw=6uz~aOL@*udiR*D<j{#x2r~AkV$cI
zH!rVeB`a#eSO#cs??s;6v#0&(^dypt_j8t#wOeO=J_p&eoS~$i@q@7M#N~!Xn<M8%
zBNbsvFJ{DczhxE{VDnu61whVM<5wmw+OBtv-RSH@tM?geZNFT16O{BWa24b=o&#Fx
zRyfBPRQSl9oT-5=26h!7Ba=R~37Ni8VmoCOy>HUM8Twd&BtDZ8guVa`0M~TOXlMr|
zKg4}JRc=_gbZDNI>52bbcYm|U>;6&1@W+{%%vGxR5S|!o`1<kVUYC=>!OQ~f*}a=l
zVt|tCk_~?EzfTr}Y0kZS<rMBQOJax4k6<8q*{v<%oENMMjh1vifn5vkm-Gm7hFBBz
z%~jm}*m*p`3RsROjkTnHzbjlNMBL8Z?DluR*U4^<Bvnsc{tDwj*A#T>oy;%(44-k$
zy-x18t>fXN7lto3D!$%9y6?WlOL9`XPTMw+R+0*iGe{YIb>YJ{&ksns<^YZ^)i%4E
zW}Eio7Au|V`n&zPOr!7W;4s?yAm&FC@jazmh5IHc0$*`XI&F#F`cA$%-tVJs$DI={
z`Ta6Td-JA|d4Ku_03lHmKtvkaZ90FKmxTp>9qEaR%7XI96;h>85uj_sOD|^hUwL(a
zaCUHQz4-c=YXdDta!bEbIBOjQw)gxuH*VSbg}?2lL!)LFQLt{@^4Uq4<8W28@2r9s
z)bh?}-1ytK<t6a!7}@TTqby%#_qDdeU<F@N#mfGpCrgRxo+?5rj9gqToJq5Zu}*1Z
zJVPoBdkvH^*J|uE^K%g5ZjK#zTIl7UjZhG+Ziuco%w<|fpVJ{BZq3(!$q;&Mrq(1D
zyVeYOa?f`9pKp6NI>fcj|C-zCZ?HCG!liYE8awak{odQtckKA_r^m0G2LOl%5SbFa
zL{gH8;M9qdgFgXdsoQO~9b-65N&I?^`#?K-E>xEt1ms4kFk+1z)gtiX*Ufhr17hX~
z#QuqEe%ynSz@jIeSer*leyi}^G0DF6Dn~N0Kf#u(z2PUS6cPrh`VL#t7A!c%hFI(`
znV4+*b+lk8|I50J2ti_&G_DbZ=Mzu^C_;5nJe|@nDLbmsSNrse$Sfb-s@|HL`liox
zURa$?My-9b`{T-$N-d>mc8gTBL#O$7X@4KBfd3&Ddu<@<_m%d^ayA+eP<w$~vgxYq
z+oefcrB#en6_`<Qb&gmBYHQBK*x|$C82~JFpZas;$>R~@SM9&IzdIGq#rR(d9b_xB
z&xE&HxSg=feu5!@-`*h-4Ju_lCxn`)_;R?(|JPcY%RA18A*z^Qlx{N`QLD?gd<|A_
z^inZWJAz~xU96Vz=XR5%3{6c;I;fa0UHUXdva^a&_`Gk9c1thsQBtq!yqA>@bprvv
zc$K}kIYCd(qz_a52nAUAe&q@}gsyQ3g4B#W`38(4%YF}bmAOF3zmnKaLTDCHA#sbI
zTt!c@nHffg!zk##c68M~{bG-P+WfClfpL2aQRQmgjMMApGW_j_&=MIh2ftG(+s$^^
zVTHBkjfqrvL<D*6<jUHxHn;r9ilP(;S3{@*EMv<KHT2p+Yz%&<nv#K&&az}zMl_$R
z&wCZH#qs^lUB7Klm`LS3I=HUr=Ap-$Y0GOS0IdN&^X2&@Jll13A5DbX6{}a*@rFDn
z;x-vZG`}VxF^1j4bCbgBTMyt#AM60KxMc{JCdy#(30}Q;{`_~JKlf5lcu^g_Ioz1s
zR9yom4*u=csYM<!95pUHS7PYR^oI}2BC7~XeUm&}{{Ta}`0ag1z{JIEs??#LV<hvl
zw>mgbx8Nn1HoI>@nc+97u*Tx#wNgJF&b6;4YfX`wsI9%lyrZPx{Y>1!PD1Kbka&Nn
z-J84}-E>ySzG%CB@tZQ)XCAE&mTlQGHo(-$!NGLqOz`@GJ03bFS(jxhw6EQ~Ns}-e
zni(a<W`Tu+{*ImM>VE$?0aTlv?QAq${Yc?;+|v7fvu<mW|9owyE*nGi`~PY=Cu}GD
zzFRO6hyeuZnd}@131L%laWPS=mHNiUrj*Z9FvEoTFUcUi<s&vUauL|CL6Hu8OvAA`
zdh0>s^H4`+m4mec)+h0h(vq2j)mU0)p4orQFbp|Mn+~;he)k<`?%%w1tBkDdsA0pj
zu}DV~;r^-sa^-K|2rtL9y_O$imp3u>p_sqA!j;8m{tK^FUch$&EEbL0JLaaqdsgd9
z%KVzUWOfHZPBB2zm0MH&R_DQSwN8StO>IoXjn5YQL#n5ny(*I(xzWK;S&j+%(hncl
zMG*;b$a%(9%uaI2ts1V~fFqfBu2XmrFPL2zwZq`<kB{LOE+p>V=3OAaVZE(i|DcvX
zE|N8k4hN}xQc{*Iyu6~NL*?Ksf(>V%D3g_)jh;_MHCj?%6`I!D+nXJL0L3#yv_X_!
zW@Iov_?dBR*W+-u^6IHwb{JrMagxa!ekYE>4D#Br>Mu67g&d9Kxb$QtNS2+N?)%mF
zcT^W4uP(2sxPfeu|4vZG5#=;LOc3KX;OfLfO!fu?YnV97?*=YQfb?U4SSZQ}kP5zh
z6p)IAOP1&WJM)mAI(bswS9HWcmDlW%fZ;}U%_@2s5-c~Bi-bWBn(<*{#>hvuL8uXT
z{rdeZyim8`rQT`x3fws=E7xz_7<29%-nE<g!_)UgcXI}u6w>Y6cx&wIc9-oi<?Rkp
zz^tg06$GdKGL&n|vE6j6jF<LLw!$odPQK6IB8oT+(STJ0PuZ$LHB}a|=f&j_j~;=s
ztzNSx_~c1sbzp$=&CR=Az3(k+IR6Na|JU})c6H5`T=h6JKE@)a;>(lfl9_w;M-K@2
z^!_~q+Bm6A5oA5hxs0}B-GB_a4!}5Hlf(sv^73ho_x<GX7iGmXMKGGS*`}s8p+_)*
z=QOhHxZR{0q667VG**M^(EAl)pp328;w4Mo1vCRu{X<;4t32o}P`eB=TnjX%)#gA*
zh-li#O*IRf^KUncr3s(AOXtpxPEHlo28fVYLPL9i8mDP1U{ameUMir5#>RdVxA;sC
zH|CB_oqAOhWYmeTYx1YJ@WAxKuS(momQ<iu!?&WVI1^25(AuZ|&@4zkzk_`~YQNe8
z2&+o_zU3$~K|B7^<n%2{(5cd1IU-$(AVPmrY3$-r#)+Lup-q{(;^`BC(Mdgyq7ru-
zTs1k$9BS@PM8uY~v`Ob(rez)<G-Lg;<h4aNt5UnBti<s#ecXPk6{4_8o;!+~XrkY_
zbKlt~;%rtU$OXgy(fNxu{#I5YGu5C&#Qj0ErV}P~_qrSxthIT?;>D+cIB*L>F_oO$
z%vuJzd_+*aS#n^$Er!Zj+$%@2kNE)uSPMx+@yC6~IYUgFfq^S7Pe*_9n1Y4H85y?U
z$>#;>-Jno`kWh3^w9Sx1I7DSZ7~~e|Z~GaQg(-yX^5x$!EQ5#z)}Sn-f<oy_iNGY~
zmONY0xrG0NyP&ukh(`?Rmvn!<?M!BZzVP?8bcPEdMi=B=$DT|^?Ft>ny|Sf1Yv|XR
zXmj>|KG{o$uvg*>8XQq%ge55ZS5Mo;uDs=kbO#T2QA2t3s88f8RCCTRGH|`T1F0W7
zH+|=IX+}Fy&*QtymS5jvW>?gYK~4n)b@hE<wsg_ZnI4J#hu=>wCA{#3Ysa6aLfU*V
zS>pK!vexgIxs5G<lGr-!nRfNSOtl|w?~pvAmshjtkBa3wH-~t1qww0fb347-aCGrt
z({KU7au|G{N(BMnmp2=kDmkzyx$`Crn(fZ)E~K~y-dD*>laT8?cif}?+D@M$Lp{+E
z;NiLzUAILeO41b{zkZ#>W%qf`-I_Q(;@j+X_XB!&m^~qV=+=S+bff&*+G$sFt+bod
zriMjD?HhO+*UzQ#{r&`hky61UNF4X9t7~>tDsw8KqCs^uSt{#{Zg5e@CzpZRa4$M=
z9;9g<pBTa`gqcNjXUy!a>DK}W!})O~_*@N5v7(HIwfq3srt`TWaNWA;^CNFA>pybP
z`}9k)gJU=i@5<t#PvR7TTLs~^ca-HPeqZumO9%}L5x?FO);BkNqmIF*tNhp>)7n*|
z0cMSrOxNYEUVmEVLcwQirDyL4o`YLJqa;osQ#ImpLy%H>=KPaSgTRH6=64H(kTG5@
z2oo8437a#dt>kj>2__gmn#wSG%LeF^WVW2oPRt;(Yh+8%=&#|XZf-nxfvruZyW3Pn
zhW1XdJAdtcP|K(z$BtRu-1wSn%NQu{NQ3?6tuCZuZT|u5g={_PK-g?vKSn5YW~bcQ
zB&pyR5m|X!N~TwHtF(_lF>(6(TC9dRG&;w>m|{o!_~=lrrmJuW;FGRRuk-RM@YOwE
zkKh7Ozgh!#xz};acj5lbOaj*&3GT}g?`q4ZQa<Qzw`jcA?txA1`ukXAeE(%<ncTDE
z_5Hv0ek?2741NNh#9pqucMYy8`f@Y`SNUH}7S^UKK948-ADlDP&Ncjzyv9Sb+0S<@
z)jQV*hok-7)#VHccd0BBEiSma)C3{my-|21sn?Yk!v1h{yBcP&ra9|V2;N7%Mx}Zo
zVT(QYW(HZgV&|?VKP7b~r7ZPrYgjFpXt%^};ld|@gXB^i=PX?K$iFh`uY)UKldjCJ
zG`ZxBs(h$0xi=%^n^#657%iVYZ1CXp>AeigtR^e|w59vxO1WB%{B(#LYoe-fa^`9n
zYnS?VPIKt=v4dsgL!3JGdCrXxxr0aVP=FD4v6o_Z8Y>xQf-!hCkyVl1`y57K86Q6(
zy_oJ9pf@r+yzF)wcr51yc`4M&h97I`Q9!HzhWx2yak?a`rUN}joB|2I)OIc-aGbV@
z#TmnnsSXE0c>3+#*uiFUz5BSuFMs?*)7@Ru-EGYK3T=l@TU8Z0jxt`0SoEV+K&&RM
zUwmBL*_^d}DIlnwsv(|uUX#$|4atNEBGl8ZCy#8umH#G!PWV|}zKkb#FuFt-KJNRs
zyf8E}<+o*F?jJsxhxKk^=qp$bSL}?X!WgFYyL2=~pM1?n29vU@L$r6y_&koZIVfLP
z_{@la=n@8<#ngA<A594>q1to|tnQayqSkkF)=|3K10Iu4THIAyXP;lK6g_9^>!~4z
zlnR2=O{OqZe7Z<WKNP`SXx=^czC~}}79L`AAN-$I_x=gIe^Wz4T#1cHiGb!1sR_L5
zBl|rA$wjNtr+@$KJMtZ75WYE=(2ie^9e?OO{n#C+X<KWo%Y#NuDm0ybGcj>|yY^!f
z_YW6j_!dyafOvN6*YEfDZ)IVz_Vy^5rCmhRh+B?>)t7zXcLEQ2qL&4k*qYd<xoJKO
z0L4LT`9VLwy88N7|ASq2SOe&Y%pLkrYU1&|XVK;&uLn6MwtUU{+o3)>SY4?q+}p|O
zPQk$ccu0td2j)%u`~~atZwJtI?%#5n-}Y|eOI3w+z*(>)stQ9P<wlP5GYsP?>%#7Z
zlvW6hdlG@P9+XlbJUa}Apv69L0I7rMok*p~JguCec!9`S4B%4(pN)#*6YrB#xETAr
z_2u5CDox+|8n>HizshcI3U^%K+CG`d#A*=ZZk<+P5r3|)N$mlNB=`D}BdeLk_Xfm0
z$ab;|7b<v0Kzp>7@7<@>cpgu5rULjQRLvSqy8tKgCYyuKG&grXR*AHlKhFrt`<ZrU
zgyFPGccc>gIPd=c`VOVu7zS(BNegog1aAJ>!6Y)B0Mp^4MzO|S+K}~E7spAuE}Uq9
znn0$%&;LSoOyEE!zlHXB`O+DIgUHU+eckNfVAK7ssJ#`t-@%NHBgKrFa(U^k7mprc
zBB!dWtCVi4wJ^PPkerm=vxpB<zUK_QRpL1?WqW3%wDXG#kL(K9wyw5oE7==QHCTVB
zb~taNiGHPuZ4CEi+Tt(PCxLR2X7q01qRWfpnYO2~pS$hFYA;n)RhAaunVtkJUELDL
z7#dC{*xsEwzq5_-cBf&O1EYOO)Yt@##GuI%&()<A?}qBS_eh1CfB0mb#geu@U3*LL
z6f{3fPk-lJQ$Tgn-sxU)^{CcogYCyv+Fb7cA>#b`9~iPXHxi^9G;$c(>YlQa2LqrF
z1QjfnWVrO^#$JZ|48xd!g;k&{t@D;vQ5`&({aP%Q`;K3$y#6Vp$vcYdQv<N+%-y56
z-Syq8%SYGc8coXgI_{UzwTDkiPl_(EMma;o@-b9Ws_N=}di3BlK039iWWt~2VL59l
z=j2j$_1n21<i;!~Z4WSk6>g>N4D!2<^j<S7>smrWx<qo^k$7eA4rJ39>iSJ3cT&dr
z%^dKjq}~Sih`lqyOHI)d_DlcRcd3;X5+}BP<Erl0=Wu2PP^Gm9>{nW7TBz!6KP)YM
zZ2C;2<V6E>ot=>MuXlDbDt}ff>oef%L%dwt_&f(EAZ|UoIs=aO(IYi(zIxMrk#AnR
zcF5ys3Aytq&Ubn30u8`x1Iy*Dv%CbkfhWYUb9V4zrrIf+`2EKI{Vua<X4;zNgJe0@
z>Mja8tX_gibJ)m`qUe$EIsuxcp9-uu^EgjDI*mPvYdm9hzj*32ns97KmxUOaVNxo2
zAN69FE@2VIbq$+8mv{}XeKsNMQ~s6Lg-NSk4*QV$<M-t3oa%?(qi(89*>mI8i4_HL
z%LlZtE&3yx?lXRXY=L;wuv9ejSutspFhJqLi#3?Mwygyk=PK*Et$O>7LVMpW6ON5w
z2-eL9OkY)@U6Zt@yEA>?mh9uG;mXcvu{PrQY9agM&|QB~;#W<rSt@il|2K$C1}yXU
zmOpjsLiXtZ-7P+~e{`;B%(Q%c{Fr+1c8`;tIt@4@voDAT;m-c{vi*uaSWlP`KeAf(
z00o7GnY6dXoanr*h0Q?+DtB1VN)(Vwwe($5VKh>KI$3a1d=>bM2j#T!9jl_w-i4Qg
zcXyaUuCZJK%+;@N--AKwfW_2d(*c1f_KJ#{iR|#gJH6T(g-<t`nu$=Ox$)SVK4WjK
z$_$K<?$h(0`KfkFa8dzh&iKhKj<ea}z02Lus!Q|zb*}>D<~sab0|!8+&BY8E|1`(V
zbx$i_m+H^3vPyaT-ZVrrp|iKy%c=9Rwfhn@-11y_xHD*N?8Je5*ypk0>`7D<eDyEp
z^XRV$pA{u{s2k{x`}UzS5Ru5ok6lwP-MG<B^S=ISTU!z5P#KTFj;fb|?320;2#3PX
zrtEEW(9H&&**s<|N;N;>PIR<yQc9fSE<B7FE9At!ndI)~mNV?GG7pI`&1kautvM-w
zKwpyn1ROg?zO=WpQo<s$yu9t%C9^`bkImNT;6r%cWsELl)gK9igFZu5)pwGrmZo~@
z4j6E5gBg#TAhpwIK-tC^^I;PhFnL=?rN64bX}ZHO(NFusht<^yVBDv>oLmrH&6ur}
z%Gi^q4<EK!vnG9v(@>ipn=@~WAGBo2?(XUhMm2fUj@${qkQb`kImbOH;P~-Jqwr~%
zxb<aTUa8{ajX#tL6}3}#;37qa_o!oz$y#ZH;~7&<X?l;VOHT_<I*jz-cl3Q-X-$p)
z6GzGCU-p&V3(x&bmAA#gq36T<$RzRF_qEy9UF>kv-@f(W`$>gSC9qh_CwqGX>^5ap
zbrsQv<qdoGL`4m$PJyp*Rf~*{9$NgJEk!<u@ili@KItD$_q`^L_y)4-aXRF;xl1I&
z;kp_(N9y&vnB;O5W~f{-v7nF6EBmcHZ!w4NS$EO@^RugGJ_Uu=R_fr-dNmh!6z%%q
zq^*lT#me*h6bB2sMznR1%Wfln;VbkRwCZ-^3}N4aD2YkC(oQgkKnjFd|9xsNeh1B`
zUDI~9EB`c(JE5!rF@{n*eaaXS`8a=m;F)jFVwr_dR=)f8t&$4j!^c<iU%z%kTPhXC
z<9==b{gqT_L{}*l1`1#O8PLxNB{-sWTl<+|b7%^7uDqbJS9_iL#e1Vte};^oR{3F;
zb^pDwm#w@So$MC(Hg~D%0UUELe;!usa7Pw}JY2NZ-f~PE^vkL<pIUGr?cu|AL2K9v
zyE@dg<i!gX$Qqo_(Q3@a>9X&Aq&|)perY?D3ibPRue98_dGmX!mc*xZZ0n^tUdvL3
zGksR@aXEC8_)8MQ*6y^PkvE~1*+(WzWeiQ1EOBSTodPn&`AZMO3<Oj-#IC&oK1*09
zoZ^c@I@lkE=2h@U31=As2~L^Rt<#`!;~YV1iDLcPI)JcR)N3O-=jW0ayVA=tkn8$N
zx3jMi$%&nVtQHIbx}^7GP<fxx3c!_(8;76Sk_xW%YW#SbT?T&;6*90&XO8+K8MOQC
z*-}(0Y|X-wU;|JVqXpBa1EUfn4Wy>c8-mk!L`0wb*h5B9vW7*L8ND7qp4CU=#&FA>
zXZ4oW&t0(K*Z1$*6d+&?;LBdW-zy^j5IdQ%V-;qH$#&^td+ht<d&|0@exdNfj%NtV
zsIFgkbaqw(Jpip+v7(%=kTPiV!C5W!y_FTj4vJy}bq~|I7S~rF0&Qne3wS3w;~A<7
z6C=tU>m!LkpcP{OF?8s<9|VMSxN`~=;9$|qe7B*9l0Sd`soG3Jd6;3+Eyv&26B4)*
z$9L>VgIETzl{e&aq`*0I)Uj~cBe!DgNtEJTZ9a8HY((KI9`32l7w#c`VEhZduC+z!
zvH8z=|5~J&1qaUNf6l+sKfu&>?8%}7b0^#9yk1pNkmCAn0H(PJk=Hw3*|WzOc?TYl
zOVLDZsW4>{+Apnc-@bhWGspaq^ZpBo*_L1}Kvyx*?|57Gf#1LX=FO>NvL$ccXsc*1
zZhU(3MqJ#<jYFT;DR;}Ld(?VKHer(zvc4$)qSvOX6ekL(edFU1kR>JMP6eCepck*!
zt)oostGE5cz(vWWiubvG-@YJZXWC{<#Zn8BJg_b1THbPxv$I(mmR1NeV7qc<?uXDc
z)OV#vosqc5`2M-~pQZ-q_A$&xmZWz6q~>7K*v4DuMRDv~oq4O4uYX`#s$ZmZlQ&0C
zLFWtbn#_Iz?#whgeQIzi70DgoY>rr9P?0iZzIjx1Wp6vR9+l=GdMvDGaOF*>BU#&q
zav&6Z^L{DH%86>6NhUI$z9~TfI=j+7oeF*i0)QMI_I%GwHAYlktQ{}8V>JGCz~nR;
zTvJ&SEp!s#!J(s<-}}^RWxWZF-I`~8eda(`&X~cY@`CmBPEI)TUcmGr{T6O1Y+A;a
zpFMsQ77T~1d3r*=Uq3t^&P7IwGz@sU+L3k9G06_Cp3#K~d*?*Ikm^1#@8EL-pRe-0
zcgd;_eSOV!vxj-NRgb#l{ZPn~^EEo1cgQTV)nz~j1>f;VRA`vXGZjJFk;ZbiAxxPN
zFMk$1sPy6JtzRwZXPS>wFJZe-MVxQ2=GZ#$bmPCkGdgqj?7?zUi2PRQs<G|L;>KE(
zLu&zCHg0^7eF8Hk^g2N!t@b;9NJ>h&zN!DMrgspZ{rcSjqSI*V+M~xKo6*E7UP<L8
zic%e`O}zvMlQ)I&?k2-1&Kh!zly(5Gp-6-bvOP3{Qf`T*B~s5?k(U;|+-Q{%(`#XF
zKP2+Y=$<<|C<)m*?m9hpv`RR=m)b51^Z7P;Xp(z^(X@`)8f^_P?V6JNAxn4S{bn!6
zws-n4_Wl_vm0<p1zJa(ihyH{X&^GP(Vys8w9%uMoi9Gw|uF_{&T4Q&eZ+>?VoO<A=
zlQb|k0>t>ioy~bsbEU!{yX7G{=m!c^L-lJ`+Fyy=GAQ&@$of^kMohe=H{jf0lbW;R
z=r_}W+8;lzWHPX<;Gyl<%{{ikZ0VWYeSf9h!D1SUpW|VhT=mMh3l>Z0J^&<J8^4UI
z+q+$B*$jY-eUYeL1W)q|`B%Ds%&r;d7*Ww08@tA&&xDKBkB{B<Y1_L*S-Lrjio=rI
zF=x)*#j@F3Q1dM{5N&+k=1kzHttL;B5ZMFE`S11_i3aRLyLEDJJWf>XqoUd<*}!dm
z3Wf(%KA@lYojy(6*V;TVb@{FrdC?Vl=GJ!1GfOXDUJ`gD`pDn|uD7vAfsp{b6P-Eu
zE&LOef&8}<XEz{5DLwWPV*9>xDd7sgG;DLe0R79+w)Hf5WW?L+E6jb80jyg?pa<{b
z-2mL`zI~f4#Tv3Q-j?^H<L-m@lc8;jO?gAqcirDFB?S#1-sQLrm#tyB92SB8{e$$J
z_dkF0_>%PXWvOcRHOH5EF5f?JmK95*Ak?V@Ete1-@rN0Qk_Ls{X-Ao#wqPqb@oguk
zLH+PNJ-I_mz`pbS1E1)U^ZNTn4Xj0efcflXDaQC$12egv=S=$eJixa)9A^U}Cz$ml
z$<)<Zz2`ndx$;YF+;qC|<$kvHiBNkT7v~ufF}AEM`J79{>7_?fy=Ix2{dg0<!nc^Q
zXg<sKx=$eh7V552ia$kjXRvKrG$d^=o<CPUl8IIW(ul4Z6_+o*NG>jNAQFQW?vHyz
zeZ}l9jPux@U6LO@e?D4OwE~XDZquemVLFCuS)vE1P+e=;()=Tv+6*=dMV)W)8a~?D
zLSjYG$js_OlbD!_*T(a%f#I!tXZ+~yjun%BAbM`F-8Fmqbnx*OxQhp_#?O0kT+r^r
zggt+&#8rR(lBd%|YTESaJTaaPwL5fwY`3woVJ$F@1EgCZaBMa`!BS;H%Rx#YFY#h!
z!lZl3eLjFez0KJ0O~qezmhWudBMqM+8TK+QtS<A&p19H9jBB`#K+4U9_`~>qJILP`
z<VJ0d4Ge57dF_D**vYZ0_FMREr<#XsXA|y_WgYoWL4n&_mTvuMriU}6p}FIah^oTq
z4<ABuc5u2`C;Hv~2Uo}WQ8b@kQ0_6HT`eRD7K=tAoV|ocf2LJA2Foeb89X4!WvFAf
zzOdvkdz%gT_TkXimOt|}?Z2;^(qAKHsPfqbKJlxo%E!4`M`?(k@6$-$kmxRAB8QA2
zNY8+TCpJD1IbZlp(3Uu&pmG%rHJ2taLEJwL>``7FVnD)AdY^2tlsU+xq|OGaZ6zg0
zEvI~Zj7FIqYH87Zb5q*twol8fuDR`%lztZ4D|A*ICg<BJ<h3GR>01jSE}b8>j-siY
zS8jT*!|fR^Z;d3z>?8-uY+AidXY<>D{WscM8yYfsEH4$t07OpGJ(9q?+nZ@!Yafb@
zwPy51!#j7@7L4xJ=gGKji{?%?@jI(&fB2I6fc+}!(q8Pi=Fw>`oY~`6#w78}YWqiR
zD8&F1wF!VXQ%!^FI=B<+IjmdgC`ON#s~y3|5d&-SGg1F=+rm9x2BoF4Ty1orm7~-)
zwdNUL%hfyV3-Ix2>nV}?pm>I%;ZA%78@ep3H#l_2=l)*b<;y=cxty`9s5s&2c~H8~
z;hmBg%VrK5HfYfM&!1`AmRn0_*k0pd?Ahw2A}3`uKwc%U$<w!VSwoPN=C~BaqJU27
zfL;?OP*Gmt^Mk^P_Hjp#ek~rE_7Bn=dTEEUk#s3;GECqwVY!U@6CLp+<xxF22ua(1
zU8E$#tfrk;KHU#xb6N_f)wd}@mwI&d0Y)bM8%BkNNo=Y4HIH9Gb8fZ$y)dTozqMEG
zjISxcO*6vhp+>l!G;n8F&V^Cd-OlOtJ+$X`T{*+_vzXI^=ytS63(Y%--z5M@ZMu|7
zr`?j5zPJzV*wL4^49AQ4(RtO?oj*w>4(v|J_8J(NeL4l2gjPyqugbjWw=dP(x?D(E
zmgI85aMtC)`x=G-gt<$F;q(grI<99IVjJ2hnicCW-4iyx9#9&Po0}`vv#D8>FYPRI
z_R2wUSC{Ygi_s6_J@@JdrF3g?v_!MPGq3cZ(b1Ax`mKkAQr-bc*Y7FrTM83qB)g|a
zO0Rlz<58*2?buC0zxFR-R+few1yV*oe<Z|!p))bqP5!u|T`kf{xx4lAw7j9Rb_sWa
zcIwQNKx?)cCXa`)taz%bf*0N;&_qriS4JletfB-S*nOC{qI26~oRI&d{~dL5(yMAG
zTZ&P4;<&<u8#kCR3e2oPmo(7lwuhkzWGMfx@GXk9x}e`$k-JBGx#7w=a&}KojQrAl
z6l6TLegPv7;D)H^7~ZubQ7vCSd4TWrYfM}H97%CwC+-Ptzw0XMCZJx!-V>j9-PdUM
z)~%$TL&U3ZABa|QnG+kY5C)iOj}m{WvTuU@s`{!8k$OeX!)Eo-+`8T)QHHSv?;}U%
zyVS!8ZS#SU6=&Sn%~`grzoKHo>a0_-mGZ(Q-`<}0w-4hzS|>mFzUB3<4&f_T{;mjJ
z*`;664Ns3J=cbyTsDPspZd<@jB-!Dc&<U}3N1HorEEJU6u9$V`Y1ot(AD_v)UQt^G
z(~`^&eSbf5EZACm9`817Gmp&&<Zh=can@>uOzFDuZ)hb-QrzJ_H5ryfIl8aDfwwYn
zQ5ml}!^9|K&)%+;m9B6Y88=*A$6sEO-pyqlB7I0!iA6svE)lq_o_7`A47w?_!1Yo%
zvm!A(XKl<O5*VEuj?Lx6MrkrC61+XAH{V<+H#t8o{n6v<k`gnI#Hj~PHUH_=ro-0d
z+5D${>kkUOjN7eiW1y(K-*X=X`K;4gIjEu$qkYZMF^s~Ui^EQ<yv4-@-Fx>xOT~|>
zcx2{``PZ%npvCK&GCyM0rCGD>3sO_BpRY#^gKGyb{5nF`tQm3?iss#FT23m>kT-*f
z)&Cb`Zvu{0`|b_j&5AN-N-`!YnTtYq2$eZBpv(`-kSPfj$q-4Bc`maMqNIV4WS&bT
z^OTU8@3-{)_x|>Gyx;r2j$<Et?<ekIt!rJ^dH$wz4C@W|&Q@kScsIysz5}80q!tc%
zOC%-3vcbv2$tet&0LA;z@GzdnolH#sCK`*e9uN`v4ZH`Ly!6Ws#3yOpQ{uu|Xcflu
z{imX~XdZFw?A)o}BDjKV56IeJ2PRwKr~(!G;PK;ebbVkH6rp$7ee^Kue*?9nAOWzh
zi(l%JUMNG$g%bmwoDG~NAe+?Hi6K8Q0e5tEV)|}vR~MA&^}zMlK$bO^k-V@23qHo6
zNz#&=0uxg*izwhAu@fIA?Ccy1ZDykequ*VEfC!}v8}X7DO2v!qV|<(ei+w#7GY}e(
zUb{N0Ozh#kjEyxcoZd#dnh^LFT=yQY&D&WW{_5VxH={`aE-NW44BT@HlIa1M-+!>+
zIQ9I}cw^*)aQJCiS=poQ;78qy-R5%<nBweAsC~9$j|1S%!^bCs<|ZKlV090ECqmXk
z##x(MPLfXCa}~x7S~7Gs)}j#pBO*>;pO{`99ev=(5WbwJDmf)Jm@!EvD%Q~d)eY(l
z>OzDnp;_i4NGMDmW#rs012_%kh$VbOc%9MlK+@Zglyz>Hbxp1;EA#RYMJwf12jPeO
zQDn1__BfH7d@j6t+|Dhy74BOoZ8xtWgJ=~&N%_vsX(RqD8Ho}!6dNaZ7a^KJ(a|v<
z6EK{e3GQ@IDk19@ny`};KprP;Fk#V;hLs}J5&tyJPj$Dculgoj{<rU3SDfI{1m~Gw
z8!LIWvqdy<7BD3umbtyK`D0FS3uJEw2e6%B5kRqr%^B@7p3aFMKOSN_>%e&-OzO={
zNU+Bt1?&vT!T`nyM|OHTgzDhh`%!dU*)1z6+4|vwF&Yx|xhSJS2SC*Xng>S`PGvAK
z?BhYW&G`HfsF4Rj07BG*-`!Q@h!7woI+@{gSZl^IA$`PAH;J<R^@_OH6%+_S2Q^N1
zBS(aQUm@JXKWp5s(3FGW2AcLgUMQLVy{NlS^Ub#$DEjt_*U!(TfRpCD`4{VsT<@&x
z@7~AgzhE;(58zMJ0M88KMZ~$z{5KVtv-ZgH^YWHstgws>Vk;i@>f-~Uh6OdsyMpnx
z5Y6Jp03+*wh9btu#KsLBJld69qX`^i(6kXYBS29*bv<avcuugDpXLiOUc@Cr<_nlm
zy$gTq{8xj6gDBuJzFD{I55AwA+8#?12Y?PXw#d-Xd>}6<1IC%5zlH7xi`Gk?3+hAF
zvPPnIOrTGa*;Vs7Lzw@bl#_MXOxgwjj#yWLG$`32`y;|6vL6%FUp-ND{BNkg!(EJY
zbOh6bKwXGmP|aia14TOtdJp56s;g1J1j#RyviYJX+)PNfk@!!f2m<;BJpu#{DrpY_
z!yHyKFzCV|j>U-A!bnnV$di9icOwvyf)k2YyrQ|8nIcy(5U$=IKOnk?4-zp%5Y<#j
z1wJ1mp5ooN-!4-{QjhPuLJ<#<6;5I7II?)^Q9vR33m`oo4F!sg+N!D}P$yl!%t^Mu
zvYbmVxd1aN<&3fd4wSvz+;?aQiqP9xhNmtBc714o^MYzUyOPll;FDCDAc)r4*-&oF
z%vfibN-~Cj-5c?w^Y+fYF^hH`6yMFkx~KQqE|}NuY*j9|!2b7A1+#o62wn<tT}xw_
zmOMLH&n?6I9V$n>I=(J|u*K&GHd<I)zsSkK^GhVFpDFOz3Gg0r7B9lef@1+9JH1z!
z?txj$w`uT)kT`542t(jfNZrDk06R#ut&idV|MqPeYh|ti%V~I3I_!Ac3bIZB>i`Bq
zJ%m93e-u{WW`L9&;0}%m$ZP?JnvbFVJqM8#*meXXVJpYBx}{^a0y4H~{9;yCph~h6
zzWmAmzzVONCX{g3HpxcVT6HU55r%sh5>^_*D~DcKLR1uJng*s_VWc)ja|VK7hy8b#
z<kcwkH8`Nmdu}KME!54-3~c)W@9wa{#Map0YBv3499TDix6^#|CreSzQX+i3`N@Ct
zjk7YVuL2CAauc#$oSWpHW%HH)fl9M2(b=mGxcM)CfOkt=`}U<`<Q^xkT$vx<8zm|h
z_K%q8)}EC*ix*IZ1~S|PClAi;0GWuUJjQyG0%OuQZr;r0g)405x3c5K7Lr3EWQ>fv
zvE^YHE)6sD91akP?|3eqzlsP5s63MCzSFJE{o|O32SkO~uRw&1z1H&VQflIHx?PQy
z_=og4T0DA#R9+qDi@Lkt^!H;Ad7bu36yl<<1(|=^^B{+F)pm3?SigQfhB89GjTRcx
zHd-<*EiC}>fAVlVeL(~uWEOCyy`05ZnMo;s=v#LioBK?)=H9?a1F{9Y-?(Q3ZJ=mP
z&CT~_tpMGC-O1j%Y<3**9k8!ncmMQs!7sxV#pO%|$4GJJD8O$3<p5F+dSs}lXI)>`
zTMsYXOsKnBl$Y_TNoC_(Qm88^@4P!CDvF41fVPy>)Ul%GJ3^{3gjHEh4G4GdssUdJ
zF!pl<)~5|x4Dq)-1_x$3KXsktv6oq{P#*H3Vrejwx&Y=B*T*+m7QgwGZL%FsZ`hoR
zW*=6WY}jxIdoe&9hYWnQy|!q^L`001l@jZ*_k*)tLCq=^4F%l{pjotDIZXA0o#>$F
z?4Lj5N$$jF{ZV#FiI=x(yp8kSv$}o3ZI(9QUm9<@bEU=EW)cyMJ%n5j0rp7C0HBYS
z6}1SEoBa95aKnQzz}thn!a(%nUqmjwB^{6Gc|c6@xH8_{Jz3j%YsHp&CGSh%GtL)y
zibP=*JG-FT_6K|tG10KTKKRBB2-lF!0kjeBJVwMEKndEB92bYeB*^eyp=h+T3F=PE
zov!$(anFKJp9-Pe#??14_0Ng?f~!<Jo^I#pGA*IY1~vlGPD>EZLF@*gLtvgRC@jQO
zN0F0@cuu!$+0twB0v7>gCD3^s1d@`Hge4W0@kn#DleZo2dd$^)(iU>{Ab2gwg2<$u
zwUp5BjOpFLaUp@qAC*E))DdFF5vn5RYuAh!v#>wD7O?V5nF^$md*wyh4T0Pt(@zX`
z(XXH`2AuVveFyg0mk)CD`kcldx-C3XV~a@r2kl9In8#veb#>k<Af0Qt`KYc$!%u@_
zMJbTC*9XEitmrLd3ov093arENHE!@=HpB^9X;_Q3E?igyWFay8*zU}28XJ$lCXAxm
z_nN3mA{~jP-~+%CB#>uG$uTm&h{!|6x0ES4IU1M?UBf6xnwERdn-eL8Y}8aO1+@_}
zh9GmuW<7fbg!l+8^(I{EJzhN9C>eD!)FWh~)uvS?52c_U?u?c$Ip%89h>r$LgN6)`
zWJj(k<zg1@9=BJei9LGy=xFwe!%mUnM&7rLzx|~jOC)2h1iV}p@yYMd2q5+2>ec=(
zw8(Z}u0ui+n7@TI?S^w6cZN)X#x>~d^+!-;4EW(4Ghj!_lCuL|2dy1M9Mr607SMO^
z*s%iv61SJOT258&hqAv5frEI0foMY~hfjyYV6MN0&EO>F?%9FffwB!pABY`5@AG2g
zZzn0YZ9^C=xG5lHAUqNLOSq2gK1VG;Td*RJ`#T`>9?NFZATY0~wG~p!%C)J7N|>*K
zsQ}m(sy~1J*?4`ARL}Z3R3<pW@i-WDy!Z#Ab%?Ou$6!TTV3#;?mkNVr?&D~HqSK2I
z3X<Nc#x2-dS^`zg58{|zTCh!MgZwKnnHYJ8NYR!*kf(8bVJj<YeSr`gG_6`?Q!CSr
zYIFND9!~5>S3vEoDK+`00$Kc=O+e`TqaBc|q-q6|L`M#uErWHs?=Y|#VU=`A%cvgV
zF9YTED+sC+$lGwX6D~7;##Wr{Xs_65X*J-1LJi*3G!4WR&osOq>>z6rg%Q<*DPfQU
z5tEy<4dI!>euXLPNYbIBr&o}bArIneqp-x0MNp7XyQQXHM;-v0L~Q(T1vYgE0~a_X
zER3uKXZU?^xWbEv*5n49&d4nA^dxBnQq9p{-Bto%2dN*}3ef!EbSD%`?CcmTD}oen
ze2L%Z_hB*s8iZv)$0$jGZZdEt#l_KpAi=1N=&GH7$}!V&2JzU~0l?G(;e*x-at7#~
zFjD~V=osJDE$g}57Wr;Wr{GmKc3C}#@>3%7T;q?Ssz;B<KfQ6JJR0l{Set6IF;ogz
zlrOm=fRMo5pfYXs$qLWKnic?<v7jJug-T1ccy)ZfTN}(YKKAHeyok}^P?R$5+GTBG
zf=mIJvR}M_1+?BmEbz~du^^C*h%SfE1(h3CBY2o|ufUc<Vna)Yu`lEI%=-d74cjnh
zqD;sDZsjd)^ZbH>A6D})NrjkNq@^_l@(jxx#WN!C&cQgHJr`d3e_8ig!YBR_=bM*0
zo?HUWx_=)@71$Ji<7$8_x2+v!`ZEV<9vXak3czS+iLYay#z)1@0N4A<++-pwn>aB$
zKObnwg8lX2g@O#gV+bNa{et*8j8%dr039=Ns@BGW`+{`;#Fjv+4qgX3X=qb`4uPM>
zL>?Yt;mNhm?V=n7ZrJ@@M_>p|1^|uH0#Q_iNu#+Li(Sp#U1r3}1=<5>k`TT@cPQro
zdM+u<>%atrpDD*fVXcXc^#C^nXdIo29Zr0%m}IQ_*RP|LcvenUY~9BYJ~d!EJF)lI
zzNJ%ql0y0&49sn>hCldLLeheumm>h%iB9+GSC{_JEM5;Vf(gevKZ75lcu-V<+BLvB
zLT?Tc0vZn#C>YEFXoioM7ju}L_oRv#cn<gUAbAIKCt$>ovj16$%Zf*B6*(Pf&s-+i
z!%u%4EAQm^-#b5|{IBABWJGFT?%Y6f-K6u*A1#{`TkqDN@U>#rAYPOr9-T1ZW<x9x
zfw@&yA}2kspkQa(2B^J&Qc!U6@+RXvY&o$J+CC)R0#`v)H?j{Z-o0D5D(r@B3IB<f
z3_~z>s|-_VP|djGNcwnruLAR0P<ks;z<Y49$X?;X1-3#Qbr8vPB7h$_0(xzo3_Z@K
ziZTGci1EBlBjr3Em6>_K>J?VT++;7>pVhfuYKm@Y<t=%2=vncD*<O$mNbTE4oQZh)
zalCamV~or@@MGoWrdVYe6E<LWCd)p4ME1@)*?jD4&-3%o>gs|GL@Yfj!hO(Mq@{JB
z1@e{0Geh4e4~XF_Qt+^r;fO77nl&JIFW8=ucB;E^`R3vJD_yS<(Nlye#@G?yCW4lf
zV0N6HWu3<_UcTIeqEPR^){VCz)`lo}cf&c1tV2i$9xrKIP3Xxt-{~0{xd1ZSpyDd{
z8sIO7J)zlp$5!6Ofmk;jROsEJ$?@{|-IFKtoc19T%-h>eJC=^95(kWv|15nRcz2M{
zCFx8D?h_O|DKGydG7<`WFi+dq-V)`GdEKw6sop2+FBuupp&6y2Q9pkk(|5|@4&d}6
zI_HU<uvmYJakwM52Z$2)v18H32RxrwAR%RGt!comGE<m%0ubkK^bP#173*ETO49j)
zKL79r=<hegg`*wLNY#@Yl^5WC79UTGEDt0fqG5@MSU}<`&IX(~x;i?*KB#cM;z<L`
zf;fD5e=YL-z-%okp_Ck-m{5Dxh)E$%=q+q}$bU-2SjP}Oj_w>4@gmkaung=JuhP>?
z9>vY{IE^LTXRbFb8U6>nCS(MdEyY5m2=f}&$QE?92&O>r0LEf~4!MbWx-dB)^ZD2G
zblE?#P$;8iYd}wG7Vo$Yw{bhWlmpfcP9N-<AfR9k1u_OR3PI<;HJDW$U7q~)>xqad
zVJ*jk!&KOiDlq8z1qEOi|H-2%Av)e;yR3w>J!ganX~>^J_77ThFx>o2AMlRwy%VdA
z)bQT9@H>bAfjZIPb|72#A3WuUg}3-%S>(xnAALIsP>GfdPT;FT8FVdZ9s#HM`1kZQ
zs{twG<GY(E`<(P9&v$PM{xAml5{rHSsOPd_JAm&A=aGuaFu;tj-z%Ua#uH3TAb;&5
z>`gvG1Pi?6=ni@meSm4#{wm4Bu^<hOz*32+EIXh@>7eb}EUc`mW$J_Vm6RxU<efX*
z;`tmXFKT<M$X7tc+Hc&8{ni@;ddTkUHp=E$luicAm+PBh@;$$a$C{{Obqo!I45w+W
z+nXS`==h34Y&RQQwTXzV`;rA_6}xP~z=<Flgqeh3C_vXKz7WJcq67S8*a|27Qek~r
zSyD}+olq~+@EiqaRN_?>u2*F0A+vhy8Jm6v$c+rf`r-=7$;qKg0M!mW2SS)jX^u>u
zyX<j#?j}ehA)%8L2lR_y)I1w%1KWtWLVp@pbYtmF+q>`l`jZKPn2M;`FW)YYQONWN
ziwT+x{QPn_<AmncJYG^nWH>Nu9ce{S&%<K@n1|90*hM7@VUg=$a*Hu8*t(_d^^WQe
z@5r9j1P{s!_81bgr()x}_&~LP4g3k8(|?@M)s>WzVy*pLTU|{F>z}akT}&*2Wd;Ak
z-7T=WPMGHH29l=>R@5F8x4Ezdtv-fM2eN$7ef?RXu<WtW&~yVD0G9)57}6tzxuA>t
z6Yv$PU$nJE0Z4&ktfwIZ>V+RGa^HgHhUn|7Pkn0AhB~awq@cZDR+A&khYbQGv@XYw
zubXw9K{eg`N}2ZL4YUrWIF@>uTzFM|a^Cmv?3y{#QTB84So!U|PrawZGlDG*=vRE@
zxhk~R<GZ1v-t^!L8y(~*P@5}jYd?a27?MO9GT3t%xMAEvrUCli<9nl-?n!ya+_-BH
zi7y5dCy&D>(ozaOv8^QwcC#h%j}_2d;AVZw306$Z)PE%*BNKDe7-b%Q!1<zEc5^#A
zYgxh(DmT;f=uw25+Y0wgkLzLV0oWX|Eg^9#NHGyi>nj}Ts0}eIfK$~%(x{9=iE;~R
z5MsD41p>a^cKAD@{P3uT-jGh0D$dHvQfIGgZax4euQ&QqK4PdG^+34eDTh4g#N;j_
z%AsM1($Ni?98ers<m0{?>~F2)p=GPZss_?e5s!x01XHhoZUE9j9f<7$mG4*=?X4X&
z;+{;yg#hpp;rg?-K8nHy1^U(}mkGY+4eHY$Oi&z*wb5dbuXmy|YI!t@*hkmjQinJQ
z=jp|ZOQ5?U`06b?^uHAHbJ2M`VmZs;=WsD<bo`sVjH74e&&S@BkrFm$6cTE54!6)R
z9F~%Lg8@L;N%0b<-;)6;^nGHiG^HdB9>t;V0sxdeh#lT6IdNbDXC^AqubrThCn*g=
zI<Eui1co5A)=8N$e!CQ}nL2k|R`swDFRwU;KwfluQ<F#}f*4oor}T1rp*@7e;rI2V
zo5Ub%l-B@v5LEc_O-YTN`?#y*+!2`n+_1e)0zM;tP%{8gpPQdYn+%uC0lnvOHF+B}
z9EP?VB4q(p4LUjYUAq7|UO#3FkF<f$j8|m7ld2xT!4pK~a1<k&T3eloaCEbp^MzOS
zai85IumJ{Mp!1c74`WvjJ;g@+Z;+iFJ;D5UI31z2u<QLPY2D3ILhqZIXWnXND(>Zd
ze7Y+UP9CV_Ii#IF16n+B;ts5gDJiET#D>s%fIKY3oL@pD@#}-7g$2SIDzf1b-Q?Cf
zQMwwqH#~0<l5La_WDBt2AOp#c8=jSdq_hz2w77T?hae-lvZ;x|Cg;Tqd5XUv(6vIs
zrg*QSVhs3Y4Xg1b1gfChM7-zMsYE*o4yXU<PtmRbPC6Om7I4-*C{%~$URcVyCA21R
z<bY<pPs2q+fr>*PZa#387`INCh0(V6UN{Z2BT9U0gxdkb2U<cAk1+=jU%gg>;>T6G
z<!E6bV_bqdKW4&VV}nl9xF!<ydT1ykc@SM`!fal}!!<Tc0|7q?FCOW#5kUn_O$&ME
ztC7v$Zpd&2zXWQGS{vPC9z>epX@rIAJ3BA1wj2hXf5y%(2QeeN&0~P+l=;$cax+G0
zhdQa<cn<DQK-0ToFq1&JE#|PaG&PIJMotIp8W(@x=S3=ID>75-^f1z=YY@s|-OS$i
zL!)E0*f-I<(gsb8jb&nv6U{aV7b&WVI6`f0FPaDd7+noeu`a#k!cL-+oGU2<iE7M9
z#g4tKC}F$#b%JQ#tI5%HCT`qgknqL)LpdEvM@Zt}?+MHnaDqb==qR!yVv356e94i2
z)@W>GghCXpE41?M{UK}s3-$Gs`+tSyJO>g1`$BffT=+te-=h%jLMvVAnV!?9S?}CV
zxhPIhHcKau;vNH>o#Dp#F-Uz8MPFI@_V^rNI6GI}br_8WHVke^g84jkY6QnWY+)`k
z(SW{?i~IZ-Pp}epYaF0pZvY;gdubbw(;!&sb5)h^{rly$n+mRe`@}lk(%5*-&JKw_
zWXMKw)<Hl42pjtn(#7vUNCt2jnhv;cS1Y=1`iMIgc#u5k$etTb`29X<%crCZXfkZd
z?65Q)O4pIkq|aB?zDTj-#Sgj6og)S3<9B5EH5G>z_c<=!Hm#)F7<^`Z_`qyM>U^@?
z`yCtK#Ez773C3I-h+5qEV&KbE)IiqEB#v_!U4VRm5xBdZ4bi8>7@o=bC4{6^M%+0a
zDRF;?nMgROQ$W0^<z`S*$$Gr;_Evyl17V)niNa4m1yNSo28;AX3=`&O2xJyEg4YoG
zY|I>WN6ahEjlsSVa9blUBlhrXwyu}q=O4rc1Y2_$rFZhwX#h{t;Q!D`c_3^7h56qR
z6;JTrzk94=XTf};>kPk0bySrREkNU%O7wRsXkbDFG(y-iK)pO&J<+*j!NX&F`%eBr
z1ML^@k8*mob@;7czs-kwZ%sGBQKn=<^wnN>RRP{U6pGWH*TmUn-M^w7Mifk63>kxN
zGaZe?;1*oM3}s+@jN7~czM&L)X|0PslT0RmIxxVpYu6H>CmhKj4B#CC_;h002@_AT
z+A%b)lId6eVZtyRSz~~$4u4;*{|JtA^!I&C72hDNLQjfr1&j65fqXV*=9fq;o<D#S
zhG2a#6P%yU7mhrgt2)G$0D!FXigSC4raq;MYx^CrX9Cw%ygu}R%|)MX54{>LRFWjB
z&&^x5G^WEhn}X`Qqh2VjsJH?$!+pU@2%z!V&c@fNRC~%5k{R5#Gp|}xnBbI)8tcse
z@Il#ofZ8-^taB>tKx4S18d)PLNj`K440R*=AdEajrU5ZFO(iT*YusPx%}X59_>?XH
z;BdE)MUQ#npsA`mG5z@*LL`Y{>9`Cq+u`wv<4?oFMxzSVSjyakXkc-W9<xFs2TwYz
zIvLNO*Hu;$kUON86M%@y&0DQp-LL_tYL(9Y@W-}-)9@w?8pytZnty&cJsWChovgRx
zb@3k{E^^117?3tdA=@2vlA_1H2uIQE@81HtZ|TWc)W9UsGcAJof=uU@+$B7nrl#VE
z@4!74^GUuQA)pw_z9|&6)=r9zghmTe_J_~WKT}gv<AiTF+Du?rty%csk1eli;<z1g
zNfs~fvmct_gl))+950}@ea8Dk59)B_oQXSesh>i_1;#NU<s$D$cxwo1KU7cTN{^MP
zSX_#I)L*Kq=122tDL8>xV#v*w5@Fba5Y-#whe&@+kL^s=+6!0{NElWa_{<kr8NUr(
zuR@Or8~{ulW9!{3pO|@hrEbg)kG3af$VO#d9r{|7Ov$9#(E-l|S8&S4O|M;%`abRX
zL~hSpOGQhZk+1_35fNXFbK(x#o;?2rl3pZ@DpOz%W}ofh`_ZcTVxQRmK?E2UECLK!
zvqIdeHTu~9f%-Dd<3(}cTdY(d8Q>2tUSFps+qthaVEeM5-aoSnBvO83(KT~0Vpthi
z!}Z_48_sJw&B`8|fTdAwp&Whj&_n@>h~?#F@E2UAheiRIvkBNR8c}TQ@Y!MKzJy%K
zZ2r68sUyYxOKmx4QV(p$mYfoU47uW$l4l(Cy<-f9CIQP}CXO}(S`mf9lL?W$b%R$o
z4hn7{)qKQy*~$rs80GaRZnTy|Ou6FrsMi!IFcg%yvBF03MFj_Q8Gkp0YNs)>{B!6M
z$@!a<ce8ZJK3iv@mmaZ&P6yJQ@ur?C&H0`fioP$T3hHJK*b6bvt7vBHK0laA8Ze+h
zQ%j45>LWU~t(>SJfx3Kqei=#)*eD!NgOKn=#d#Mgc$`1+nedOA41Z9yHE<83U<2=;
zB#AU=S~3bH+|UaLV-iwOU`l8=Y*!PN>bNvmm$vxf+v{^~bI;4ZUJNZFxaO3wV%2zO
z42#~e19{)qFJ6oSo`QGAp#x71Cl3_Fc)-RQ**lRR0CHH4Vjcs&P^g@tm<M@F_>C$c
zSBRiM=Z{AYo~LtM!6=q6OXw1`nfqHxSHg)risiF>oF)iJ%AGOBuIqp+iGVj0A?>c1
z+8wV>Y*WRa*$%r5GF)h1Dpt5IJbHEcPJhV_6{9kEhSsl_x1Y+x$0f-H7w`P#MT7bR
zOgbP$O#7AsAq`yrq9=4bsBcY2Y8#>MKqc_bq{R{P5V^)lfCbrzqvYpTLay}2jVKcb
z(zD+Xg1V+AM8Cd-5YPI)E2amc_HJhb{O10%k~UubFs!<W<!x(k?;99E$`3o*I*dT7
z?)&ZX8YUKCUI4uC-eij9W>QeaE8CvinVDCxmq&a_<6`yojjD+Zo%IRPyD+7jN%yiS
zUKVnyl+*a`L(g36e^NDR>+8mc#q$eras7a<00%yPP!mFr3y2%fbF{>j2oS9Me*?F~
zX*P&^UJ_Lp@qF8{ru91Uy3T1Ay$!k`KEkt>I+z3)f_Q658-t=T#Er&<Q3fL$3KFos
zA3wN8EO7+W(b68LC;}#nfK`)Vw2Y{_GYYLz4w&ieG#}9!`|h5eAp7-qpD8^bjZU*6
zmocMj{@t^Kk?S(Z+2c^oQ*{Swe-qC<J7~&G^%3qI6mYtxNsu7jjK)0It-q(Im#~E*
zSer;&Avle`KHqX9pv`ba0H*}Ze|DjE8Vd=vfD_745QN2ABXDd&Jb|xwoB~kr5o9!B
zI?{3g9j6I4d*rztfl<K8sQ}}$@2?EmP*CkwTl{tT%a>@X@^XptzQ^vX$G&l>{$0&t
z)H4$6zn5L!xZ85wcjvTMA6akN*m;r1LiQesloXY9*Jl1;7Ix;zTf=;n^B;s*53SE`
zIkf&e)uoLF-+BUHu)JkYEw)PF{zd2ev+%n+no44}34Q<rp*2NtF^UW@1n<*69#tlh
z)I~K@TZS6H9=PyiqHAA3eH}lWWMHY0_jKb|;>jYVL@%VK1L;DD_AoH&`TKaR&`II=
z`U*!cjw9h~yGe0J>H`A%vo2J)E}vHps4Vv`bT4^%K{C{oAjRqenoq1NwQ4;G<>*2_
z#M$^1Ff^RouY=CQ&CLzHv76g#<m3T7281=@E=3|)T+CQq2_qh!J6rJZEH6y1-=Gi}
zC{T?jffU!CZnbU|OckCV@;$gFOG`^JaT`srHHL-{!61&(5ak0Rf`+=gQF{1D#ef!P
zrCj(xd?m4j;)2WH^UQ<v_M7Fk<Ij<A5uDw%l@o3(WELQ9mMgO6OHB>NkZ1KmRu!jl
z=-RcSbX~R`tVxV4sVS~u1vZ9sbPqiS0^fdPZ<fYO%9W&#P%L4b4Fx9<d}n)mO$M0H
zR*^#mYG@X+JV!!I)zn0&V`60plROSulnn?S0h@?=M`vm*Ac5;k%(0HeSy7c!jOFSJ
z3n2fi&1#=II1KLtg+e0nXJ>6%!tJEs1Tv|+TLLzA^ucema!(86wF@C(Li2_xH7W(X
zQ1e3jf=9Ho<VGQIImQW_$`xDUMbxz%qsfLf{lgA=ltV|TNp<otGU&c>4cm9{ATYn8
z8?#yOomG<02M6EXuzB#He%f<abCFHs<MXo<kn_i3N)}$@7IrT<jCj0`su%Cf*^hZV
z8G(>lcv^z3YvJ%%inqvC;=x3&v&pa8TC-TFc4tT$-&M3sODLrDp{^~x`3g0bueZXn
zpTylHHLJI-hcv#&>j4NOfz%!x=nnV=awQIuVHYfhl&8SR1gPN(s~PcdwlhC=Gfmdj
zA>Eh$oSmIt7caEqzn10`e%bm)>o4>CeMv36B;aQp^?*RWeOo3HUjzk1zF$D3Br7YV
zU!b7D6*BGDQ?GzpNlv_c-@*5$m<ut)gdH#Q=^0UGVY1fbYggs4g9;~uc3yn5U-Px)
zKH}qwMHL@3f=Gyn+{8u3NESOti?9tr(TNhs>j4-+j4Xh0Qpz6Z7xuulo3BkDTbnI%
zx1v}0+BHWq*&%Q@uTf6Ip;X_k-)pn2f5^X28}T5lC#1NR;U1mO$UwyX#|Aiuk{`V5
zZp?26kv;DT2m)m+Li<|?fv5Tz*yXX`elp%4b5!{2^IIC43H+Wv?5oW_cP}nz78lP%
z?;&2a>GX0yGDx(S#l?vA$wpPGcKURZ4GTnb_yG|nd;Pb`&dpayB<<?*a)oxODX-^S
z=NFpx-MwAH5UDR=yxC*Y=u>41X9Hob@r+0-EIxnYgfYQ{KU2-j-msn|rw`JxkGRhJ
z_iHm8@}XH)aXu8jURT%s(jJ{B#Imu_94@|x$3P4-$rZ6Sjq}<=$H&wvsY9<Y6}@Em
z`Vz7S;<KKi+2%s>G_wMUkF5b5#Ba7EUi-O!Zsz>zvBe(|4o97{4snscmb?-OAvA`b
zmTBz_3&=c`im`zT%JC>LDec&s={n9BD|-1*j3z@6jTSkIqJGgiAnBU<%5d{~Ny+!>
z)AC1&cN2RP!T1Zzbb9Z+I)KZR&7zXJ<`V|dHy7`zw}n2)3N2T<P<U{8WwGJNwR^Y2
z2^k!z&f$_BK4SmfcfJ3>3DjOhzuZIo_lOX@mQl{jTfgzCT=&rZv-*c^@8KumA&G|+
zwemh!`<L1$H?`?`hZEb7oPKkLXDbqD;_x}*RQXj#ekiD_KTuThd(XW2M0fv3YKjMX
z=hOBRqG?i!5aI2Q`y-Z^i1IzpphT07+up7xQ(thr{)FkdwbxrFufwl=)T@oLleDgT
zeM46{;H4Ueft;sBx7^>In6!P}rlTa!^F~jfUAuE9AH(t(lKB;l%sv|CkPo>Z{darZ
zz~8-&)I=pJq_$T~`^I;fyXq<Yik2qYQZ@$dQAZg5{Vk@o-`bp*BtP&h)&Y0(my4ef
z|FM#W`KeO_pW_ejI@=FXHHox+JMorMavnfXNS-^X@&Cy+@`YMPOINhNKF2M7_A*t-
zK4#IorIw*6=l)(APnCGYELap1-<D0{xxpLlwQsv=v)_$B+UInLm#yPxe)OZyKa&2I
zT-*=QT)q8(P|5e;rBq24kdbbp-02VRn73c+);oNGHD*xtH!!cRZ({GQqHl>1|9qkF
zQJMOmKR@oXC1d=kK&s+fv%F-@RK5LIN9E%rFkUONhR;ihhi-|cnPK@vSxvygd-i`s
zJ*71rb^F*;VP~BQYSC2e4nNIzB47WJ$uq6x)c7}t@OO8PubNXg1&?OOxGvMMC@X*V
zBq_>!JzTMUl%Frge{Z4IgPH2Cl>UXx?(F!5hpTFw?)NPHRyco{Dc=0g(VhNXy5X_r
zTTJdub;(*6R^41*&*}Bc<*3``r%y4xub=>-RmY9@uFnT|Rc0cx&rdUq!|49v2gaRg
zA|;BKLBY`ZLTJY-O`EcPr@!f?OL7#n+q}Uu-;dtWgbfje_hn^eUnL&P(Ti~s|EY)0
zUO(A^u5p+nGZ_#9=?n{QEN=h6Ky2uYPh<0eXxUPS@r%<w#N%%fPCO}M0>!dlA0PNk
z&B5?7j;IN-iix`>#*pkd$z|TFc>kw~w=mV+w>9@UX!hj1(ALlWyo5jhs|7G3e!q@q
z?LKdeD7<PXc{Kx7+XK$mo+P`V-G2{yQiKTcVFZ$CfeO3~zB!Zjka=<-b_#{`moNYM
z6&W+*y?80jLye7&VF$%`9o18wIVdhFxOMBkm-Y%jN0(q{Vi6?99-*v;562ZLVZbXg
zjIW|XeoB3zM%>Se)lgeoz5-xr#?^2_Z1UnFT#_UUTk&yf-R+vG(`e5CIwL-2T&Q#@
z7bPQ#z-Gfw0uNC=fZQ|@0Zqpy&4j=7e7b{J;Tw#8%@b{#)0h$e)!S$64x5Ua|7H?;
zIz2709mo%%tEs;7bJ7K363lm3K@4+H#Rc%fXoCoF+x4`;CJb0cYlHyP*e6fg+KPzl
zCFV#xbgYUqiyt!eI8ilJ1%qdo1bVOHoQZA*6b#s1248s;uKxG#r5Lwtmj~i4A@SI~
z{_iWK2oe``CUarArzIm;?_6`AwzY&b9C4`J$??CE8^p)hm>|GYqz8>c@+^MG<%C~X
zCi<>9+Y%PhN7@ryOkuaJTK%XwvW#a`MG@IvP{8K}%x~c5&!%Nh;$zl{5w|skgVoN?
zRn%-Sh`ri%lB^L>_9csde+IF01J#P@c>A_Hc9)pgr!m;)-av`~AO+tSg&}Cp?!G>F
z+jxVO7}yU;N+P@~-_UK+8`=e^KP11qT|-~vmoQ8x#IUV8lRPmn07?Q=0CdPo5ZjhG
z1M$gUBu)m>`Sqvc@ks>xUpYrbOXsUa?a@q4U`9*$y|vJ~f#-e3s*NfT{8JQ|sRerM
z;S~n|a&!|%uk{DGnE=Wnlix$nfz(@E*~HY;bK~lu{=xAA_<Xv`PeVZ@vKSd*Xy$Js
zHjSgyX_9zXbt~dFSq24%RiviL?lLtg%U1Sy8Bq*86Xge|38YZJJBcG$cLE{@KwaZa
zsq}Z=Apvd-ARC4TBG$VQ^)jNJ?N35)W{JVRSbir1m@gv)7UVIYU;H_YkyEm(^R)m9
zKoRTzFh!UG-L73kunQ!#g&6n=%$VlAwWp0F@u?!V6SqEpQqAKfhldFR<37LnXR==t
z6tvTIW@6CyVR7qYOTqrZnFRL@N{<KBhwXn-aF)T1ixL*FO70kd?<FMA<C@(7YvH%a
z4gMhTI=FG4jB+SA(Mw%;VNvh={U#GV{S?+F@IE3c{otgCL#FDg?45ThhvzwnPxI-W
z-WHPFc3*jTY}lj@+upg&pfeE>VXF1?XGPhs_3MWXO15m;#G;GAnn!HA-=c8-J->vf
z4hQkXZ}twLO7dJoz_%d5=}UJeQ-oDJIyv!qK|Q2%Gz>DP4Ay6u3j<-{gBmk>^5Dsq
z5jsMlPs+bbi<8?jaq+{Ij-*Mjrp(2%QKh9xT6>#$k`a~j^bFgd(t&}z_EG3?1YN5E
z^)QY_z@+FVt~%0|(~Eu)k~IiQ!2h4Gy2-22=t8_~ia2p`n-k{egw2;-foO)7-HVF8
ze(ToED}byfe}|DCEY#S)6U@-^-|2GzAs4={t`Gt43bhZk7>LnvC?Fp^AP^ctnK`fk
zY(ntccy~9=&Yg-BK)s1sc$vM#X&=`@?7`Z{du!Of6kI{Mu42SfP~avjd+XJW6Fham
zHdsW>w2Q01fA_$0gGmG<oI6lo6TEM!q>Ri8AbY}`e){x&LBaGdMM&>|WTU34`dFya
ziuK<swSG5AF5Y9|gSR(rxqjp75cR;=8CGK!i5)(9VBr8VkTFE|h1)s~h_DzK<Y#ZR
zHZ_f+B!=`EGnD{HAA~j!WH)QHQj2ZihaR!N6_@r_Vu$2i8KN%qkUg<SonZ<~vONz<
zm9zaif!1!)d-t9!m_*}*U5p#}bjQNtBG?r9QwOU+6A4J2N5CuwN0yJ-$R7Liedv4b
zpH?yCkC8~0M5*z%^pMJ_eCplSWh=Jg<vq49FCGl4z?`oKSI}xPaC%|j{IMJ8xC;8Z
zxHOK&cSF;I5|gnI*$Izm5^y$Ny=oBl?>&7Hyn!_5HBJo>pn17xw`^9}itUVd0za7|
zwl?evH2Yj$(5C2SbnWv?NFPzbQ<rUYJr<0{?%k>scJ}rUYwl<3uT_~%tf*;8WQxm%
zMMj5h<Hj43KaY-n;%278?OK5igp;$uX-qp(KOfVWD-9ayTN@i1a#wF2feszxb`N$7
zvio38&bfTc(4{P-kOAYlB_~I54s&VPdj)xS`lr(q#a7+>%iE5TUN}Su3r9JQqzzIJ
zbiXxxrSr2pE;qxntg`Z6?Eq#5Ru*8k=n`TvfdwLF9NjNit!jwU{r#nFyX9bfIo2lw
zT1`LzDkBB3S}PXV4{ky+fxm(H6GSU9Kn5)k9T^Z67VCw{QStj>CubiFOuE3o1qsp>
zL`C8sMwYG<H%3ei8)96a<uooI<Zh!lP60-MLuP>ogwq6tKDuUEc*Nb@mcWRa1S5|e
z<C##ibaqD4h+OFqgF*n&j;=P?+A*l5z~=jIY?Ji#+#Y_&n!<$3$I>=udIcY5zLmVz
zf10d>z(IgbR<85JyC2y2v~Gjksj|<Nb^66kw!bd%Hot0WdI1x|>Yp(_gA#f&BA`x}
zKCodbg@OnHCwR#x5fQoV*1&5Kmkw<lAp$VGITH$$0oi1DNO2U69@xLr3R)Cm7?rI1
z(d8^)Ay}p+;Bo+0X0_2|J)LAE6gU9<A%n7(=!6agS++C3?AES{7M1HzqL~oJ&6ewa
z>a&MCgMt%cY^aJupWR(}-(w3>0;8UYIu0@fZoUv@4Q<TmvQA{3@j-qY5_M4Rdc5Xn
zqVO*lIdo``7sbZg(3F2v-C_x>>Mxe9&O;~ebWok`CHIxhwkOlZ1xmO>braW7;Z*c|
zoN#`*$HC!~S&b#v-}Z&{6X&+)PPzRpJ1BdtJD=KmUnk#tY8GB7>CvA1wCMfXN+w|;
zTa0HJN@9vE_d!+Vd%x1NUDza)q?~omX%b(IB<_ukBO<%AuxhwD;AD^(gU`LFD3@Ee
ze#qR^4r3&KLoO4QqNkZs??ma53xy&;7e_imOWm$F8eU#|kUj4bufS-gxY@j7$JldZ
zA@lnMJB?Nv<|cjajI`iqiL@cZLWwCjtH{IM9b|4{aTqFk>_`%k-`cJd`{P;*;c0d|
zJnUKCwP$~Vxv9#p*e59}CZBrz=uD(ys@e=O>=Wa>t7bJY8DS1s9&$Jc)9XFowZHch
zW|Llr3drcu<KfY+D}&u}c{Z^L!_f`(`}AKPTAl1|XwPd-aV3<B1d5PD=?bAP@U}&i
zc8XPNC<N;!#qM=-(Z1y!Jv~uZI-i_7x6N2DBiEOHzB(*j-{U3rUyKIh?+|3z28A_F
z_-dmbL+E|bCavTCMT(1A>j^*H+JL(DtX_#Iv&pF5@XW}=2Q>NVfU9r-_fEQ~tjA%5
zDO}mPZp5XHvm<J+nM4-)#EJf=k~fPLH6JE0Ym~Y-;C_UQmVHzVEox500&O`3i5ebF
z>L)H+NyA=@Y}M`0Z{6&@`;ghL@amE$si!U;ZMwjz48qi+%@PtDn|&MDJ>FrT9Qw@R
z_;Y`Jw6cKr%+L`%0jXvKu`q_2uV0VqZ{x^P0d)iXO=EZO;9yFG$=V;*_psnsikXv>
zqx4?|EX-(iL8;|p=YBfCb;c)s2X>l1wRJczK6G^%n65p#a=k=Pd*-2bn&a@J_&XcS
z8~GG9H2Q@1>$E@GHy^|?wY;3Flc^mUD*zoJq<$w(yz3=Fpot)e%OaDthw-*@r*28h
zk}T}eOK#sx-4ZElsO;<<?0@QAO(b?*O?vvX{ZEY`qTfbIIRdD#XbhWEsmr7hN*Y#P
zLE>{;Ft}{Sa-`8YP38PybBzx*&i2KbUUZ}9Kj(c1^>p9Ymw=<OhfPgQbwViu#vTqY
zjD>4kOHuf{7Ws~I%f#fF`s<h^J!DpH%|iG#a5aJIG~Y{k?a(zq+-riij(ErW#0e=T
zf~lx`d6E?HxD)>|WfMw_It%9c`KVa;hOW|BP8xo{c&hC{&ckYA&cN#^vR3DLKHW_G
z^?JpBQvbZW6ZP#Ci?=sAbN1-mb9r+hz!nzLJ`>ClDefQQ;->M2S}G{mwPZFbsp&F*
zFxmDouy4~wu?fb0;xQmuZp4eF1e!UBN$&AvWO~M<`18km%58L=?Wz0Mt_~^R6A!EQ
zt(otu_JAGSe#t4HJ~cv306rvjZeriRd^TM+>i%|*a=)Ez958m^oflBN$el>l@p!p|
zeeP6P!r%5$E|mDs%Zn&WZ*2(36;Tjgo%WfUjDL}THXK3)JQoqfC!SH^Z-Nx@OH1b9
zzi(4Y{N1rsGmpc%x+}$wbfTh+W^3R>-7aFIagAePuh~vzQvOlEm$vBTF9~ZJk3Zp`
za@Vf4g(~sK(c1WoCC7vgsW+9D{^wuZiGMlp&(gA_bFbo`zUs9S>$SL*m!U^yZU*sz
zND)LorEdIrcH&59p_PvheU6?6)n9~!)Fedw+lGMK>%WYD>)1jqFhXs9>8;_56zi7#
z@O=Hn`blCo#5HpL3-8M=z5QHsmVo!&XW2dG?LWPh`-_N>UQiIfCg9-la%M!lq2a$;
zZn+I7a7jKr7c%U>ih{QP{zK-T%HX3O9#~=jcF1v?h^rQAeC+!BZ_kfUt=!}lHo9?x
zz=htDLQRuJ(Dy-eh;#d1P*9i>4_Xp##eli-R}(;RoByt0IZ9$r|J10Y)V*<wmHulb
zeie;?q*F{QpF#;N?NW+}WV4c*S}a1%K`xsT5i-QYHE`Bg*l^ls(_SOiu;)%ntfnPd
zENp_P3E7KLc>IbT@f)x<a&1blWG(mL!l2|v%d2MxZSCW*uvRo+&_kY$A-ZeSHlIJL
z;<9g4Jibnnzw(#A(2Ec68O}~Ty0uPd&z{x#MC`7SRx_#;Xd2{_o28*<utign_IYhR
z66XqQBcd_y%VM-%wun2bl-xN89vDs9UZS8M^gjlcri25Qp%2sV6w1q`WS31)J>}gf
zOu%fg0;i$do30BB|D};<VorzH{~0~kw+<$({z;cK<+UXzr?~xT9jYu0)LVgk1HLcx
zE|^iYTI;c5Y;NA#cK!G<|9}9*nub3l`6V3GFHXvJ8-||=KX<B3U|^SYI(YUF^I`S9
zqFOyyx8KKw>C4vR9IHF-tt`%(svYs*LEKRo<Du|y7-=~pzHjY0t&3QoSi;tYg9P@2
z@Q@H=3yUXVVTdK~4HeYA>*Ir*1@&48_(HcJI*Uh3Il8?J{^}s4g`+>XAAOI^8f)e&
z%IJ|s^!IF%D(*`;-r5shG&y?vQC?E@PR$D_{r!RlA-iu0?ce_u(<SA(P{Gw9B_JRm
zfQPOMwKM^+02L-^5Ac&Y*SuWgl9Lfoi^_zKjt=y20{?zc6gxBCkFtiu#27$iww$n=
zI-Q07!?{v}fqHIU;RN^PIRFfA<`Q<-rel6B!I>Et#Qc7~<3K;#+RkMm6){b*gpC$$
zAY>!3VdDKMIMIecEC<yhgF=1ZF0l=mjxh%Lcd8}L3YuoG2Oxsd_%r!D-(9K;O0{Zj
zG#+5xx^)eqU`<Ya=^ExRq=~eH9uAKeZ-=8LVBPIo2R%i+hw<7K)=pJs`ZZ0igQB8k
zZq~JR!)B@nW8P+-RVGlD(2)^I7{}j_I}Wvt7mZYIJp`sQv1<FShGeNUE8m@8DM8VH
z_sbWwYJSD9PIVRvZQWY6IwFJyAa8X1pqfBv|IeS=#a7@FSXElv3}0AieBdGIGg92?
zkb~XjMs|=HoIZ1=`U<)%9xo+^+0@PXy*)k0DUg(`HC;k=?63_q5yr-eKM$k0hV0A9
z-X2<X{2;sOBCsViPoKOoRH=@;6TNM=%L4>W`**Q1G41Wd+@gQ1tqE~Fx@;7|j?k;1
z%>|QDRrL>yNiY)RDuHc7W(ig^9T|t^&<V`MyRgClAYvBB-t_cnOJ*bzqdnJcfP(}H
z9-o$VD81jS=*DNei+Is&4ytG`ZmxTGd~wb$p?heY`r<h5bFqb0;klBI{=C{Z^Zk+7
zEBF@_<6gWV@`b*;O>MtvU~X;>cb^}J{r!Uy7zY2Q0r-}->jMy7U2SbI(Z%66(-rv8
zd0|Y)8>cZ=&Y#HYBOnS$P>GQjLq)~T6L0K#74P28KEwxyE9AxsOq9fRt+S1o7X1Zk
zHq~JfoiPO`B-1)qr0P5|hyckIsN)wAodDQFhbk`iiS4y(TJ{fcSr1|~#L7?~9zXm*
zRtU+xJ<JD=wRssCpTB+6*TJyA_j7AMkYz1{A%R*w?Kg!{jtepuP!k1ze*&@sZ9YQB
zkfZ}hrOr^J`2_Z11UrFj59g2-IPxV#Bk|)=_b7Jy6RR3~Xv)@p@J@ZWM^9r7f+3fi
z`Sa&-b^-`-{NQG}-Monq=d-W;VRT&r!myPS><KaZBn&G0*BCm@X@dPY*Jj%CkuL(+
zD==43r?1Hu93|%45X9ul2#KCP1b6>TxQK$rTA%@`E-u1honb1jdYyg*x<Gw98<^T~
zWdyGp!bRot;h$eWe?He+bt?dQ3hF-XXg+KFUI7>ZURX9;qPs+!f{%fwsqa0(t_9Dd
zp{mq*4mHtZ!;UeirC~_$qS@(B3_XGfXs7=SfC=ce(9ED$f=!6XlYIv{LFjHTzQwap
zssLo)5oc8Z)C)D1^H75+)+c0iRn^r*9S?v+fvX;a>~Vxt8s$LQgh-)<57e@RFw@X5
z*5F8Nl}W?NHYB0z*Wm-JMoQFlbVR(ngou-6Fk|5E(PMVLa?T!27gS)n5N*;05SE>}
zQb}23Zb+1Q`r?9(;p@sTRXK%)9O9Obu;NkoL4^jn2^>Z>xUjNrbLu+TI7VSYiCFO~
zgIE>uq5$6UTZ@QeN9O{q*gLblR@nWdPs|9lpwUH|0}`8{n1NxkPXwc8;rQh9!q}bA
zm0fA!TceBiuOgGYyBi^ly?7r-%*}-LLq%uHTGY=pJZbo0^p%I|0{HRc<&w>z5a1#V
ze&$smcS^vh4TpdEduoj{^|JCie7HkZfL5n${~9c@VgJrkPA9hwD><D!FDomrauZ%<
zP`jU@52~pd1GE?V@hE}=ZA|g0wWAJbZ~4akt*Z-?9AfiWi4F(~5@ITFUv?!U2YtJO
z#_IBf$Ix@s#8vCdj3&)vZz@tiy9G$HVCe>AVr|g#knh3AFSjR-5c$+^Nu6dI0!}#H
z1&pPdNZM#%LlcH*cYY3nl5}mwBWS`lB-QfvZQR#)h>fM-#8*UC<nK;c#D^H?0GgSy
zA-{N#d!n;YP3GnKDjM>>`R~s!%U0djz&tbLjaKb=7#iA)qTE#%45+pA@~C7;74fZ>
z;I7sJw6980-_+zb)qiH}pc4#_(=2^Ka*3r8kjg<on_unq<|h8#!O026_(WpoAT_Cn
zxNkNfka`SxV1xJCEDs40vtxzdC7cx1Yau%BD6~<d*eBT=K|`)BVg|90Fbkhk{a6bW
zcW^xZbiO%R&g8Hh8x`dS9$AVHp<d|PkzK%BT3=D&<F3*7-Vac~{+&oz!#J*4xOaXN
z6RCza{Q>xTVbNr_?4Rq)3$Z2)B{lDEeJxhV(&vtC(J0K#?ZDbyYO}}L2W?lmS>9I8
zT-q6kW?dgc)IC*kpKY&)67=ZMS%I*E^byQPhknNByWjRfPccTVUT{wUyY%Dgz7{Te
zmK?p;Fni;JDa5WFymDGkH#Z)Ks?U98egJ?mEg55|4?7NPyUOKjf>%j*;Juck7Zn@(
z-SeDITw>yh!WDsi`)GaT*Y<@l;SeRBX|)PWT}P7)(Nz?1)%_SpNch{l9%R2+5CG(j
zNd2xa15>85gkP18jPDFjtor1ZF{GJJOvHwTIl~0IkRFS0W>==5igaeR+7>>5AL~=K
zRkmm`C{REw9uyvauBfmwtr6i2?FU^S%P45{-8K!%&^>CSsJNL^ULf?7Q`OVDYD4Sf
zDhnPyOPh|ez<?KX(+<w%KCgLOT1;0~$gbY=OKv#9Z3w(@^dT{`50$A0M7!AAaWd*;
zo>!2U#~e*^B|vCFKAgT#Tg*ZN{0%p#piq&?%G9(Onid=rf47JWJLi!DQ#W)U`@)Va
zcRtP3M^z@8u`(f0Hp8HiJ{P+_B@r(RwNZPvU<y<F=j-w6mA#X!((EbQ`9e~&vkfm^
z+^_STN<hqxKy&~Q!RcXAm>uVY$cQ272FPAiHdu<iZfc?=A}TZVa<~w9^Ytr)!Shfj
zeClntc1m6jheY%2Ys-=hvj-~4dWQ^Y7qk-i{Z{S`t|tmshzX`gt$wE#MimuZM-$9L
z=R3-Bbe{V0?#i*RMS0*p%he+B0o&SOS2%R!$hkkE$II_jR;nI<wwL45l`HzfZyI9r
ztrH){;-dc#@iE)Cuv2QvF%BZ1Gxh_s1}HMUZ7NhXEx5p@*03Oh&_Ec`78y4PL1}dM
zpNGI3rR{!f<m5rN2GohuOSggAV&lwdm;;vw9|(5HtcH@a2Kzb7khTUi1S%z92orVj
z8%lzAKfs7!b|@IxFD{?{bfxh5^S6eIjy#Kd*Ea=@vbccQWxs0H#5ArHdCSi^DemcT
zpHEv7rHf2#DR6Q)=#XiKk+g*or>@;uuy=3}laPQY)i)sEQGUnM)>w=o(EY7gY!?Bh
zA%TSJ3DuwRT@^DAZ9_nqfWWcWArgcC#-|dADrQyyNHy(6A%r&L?tf`BVs}SoS7H&*
z|3L;UW`i`uc7c$@MjmRyOXmx3rs*k%ra>bdx=>B4u5}7x1h{!mZqG~mzHOW)z$J54
z;ugZ;1gxs<%5DpxBeo{F`(i`uSFokR=>uRJY?k%tMLEVNr=P;sgimI=rXhkeQ8mqS
z&&jjRi~$DY1`^yQEyG>1`~S*pR~F7_Y31RKqoo`XB#o(mFWu&U<8RYF_@fQod)&{f
zmeK)&Df3-Ao1TQ}eh1IC>zYmHF2Qox?3QSK*Wip$3#<VcT!4fkowr~nqhp)|OGg!E
zMB~jnsxQsx6Q5Mw>P)dGWMu-JWf-raF+j(sb!BJI9ZTew!Add+X|I;1CiuXio>_>!
zagtxCdIi<xBQUNpF<Jb3U>4btuDZi7Mm3eNPBmD)Z~rf@;7w9e?9c4mkg+nBcTa8Q
zq~dLs=6ucJ&L1w3DRmBGl60gn(&=*G%2yB{xcDvbr7j4Wg<3l2pnrgD2mUK}G#wan
zw@mUbDIu1Zsc>^`(#17m6g*Ok1Eq_v{BMK{lka8hbvi<T|1bKaQ}R(975>xMkIE1q
z6qi3K_K}mL<5`*ME2gH7co|f34Go&e*tWlcN(Ck28H!+`_5XzK_@-O*#4Quo+>hqH
zSssdo3w9Fjp=#w8iYKqNW*Xl-Z~jKp=;X=Yzc@MN(;QrjUK3T>^3td43)A~J<zex0
zm%zx@2%3ix5h@zm$+A1JA$r72yZK;yhq?|=4~{~@QE=?o(EL<iv8uK#J#imA`;&$O
zFM+=0XLa#0I}KS6^0B-gfMkR^6ux)O<_*8R9>9e|ONOq1(B0cjfTx7yUx^}B?`eZo
zd^PrUht}e|a<j3(V6*k=G#M0_K8HePGJ*jVvQ8NfhxNZcQ|l?Zrm$s4lUO<!$|Z{m
zRubv27U1muXYgvG;@HXw4+SbLxGQMM$dOr@>kabT0%ckOY&XKfTdAm$3Y*4_puNT?
zqu_)zyaMRJyqip9&7bM%M?tCo@DX2#RCA({{T2e);v;2Goa*iGCw2{7Ebs)H49(a)
zR4RP>?xV^<r3Yh#wA1JyjtuyctXm;8ld$PJO@VwNOlaFyk^_MA;=!<r-e^okhB52+
z2siSTj>*Wpd6SiVu`wD)EOvWo02J`=q2J6M1F3?8(BaI5p*k`1*{f&DPkf@YvRX+y
zy9Sgzn4PeSFu`<s_^63D3W}m9L#YVm=pgR~oos_MHcKc%!JI4t*hc4r)8L4-bhRPI
z(%ZnC!&wFkK_JrzCVDb4t(}972xpQnlK3rNvqpHwUAHLSbo7NDPIt_mhA}shBNzi&
zGVOCV?DS`3W<Crn8>}pNmermdipPVA8Ed$g@W?xG79*RtU9=71_El9wn8*UE8_xnv
z2iQE3*A;N@9<~9IkgANuQR<!k3KTv6AF}8FBsgAC<@_&J$&ml&X!}%c4}<nZLrdnV
zu>Jd@Wul)CPdm?}{0Di7!E{OC;q55?h~ftc3FLc_F{1JrVn*O3hMggj8?z(b7RL(l
z8L`hHK@4#S2r7=_(8M4_!aguK$Eu%U-Hg%{s~`4;P3zxbwJ3YfPvuwsd$E9Ip9#dc
zPC`=v8n)O^=grFN&j>MfpX>1fzJN{+&oR^@7YowC6~GYF_w=$&X92;}W9x^%?fcEy
z=%p+~rn}DM7Z0}Q?OncYaON0f<LLE}g%OFs4I(|fx0gr(q%=FkjMM-~(Wm1GU(;Q^
zbKfefXa&ZpPg2l!b`HcINlcQx=!o|B8^TVp3h)Ng8z@B(hq4tTW4<@Q!-Exy4s;LZ
zhXP;uUCVliIRGC9TzOT2_Noy|OU$@3S%+O@xIVE0?J5N)^1J8J5DPxlPh11=NfACz
z7pacEFv<1lf;4|@a8R$t86?#RKpQlpqYcTFJ|7_Q9coH##D~Os*Uxg@>je`}4-7hl
z2(zn1I;=Wb(!S^`c0JVC>EmIC@XrvkKM4=dX}3ljDG}>hgptR*M~}Lp`N7=*|AiVF
zSQFwR2>-IK;Z5-y{iipuvtQYqsAZoEfQ=#YjF|XVPOiPrq;!Bh?exz#sd@;HRRD)9
z_INyrxDM^59;+zuhBn*(M&=>|J3G?)e_-$*ibG^j;t3^!W>B_lgZxKQmpos-FgFKo
z0075X>)L~h*YVgtPQGM+{JzK<sAI92`G84gKPXj3!+rQpVJ9%rRiHJ1H6jkyY25ja
zwzlW(*6oTrd=fJ=&&aBO>;4}?JwC1O;ZM|5kM#6XxOivG|5apt!rg<o>M`u)3S1h#
z4ek7OCcx2Lw^QDdQ^$z&wjntpP8Eq+Fg*ClCw%;}$#R2pQjMp|gi`!E?dvB6*oM=0
zfTY{wMYG0phJ2KI1qj}G*|jBISH(#F;KA4VR^`WBQ~t%^Ei?gDi}AJOnSYJjQzPyC
zsO%nElT91%ZQic=S1(7Z+X!OFVi&8jaukPqU9pQs{kADD8s&j6OU>z^_Sf)H(g8M*
zIA@K`%&!&LwC<qf_KV&7sKBZM)fsrV%9)hEDC%RXO-DV454jv2``~lrm6I@oP`|ug
zktxjQH*b!-d4IHS&TR#N3OYty08lZhnVH|(Sk|t0(`LLuD2?w6kH!1+yFP!e()f>-
zy-3s2I^+H4h&*ghSV#nfV`^$Ty%TXZHZ~JDwy^1mU%k4vzY`2&+%4tg^R%>4-Iwl9
zJ9rD9etKzC4{-Dv_TTck#ACG!5?U<J-)u;1gpCwfX(o;|oPrWMe_tj4t@Y?eQiN#D
zpv9fbAEx_jv@}cnFJ9c1Whk^`yTXh?DJD5%<%cP&sKAa8<2SYT4N|y)>Tkb2Go1<v
z#e#8#z|i=?zY65{_b42kvz<|RyZ!3E_0SU-5<~^sf0-;&6FXSwCedjht8GLsWb_m4
z3@Lk9i0|X+dJEsD>yWHlw{@qEqsL*{QV|=Q_wT>I+58BNj49$_?r!_LUQ>klIZ4Co
z5?|{5M|C;>S9Q6GQpLO$`oOdOEg}H(&a&KbCzwQ1oB>uw(;3Hw5C2#`zJ5JkHcG$|
zB3Xkq{T(S{$9|(N>K7Q2;r2Ov$rA|WwsdERPsd|W{wOO^q&c5O7CENlXwyL<Svud(
z7TUK)MyWf4j<D>FU9lB4vq(zf_XDp21+NBS%4yJa3}46TMI$PjhwevUD<&RjU@trS
zcuiI!_YE2axtixOvIAG!;;s(_f^eMZdhJQ7k2?;+Lf||!4G{5Trx6^FK;jaVVZ&N$
zOic<Uu6%R(X|;jTj^vv^HU_IOmMhhKb`jK}_!xB3<E0`}I&y5j%LoZ&qYvu>!h`<{
zy9T6V<sdk)kl3UNu|CO@6`YB>xS2tWszu75H#e!`)inZ|DpMS~UgsXU#xHRpwCI*S
z_8OL>F1B#P{Y%QJ>*6av`Z4L?G|@lWQO<u*hk^Y6()DSFqn^Rf3x50>1C6iV$-qOB
zSALk8?h81+rhP@h%ZM+)sQ##>+&}6*oA|>J#*c@XQ5Lw%k^U;Ap&-JE^h+qBzW(cW
zHa#K3zkJpD4dusfr2YG-lkgg=b@%*xuYRq3<9b#(U~a1R`nq*o=H~xud`WV|xBIqY
zyHxI9&3A_D^izRr=}(DlB51YB$}sK=#c>j9704LygD;4$ni{5sHUg+pI8839vxzs0
zSq6daCiyA;YY?;}7U=mGW_26~B_y1HPz$Xd8<0gEdF<FSB2obbSs_jawTd`G>=adK
z)-J^GmTfK;@yu>-{=}d7wD^JcU|58tkjIxOzD|yDSZ)w94q2)f4Wi<X-`R}YRD)=K
zAbLQ-Af!0Ddjh*Wl1-pMtHkNt*fzhm2IR)#J$+n>KxshOjoA>&nFH}YN)R~+Bug-y
z=*TBrmMza9yXW%)yT1S8%jd%u!AgsT-w$6Xz$2#4oS1s9zfU3a!ZXv$$NT!4rE)fd
z!}q9fmd5BKoPS^wsNllh0F0Y(pF>uN#K+!VdRp2ZoRlMtIB3d0erymLj&KcWO*n`E
zE#N^4T5b_c5x3xq>|4j60M<ZRys9$xte{9kU6r!)kS0uYcQ|dGYfGK8d3d}mTUGPt
zn@O>?hR-jp(=UNa%@(ynUz#)d9^795MS-}mP~DZ=17*w!{V$c54*kXV(C?4f@5-<H
z&%18I6{b16^A})k!t{4WM2y(WdOpmUeO7PyrvLMYrtR4I$(|2%f`X7%QDO}w$;u%j
z0?tj|V9edcuE@{H0|x<G?09;5UwPmi2`P7FG}tFyj$+bE+qtzn*#yl;iaEeiuLm$T
z!0yCL2=jBh?>?nvdIo35N?T3_lMq9>(&(t)`}f<pxynjhQ}-lGJyhaB@wC@#aA+v=
zT;9`9a|E_bR3=TSfbB!B!1W!!B^+J5;p732<r$`uS%A9JJLA=rY~Yr9;)VpA<eZ#I
zSahof;Jm%il5eGpNF`u)fXzXB?-iUYPWo@QfPDg8!N`c9w64PcFDa3B&WnH{5o`wZ
z)awD}_#o9=4+EW_JRv-~A3pSADI9R#rNDFw2UmN>xjb~Dmn|)ayStyG$^f&0mu|E{
zEN_OY%G$*d_crg2j{#vKj@Abpo`{G&_E*362Mt~;QLsokSlm%fGJkWUqP<8fbMeDj
zRAncTiQx4B%e%35Rj?H$9pE1c3l*Gp@LUVY+*IIV*oLfT{D3}~g}-7utC*6^VF`&A
zBoVB1df^`62dpLf146*E#lQTAHziZdk!y<X#{Ub^Y7LRrF%Jm5as`Z_Gr?=jPWX6x
zd;9nxAMNaQFI-?ii8HykWzoQkiA@8}LQj-pwqxxJ=HFSrdzdAp3i#JI6km8J2F71V
z$pcf%?FB!Hg!;PL%Ie;f>mG-Bp+iwC<=17FOaOR~F57ARYkNh-GhkxiL6PNxj+iUh
z0!a_J>E<~r^R0YBLbZ*JqO!8FG$I&Ujh%pnmGuoASo`<G_iT$L%Ak;GP;%bsCq8sz
zGiC?P1Cy3^94@<b2tg%;h6nvLkVr^`Az{mwQby`6L4eXUp!P-O4O|T-aKd^>#Q)>h
zK~`eRdbUIX)^-OK6_3}8E1iGfv_{a^08VUtFhtHnMfUmp&9wuR)Qn<dIM&Cf0@saj
zAJWsqF(bISPOY-DqeH+jRYp+I3XLQR=4o}JW(Ob*dCgTTtMs4x&@U`bo>bJJ<@8dk
zto4fG<%xM9Le8MD<1q;?a5CrM05N%{KsL%AK-siEK9}TGJ`rfkh$G9UbeZo$0YZM`
zQF2SePY4w-z$F4xx1k1iSX*l@5)K=lgR}E9$v~(*5x@uHB*5Q)ap+m>#~==&#Trbk
z`XBmTQRtYbs)b64Ff*`SY3JSo@uekWS!3M{9$$ot<-qwD78tIK&I7>k{lGvcJ3EbO
zC%=Tj`Wq06qWeI8Pkkr48BsjPXdjxpr~aPKNAkl(sVvL7i2Rs8KO2wE`wedChFI^>
zuAgWo(en_F_n@GGDfH{Iz<kl@flBSdGu<~4tb$NS$<fvzHU@d!38Aey+$JoX)D%^`
zUMDN9$UA5owD#zIFiQOg#QyxRf3`0B3D(jV`E&|m<eo*`qbH~Pab#4B>k}>e@(&b!
z)Bd1E4Bh9wX+(@aQk|Po6%`lXzIE&6L(E*k_YQUtHfVLOV6-fk5l#0S(2!SkpOmyT
zJsHO3^w|HWw>J;Vaee!~6Olr(GG#0xjfPAqtSHG)Dn+SGp+Yj1LZN6?W{F5hG)SZ*
zjmxmi(u53|3WZ98G{2uCzkAy5=Y6*OeYR(NuYZ<RSJ!o($9Ww4@ZG=rcOSm3SN(`O
z(kA4a2a8S+2BR9@_lh`smP(tn62+{9vM_lVdMp}<Sxorvy13?PpFIuqs=|rvtOT7R
z>Ux_!Q*$tp`Nd}R^zx#Ulr(^-1nl7()dSXf^*CGQgVd&?U;mM#6Bh1$v_h`#-~lDy
zbKMQ3`_8#%+U1`s8QVT@@nW>2g+kJ@yWQ$no+$`it?NmU_i=?X!1AENKB+;p#iwWr
zt$keT0irG|J>yKutDC+WNizdPDc$e!3it>_1ijhW(S4w=-3|RT5fc|6_sIR*3pcPt
z-vg~Jb(xYorsjCLzP|pawA1rTZy<a={LW;dMxt?-&^E%@)V+&k^mU}&-RpRIEfwCa
zUT6!$_$_wNxmCfQ0;=(vSC~Nesc-p$wTl<8`}Sg#XURGPf2bU2rVO7GTEmx9$ifME
z=37?(Sf=Z9JY=}{$-TbC-ER&p9;$ST?1m<gc>A~Vqc;>Lm|Sp<?Otf(QnT1AMT`{@
z1+sDWFbTwfw(CdKn)@8@`Aq8Eo)4gQ>Qi=VW;3{%JU`%Q#06C~00*!cC8hUJ;%h3*
z_X?@b-0yJHKs*!-u~2HB3A^m)Wg^-X-0d0%8rk%?^KINS3Clskz4BgRTNtYWX(P?!
zN|TnSefpvIOd9vkWMGc`2Ru$}krR8~?#fa>B<1jv9v;8C%$Ol~*}=6ZO_hpQ8b~Wu
zcj-D~^;>(@@wO*UP1kqMSM`j!rn~Q!?SchQwE-Zl++L%&xe^_wl5zjsxifZEySJn&
zWxhFME5TKcXrI#ljS2rzg<pVV(Evr|_U-8n6Mx!_0dm`z`fbRZnKS3vXYn+z<fh}<
zqh|e%@#o)E@LVO2+=5ol!^da6d0KETl+L}b&aN6bW&3Ptq|Wfk&Uey<O2*yr@bbcC
zUp|<Nqn0vVI=!JO(dBuEo=@hx#a#0|xGjEOvtOQb&1nx!aECHq*aVi<KAfrT-RY|Q
z_H8THibCG3p@*P>&~B(<ccX>^IPBS2#~VHGeJW6|-#NA3y;H*H-@&q?boqocXRfED
z<QQ7RRzRr-<bFGjv?3>B&)M<0AL-g~|FmvG&pPMeBQKm>-VPb6qB$q0yXugL7g~1r
zB4)q<7~80mV6(3(`sJKOV5U?-8KDKMM(BxLvYS-$14jq_Jt}7e`DodJ(*~s6yzeUn
z6_U!>KfJNsM4#OaP|f)*zQFucl~-OE-Oyo7qA-fTykOtz<LT>7x86f5n^Ak&w7YJk
zj$}0F@b8SUdr;Q2X>J=gp1Uw}#)Gbh^5itt`)(f|F7qqEBCPM$$-{@<zE_lCv0|9g
z(CCCZ^F><TrMYRtO?%YT-oFt2GC08Be{lGD=jWxm{n`iKIh(pKaQfNPr%$VWe136Z
z#PT)c0?w+fJ=a_DqI*q56|N!p)Dd~*U=vCN)0NM(dX(jlFpxg+crr&B*s7RT2!D->
z4Qh6qcgR<vV%^~P?cw==FE1x6DgCJyZ1tLP;tqLjY|J)hpN;~w;&?-C^>>UpsYCP0
ze17{B6BFiEL(`bnsngtbBO|UKJ2l?iZQRxCrr}L9_Wx41rmHmRSWKM#dA!`xr3TUy
zaRjKHL`4l8<~!ZcFidZY`Ta9$-3G7=a2L;Johqp*D+rW71RII+m*s2QMT=I$>u3{i
zG<N-W7X8+LehWQ3)7imc`s>bz_5E0Hhz-(8;oE-p4S6`frU-7w$!-}mOr_Y-hF);e
zqAfxZ{WfRX<qY|b7VzwuGnK-3zZLPzDwTDL`cICZIyJZT;I@*0_gOn`%nBKO^=hf+
zeP4Nl8~2JTOG{n;j+&QidiZ|n*!!M!<zW^*3j@l_?qp>R*3mg)|7jc4F&IToJx`5L
zr9Ha38aPi^>D-63ANZTRJ7sVu9&8J|!sT{UG!=cdxYb$GKWb|=lcP>8(D+$f`}pC*
zA-)LF<n=3c*e$doOs8xMHB+srtGgikOL6Zl<YTz#MH#zM)2sgR<IGCci#ACn_r227
z(!SW&(CAn7M=8Z^FcG!wZH=H3W|@^;^mI=Dw<^ORAEbHaXS_R3shdMXm~v<Q+Qh;H
zzgnlYW9CdZ)*vJIJhSfL)iK(GeYy;?FCScGk)*C4W~mlABTjuy=kvWzhK0%XpHSIc
z_ZHp`AtI5kH&HShV=V_5cC+vgBJJ9r(Ad~$yKrI7)pAGbN~pbZlF)=iN#;{_xa*Q7
z2dNr#?~bCZBNJ@BzrESAb@s~E)*miMXOBB*xb!cLqc#RPhhF?;V`FO=9Pnl7<G8r%
zQaLsO&*CC)KdV|3C_g<ms#ihk(L*-2iRQM3-wc!w8ZJMfQr-CRbzRf?mAyv|Kb&^T
zyJpm-4UZ>=A0M%yXoKBnUBkHw-Uc#$OUX`&Wx;&<p`JH&s!x1;Jh2<C`M~WOZ?+Ad
zFyTdF;&#$iaMDDZ?+28oF({4v?r>-*GhZSpLYGlI15sv!Q48f<{iv%OA7|qriBtzs
zE$S^-z|Enpkg_}{WUD~8%on>ijW~Xm$6A0?W-yP2iUM-UEQ9gLI2Eo87h_6!GvOt6
z-L|xoRB8EcHt$rwAz2?sm`7W}Ci4#pf>ZY#!)_X;t74KWVK4pu^XFt@GSPl9pidv0
zRbM+QmCGvmu5!3A_n*6idfP9tKja;AP3Q2GOtgEDS8)2WiTmH*d{b?&kB<aCx~|BJ
z>)T)x_mhvskvw_(H?Y`Y)5%*(BSU0@7SHOXZNC+T6JoQIBxZ5RMGrEa2oiJ4+<H<d
z*~~eTMs05Y{<l|_&1gw!kn#JlcYM)3@A=+jaLnB#EAHN}-z3z6n)E54U_c|UG?3!3
zL1ukbJhGYtgujOL#FUKLEkAd(??kg)LjDCT@&s*dSK@KZFDn1zlK>IEa=g<N-l2js
z>y5yQ%z!}e=Be0~)9i7TcD9I|1sCQHAnWPTp#JgWjOG@v4Q%Sx9v@z`uGLq$-3c$7
zb~0=aKX)s&(`&I@^AN!kcSujjK4)cR4Vx31e{o?R%FG}HWL?ON=<=9mX{QhMjk35P
z#_rNh=l#k1Qfg`=kxJ3Fx;I4I&o+KfPESwQdD2wlT+X>?3^6GP^73d)m_Z!pN1c#X
zC6(6GTXaJ!YkAIDLw>vEy<cxVJv|-8y(o*I-wL2bpFrh@KyBLhCs8+t+T`JJq~s~C
zdIGG|b?er_6Qgg6ZbJmKE|atn#e_IYI7p9T_*D4zhG_5B7Q$!>`Glqmp@N~mF;LHA
zu&<e_0*I-`T`v)c9|`F0p=ZaVV;UnJV*>g2lcyY{A)h__$cPcw-oL-rZ=7`8=7aL>
z49-dCSoqH~*x1Wp#O;|%$_fGfBj!nua(Qp?_3YZwE32+sBx#14N0W>67759rRKW9V
z>yWa(>U54-(n{zAQ;EqPS}7^p&RHi^Ii%x}UK3Q!;hY87&Uu=X<3>h;yq8+aikY0Z
z28A0Y1_|iv-3^;I4|a-q`25#_!HKI2M~@zzF!dw(ou51ju+-590%cZ{X*bZ6u_tA3
zGg10D5h)uJXzN{6meua_#0?5^gG{{p54|m;biVVMo*&eeGDg>qQH01x-_>LgD08{y
z1r|w4zEQTxml6~70}z3qvaRa=a07qlnJe_E9Iv=HSFI*DNpQ-u^0E{>^){UkEOIG`
zDPX2gYvIqgeyMSF-$q5c-yf=^B|U1M|0fy0belCU%krY;MX<R+5Uy*K+Ewu-I}YhL
ze!GgG3Gt};`@rvngkGgu5-M+%5gd{5=n<3^aq}BJzW09iw)Z%5ukc=IxOe#|_qA7&
zwR@laOa8rf#<7Xs8E5829FnuUx%%_n(3q}Avhwn-vOR^hWs<i=y8RTyh2dhqq4yA#
z8MF%iFwh~#{g$aE;*b4-5^{YP4A6H*a;{<j6q7Fz@MEfX$<?S7jT$)eDyEsk_KRSd
z#Ez{QSmj!UtaDJ1Wo-Qn(T!s&`#E%^daa-A=IYNb-3Dr$*`ML~@Z!V`iTYllacY)Z
z5Hb6j8yP5X5NWP|e>)n;sI_Mjcafj|#8P4-k)k7O{PH~h?brCsCsjfz&?82=a^xS5
zThrd;Mj~^6bFu0rQ~TzIg|v4D^&fFc!?B%IjY!B_TY4s_N17w!E9{f?y(0eBCjpxn
zZE^1W%WW+wk3$^P)^7;>x@_a-C!rHYeIC?0a`f)byi3ARNm$P1D_>n)xGZ^nq_Tec
z^~pxIuTK8>cy-vE(Ym^!SI3FBUDoK}p*O%Y{qa%F&4Inu|L76$J;Lg92PsECzmogz
z6+g(#))pp|kk1lT19pF>>P6avvRO$?Te%}(n$L_oR|EE4Ykk+gHgc9qO-_B|l+Y<}
z-+f$mEk<Uf;$9l)I%C0#Q-`&Rwr+5Rgyr`UW>T|WA}+vesHC3SA*3`K_fJmH-ICxt
zS$9h2#gq-32IwA(i%|xJ@op%+;P*<*WqLBHf__J<{Lqzcf2nk2eQzH%gH81bw_aSC
z+uRs(ps4HPP{)h5AsO%!OPeu0><0C<F820CbmYYsI6q$4Mgln!B(Z;ZWYGTjQJZ3}
zW@gr2vb$4zKh~jjo`;9?#*O8t|NL|@ZUUuCyZiM`<e^{}D`MMj*5r<Y=aKM|qCsFZ
z8zO&BWbi6Jv~PgeXg&#pvm*2q3Jb?MY<H?FTp6%Oj^~n?^tk2jl!R#pL1SG)uU<=_
z#u9a4vbg-k<5>CgX?kDvy}o^ZY*Cb9w|Y?IX!kEqAIVKV@S|m&N-Wq`^fu?`<*(m^
zGXLfRTw1fc4OcrIbiL!i{eP?JzKotBx97^TS;iH*-mQtZ_KdNLi%BfClg>df(Y%qW
z&f_k@;@hV7?f!1wFe^l2Ybz=yY+TjZ-TgUak)L^{rLU5|mZYY7^_A+-A)nf^Z#Q`I
zu7wGag=KBGWzHYU%wK(xHQu+YW3*|Q+NJ$xobrM;ea&_Mp4-^KU9G>Gl+=vrMm#Ct
zwGH4GgtZeXL&f*1_H8%tj(Jgv#gL)5Q*4tDmCJoDJ^rI`W%P`FmWLW|1>1pP;o}-<
z<^Akxsj5J5(g-^j=)6R&Eujv$t-M+FrgynqU4eDL$&()%ET1&q(mx8C4v-<7H{Mgs
z@gxfc_bUGWUDN$Hea5$}$GfqLdynsrGLJsj3E^z2)sJRqk2l<Wt)REVs-s%-&t2&K
z{O&53*b^iFoa7k5Q-1CG)#4pcg6*S6-bprbW~vu8fA*F!{w1#&3L+A1xpil+2Q~Af
zEY<4eHIJA_D;O}&ex}@FH;tMgg?cw5o3WPpd!pF#rWqj*vRCO;xmt~=@w~ER*Rk=Y
zx9|3TH6yN9#Eif#)xN2zeUY5wWqCp(JT%n#(-5@LO<c~g#f>lI9k2hg{~=Fy^Uu`<
zezI)`YNhsq#jA2|uAYDY3G_~;rfX`>DK{{ii)k^5WSmr$lFcP_6U%BFkX=rZ{c}rH
zfpe!$os@j7V(CadlG^{s9+SuzybYx?`}mO5UPtygh8;X>eDa{p4fAYcFu8{r$z`Mc
zC=O*mnlk@bH|A!n9NFj5mM?dgZmK%hJY!Js<(p~p2G64{KAoJ{x^PKGkIbVo!<$_#
zj!&6<r2C+=h9Be4l#G77?c=n>jloUJJE~khGRFLf&zM}b2!qMDc@p!Esf~NKHs@>)
zt7E+l`h<>q>{+27X#2aLF=n9T)=lTqyq;_6y&k3(<NLlw&mC=fiw#Em6k6?mOBD<r
zBDuBptnr*A+bcI$e77-NyVh-6ke1fM)~1zDUR*iq^}B`?<tbfRu-nQ7Wvn+B!jEqC
zzW$%=otS+oF*TZZOtn8%drE(E^{42yZ?&%NH$OgU-Qs6yTRNJWj<IS3t5nq`xAweH
znEHJC-;(oJukO4pv3<woPeL76<=$RBUz`ewfxX*hJb;|IrZr~<r|QVl85LhoOkBC|
zlEv_=^V^C-s)P89a}ABIjo!_(s;jeAYDWF~wfxDcxNScS|MeD+9vyzE`8Br5ZcUwU
z=o7E3;N-M9eA|lZf9>@?VPRK?&3)9l+qjo6j{P>+S%-&Y+3Fe2-A5$7S>0MKH#Yz)
zbLD>LZl8YZ11(OGt~?s_uMbW7(<j-|DktaXhEGx2_BgL6Z>Lm8?61+ZIToiT?jy^>
zz$aO+Ebqd1h%CC>(|jkt{@sKY(?>E*Wu9Hi(&Tk#YRli;T&2OWpLzIV-P`PK^^RxE
z{_S!s41JQjJn*XfMxQh|;FsDvwe{J}?|1I>4c*_?%^IV4$llTVeB-T&6CM40*Zu23
z-IT5jK4#@o{doN5LEaa)R-S$_S#i$X#fzUkI=&^z`S_L5;^KT#y1!FbDQ#IdXnPdw
z*%y;@o8NzWGS*|uzx`g1PuVwrV`B5`lrq`g`1b0$?54_T+a4ZW5!=>dI#^qBtIFB5
z4R@E^>=65QyE+Syt?g32$Zk(J-Vtnt$QW%ea5!L}l>i-8RJh(!mgcLc7VtQx2d!Nj
zGs!*R)r02uiw}i`=~F472J>~JuEnxAn<NV0f)oF6K_79g7iNrbZ}+P;=-IP{OP2V(
zsxB+=QW$uO->2;Ac(CZ;;lpD9REb$Xn^qY3lTbjn(y~c3zI0REsowk!zaFRzANllY
zT1x5ue-y`(*_(S)MtnRzC3VtJA#_FJ8Td+Z;6SEf_ucSJn2#)R`#~nbl!4_9R_RQ1
z%{cB^X8Ek9^!1wo!#vX`D=VM%?Pe6=#!mU|zkCWuM%k9GYOH-2I&sYBq~}#-&#?uT
zc#dXwxNG;f!CqT&stp%A_D8#n(GO(02B&tgG$&AYEM2Cf*i$1!T&$%g+zXAr7{P*j
zdUkPUo9Y>dSa0!6m;je@R+*HO5%;TBXuHUh(v?0c(X?wI8gEiQ2Cer0^b5cGce~_-
zOxEjXINHpr2kM_L3Qy~^MgCno*^lMrEpOc4jwPC_JWt#CMcd?kvhO&5K5Bo>|I+TE
z;NL)|Z-xg4w2-=``n*wuQ_k5tWUVhjJKdec)t>6bg&oXy-o*R>ob_uz@xzgSKsONo
zSlmt;WeQ#kBqV<~aQE_&Xsg8gNDa>V*IO3&4UE#-PCg4OC+T_|Cil&obc!(~8pkKN
zosEr+o#-OoYNkd<GlLR|H^3O^Z@58PMncA(nex8dzbaN>($>aX#8)Ds6B8!N0bu=8
zmw_OFB?#s?br#eEgc4rL=Tlxk3s*ke1{?F*R^lKGKBFN_Ew#koD|Ki$Flsm82g43*
zi_gnVX<7%n2d?L{(v9GAuQIl!9{+t^{khv;tqh8Q;BOSs-Jpq*?)ngM$?sZ+WVQb_
z2s&!EUUiVXGUJ>^jF?&8R@-UnWbvh)0`%LV`ZzzSaRcaCgrP}CKiQSZiWTQmQ&W?Z
z>j(!MYrn26>))%_wI#PYmb!U*-aube{=+W0$;Wh@4js1_7i-qDJB5F$h+~y}QF^C4
zA$nf?bJI{ACxGdCFO!)q1n=Yth1b?FJTo}OMHtZ-ngzMKDV%AhfXOS5968dv5e?T7
zk?xcU@MkpUXelc%r52gHwULsoCTLMv3V}Q-<s@I7mDr!h+uDibd-PaZZY@$kOU<CH
z_VNb)Othxt`wi_D%Hsz9L=@@Z0ris@%|Nr<xYgGodC+rX(MV$9`SaTrA`DuveEs}7
zD;+V3EX|bhJ6oyO)>U3I)rTiZ$rbz}^LiX80P7nLd!L*?dR}aSVwaA94U7>e=Il#*
zc{KvFJCNg%xtU&B#U-77+4d9&Qc!8k4J)3lqxhIsoj@<<tXAd4i+-%Aa_P#cNqN*m
z!HyMH<$tY(#>I{Z$9Sca4GSi7QQE@Ljt<9iR-s#h;LkH!8@mmfoCZDog1zv5X5YKN
z+u?jSJKGf$+;A-sgB7VaxjZ>bD@8t#$`P&4DF0%n#E?7E>EXlgYTOa4_DfDu&!;u}
z$&<C<AwuR!UYVXIZyyJ`U}>D_+Hsfs%?E<FX|-2ztmmiWD{1j)74;8MI_)XiBP``Q
zcu2HHa9z2(&@{IANUlg*$Z+3PMQQb<!uB5TrD)O1dq!pIPTE-Ur3KyF4a{o2OiG=c
z46ZzT)~qpOsWZ=)K9jQw)~d*pB~1_RE)gzSZ@yu?nbc|BF^YSk9ADJ15X0c*Z$5k&
zyUvr87B=jji4Vn=j~ULLoF&$}S95({gpq+iWjd8%!#-6+6gV@TRz>QU^4<5fp3N3o
zg8v((g@KS^AbocOszA5$b15l_xN5-Y<@Yd`)UiESOn@m6tW8*2pfLk~imvG7A{S(A
z%Hwa_YJ=cR(!WHkJEm1uTyIPdE@LN(o__MMY-w;mXamNozv59JrRgwuQE#Oe20x*P
zeibvRI&|vfU0Yynp3JI|irILsw&_A4$_*$3Meq|T4kB<dOTug_<VVbqEB&+O)g8vR
zt|9}{POIFxXU`s^2&UOccj=-B2C!2~)*j4)Iuo1{F#=CcF4~*G9hXKUNp#MtvS)E$
z%gUTvdbgbz6&a_b7_|yOnibn4xkEq}1Ap)`=OmjA0~LGT`ORimQ85%jDZ8dEC<SYn
zT5e-DmtX3JX|)D)xjfi8>E-?VGo=rnIN@?5qvKISQrk(B0D1AKDK39*S!ZM;wa&=7
z)I@*Cwu?KoFs9s-#AONgjd*EDv}@D9kdWoPi63N1a=YRm|D(4!r^u?>&7J#1oVrtb
z^<>oWY^Kz`-hd04A8on*d0W`#Q*ZfGg1#d?>Vb4-%PJuUtG(#@ItZ(@EKo<DFk!+y
zy8$sfxqX`0Zd|`!87mk)J`XM?n4WP!WQxNBz}2^VcR|F}&iHi`NRbl)!R~OX&Vp-5
z*kGwEG3_oqJ``0I$FJ2xgBVaa<*Qh2pR@DJBw9s}9u~xCuOg%y$OSzOuBWDEA0IO#
z(p>OTgprn963zB|!F~W4)T5Wc8Jf%MbT4<H)*;Hu#c^#-NdcD8?VG46`pKg*T)*-m
zF+AcicUS44;A@_dyZCParrn4kgIL}?QRCT6rtJOYsZzCd+j6MH=C?x<GS<MKdRwnm
zAvWjYvOe0RZeBxi*UfgX(MRs=@xnHTdBiZ%{&qR0%|P0zYU_6MK&Pq=+u6Si&D45b
z#GZ6<`Fg=@*_HgV*0f~By++X&^QRoK91lA|GvyXVSulBk!~*9egEN;IfnmBNYMqRq
zL4)?7vnA%L3TnXtrcVuy_v%zKEpbuq#2}-HCDGQhcltW@%<3B*8T(>PuE~IR@Z|mE
z{j5*K&x_bCuS`L{(gJf)%Y5nHYavO7)M!+H)V6gT@+SZEa_aXma3gCLTx#85wnOB2
z=Fy1>35QZ;^tvb6Ja(8CxWdIn-MzDiIa!i!OV2n-wc=?P^D}RiZq%KZ@4lO)NuK^8
z>)oM?>mlQ-ooic!V-}GEcet-x$M`ggqFPV2_NMpVGUyjb&>ml3g|*Z%4*ygA*eD{$
zc$yJ(l8(FnKDo!}KGTG89S-Zm>Fse`vBFtLG0DcRc#+(K4peBJo*X*f+7Ka0`^Kk|
zk{^L_e&%Yyx$%{r>8Ip!3}y~e9zN!1AeolW81q@yWqqS*$0)3`if*6Ah($)|394!0
zEKN?KU${92Gr7n!c3mUYX_;2}O3UM)wsMH$tX!T<i&##d=!l6~<DGKp7}B3K$@a#t
zXOCJxpBCS&rXUlMb8pY86mt9;PyO)Ie&(?S=04;5blK|P%OuiVT->KmCe_MI+$Ic(
zlbkRlX=CK6gOja%;`1%OfZ6)W!#|gL2|3A69*=S8%&-L+^t!^-Ir4M~9%Y4kTjKQ~
z%Dj^Y6fC9NxIfuM#$~j*V%2GLE^o!I(Bn<Fkw)6gGiGV6EscW4?>%1_0J8+1K5>(_
z?*ynIEwY~xg!N$VC>yk*LfgUV@@mk6Da?^ty!f<JMgYJmI2Jql{M;}#*EMaQaL!~~
z*QTr4lL5@Gf>MdTWe5mr!D=Y%arkw^F0o_wZ;F;M^MNoA@_h|4cY5&3HtOQsmG2e{
zOLY2BRDL^q=~0dkj=*3V1i=L4%--G}?m?b&68J4o<7ldmr4+7S1YED%u6}B%TCIYo
z(sZC<rjPspNdXU_%*_Vr3WQDF#jt2v8{Nn!CMIEjmaGy}W`@l@sp{UAVM*MJ=7-jQ
z;6pki(}IHlRxSf?<OYeFh43u(m!2L9RF21wPbMlj+B^b?a?^$lC-l8GOC6mz84eX4
z+_n^Ne#gf?dK!y6WH_gy^umO1hMWgy{t&Wx@JC@BqJ&9&JqgD|EOO6s@Tp5}{oEdr
z=B1ST-8IE}s*ROQJJ}6r_0tJgc9@dy-o0~(SZpW^eEW9t$dQfIJ}|uO_7;vOq!CIb
zfZQRe_Lgb?CD>j6To}dE=0ma`eu8}#oSSNJ=Ge$3HyIa7BWaa8!Wtr$LR!~(WFO{2
za*ah)V^U(;?AN_Omv<!xshs!dp&VI99bjus&J3yCRQa0#@N5FA92ggx6$WjqIj9jS
zzDh!^y=0mm0_HTp*eRh+H-u@>CSN9!zR`I@6ApX;TJXjc)a8HKc!NkIB1hF-Dbl0n
z%e=XB_r0F4cM!r50ZvM+_%O32(v_oWCB=m+7~ngOIPR#!FAyX_=$c1DIU;SWl9D(A
z1iwG<t;|qHNDq=38@vMkG+{A?P7ve+&$0;0*5CKK_v8r|vI(Xd;`pDwe}6;Xz?ci=
zaOJpxt$_JZ<*3BXDXVz@Uci9FclNMdu9O|2g?xYii3x^*z32)dL)rd0%$+ip)$(lG
zIH0_-*{(xEWHWquol?z>6IU-A*(V^jn#hHiZ#~ZpWR9^W-#>ru-M6okX8|oJ<u8+B
zLqe>{IA~42&M%espE!8%3alU>AD?k+UNWi_{@*6<7P@2h+u<4-j4t@jb@ntkci}?s
zE&F{EGP8lv=@WdF>76P6fk#<Zc=ZBxNlJnbYj%r=a=F8z_S(D9kJ2Q{kRC;p#7;JE
zL<dsT;SARjZ8{hm`zMYUEnEaLH>eBZLItHQvV2ke^VY8ptrQ7@qf6Mv&@)Lo1m+mH
z^#P<G4$kq`=e}Yz8B-hRaM01&2-U<doRP6(s4n1|OZVHvL7L_DGqob))s0OqY-W?V
zJW<X9;U3le1Bp;Fqe2`RGFK<Or+B7sXP^V$^RRVMbTy?tY9I;F>+wpGlLYCG&taXE
zn&;U}uY@2Vnh4L$JPfaTJyW}!EH-cciILt;3lAC<7?aO$?~BofTq<qr-pSzT;^55K
zt2qDXr1!xexAihuO+3M{;9&24BFy!<Yv>cQ!)fTxD?445kRCG)?wq_qoRwU^agVaI
zP2;RY(dmS1Mvilm7_H)v*@B?BNzr|F-uN{yhx#Im(SP)EV{SlHA}A9=fSJwYrAYIh
zU&0R^8jS$Rh_G&bMxG{o#<{?VV_9MUvWmd21T7ks7o}u-y=ICs&w?^1*)kK5UQw3g
zk!{p{L&C8QftaKlvs=~0mM6+!+YVqTePTEnGx3OR{PWDl@>KGd`D(3WqKnysM{-kP
zO7(M1%ZZvd#LvKuq|7y=ewdaPpRf$GEhj6hep#?E6}#+a`_ZRLa)&bIGBzrr^>$cB
z=tKl5<2lYEoo93ciWs0Zv4s2qJ=Kdmzs6DOHTE(_BzUZ|C2ESOs&W>{v*Ex%aeNl<
zAXqU>IwG5$e`D2BwgB1JBFY+&qe(bX)>QW$S>TXSS5FfX>%p{6)bg1@DxSl`s!MGf
zTDpTrU&zHKa>Rd&%EM4cc3OzG;{_@NZ2MW?Q5lk!k?{m|VA4Au&mhm!IjaI*-3MGx
zUR|g>zemrWjxoa#r-i%vXo`kwkAVYEg@(phs=<t5*A9+$qqT^GEhi(x*2_g1qvfSb
zyngkgG4<h_IIT>m-pZ%L;4+`Zi4pZ&idN%zkmJT(!$vr6Dc~eg@L}rr5vM66G+VJ3
z_!ObG%DFqfkJf&w?Jq9Bhi*atz#OkW+zRZWc#b4(`#0MaEgo4sXdZ7*g#`gv<15d?
zouF5c*@lVHARo9vc(2so2_w%-h+PJ=VwMzp@NVK~kU1E#WC(<aZ_Ir{5^mHu_!-Q5
zmwc!yJ)68n$cbHdTlRizcj0;$=w8|pvO|HnA%^EKVt$jC77sF``>>Jq@{>ns<9v;l
zAnkcoq!0JKzUk}HBiVshSLG;|I#*^-Zq#Uv?+iVHp|Ju|c-o4U=!X!N0nPr;j;N_%
zFS$;_(F=Jq-%p$pX<kp}9crFjjNa~%feP+YUQ76E!UU<GfAs02st;tayJqVDI`fxR
zT1%qt(TbTBvqf8ua-VIxpud=!kOz2mt`o*(^w<_q1Ow?Ln>X-;n!=!u5OX8M_Y5G-
z++jt^+m-3|t@nqAA1LDXLBu4q+vjck0m*xPMKPwYw*eMc^s4$Lca)N1WbLsV!D)xt
z`Hl7UJ&PKsr75hUT`bcowx>Z_^Uqg)jr<~Mb)mVm8%9%7cz9cnpf|T}Mfko9D5fZK
zbNlSx+g)@*xf&WJX44>9ITr&N_{tl8qs;72<G-h|E2PGyq^8c2BO;!?H`gqUYJ=A7
zd*~sp#=<);ymt9@#Rk0fxj8&BiI{<i3``Ap_91iEcd2X7xLmP<iahN`N~ND%4?c4x
z#k~?1X7_&DDxFZm4sw=U9aTG#^PIcibSLhB_2R`2P3}=?WZzJc3@1mT{s9c5oi^KZ
z$MK>bd9_83*2RXMBFjm5VDcBi`NU9@yA?55$W6F1yVHhH>WPM5n-#lt>Z(Y|o#B<x
zt1rWgfd0<WiNr3cEV^g5fHkMvDq)ZaEqol6vL0p~Noc{1-28Z-lvv<TPert*+&&G5
z=r-d=Rh2F(9i8i(#;kGX^dB5R#IelLF*2`IMz4fJqoRE(KChGntjctwcK@J<7ic4P
ztmxmbUn>7V`Is0`zkbe}hrc#VgGzC3(>syz<CJ*>GD^UCm=olnMRQ@oBUkOZ#zp&P
zw&i>pQWjqv7b_M{CgSZIZvJcMck_xrhG_;IJ#oTzQK!6Pmz@rzNJN@1d?#5&vy1tx
z;NtzwD*ZpW^|bGqIf0O<tSG$^T#TH=bz~phW>0zfk$^)S4FPHz1dQ01nb`^(0Ku$X
zJ9mN=sQW&3P7*Egqm~)58;JWBlns2)JedL4ArXsNEs0(`x_k6ZT>Iwb<HwpKMmS*;
zD2<<&dK8Gv)pk@PB=-^$1?=CSxvg=tTlOI<D;Kt|+W<xfpEgqsGFs*qQIi|@q&gY4
zSStlWA!EeZHo3IWUVJSLX!2)y^q6-~m?c%iUjZVOIQ7P2BI%3&@>aSH$!Xu9DNOq-
z!V(f%Z)y7VgNP(#=QcNul5HaWSUfi<>*3TmtNs1P&y_xtR|;Lr!MT;y?mf^r<Wo9|
zt36YCoSJ*#@UtlG_j5N<iZc&F414y$0cdSHEI#RQ<47~;!kxXXP?`H;&>g+{uK5qR
zDu>Wi@?)-PQkn>)QNg(3oq7%!nedK0!XC*pc@}htuI`QEumukqMfo!e^t`GMBgGU>
zVG?dEy%W&m)2#})_&DwnueIX(wSpbvF>cz5?lY&jO@jXO;ENz60MEhR&STaL2(pw7
zfx!@8V;C<44YGCF_wLaK9BrFi4)w@nLR4SmUZDr2rO7yWx#}n~YkU6%_b)h5k@crE
zq-SPk*3eWN@n0`&-TL(+rb^U$nb}es4?6TYGvvCMCdF|_+{nP6bhhUUG2aeyk+jlL
zEjycc_HN!l+{44SD*G~)S?7&Qx-rFq!<W<`?T;9Fa61lN&^mUsXs6d4N?g8;M$jFG
zwKbmNAIylAUy5T<$!NGR_r|(+6!>#THx9Fj+YFV)+{fGG)^o}-Jq-Hmds^O5+Pa;9
zBzM=%%$P5=4NcyON9sg7q0dCOu4$`Ly=p12c1W*&G70B{3Y_S>A?8J=yrQfO9PSMs
z{iMFvI0~&$ex6;#?wVXJpOf!g%YqeoeB6t~iM9O}m60Q#5==!~(q1wQ;kMBAQ=r5h
zvK|Yz^MjxPR$c$zz3a&6@V1G#MR@4~(_8n>2O{E_Y`qPoxhfXkeg;=y{&41L>FGLX
zvfkjqS<jz8r`-mGimh>U*mtVAz3pw@u2fdM`)^!;M?78#W@ieV8Ht9ppu(U*;anB*
z#x{ULc*e=ULDSi@Uo*`oGt;|e{_F79Y@8Hk9>rOSJqVGO2Ye$XMU?zGlDSoz;NBlG
z$e4a3c`F48fU2j+r~?$!SetELw2}ps_%z)ogkj7+(7^DeWH1xsti0L?DsrDI-j$Zt
z<CD^Aa%V{y9a6!?i|PzjgV+LRTAT5UMtSG?JS6v6hU@u?y|fk`J9)ARq!=pt8lFBT
zIr!yR#mB^nb5Bk<IE8GNA0}g5!STs6MRUN5E4LB_2JqvbevT<Itxkfj%+NGAi>PQ-
zXRokURUqJOX=%xQAq2DF-Ep(_JNA?p#6E|eW_@u1r<=1TyXf6hJOT%($)%J7X5%@y
zd203Fodz-7Z03S{|9%8V6KWv-Oe)ri;Ab1j8szdgD_i(>EII{=EzCD3H!$#L8_`o|
z=6Uiho64)UOvIN=7yR|h1A1P(y}q}1>wyFErl#2m3aG)PlL?BriZF~*v68|Gm<=w8
zM*SDu|IOI9vPxVZUA{sT9Xr;_mN$yH0&Gn}NEU2B7!=;5A1x1#SjHl-vZ6XeYRH-~
z7+~y#2@4dP3GrHJ;sHhv|7UL=7&rF(GG`My)+8G45y1f^CsZa2=YeUin~_d_Nt`?h
z&%}gR^tTu{op@jQ_w!#7p<287!~9!hCy$y;80`Qf(NCVf-hAhzPkK6P0U%@W%s2i$
zQR5(G9`k5Oeo|}wrDd7tbTO|K$Y7|ulB!&i&37tGzW4A4o0{iZ&P8;nuyCTjzTJhq
z`*yMqcKrvU<<v4N&HRd(jl=7IQJ(uBRPg?HZ#iL0CiJ>Rov1gzWz334BWw2u1dN{M
zC%(<c#9u=D0_QgP_1~*>2%etn-4Q>RP#XihVbZk{_DKpup|-S9fFvwebnw`+z}8k)
zNo0~tSn6?u#;Lpc!th9#g>`>dk^{N~7#Nt_@jZ@6)0Y{hA;%xI!48&I{IP2A12&nA
zl2BOq(0~Gz;8Yq&M-LaDa9S4-WZ2-rQ>ERAGQiH1CmHXmCRcLBx_54zX>5Fq*qSUB
zJdqa%1vIRNr<^)9nTXhF1lVqCM&(>enbuA&JBF`VzFfohn#SP4rwMRCskqMpz!HGP
zf$GBuy9wEVWLznsB`1rq5%gWk_7=INgvTP#^r#F(^_XJIjih5h_Uzu>t9-DXL@S)i
zkjsbJNq)Y*gcrwbF0q~XEF3Vjy+-1i?T2I_@s0&bqIcTjArU$>@1~FYr|cu;LM>3L
z!6D7V4NH%88I{CHSZ;R~)|@Dw2jp>RY1?XqvKmHb%)0cSu#UouC=$MJ!_8+6sTKt)
zt60tTa{trB_klHI;Bp}_(&9u1h*3}<$i*vNT+n|^#x7G-C4D=CZOR=gNyqyGsBDpm
zQ<yP^S0+ONsD#@e30)lx^;J3$darK%y-}nq*Av4$<zOO#xeogVqF?fMXxB_3^l<m?
z@L|KwS8nLq@z#t%+sL`suMY?c`i78H{^X^%{V~rVUiyS8#KnxzQrvsKe-iTv-w`?*
z&|%|G27in>Dg$1EIuQ8t5I)OQlBm!KVG_BXMS(MWNKv>4n7n+N@C|t1`1GGI@jVv%
z+SXA5WaR?+4uydk3Q4`v6I-@+uKtX{+zvxsX*#Lei1b;Q6G4{lti=5$fe}>|{N&O)
zD^XV`;j2!DtV_wYOTTd!324s$gL83xarsZ_QBZd+Jjd5kU&1C+27;6D9}5GHZ9ZN5
za<68LK9vkrUsd_t)r}Ax{N&k&eaODw#}OO>{GMN1GPpV6^z*Mk6PAlD7A$x{!VWI5
z##y8=I@zsg%W0*lqmKq|C!@r4!i5>XZ?1#K@tN%q_^QSsjvs4stAc$_kf2unmw7AR
z^;KbE&!U)|Rp*kv5?jy9xQT4W(C-uJuVhFw+j0u(ZEn^dK^W&qQ*K;?<yM{qy>yZE
z-|K6{E;L)q@?E78h!;f;j|coUytU~<zN(#s!_CUnU#>o4nCzuppcV*`RJUqXE?|!*
z0*PSz1!uJ_nm^SO0!JthL9Wk8MtGh<=tn9bg<f<5_Rfv@r5ll}!itxuw{bgT#u&YG
zWMVRSn`)I^M{j0ikI)R++I}KFrV6YiSQI9bUnkW!HZrtF11qko8pVS%kj7J^AWju)
zV_AMFd-(h*!R>f`VeSQWcm1~85k_kG-b5kXp1aZA=@%p48wj?yb|xl|YYWZ?@JWtk
zRx$KzD8Cx3I0t8gLv@1^DTbXbxbh(~#J-0C)78-j5sfuvK$M!k9_+irCPVCwx0Oje
z7IpBbC!@GuP%yvAUEcnN#CE(Z6?Na0V^_a0lkUz$1JXW^*tTF>WIa3P7<PLztL|iW
z1*^p7JX-!$%L%V*X7-ejoA&2nzPNE}O-IMg9KQ!H$O^Y=NwQG%F#lex`o3pdozP^e
zPe`~)q~gfMKJrWPv!ozoD?K9ZB(!}itE>MoV1TzOeCx_*0#t&9F8!Xwcd-Ry7<csA
z_4j3h5Xrq9M8NYfobUTELFAWXeb03Q-Rf?DwQC~W@oPltgiwWq$sJRmlBp_|cB&h6
zot+y~yk^gSM4g<RhM-fXQ-$Pi4)D2|e<<C?$jQD|FBiE@@nKV6c`{nA<9z;tSSw90
zFE6Fr1Zy&j{4)TWDAG$dhV;r3Q{^{K`_+V^?72?ol^r}wBT0qn;G8y1G+;D^)&AGW
z`QI*OfnSFC7l2#h4Xl}<RQUE`@`m}s2d?VQNuFG_YuXPZI=j{5(|Yvj^X2pB3zllk
zWB-G>LveQFy?(ByY8;(u9*rK(LlQBLF96j@WJ*bp@k~6U*2cH3R%*etbJy3kQU$<!
zTE`af^yvABVg->_z^A-Ik1=tppn=YDPq9!{0E>hwO`G6L3bgGdC2=tzA4HB<c%)9F
zy=ep>)=_H+Yr{Cpc1k#mh_Z~qAVLyaXp*B6KgCt(0Ezk1>_}1r5krq(tFuPXAL%?~
z(^uEwU4GD=L~Ugq895a(_15#WU_3g31cSiBP{K`HdPgQ~E+F^9#b>vRQ7c1+MEaGB
z^ZknIg0ZnN&BmPis_gq1HrDDNOw7Zov*JNZ7%r9B=D6g>mvlHKgsU1KuhMVq&B5c;
zkEOZ>T>9+UHG()`3<ejj+fk$CyAD}5CZ_`m!<f%D$t(4frxXCR+!xl(`!HmLzVej%
zGEA8KobnGUyvSma+-CBj6v7njO*24ZVX@+0p=?J&&myC?NnT59j0Yx#$sOjO722-P
z<n;e8g4@_vg=}#6aA9hx`Pt1<aikw@*|H&aS#q<AKYTcAsYY3i=&$&aR7vswOH4O5
z&P(RuS@Y<VVPScBd9oU(+s~R;pKUx(?TN9bUG&OxxtjMM_>ZV5utp7x)m8sp$?_S!
z#A;02CHC^^{>!AeDBX)w%C6u}magtEesf?@yA1OQ5aMZ;6N=2CHUP6Lo}4VCQ@U@g
zq<D+vz}aU)-ioFS@XiuJx5F@{H<ZPS5`1hM?Qv?Pflo*w2ANK6i(;92HYtd;7;IJl
zug`#;eO%M-Mxt-ytCEru{*`~SR(!aG+HfL2M9BmRV|_SXLmNJ~74an6+0O-P&>nlz
zCFv#6R&{<`q!g7%1YjQfaAR+Yy++lE>TPz&a(xGniAb%z%E`G>ECjD8iH<gHGr&G3
zffCxogH6T~TO4Zag0=x%%L{G1hpCYu`U??%?Yyg!zu~6FuKfKs4*>th$jR-ejRAZm
zuJSx-h)60}+OfY2q&^RJOK2leohc|xyLXRsdG_dmc*OvkkbgH~?m_z1-8YF_^fCE!
ziH>5)rN8?J84a;M)OH^l*2mW_vyxiAVnt}ZKgb8jtpS5e`ls8weu{{|dnu}`t5>&X
z?+RGz>Z;wmzV1uT5I7wZY28Pw0|luzTC$}Yh>()6eSG%)`v(OrnVLJ(b)Gz7s^I$P
zJFDIWCCa=D%Fhk<G_nElN~nmt(cu?=JSKZA6tvtaA(X0P3qq@(MyXPCj<ed3>y$#^
zqafp#v#J<k04Q9?T{|EOL>D-Z$f{Pd(F!(U_g*M)-lQA$I_b3De=SabR{X!hEcw5~
zPx=4-Q+#a7B_B<XwuDu1yxN}{D2pqN2mq@&QEsoyavjC`UmDg>#%p}!a@P0VYt*7>
za|w*7re<SlyFq9DD4ZHa%&U}(zVjg{(fM8M)3fux^)2;onP=o@du8bz;96gK@>5RT
ziPt3(1v!*&wpJABLj%E&lH-{Tu!C1aQ(X@OK+JcPF<`2n*8LEqaKP@)se_~a{twtb
z>-wc!5DGa^su<2Q=Y;OWiTCZgnGFb|hdj0b76WG>#id<L1nfT`9%&`y2>Ok69_l`E
z>ZaR$M+{C4&>y1mZ@f=2R8$;?|6qdtw>)yHrhB`Bdw3X_1TSeHf&}=GJG(>cH*`Fq
z#zA6CVeZF<Y<)^jX0#@|+c8gf{NTmg5!1$Ei`0lLi&YZL1Elh-t*s#f#bgKsGf+EY
zLCShZjHIk<I)+}9G;wQ$UI1?xEn(Hj;2mF2Y*#cp(Pj<&P9mmM`O{^y=FUaMv;{Jf
zpFFHl3P`NA0ig#9A9N`dU+NaM`B%pdk1D_^sEu*{+Q?a@)Skw>76t9wcM3?7r&q!o
zKfn(RKYZAt_&g2^<fdO2M1t{z2#_+hfO_~?>@n#@!ncznYF6~<+ZUrUhw^ufADMz(
z=Js^{LwO(eMgvF~lsCbXWEgzPtoAWOGTOBpGi%NqyOi~je=iopA9;GPmojL0B(8u|
z!WVx2^yz`!U^5Eaj))3#Cui=(QEDVUhk(N3yrOg|Kq5X<@aM!ww8Fej=)Y6KwHwU8
zyVU&Pag}Iwh-zq|tj^rM0;&z*IJwjRguQ|9Cy0QKKm>oEFG{cMDt*iW7A01WN|_if
z=;)~83oYq5VIOx+NsdZjv6g()P86O;l4TP_d?mfszLZTY0;6XC{%jZuIBH*cwP0uF
zpc6L2#U@#DoG9O}C0Zb6lp9Z;Jo#Y42+6F&6Sg;9m%quJ#n#pJSZ|`oK70Scl|Qgn
zH2cow+*5|&CH=m%^myjHXiL-sq?OnRLrfP=4Hh8|395~)EtNluGiC~fGaBkI0*g!a
zS^=ne)yXCLddHS%QSXL-gU5sDSaL}!s)4**&36@cf5%TEOSX&sOPMS^cE$2}AX|cl
zcrj;{j#TG3EA_Z;7R50(5~rS$u@F!V*B5kXI{g-0-)|pJYjyB;lZU3EBZk=ZaexFJ
zehG+OY<|d})wQ+3sl86yTn8qRKU=AmY96Sg+AgaLUsIe*MFC`$svAlcj+_VcjRJb+
zDqm48q{yBk+aV?z0Di{w=_uyMOl=ghFTShD2NrI=DcxkTkq8neR}k;uU~^o;r!VeF
zMoA`qoI`|BSEcDpNa-@RU4h(NAD<vvM`<v<Oz12ceGxgzZ@DR}ody7WXHL1);6V6y
zglf`C_FGDV6KYm=XmJ|FR9w3a1kwoFlxTH|(mH25ADUCIRmoSc{$X&rMr(1$t?eZf
zP+`6e%feTlsBTWE&cW@^TL3bcds(d2HOssB(<gy?@~Cvf)&gkw&G-~}Te>N)3Ww{!
z$%bA6_97NY(e1ym#x@M{_`OCw0srWbo;YCwRHC~ZyFR2`NL*P9$||@lj*mJ^oH~U;
z6?Myr6AS<OX9e&m;JJC<aAO~DbE%`MUpNkA%n&08jEh*<=k{K6{(0%+eDBU36-~|W
z5Qd~)M)o4en;0M&)13oZC~Ec6-TU`H=`9sNYo-$Cf-FZ(%^4t~YJMk6A>-GS+1i-N
z7k5^=vH80$Y!*&nNONh2U&CwJX`eG=tZcmA5%goG&`R}-U+(|cU(7=}AioqeN06T;
z&XS605C2qYI?V?=egXw4O(!*99@RN&uGk&HREAnxwki1{8?s{#8O46)m;?qs%+%GF
z*uYrS7U1W)cYn8s3J%8wsJ+RpZYsA-V5Iqam5Yllk-X$yDm}m<@C^gxbKgixSPFre
zDM;s%cVjsozSXj;eU6UU@iNI6S#EucAiD)Wz|GxEK9~Eo)<7BxUhbs9fz$*+Ss=BZ
z2Jd1UZm!Sh+PTuy59!(Wsh|3bDE+1%wip(8NTk|n-BoWhKr+0&LFox#^2!#nWJpb&
zgoayJ7m}=Z`oYJ3@&`ir-<ldQ_D_`48h(C1;hopU)qLdLu6R8uF~Bh;9VUAQdkw#6
zAa~ki3)LKGKSx7DZ$QF#s#N-SdKy)7GKTfNrRmzmAg@z`$bK6j3}uPozE8i_Qg(ew
zOMrQFf7DN2OGy^wV_$vc3FM!s8;Rqj=M%F6${P&|3>&;he#9a`)i__^{N~LY%!#IG
zsL<O4O1L*L@D+;zD@ZT|;B%ckxq4(DOSJ;Pb|#!Ex$l^|Y5B-L!a(^RgJMSEvmmyP
z;SNV{41)C>a~?pDo)&>|^EJo@VJCIn%llJ0e<L%HVzw*AKeFQnY_*W9I&!2KE)zx`
zka~-!=L0)MvpH<ZV7H+3<?s@J6??UX)e@#XpZ}Zh{%@Z8zqfV#pT5QCB%lHffvO5X
zZR-mvy<6+N;rx?X<_==Uh)OWGzC8Nw<HvK(m~E~6yHklrxvgt@3CncgfZ2Qc#|BRl
zy(R@O!RuB&I+jB<m-V=BH%samK9{l!G@xRw4iY{AGvGO&wmG5WWbWzNK*FQpuGBMv
zHK4eG)kaSrkOv6rZ*eD<4&=}xHwBHIwDHSecg>!k*nrk*!6H#m+$(r=Qw^0_AG2+s
zGzQY6P}xb5jW&kcKn)@d){*Iy=AnYUXGu>k*Ow1P3<AbN0&!}454k=s`nB$Od!GV|
zejvJI%(lEiM{?-U+o+g<Yl2X50BC*Zyx*x6m>k4nNROy}346|IU<B0+u->Cq!KQ>F
zZ|eyO>11FFUQ15aUb_c82H``e+;e|nb5S(Lx|9Fe&MPR#2OzeAsZY9c#R<mAl$Ia+
zZ-#j@=q&mdmakU|a?p@NL<e`%UKAUd*$GPI(0N8DiL9zVOe)-3<C$KwHx2tQc>)yL
z3xMh7+c)@kf(Y%jSXf}{!BXC#t_3l{PQ$jqv}kF`)~`D$zZ4fH^|PAt%dBP~K{o&Q
zn^I{sDpAQZHIJKnK};Zi|AfSYhT$+;f#ln|Q5uId|44dnuikVBBB`>z8@dxx<ETYV
zlNQ7{<+sd@);M@0(T4zY`DM^grxmG25q&{C!J`yp?qCDPuYPg5z?m@+Wcp#=WsTJ0
zm~%VKmpmS1G@p3Vy}qi1jIV(v*!Z+6llRS25ab#*%t$&1Tk^rV4@7?gOZGw~h&+R0
z!hnk*UCl8zX1>cxI!pLtn{sDJb!(i4zAY@=f-{0!&YZJWip2W|Od5w7XsQdhj<zIt
z5dkVyLMV<fIa_2zy;+F?Q*e~j%(k-igi|HVA*T*`-`r6mtH-@&S<>+Wn}b83DraEf
z45Y2CpF>FcT+!P4GWGef^X21xKAQe%k`y<(caGHY!%ngNel1mfbUa8UYTlj+A9bds
z9atRHdv@Aw$NNXkl2sJ11x*;R@YBe$vQ0+)R~UtPEm5}ry0_=N%Vmit6qhR8@6i2K
zYoNpC+?vE@xsX3+WrSPrd0SCBF}t*1RgL>3!Yjou+rKA0r)$MBBKz|!Q`6oDQNZ&|
zx7WNL8R-Za;N_I&tnR-!3;xFD=7+LB%bLvWx1;w#wuX!8kH%*E`0-=LSZr#>xn+mM
zz4~r7M(W#UFZ4BgHQ`GLah>*C4XNCo>2TstQ7Sxb32k;$_rCDiS^|*<%=BLn_YCQ~
zGHjK<;Y9|h4=$S!_5{+PzcGrgP}JI|5?R%;^nv%Nh*0{77F3w8n^O(X!7_#SD0^*z
z27|vl?)tOUA1fVb%q+)y_Kq-1;41Ohp4*t4n{#5LZTfe+TkZID-2@aX{a54Wc#F+C
zq?IW>?XtQWXbkqoE&6|B&er8O-#0%mDEN!v;AbY+14{iRxa(ZNt@ObX5?91zJD(^b
zvLVVq#gqaYieT5-X!O#YJ)Pzv4P~4wXSTD9U|chRwPhF=-u)oNnd!#MsDSM#NZovk
zlkYHHi~u_?w{hGnwm_iqy<4BDdJqBcRQ#hy55dD$rpUPC>Q}mns3ZW^m4?S!J#lfB
zk(eqcMruyYPucuE)IX@tsIxfW27AW3;^ev`#ku1Lphmvy(&5U^r?wm*5)LrFx8}1B
z^yJQkEzfLS|1zF2<F1&ARk+&L#wI`FR$RI_w94AOWy5ztH|wO8W-Zl(&7~v8uzAKs
ztJs`X`9%s6J_E(*MxUq%lMkZOzsRRdM(rKN`~IUREKk)n*IM6hkk*dGM4e4%dVI0G
zrZ%B+*q`>>>m9d!M8~aQ<o5aVgPq+{7M1N*`kbvEbNK?MyrvK8v=Y;!rDOA`t*Ize
zI4whu?V0oswF&-;ny#*{-upU`QRN%HYgg4el8Lyt@BLu133jL9AEt&|PmhU-Y1*FI
zq~EkQ$>xB5O@@R-$C#PZ|Fon&eM`TNvsYDjLQ>*sdsK}j<TvF%TT3q2r8hJFx^|yu
zk(cBV|GeMmUqG-8|4^p+>mKG|$hq!X?bayRsXUEd%~WN}(o1s&N$2M0*ZlgiY+Gum
zVMVPcIm_3&Mp3R5f_ie+;f-4}h@l_6{ZG$ko{HPHZRz9w85o5M>zOluPG3Flg!8;4
z@k7P1ZsraS@vcq7rJ4vD46elv_(sFQ`t|EqULAXUw%~be{v(|nboXlb%?U9Du`w~u
z&d&MqJ-dI(#-vUCzR8s5|78@)!4m$uWh1Zg<rwZVFF_6^F8Ry<T3{oQhhZ^+F4Zs@
z877+j!`ZGGKksX<NGPP8wnB-B>DaCKhe!U7bqMErWUNLAZF(`7D#>5a<P1*O^3f}<
zZh^C_oB+)|IE!lkw6&MxSH<^Q;x_r;Pr?}Y@<pj{w3Ok5bdFh16|<J${4f8<D^4IV
z;>_?qi1*?CTPfc<yX%%<$A$idp647AjS7EbJG(bng`Qe_At_&C{><gwnM^OC#zmtc
z#}npr!`4D%hhy?e8Mt@}rT;z-R#uyVj@Q*kEv)I<wJWVf-4E{EDcL;2T`~ARy#8tB
z0gQ^*khu~I_3MkMw1WA*f%$VHG?X>&rLo|&QHtrw2|ox1V$%!3{1V@kG87k@Ga9~c
z*4DuYS-5DC9s&IB-5oFFZ6P}U_HFf82fG2W=a;WsX&d+K=+Rz?zwf?1t%0PP+#S*B
z@X*k*^(inh&g`2X&Yf@8wOe^L>D6~eG}FCVTU)zu;X(j)<jZ)Jx^$fJIB4);hZ)Rm
zqzh)&?AcqZivO}1I%r8tnm^)7Cey*lTTL@JN=TdyA0#2MF?`>?Znc9%gwUl-4N=~d
z9b2h16|DI7Cy~#J(Z_f0q;dflH~td3gpwZ%GRmpWcD<2}4K*A(W=!7ub-uF~Qy2d;
zxoq_{!U@ulhQ){PCsH+(oBV6)^vU(be>6{uIca!%#Z|+Wr3DqTxb}el9VMoA5o5w7
z<~ieD*Cu;G>K%9DG28j^iqtn0iOp?o8>xhreuUIwYhl6UUqLUsT6i)zxMZU)Ui&En
z2$d0zn{;!|2vt@2gG}u@U}|b?tg5N0&wL4d1=e)0(J~w`88Q~0VJa=BZ8wdxMMW0J
z0j}*v7|u{Zqli%|Dvo5wS|KiA72PuqQ@M)@ymzZ&;QkAAib5i@`5Vh9x43f8%S%Wt
z*Yp9?dY$NGLAXeY((6T0QPE`Nk+6xdQ)0r3BOIv4K!7wHy8cPGJ<F&Ni||KVyW-|I
zT0C7xN%Corgg`3U9xuUdJbj538=&W5HeJx3PHRQvT!%H=z5e~4K{tkFdz-Ftpi(!W
zzIUe!c}(4~riyOp)K225O517OLfvw0@~iKgxm5kAhou{1+CwDyyrY+<4U+#f%jI%~
zJ^yN%1mh(tF0SW64&!ubfkx{O%m%5lhst!xq^pF_CHabV6K+hR?XNOEt@5U<RtN()
ze&93#4Opq|nVHwF37uNIbkC%>6+$cq{2wfH=j2sa7F@LYN*+UHm+5T<n)h2B9zTAJ
zP35RKySe?Kd*acG4M0oKKyKdY-K!VlNWtLhNI2E2J4n3C{70a~o;!h}$a5&6m!>kU
zfh?T5onqCl$>E}#mLpf?q1Qd;!`d1s!5!xAZv7Im|JfzGep+Y%d$J9wLBV5i3R%u!
zecrag^#7zQ1Gp`j(1?*rbn1ZkK>nBuvzBt+wQJ)R)oLlz?#+Q>^C|_lMlGcoBQS6Z
ze0Awj%=L!aF8VLTu-z}CX==t<;Y~Ip5D@>IfqPELuTm!skdV01m9etoHOgm1Tg_VH
zwir_`UP~qZ{a5%pUOs(<ujMth3$ua6KL&Rc&V-kId+|tl`L`2pjMx9Sk7<&7cE*iH
T)|1!8-_A6gJN=Tu;+_8wPBekb

literal 116061
zcmcG0c|2C(+U~0nDN0GkM8-&D$ec=wP{u^cP=-XvJXItlB!py$Oqt0{B?=ibmPnE@
z^E^A(tJ>!~-}%m;XZ`m6ZCl>;uJx?@x$o<~uIpajDoUrflhctC2!!ot<WF885H`yY
z2piv!ZoqGxHl2Kg{~@!Kzi3S$FnSUHOXA-<VMHM8C!9HXT+RO3aHo^b!@)J_akCxg
ztYv&y*u3}1)`*$M9HfbHeimeH{<}CLShXhhwvR_e`>6*EVjo!&{0@4xR<ne}YOC4E
zaI~lis!lESEIRj#M7<1q*{rCV-p{=BbaY9wZKZU~<aSwq;plJukA?!o>meSu=~dgv
z{`^Rg=~a2N@z2i#B)iA|{??JOS>n%+?o2Wylz)DV-unOjPhY1|H#~mz>QI>C%UV|Q
zO&gf}Pm?apPrT4N)Y=yl^KyK_nJ+XVGBWk|$>?xJMa8FzFBREqs;e7s^QoMxtgPJS
z9Ty!f)8;L6JLBu{%X1OCWY3;GJ2^FVJ93BGR^JK%Vd1)(njm?$!n)1L8|~s|2&3{d
zy76{#wGkq1#X{5nUlxK^AWX6CnDL2Qw=zTgKVHlJdPA7y<RO~S;9!%50TtFSH-vpF
zxVgAK-#(yn?*9Gz6kF~M{L~{Pa^YtqBcom3CyBSiAEvm&>jWtmFK>lcVgfD2{g@b*
zo7bi_%ZpiSYii8e(ilrTv$L{%{QT6JqXGjbdi<s3PMtEke*LgczGY|O?cmoIrlw~}
z7kWeWIyyVexXsmVoK9EwOpvljuPskeN-sy}-Xr&XclYjHX1C?f-@aWk+nOpF{5{aB
zK&SSSL7U=*+f2R{v6|=1u;R-NlWoy@PZinky^VI8m0_-0OaI`N*fw4kv-)^V8eulU
zGQ_pW^7ERUo6slm#s?2b9SW9yR`|}(&yRK%rI3fnvneSl5x;ijh=@pDqg(nfl0!5K
z+t%`1JjE|WE(>0^TeXdQNj$Q&Ub(I;s`rctIi1$#NP9rzygYxt#$fxV4J0e2$r%|c
zhiJ^E8dOy0ouxu+YHMp-h>OV1&aTL=${y5xqFL{MjEVR9w_III)0*xKUCGqKUu@pI
znf63X0$m^DV5Y%`-EI^V6lzU@PQP1uwM)ds#fyrHE@^40Ier_T=sCS8%eKp;o9P8#
zM<C6fJtl%A+WC9-?rjUal<%@MyRf)Ow{PFUg9mZNSUh2i?>bYhbk%c~>-$G#%w>MO
zyEXNaN5I>ATj}@h+bOFQC6&s)+FKJkT~T=Z_u|}`@`uWlw)|T?`Rhkj5J%g|YJ$4^
z2B!5h40#J)uX_>_vP<@*Qlyyvds12XvuCTzy(rZy6yh%|%5v^DwI|(jk7YNxr|i{W
zCfzNGJ2`)l5I-9IJbGnh!<{R*0E5bhG4$o-<<cd+Sy@?gV_j~J;zF_q0w0__AQ|aj
zoRJaWT6N*{KwHJhzx#gXxT)#T%F-Me%L_TbpA+`?wo=~g$UnmJYxrw+b#GW%S(yr_
zPNx3mmCSPy!m=?9H8sAgUy{{%VpM4N?K?wSU^Cblz;fc#r%wZWN%S+H&c0zjW_9wI
z7s4T4A#`@I=>z_$y4tik@qF3JY((<lhZiqiloS_NK4LU8H#fg}^{ScK@1H++?c5oo
z)1|GYMe6xZA(YSiWGp_deeq(hSz8@WP*-=e`<-X{<&xw^Pi}7R_9DCV7cW@NuQx;=
zfByV=eUx;%pWF^63C`W*I42oPOiaw{*RN@4Xb@x?8X9$Vb*KHA)b&1{K6R>~u+Vm-
zy#?oNZM9upUYJZ^s!sCr9qjErxSQO?#f6N;!NFnLx+~XAGnh}U)|4~o%lZ<_tk5tq
zDQ%mX==ne`E6Pwe-d%p;#EGV+rnS|jvDK9&%@?(A-n_v@n|_fja+(|6PA~k?t2gw?
zla3<0vF{~LOLJqCJ9fOz$$4`6Y<y#d?_Mtt56)7_BgA83s3pbL)|Q2ZWy_W=z3qqj
z_-w7L9GB;hteihY!@)*8o=LeZHYcfW+`6NgttvlXR8zA}#NzwMy!Ri7$JWi89UL9g
zbPBRU=qBdoy7F(mSpG0PJUp;|L`O&WeDIHW`joTGs3wH>+iUZnHTiU9dwY9J%Z#5@
zL*x3U%;fg#A(%)xFu!oEHheqX(FZs830_k)Hs)kulaMIZAz53Q>pFb7SakpXD=o?D
z9s$3887??>T4iQte#^74_+DJ#<#l>#b_D-P=^MbU*lc&8rVjF^=9tCZWn_#u#w!E{
z23A#7QL`(~|4uKX*uLFl+@fqnZkv*|_0(Wf!X<5O0osS5hxMLII0$fY4c0}9i`xzd
zP;L$QmT~ojm6es(4n_$Hi5oX=SXo>9GfTUUb{z9QDS)s3^5x6fz1%f*bxh*6H<Xli
zuoxoRuvABdg<Y5DZ>gxLT)zB$cI2Cpz`m0GLf_;O@TeZ7EW*OVtfwCx6%lcp{8jZe
z`#PQ`vxNPRzP^{sPfuuReJNX8{g$N4VP|LOnDyZiqqy5@u^P8-#+BnYZoG0fdf#63
z^wR4aUyB{4UYlc4zqQ7l3G(pR6q_OTLMhsP>-vJ0ga!nR4}TSu{phwj#q7ZN`^S&{
zRHWqFw$(qEEV|VrN8@|Y|2jV{66EEFnO|>IlsqXbbBmI4*|2frVC+-V2NnA5B^{+M
zOFw>G*|u$4q08d*psTBEgouUGxpR`X!%rh3=sTIHsATT3va$}|Sa?!Yw1nMwxzyPK
z`K>ox@Yt~ey)su0rDxIpTN<c!aP5|sme;Obla-ZiOVe%XAMEcB4GJnZ6hNjJ`}OOj
z$Rm1@(FlvuFtOXF&d$ro=p5(5n~^ue%{`<}{`mRxXMg``@p7DSzSFL&_URE;cJ@9j
z?fLWPXJ==9PA|-Lxs@$X)<j3|U#Qx#Prwjc48brsICxmE^hutc)83#}$A+~=J=&R2
z&28fpe_+Uf<bu_$o=U$Ui=#)6E-(I0Cu6yI@uH1G0e9fp_y}S1<Qzv14+b%t2ah<2
zv0dz#AR#G<wMY7KSrYC%K;x^Rpl~VgrsVZHhKjpER0pU?2gdIpLVI7JwCd^UIlMiC
zYrR*hGGm#qgU8=p?nTCO^XARTcM$@Df>Tpd2K=<UyggdWg}#$^rieZ%ahU4Ej_tP;
z^$6g-^!kY}o)H$E&n`>#rDDwC%Vn<T^ah6XhsYiOBw(4b&qu!W_4VQTobMRLI@(1Z
zv+hetFHOD>q31l|nVXx-9@v(y$Kj!+tt}=h`s7=Mp@4{a`;#Y6*z}Sfm3P0dtV}i?
zLatA?of&FTDDdz6^hq&@f{X=w23eH~VJ^{Ig41Wde(j1N=q$1`dl883qp@&&TG+f@
z@tNomPR;=&BGjkJ(-fpEB~Ei1)>)Gm$!u;PxxM)>-W$mK;LqH|rz++W85?uLYjVLk
zP=5E$o!!5GD=aw)HK*x{Z)Y9+>@+=aew54cY-zNvr=h?_gbQ$%zWz@?dXWPM4yc{d
z5Y5t^I-O<mm{&W;7?JNu5-Dznt4dlN97ySuM`-gr979%6Q3+6FS5Q(KZB7bucEa|3
za*s#yVydp9?(x6HXYrhr>-R3=QTFa>ad9!d$jyeGGjxp^(i+^sNZzJQ$_Lrl@Z<*&
ziD%B7Np20}NjT+86P}m;kS}0%v{QUR%bolT+vDZsWh;Xk>5`quCFHEVU5|@hSF|(T
zEt+2Xh0oS}Uae@}^B4U_e=rpI8YX!Cs5b>AW%>n@bkW#L+r3Z9hCXU4c5=hDOG!!D
zcw24Ucn@ERoMb!vCDWp-6nRO?W<bSrEIGqT&w8w@td0M|fxu={R-a!zLp6K8hu704
z1o<ZM{H@3`{+%UGBD}nQR6hp>M9tfe_PS?f2~bi}2Fjyg#{c}O&lh?rn{3>^SR++C
zJ|Tf8%{Q%b<Xav#r2^?-M?rR8-rSG3_x6-=%}>=!n+Z!6plV4vPA6Wkf9|$C;WbN-
z)i7yE4xY4Ku(=fy64LVVwB}w_&A$-dohisa%Ffz4D4n2J*AR1}K2ls$OA7@%Y1Mmr
zC^IKVQCZpF*Ehv-EWK>i5%E;%N8hBT94U5+wA)e?MY(!ofr+w}d9<<%2M0%uVHWvP
z6oB}^FH2L$vXst`s3%J^zY`eN20pVc#sV9KNBKTW<9nIbJac_|y<QU3uyy<EB3V6-
zT72)Qt5Z^cnw>4^a{PK+`YoHGr3B@u)tNTEUfU|etQV*K0n@6T?h6VCkg{MGRc}sE
zWOte$ziMtMPl^J=;$e9Gx?aBJaR&<|zfeA1#dGJ{UR_kB_Z$9}=PzX(&57*C$@ze)
z5moz+j7o6!!Wu7?uFijnGk^weaq$Q$foq>=g;pxd?{Dv!u=fa%UY&n|@>toVm2rhc
zm(O)!vKP-<y(1Sj`~CZqvLEpb25H1JB(SRl1yzq+I7H**<kS?NEnN7>i;{V@Vy_;R
zEOHTwP*T1Ecd(<QV<XSt(2zF0A0RH@rPqp?_v`EGX2-`lxwu$7UZFxea>VEQCM}QL
zQ@TPbxn3g%coyatkYba`a7{aQ?AW<;C%%qY2>Gwd*1r6lW1@Qa@L>y!F<>feT42ML
z%1Sx5UAhHUo@s{H3=9OFW^d3*-1w5=?oP<mk^p==s1SlMJBVi|CAEr_>^3tLEVG~U
z&>`TShYucP<mL5GEeHj(-Fy2b<7)ZGKvrvOYYye-fDC{u7J`D=$W_d!zQ8FSREXN%
zmd|E!JNZJ989oEM2C|-hZPqpl0NdT&ZPV%w6z@lSB-W8bJ>ECT)YN4LZ4XNviaszT
zQp3&0xHF!VQ)XNLqKpOw+~3~Am)SgWj6cV^t*s8-i;Rpk;bmlEy4<r_g5`aA`MvV)
z<@p|eW+~?{xn`k1)A)1?hr7!3q&gQ7&HbOksHvy|0s{E$Ms*x$@4X!#AIB{vbhc$*
zuYZjnIJ8tt=YRTUWMm*$l%PZ$A|8qp15F9aa}yIu@p7oIk`fY3687Ui-fpZIK?w5M
z6(q-<-79$XXbzp%z~>iRUAjj{l}Im^IL@HENr@ciko}vq9JvGp9zXK3v$E0`;BgQm
zf|)+6f{Ka?ELweSZ9MCDv({G^vlnN6=V7;s+l>}G&86W&qFiRbw&}S^I?WDSl`rMp
z_$pvC&;aDk!Oq_Jk!0f*05Y4g?~znLySsNXFzBeMsl|Kt3LUll@g^}b5efgaoZQQ}
zICNkN3N>lEMZiaG?d^rLbc1c_U07w?(T;@=%x-ZB2~+kGJUl#m_wJoxofSI&ZEw@t
zx83*`39Dc7*RH+v_C7P!S1Wy_TlsCY;HFk5R9bB*)jKjW%-R3F7gu2JA#pL(*MDay
zU?gx+Ln9>3`i?2uZ1i}!6B!4LNg`@r&V9hcTv%8rRN9(+uR?zxJ$-*~uSx!M!DgX+
z*3_-jE03nJiFh<#?e}2g;As2NLvW016mR5qVRdKv3p%OeE|@4Pe!|1?_4WOhV<Kqz
zLncNg@pl|yi0WU~cG67KxtMFJe)g=7v+tcdcYJ+o&{NR~nZ}jBCk*X4#QFzsiP6x~
zin=V?R18H@9SD@iV+L^Zy`Dti{C6USbiK061y=FNTq#V_uEoG$ya}6BL}LB|Hr?3Y
z_9<kdj?>0hug0`a-l1zouY)!bjiS}m_=%Yx`~F;XLp%yx3MD5WU#Z(#X-P?mE&le`
zI^ItFeQ1dCFWcbGSAu_Te2Qw$Yq&w>!+*Y}(eUvC`?+vq7HW5~9sjEHg}j@c&0G0t
zAD%TJ&~5n76$5jeXXd~1h9D>?IKsA3CG<aMu>zX2vbvC)cm9POz3`2OYf^;UG}vX5
zQ$PaE&1y3Wsl*gB5DGM)^jzYigA>ry-Fx?nYzAXjo)CuiW8H*)5M#@qsl=)mW!^`g
zEh{5qemJ}SrDB-CjW7JQ^qJY&3i9$Q)G9X}8<Q)|E}EE_08-<dWo7Sb9eSOSK`VIe
zQ(l*AN)-dIDH;+=MlrQRG<qeD#58s62O^xf2V7SiO-+;W<O&K3&;mw8M0k69Cks>W
z+O>mG%m&~u&!X!!(t&k<9e^3S(A6gr#7dI<zM;W#ZnP6yd&l<eEB#S!xJsn4(#4^a
zGePVV_7b|+x2kCJ8w!M_SvTAVyTlunB(pzytrcuSrmI~D5e$EhJ%z8WF1k<2Tx!<F
zClvqUnIgLxpv##v-r=+rcZ-uJ!l}U^g*~j$2Mm{rs_(VR&dyG0m$$W@eZr;6B<ZMj
z6O=(JN@Pocb-yW_oHf0ejj^+{1WVkz4Xe|GO+1=uYD!w1yP1wz#VJQgRhy5Z>%(n<
zwCb2e*ErT$6k63hc3kWqfdTqwP;74^yMcj$_sI(?D$!uQ#J=3HZF+f*jOBmWQ(SBD
zgpvl!QTgAQrd@%ujOPa#O|X%LEVfIIkDoy*L*pX1#w02x_K5b#-TMANeW8(|p-zES
zMN`uhvV8FCppX!`lv}4ye<1cN_Qic5C8EN^#oShxZ~typPxk~!(bUugF(NAZkWsw#
z&71w)+(T$Q@yw|?l!{%JUIzxA7t+3*cat&oI}f1a#7JJ(F4=sY(HxT|$JybFYHGU~
z8Ij0WS67_Sq^4-5het&4l=dR=#V43ty?O)$1)xlPTpWXxv*@v7$54|R8?C-(8MS<*
zUN48H8_DhG*$~aqEk!^<s9wq<V&dY5QKFGn(e)PFPy9sb;fcIB-IAg?UE#Bb=cq$v
zQu0Lm?%lg>)>d6)AN<TC0HWk&XJ5*<yecPmKRw+-h+V$`bqf1JeFR*RZmIJzmN@kk
z4S*da=6#(~)_tFnR1*pd3;TZkLi+AQhd-IehTJ^;Dm4|1il3j~psk(V+~*gkrCpc5
z)Yn_!a)5vnO_SHJTZMkw$@$QL27q$vr94s=TtKMim>NQ0Zusj7uN#}UY|&`n&HHg5
z#eGjtPj;^~(_-Pz0^dG;8b)qb>yUDoG+eww!~s4&KFQ735O}etd=b^j7Caj^Y(Oxb
zCjI)S`57J=0SG!vT433IciESc?xMDKbvJgS?5`Ga7LS_V#<9KC)zv7OYVqq^h!8y+
z?cT-#Vna4Yg>!eBR%YxwcQ7lU1Q7=4E;Y6-G`L=q{pdkP{bVjYJlu74q2D9GZFM1`
z@ul%q6YSe=wu9<^ZQ;N~V%07v=R~4nRw+W*kLtiK?*pu?Db24E60W0DD05u_1<6D4
zz{lr+$GE-3>C>lsOOPL;52Ie4*8A1pzc%rK*&_fQ*f!61Rue0WGvc>?zQ+dPalUam
zn2eg8)KhlbHeWOYRz*L1dXQxg?<P-BiqzKDPIib`U+Mi`8!kry%znO}(A3!YWDZra
zp|orBHX6jB+_nWnL&Ju>jN-Ph_}w3s^6~Jb*lanj9RMhAYCnMdRf{fb&=>WVa|~@6
z@7|Z*sQ#CW?K6=>LEIQ+YVD~+5v7(z*P+s#V~zbfo&CViFaFZa&cahsE{Wqcdu2ZY
zhInnKZw272oq8s2$F~l5zGNEse3N#b{{aN3qSAo#f`v#F+l;GjP%$wX1jo?X**P*Y
zvK0gi;w^|nS^trk`+dUZ;^#5cuhq5QEX~h9sQ4Medx=jwM_#;2R9t-P`VkQp#x)jP
zWtcNoa>8o|#g-E;syX`sRaa2>!DY+>h~nk}qboZ*R}d>vOzA?P#S#v4KQk>wGi#&d
z>Yx?Bx-zFHw%ZLEruStuPn!VC$(Bzc@$vB>Q`kI+q0T$GXWsa^K-eKJE-o=KF(D!K
z^7l`UiHbISUfQ~YA;2!%x8f&gAz)hs&g^?{YG<jk_7S4Mz{1AMYYy6FbaWJ;#}9?v
z-@gnS6*0`f!0`C-W1C1uLNnj)amV{1mere=Q5EW+9XmlfF*{2`x~ct4C6WNJ81nwL
zY@Vmvj~qHQ4C2%y0J}|-xNBGs8~|mg&=$1Uw*tus-@%A>s5+RZM)K(87W>qyPoLO4
zXlZE^wYSG&SN~cQ`^0lFP`<A&l4qAU10!RL@}@7Tmqglccq%JPIL~L~=Rc)7pf{qt
z%3rL)iC<Wv^8<0AaQ%A!<r2r+!(VQpW*|L<8zw$@d!@l-ysOy3y8GSc!il1>QeSWH
zBLV_qw|;KwE62^xpF9v~{PVpxC6k28_nGB|DMYhDubTVfcH(oD90!W!6gg{J8!qG#
zfEYT5LM#%tk`8Vjj8hPmzP`S!ub2>?8pkegQWg{{>?5pW5%MIK8M!E5Cry!E96&Hx
zJtdhj{T%X#2bEM*x@iN+ilb+r9+#e8*OxC^F)9ONW9d>>U`3}lubfj+yKv!v-u6c?
zPm{)M6PU6u#+G+<aNr7AK_&p&XW+jgqp0|h{S0?F7&k`!42v!)*^eM|Kx`gksKX9T
zef282oeod;NjCm%fnUrBOAnr<1vn$xi1Wt`#mJpd1Z6@fDu_gCuCuc<xAP`#r9(8n
zM^9*8Ouzr^*)u#v<uHNx{CutbRL?|j#Wo7n_m;c`?^cXTkI>c8(YaRltN`UFF_Cl8
zeFX_{;Ne`$gLD<py&ijF3_hax^Xw+yyn)2G;zGw#ld^P@^xpQVbP1P82Cs4!a?cz(
zbZectW_wu-(QxK)YZ#zQvGG~L`Ok$P9UKFVH$Op4B-jwr#tqTvJsbHK+Q_HrJz`^H
z!P?>~ZhkL5CME{R-lmy;IbIOYV(#W>M}a!y$2YxJ*Xy30Aq85Qn3w?DCs<qdZXXeu
zL?<KVHg5$rtIZv3%CV2oAoXfv`gLPU3F2n(E2Z(R_}!My;qj2+SXQGzRa=`KE-##6
z0R6%aa`NO0H!mbjqbP&Nig}TaW-A}J@m$hLW3duJvV!i|g9?ez>cs>vFE79mkyau-
zh9G!zXJg9FbfTK%K_%oSIw-<=RM8KFXkC<a*{fGK`aXrBy>0r{fIGIkb;|(^9BvD!
zH!a`njXMeC5VX{jryF{k$RTIwmbvOm`QAZsi;9XWC|IbMUVHZV@z1d_y_QT005LQx
zJAEsrzj<L(C53YkwZh~$9E}Mb)>c-Hjg8HpA~q5niIE<-W_`$wLTdm1ANW2L2<TQV
z7qva!zkk0xUQTJVS?I)xa$qNfpt|<U3mfjpv`EZ)1RO3~&ca{bx^)XXFy*qdC+O$+
z;`V(kzmQ_lSD^o*<G(_PsD-vRJ$TEZC;0ic^Ww74M<FQ$F4`hhhJ=T=m%2DDjyz-s
zE<|Va8o`9zTOH+Q))?m<5%CS&^Pt^6&{r?N^#(5kFFfhr|4iC#73l^$5QySp%Z5^L
z;}EVO9e4zAslIIB;KHr`XfhGFt31U|XgG~%f2dD<v?3yH_4WQ#0^l0YwW=+lnhieb
z6Ov~Sf<AP!tF+|RtA}r|C_d%C3YvURor0cuq}XjAUtfhYXLfp@3|BNrJF|?2lBa`-
ziRpP%R9BhX8c@vCKk3sU8Q_8414ZD~t5;YGuB{EZbK0+OB)tsu@bJjU;Fo<66hy82
zL-&!=ureu@bH@tfKxa{&GmG60uO5_^k&#GnhEJb<fzE`rXgt0I`GJ*u+gMeUSw?oY
zsuBYJO=Ts!q@*MtU)rlziYwOpdck4$_0>j9=eG8#ZXytjNyCNB0V%k3`IJ`7w-5$|
zkhfm~ZOh5YA$R21136#1c{43Da{^0yJGc?8*epN}>EM@4;+KL@Qtk1ImRt}Bd-ZN`
zR9doh%jet8q%7d9sy92|yva90uPOZ^;^cvt!_5+7``zf)utBR0hR>}Zz0xle687%i
z&B>%G<?;5*m;86LCX=U5okG@|bX_{@vONE}qM{c~3(6pSP^9QBeLz5nQ!V2o#9g^1
zs46Ern`bUs=U8iNYp_z}M#AE@pf7{W3m3+*3&7%0vAnOTQ7ETNs^rLAf~;XOBjq?f
z0PayO-BB-7zXBQDZum<Kw|m{|_QKm{?O{4C_U1#KMItW8qZ$wQBcZ%JY`SdrC?_We
zr11A*hwOLfttZJWQ0NAH(JD0TP1cA@Xv+Yg2zR=C<@-M`fachRB%8Wt$B>6agoNU=
z8+=d3A`z<f5?OGWNC22bO09g$7+V1nLZl2yy4!%kIGA&^9q3KaBs~=}3%NfwGc#kJ
z5C(c#tzG;4`JKAShCkli*-FXOJnn60gsUMVBb%6+i7&6IahM(6$0))#cAW?{K|Wl!
zdFp6voah`YdTh~9)T=yjZ{waU@X6_Sm(6^E)EjxHkAj1%B#JA`^Qs9-!DguttZv?b
z#Jsk;GAMWH(xt^2=POsDtu`PX;?LMU)J$lIsu$$w3HxG$509|P9`T)X`|*Zk!JTMG
zm_G?!Aem_+n+JHKhs@HKBNhnmAu;Z4(m7jKquf@#A3WIa@#<2ZQf;hJE;m>j5s}co
z=x&2X(DAe5JW{1XAWsWxIy-f5Rf@{WYiqXyQN&$HLX3xoYR>%3eVFF7hW!NUt6~IM
zQp#!e5X<|P7WK?aMeocmsDo{}(UQD}#Sp@jPnQ|w)w&1~f7^s~t<1X>`kzN`2s_~c
zdIU`M*E5S+%48;C+kva`uFw~bNL_fb4|$oCoLrxu7SJ#kROFf+qkw<_9-`{V6z<R6
zN&{sImBxy)5DMG;{QPi7foiGYAwsi){g%sX4NUgh!`m4|Eg{W@m*rhUz=6*tV>v|g
z`%{=;&e2pAsIH}Jt4@oKSmEjJ5tF8u3rkB)eynm2X@yLiJ-6*$g}{i$1^Pu#cD)qO
zp+nASvs6^5T<o1j+H)tR)d0-1@G&HIN%RoNI!Q@M$YeK!mr&0@m|_zmxAi9fBu_u1
zeuVZR9#P$MNpV5JrtjasmzTf5gNCs8;Ne5(`EizY?F_h7!?_0A01#~o99}SwCb5eK
zR|rlXX&Q_21@Z_P%eK9IPyBN<4@*cmqtEdOK#}lck_^AeAab(<IYA0%M72|TCYtHk
zPej4hL}mNxI}dyq>9-hbjgSEp?6)c?tlLH~WhA#>VU2q5yvlHKetv*I2ojF0yom{y
z?!={FNb5D*?A2W%L=xpwPfq|>bENKDp-4mbx6@avyJIETcX&~Q0y3(8(qA7n_BGq9
zZl_Exs5VZ{L_F5qTO9zD={g0CZEYf;bUK9gfdhi%R&ah7!wwE`$bI|v?1>lcEV*;%
z?c;+A+$+tAc<{VB`6}!|025~>co}wFM}@_*!r}m0-H!X&C{eyo_A2lG(oQ^v#C8%I
zxG7@Vq$lHlpm8xZH%~2BG$zqwg=~yw3uSt2Xoy&`nEW}YNX6_%CG18!P^@`0dx(z`
zYK{&NDE1t-M?qlV&ZzpltSs#;!z%b#2FE9fKRrgp^kWo%jYkI<0QK><l@&P|+1ES^
zX8#W<{Pp-}^N##X@DS);7<NE|zk#L=;=-jun<RGY`7*~DQzU8q%zX3^)#Y2S)jvNM
zH8u~-VfJg&I)6_maDP(b(#!>rej)X0`>7fp52B%sn!J&nrRTPC8(4P3ogaWnxy_;5
z)P!2^c!^cKfA3LcSZyd!=*T?4y3T6IGW(H~lwZppJizro5E@tIE=a07*~LYJC~UF&
zkE5ai7z$hR#Oc#+bDegH-~Y%7bURRju7RIF5SWNuZ$o|yi;ly1KfiY*Q*=)xIXOA>
zpe6LuRI<W1J05t;{r%wX={LXcpypv+wJu&9=_m-I5-@8$gvL~tgBX)4caX?lqjQ9N
z!1Ep5Gu9Q{XihB>bF*4R`^g~9GD~H@?2L>bQwzw%O?}5nqBW4eHrg;+$dZQE=w=yu
zr>CdioFaamO7A<*ZIe}vyQryqe*UzHploKu0+#^TA;@n>lDadMQ#neGLppVtY0<<Y
z{{EVsfK-~+wEpQq@JK$=_r1LAM%vXit7IarAYVhiWMzF37ssaCzactZ<;`yrkh?f^
zudjc`o03@yd(D?h09-aSiN4<6NGX^0Z{L=hE)k6!BcCif@+T7Vwh*Ed_&wI0CZ?6p
zzs@Qskg*_s7?0?0VqrUY@YE{*kAy4yw9`kdKJ{!P49T-~xD)O<&DlhrHvcQv3Eg`X
zwU=}g`e$arYZRTdOiXD_am|79Y+zd1g9gO{hoCUT>4i4$baXiqBNb3ZQ6DMD#2@{6
zviELF${EQO^}nVQq5VQa!F9^Qrp=s8{v!6{SHwD2kY-GM3YV9i-Q3)+U%yUIPY)>%
zUHYeRAs#WYLY=2thcpDYx!Bm`god`E(N0ZG<vn!hhPk<}jt*4GC%-R2AdgY8hwKq3
zpJOw40gXQzsIaiGuV24n1CIk`sf+aj@dMkVw{5iDvYqbN;NSw%T)Xivw4RgRE9iV+
zAi;T|FroQ*78&W!B$-_LwCq;T{gf2!D&1OaC3AD+mbJ;=8gf>%9(~|hGEZ4CYU|6F
zpDXqK@>;C@@m=!8=NI%v1(}yi**r3srH&R`9&bC-LN2j!-D@YbUqa#;%@D$j!^2Ya
z85%9G`}cu)h!zXPiKeRp+A@5dlF}pR8ea${#rC&gf4P`_&AhWPxAETpHjvzeOAbe|
z<4kg7q^`XDgFAPQ7udnFB!29er0eq2RZbdTFswM>MREa=cW2{PQ>#*d?X<KsV1+;n
zZ-3%JNlrdFJq@DJh*t>MAGa&F;E#3tynY-%ejKuKy5+l`DAz?3FgBCSgEM51g?8-F
zkhsWrS<&6pK+>(QtFUFfqrmFhg7lh>fV1fNMZXriKXPdFmWGKbL_7UaW2#O_a8DZ=
zRAlAt?encjQ1zB@W4*tASr+K*-5Uy@3q(;g7@#?j;2EWy?a{Qr!81g;_|2m1^WCP7
z4rin;NS+j1deB^=kySi<wh$#44h^tI)g@+Vf8a9#U?S>5nVD6!wOEUJU}d&lRY1D>
z1_ob;&)$3tf*0a|LYX76(4whA4+P-DuzUBD(9k@WCEGB?*QPDWbQWwvLbpMgcu>*P
zM?m=kpDFDyNwh}+o1x{mw6<1tz0wT?3*-g25t^oHLx1p3tIG~aYK2$c-bHK99(3{B
zUTe5-h=37IJ^mAZ)Z}qMfDCCM^WJF3rxxGGS5^v+ACHUe=6L0{*1Ki%p8D|knPStu
ze?+r+HJlx#m(^prUwSQ2vkXGvhC9)JG&Vq2!%Qt7ivq`)A=r&bSx87oun0jGQT$i$
zrly*oKmU1Qs-K7V$HomCNLe5=`#*em6e5-dtS!s}h97G}4_kB=#_L$Kb8yW4ZW{x!
zGkI);k%58ckQNmK!!vB6fB==jMmv-}a3YHA)_81y7&bj<GD;WvB0!v0e)vH2IibuD
zskg6kXnDw3`1ttnksLv1up?bAbDa<O34-lFm&CYdkCBB%y8RP0HOKF5VU%|5XlMwy
z3B$-o?8g(bvW-nmdZo@>7G<-1?5Qn-Zu=<Y3c5DAC+lA`ns{x|&nTt%USIu)QEhnD
zP|vf_`S06^#Xu_9d0`Glk}458IyzUh&>jKM0}6|u#Ue_mWWU^)g*j$-9>_GJa{<=`
zi`Wn)_y1%ck#Q&5M?~Q61h-<v)A;|ek6e0Q92LcU?`?oQTMkp&gxtX8ikM;4X|#3H
zASXO-e9em2k+}DE9@fn2$YDqc1HS0RoAwqKJ|(EX<K+H1(r(m|cX({BOI>|TK&s4m
z;^#A`xwETdbH!4}Dc9{W&fH>F-#R*qQO$k6MLv5rkOrTiGwO#=5gNlYq!8L6-$c%o
zmFol-Iq<yiwnT)hkF1P$-%Hj-`+Q;4Km=g#SF}gK$B*)vT8gA$A{NglzhkRuUAiO<
zQwf|7xt~80C7k8`j!=l$m7dYAQ|T*@AMbLwF3$3+=5X2Y=NBK42KMyya_~gysc7z{
zmQmeBHus*aH2pBNaifQKj%WJ%8h_2o3gOLt8m1Amb-ma~RUyxroaTuvRr*Zq=$)ll
z*bW?!T`uc1B1TXsvN<GZ0Lq=z)T|nk`(IxG{m9DNgIGe|>2jVrck-l1@<(zg&a3jm
zl3!`?=&*m@xr0xpmTgu(tv4b^6dtZRca_4u2e7Ln?Eu*I&uc<BXPKsRgh@2xGo^H5
z()S8Iw@B+H+dENul4;+`{@8bZP*;7ZUs^vjX@f#T?RqM=i^V?A%E^IORHx8p5cCP~
zjLF<#ukyifd1Hy^U$84aZGbL+?$Psr0Ly`f80<Rsg9ksKe$UeqdurRccWN=);)t3X
z=k8=OIFi9*^j3zrIkLjtutyHs`J+<%mBn2B&uK&3?vk6V>2FB9PDNd=msj^=tgPj|
z{NnV^mFb{3J?T`V_!EvQZ^~e?fC@wO8seJ<v_l~Yn!t0HM(jZdgabiU=ty3iF+@t$
zCs8KN%*!i7LPW&_)SZN6cFoWbi~=ww*616Q3s7pvA4A_dvab`fziw>Dn>Wz_y2z^7
zy>L~GA+r*7yzU~nk*{%RX2Kt4M#~i_Z<@RF)a0$1UUG}t>kpf%DmJNjt_<(%`ru_f
z5*|*Ql$&c@OE<Ar_u_n6%ghETF)>GACgkxO!bNCNP;t(nsQLJ47ulwQd{4_JQpd&v
zkjf$;Op)HQv3VwJ9&ROb@nV>lmlvGesN=V89hr0M_w2J2eYAdf1YiRJz3D**0{;tg
zb-pjcBKE}#kO%6Fi{LL&rw@AAjCOqe_RR^LvQNG1+CqPnkd03-7|>BP0B8!A;i1my
zg2DD|ibm>_yqW~x?3|oea#F;7S^d<aYckJ5Ggpd1z4fJ^an)meP0hwPG(w3@o1K-E
z>cDUSJpp$h3a(taqN1b2Kt*M_Fli($T?TNhJ`uHt_*5FC7Ums9L_`+I$9pLl#ilz7
ztU+O*ab5T5s;a7zvG~ymK~8^pY2BZ22+fh>@RuXX%Jr}_LGy>ml_EBc+)B!V<^-U`
zVZU9{(4{rylTnM<J0VquduFBKH2+-H-8}88SrmxvIhSIbI1ch_z?ZlXu6$O+8}gw&
zc5Jz`sGw}6l9TKF9#+94J9)KKmPhikwp^rW9UMFeCH&AXZ?q!AKYz+@-F&M6*V+#?
z_9w6&NSi_>D%?->@u2^1>2W@0C&pUhY=9mZOnVB$VZAB|X=%OM&;Q-Z6C>BJavOe2
zj{tZ~&^2<7)z;UKHNq~EG#D;qX8Ix!5q>7V(TX0QgioE6MYqUy59ucLV-agz8^HH$
ziXK?BWCu8~9l4-hQe$5nTj85+pzlL-_~z3V@uZ&3%ZP68#`pA!>_KS?6MePxopD5T
zPw=3U1?J1j&L(5Qx}R_Q2BQrB`rrW~08PU{C9)s2ENarZy&jq;&CbipqmH#OL_XT1
zeCABma+h}!G~v8Fk<?>mFMivjkB6aeGs}A@s$k4~ioiuAByg0l15o9DCR$6gjB0k9
z%?}UXprrIi`jq{MAWgBmpk4H|x=Tvt1V=EuWqwPud*<i)CnvcdUCs8r{^_H)H<wO{
zc59rs?NrL8p=m0|LNQ~=>pwppK)&V}QJa$4O}rW7JI>3?0KY9B(7m^A^?LB&0i+d_
zZ{?@_cc=t#gMf=Z6+d^f$|**Q<$}wWjp3J+TmeA>?8@pfJ=PUPvye<dFMMeKewjlw
z{y+{@j}Ph=+5j*C3K9`0g!_Z*7FD<npt?Y8RIlBlx1C~(0U!msqEEH8krEDAP;0Cc
z;4sXPF7U_!rosv(A@M}>bb6p11Yztm^gwct7`|e^p#FMzc{M#x{f|{B-qW}08XgL^
z1jHP)W!6FL=eV7pGSg9oeHwqadG-GGLqY~A`yM_%VsJ7=BiMEzy5iGNnwahzN(Rw!
zfCR+c<lI~oE=7gY8SD<A8BD^k2ZPIt`Muu7(xqW_up+x;QSLvN!3OmfNYaLPVtm|E
z`wcX1XWTH%2~XtNPCsHeX7%fVzB2$nevEb9G6RLLE=-VP|Ni|fEXQwgO8s9xBg*aD
zQ%Aiue)t|PCEd2|621a4SpJT9?Dj`>wP9wC(l%w|=Wi%4?;%QC(1$vRR3l<red%G2
z<&j>yF$g^#0m#%q`T6Z+_eu>uyeDl<7s+?hJg0b1n_20^?Lv2vqenYXBmo-H(E;CV
zUzo+|KqHpD`zi&wY6i%d%K~`mSX6y_iCXaIbXqbN2`0**kwGruh}tXsrsLb38Uz;8
zH%Xl=sWcSWPC+qFHpI1{XDS@wM|ocPGi6|XKN1Po7|zAbNGW~w-6mM-X3l{sx;gQ)
z0^}=lZs^%~o1cXMH%0mRc;KusN}_E%M3aYn3$zaV4zr|V&d=J*Rcn&eHN94Q+#-n%
zO~8HRTcT(W$`-|Yekdg!xN;mRnOb%gW?vZautSlqn(RIcA7gklITp9<uw><>Qi1LR
zE`2;(nBfpOxr*$+zhu71%_Tar)Mo_u(b0vNU7V8pl|6@NDdD<ok79#3R8&xalMX>0
z2Dbs^vuZ}Ai47q|tijS8^*s~6xLRf_7=Tgz{p-<uAm1PXL2<k9w(scqwNd$f6BEOz
znB2|=@rteSzSAF=YbyQwcfC`mPM|cMee~owN8YP!NZtrT6*V=(8#g-8Q#^e55Unfv
zbdXFlBi}TSts?vZUi>C!u{!FDcI&w)`er+9ET9O;Sish`M7-P~Gxs>CYN&;c^MTrQ
z<PW!tT2#!;0=K!Zwhk_|+r%r)y}`)B%4E%9Jc!HL*ZiExgMdXqYh6b}f}b?kwAS+i
zEIyZQg(qeXirYQ@sULiVo!#)Oz+|VLN6J{_Xl|m?-f<wrgC200aVp7(fha&0##Y|6
zc{3Qaf{sfd;s%PU0VE+l=7yUeo4j=v`x4Vh5nlJh-}PES4n%DMfuW{~UHWG(OI7Fv
zSF;01F?<XJ!1`F0)*&$+OzrfXXjWBZ&xT-t+kARVryrJv0Z951wPU=uPHR&REHQGu
zNc=e7rnlX0)G$$M^S&rO0gKC3c3o9;BJ33vLq5)uYOuH9CWS0gM`^|(IuKP15AY)i
z9cduZ;Id)k;yQE7*3qZF-)^w+R6#D$`Los?D9x>*t?g6h;OOJ}V19mx#QDLZ0(($!
zaBw$uvz<T~1#|54SnH47*Ix1k9PVZhJ}a~)YDv7L2}YBPVOQlJ{OFGg+qaL~;LGsO
zB-K-wFB{o-A90a5=hyPlOUrJXJ(EBD!?40W+50T+dR_k%&$ACZh?$u{u=%*3`)r1z
zkhs+4a}VEtFMabNAt!cKQzd>>iQ-MSC8S&hqF$p^po_5f2tc<ww4ZnVlc{Df@<y&7
zi}5oAo4WmCg@X@v%2BfeupYFvEznH+<#|8WF+I8P+eT&5Fr+*fT4ZDhUgZneOn4k>
z+$HuC`WV&-s*2-zu#O3vo&Lb3!nzGIjIF+Vo%Qs}t^1qbKheJE9m30V%c@aZV@5HO
z@5fdO+JGldb~7;{^*jP*rK6{Jn;(w>@n4}|izx-f@1Ob)1Q26T>Qd8H8cDzV<P#>h
z5FWE8xg5hYnJGMPIUg&lBy6*n{fPZmiU9z$D6soM-1`3M(<{_=xQF6i{Qg79$!OAk
z{kqI3p1+k+V7-fDXFv7)d4+Y~M^OgzK8?dbj?by|jf`@g7p(rUTv(>H3N%ZjUG-hL
zzA=s^m-qQ=F4u|x1Q^N5wSP)wYP$D#ii$8qp7g|NS=8-!YjvB~rPr@s87c5y(^&iG
znIrfa=S%6t`1|h`{M7c(V@FUSvsHX}WySGcksaXo(lnP&wp?RbS!<T5@xQ4^_p`4&
zHm!#6YB8tr{by)$&y|<)6D!LZQaYV?{$0T8R#pzLGAEI_ye`^<2b1*;D=T9?xB0@W
zf6X|twu4UKud$CXu17rAiNCf|g2$8KeR3zq+tQ5UUnW1iDxe_e&!<!L`jz=5-fLUd
zJpZ0PI!5q;&#qS&%OClwQs0~Dw)y8Clqs($b<@jmEqSZwmgF5cJF1(1%@2)RZpa=r
z(!UNvnG<oJ7cC>jyh9FuShT#o>8~Nz{r;&JGu_&GG%Mp9TfI_L|K0M@-48o-Ru7k1
zx~)Z?*LlU=SN6Ny!Tw{uj(Ojf;D2`52Ev1i;&QKrOL+oT*6Nj(qNfC}`}{dS^}q8w
zp6i+EK55n#qVJL~M)B`rO5z6>Eq}Ku6xp_DYp6*6waycUSY}t2HZU{a8!#3+qQ$IK
z_HURbQB%v?d$~4ZdwmoKd{Ivu>!qkuwayD%qqxeS6Grb${%>JtHVN|jl^JcDYv#KB
zyFAx~Rr~cT`$@xBlnHkKZf!u)uhq+}T`k9WXRB9}-oNjj4%7N1>F~Jh7ONnSf)Rz6
zs)Uh`aZ7U6wc4Dg`<nUxeZgW<&J%iuPhDA_3$v-!v@|I=J6KMm)JtYZ1QYlC=WPDR
z)t>xFi273h98cuSx4mhXIC5XN;$~Bx7Ag_<{t&OSvhtRDZ+oqxq}|e_tRS&q#_rua
zSsn)x!hp!Vr?g-a{n!;%R7(^VXKRZrSq>lF-QwhO_pf)Bu;M<`O(**Ax^6QHN=BBM
z)6dP{IuM-f!)4ZTHpj+fqB+UKPd454T|aS%87#F|?=MnVHT0NK+<#h>zFoAOq5t=&
z#LgVL79~-oQ}%prwWON%6=hq%`ue*sCPhcP*Iu)}{dpf4R<>Pa{vQl}kf;$7pJxe_
zKOkK#y#4CQ?%*Se*Qy3S$40qw3yHMpUyD|F>Pe&lhGNk`Zu5ly%)al_zTOhBU~x`+
z`t$NKGN{Ns;bwCL%y@dl0G_J`z#I^_a2)}Wc~tJ+vj?o_ZIFQIouRitU!6u8RqLf<
zWE|=36|x=HGBYy+_X}|n{2>rAObJ$jlP4gq{y%02y%jbPwr}pQ-qVv_YC7Ea@}=Wm
zf7(FmJ;83{`;+IkY?dU>Nvos-vgp6PyBWRv^yxtP>0ebrIA1A#iA#mr1(G(d2!#z}
zmXPQy4j)d3%El#pAowgG2*a*jvH9!6egAE`ZwK1O;a|U|=O+vnG54$_PdwmQCboH7
zb}(r1a8M9tNSEGjyyNfxHd5SFKTgm2JT~(CRt>X@$Nwgop<tk}(2x+=B9)=7fj@=b
zHa<58Zxoa@{|67`m6bvLy#ps1an!`Bw>lVk9jq~y4?ICOu5(h(b!kLIy972|&`O<+
zjgQ}6DRy4K99$6kJ?K$iGYz~Tuoj%K_>#ebDKY>&7#7gM0U-biNZXA*J9z*+(&ZaB
z4$)7^ZgeL{c6?ju_iWo<)EL_#wSf^{TbpKtdfSh5QmkilAxc7yY0KGjZ=fHT4rzfE
z@R1zCg~9&?!#s-zOn2$pxz(^ljShXwONUh$mK3c+#K9tc5@rm!k(z+MFf@!q3+;@t
z-Tyl2wn6GYlWxWkJvx8>_>lqT^~aANu!MkPL<`c|Y_dV7<jzdDjP?9H3Q=?`o~+4;
z7C1nKkVEG?e!^d~Z60RiL$Kc6B97@A-{c~v3`aYKh=Tf9A0~J@6N6}^o9=?LEOA~C
zICkvfLW3fEQ(K$n<q}o697%WVVE4c}?m~*=)Za~GMalGQX-SZrTy2YxDVyj7f#RPR
z;8GUb*5BnMESuJoVY8r~>w-~1;1J9oFAr~_pp7=w1P9{G>Q(NIq3@Q`@V6U?Odj87
zG+aj&7zlK`GM_q4vbi;}zYL!^&7d^#gez<2_x5kEgN?469XXa*w4Uc=Qj2c?)`07%
zWfNLH#dEbf2LZr2^G=n9L|F0G8fugt@W+Hb%;|XOkP$%puaj-SCfpXyzo-Ze?hP~D
zt?Ad*j|TLY?QZqznOe|R+%Y7!o^Kjd)RICFz@pf*fc8O;AA|E1a`2tBgBRb2A~trD
zSzA_=lsYwQX5L|)B8~Rch}NtPl6Th&S#&X)0?6eJGvCpT6xxm5G%-QUeE_zD)Vu(u
za8Iei_ZCWm%8sS(6U5Ch@(r7z#BopBMb^yhCT)%AKe^m}75si`NhMKdvmMcvmB;PE
zj*n*R2JleV64<ZS?=d9Nt5**moDeY=kZ^nc(J)#6thnv&RGo{c9dGB{qlun*0wJ-<
za8KtXc)XVHJOAq})Qg`cc@hMf_@@^yg5CHYb7H8~OHgQ!gTqx9(FQ741$;DG$!TfU
zynq)$4MTCR-TQ|RDP$jDuUO6S1ywOgIcqt2ZE;^b9yHT!xZ!waM8wA?)z-v7Ve5@0
z_f5_dqmj?F^;MY6{{0A64`kO0U1$Wb-6xiufOB=1?1c*#FsZmF&$mL~!oouP%?oBG
z0yjI2HMj)qp!xa4SKp64gE!M^wCk`=30;f08rh$F@&!DMjAUeC<KwIQ`gH-F6H&m2
ztqWK>&-|N4idi;hl^;@Ji1LFU{a);by?cp9H8@u(X!);z=$|(8#3n*w1uNCAu#qsW
zcrmQIz`FAFYd_cd<Gn+pir2h@RF!OOK&oDx>T9p19NI_`gBWlZGetv&uJ>bgwU4(q
zJS#7MrU{v~HaPT|mg5QkA9Fklwp;FDn&>L#oQ8(Zl7>k-y)G$nF*Y{ljVB>Ez9(Yq
zn~Y0QZq4tnxQNc%4BT1&f@D~|+(7i--MkHc1G%LVG8W9iW9f**aVnD|6c=PGO!a1j
zh3)eyKO!g?Az)Mu8c_JgmvPA5PWJXk&D)zB8X5poHJS+Sh5jo8fuG%04US&D8$P6q
ze(kco{bgg$WV`v356Q3Frn({D=QzLIaL4^a&rVq>*JVg-xemsOT+UKaHNDraU;iz(
zGm)K|nwmX`E2s*8Y!kUnEeMvolLxp|BsWqu8~x!zJp#Z@(CssQj`ul00faB&mgqBO
zi|t3>>SsRm^P8HStUh;wu(-i4F4t#ftZdnjp64QS+KH{7lhvcLq!|9a+9G9=fAjqY
z(5UXN{6fT|<u(mSudv{XZ$+W71Ds#Ul}CLzr4X{aHVkkvdP{fcpO=Hkuxc^yu-Wzx
zj2W@*8ug|Ix(XZXN_p6A>64(i6X1${x9Jf<W0`-jDN#e?8ZAv_SB`gZc(TC#szYJ?
z$8+>gnS6V#{fUB(;U5m!(`4h4ALaVkb}TLTRrK3GP*FP=)5?$)ARy-C9A)xvL$X7Q
z2$^wE>@dN7?XByy{Er?3%ikg4v?le>JySFlw(LL!X?XJoyN?ur`T$R~C$tgzTNO;U
z5JPOozMpF=+CXUFZLJAi-cAotQ}?aNaC*QropP(n_H`sCqX5uw6g3uOm%X}(Z7}kv
zF75T?1er%+k+t8ly^1WDPt<~7WIwfyW_N9*0<jU<^5?GITkV~kG_^`pfl_=PK4cKS
z@l;rK10kArVWGIQ$YE*r$jJlhN3tsYd};U&g^O!e_L~Wc5RSb}BaSn;Z|MfN10XPD
zfU%q0APT~pHI6__A_F+nW6_nC<tW0>`TPhgOl5dP(T-agW!FU?;}h`3#vj2}ZIWe2
z^K&q0-*f5uxrwVKuBig|-*WB~l*@6vo#g4+I=MmR&vdWx3Fsw$?@?(<>!REyU7Wut
zM<sjjq}+EvJ6q1wJT><fq4`g*(cS=#@~i<g?%V(}!cj}a!So-QvEe@QFWw3w6!c}M
zEcEfXTxUWwiGKXcNf8TS3bG_amwfe<vy~5l7*)1j-6f7YY}g_dB@L665Qzkn1<-|d
z%~k^4?!F>BzYt!*(YAViQ7!eE%dJ^PZ(CE3<=qSxJr?8WRY<muv7!%RE-c?|&DGQT
zvtZaK&|<kaJGGZClE{FlKJ*KTqbu`XU{pG{96QzSq5|*IwacyfnMj(6-0SmI?jhp~
z%ix{dmPYd6gYo%>xu#t-G{V4yx(BxqW`eXnEv?w~94@8$n;>H?%4!|vGc0FJ)y7oj
z{wP#AUNya|@bcib6Ml~#xh~D7%IqINxrauwk2aN@AVm7R{J7+w_A4h9oCNx6G}VM5
zj(drmZx?5VIh32<%MXuiYkL(qI}n3qHvQLEB~pIL%p`U)5hCa+rPqorYw2E5ZgnL8
zyXgi*u9<SXeb_OYX%P2HQC;!kx7VIEp`nW59{c)^yZ!4}T`E4+^^JTj@67XOp-J)K
z5<J4Coe^z6s<Y4QMQsPcl;JOQZ;+-4eBks$=HT!0iG_;MQ9Rb6?SFS<wC11f&3ozT
z*!QG|NBB8yZ|&sOyWAS<Nh7#1QTK1hBBP9Y@46a>anWkpLa7da>5$hJ=@p*{IWaN2
z$KmOP9vX2PLn+gnF)amRlla=646Q?3?%nZEXL{q6SVl_$|K)p*#00HFG%k1C6lV6W
zP<<V~Z}mZItvOr#!mJ;CN0h6+r0Xkoxy2UEir=ll?O*uGuA1^@w|j^;I=+yTI(a{&
zMq4+q%8=rd@Oq|-_J1yQkVo_J1Jc_^N}PgfBfRh3du_F$;m-XFZ~A}zGG2Hf!Q@}l
zyGs^qKHDzu5dR%QcR%!i97MwaGgcxysLb_bJkBzdj+s08_zQvZaet?+0OZM7;Df^3
zX4(evVj-<p_zO&_;EgNr+q6uqeJPsCr^{X=`4Lr=qfn#Pu6fUp_v!(^VTEd?-99`g
z<ZQVglHb$6aj%l*{LAr!X0K0%#L=p%HO9*47?=GX^B1&svl|`xesgKop{&n8q5s&N
zM1Sj(e(U;UQOe@>2jcsFecZaaDpHEKKuV{jI=F?8PsW@%Xmst4a-6wts&bsEscGX}
z>v@`!&}6QyCf;=u=L@arwhW(YP0S2fGavuZw&Mq7=DBjc^4Gp)v)kPsm@`U!%F#7n
z^v<c9>9&_JHXA!nH-EDI=Fw`(%om%+Nm(F>kj4m;{4Qr<QxexPyo>+K;t_uO#*Ja{
zpsvdkAE1Z)9vs9_ky4m|5qbw~1nRlRz*iUq!tq?}cJi4I8d_TUrY+of@50878-Ykx
z7vR0X*fCJWhlU1ze*TOzAKZr?Ry?L8JD>epV{(d`zNF@(T0tRA#fQ+PMg8eVVsy{$
zoL;}2+ab@M4a12u)WY*N2>t>bsF$$m<8>)S-rHY#^~-_s9xJqcb$<T-9D0*YEJwmV
zVGsoaX&M6(rVa>o7OpD<|Mhzrf)phVIYyD4n((W)_m82cxw#qP0SJXTJk86OSAm02
z7U4U<l;A|I^gHm|K?6h`16>aLTBQ?YfCHbGFjE7n`ih}p^RmH-SWLn`SzaF4^}9TC
zc(~fd<trwOza>(cdh)yfy%FLx@vah8&fU;{V46%$P9_(|<o5!kCXDA+bz9O>0Aas}
z*=Ru%lw?p)P}T9K?E<W<_Za3E7j1vHo|}dnaSiscHk*wy7h>b%I7h$p+AU;x2s2nu
zSur#B4ScQ@=JwF$rTDRv_}l_=FxZX(^fS{;3=DYR3uil=xJK}R!l^C2GOc?1_6$;4
zx^7Wr+1VJCqe4Ovj?b10?O>0%+rMxz`iP^&t!7H*-E+f%($;bpTtr<~$PcF{7g&EY
zauM|yB5|W$U(tXZxVs^YZzX`fPCMe*5060mgPp8|{TMF?^q<M!8(PB)i)EMSc{4e#
zLu!*#s`N+YrPxmDOxLMu9-f=wbzID;ZrSs~&STzk3)6ZGp7AX22G%9u#Rysc{>rH%
zP^(hY(<up1bon7Yao{as&!009hT#N&A8@%6rY(3{;k$ulwh(dvdOG|b-7F0V6bov=
zRy5}DWaF5ew9?hp1sxBY^g}q3<z55|!|QXY%vEyjATD^TJBVcXbz#;Ga(n)2gBHfW
z%F3kI42_I>@yZ&=5%^L~b#+Wn;#axQE`T7Kd%k&|Tv25n&FkXqD)ly-QU=Cx>$^SL
z@yy^v;<-13zxUOPbAvr6U>=jbR=|GUtd~;w9WSyl#JfEtn3TuaB5H|E4*Z>5_K#0O
zYVKl!@rC$*Sy@%voXE~9DLqHGj6Qk(I)Fb8*%K%5(j<V^hJd=A8CsM5bzGwzXKL&B
ze#20>>Flu2?ce#@+O?tr9__pSg@zj5*7ETV>_60<2QbgY$ti#0#J#syo<4m_!5|v&
zOxo%fsrUEo+qZ*}02JWbeHJmJrKRQQxCs3i%1zO=S~>#UO%c*=t_WIS2Eu-Hyt=xP
z#4*Arh;J;44LEwj_+S$UAzFnqZVk*8UOI3={pLFlGQ5Hb=>qS4!K;3Fxwy#4$?<wA
z2ue3%?K<lM7cEIha@+(-Tiu(zre^z|h70$6q%w8-x0NnbtwEP7V=DbT@-_QId%GAL
z+eyys6@WgHH}BuGiHT`rAi&KPwmA69_wCz<gu&*Xpu%~S&aW)?^1I9{Ye0C=t<<=M
zL`Ci4O8WH+<{Owtehm*}z{LuldE^fpIH=Ejw_Q3oH5~TyaM?!*C-#`_W?wQU&qZWN
zF~~N)@t|7IQ1aZM=h(N(eik~9L0zsbZU;&<;2eiCCA%U>e0kf8l$76i+_WVd-5)=0
zM9w1YhyPH@0tF1BBu1~v2oU;~2O=BV+9)X~;#4>>Bjx<SJ?^B_l-$=&U88rK$UciF
zoc3=t_;B_B>k-G(@@GXi{>f)?vI%Qb1MseX1xNHU-xG-g;ni>kzkn<B)?t`p6*gMX
zF*4Q~3XoiZoExk8#~cY|b_ImamhXx8<`blJj5(Jt<tR5x&vf^Fy0gaTDMNbZK8XY|
zi%>oGDJ&E_wm)b@R#W`;P>#We9Wg-=jNzWRGo8mTAdsB9-Ty;PeLdb~p`@Tt1jMGM
z_GEgc4KLo<NWgxBPsa2Hao!PWw9H}Z<sA{zxkOJ-@*+E)XJVTbU*-#UjFqP7<R)8v
zmk0<pughd<mim{j<KFqFp<O~p%4MT|{P<f~{Ra%;%ZS`yf5A3Q4rX?|ZODG|*BQKA
zLtQ<1+NG8Fx-!D}!qoMuq?wC{G)zpAva+mLzZZ5!t7%kk-D-f;6;8M1PpOmV250q)
z=v2nMWnkaI+nwQj92}ZG3y@<M7Zz@bG7tg+1Mxncyrd)(+%m=-?STm(S2Pdu5fMe)
zjVk?2I;$76-nAjpP8!LJmS$~MMK&gtfh^5CuKq1K7>813K6kDb&IQONz5V@mNxZZ?
zTwD(D@0O1(VU!qe5hBQ-s1%o!U|aLi_IH&y<=t7K#|!1y*=NVP){v&R4<{S2%yhpt
ze}PfJ*V|jYP9&bcO5;c@*eq;&#3D;AyJZP@sU0kJxvz=uu}Dk1Iq!neWD*$-4PgNR
zE0izU>}iIHu*SHPKIIrN$4mwtJ&ct*yqd=8b2$ViCb)O(s2&~N6>Argd%G%_C;r2u
ztJ9BE4Bq}rUU<*mxl5Pm3Gd3wfvQDr?L164xjuR$RDBLsu}uWvY>eW{hT2Pfhq6xh
z#SNSs3DJj$*L=ll0uNxg8rB}X+^G2teJ<s6+1jXtSxEK#gn)9Ck6HO>%kN9}_EouN
zua4gfx4uD|>qPs8b^Q)al9H2|B<usu#&2-PTjzp;RBdfvt3-f%17(n3P(b4T=#f%6
zUUX<mPC(DNu-0jWfCtrchR_u-I02~_TZF_NhBB3EN#O`^!|HgQ9O1&vC5sP<1qG2B
zssD?wHv!9Wf7^fWlw?k1h>Rs;87d@7#zN8{G$0|7h)QWfWJ)PXQmLduh9+~v%9J@3
zBAF^FLNYY(&z-gY!#?(YpW}Uh$66M7p8NTJui?DT^Sq1|$Ljrhq`1l9a((@(?b~l%
zavcBVOE_=BS@v?fzrnjOo4b3k|CD-UhLjj+Rt9+e$-aI26fPwsP3Dl^ElsWjWLQZ8
zeIYp{(Cl~Tv53*2xzWQQ9mD12DRe#LjLOPoIEi2uWNvIcLQReP698U}ufXghTRZHM
zla(FAJfA2H9h>CVrS}-!Wy^H%qe2=&xz$#X-Z|*l*31tlddLs&KXIUUsm#pANSVm1
z?zf)CneFfO<EHK488RCE{ypgl8<4p=J3G@>|3+=ky9Wz~hq&wlkuaO_`ke!Pg4G6%
zzfu4v<5PLL3xy9B5zu&>-5|a%AHz)eZt)+G6#fB5b>*gxgfJL`@R5p(i|>lBIoZtq
zFV9K&m}wPbYB~&_#077ixKvZizL7CDaelMQl2>V;z36AuJ?G!^R2asz9>S?eJfoB(
z-!mp38e9}Nw-=T(FjxQpa_`@l)TOzG132wq-}k(C#Ys-K_)8%6foPqUn!3A-;=Sc#
z+;P<tApOGqAOy<YyZij!vuB$4kHOt#q{^Al?*F);AQd_ZMI1m@=YHPdzCYi|+ItR@
zGY&{q!3oWw+*On>o;V>0yrY%l!XCP<D_G2op{%SdH+@1u&%`rlOcG7|2mD6Ec%*;F
z(<(qJJ@=a}TJ-Fx<g}mD;`7bS%p^|W5@6a`tZ4txP93@YQhojQ&vYDPE1PDF=-xd)
z_j%}%puy3dH`i7;>F7Kd^9E&BYSNjXdrhI2*O&ZP{|u__8a6xNaQcS0xevZPR?tfI
z$RFy`(kzyDvfKZS%X%_)j&DX(7Y}bO0gv~eGUD0U#7OnTqFK9S<$VA9O^ON{Dzn$#
z>2hU{%v+~hY43XXHeLV1Xo~X)ql>Z{dZVQ!x3*p|U#($@l{txtW}_^2i%+HVf4Mdx
zC%QFoj;ww)d-(8f&v(b4FiNvqc&nyNTRx<`UthgB{aR~^XqS=d>RS-oX;>RWDSq}$
ztXiNY5;I>skd*YPsIY^%FYE~jn4qb-p;=;J!KO`DoV=S8OTR8?s?}fLSL?>TqANBJ
z70#u*j6c|EkZH%iKlWgPs3<*I$~gh7b9HihiI{hI-EO1Ut@X8Qw`|$6di7BMSwM;5
zJ9kdUB4w$&k&K`_bm%OewCcsF{q8?H_uz5B$_*v!WA$Y8+qZt)gr@K*+g$jL>y{-K
zeWnP}Cc#jc_3QEx$sz><LxkQrvAY{MP5c5EMQy7S&=6UoDBU6a`gM<v{v9QzOtoHH
zwJ1zwfl}%U3yV<4mp2u(FN+WVboak+{nukORJpnOwH{cf#K(VxiUofv4dEuEhw=e^
z?H2&3Xre~q26Tu(=0sQX=<#En&E=hT>8(9<(b+CHw?j$ExKLGnyMI2F+_3qw^lVo@
zqmh0y2NLz858f{>ZiFww%uDbApqR4t+ljc@zAw}9*UPJj_)5~iE$ZQ<{xj;nN%U0P
zJ=Ifb##hTJH<GimVvhDcP}0r5>F<}Ul}b+pg(AFP6dLSq+g4v(%!$|Tf4*D8jHEu{
z4{^UZP^!tB+Yp&O&|u-Apr|O1sHq!khi#mwk^U|*F{Gis=M>MSWtAP&myQpR9c0?|
zZ`f9RePY_q){DtRZ`8<<-LKa3kYTIf{dco9WZ2OYkrO?Cl>6K1;7`DPp-4ENkg%BN
z6c)AwsxD0$ZekcBiGHPX#~)>=h7W|Dy}b)!j&b9RzZp}O!^T3GNY-?vtY49{n^4@|
zUrAGQ5kyK}#q-oY6+e9{r?bRv3#i%1Zp#7>Fj>p`>R}RVx@g4bSMeJhk}fz7Ff}WD
zA)fCm9&Rf<Jj!ay7Qa@E)|C2tMY0+C2tK!L7_}IDYXqJo5(3-t<41}HVk4*`Tw#r4
zqPvM_so_E_&{+HQq^OiUa-@wAoa^zUeE!z0e_)wI!B*}Xd-bY2DstXWtkJM-zpGIK
zk7zM=a!U61e+YT-!-o&dE&0*p6Ogm<ICyadeuNjjc#&}CjARzb;migcrRaIv`B3}4
zH~c)SSMT17*Y!nCzy<8I^V5mTmRIhF7Vhq}kPh48b%a4u<k;AXq7<iP%g$CGGz#CT
z|5>aar1(UG+rgMiM#6Z9D|Br9nl8c<QEwE_+%*4FErAwE9Z3Tr<&-Hm`1Ov?@|bLe
zgsfV<IvvFs&!3Qr!{#W21Gj8OFG+@gv)Ax^-S?fjwxXs|stzXI@hRIhTa2+zB7GxH
z70^LvKYO-l{`@ZxiqTdh0|TW5DWt{{bYt^i&~8M{!JjlWGizU_jb+5(ux9y2J3(U~
zG&)#Ig)isvC-n6b``pf#&<6r9ejMO0hOd-=!xyFxfjcQXgCx87r7-RijLgiYI9K-2
zIUU6eAtmJpka4C*T{1;2tYO@)vd__R0&c<(o<wvv4t-Qo5;-e2LdScmj*gCoMj8nu
zMsj&d5xIt2Bgm4cy_-F<&-`B<MW$Oi4oS&&d4sWxz{in!$%CU~V>#*qS6}>4c(6d$
zK8WV^LKLM^8(p6NoRX;pdywNyw9#DJQ9<3#XndzrvC)kFQ=QK|_;5$=<wof)qkjCn
zRliXt&H1u#Yf<t94^k&c#`C(6mFH+}!VJ!<){tU{1KLHPoDk1@fqaO;>noe%`DkzT
zvG8%CsvYF+sv0U9CLEgiH6tSfx%4-C=*!TGC9+<>_I&D{Vp~;J)lOR4cvz2~`<<y(
zoq`@b-g$RKJ=3e?Y0F=xt_{lQm0*osY24a$#;K?J>-*`=op0viYOXlJ(71NPhAtl;
zht8h8k`i*V&);<9lR4<Pw`h4PRABq~#CGDl&!6X|9}8Z+W7O!;Ge5pSzI^NTvHNAX
z(Cm2N4{vYOvJA_FX<N3a2j@!pfFSM3C`Lv3&p&O1NNqR0M;ua0J0XjS_mWXPGs{rK
z3O)vLU{f=TmmA$U)OEg+c+!@3-(Kl|EcE%>!VzgJa!OT~^?Z(E>%P`Tp}D@}jP$Dm
zeP0gy+ZanJz3V3=G=>-KQ}Ra_VS;0kgWAZEuW$zu?ml?X<LZoO7du89-<s;gcoJ)u
zgZ<v{HG(qhb-sJ@#*rgEy(2PZllxnbQJfN%phmtd4w*l7NbhmBdjN(nXP<C~NPe<#
zk>7*t>}WUX{0ur>`<>Fce!c}u2ORuT-kN$B{CjrM$8PWEfy}YgN9r4u^?PeQ)xACa
zaJ4y|pM*<RqG=@!oWu!i9bg;t21ZZyZazC)I<fa1n~Q%2On*^S1pLrP01BBU@~z@e
zo!ar|=MN4<1G?Kvbr7Qg<9rnMSGLC~?J(lqix)RV)H_;WvpQ$A<fa7EO1O*19JIf5
z4^AZ5Cw85E=+x9=@8F4peP2ra`?$DTL))Dks4WyMJsUpn(js&7juGeAxoLLTl}t#Z
z=L@P`-AmOHW_0xOojZ248Gh}(+EK`)%4aIToo1EDEYVR@QJJvm{fxX`;5C~;2B01`
z!ZfZ@->P)?&6{OJpyEPE38GEKq&xZN0_!*KY&X(v+uYM#C#*I89y=R&FWRNkzcHHR
zBIts8P*Rp2NK70ql$Mt6*{1K}qM5m5`k#gdDpVnhdz<UmhM(7~_}T_rTPcB-Q7xf!
z(UAp#g)f==!=3f>e*Tz-?AU_`9qWF5=__o!QjD+^69=a&#nKsx2lW>$_zHDfQo56@
z7FX^o(fY%O{K2vI?QG9mnWUs~3*z^-c-kHuu&hU43-`~xXFo9P?rYTGv#PW;b*qAn
z5%h#HSWbFrxnqWmPjPK+?Y(;jup+f??_lKocSk^UC=hy3tjLtx?h1!ieG(BoK0MrF
z<Hm24&b&!c2jLXC{`BK}_rOpz48Lx^Vma=i>g9H?N=mGWjn}U?(teqa2=OQ$xHK*D
zsFrgj5vTdjhpxgz&KP-JbjLRB-&H^7#fv}4!huW3wd9tN(9o}iXF@q^r%tV&+cjx^
z48f@Mqe}N)D$$*{W%+=FEY@j!B34#u1OEcHTB!sXa#Qe1#0|}Euy34xQU%7GCr&if
zlywyL7+%u6$zP&>Qlth$76}TrU8qY^Fj_J(CN;I7xY*LdVjIBGj*sa6X89|$74Q$X
zZMuH?bS0>w{dF~B8Tx1#j{514JL5oV4)3u@IAFgLv+$}atU=$w8&S2qEF1gV{X|^c
z@N3u392}rF$o=D%szvUf7k~d*)%(Vff=z!R24_ozdi-rT(MtOe{9e6$S);~R{iWe+
zW;-7$spzGf(BSf<+X;Z#UfX}q$^&X{?bzMiCg}`B?Rp?2L|)kH=0*XgdtplW!(TNu
zgXQlsz#dZ2H+N=KNQg6)YE4|JzO(esO_(1%Er<ULhyYpM;iT&`B^-Qe+qW^YnD+4D
zm)akX?bFkpLe+L2HSHl5Yn~1J`fDdp)+2h6wO{f#?Ay0gQ<F}Ndayc$>{UaV3T}{G
zv|hl(@-fG|?J~Xie7fw?_wS8OP1Tu*f@d$WurNt1pvyvF8FcN@ojbD)4bN@z(?S@4
zT_(rC8m9lPTdSa%_2{6Cl{a*9j4Qz!sj~3{^2x~XTD`iQe)O(^7v1Ml53ka@hd@DS
zxOffpYPx=c(txl0arlJusj0O@Wmuz~J9Zr6TpoUxzkoioSNHDIJ4j%K2<L>)?_Y4c
z`^nf?WZ;CQ$u!E`%C9fes`#T=->ewZ`{XpNvBa7Rt^q`Yn3%yB#SqcP_3nh?Qo+$`
z$r3lJE(9^?s42N9-pS&lt*<YxeU{*6v~R!s@2{ac?xvQz-GsM^rx$H?Uo^J&(G~rp
z5B}E*bOhz>+*#h$861!#4lJ+t5jgl7Ix^k`T{{maFi^HRj^|1`)S5I&<CYyHsI`uc
zkKb*4h8HT@v>`)Qu39DErw@lIj6pW<qEDY($9d6YW=oh6Ga7&fdylo5K|T>Gs;YuV
z{Z30xOQZ0rgc4g*Q-dG?f5_f)a)VW+H=Mvl3Cs2%d8wNu7wXDha&pOF+VUe875q8P
zo8f+S3~5g*6*Yd1A^EN4!QUo&8xS&DP9EUg1ndz?0l2^2Z`Vc_z>F8}9_Z;XLr{44
zMMKzstNu+(vnwbi9z3)_7mC&+V*_ggr9t_rtQ>bl(K<HP)GW7MyVl-Mm87Ynw(Q^D
z@6a+cU!yyJThW4U!$*$fYwwCeld?c=@GFS!c)kvvIC0|8p+hn4xbxvtMFshYPo<!~
zB)JMFqpR1hiP3EI=q~_Z`1A|Q*RAW@z58acS55~Cnf?K5?CqggEC7UPKfV-6nvzcx
z<E7f4UpP3iF)=tpg&jGv2<{8hdn&9zgP;E5Y}U3&Cqh&-UrkHXh%m&x&<%Iv;M`6!
zGEegJyIX09gytSSPDV#(FCT+G8e*M{K+whHLv6%}pwLjUG6baBu|p5E8^E8h2aI+S
zhHsRJUMj$==&O`o{*})r;V(pANAo_{zyO2~<O{pxzQ3Cc#4nBr2PdnCb{Jk)$B%pP
z^}O+J@dQ`y*mIO!be@K$s=4Y2gH88)w(nEEo$`F*MVtFQAB+i*jVV}KmgY$?UN=6|
zgt|B=<u_k3wE0Pn{tEjUK`>BzdsD{VqM)!a(R4(>9qc^T%?)T<7<}xQ0!u~sK0a+f
zQGX);lkK8R0105&nYT9r0g^@~vw$~wgjKaKxlYg`xtVQ*U3x_fw7~JD`cm4!u9Ol)
z_h~jMM4us^dj0uEoxOL!VW)}x@W!BH<)Sd{GuXIWXV9P(G%_em#;U6`F&pdsX6h0>
zwIM^sFVFJf=48uvr+2nMfFcC8{nhw`5e~v@yi^D)N|S!Ihtxx%V4PS$X{_HSJVErx
zYsKNwQT$*U#%+IU$8;(U!BqA?Uh_wQGcd&J*QK2Rv`mJGJAZx|`g)pnB-5wc3%VGf
zBu;&CD);8iVGb{6EHFIsWcR8y<{8c-Eo}Yv90qVv%u;!pe(sryF*pCywwyk#7)Y9h
z$ZQg@)w(+Ct8n7fDHjF^)SQ+X9$|Mrh(Zh6%I4-~C6gN`u#C9s(!X1`X)a&4(Q78P
zx^RfClg7HgHTUG!rr_T(K!YMWoVLOit=DAaZ`UceK?4Bb0OaaaaLW2`1Q6UD-ex{1
z-<UBQDTJ_rfujjs=}c9F{R#SxB<mZ;a^4LaInt3|#9wlUxqb&9>ii>z51(C@Swc(n
z^l5U~>=%n|Bd3}*mh}qzv$eg$;+@iF>q{Oar*BAKILLRY&I3Jha1j0L6Fw;?Gj6DV
zs<KxaIZ||WZ?^x!H$Vv;B<0n4hFpVdWr6;%sVc`KA_@x&>mEuGSNCq{Q}Hyeqd>3-
za$NR=PYWfo^uvc7o$iW?^H<x1j#@I<ok+0mX$v+nmA{TC9LTjLRBMKhR#EBIzrSja
z8<IkVp{uj=6r7gWR5C7%GSQ*Dn`Vcl_NSAtUAuR4FYvSRfOASCIxr6ml>J+07Mw%G
zh3CR3W#wH3msm=0##QJls4F<|jUi{Z?byV?30ws<aYD8x^-Ehh4*G(GSJW|*J}A_=
zF%|Q{yrDcw3wjM%5?SDtsZA_pn4<%~v=1VGt}8X%t)$RLckc=UwrZP}behw%9_XTA
zYuYEV)-U*iKtD3=PnI(ta@-+lA6nDYq$Z7m>(>o&K<nIhC{WyGO-rlS$yvHUHL`N;
zLZ65Lc7hdlIXbx+t~MC1fsl9p9GtzJoJG-aaq(GlH!e818lD}b7$lUCo-HNGa>(KW
zSMR9(`EzF9@MFh<3TpWrP%(Ll^CQvm4<KY-xaRzgxl=y<Abf$I$j!QIa}oXN%yr$>
zBiIWe(4#A_?Wj!AXTNAsPvPdRTX^s+{T8IYs?O5Nit?R-2Q&Q1*;;!DSVIPM7Ic9i
z-u{jU$|ViajB^rgCG{f2RO;O@zd1lhsnE~F<jmrX>-KqrtyhMwo*|Q_@^>5+?XnIV
z2_Djh!{><3ZuHPRb779Yei%kv7ys;AmUdd>lB)kKNg=PS-pM9LbWRrK=~~YBii)UD
z`U*$ql_M*i00A$js}9Y%-<xFCI^CP5Y4Ki5`19xc5d98s;o{y_%ceOI=cLTHqAQvR
z`z%Sx&dz2cB;Wet(}>=o&;y2DE+Klo6O62^ni}giXNLy+?(%)oWrz5D2ZR|N+5U2D
zSk0h8_OuoKCJxke4-HAOOilR4hMQIQ$GFdC5H`MbTx4$!cVOLy4UxgY`Ij7*WxGA!
zjZ4Hed{&V=0z=R}!vU{+9m22(pNE5g?yXRIeU&#<>q)mytB=pbcOek2A=7Ab0M7|e
z`Bhj+V!5F-h8c~)ym=Uzw{WE^r!>#{4GrAs39+~S=RnUT8OSyRLq#*nyMcom`%CTo
z#ikeRKeN?xrzy7C+GX}sZOiY^mnFKn*}QsUXO^<s?5x#&`Mcxx$m+?x?bJHSdnpfj
zf~tuMPIE)8VJ&73<SK|&yN;c@6Sv03rj0=TLMazxPB$HPP;>nF;8T}cQ&&IZ=MiHx
zHF?IO6)x#pH~~#>=+NDkgLl$*$X)8-gQ#y}KjXVeMS=2P2d)nGPQuZMuSG%^9E|Z3
zXe)*fqEU0FPLQC-4bKxNi9x-xuy*hmSHrG=Y{R>qwDkwvZPDF6P{M;lxvu!1fd(MC
zg(VsJk9FqmDOuv;a?ADWkuR_J=S|j|<Noh(sBn99^gCFaQYmMxFi1NW5g1B`$(b^!
zwP#TapsPS4l6YEM4hBs%`R?5y_uNO14)(+ED`+PWY#Z7Sk%>3+6|{LAJcwV8FH`kT
zyFQ)sr|i(JxM#s<pCx+;y7y`h1=|juGH6hA<(pMsC>fM)45-t@VD{UkIsB{)uO<bl
zQ`O2RRFo1XYMq`JJ~aC3WV)oFmZtMcgJVhxUw6|}?D_9N#+jC?XSZ(IQ6o!uCVUO+
zUO!W{qk5u1m?EIPYhSnz5drI6CJTUAm#<zm$hY83OT2{-dUD|P6GRE8td>#B!Ikh8
z5#E-S{e)kVeS5v;tNZuasUs`gb*f#R6ZK6cyrt^&xm9&@gvYky>ek$?y%rf+g4c$?
zH&&wt(fUBA1vT|<OP3xa{#IVzG_K{`rCzt{nj4rl?>ld`vgx;pG(%vy>c51X)8l8(
zyanjeZ;Jb2k3Lw@8yhFPezm6FU2ARlKQ|r86W`EuSl-hb1ObO}Diap^17{|A^nC>d
z+>Q@?jdSX+qF)UjH~?FA16|!LU{|Q)*lbclPQ>Tj(^G5&=t=_J)=TfqOL0R_)lfaz
ziHoKGOr~HjkqvC<AGP}vsMyF^K_5{mtoZg-`VT)A?h@}*9DbS<2mky*^P<!EZMwf}
zyjNqw<!zT5i&C;y@9#Uve0IFe&&m;N)|~k8VS-*l^EWjaaTIRk8F=rThv@GEHcGRo
zf2w{dPFpUdV=nXRabaw;;ng@7Ed|GXi<483lM-fai!a*TVjatEn~N}cl~YG$BC7z}
zmOEaQ&f>NM2Wb4oUI?A&fdf4}O#T|97?)hV$exB>{q`(+bS95<hWeFO9#hch>Q!}A
zu+m-oH}aJ?Yi=DUiFN5f0hnnHLEgKscKf#woA{KwuEre>8QxFp<OT1{ll2=<pO%k|
zH0vb2z}ouQrQKp+kZT<vM@wTKJtV@goe#`-P}z5kD~&TY5}ELiVvF26e6|7Q6Hc61
zZf`G|)=<059dM&N;|=-~9HrbjnG^{jhY#~PL{7DwoD-DyWFN~wC??qVW2=dc!Qqe)
z7z72>o3x`aZPdn$2{dr1o;^cul4KT7(B8q)$OwOA{)#q1F<mRojs%?bTQG0lS;nPm
zYAV-?^%b`U9c$*QGM^yI)3!+Eb^dm8YO#;RzH2V}30=+0Dwo!l>I2==<NGSKee`#J
zT<FKp7!J>*>gVU9z)qGd=`+0B0Csi--RRDA@V(Yp4GrbVCF{(pK8jKbghqktR^3EU
zzxDMr^W>+aVPT)Dst`ZE1NY%qgG`FP@4x!<Q@&rng4j_}H*PfHHp;F-us{m(@P~`c
z%#=eq&QZ|T3|Aa5U_rDIoJW}rJ|CTMnd;$Px9dp6XqP>v_CIg#l@Fh=`KDS2U72Y2
ztiP40#;MB6O5!d+)TLV8V}*1`Ec4U7hkj!Xh>7(u4aL*)xjKkVdHkcVDC>lYy8qm1
zQ|{H!eO6Ue2)fu#EYZ<nS3>e=y=BXy&ffN>Ft59i#npN7qIYvr*Z9xXs)O8Be!epZ
zP|T@N98^&87j`uI0Oz(D$Y%b$d4RImU7^3oz!J{O?SjjB{u64Vft#<xo&hzv=Jo?z
z==A7KuCC@LCXzmQDpAkvUFQOgo7T0B&!I!2uS;%jeMp&ThT0mvcJzJQK6zi{fUH-q
zUgdX}Gd+IX^wj*=mZnRWFJ4SK+nm^S%?fOYX30cf{g*Z6znm@=0S(e*i6xGm1n4&0
zuFrgx>er6FYRAZj^AR-T21E}qMAG{DPt(%+kN{p{w6jZ(r8RH^r+EAsW1*DTKz~e?
zs&|mO?!n^+=9$SJwier%vEbG($P5irIa7fr7N%MfLKp9*IIRuGA7u9V+RE&zcNR-J
zd*j1eg`*s<5$#_-_-aZuHf@@OKo&tdA1)Fvhm9C<k4=+KmlH-Co14cQCo)L;Fi%L=
zft)a^=NZ)0aqV{4cSBN(^WY|PjHT$i%O=m0JHKyV2Q1|2^Mch#TGbyu3^|;fk#r)D
zVV6BUVB_OeqL*#MpGn^sd(R5!?_gs?ixeClF77MzQgD;r*-ch<;Gch>QBl1KJ^;i1
ztLiW;K-JJAJCs58PwLkDWo>+pAGe$}OIqhKh9w8%;*=*HX*W#DNOt_&Mb-nlM9q06
z;W5~u%HDIZ@~iVFeI-0hvUNhg2+dd4SB>#;Y{&n12eFEaTZPH1uO{VpC+XOt-39M)
zP+~SqXq%sLTw~oLny}&BmO%)MiUQ(|kBcj;k9FT9p&fc(TqXf$j7(=`X2yP_HFSX0
z)%?W5Z{c82qKVl1<}LSga~%lWAbc^Q%-O^dV*QexzR9AQ<hJI<NkStfBG0KjD*kY&
z#m}lxGs{lGN{MsN)#W0CjLR12q}6o#0lBry+wN;Eo?n*GR*i;;SzwA)96BaHzdH;k
zxebI|26^R=qcXvaY4sJB*|5PaG4TqkAlBBm{Q43OG7kLdf6aPfN*%zt((JFo#wK_F
zZbsK;#^-|xvTbO3MyHM)H@moGZ2BOb+im%(RfH%t#&_65rT;Oq27)8((OH!cW_cFC
zimjZx3cmqfq_<f)4a~Jgks}M2wBV<+YLNQ{g$tX0d?VerZBs#a)ceNt8~19yZ@%Bs
zBK^m^DyOrW>QI&2`!Ws`_-$*xZu8~kvwp<9mr?0sCB2EqVn$lM`d)5s*wLf-<n8ZK
z7uf$%R;+v0<H@hIj&FBdLWN~<&BcM=<ZrH_uKtWZo46bOgE=hW73f0wP(W#)KRlL0
z$DeYWBN6niJlMbR7Qi)5^~#s>o<AozuaEqmI)d)hD$PEm!g|A02Zt~8B4`IEd`vj0
zq9w_;43<7%#bLa18y|^5^`XN~4qX$wUUD8edgR>o;pvP1_n}j8mCn$pF6@R+mAJBo
z3Ro~&v<N3+kMHkivOED_J^hkybW}-3c6Jpwo=@!8MXZkwC}<cgTb7B708kt$c=F^1
zP#Hc$90<tc5%&2nZ`$h4-Q?o3XyL*osdk4hPbmc<a&mDgK^Y|4@>pA+i;f<PSnXLq
z0rQ}RmbG1kR`K4Yh2`(U!|>dDdSrC%^YhB7D~gIj9E<*`E<7X7`t|gdZW9Ld?D>ng
zI(l>xX*ziA5mF_cG=0o(Mz)4gW&?d??fV08THJ8g@XJ}aXp!jh2Jk0!9{K_0AC}%}
zCr{dP?wO)C&RgglhrQtO<MX;GtSC;%;Dp*cOt-glokv<G%)&}YxhxHe0{ybPwg|Ey
zp+A&bne>norZj=UHj}FL({(o0@9$-R!)w<826j2=siV;4XxPq4@$WFm*S_HN?<@v^
z&=Epr#o{STKN8Zq{{hy`py|6!&CP?8k5ln1l^+_MyL`-x_34BR45W;BEWj9rg+B@F
zSYqbdLIu54@;k<8m~B^&JNk|#TBO9;=$4(r5#(q*UcNM%i9WUax>`_b&@}FnW?rCz
z-mw@Thq#RcM^CYW;pqH%vB+>-_gt(w9gDNXe58<CDKz83K^cQpGgH4EAA5~nTFVU-
zSAP~{fnx04Ootitcs)^H?uZ1V5O??l!bU)WQ-<S~Q5Kt-6%MmviV^H7G2i!=jhU%w
z>J|A<rX!qjtC3Ua-M4SY!Z+1M$x7=JdY-><VI%C7q;KBRYrVn&(85<Jb)ME#o0xyv
z>_0@_ah$ouwNb+ohH0j|KUK2(`Lm6zp5lx%UEXq*ItZoh7X$*eUAf|xG~gFqzr7dz
zsMxAgbPm5#MoI<m;iBD4N<i?2ofR&7Sj}F#c(LwNd=$pPQhmpSJy1w_5^Va5XXXv!
zv2*86nto7e@+iZPKcnFb^Hph3*#Z#od0FkZ@WPHBn{bB0;R;_&0Vy8n+E`6Z9vaqJ
z*J5Ko0LtHL`-3JEq+=7^Jx1Z0n&}rWBBOJGs%f{a*$AEiUd61XbFK06>(nQi)Z~@B
z&W1C4EUlY?u^A{kMrIIfgCp5&fsLs03YXM#U+P46ejK)eB<UP6RcUXoUof5BkX)(4
zr?Il9Vq;C#^|i8ky}Gx4lhM8V^1r^AG`wq97bq(mY#A7T(iEme-iJG8s;WsTS10Qw
z<k%N_2jSTz2w)j}a*!PQXLW^#1qEBcT=X{AcrcMTc{E-Yl$v02O0n8I&raQG;E*Bt
z^=GcfDDW>Gb`55g6ojnKJ9KWAW7qim*B1-n=(tq|NewaVSGoMpTXlueKx}~gjpNx_
zzzb0}g6<exHjK7uMX1)B7%o__j7)^Pa9wgt%zMH)c4_?LZ^+{tH@-c;fRNP46pLyP
ztYYBA;Cb=+5<Ys-xTA?gKJN^kF=f1}YBiK}SUA*kjdMW*A>g88+%aU(PF&m4lajuG
z7|a>FaT;A)Ufx(Z6CU7Yt(+ATaeXs8C7K(`TJ8_t4))bz@twbsaaW-CyuFYLcp*)M
z{|H_yb0huSmtCS4Hw7x^MMM-nT|IK-84wBu+Y3fkx0g!g_0+H(*+6u*(2Q0LF=#8$
z3poh_9m>|anmFH3;#Q~r+LyzkuAThH_MY$Vdvy&a9Xi}hKNPa+VVKF64>|b->pdM$
zO#K%@{S|%DYfsYE1I2_gYGv>w=%#1RoT+2UR~bdtIKYa$P^F<m#eF5UjWk89EjzaH
zEOA9(nDX)S=S~8Ka*Ho*u$tmdm}eaeJT411%Ak#nkB<)tF{3d@_Y1W_&|R=VjU|0I
z<>)hPt$$MzGsNs!={qYHX(SKj0b2}BOn&g^>Bfe%9IN>BsjR#lk0w7`nM?|f<;w?#
zKRgl=0v=zwU`&7Qts7Y@ME}Rk2xcGv4Sm4Po28(l^mi(g+p-PH>vy$Pa}p)?{0?tP
zqJcIA%*8OwV(MNR>(2;~z_Y|Ifr!R)<k2_aT9n%`1=I<)etQAdW>Ha5X{;|O8$i#@
z+Hinc80HVk#BH_pv~D#A#33*dG&p%WcxnH!HolO8^z{DWGt=`*h!ygiH&Ji&tzONg
zl?@9GrF1wE8@p3NQeb@!=VX(zymD~!tCICpF4XC&D?aXzidvR5;J5k&)n0njv!C};
zdgizD%H|bSivkA^AHtlrxAjPN7?qNrP{A`)%4})A#tA1`iT|{7XAYkZNP5Pyn)kOl
zo;lW|SFZ^zf4b=<#J+pX1k&&Mc|UsCL;V>eF?kuhbuGw$=~HTryM3BZR)bBf`)msv
zVaG=bqq>$xQcFkcnx8+FKZ-IF9XF<*UJZEJvw5{S4z_1<YJUNDzPgWe8^Q_lHFeL9
zCSaJLwJ+dt6;3RGTGZjjGOz77DD@2f9Q16e4=Faau$Zi-7U12Cep^~#vm!H|>|gZy
z^loj$L*49}-2cnM2na{%UXRx@sj#-T=1N$qUW$8jv*j9By+TjU24r(b=I|uFE$ZjJ
zz5Dx|Y?*biRC#E6tu7zeCCR=0l)4us(A4+_&bH)X>Bgc){?@&FC&3n^(%oeZH!zgt
z2zy;_A-b9rMJL}WYz&z8{A1r`G!(TWz(Xe2BtK3!WIEB<XKmA`!!N~M96^$biV?d#
zUJdVYkMABC$yV+D_<^<+$*G<=V2FSzqZBEj`TeM#x|gm6{l^8^uk1eA#3Uah_4bjj
z8#gXavzHe{moiG7Ue}rO+9uDL%0La9>r&Ubv`C#0YTRcvMoaZkNUGuzDUoGVSVe{x
zAf#an3vGUX&(T3_jrbfjE0*JmMjgb_a7zN(2t*;MKE6E#X=R?)S-=MF{j=qpuM`6y
zve?dg)-24<$MbYg`q9-;&MOLdyx|Wz>_yV4Q|07t7EUY~^^BecC9YM#FXI^FMB6I^
z{c^g!JpLw+1z&F4wlT}_H@_1SKwRgsVqO)Hk<rW<jAz3f5O=xdbJqPq0wk_nu~cm-
z#Ro)Oq=A}0=-eJ76BCPzHoKs8zjZXYA9MT8{Xg}_1GEF5>|W>au;%y2moM80sdv0b
z{huQraJx_2))h)HLGvJm06Oh*X#&N8EX>{ut6F7pxW=jZ^$gUQ_n34}9m?#q^zkYJ
zMLTc%wby8)*1~gaiQ1`f==kw|^q1ea{+YgBv++(xn0J|uliNGu(Xiac#@O1Lh{p5R
zH-{a5lJSG`q@0$jrB+s>w@!YVOZ##9%$c^X%b~UhNs_vw=hEltMeNLKC)G(B-du;K
zu|%|R&=pEu5)J)?{>fuUkIp%_FQbVz>(Ir$T?#ShxV|j=SL1IB@P_K}K6p(6ya=MD
zEN++g9bG_}$GxO5+?SD#hKl-<XmrruU)j9srL{q0u~g##o7w3$w>m)Qe&f|JT2=L+
zhh{Av<oHC--IMs|)dWE-WdE|oGnvy8A()!*37>R2r0qWo;n(@or{8-=&a8#jMHV22
zQv?GdBM6080}?A=|ImJTs_WMtmmGqxIn8XH#Bces^x}%?X*(s2hCT6WX}PKL)J352
zyt<VCSs<K`i{rcwJ$4>FNk@So%>~(TZn{kLvkxp-8q$u+fRZ=KRfpasMx}02>d6_+
z7iDj@s7$8^FpLnH?YDo39tE8{mwn$#Tp&1izk2V(_nIF+fb3yVL7x~PFRy&Ag{Eg;
z<k_E1qEQ|mw~`{2vO}5opilMm*rIa9!-2XIvCP8-YRjJgO7Ozqn-`~%#n2E`Am2EZ
z-tD&O)h+&oZk!U2J4RG(Qkyz8az+iAjPVX7&4qL4s#&214L{t?&>8~X!lX|l!u$NJ
zGT$}buSb^w(%-X3{J!>eg!#F%TQ=u(`hg5H-mImll0ZYN9+oU?YGfn`oJ|qfGCD)(
z>E$%gv}N9L7!#W(_H%Qq!{e|+q+_LCv~$xapR+G|-Rf$Eq_IxI1K&vsZ$QzY$Vk<)
zeY5i(E7-$HWj*;ghU^<QWUP7IKl)5*Lkp@Is>t>c2wk5(6?wH+)6ZAg9YWYOskvn#
z`Sw(v9YZ1crRFf<w^ckD6yzOQu<q`|hw}SM@g$&2=VdWn>5YO>hNlS|Gv3|pW-xQt
zWzFks+L|(DWG}Vy!c`d^X)6L2DQ4j>Sxvg8vF`_m=50>TC8{q1V@6RO9Q%jMad*4A
zNg9_GtXLQ$F8vg&kGZc+lF$k0rDLQ6*BNi+B|Sz{`LUIjsK__j1@lLsbR<Cu7o)C9
zTM^9o(=-sAOA&9iayK6sZSc(dkaM21Lx4ga<Y`<2_|{rGKFealIq8PH-NCFc9@9B<
ztVazsnL29TraN6u->ly#4pr8_$ZB%k+<`<oGfB$_is1MO6R-j;0=Q-+1SY8uY%Wr+
zkQHh4ClJw&^dHE$wFS{4PB1N$EleuW`P@pnGgOq|O`~-<wrb1I(1?hK=SQpTIjh}H
zAg>FemneXuaDwqz37a_ly7oGdEp_IhhE%{E3Xdi3?pnL_;3UZEPg8?SJ#6gQi{35G
zOrV2KJ-(Vg+kY;MgbQ6_Y<uUnv<n^COFr(h^T_!aY;h7sYC616O)hMffvD|lG5tj&
z5>f6eYz*i^@2jh)LI?sd%VkTTMf!nEqwWCTRb!cF3(LoarJWl2vDY(&$u8?n0#0vd
z0Jr3Wo7X)pxtY`kk|N?{Yg-Byxofpi-;pgpn5mNYr3Ln1GB4QT?{N22J(cXHNso!;
zEXC-Wc<#%>b-IJ=sIIYAX)7fqbGuS02pYN#@-#Lz`~hgcx;oRbyY1KFg=v?}U0g$D
zZ;YRPa8!ot`Pvs_A+hUynZB;;TZp6FdYui$6FYbIhbO=Sqix$7h(E82iqdMI&?^?g
zYK^en0+}QeDC6>KIxLm(PswkX7dd2nFT79fGV{I-RFdK9>OBNpqZ3b_lxist!(X$<
z>-<L_zSAP)eLGZ4k3{DnEO&HNTRJx+z*ZJu`#K)`6~tm=bSvN=;=ZBtQ|!&QI6F^X
zs<dFiJX4mO{W!Y>y%Qvw*O}hAxAwnz6tR-skElasqK6i~l-w#9(T}sf<8QsoQNBP_
z&bwYKcHw)exxT!~jM9r{B>=Vjv)=WQ2rua-O1DyNC3)Z3#`KX;(e}-p6)Ohq%pO>l
zj2=|iFlbjTNgYn(4z)Go!Z=7uK8|;lTu`i_alT>+=JAftt$gx*-~WTg_Q$c?`v0ww
zV&0dyY+Herv+tb9U)2%G;o+4>$Bw&r@%_t<4;~mSI6f@HEx52!r{4?!Tl>OoqVzaz
zCP#>cXy1SQV8#R8f%L)gN`**w;J~~M8>*Rb{+U!sGlpj4{=r2GRIdyLpw;xBu;Dnp
zF2_OD`ud-DIqihrtOfAyn1!u26Vykx<ERUgDd*Y$z)YVO2<nJfi?s3_G2qbweSP%k
z+&PMrd8b&DfD{Y9Ak0#LTKFq=;}Jd6dzcJ6?BnADLP45ik4=7F-X=zb_4W0n7Di1N
zcZ?QTwHw*;Xx_85DfVS^lpZg~k++7wQ|W8)rS$Z3>#TP7oHlxt>BOxUe${W(X)^uv
zP9iz^;UrBOf6dm(q`e1q8K=jRqs(d3;RaFMMN6akWP#~o+ROc&&YxpV9J+7*7f0W>
zb;Rjv9Qa$x81uX?F6JXg9y?a~?p?MH69nA0G3yD-w!W}@I~yHl%$N~z+sN2B==kwY
zf~D$nw2_u6F!ZH&N(f*d&|yf|<RAWl*NzVqz<eRyJyuw~iq(uxn^{A5O;LRiXJ5c8
zvLrGq%w>3C?oa!pT5Wc?#;I=QogTld;|?D|pJ!rx+R63&x0?6Pl}al_jTGTe(uY9~
zguejq$R0+J6|6Emu9OA?v?jOC6AF|7vcRepd-kLTyuow8MF^%MbRJMI%!SqyDis{!
zkLDBd0W@lHt0h}k0d<l0`OzX9t0JfdK*1TdA&3f1IRk)aPoFAiZ5g&RoBd`J#IK63
z)_l>T%BrdfITuLKD{XAJLfF93(?9(19St+Fn9=Ww6)W<-{YvP(_WK-M!HTHr>Fsxj
zBphDoGN`K@)~>xm?*ZxXyN=q*i=~ev&csbUeqdhV=OX#4=bJS&`no>g^L1GGvZ71E
zRk2R|%h57E+RD-TzV3$6x>Kzas|7AzG3j%FzzyPRwqB?3O0QjPGzCNh(g!rzo>v;U
z>@Wzs3Os#;Eu+Wn*7P@wKXdwY8(j*C#6MR|3I@*@;x;vV4#oB?){dM!35-?PU=9Cp
zle4pOz;w<eQfkUNYDbbK#}`<4q?#H&EB>O056DLB6`jn_4Xdxq8#n&eues5tlP8+~
zsdqZyAJ<R6-SO>r{*U6TN<D7-U$%pu;=#k4Yhc}x3LNgr9bhsTSb?~E>(;Gk7=t{{
z=Ac`wq16y&@;UGL2KMZcc~xm2mt^s&CE{GKf09#1zPTR8Hy9xu%O<(>m;#OVZHWzX
z|BmgQErW*CLH34fcCzG56$sS-Y*mmxY4iIG&)_F2D2!3C97%GpdNW*LW$k>+wE?pB
z_h0HhUDzsRD=0XbsP`O0XX9N5@0S_me}t~UV-l9rUTDnv>M<U9M{H7OAt2Lpu6+LP
zb<2(1-|BvU81C#}a>a7Pc3&$(Zh^~x3gqUf<w}`wLFt(a)Pg>4oCvxsz`6g-c;;N6
zv<pVLj~6(6dpml7j=Qi`u-%{f?(`XPffz1B5)FvY`$7io4BP<W1Bd%=UyL9TnwX5&
z(ozu4Ez8`%o{RJ6y#cqy<O=A;i%R8m)uowGM~fj?aeYWI+)Ib`>)FJ_ZVZ}Dg{Py*
zXdV8Ni!_K{FfV0<Cw8f*67@#wvRgSPXXW2bC$R4?!;;~8tSeUe`gerxb$o%gXZ_)m
zubnn;KC|haR$Po8yq-Tl-$_e!l3P2yGRLYie@)TLv>p1_$>1EQYmiQ<>iJkPx4--O
z`kuAEadlIE@022O53A^(ze9zhnis9v6x+LjIxUSRu<*Nmg?+N)?_AMd6Z(s{oXwDe
zi$H~L9HXsUOUuiHJZp9w(k@bQT(!!XR)u@eeCDq;LB&3JVAU{MqICA(sLK5LZiCX$
zpdkM(TBPVtfSX&inwaHrpXUu7k6-zM1&;i>-g!d<0|Lj>Cr?;cL`O##u-?~EDvRQH
zPVEL7DQuqX?XkN6Zxso1`S}A#Okh0VXzKTF$m-4M@xR33VsqSz3|_QTiV6(;_Wi?b
zrVCylz%86~M_g*!i8;aV_#578zckzYeD}ZUE+XL_Vfo2i+F8O6aC8HY37)`2DoZaH
zK_C=eOi%C9zI`_X_cyxC%i9Sk0AQyu({8S#Sz|-f8e7|&w{P2^A-C81UjpRLdrPUO
zjvPJ8E>|oX?h?B?!d3iM_om3%OJ;QU?y*`<CzG#D{?gRoA?!Qa@?eYe+yKQPg|G2y
zFUj!iD1d|DGjM^GVJztuY}c$Iye_iy{mWDI?|b8a@GEHjGWGrws1|njHngsli)enl
zdG)GQarW+4Q3XEO+3sm=Zug^ey;&|OY^f|*3*9YdG|vFOsP*E-e`pb-rO8d5SxP&E
z=3BHVFo(hZ^zx<YGJ_WW8yTW+$DccA*yY}{Boi$Cn#0dq_j#L@c}H7AgT33M)>W;0
zeqQH~V<3$1Z31<*DAKvHi_(+oS8Q_FhP*}JZj!Fa+wWVce5ij&8B~YJtL-Z%5B8P)
zPZRyUPfNI4b79EHXM@e<C62V&SXekRR5hvQ+rX6S;n4niXZ2N3*mrF|MH1*-$^Op?
z&!0bctmrE{L_>qQefI^C+UP)Nf9+$g4fHkg|9jA7_5Al1`1Qpe0k5^oZQrUs-5sB>
zIBnw!rVHAuRyFJAbuLZoxqXOPf11c$j~H_nx)mEw{M*nZ{#O0}ZfNZGh4an_{P)%m
z$rSjdw5ezl*mYyKw?lloWOinrwm}hz{kJ$q#ehsbE0H^^Bf?NonO)qOu2(;0mS<lL
zQ6BOC>MI(LzU78L93$;()Xr~7MXPm;(a!H%oqDOAE_$(LY1%$h*LcZM*vR$Fa(}v8
zFEMglT*dbfP#mKx6b07$&QN(O!gm9a1|}yP*GA%AQJT=QAnxkW?*02CLQ77j{($Kt
zf2>p`zlPv?^1mn5d9Zk7?GKOJIjzBQ+?6=tsvyX=kzH<M+(XgIy;LUtRIL;04=#Dk
zM{FxZt<z^wO=6Cb{+#OHYyeYlv?{0s91)g7Z&$AQBpGmL8_{(90f|8c+5b5(RPH?z
z6!W?I+&Bk?rn0$tXY7SREmJqtU-#D8ls`22y_L9rn^238UC<>wS5(4;7WXa+Y<C>2
zT=s{JZ5-e(0+0s%r2o%@N&`Fo<dFgaw0+E#!Qx0R>w~toUV{F+{(tq<rfX8qj=JUW
zs#Mil>bqO{#M`$eCv4C*h;B^Xg7e>j{rel<`3+znLH+Nq=cA)pS1InYXUa=GD(Q3S
z(xtt7bLmdoZiB&%pN`X}O~Xfso0w~W+SxdtXlhTrh=t`6C!|ODGlp0BG6Agw%VPb~
zy~D%lxA^Lzvm5B^PhzJ4^*7ucXbHSynDY$054R2aL0`A>KW~RZiMR_!-``HF%kdj7
zT5|08`L3VDmY=mW=>IRnfQ$djFkpy!%KQWWBmK$#`UI8S|KC#f_3=rY1=vJwp%r#(
zxl<Bx$0RJhtC0C<Q$LILafL?=){ajc5}i74RnV##(GTavnFSC15*L)P=b3BFxEXy?
zO}aZ61g)|^w%c~?!a;+I9@jT~h@6x;e$AmFb1!E^{r=%~u;EG+|F?Pfqo2k1<-6x=
z-O!V99w7!9t;-<R^i*a~7mf&d37L}#wWQ=ufK?b0(-U@JCPAOTv}KC6_R`Do75kNx
zlqOD|EU61ij{j!i+Aex^&%S*d;Hd7IG9xw*dOb1%J8ZHlD@RQ1*@Wdy^0jal`0^cQ
zUhjLj(ta0Z^g%`B6#A#dGTQ25j7yn8WRP5DV792Ufz!~}{G4SjoxG|lL^9AQaWu%)
zNl!n5w~`O|6za2}=R1xtS3xokdXH{{Q%PZoc;sk!__Zrn@Zj>Q^VBFIX=gV5dd2Ud
zdLZ)AXXZn7C@gfoxu=@|!Ae11-ps<{m6LeeBSepgXhZ+|W#TS)Bro(ynXFzQCVYGt
z{`^@4%(vkeiasM_;l-?pp%ZWf=*_u0wIP863~dzK*S(V&5u)V)eKG%(hDXHA{WCV-
zmJ})~D}&^}M&YzE?2~MSp(uwO?dUN`VEc&K)6JN9^7?vviyVNDXhUGMhI>91DPZ(J
zjLS&Q6dY{&4LtPAF{5rZamyiZns4($V`C=-KbfpMVpc;<%^0GxDCHy899-E3(K6!R
zy$*$U+y+PNBV^FyEY}ACf{=?8hj}oqB}PA6TADrim(|yl0#vqrw+W4yyvq7|;9@_$
z#7oJii>`*`XL_et^}<3+MkJs|v}K8x`h^rqVU&ngSDbd;vib7jqGT`CnDl?%ZS?T;
zMB7LK{%03eJD?L)46crJlOR!OPS)iZ#-yW^9_ljgI>dj@1jQ1dGB8k6V`J;sgSx$R
zX1=OQ_)!^baaP-^8~JH$viX_r@cGb?aF_Pfwf=shgjOQL@b~ZEG?ZHC$+5lRBsyEe
z&6tW`6O!#Me5q7A=&UYrBu|`pOP9|hdUWp|JDXq9b<fcn>H7!G2nhn^<IUA6IG)Y!
zFuVsAKkoSohuwb7+tlKPXuNMiPTuqLC09f0t}OX3B61zY%p=8G^V3?O@*E~AYCKAZ
zdY66q#smPXi;A8mNyy_bE*?c6(qY%rdUMCNB3CRAzNs}*58B1H0UKVXehOXYRdi+Z
z8ZH=Y4jfZ{AFQ8xVrYIrfw*(t#pM^@5N%xe@g@HC<^DFhJ{~iq<`}MhR9HalJGG%i
z`-E}$OV*$38RukPk-;c~qNzrY_GtYsZUo>6k_5~&a&F$VCF4>6*sP;kM4P_;_JKaI
z_W75{F)MalyVd~n;-<(gK^Y{D%yBGy`V_=;^QKKKqUw2iR{xPG?L}xX^$?FZC|1vW
zt?h@-PxA5>nww`bfPlCf5>n_RAt{Pe($YXC<P;R{Wr}MW*&;{L(V4QG5lN_Wz)RjF
z+cb*sR6)?nxAfu#JHCF}ya5kZq*KHGgvuYk+Cvjec<+q6QMjlnEIv3A9m;KhB`d2N
z!1t7r-L(sj24DuUt<*mx^}-HyYTCQO$2(j*L^zvi?Ynz-hnZgm#Pf)E$|@>k1iIKe
zyrIl*P6^Wod*7MZ&k*vX!2GRdz?haxW}+{X7*cF*%AkdTN=H2RcyRyB;u}{R5*Bt?
zv7!h)#)g3YW>crG8a~{)W6$Na2VviIn$5MS7@(wp%S1TuydsZ2AxI-`1{e-V6eTQ3
z36>0Mi-8DWX<VWAc^8UZ8n2l%W?a5@O=MVKaKgof3HAN^_v@6Y(|7?=Ff}J%wFFJJ
zzkQn|OIYMS>3hM+5}jSa22j}WEF9LXK?pLvOX0I;Y$gL?XOYfIo~hoKn|JOI3=cio
zahD%_|2}h*S43D6{0ucKbMrM04o%Drg9e1nE2k?0OYovfW%e<@;$bgi)Hl3Uz9uYq
zSQX4J*yiY2Ui0WAxZk{a-XES~dS4jdYmOe>SrA1hWgp(zCM7fR!fQskFbwXSt*w%9
z^3*9P44&`gEoPgU$;;<lUxblApGQI$tcr%Cvw8Cy5L=dhJjz@MRYwhHA%)GK7YCNk
zUZ_3u*4114p=-R;aGp5@{8+8k&qYFD;ap-uf{buiCO(j8#GIXG3bo^ltgN>{CU6UQ
zTq{zFC;&VleCkX^nU|85raU$!F0PF7fFYK+ni!G@a09&&>oke@!S|>cdF=c+t~FmE
zZUED!4Wf8dQYxi`Cmoz2Lizj8&#z8!>-+oX&$5jZ>2N>?#Q(}iK#hySu1gmtOE5-L
zFI|da9eHwc4NW84E^n_+!RT3}Gmyr<_hpw<)MQsx`c?II?b=E2oEDo$A4Y2hWOwP3
zE76T*>AtF8aD60xQRKmX51P=ztP!{ha}r=Rk<IEqwE4}rp}|)$zDn-BmGHtFZ=CMm
zaui=an5!T;_=-HlyDU15NNqV#^6QQO>0<|O*+}RD5J@_t97_X96!4>zwHQRe9ySa^
zrrG>48rgc*#!>sIxMX|qo8eSZtI<4x<)o+o;<tg7Y1!VDy*PS8ax+YD0^LNrCnp%J
zBVH7>i29v5FVlB-VmX+ri^~!9+_s+4WKG_8HvyEauUAYh3S+o+iXzlJ=}w7P?VW)0
zYF7Ksoon8ICw#~+sU<~M+jg{nmh^$YPHu`wU7XWex}=dDnVFzUtK)qLV{C5j0Y=-i
zh?5TS({jcPDQ}gJyd|3T!R*F`h@zttrDI$9^(!Jd!#HC`G~`A7=!l5dOe|1$r6ebd
z98qj~jaw)~7@oEB!L_9QYWZ|Fn~fctn3N<kHiISqlsg>Hc_(u)ApnPAz*H##EhL4M
zi>s@OA1pPlcT-an*NS?)PD0x+hp9*}0a6K3iMeH<6YhWbVNhjw{`r<oFLnF&Zo~MF
z*7YQE+5@BrqgozDQQg3>rb>kY!R4F^*D0?{O3`{i4gRh+VsX5G|NgfN6Q>SkX31am
z{Mj>DeCvqrJHEh~kq|`ea(G1ukFuITf}BuX!bsxZpOjAuB*4wFkKR?jS3P>FASES5
zlkV3d!k?`WpcBMY#@ja^ZJsvY1V_2>aM3IQ1nJ1(2dvcEUXPuuhdKc@0HE_5u7IT<
zC9{_LED@gXvNsvF+;Q2JwI(-Ct-ily0L4R(jRt-75>md^)EM`Qj*VT3Sqk4c;gUyy
zqloqTWaSRK-d9zP+_Fb!5j%tD*2Ivr@wjU1gM#*Zj-DR50Q!MmG#&uW#%q=(jRigU
z>kE+lVT%8(aB;!eD()%w2&i-D8^<4-=VD`1NKV;u&7Cm~%fhnlTOhEp8|0yNUMuF!
z>l9X8RbBlkpkyzHJ^15L>I2c>=SduX5=`oSOB&&41opDIH0Z6<6)c$d?B6fx0|0jA
z>eUYYLw>Cs{Nab7t5AyJpNeC0MM3?kLHXT<to!$?IL{)+O^ErhKnJ%M7BrwV;YWgV
zH3RGNh$VzD6*flIY2->t$@$}5PmRQSK@So&pf7v5aEp@dbUoZ?fH?J~wG@^^`?}`x
zh5hAAn0j$H1-M?*jrC?1^oA?cy?*kuOz#$Ewm7E-E-1>*p0+V#ZvtKlOqqlib7MPO
zWm=cNSvFYN=Yy~#Wys?4@~90Zz6rxrS5JX&e_rySs1ImmN;Bp4z3o-UDH@{n;7oM%
zH0y7!YIy7e%Ly0QHIz52cc<GFnUT!vf`ar6?pCPdjO1CL`b74kpWp1@uW`x>1DIYs
zndjWBDy|F*>e61?AFNk+!l*X-ku1yQSc*mp;n(!!6}*1T#I+aHDe&SZSbcfX>(}ZG
zRq)pY-GXyM>6@~66i>{U<m4utFivc2*K?U2MnIr(`}gf@<J0PjrNkdVG!Mv2K?hf5
z&QtHG866*XU#sUSh%P}slta6HeX-bo_WU_>jvw(+UP~`5pN0ex%DatC3}nTW>yCRC
z%%AVTPz#R|*5_m@c9#@9h0o2eukKjTH>-`0cH(I~^4mP0Jka-v`Tt$1>uH9QazSLy
z$@l0XB`w{Ia|cN%rtgN%7}lju%d@vtzYMMAm|XcWT%SszA~t$HBL1*!m;eb~gkdSc
znz1vR9uyAdD<r!0WrbP1QF|RdS^v0O)r)6o`VYW-fO?{0p;pAX`1sJEpnswsWkC>A
zS(>?heqK7VTZF&XWPoc433KcwsPQC)3+d?x4;(-_f}CM$OXlF316c(HhdYh(p44HN
z6NidP3r{BS#`hQNU3eT6BCNkAeuK2)+^~{G-<ng7o^%@wEll{v)s$QA?b2okWce^3
z-P?t6Norfw2M+XUxh^TtUw7-*kHP0n+i~KzBjJ$`X`4-19pzy|koXn@620pYprljG
zgA$hC^=|j7Znw)91@k5Ch4n*X%PYr|;mdbW-zTJiS_TcR-cXdxe!K4omRRfzeV`u)
z%V-yDEsb(HrCFXdFF9*Vaku0&p`)Vd01pv6bqZ?nBzQIU7Am$}1$7bWJU6E3q*!%N
zwfnF`;Y#*Ix6O10N@kS0U<2LecTu_A*?J}Itx=*W(m=Lv;<bcKpId2}szNdqvT)>B
zvZRhcOVOn);eiDNWm9xr`Qsm~;o*f8i52k_0?c=|*>1bVou&}y1>uiJJf?T2=<XJF
zk$X~q)*Vr`8qxbkciX6cs6bFXsp*@RP*{Yqg5Bt4%CN#hgYXF>eRsDB4r66=Ou+fe
z80$7SrYQJHT3qsS(-B9DQkJ6g>4}7dyKnkT9Xobms@@QlLlEoKYxdb@ZW6{!m_R`A
z0t`w?8Q*6Ztu*Z{2#Ugl4gHFCCgm<o-L%$IbH0{aFkT_M?#0c2JS*CWz82~J?SqOP
z1q#9i1_p9xpZEssHDo3^RZ%!+eZydb?}I3+&2Y9<n7&@8#BqhjPtIby#4^6&Z$E!3
z>Ue3p?I&kZF0eMx*KqA^x4}YIl+Qp1&y0n!v-T!AidHk){c%(=w;6fj$cc&%v9kR-
z%^NE0DWCLq_=soctgV{At@!f(mb%AjjJKDK?I$Yk1Rol@pkDp~)ic(vTc_h*Ygdr4
zkkP8Y&VIxA$4AHRT<_S-*UrhMZxT&isWMnuLkH8me(D-w;FZ{=3Ex*opE-ku<`oV+
z`to-Vy4rM))L+F}we469k!F6Rc3954^~IOJ^IID8-k$QD2S&@mhm%El2^r}@R1~Xs
zspopR{7C`nm`D#uh7$M6i6*)F`oK=ef%KXjqumD2KYBrT&M}ARq@?-QC_Rlo*UJ>F
zUp(r{6Qi0SWAi&|{bd3>d|NO}B4|MM$EYhUUK!*>hq4X8oA%m&Ke7M|kcu?0U;7Et
zOIG4uGg3eHp`~hQ4pN?eH7n8DqPfYPWk{_BDg*PNos^WI3k+RxK)C4tvAn!Ph<eiZ
zBQ){ePZwNtdxV;`DAxM(m}G=T;j^s8eY*^<q<p}^kYaeXC0^LjT9jr6qML!lS95uq
z`Z~{G7Q>7F6(~?Qe$*L6&S#aus820)Oen0|2BbgHJ~VXRV!Pfg`UVlYbM_wV6QJVy
zDd=?Al2GrDJ=-ft3BqH+J}K$`*Ei>DSc<G2F=*X%m<XN`BK#Tc9k}rv)Y070$QJ~0
zsj*SU*ck9rC`{ZWB~LAZG}c=|0TRZ%)N$V4=kW%Lo*wgIB`$^ZOnSTKue)9rsPT$?
z8{N{SeFxZwjo;o$3P?Lo^-;TutYqAV>u^GFT{a%}<<5+|xw-QrwB!YboVkr|^70|L
zk*b@U^YCwoD(d^8KN^M~K|m4qpsSp5g=fz`!T=3>)7+1NreB^u``CTn;9+~-d{K*X
z*nTqz$2QH)tDm!9PpPq&p4XiciJLE8ZLa^_OPg)slM`2;8dzZ2Q`K_VvC0s}_w2&X
zUv$n=|LFiPGjIUV2Oz08NAHo$2fHf$dNa*eq1B6}zH8z*$Btp-RO(vMgsce&YKI@%
zgKO7Nj`Wc^*#JXUN+8%A+VXw}iQ{fZFWp!0F~+r_jUKID@af}66)n4=W*?HWN=m$*
zoR}tXP9=hj4}PW8QCf5P)TyC5+1pZ}8kbyL*-k*^@I%sxFMn!sy9RC#B**z)>yLOy
zNO}J=smMD+!gc`$Zyk>I)qmMu2GeIj@NTY)e^F1VFj%d=Wx|#j%M8nlQqJw~>70^!
zxasTpqXQoW4(c-d;FhXqodx)X#KSu}{4F0D;IB@xh&?Z(@sjw=#0e51=73N`9;a7#
z(R1kZoI2!*4kB2R;*9*G8vS?aegE~Vx%%2ymdH$8^Qa%O8PP2pZ$$01w)XLP-7rxo
zbCk~jKjp2ndVw|<gH32Z04soJ{9=$|IoV7fg^cNlLWTj9AFz;a-(C&2fSv+s08I}-
z+T;)e`D5;*d2BwjVjW1M^ux4*5IS#Q|FJWh>H|(KEOy(kS1F)18`4XGPP%mLC=L_;
z5~jtausp~NRIZDrO^XJ!Jln?ZcHIwQIj|R&%FC90WgVNtOZAJnzeMSHEs;hz5gnbE
z+HHMydzqNY-yr-Q<m8?`3rtz~>eaROKkK~e%6cJEiLNIcl-9Kh2_=UYjGE&$v2Kry
ztXt4OcM6{*Y;rvxKH1!G*6js`;<jR_YNcOupjm-*!=V3G`$@!8bejK23dg?J_0aSG
z_~{e7>ato6;p9kV0G*pG%aww(MB<_Xa(FS__hqY0yS6K?&9qv4S!6dyrs?z7G1kok
zg@NZI8hle;Nv%nTOu{4*ACT{JMb|$d0br)&#b=RdM^AEQ#0?Ko-tvWb8R?Q}E|Jbh
zYa1i<Q<?s7{)X%F7^~BT++n#{uI+Ams$ILjccpP>={$B+={KF*x~8B7M?D)-x$?b<
zk&!57V6~dSaR{`k4xbC-kn7gFM~_-+F%VEJa%ZS7w1hn0Ri{yRsLWZ5ONK9->e`Q!
z7&K%^4m=U$9FU(Ir`xCuoMREilHKK$-++}T@|x?Cr@i;6xv-jWKEFa+ZszcjCfe?%
z&3`uSn<L$B+Rr0L%I*aGX!g`yy>1<ff3$BVix)GEe0*LL+=M)LNU$EWM=zmp)Zz=e
zpUY41tcL9WhyS)7yQG$!rw4IW*T7fcx?NSq374}?P4Q&&(Y7GEq)|a^dpvVB4XI~Q
zmT5bc&VdY1E$pTpNi&D!D0H+1yrwbH_u#+*?M}iLn4?IOVX<r(tkFUuJ_8)CLbOa0
zk&U0pU=1$rd+-(5wj%bjmbU~_peJ!X)&J5#{9Pa#D^tdd9s3>F6tN~rpXDF`#zD!4
zJ+D9rM7}VLmx{_qpvz60H%ksDWKgrqWj#Koai-`^7E%Bf3{S)_Qb)%dz>l{N@8@$~
zY&*Xkgnn+^Cs3B$+H#J`e(4APN?)XWhb~x*PXSw^J%4=;qafbBdlo!RfU}b8Zju6}
zMcRc6b~Sx>$rv~!d2GD!K6~fk1yM?c?}iW4br>yN7&`xbldJoFns^*?keN{w8(UbM
ziI2a8=_Gx2a;`ESvb#EUl+h>wHznxv;ypuj0e*OOb+$3Cqn+b_=q8LE7B&CDtFagu
zq}N1@7(V>a?B_&gn{i=DZfgrHIs1fy+_!S}0YoQQ;Q@ELzdU;_2bF96w|6^%cH2dM
z0sy{eb1}nq<&JV^@+c;JZL+vA7VoQY#VX^f(|Qb9tLBb#0KMh<SLyS=O@E%3hebIB
z%MJ3kQ4`2c(4j_A)j$8w9w-Ng+RNzu%$4#b>vmKQboB6Swh-S6<CSC2R`7^XQ2Zm2
zp!Uz|`%#l(FFOuMyS&&&aY>PuO8l{(Rab7f>lgka*JG=aM}JGqmGt52F{ve#3`yy9
z;{5z=b+fJlND(I6=t9pw={Gx*&9vz9gn{5dA~thELp78Z^<{@VCrNLw@P!TavtHUe
zdQGk=ay$u9$<AZl<-Lkjq~<`MEOFb~?qEr#9=%RmpA|W~)1JItKxs~!1NixY_`iG%
zeC&~}E{vL->$!99b0jF|z$mNk>ueb(vEtz|4JaE`u=oI%L>umkimh<Dy}iAoD{i#^
zDX|>;CVGKJ#A48|2eMN#CT1KL052k3mJXC;N;TCxn~hIV)oCf#t$lP{Gf8`R;=MP^
z*4KJbq={ZJ?17)RNA~82?{`&eZNEzTW{ham(Xj0`D}$U%yQre1BxFHk9l9X$Y;^P{
z#7Eg$aQ^YjhGN89gJ(j*KtZ<8-(UW<PcUtzNF&&p%|xg-9U>4TooHV}FT^qvVJLBR
z?CtDWDRFl3MZIC>GQqj)m-nitk>isC8rzFTLF*vC^9?ma{I@xe%nSnia!%^<<v8~-
zGAV!m9(voTL`VPYP0%`@ilU!=nobCFoJ=p6cfMXX-!7^BrDyHm^$?8=5*CWChRG@J
zKQ=5=K@ZByoA%WI?3=7d`=Rv)$z5@+dgj31H^Rr-QnQmuT?J4lXx_ZevNrkQ@m-1`
z>z5vT|0p84>i|t7zgx=-wz?-P6rWf@s#8)l$I1X4S*okm^{t0WTR_h252_9+IL_BE
zBia%1n`}n<?%G9`G_-p*noS-z+?}^<NnO5zhyyOp&47u9pCXN0+W_`d-+c8dJnB$%
zM#fZv2QT6fL)*{NF<gP4koD|am=-Ys_b)P2nL4!uFlXa|DHu-j8Z~M?Q$rxJo@b9A
zIikt)Mw{k^)us5dv^)+-W@LD6Xt7uR9?u_f%7sbJOaJmKI@i^H>K<aCY>_M>3c=^L
z`zF)=Q0}8e#g`UT*OW<DZj|nNZb@cF$;<kbBqoYW7cXvXZum;JTbWYC@Pv!)O})rR
zx!dLstf<r{2ZLlZiYOq(Z5~3(>W7Tu;_Y^ch0er{B7d2gIoEi?xN(EY9RpO}ok5p|
z7?ie^y&~)@0I!7;%5On4)K(>n0RFZ3N_T;H0}f2?geM`gUSQzR2|(+GiYB8xZ<1@Z
zB^Ng`vILMA@PI<8cmMugW%uMY!@(=>@mKASICZlZ<_qZ>(Ru$5V{ZbMW52e2UlCGh
zFpE@@S-3M5N<x_m8B-ZTGL*42OC;TBFl0y~5<(PB3aMmDhSEGJng=Rre19iveebh<
z-}`Rc*Y-SXTkCo5Ro8W%|Nk-U$G-1}1z<Iu%)Dy|&gajYN5RDOkv%Z|`}bcK6%>Rq
z()HL_ej$e49?#Yj?ja}9y~gq)0OUwn55vL=r;e|xntCE=s@dodd254@{4urt<OOl(
z!Oe@-yLZdB<jMw~)n@Nd-P1#<#qMOTEmc|3QF{Z^4i0B;CONYOY0F<q)v8*rM~@DW
z!tHa4<oEZ0bt=1H7@Rst`hxIGGyrKF;Mt4I17u{LuAe|jhk8!@HM=Lf2!^EwdU`UK
z?6E{H0_*+${X0Iir-nY8bp8zo+N+r(!+8&KAoJi@1_PIw9JmZxAGugvz9jPy>gT44
z!`$^)1e|nI`~mu4K#&E|W|$=-e-{}pUD~~8&&d-fDwPPQ_uV`2boNyoMsGuRJu05D
zH2`ahWB2ME!s1CJFIac-0-lXrbt~IylD<`(QP7i!B~l$4f(h0Mz^<`>D&8*T64beZ
ztfsgafADdSI$e)lSNGEU=+A^YqZQj%hTbxcHnMR4ohqmwAUe=B7#V#MuD+)y4qj7`
zY1hQH>>^X{S@!6t;XlgRpFAdmgHsB?^gllH)s}~j4i0Xe^bju>S`&9$*s&xTdWold
zVZsI-C#QD8e$x?{eCAficwV*FY@X-l&Qwte!c&RO@gs$0F+4P#7<(vNK7lnH?I|)(
zlV7~>+Nbb$Rm%V}02(*|EI9B3<>Yi!oo62WAuh5|&*BYt8MEEOg0&Ph$X6*t7RR-z
z_HOyMaN#ZEiZ_3_FQ1b97Hwf54GKr+hyD(JGu<dhJ}cU+!w;J=%r<AwM#=dMzwoSR
z&6~L~z&}J(cxx{km-3-6kt_|g+Fo?nwHY`?&T${vx3AA2%7ahi)+~EyH@PO>?3Mhf
z1=mqPa<rx|)*KR;{)fpipof-LJpaez3PWJNS;!fpQ~NRMSOaA1Kp$$-&MuT%tiy5Z
zwKT+N=+FS*pR~5m?1vb3VF9)D;aCgXHMW+PqeX%_jh4Qt7c(tZG{NTLRI+XZ=K~S`
z*kC99H0JJ8G|A#5gjvOf2SrD``)kv?v#c<|op7Nxa%90d7Kk9Ts>+F4UitWZI+e=^
z_0g__-m@}AkU@BDtKV0>grWO5T7K8z%qD>IYw0;s{WUYpmM?FAGcd8OC1mbgxr<lQ
z$}jnEpD`m+?r62FWrps@mn=A;dZ%6F4?uqzYZ}`Gc?hYfjK)%9<Ccdfe%i*uHH4GQ
zrZ`&a5g5kb?36iGur=)?Ik$eF{zXk;??(9j!u>T4NQ4!~9rItK9sIGOzMdD?B%j43
z7wCbv2AL5$_G;)M_%99^psuQ!veha+Wzf%W&gtiJ)&>jU9*Q2{%a>t%s$KANZ*6&N
zJ_+gYd(|Ox7Ejrf)T6zB1+@dL4J#@}U^^}d;oTUjH|7)BjrV){<Df9o?)#XLBmD>?
ztJkbqs*^s_b@MLMl9I%2=f@reR~XsnIEVs^!=TWkHNr~`syi=YeEn%Jt)g+|l%Je=
zem|{W&mdRgdGWSAofmt2zGtQtQZA(|7Rw|xND?qqs9FymES^t*e2_0#DF7-i>iVVZ
ze}jroN@q0A%WFU9)i|X0ZBp>`4S$$VJMkAmiSy`?T5n)`<$<AwF$rJmVBT)BwB*6~
z!5Q%N{YigD)k63<;^r2<zb31zXQJh_Zp+@q*vR@^vzc;u>F$yRa~JiPS99~@pb6_m
zDW|*w0#c<deyddFTGN10%Uqhay9h9OaIdXUS{OybIQ>cU*$5*?A{}9)%Kl5?bK=Y<
z6_!3cn4$O8wPPj5UMxJ(Qdj?!nR$N4)o13DSV{~lLY*}$WKqe>c}>|Ja&IGcA2oWM
zzO7I~cwvu#P-purT?i;M6thR4$AOKB$h%ckdJ8a(Q=3rHE{Uwy!1)1GwNT@n=<KmR
zQ)X!m))|sVU7q5}`9p~8_{eG5vJWHxz+W#VYLD~k+kDpFJcw<}vsVl{1kuUZu_`TJ
zB6^{fhtwl^cUW&71#{^^>U#IRmIVI{)&!$rqcmIaZgX4(84<HPo}_;Ln9o(t>EycW
zxAMc&2OZq*Kk&~?!%`lUqIt)V2Qku?X8egxojc3lybsvN^l)WG>Z+>~jb}>r9JQyV
z;r5STY6gp5V7UPL<P%#8XtAlhgL5e2q;t(+<tV8ezL8CNM~NsU0<GO>+>(zE(Ef5G
z{QI`W<hr2r?5s`0v~y=@>C%3la~p4d?^a93;fLm-l}w~T@*w{dhi#+s!M0|)N>$ZQ
z5C8sN7cY(}d8?msfoFdH4_)u(PW5kVoS1x!sTZGokC;zN>pGg=H#c(kwa?8z&&b4#
zIBV_iY*zc}(=D5-(FTh{jqhKn%gmK6uSL}-4c2i+t55!YtW8CL+(Iq#UWt`00DcmE
zv7!{QHl&RHT(^QQr<r>7?ZvI$>%wl`!dBF%-h#<}=9s)MU;Ix05A^LpRMhl?&z#>V
z@u8oP-EX=j6k-T@S9a>B`t<)1*RD<UDAIk{aOGY=(6ceKg=Iiv>WgoG@H4s4O;p>b
zk0q*>g^NQ?;x4({s|`~r0CIpNB3i#|*LXFxe81%{k2O;<P`7n?%@CC$RZo1(G#O#`
z`e1n$$;R#ZJNiO&hN7oPFyrc%h=0k>_DySG-Yl_L0}D^@Q>UawSlo?U^CR`g1q=g4
z+RK*7rCb<4Qd!wGcECSddK}~{fG(P_lt$29uNix`U=$S?EQoXE=|MXe3kk})=bxiL
zk=7zLou$azk6ga;?XzE59Y|y7d*6E!=eDn0m*cy@F|}lh;da%-OP$L5d~=`UpGi_-
z79hSba$U8+m+|posxRiUB-r=<qlq_`_}i6#8l~#HZR^&@j3;q%gKB_m7(qK6YxW?n
zShb2;ESV)B7cb&*v4X1wp@YcJS%94pN}4%zXf}upoC5t1lcDQv$!X~+BwQLyV0)(9
z!t_Z~LxZ<G`pgAyZxE}cW{4|r>BI)&UPy?T3+?ARJWo>0scM0a;rD_Hph%%kKT}z^
zWyg+1%7v;qv(7A_7H6|Y>YF#=VPR`<wxLwxfK&e0#CI@6kLuaprML3V?aOl4ZvC?-
zE%4DqC^T>xjxK+(Y~@Ny@Lg}m2X0Z)Shd_JIP1tUh5o_CRpWepzjnEAv?l^E+1M=l
z#hEiZpIla|Z(ym6ap(C!bggY|iQi26S1>JY!!C<juR(CEjLPgInFuVFljA@)DH5zP
zA^cH}h+J;p9^=0xA5sh|Q;1J0UoKEMSX<xUvA4sA@A!X#mGJerXeQ7IDai)2Rx5Gi
z5r{*JV~4U^30A4|s#evUc5XF;44`h<Y66=3rHKcfa@O|5h!uao_i}+N9C7~XQ!E}=
zgt^}|x`p?^CY{;!5vbsG9)2nlS(KIK-U&Tpvd(<Z9?{dj+RK)7K^}L^&u@!#51H}-
zBu<7d$nlGqW^#HK?8mXh9<bmauu<qpjQikT(+0CS1*#zALWa<{82zNAbl)<`2A-Cl
zo*&CfI!*(6{cdW)E%9RI^o&XfO+Kejhx2=9AMOfqb~GJ*d3ig}jV?*){7><nnGZ^s
zm(dd7eSwI4IpH}YBVghtb{u}8vvE5ctZ&~!v~<K!1~!7=(5m`WSl<WBeW>#w=db$N
zLde>-bk;@t`<q6N8@IDWa|M0~Kb;9GpK48KmEOMoM*?)p-Xg+`TuxL<{^~eoe{ah<
zTsAXK)cce5Gy-|?&hg-Vgdfku@_ipev!|>2PDd$$#z4O>G&|HA?=}3PG-^1t+HTyq
zk;48jI*_(|Zqk3b0A<C+Q2FOye^fc(=%STpKUC+|H`%7miSJz5IkMZ+t|Axu2<pJ(
zw{Jt4)Q-!#euKQWl-qpsrp(&s$Frnb=Vgh77f$^^Zw|sh#!9=mY2nFlJ41*3uAKS-
zS(bX-<Q@galI>BIBK*H<_}}?dK*U=!3Uj($fIo%{XId*xiaBUD6sPp2G78J?0x&go
z{P-o-_ivYAX~|d-<yYVydMTh$kfF1YBO7ux=Q=xs=Xzbb>XCiVEVKM`7vB+M_LV7r
z{`Dx`UChDWJzq~#)H-kU88R1z2u5Pq`0T666ZWxE+V0595&Q656gP^xda;Z1_-xG{
z1)8EB1=;tYb3=?luX)Z)=lU|N?6B1`-n#h()uX&$?R0kYKu!7fMEOteA^ZD^;sO)@
zAdML-K3`aIyZr3<l8Ca$DM<;UxF-o#K|@4N!*<4!5(1pt*OWdf&yD#-4GUuWWuCv~
zH<7~BF!T#<hCk(*jG-c?wc+vbugJr(6U!l`r$=eHC)Cx~@3`_Dn9+)1Th-(IDX&XL
zMnwFWJ^Q%3^zzZ8|G;~n#+j^rXxaCk@#8H~BjMi8COitXuhNyj0k6oTl=IsX2E+~g
z%We-mI|<?u=lpX<h9{w5&$8vqS%j2f?N9EUGDnnJ@+vWDboCdd<%Wu3z0Rz)vNC(J
zQ=WJ&5!jDKU11v%y6SsWV_Zn9#Wm1)`oliw$e-0RLl)EYJ^ZxIZ^y=~uY&YV`j4|z
zwA*_AqI2xt;2)(+1?}z468h%Jv9Ac?1>C35^nkO6|1`yJ8W)o8G=%!bE^B*ezI>U{
zmN%tatPEbiIc?-Vc~?&I?NwtoZv4{Ryf}_|SYQ%c<T`fv1JJE3)nAw}N|3zDlv@mX
z_$6+eX5s4@&UYhi?&Xxc?)j;(wCpq6%dB?hx-}AeCO@O_(J~HZ)Y8}P!-v1B0>k<6
zY~}@jlvi+vOLW|5=d%2`6*)1hW{tj8*lImlero#L%RR?k+RAtIQiQm<CFEyAg0l%9
zMCnr+=GrmQ(Nhd(Gc-s^3GohHI?Krj`p)Dj!hXkPEUaA-n)fJl#_;R!(qu8Hma*In
z^E4oI6E9hE=C*DA!=>Rp1wN{dzkxeeY^h3(_58Vk`_ATj_v-H4Q&aRqsc6I(r-;IP
z8yp><z^8L?h%HsZYyZUyUH)R~<LBnjJKE=GMo10vv(*nyc(6Hkh^F@35~=2vhUd<b
z*^^s??{9i)mXHeKN^(Ur=VLURRV4Vv`H)QBn#u;WO)g$#VYJ=9SnG&T>cuJhFB#F{
z$k1)KN@~O$vw!IJuB8!>vZ9CG^Z-&#8Tul<L(1iXC$nMqgl%t*nwO5zy}k@Ohsfpa
z7D!r;ox0)d$7z>_S4)UnJ!{JwKT@j0H}mo54wT6`R;+w=dzD;;VaN`7`ES*^Gj>*O
zDXm`R;c%>J^NqIW|G7IsR5q!Y@x0yQsj$q@@E9Uf7`YX{BnQ=<z(g^mE_DA{Q5;3|
z8NpN0!(qDsQaLs32>uH|%fHh*6u0HSaa1rb*l%Xo8|MthEIf-x)e@GHgHGJIdbN`%
zG@zTQjC%qhbM)A;m?o@_+A};w+PR^8Mc2bhWqmSV=w7eNmjALrK?ED<bIgG-^xy3Y
z&Wk~zpg!<OF$gsN)n#%Qr(sp1TKR2X*_@QjFcQ$x^$4XF7R<GnkYwFj70@;>i(zom
z<D(b92wRJ94x{lJ2osg^Tx>nYg@98cYnn--UD~Vg`hCOI)G~{UyACoxHE{V8N6qay
zk^RaiZ~<-z_Sk$mpbj)=Y}BNS3=6A6VuU$NR`m9_|Jl;FzaJz`&@XnW(Oj_`A@yjF
zfzP#6e0{H#R*bPJk<`2VAbFtr%`IP(?8c1o+VVu1U%qk}X5r|u@kk2aKYy)cwa;<k
z(!JAJx)g}_i0J9pIy#eik|9Tqm^ihs+5cP|{`c^?H6+v9t!=Ms@<Ja!N}@Y(Gc^8>
zSj7Jrh#Zq*5gd=$w{QO3k4Zk#Qk{3VD2Ry+=3ckUe)Pj#x}x7$(V!(Jy4<U?|GAnY
zK3tnG?IU)XdSgkzUF%Yr>vxwptk8I}>zV71)2FjO-#+-`M`+eUp5=eG3Mz@iQowk}
zk9LoSicN4^LCpS?kXc00$NMub$e)Wj@KR=ykAl&v5l-{hkB-R8if(;&c(k+gE_Ony
z-vE8Qd)EoGiohQ<`$#+-f4-eci<_61_x9sn{*=GMJ8>a;Yec+xvuNtn#~6r8^y^0@
z+ow+sbu119M~@zbLUMtZB)|XubS4U`vYx~wKit|V@)ms-#fx_Qwi<Hd$XvE*UwU@<
zqC?+)o|BRIy>Wi{-jSi9RP?_*@HVYKR`!AKD$rQ4{y{M_klsj$U(B<RB2Zd@GBL>_
z{xsuSZoO~c3_7fsyie!OdIyGY@2{nmGtVU-jj6jRz0*h$D#H5G@~Z4Ls;XwTp{%r!
zdjhE(jAQlu`QN4FdFpO|v<pZ(%9*39*a2w?z<11nP^iD_kMJ+7puw-~R6-&A+N2jR
zu)}-2!s_CpZ@>2rx3X&feB1R%SJ85@BC)*AZlbY6W<=V(?)S|mfAZ_T%vMu8VfQq0
zsn=u)e@Y|_uz8frsn2m`eEFqiYQDpA>R{Mwfhl=R7X@=HT2hpkko&3cKhN{udDZzx
z!ug9*b8B84bvCPHkzUm9Yf)EiMC&_Z0-vBr*c;)&lN0+RdNcf9TwRTZP|uX1Nhww>
z*^u!-&EB!!%;+QIeq>J~FYpU?oLE1^>4}nZxW?GN3s#&#Hr(^O`^8V!J(uRkiE3}%
zseBMA0Y6iwIqN@5fHdfdW;(aWS{&p}&|@<K9A^}_D_4V#gasLN@e*f$;cIAOg8R3a
zuu*FYW=dF1z?YIphtSu(M~_R)=Q*MX0Y)rBQCHoL*zuVL$9GtE6s7vHD&`~C@oxzX
zR63RHyKgyX0LO=n_wTO^o^f-F1w1va6zN}sU4GPb92XP0Ff@;`T#-4{dF+?i^UjA|
z$Myd|DhPBO`;9^UYGUFDT2!12MkrXb*xEWY>IxBQ;@@Vj7w#)FM~!0V#vS}OInE3?
z(H8B@4cP+IM?ltTGp1Z*tJqJf5+czQ2T9MwwGoypI`?HHCXf;Z5%ofZ{NSFYU6*v1
zlInf@kOx-<AnJ)z+XXZ74?lfnWecmT<F<WNU1kBj@Wu^(eZT(wad=hx^pXopD<N9X
z>I&esWkXdLUlL@|jNRi8)Lb5Ecv+>kS>edW&6`(zElWxJ&`AVVu*{TMH4|G3d@*gZ
z<zMsWjyt4$e(vADPqls*UC5zdV!TkxzY@}N7lh5&Cy*k$NYTlmvH|DLS^31$lu17?
z30#%;)T#rm0nP-&6xc@kBOogL9Fgj|f`>_9q>P*LU<CJb7Una2Z|v;=yCHGuzQTNP
z#NrFO${U#CfgqsQZ|8K=Zldyl!HVt_KuI+hTXSUA!WyKC@^UP!x@5;&7j`<RJ^>tD
z-|-4lFGuTIJ4B*ACoC=SS}zJ%zjbQeIkP6MS+=z<9nFlpeN`79|NldlSJ}Ub=^>!@
zEOr}4-Gd0m+-`F1d*PY*NdCD$8W-r1jn}N{e)5VOxD0h9)vMq1y<0BsyK}G&O#mNX
z=LnC7cn67sL7!lVV9*M3J}cXHQ+X}sEQtJ}lLs!Y<bmRt$A&l{xL_z%l2w&L(^bUU
z7PeiH@fhbxW4|Ecz`?39VJraF73i?vd`UaW5r;`-yZvkD9c9~2Wn0TD(P7aZQFZI%
zPTdPWM;{n>`%Y{}{<d1af(RiC{7kwt%8!(Jj<j;@iB#?Q<|`OB95{d_Z63XugQFwr
zco>gg9=qxULZ3AZ=@`2-D9r+}JixItm15w$=aYpc4jN4F(vQun5qD3(v-;e=@F9p1
z`k%=AO}S1MR4uarcG#M&MJ$M()|$3oFlNNFexw~(^)#NX=+4@R1s28Y{s%96SvS$k
zCURk7V)vD7Z*~4r+?^$7q32Rk%Azg2U)kiR<i#5-82hrSbM1GP*pHvza=DSC37a_*
z^S8xAs^cMwnDFSOP(lj=5vWhO#qpU1jLx1=o{I#f5Z$p#{9e9+q11C>eX^lDgZADJ
zOxOBwLXYRhhdNFtA|>=i1(aB_CsBbeV0+%Xf4Q!BDC-=7MyfL3KPD!obQD7{cHF~Y
z207tfixt2bL9uJ*>#!P{5PK2K^u~>?EULb%vB8z1Tk(FalZHq%LCi-i66mW-*CeaF
z9wI)`MYeegd-2f9EMryXnS?^oQ(EOe8&=>+jS%Jo5D4dnvlsqdap>2_F2wSPNq@57
z{*kRN_r+Yt1Z0P#=51kHDUDsC!lWrv+IO@7oy2uahEJq_hC4b}iew{ph4>7JxG`1v
zXJMDe2k>9%>Oul*={R{aA^zug=$I|_Wt~N*%gf0qN8CNXxnC65KHKw!y!<&Wht*SU
zb}d*tD5`|u=<VdQ?PqI$08y@%vaIHsZCb1ppsz~n^|5RpK~m&Zp9B+8XI#%e8Liw9
z{6ow5JI4>`>+EvcT1+N0E&qn{t0mM`tcSN*y3{+UH?;y(##xg_zQy!xKO?-b2HakC
zrBING34fxeRUsPypactib`cSnD+C4^%@QywJuyXgdjrsmFcvZhxpz+M;IfGmyq1}c
z51HI$`t;uW8?vqPatim%EH3$DqAr?%ai(a(<jIZ=<$XKqdUh|MlrRYX2smIz&4KDg
zRLbhUphj%WINL8ng#tBHT99o?>>sJKA$YaU?TG9lfc*LwZ+7>oUkU#g*IxatNKoN%
zAHUopur7m-<&L|@^5us!RhoK#E8C}Wey&hRPa3e~TPni!w?5uWq2T_gKRw{d@{oq~
z^r61KZ~Lotb>ZZKX2E>(T~TfR;gGP15$$pi+3&`Uql^6L%OnMUy3i1W>wfEXZ0>=e
z+l=3f|9r}~Qh$~?4tYg(zi|Cs#=km&4wP&?Gr&H|^5f%!u0cmIPoAbkmCVQlo{Y?B
zwi9?A$Q3((@7?+dwlk3Fn0T1GZvVl`Akme`NWUPF%FpjxU!^n{M=Y49cs^F<Y}D^$
z<Hw^7E*D#D^Hx8^swJWlT-O~_j48htWsGR~5SCd<I|5bfM}bSg%J=MNSk&)jYU;W0
z)S>zoY6na^3k{>>-)Uap08Bmxsoy*#38;Cs<@{sK))OCxb7xNy0E4c<48Ux(xPj1T
zf~!ZaV8;MuU>Yb_S~G@fXAXQhui}nwy^Fzm-|4geFkQWN7B`W!7ESt+oR~QC<S|9D
zpREvpOpd%=Z?rCInQZTbcANe)fixbh#mMITn*d?H?j1#LXA7xl1{vf{0K9pTvq#*c
zM^mQ1D%pcZRrekO&6f%d?5*_r=C4gSxzO`(cIY#3;CY%G^$7_M_@{2n*ztH6{3=p}
z#g*sC+R#A{+EpyLy$E(YWHl6)o1lpCKfUFD0@~6AFUZzsfG?u^Is9d22V5gextN{<
zm>+nyXZrLgR2rS0e)jVEGI_g5yKjN!?y9|Ad)?t&Y+5&Xp}E1V2TqrZ_q#jyf7;u;
z-%Kr+@6z$Tq2m=N!re@<vz-lOW&X>JB1_$&=1&~_(~Z#W@7OVT;6VMW%V%?|8QN0q
z@BDE4+aBU0Y%=e<S}K?0E4e8Yc&%H1bz-PJ-L8Hl_P2s_5d^!qe&?@IWnd0eN3z2|
zG<{brA=y{k20kIkO}`oFWC_WyU%e8o2Y+>Szj^(7uN5mqwO4Pb4iJe3i`*2FtXt|s
z7^KA8+{?=v^rFgZ{@Dm<_<kyVpzaUukU2j$z}p){*ID$G(shh&5qaok`%f@kKC!Di
zVVW&kP!um*pe*z%I`QY#NLLHHx^8wJghB=$61WxfPqTIF?p(dPkuJ!}DxdE2QFT^-
zeBj`c^*`8#Z=*1C`SuMP{ZY8(CKOt1-)_8mwYpc$;+ISWVY#oovsxH^`T8cGkZ0x3
zB;+Ifb%EvpvJLQ9M;sAbPp7~%m&mpIYYZhjN*vf!D@{#zQIc)i#0Yf&#V`HJQFnLW
zs8Sv|E^2MT?ZF0;?$TohYwd!HCSllFyD>9h#Cuhstp7l2qeiMbm;r^p#gudhRmP<L
zE{elyRc1fINYmr{Z&bt=FJ=P)iwUz4iYdk&*Z)gCEe;g&W8$tXcOj2-`kO}<ao;#3
zt@>$IL(rSu-!KRG`Zv_|iHI}iUF#`0FV+?=-j$w|B<zI6ZmJ167lTo{7hdJC)w6L9
zOitd)Oo&|dVTyol{9T_w_s58be`aqhM(3L$QgqDtcm1T2VZ-`5I@}yOOXQ;b)J7C^
ztDnoj>4%ow+q>^hX1GW8vH25#Be;a-@-v{~K5OX1E9Hl`B4T&SXiCj})u8py6B6Lr
zzW!Rrzv1V%I|DAXb7%*yI}VkxWHXQ~>QC?T(|pbaYZw0%3<WM8{%a_3G0>T={_pC=
z`Unixcv&%fN+f^pX1^epcl+I>fjtoNp-GI~jj$M84bvM{-%KX+#0xGav@5xi#`4dv
zGcs1G7w!B4V*!8uqOV_j_wUcPsqG4<f^Xl}gk{0t9SF4td&KjdmS@GBTWhFjn|{Z+
zv+Y)q<e9Zv_6Hh{_FT8^z<9qI2Pbte=)%=y{%o$Vb$*+gCG0ooCY?mQX3A`ZQyGl(
zq0J&zqeo1lT>o@c5QkyTJ^IY?>}6M$98un8^`+O1|6)LY_G*Bp>)t32JTRZw583C=
z2i+=fChTVx;Oj{a-kIC~lP(XOy&C*$sOE>W!Z>IAIyh$%BHkLGp|xjm{2ui|i|5Ym
zBBJx;+aEdkN>z1q)tA}!%|`Vvu8D4oL<MPn6E#|XJ&a#}e0Fa%2CTvK={ZY7<H8P)
zjBRAZo40PQG&1tKV}G+JD_F4U=-#&-<p*y`SX;S$?1R9s99R@oV;beRbsKPqL30P+
zjTe2E1D)=;ZB5gO@Fy*mQ<sIL|N1jpX6j>yfej8A2+Ab!)Nh0IK4iV*)T*>$*G*!~
z=jzDD$%sS)Q1k7s8Wi(6x<QQy%?41oyuC!dhR(`K>1<oy0PT(m#o;58aOU(yEPPO1
zF*pwjA4tZJs{=JXWGuHuN+BRtsy{=S@fiuNe*N8`pgsx;d3o1t&gbOZ)oQ&m$<eNJ
zn*H$`Te^xwUt%J|SH`>X()h=vM9gC@LMB+5t9`Qst1T#7{A|leWMlTqhbNw50L6>~
z<`9De2CiMY9O3fP@*2b8TDb6b)LnEeq!D&39sV+hm#}c*cJ_O+V4`Z~iq-b4Yh=-5
z4-rF9io2b;{=G_4wK{~YkRHjD4-k~H8Z-<nw}rUI$Lr`x+)q3evP$G~VQpjV!Ml%l
z!H&V*0t3N{#=4HQvU-@jm3dF(MU?Rei2aYsyX(C5d|d(}z?pcwe|~#iL_;?b{~p-o
z53Dv!!sFNVVYQ57r0L#K_Ha*W44KAwdcNlVI5(FCvq6<f8?RUzDdp$SlOgBrYm*0g
z9yaA7i%o_O9U2c&iWoxi=cK<~;QB0HJiKMS(x+F{*n)?Q+>`B=mW&r8hRm<&+gIJd
z;PcXJeOwO%u+7+9vixK4?bByYunh5Duz}c9s^uMLTUP!g4N)d`)`sJ~M*5Ut4BD%q
z^d*p!M5r+Z(dW^lNl=AFE-B_dvbIjUq-HZDMGv`Y0)ml2Ot*#BlYtR(<uA8C{omaW
z8vh5Pq55TL9CIWLGO;Is!3zD9YlKO2Vp0;D-JFK-|3O;7jTO`8GDUe4^1DC6d?}nI
zK#b6cUHoj?^p*LnXQQy19Q;}3nmMa8@O{KJ1$h|QC0l0qvJUxK#Sa+&UB$(PG5df4
z46Pv+L`6ojr`*%gS;o?I@!&(>>ul*{*{ZDgyJsKWVBr~?nwfQA!!`b#ln7@BelP^2
zjE3KTjg2D^m(Cp+tn{hOW7n7E@T%Cv+rR&+6)QN(W9H9K=28If+=vh;+Djg2_G-?e
zX&Yw}ecF7;P*!I_ucI8#ieA3>xw(#^%#Qr=AGIZ;oqr}wR5Nj9=9~$S8jCvP5MMb+
z?E8TTXg$B}iK5tT`1Q*wyIrw<I0+VvE&b&X|5Jmf7gbfgZU42~n4k-!!jHAJHTxFc
zXEbqXkPhIta`gX$n}nT{kdmS%^fz<{b_a`sE!suuuRkFHeFHQav@Wcq_cLYzunj9f
zavH{wR`~hL2;KOwj#mBusA^{3+T@P@09uf-sp&g4w&3w|($f>mi%h8Vo9vDm`c5A1
zf?xhCZsH|h-_9c1ENVNhEI&vYsOf9i02-21vIoI=ip<Id3tn;Fu+5<MV?v-#+QM0r
zR%`P2P(?S5f%@`|wsv-mF(*KhiHsENN;{yBtc|yVG)hSb14s9=%SAuGO9*TK?fm$k
zb$8b-`q$^4TI*D;9B0FjhF-ls%TKYMJXvf#d_S-vQYk)6Nb9#fxx3&I(j|sZ{GSoO
zN=u(2{5;-|W!iDXg;aOo)M{%U1UYxK^BsB`Q7Pm>^Ag<*2|>EZRbgMfS<HcPN=m}|
zzsAN!^a5VKzQfK?>f!UOKXUuFZ46^=IW_Q${!vyY$Q4lyvlWxCp>xibatXpDh*%-S
z^e+MM{(VbL2y#>l^g3({z>pLSV(Qdi4AY4%DxqzBLe2vK3O^+M0Cri4jLNnC@nrcZ
zIF}SO;<wM99eTDNH$T1vUrW`utg0#(idJA>4Zb`o9?|;s>uag?2%L}(i}8XcXpv;e
zw?ziM;ZopIF-lDhnkIXNHcl8oVGQF7>K9kAg}lWM1ndmr3cl);{u3lsfLua|b<kKb
z@nKOBT`0D>P{2Gw7hPY*trLPeDnu+<2Z%t{C|Cqx<d7k4(Bxa1n#P*NUFvwvKDoC#
ztk}e5f``3hN(<H6&fFnxL8$MQIGsPTo<C<JbmYuxwYl*Jm+#IPzjn)(&inoHos1vu
zl(#Q6bn>7IJfGOy04$2Ty@a$hmbKh(p`@u$Q2Nvd#J+gj6^h@4$>$fT4m`s6*%|LQ
z=dHecV#G2n&CS5}taegc8-DCG|CtBC#&g!762O-UQv>r9ISdHs$vDZv7E6+m>#FV&
z5~yzR9#VO$lFx~Jd7TYlr0yg&>fbuz7zl)5lhesoZkAUEFuH5S=jnsr0RoE%gu~U<
zPj{?J1gv7cH7Asc6E$c=SXki4vet($%SA96q3}<7SyNC@`_Vva-0J&^%%XX^6DOwS
z<O~*Zsq1jE`;g6_{$YIoe^@R6+})h3-+u;5Rbd$kr5sFo>{>Z9aY3afcQ|_0U&4l|
zsin57tN)h^pvE?4AfMYSL$mI0IDC2td(=|#)=??j$=4Vjnz?<>A8cIr)^aMQtnztG
zyGU8XMwnhz{c1f8u77K3F6A>^*xb1_;qy;La}qQ)rpP>bIRC^gEIdADK6FiMSyh3O
zTsdqu#uy+Qf*;lmIG>0z<c^<BM69pmTZrk@uyD1Im>5m{w@*{&ni(3_QhIlKdUdX6
zQ<^3)USUm52mPe|x_`?$rnPUsu}x!y?1Jwv&wrN9yxLi*Z;4^CS7=}5)2m-tdYqoK
zBu2;BLuNy7*=-ATc7{E=sS*)Vs_6f{spV^P-1n&g=eJx~KBC30J?vCj+`a0S$rnOf
z_UC6b*vU$WQ(b=CbJCXC+O(oT!;$+mOaSkYTFtbdFEMK5NI09yZgdInON{w42rkC;
z^80%EN?6#$B{MNhQ!|NvQFmh6mc~)dggF%DFAlcD!mUjrGby?0srayDJozyv50Q{W
zd6a9l>bg!fm6fefW$+9<KD{M3J-v}`5^q90DP_J-6N#Qm#ZAK2jY6t2yGw}WFunbQ
zXWyz^@A|2-5Xa)%?4729qVIecpp_4|2C$jjg;mB1>zgc$0Ymufc|q3}ju}5>f`da0
zpYEAe84Ul9M3g+&13w~MSP)9pa<JOHHz6+~UXi2a*=$I9&a~;ita{|IVVyf3?ls%A
zp{alio1f&5Yts)(kD`Se+}>0F08;3wc<eu3Cbq05Z&c`-Ym@9B?VH#2-o$;K58wFt
zJw7PNFxa3b<;x4i#hOSI-2@`dqD70WFajU>=IvV`a+Q}a0Z`n$XmU@S=$SlT*_##j
zKez>92T1Ge%TaWrwTzVBY-y>1gNmEsfdf;L$5YG3dJZ|iNI+9j8bC~lI(2l6on#T$
zTtbOLXoY`6;659h<lC8b6AqC3hyChgpKjhl_*i#NJU@|*c@&+44<o(-e11ZKCN`&w
zw)%6E!T0ZuQ*r%jIz%MWlRW(F>C^&xOfb6Dn|J8mv`koBJX+H~y|S=|XUb%2YecQZ
z&Yj?EXX4{Km|Y0{6kapyPKqm0Qjt11;rJ~J-Wa<F&v%<jc&aWfZNe*{744NkI~qM2
z8Ez7W??NFsYlF?ceS}Js58uT*71A#-%)MjM>8vK+To;fgkrrS^+Ngz5(7^dnbb}u*
zA>JuKulWc?EQ8CLC>)e@6=m4!5rcVhUatibFQPS~J_dtq_FXzMHfcgP>0J$V+l*uS
zg}q%<rG|gH-|5qCG8))>5LUJ~sEOK+6nwt6JK+fK2M3yO?<Y^?90rZQigd0ljp+`!
z?)GZUM9OiKdC$!!QMz&mU?f8S{kLOiAMv=?JJzp%UuGvZ>HH3)J-AS>T$vQMoqF3}
z;r*L8V_6@OV$QZ@K@x<40X+?8j}?I=Bu?PN`}h6Sx5-5fZQs+m&B!@AMp`85q`|%%
zkOUbWvgcNlBNC>ck2>!k{Ooq+{jVz?y2o?tK9uc4HwBmj#D?)?G6h%S>)o(-iC|1+
zfYeB&a94QQi?5~OnH>jZvh|qum#Ob|(PI&jpE_!D8tc)hF#gJgCAQ#tb5m3C_wPUL
zW-SUdM4f!z+xw7=WY3=a_wL<PB^R6Hefsn#_MV`d;};{Aowmf=&+k~x;rsc6N`q%H
z9xzYwiSKBaSJGAW#RoJB87}X`(Cuu$%%9&&PALpmpZarf5MG7%f}2|Qw)bf{g-nC2
zITMr@`vjs#Lv3wo)iIIGNzhEi1Pk@G6xBL<Tp_QzX)D@cKXi<*HX0of5`w(mBjQME
z9kw6TADCY)T(Cgbc-qlxML*AikTWre+Ui@^+%LWA=z|B7Kpi+p4w`W|piq5iX$=)V
zA{n2t^LMJ0m&>|{UMPi$3tS6Eq_0OTghp`o(?F&Yb7;@?^-}~K==YD!v58n`#mn5f
z)ia=cnx5Zi%}U$Pm4(vxJh#kTiYw4kLqps#btItvwbg$#VBnP$Ls=Vi{W|3Vn{9;A
zaQ9e?5(gO_UVvcB0E4;W)@$U)95TLtD5&u0(Sx!xJ8c<s;#*Z!TW#SSxA)XZ)3;2J
zTv2Vo+2j^;d>F67^D@`!W$H2kfNIvV+w08f;V5>V8-y`d!zyT6zZH!pMz?Sy{?ScZ
z?zhoiud_GC8qb?$E7NnxaSe_49jZR{^8rL&R5%gU0AQy(jS#(0?!J2Q;RV-=(sy^7
zwWwdLpVXNJn>&0m)X(C6)TLXuuV*Z%m66T8&&U|cYNqq&GpU1k1>)0JolB8u@93ZE
z-d}wCf*}=0UGahA^)htfv=Eq$OiUJIF^u=wKmQ2T(&ZT`Pefv(?mpU`0?wR?z)nVN
z)m$fe&%R<J$3uTt(*}qM(|HyII^}L&aS<E$b)Rp1T(Nho$;!-8?DPqLwIq15eRR<6
z+cBG8_iG$SDLd`Z)Cm))H8$r*5SDUoMaRZoidb4ab%}}`hafZHyrLUmAE3|*Hw;ck
z*>u9fo|??&`@11-?+GDKo;=C?bzn`k1yxIAj_ln-{|2NYTg{0*=sWX}+PHCvlij#o
z2U{OD(D&_U4ZyR{|90y+sL#-$9n_u;2Mjx{pR~EXY0j)^tB-H*RNiq)yVOrBe*dps
z&yUF#TNQ2i1lj<A+(Zw<teMvGq{j=9gFKoa*ht_tUsg@8!xf1QGE*sHnXP&{H*$h_
z#>#Glq9^#9fuKM5{@eA}X;w|%qiie4eJa=CQEa->kbRAJ@4lO<kiV&<so$jYGyWiT
z-<92no%}JF@w^BGhEHN+`HxpBR7z(RJ+ij%Xn(?uammyaiRyb$)#E}!)tlaW(EE0l
zwM^9CXUwinqm$>$?_>p#Xun5DJ(fw<brNdYoKYG92T%3rnvj|*ttC$t)=rk3-c`c5
z`k1bAAzqBtIUAhq_I^$3=_cc?hZU95z4gsU7*Q5f(iBvn#b{HuKexnSRan^4wGWRP
zG%ksF(EOuz!^hu0>#c70`3c3lR_dhPS&_p+>yvArUh@>3s89v1j<>Jd5dtx@)^1kv
z`U$1`G;DTeEWw7di*U863DK0K)}1zedRWnomO(~`X?D)*donskqx<H6N(#075@s1^
zjnH(H)*h-g;*6=((K6O7$HWK~E<qX!(v2JS_4~xGb<q|tnsq4uvDMJauCCv;%S%hO
z&1^*4sn!fznc5`JgSH9Tb09{CJn4)A*{EARon2RD$_8zbevBH>ddG$*qpH_Me>96%
zj^BOq(G!OyhnVq8t;(7&0bmEod+!WqnZO)x?664vzy;hJK-WW4i|oa*vG<=fx@^9n
zqNjr-TvlPu^<>NJeBD>CmrL~ewtRzvlJ3PBQJa758lB(Cz3m*og>{#l9g17X%1s%&
z#jgZRVk$;PA_CFR)6f_-n&1)6%EQRWFX`$1te%~zqNGM+__d(m>2CJ7sp7uf+-wKM
z7|`41Ah?C48{)hvzlP3I_5F?fuX}qxxVM1QG~-cL5y{EPUyZ9&RaFNg%4%4}-0Y<0
zSFn7toJ1fvV9~cssFTzd$R(RjcGk=pN&E6CU{#MpaRxc!g4R_+3EsiU@v}vu_qjI=
z0)&Jm*3FYEzg*cUQ}({9SAlq}@#r7&-{ahhYA`6!RZdGup<@WcunF=uLKk@(6`eDh
z9Bu?M+Eh$?_Dot*ASEejz4%%CEN2BJWKT!Fdp7QVaa{7q;clDnn#_sq|12#cd~N5>
zn*G$~wZ)GTsWBL&n~YRUq#RH@)Ga@V+7J)b8nZYD2rYusgnhoSdFpH2h&z)PlP3_`
zynFkWbs8vNC++%tGR-0!cEi{75s}9&&!H=)Ojw6Y-ty%MP6obgL(GloKcd!umV0tV
z;~HWKu@?m&$jC->^B1O*rqC6tsE8|(Ga}e^a`Gz(4C3%{bW~>&zQ#y5EggPa{}p2k
z3MpbK%StEX4%;}az-Z|H12!U2ceyc#Xc1{o=j_;gRV2D%<+FZ;eA=zb`jr)<yz?XP
zS7l0w>#Sbw$Fg6h8d#kl;BJBvl<laBn>*5m%fqG%NMVEvpbILi+SY@#(bd5-wTq55
zH0%jplG5Hh=8%o^w4=8yvzr^Lwr5Q0=sfN;2?WC>uxUi|jnsQ$^4mMVKLwjuu#?M)
zqRe?O2xW5}KLqKM!07dCk7pv;Se>|8Li{-?4u`3q6eB>E;A^O*+3?J2LPW)~sYbsV
z8>_>MMorb!jDcNbIPx?=AAV4q-}p_05>KPg%+rxT=IW|LYe4UYWEA2}T#DyERn&<<
zM<*0!vLba2;GZs#Y!|%t)z#JUV-P5b+;d&!iGc9q$M@<8%f_JZ*HB#dGlAj2u7EpH
zg0jtK^u&p>=Q@+MaobByAdRr<j6$grAR;%HMfD&_6kj~I-C`N|$e=8rEvMB}q%mS;
zr~D5eT>Go{!uCq1P9)NPEJK9oroSSyY@(0dSCeKIlUdt`!ic0N*YFnx0CWf|+f375
z4FUcGkOTQ8MfhE~K)H{Y7%@dZcHDLgj9Af_Go5WD60sv(C-v=<m2qcV!_Q6a`3v0h
zCjgBPb!rmpzlLcdJWg|~`Qk3cESK|5DK?2YfJ5jWBM0O!i%|DeJl4>8`Q=L!tOvRh
zC`;(6wgz}DRFpx>$p)(~Gw$)qu<7NogMEDoj_WNUj<}V<L8}VG@6RxrU{gparFX6s
z{-#p6{jXi?oD%c$+Che({n2AWp5f=tH9R~pH$Wt{x0_7w-e0S#`~fL1UVI&w3&PWu
z{|rbiHMIu7gk^h_+{_ysYoP)VWCE9VDP6fSbkzvo6g~t?HgR9CU#z?;qEInpwC2ot
z%dBD-@^GH3p`T#AtcQ#pgB0S*n$EgoY{1~gwz>=2KTw~h1D$O$>{6be5N+iIG3{f}
z%}R>G5jvnpxpsN84lOuy!egMt#;Y=e-~3~L?(8nJu&cPboXw5t(9G0)wB_v}ho7Yo
z5@jB_)<DzX_3(o=%eiWgUZNCs5Pqgahx~JuN>$YtwiWzz1hGD6&mQf#aleId^6}$G
z@Gf#?Fa_t$n@~AlK(4xR{rVAicc~Tf6@stFc#{j$_o`OJQlDcES;!sP`u0LeRaGcs
zg<-=I*7|JOv`Nq;1D%lVp-eV*J5HnlG-kMvfW<f!A<8IX5T&JnK`nV`<K`6=^_%p9
z&oTuJm_PA_z7#{!er(Z5^V$sF2C-{xxP_pWg!~KXS>VZ%$3ec~FRUBX&!u`;dT-lF
z=Y@Kox}5;$z+G$_gg;rcqEIz>8J+Q07#1>h{As$zd|C$nDR5y^vj5#FBi2T-=3ot)
zbvHLRmL1H9Fr7R1ddu2MaH^)}=AtS4$3}j$7T~fScNnuUx>((}!H&&q{0#TjtGxU9
z%=3EpmZZ^dq_FKhNHdI@^Tx+A9x=k~s;Xm8o&66V?i|o%0cAKSAEc<9n3v|00<2i6
z_J)q5%fB1gYxH-Hx%J#Is;(yK`*aKsA6D$k;2}qO7XEeZUhIDou6Od7ye?DO-?1e1
z`A)D+{#d{A0+HVF$i82BP|cKmef#y}oRX*)sHloWUejPOf($CV;UM}vZPE$n74j#`
zr`g_`e)mB_oIe?W<uC?k{PUE5$oF?AXP8=`$#acKzt(X^Q%3?v<lo@T=0j8Cb7Xg|
zspn{Z`Lf#03)i72`?JPKec!)(*K2DHSq@eZIyFJ%NtJ{xfOT6TuC&{UXdrL7vLa{U
z1W{^%g~og>_)u0h`q3kkU+xPRw>35=bl*-1($xD{f1COP05GC<;}@Yi9hd+Ep9xbt
zZi3*BK^qP#1+CO}g_fMd@E)K9UJ@&IYJ{E~ro(p_zGwQ-iFobe35M&_R{mIhZSDQ=
z@IFyFX&?cdKKjJk(Js3#XbHdj?&>heXfhIRjN^#klwC(0Nz`8f{;~B$NsBYAj!q|=
z{t4~JP6ynR89cVl*(${PisLVn*VVa3kK{gfY^D;S%;l@GUz@|)yEtSRA761)k!o$_
zN5!&zb{w6q`Kn=FPq88Ny#N1vR?`t{rw^?QTlH|Efx$WWAfY+luqA<D#=BsnybZx&
zogq{v8pYj>3IhkCA9^u!3kQYd@i{GR4GVdJoQTbKCDUWYWyg#zJ~hPmiLLC!0bmT_
z;cMGkv(Bl_e?Hc{?Hi`_Ny*9o4EUE^#fjF@pUVfFIWH$Z`M8dc-g`dmpuk&MwYN?V
z+i}J}PTOtItRi`g4WK{B=rD2T9gfKxxi;z=pvc$$03y#DANPI%;{)=<F&VJt3U&Lf
zY&uFDQ%k9ONIs--{)=Y2BLxMAmfv`p^rH4v-oRV8#>Lo7HQ)O9-fFQ-O?Om}1(%m0
z&P^PnDbj@UZ=0>hN77N+m#N&8Eii5qYEG}1R7`c)CvV8U*=DL#A!EjPW|X+L>tXn+
zr-I-V@ILQkmtz*-!$zYUh$g^4^Cj9Kc|!Q|t{hdmubGoW5s9oE7@4G+<m`Yx(7|Db
z&dWa|spFfhh&j5tx-9CeR=9Kb?k;vrl$F6eDP3}kTz|&TZ#DQ1Y@iPxt^$T)+PWk2
z0r~g-AX&yYg5$>E{FjyxNOTr@=7cQY6Kv3X?ONLt8%B?Q6l#|L?Q4ROp-f}bU_BK%
zQxBk)v==eMOq}bfRo`BoDX$cC_wHo|8G0s1fmS${%5KbA!ik?JpSU{6eIZu*QyNoZ
zU~~$T2bu%jab;=g)8+2WqmQ@D!Z~7~+CtW~6~`I>;xtD`Mk=~7%9u9mYGqycqemFK
zKrOKubPA@~iFq;m%6zaQV&Wng)POY5#M#-~+x)ibWG7nZy|G8vYai`ja8n~UKoMM}
z5_e(IvV!B`@fAOg8{NE7S+pdu8}BGOo0>>g2O_htZR1H9e31vnjs8(zUpWb=mPJZm
zEgf>e{A(&IUOSy=po3*Hk@(`pj{LY>MA?oLPRtX0zPWB+rWQGjnmqaQ^XEl?hm~z-
z6X#xkWg0iuxm2d^<A}?fg05%h=1>R^T(`KNs2ir|C<}?W#87tg1;m887U(vqFcJR7
z7vJ*vwF4j$;5<H%cC7mJPAC{^03BiJQrQlJElf8-2c3&kWJ-mtJ;Ks%;0?4is8M+7
zJ9j>TMNT`O?#I0+BR}qcSX_Ixa9>5G9|_mEM|ebsIsl{fbE-qW>bdt4-ON`>2^G0*
zX;<fRvQ*?oF+yOODAAOG2g$M0st^)U!gL>yBkBvxV-9H}K}4_0iUq+VouJmK6T9ib
z<2A6q!d{Fb_`!ocd}JNHW7V%BwQ|(r_3FxuPqrjfWzGF=`J;y|QxI!5ipfYob?pCr
znya(ieJ1&Z-+D1=Tf<sH>JrLLJigYfUVTjJz%h?um$_=#K*h0)VEp*IA8qe5(Gtwg
z6lNexHpb|Kd26elNM=5fk4fxZuj4#m?mQJ4=roBqwEw2g+JpZ-3r~K{0u?!`EKCtx
zWPH56%WjBY{@ncy2zc?~_ht1Iib`%Ia8%1Mj*mXZ;XNuQ2FsaP47s5|&~V*}N{9QQ
zEx|nlU0zQ-yECT^(I>SyttW;3w8hV(G}_e;X_xNPnXLBV)X*`lKcvb^7S7sq6LCys
z-r^Rs&+`<o0N1m9NybBW_pIr9eg>U%j-{9@OtWTTm{S`&)%)eeA-(~R<(GQ&)(@^@
zU31UD(@0gjC=YY$?hnSb6uUiD)uY!OJByx>1+2|$Dq}D~!&d3pldh63PO~G$!MA=k
z9br6Tx+Eb>S#jNa39djv4F}dXSM!nO=UYm-|J-Z&*m2{u%>~>0XJz(MM%lNwPIvP6
z@^ZP?hxGI2Ro+Ar3Qh{3&=8@hokG_}NGj<D8p4H!goHz`{kx+w=&eXoxa<P49iCRR
z@Ya39!gFRXX_^Pl8T!wS%^Blg-C0v~sWg5E%<YW|v%eVV&>158*TDwo*ctZCg0cZ)
z(QB;vS#GbY&~MO*esXf1GBrzwt<KfgoE!X+iNzRic~epSj6ylv^iN~Eb?bKNsg&j#
zDLGrd)pPdOkTDIW4!NY;?2L{rD@(Wftm{7qNs5;O_FCK6NO$iZ!X7lfo`SZjFU|bH
zA)${Rtqh+qA>_fN-(E83U+BKDpFGLL;NJe^lfAZX{nT?s+o1xtHWk!ZGD_-DmAI|e
zmEd(OOmV6$m~D`hfben#LxtT-db)@UYU<W*vAn<Va3ic6wibzLyRMg<DMA}-L0Lsu
z=kK$3^eRH{U;ZS5zVS`1zyGn3)nwj4fxBnNA{<qJrfOwAj&<%(P%>wqJcBn!$^QX_
zr*Ff}^&wZ5TpJll)<}*ldm<sx(D38Gjb!My|J_J-8F{If-Rg4z;;s7f$gYFOi~{r;
zx203+5Cbv?b|<E$!rIs04wG<@Qn57^y-?A+n2HIOUcqc<m35s`qkq1N&LNkdwbtbs
zi*$(2&@mHs7DqQ&^&C?N-t+bHIJFTYl4si7K`8N2{Br7vuE?)Jlzw$Q{&yV_jQd1k
z1;A`?AM?^mOi9;b=b4!{OK|&po^g*{Us96o+!XjZ^;AfZ%DX6+7v0<SpAJ<r-Tae=
z(r%hUKTd?lr9mPO93o9l{#Kj%xpUhE>+Y?{QKvl`c`R3j4OpVmxp#(FP+>*>>u@eI
zLu8YwA*H2ORa&}!=X9VMYwHQ7VI0kw5WX3U6H@%n3rhoq-9_)GYF~dfZ~p6{({)-~
z&wn2<DatYNI0)vz=ld>nQGN+l`0}N;#JDvNbf3E0>p03a1%JHLpjcFJY%6wxP$SbJ
z>`a;R?Pp5m!=B;KoP9A;Ik(+AdjGEeqTFEj8}Fx1o*arZ8G0#_`OLlZ_^{eCh@y;^
zFK<B)Lv)6zMsxD}l+}(M4tR7V&*LM4dmTS6X_~*n)byl#p&AG;hu>3X5W*PZJC%9L
z<&GJ4^?Fy{2WqX$|A_%*&cA3;E)6?mwR2~Qf?DMW6r1dPDrDZ_e_7a=b<+ZD{KMuM
zTwy4<Oy+}@g}6FR5-Z(FfOohxUD1u2S6H8?;Mon(BhvGZ?d2H<_b^aF>Tr2F(=3sq
zqMV1D)PrThw7Xz@GEu~{ZMMdZ;9z$dEd4MJ{@UJ;L1=~2!{!tJZ{V<gnU_wHP%$4k
zuwZ+7ZAxCqKU*C17w^<?R(SAB0rjA#nIIpnymjwhvRc=NZx72f`np(2_UYgMp8-9p
zAEHaDlf3g$Um4VKu<?ry^G@&k+a5efMD$7(G5OB9*xSWvey5@9H2j_FOP4aq={`f)
z8U^YsabuMj2gDavjeAe|qcYL(PPjF{-?-z!jPT`q-Y!ea<1&`Jb^&*P%ab2E^wfyG
zY!WCK2iPW|1l_1Q&vudOPvoyfy+e&Q1b(<Mx$`=c-}`KWPYwJzdHmhFtjVR-c?oQK
zIAT_?#j0|(Rywui!zq$2`b~JN*4kVT3bL+j=Bhg8nRJdC`tD;{J%!=k9euw0cGu1i
zymgCBNr~G-YJ{rg&g!%TC9yxncaN@{2;UgFM3I)ZD@aO7)j1Y{=IAMJ5e|_L_X;7#
zk$@yD1=XO~*d}Sl5Enr8!3y2HsF{*YaW>77Yo&9&^R0F@t*oA&gbucVp8Hdd1zZzU
zY@a^em%d>*lIHEjwq?Nvnq}k0b>8~ME6VX=hJ9c~`4s<|*<P1><>uNa(!JJp8%0|J
zRu^StX3u#L>Fdrmx>H4pVw9*Wff{y+dCYC)$`Q}rzU}#Vt8}kkTLSd*^YRpAWYE}5
zD;mg}-FF+{G$bueAH%-M7LS?tZ0YoxoH>Wv*+9uuqZqi<RDL5#Lg+uGq=spCTm0V2
zBc^M+FjlC^n=1jR{LXLUS6X0t=-Cno{(Vx+2c0+z-8disj{X2F9E*yS4|(z0MfwjR
zsR5WkwHVoF0V|0yXCk$utSbibWK%|6fOC6FYHG67QGQzpXL)PxLodA`EStgbNN<>V
zj5UcS)t=T<K~9mBm{>UK7vhSIZ*^&KW6)Y1(OJXDmbnUT{YNmT)vIl>BEo(c7bWIj
z{8xcVY0yFFWGl*x6<|tE{%EmEcF?K!BDZOh-v4x!X3!epm?$aUR%l198;vIjpAvI-
z9>H&h<>h5%ZDa*vsjZ|VIpOvhX}NY^_S6mT*Y8_N$%=LBFg~qesW#V_?1W+#yh*@c
z248HAPZfFVh2OuAL%rbj<?H*Adk*~(-vBu|IqKW_Dst0r6zxs^NLP%O62k*F0ae@D
z)dutv1(sD*P+3i8um+7xClCf5KTKty2)^H3xUQWhnX-P9g+)_Qq7>Dwn-{c3Siz)r
z9BPlq;P43m#<Ee#9zBY8UC-V`v-!dy_&L=A9H==Gj&03-vyAGO6P4({@2v2=l8<G?
za5;}fs=oFUN0+MET{$s0hST*Q+kgLR7GOjDI1kK9+~iJSlQm7@5oUYUp!hM4-c+Er
zwvjl;wIorq&@fv6t0E)8(FanyW7o9uV{rcY>C4-j>g-urwORh4OyJXwC+*H{g;REJ
z96d(r*U^}P9=doI%$6V{!f|LgFl}ZK69B3%|He)K*_}PjGEu5o|NesodU{XqVY5*9
z{d@9K>kZ_1kRbMijM2Gy<%%}z2CUe6d@Ukke{g(PA%HJT2Ozqy+9)Oex21t~WXx8Z
zc<27_7Zx0^mlv%HYBM>vD(LW$GTC8AoOgTDb~uu#VRcbo+Aaiw2Z)V}YiH^ut0Ug&
zSyomvuG(A+6QH19KDREXZCeqzYp$v<cT!dd_SA&TI{~0%s@{4tI|{{oaBX$}{_>f0
zSO>5wM9!H9(x$Vzgrg3Q1CV+0WUoOd<|)ds-2|22CvbQ7qeqKWU`Fr8|ACty(B!<z
z4?9h@)%Qh=8$Nx?%Uh!C9Uc+E{!KhpHtpK=ZufXvK7Jig&qs^!#>N8>I_XPTo`G52
zKzz%lOo<yaYWBY96^l<)B)GesJn1cwqHN@9D6ca|c8|8jBptoA(T=}=+$gnpl@oQd
zrtbwMIgiMj^8QOYYRr47$fYDEZsy+Fewz!2^qA~0;s$2%1z*1=m`uH0VemZ0=I?$&
zsCG@lo%{#3AXA>g?a$yz-~snBKfi+jO{p&4X#>E_oosO~1A|eX2hkqzuc@mqz&A4=
zBJ%FH4usS6l3_M8I85vxUj4wFavTB}i3w7hl$7V?re9rjT*rK{q6=-bF){Uq&LHKa
zzB}M4LVrlIDypgv6vl!WLjxLb5?#+5yR!jWH*#;zy@q|`ooT_n2hKRuP1&Wzzir5j
zLmge~YkN1vxu%?sax`v#Di-o#+2u_VZoNhXrw2HdzwZ03<Dx-<y@2FV3{2(y&FUEr
z=NPaX@IJ+LT!fd8&&CLE-*rJ(zCZ0-HdS<2uSGLu)}j87j}5WeKR;(dFmffwuS9ei
z9R}YUFZUu+Ha*g35zf<Rv&-CvHksELEqtvet)M=4USmn+kbUm7!N}S90W*jn9OJi0
z^{$Q<(xO@OpH|)L@+Oa1TR$1Z$aYd8|LAICYB%%Eo0;=iHC}XxG+hYj;TKFlDvDbX
zlyUD&e>H^(R+sBLOxC8JoU%dZ>*w1iGVXnBQr%WyJeeQ{*&9dsbUWL#jQoqLyA$hc
zPJr`>>cvx3YMD)JUVLJhoX53!C%~*KON>#*4t4Wtk}7~x3EaUF@NwUE^N+j49d4?L
zKW_Z^)|V`wj~_rbEG(zEyE`AOb<x;+!!MX8=6T|T35yV+-F{;$2|yiIgcDD7w-Xdj
zrUw<+&DsJd`!jME1&w_#lasqCsbkA~a*v1bU4|Du@-_z9rSE_^L9IxR`d(4tDRY6j
z4LG(<<rlcL>~Qje_D$KP=th+={9P2O$kuk+*k_$YqT08LpE6ztXbf-3wsJe!%XE|9
zty`fL8IL=fhAo%dTuy!H)Kut>Er0r1YHHeh0UVq$V{pc$&W@Qhk=#3GR{L2#1XVh$
z=0jQ<H0W<P_rQ87FjlVaZ3_g)B*yJMPZCJ5ZXR@UMB30wjA1}>kH|Mo-CIctFQ%{B
zx33;p`9%)|N1{?GiO|p=g@w0NRrkos#y+sQ;i_{x-q~^2uI!`}aw&Ph&<|P7&{*kT
z!tlLGc6{mA)2C0fgxZ>wfpqbIK=eD5!z^7!K7fk5w4&nI0T+ICw6}P5+PwMm4KApc
z5y^UV?QAs=;uBCYjI_M;^rOpd+}?xUVYokc)@$r_LBP~$wNZf8Z9+dFvg>(fwEgb~
z4i+QBZXEqSU|ZRNsgKilW)GO-zM%FE=kMF{p0h@cD&rS2zwjv-INOHOm4y{x4Ek2x
zc<NVzJzB(>Uyd6$WWWH@UXtm-S*+*+&cO~0<~I3)-j{*@rI*3N6o-NlLm)<SiPC?b
z&5%Vm-c)6GW?e~6^qH-VOBDA_n?8BQ!4HEK#%ESO!DwmbO#cqQrY76&mu309BA?+Y
zE0#^_usINL;?IxnNfs-nP+m}HSeSL^0|odvwXbXc9`M`9Z@$FRw{O<|stmkve*b(C
zXtXquS?e2=nd+0vg*&9J{9-|Y7L!5PLNa^sEnT(hoT1(&k}kv@VX@4{jZEu4FWxFt
zK+ztbE>6WM<8@Bl+pRqu>r2yJTzXARzp&pUwfJCQlJM!SJU({jDL#M}_%9<P<ZS_x
z`W*MjWuS?2L0E1@44j?xy`%&!=WZ(jA94}vtTM!38&Z^gNv!s=uF?uQ*-D}?842+w
zrN-@il(tCq=9XU;zjp2WJ*nC3yx)@2vOqH@@3qlJ9(qRQ?^e(IcR*T+pzx}0>X-|}
zJWp<JleSe^sOLz<+17Kj@}DIxavl*fW?5MJ(}nu_IJxvw5}aY2M&`gxsFr_>AfPFM
zQ+`D&By_HmlSREnpZ>}J*soeXC;8O1sa9Swn#v2VjPB`7(MFVQO@7lSCS`{3ZKF>`
zZZANm$gJ4l(zHTWtk=oggJANV5A5ChkGguFT4RDWDe&DBe<;#r`#k1lbW!%!V{{u5
zvu>|Mvuf*{%0h!)UW$}-t(}~XwVlHn16%#tgATdfBEn@(4jFdT{n|aVgD007uHTjS
z=1X+nQKOJkpK{P8|1&8P*41qnonO>gSLa>Op(ftxke-UNTXVj&Y^!v;((}Yc9bHxh
zJ%5#&YOru2UE)J}edG<Cobd2{HAlt&QzhIHbCsET2#2!iyNlfZ%C;DO*LFVWVT&+W
zN5_@^58L7{9e2U%s7gHT?IE(^t{Swlw;O|4sEj%{Ny(Nq4gV>?X11n?$pziI1xN<@
zxmBj+9`s-OJ)U7+FWm|17b>qW)TwvBmS&jhG>vf^-UmO)V7c;ga+3hVy0>@eD0|bF
z$msNu?6NZB7yOuILbqvYd39OHlOa+vUDkO^-i2HZ{K{0VQl3d3mBx<k+p~?ADn%++
zbwtN1_qDW~Is1+AZXT{gAbcEt0qcE8v`i8RDy9{EsnkamctgcNqV&f34YMKfHEHfL
zq`->8izn}xodcOaVs@^rPy7BgW@h=>heWbrP9}8f4F|?*>$Z(NG$J@?B+TK4>LCFW
zPp^5n*7a~%bMwr8eIGO1#`P8_ERUH(v609^Vg*F2Vn3ddb6T2McL%K_y_*O_dIc?`
zB9;5T7sMHCUPn1sZKA7O2QqHL=Bw2CN1pZF=A2Q|H^i`eY@|U!RaI4HJU{|J1^<}V
zikmK3ax&(DqrJWToH?D(FH)c{(z%&2B3wA09}F&!tjwO@T%iVH2&Y%POPZ=db=XZ?
zfego%NlHd6%ZcB>n3_OO@ag3Aef&3ziLN-;{rXj-4vsx>z`yvVIkEQCa9*$aE=BDT
z)p7B^;F1V*5_xzPOwOl4h@p?)rJ=?jRa6XlY*jL6PE4uvR7<0ZM>}(~hAt5oP}xZ9
z4Oa7sBaS+|cM}Cq{q6yefVMdw8Kl8nagjmMjr<7#k}!^)Wx^gTYkw)5898lM!}kFR
z?vHn6eu&#KQ0~UGo}^-hvw-&{ew2_pour#P`KbwLyu_i^W@<L|H!F`wiRazcXU?+I
z_)+*h)3++V_LHXY4=Bq7tz1G?Cav2^`<%+9r&N?27?h+-X3yS0lgLN(uB_wpx#a0s
zZ1+9!@99&=6xFBv7?g0@cl)m&mbu{`qsr26@tdMO<f=<kGKgDtigKc8l`!#kswakh
zdreIhPR!f1ljuVDl<L*%tdV^bV14elm`|wVL|!unEuJ!7@RS(eoPh48ye4&kMtHYj
zL}JCeDV>Z4^fm7xn&~pKU5RapKLmHkY13dlxeDBG&Nr5#o7jVaO%vsKQ2;YtQRxN&
z2}w^s-rnx;zy@5j=fHt)-KFO=Qirv@Lkxfsa>7;@HkfdI?Z+yJn!L0-#e>!YcDs2&
z8f5^9BSSk<U%@<x{{obH6*(-$F**ja_HZ80U%8U(g_dH1bGvBOs%0CL?oBjW{xK;i
z=-%$#aH|z01gigE#SpipWHQ|yzCHt_jd|gmOKEZPY+L_B+SB$FAA{w24}Ps!*O@;S
zD!7dIxcC&BOCz<8gc77CN;OIyx}daY&o<Ecf^D4g@W5-9?rE;X@IQxcfugzai7t9e
zN}kTmG5YrI%r}z@esG5~nwpY|&2L2=G)PJsdSV{Y1TTa+5{u7ID&f03jMwMYuIE><
zjEdHAaCVCM9z(<T41es0j}}D&3(T4_rJs`ejxYBaD{ix}u$CTxM<#s>!4jT3*(-ki
zvrZFJ1>AdIfK%?!lh4-QVHqKMQ6=+@8~4CLJ$UeUUM~)Cb2n;vT20FJS@;ML1@-mI
zVrt@yV{Iqetd!PPlbd=-n7PX6fPq)3+b2`wQW3y3qL&)>t&G3mvP<XYy>tD8v~s?^
zt-5^K@xInz-xRoxKWT|&bn+d4<pS|wHfFDhw^FGAS%zh0*!O+N>m@F_{9<td#hR=R
zjEz=z2s?^y94O^KzgXna`>Iw>(up^5S^A}@BKYsf=~y7WMSQya=qwgCik?oLpCb+{
z>L<V^tgu%1Gw03?b36LK2zwJ~Ecds2{7I2R%8<yA&>&-(LPU}j8A7NK8A2j5W-J*}
zN+m<aBJ-TgLnLKLrpl11%yWnczkTa`-}S!lf35fbth3HKr+A*{zVFZH`drt(_THDg
zwE_Opo$5BH{X|5iP&sXCs!I@OaL2cA7dDbHSNYCgGO7G$>(--AlWg)=V0(-_4B~a{
zsbol~Plw>jXel4b<h*Czq6n@}M8x^^{E0Njc0KTGLkx(&H!wazDkksNmUkSUf4pgP
z(5s%kLEH<IgW*5o<GouF`K-TIh}_&yE5*jfA}kzHRkdvQYZtovbadXdEO=VDImA4;
z`}LTqjRujU<1S31gUk*ionWsBiDSW<Kg?}Xl9Rtdd6k#f*3;vO`D6r@B_aav1HrZP
z+NyhEO(SN{|NIH}^;^Kf!o&5{)zxv}VPqFh5qxt{oZrvD5h67?8Lb-<=M9{ar~rU3
z2lW>q$Cq81dkpFmV8EcDk1!{PDbcUN23CVLmdYR^Sh+?altY2`y}uuU0ym18j*hch
z`=j5RON+mMLnP&+z=a0rsrnWo@9O6+b&kuj;fSG{rR0oAwoq25b*O!^EAX@^Lg3LG
z!>}JyQxiILDEJG51<U={s7+y;nQa4hJ6@`1&cJe_z;Pl1#HAXiy!?Dv&SEvtd!C9@
z2U!QNe}&h`3Zvbp!Qu2+{cV13as@dr&}I0~&Vw%S=<a6D?H~nVbdHIz<ZE>Hdf>HJ
zRu-K%u-@1#4GaiqhZ`8CXoh_!f@qms!KUctpaR@2JG-B^$f-eh2M=rovn)WC!VaXT
z^Z~Mk>1UWL0h$fvIy0$+eitrySm5l56F>d7e)yYcwk9DXmtteT$<G^%77U0(n+Xyg
zdeRd27DT<_wo7APhpsU~7nF6I$l`FNM~{}l2n5&x%5rK-%HBm{rUSCgH-K_FJ8y#N
z2G(F9_6uk*$yzCJg!k|`8ZfsexCcNM#<=^?vIv{Jhimmd)kqZEu28EY&mk~V-st~t
z<8X@&j`!6Z(>hxD|DzEK3b--*$*-W9k>ogz^_KZH*t}xq>%q_~U@W*hyg%vQVpjM5
z)+3BD|0Y$#>wSOn`~T_t*R8ry%n5<Zroa{a!Q?-A2=W?W@C^+O{Tz3J61h1$%T8>o
z$pK5VYcJ7vMkkwKlY4%{+>l+e9`%U>#o5`{U0ez=jRMUgimbLiNC&fDzXo3^3Jg^1
z*UKy@=8z&$a@yO^A|~@4IFRj_4htw0Xq0ha?II84@(ELU84{vxWE6+@0T&bAEz?Sv
z5IZsw{^-%Ia<m{u0IDbiLbV&E_yUz2WRpM?vxI&owu}HogZx-qLjy%-GzK18>0qb^
z@C<AtaGr5DPVaksf_aSt>UA-UMMI{<8`9C4P12ZK;A@7o*g(V>V`UD4T7`TLu4kVo
zq#bemh#fpAM~SVM4%3kz%+vKRarrn<G;&T3HbNK=K_rekOlqP8z+we}g&_V-zd-ai
zu*GrL?zW!~H67`jH+;@$*`z<!tBCLO)%o?2O(F{e2z|{PJ9N^9bK8#;qXu63{rgb3
zO5{q}%>_XTiLfu1sP4!owm_eQ{=nuxJ-ka`I2Fj_H0CHM4_4ZP5-rDkiZTwO_+BPF
z@Wm5IY5!N((sOy@@Z^94SIzMkO@spM^xWP(drI7wcW@G_<zH9XI89(5!p!VX3u}g&
zy-O#Uw~>aQ0p10;cM}R<4C+N$b*z|gnuD`z`w}2TM@=?dkJnpw)|Eaf#`V!0pexfu
zM+7Jv)eQ1TAZoB)w%vy56Hq9ZY}LMSq0TB4e$*cCKB{VJKE7Osz8;-CK!GTj7uN_P
z`2a%<IKR@KUFhc`++4p3{)MEx60kkYYU}FhpPbx9PY=)eHT0#V<*?$t$}rj28oMq{
z$U^iyx_ZE2aPq_VlxWdoe_<9-z(Ox6Hj_@!8;{Zba5WM6l<04BK7O(XHYM9*6cZJU
z{cazx>n`2$+V9;8`Rz9-vOyxyfjuX%w5&@T9)AFz4oxNm>)y4}|JB<>*RVt}$E+lm
zn_e<;s_*$6djCH>z6Q|}qc%^H3E4^G6zF1|X~%39^p6V{&<cR^4oC@jU`_8fl9tG@
zG(m%&oWSw|6pdcK8^j87a&n+)S`h=SFy+yMv;9)h)O*s6Q(e4s#6CCVj*8Bp-LAmP
z(<|a@Y1&55rot-nvgY}tG>ilRSB#PqsRB&C2!D2P2hqWnmM&?rfz$+nx%T5nLA1kc
zY?yxVBR!=vVO9;ma_Sy25ZJYFQo-MExleX%9Fec2O!>kE5I3K-Evc^G?W-V9;#k3C
zj&7_}j2?kq%p1##XLi#%iw)9hF0k+P68tBhVpn7E<0rs_yHhN1EcHH9sQs|BcUYpL
zEd*?TkwX?G?!|{V<JQcuY#}@VD*rhYonC|n$~aIC5oRZ6fXL&ZCIk%#RU<c+00l-y
zMq<dVE3!h!lxL*^y{fH#C-$VBp6{>0)by&->^iPIX9KyDC;hJAsBn)2(5|b??7_~+
zxOY51|J5s7rw+2%MtVVSoX~hx2iCkj7N?F5mFcBRn1MxpXd5|h?QWm=yLX>_PSXe%
zIT^8#Zd4RUC=(1PyAxESIv!^`Y{pUXO9(VKOx5z9wZ;^YcWcFivLYh(U|`>n0({W~
z!452HJn>)H1DJIU{)>P|x&wT*d_DoNIP&eno4mWna%KQGIl1n5ih)I+57H4!;V7aa
zraJ8Q!4}rNuZ%$!$RBXh0q3{+0d5#rr?wmFK%zwQk0#Sr<P}IW!1=%@-r6%8wJRLf
zic848a=^U?8>*=(9QNkO63}I=8oL4{6=Pcns^3w!0vHLwxgq8a7h5@gxTmKFOixWs
zd%Pbys>|>{l(<&n0s2$<3p{B%jy^CH-`<q`*xcHB<8S{GUWDkVB2#e3=0x|%?bhag
z20?G2K66M?L2Lnam63_*OGgLq_-_#2p%WCb`|%z<Tu`QLkcE8Z#_cAJ`*X<hf*tg!
z`H`BF6Xw6!J3Cj!c>W5UoP3UNe;1BE^1~?W5Ex8)957bLhc;XVG?^ky^sTYKI>RB;
zcQGGm@taGwQv$saG@R|u(-XImlR>#KSqkmPm&h8K7zHAtG1=0c;SE`W4Z^W^9AKx^
z)Bt8?_Cy6v$dTakf;bn@k~M)01;+?(5FRyH(0Jj|0zQp^?Gg?s2@2GBpmr4%79vaQ
zMabDsm4HnqDERT+yM54v{rS%AqZ%!3VQr0=22RuGrY3K`8<XI);OI(I<$CTi(|-rf
zwtz0sQ$kM+dW>I?%Hk41ga_}<-`=Dm#@4DA<9o2^;p>N^W~TSeMtmPc3Jk#n+58@h
z*xfG*#lQpbBS_fbBE#Akj@dqU5w*%aWSlq=p%*FClUq@Dfbfns92nruS45F;VkLs^
zh464g@A4K(|LK9iKLHFm=8g2uYJFx~*}fJT9WTy5{MhNioXtBlLm}t_T$*{YP#B2g
zxigFZB@Bsh{3{FzaOp<a5#;2ILr)iqix;#jDF5s9pQz@{AS;8bs_5avFhe2cS6#LN
zJy6gJW0JT!oJn%!5VJY|>Bo*{W9KRago)L?+h^aI2vKNh5zxV7hHWYsyS#@FJHg$t
z<fr*6*l*=m+l76=O#*8<7z(-oM3@ioFoY}}yfMP=2>NQ>@A=l9J6Ku4N;r6_0u#i*
z*eBEq5fOL{Em()Pq2!Fe7Ez!OuB>$ONPmz@fj?qtVNnRy7g89!QSay|@#CtY`y58_
zsG%qa=?|`jkVuNa@9*(r<S+qu?-Jc}WV(2Ns5JM}1#KzXbA#>(z(a&DS@d%8*8s>+
zyrJ;sl80QJ=n$d0y-h*GB4NAxjyw!Y@mgN!@q~Ou*Qo_mfWhzIi5d1A#`+QQ%OAt;
z878Db^q-+-zzd-fY@d4k*<#vHMi|NC1-A5`2QPPSzuPrcUdM4%rMG1<ROQ-*!W^-K
zS)TKL44b2spx1JvxSj1Vb`H1<sAz;Nc@_H$J{Hiq`fg>fWl@V$hqyM7!R&f;6$oZT
zw15y?Lc$pbEsRy1QGFL<#>Rcc@61B&&i_l%Vq#|Yt^<{wUnXQWuXHEI?ywVEPj_Az
z6;ct~z5NgImY}?2MREJ=5IBz->Lqd);<kW9A|vw+orEOJ?Op=$mR25`7eukNXOy6o
zB6cBVgf0QQF81+b1_K^qekw{`P9G$o5U}XxdSI`>b#VyN#Q+yKj5x>OTc2tMndkrE
z0z5=YiM4h;YP-)v%&T{HUPO}}O)!``6oBI&*S+s4mIu%>pHEOe5UR6I>=*Roahm-?
ziYG_bb4DRe-UdDiMn-6NUk983IzPZ!h(-5eqURT95I}Hk(2PK;8&Likh{;kK1}X#Y
z>hC%ZtJIG~_LGZD4d)zw3dkwQRshY=H>|2U9?5`m`+%tE4e<8xPV{$6V`vLxFOGGW
z&Ut+0P*pKAGlSIoJIX0Sehbkx8ZMkZ_@eL@k&9Eugo*+LGQuAYF!7EZpymETg@Tk3
zSvRmAq$>z*TsKZsGREVR5tf1cLCk_<$q?xhLO(WcXMOz^okk;ll9Ko?`T$sf{pjm!
z`t&Ia>Ru=oa3IkKyhYuMZ24l~HEf@>L=~d`oTE;w-EFwo_0{x6a`K1T?rhvzw7nz3
z!!>xG5H>l_Dll8Ims|n|%3v7TrS+5(W8I))^Z<h$R6<T4P&Iurci?D!Q)cETw;R%_
z!`$4rFAI2O)jYr^!|tGyhqvq{)2Xk3SfF{u7J?}1gmUlotGIz39TDyX#UwLx3UNTs
z$<F(2K%1hfvN9h=3c_^*X985DXog|oqn^eMLHi0l1stSdsOQm&#kqAvKpq%a&m*Lg
zaL<4tCY<R}3V~$t#9e5e<=c$4iR>7?prgPy5lXwM@?z*;$-*rBRrckZ{3&IFID}n6
zz=Mz&4<00h$fziQgb2Lg#;R*+y*+9XDYqLt>4KhKpJX^ZYhULO&3f9tco1dQ#g&;U
zDR)aBe8cQoSf<@&aE2LGiAMhnSfv;!cnR`17^|#9SH5}5dU}$WhWx_)RTh!?a9AD6
z^qLUy_%n0z*EoBX<|p6*xK1RZK1IQPo=wOC`~QUB#yn}TZ47P&ZSI}HF@oFH_5Dlw
zg_`VlC%=CG9<&EVFO+U<w*wT6;R!)t6AjbTKMsKP$jVwrw~gOx>e{1IDKIlC!LZd|
zj7(@18Xk{RNB6)&G~(X9dLT&!Yr7sR#bbMK!5s$qU^~?SNM_>}8+#u-d5pR5{0;f9
z1sdMQ@8ub0O*$yF4-zbYe@f6MDQ?p-GfM!gfS8gEZOzZ0#f61)Kqo|pA1P}%3`3#i
z&*+2{3)Tc}V2G)*rc+-*t%8;-1ndo<p;}p?RWkSU=XnrTaikM;1n<(RC=C7tmm@nH
ztZo%h(9+UOgu{UwU|S&jjuL;)<S0#weSlb4DMUf}`l=<N>c4haL<IVOxXckMZtN^+
zFMtf(MBMowN&VH5$_&(dFJDp+6;;<rQ$gqqwS$`vZHC5S=zUbDuGC}fNyS3j0VoRy
z+d^2lZrTL77x2Kw%1UDd*n@3Fn=3TY8-92X(M|`O)d#}-_K|NluEH@IA~P2pN2r{j
zppuO6-f+z)-zMbNEfb7(p1Ia@=CJ5hnk|2P`Ra3TbmaIy!_inKilHdxHDzamzu_ia
zk7%gWc<UX7G!e`D$Ty6kkoZM8^Wz9xfUhq_O)h_2KgJ34Gj0n6pW5J!;BKQD<fX#W
z`fQvSVZQs=I}|HUMbJfTF2|4v22SrG2i+1rj1byZwRU1{Ebxsr`n+KHXz^;l`rdLy
zw2NOevNOjGnJ&z~7vV({5;A}>jYC7+KJQI(pd|<Nhyw)Ye$n?5(m$Pw2$v)wq+XPq
z!r|kfxO>LqzQgAW+~&fY324<KJ(4(b<N@npRdk0tUV-^}4MA<3ycAq-uszU0rjrM>
zS|Dbnpx_hVqdvv$@keQHCKxytIAc1$Rt<-@eOUlb_XS_Gx%rX9h6Y$XTxVQDK0B{?
zh2*t%7u=-pOwL>n-s{eecDgJa9@+kQLEmhPPDsriab2|(motm|=>ym_|M8K+h@@(A
zh2p8;g1^#t(6SLs*I9h%`=fybkw+Mi4?M5pkrKJC{nK5C7mtAb0_j2At}SSQBM{wD
zuEQ05m%Z$kcCirLzYsQ73(*b;#gHz&E6c{&fk^vbJpQiK`7&*?c|T1Ji~0GL2XlU7
z-`^p4gN{y%_%S^@dwD%e{v)AOKN11ZwFj9htvs%jtXq&@NC<qpq3(~PK^d8uLRutj
zafmP&OQ_R8waeX#1Oy5|P$$MTr2P;pgY1^%*$q)&=<B0pe0RAlh0?i=e9z+#>m+_B
zLHQ&A@96R%G$BU5e*L+w4*!>0$H%ps>^g&UoChz$4s5E>I|XEj?1?F>auYZb?x-57
zg$SlW2b~A7iO5ncE7-MTnJ`fL;JYp^E+T_RN1pA7;}#}gID3A=VFI-&UTEkd7pIk1
z=+75^aaTV0{)13iTd((t+i1hP%5GSUsmUJN^lbk>Mb8k^N=sL8`B)Q(NH1P|2O}4}
zRdD!xs5^*$ZEfWSQc)uTsY8E28zFP^wU<>3{u_(xV<`sAZY%E<D~w{2ktv0H<lmCw
zT!Sh>A+?$Nerpz~BF>H~>!Bku9Z2&ecNG$Q9r=_8p6Z4NIhkO`6<+`DixFTzkiX_T
zEk(u@_t2C*h=u$QcIqW3@H8OB!B`En5{QBt<3QEaK;b%-Xui3BxQQC3wXd-nv3CfC
zY<T!gQ6KK#%GR;1hpi?evvAmZ(W1pk3_SxviYo-0{;+iiQrx>!FTvjL4OZaNIDMMR
ze0&qxF>r`ZpB}wXu($To`>rw>+fl9R^b0O#Q@}r7Tt4pTXzwgxGUZ+i$_D%jloHND
z3~XX#bO>Y;;4+{MVh$C^>A)tgC~hPT?*lG~p{k2{iHV6`J@YU%>OVC&jt(ad8(Mi3
zncC{=w7gp`uD(P-Y%$)8{s@Y0q94OhQe9a|#`y#dr^k=0WLBZ6Fy6)S05gik#&&L`
ze*6K*WkX}*P&CK=)ZbYzT1IdOhQ1ENIstlVjr|;#^Lj3aFm8IZ_Zrz6k^uCV@KCvY
zC^y}LF%S$|(Q|b}?F~d2)Q!!CgkYTzEFw_>X#zX(BqgeM9HZY)ut0d;lRa1%aIQQ;
zJ`R0!u#><e2okeF!i42R8HZM$KZkkDUZ5?;*6FXz?rXV7qEE$_Dr8}CW~gy(b8+#4
zJ8mm6VjV3RWOZQWgsYs>)O>~<3^?%`QCe^-&Vg3qWuu8h6r;*KK9(SrLbYH}VUw1W
z^z}82zqVpZ^VGz|{g%fd1yi!($HqX4rISAy#P&v@M;DNq?FSWj#}P2!ecX5FHmPfo
zACIoWnHA(Z#@R(Iz?zUBWE(e1HvhyS3FwmB$LwQdK>ijMQOX_ysb{%4AngP4Cv)^z
z%~mEBFvovfIro6J(n*fCGkw!;z5ed4H4XHgot@yu6TDqn?QyiakP!fcGBz{wwc>}r
zBoxx98_^*`E$ZMpgdW@y)45>DKQ+3#Cc-;)Y$vvtC9bA(4Gi**cYZnHDRKGx)7ag-
zI%@2n5@`e80_e_C3)>T{wD^;5^2<ij+d0%lz+gXzin2iIAdb>WM1)~~9~uO22lUDJ
z@bQ7kcR^kKAc{eD_9zsCNO}5y-$1(`<he>kC9Yr~+K?=xEZG5V_l*(4e1t0)Zk6yn
z^z)=a(H%JdsT1gYL&F3%8B%HxZtNT!K#5-kER5xVU%+EKd5_t<F3{i6jgc_~IuaPp
zqHhBT4DJvh0VTuV3U#gag&RmVoDcs0@>O?kp;&&VpKW{segt>qv!QWxa)L88z8osq
z)2gcP2F3uFa+!%8s|YOk$IaGuy2o=3<bNOF*s4R}h=V<7%8s@sV<(0I1Su{qMpG3d
zMNsA61p5WC0zGP!_gN925f1|lt)Bt52L%{t2@?}kgUoAjE5DQ&nd&&C-<Qj>vB*Re
z-E?{p#lA%GnP@c9f|1KM)+C6cyk>gZHtu{V6lBOo(KfM5lZWjbZDDpGf+r9P6gWug
z;QN4=ZZSR#5*W!8Z#GOihvW#Jc)%+$gc|E`XxdRYoY;9lBXagnbFuYcthdTOu<%n~
zzkd4t%>A7r`-Fubl}b!P;fmrD^&qO@HyW0Qd3de>VHv*w(m$FYur9)ELCC))&E~;>
zIV2@@{?$pASY?p+3dX%56^ha3!hQfB@+N{2PI>%M_i%5oE%NP)*1*}9ia|B&XCgbd
zPyh@)N)<^5LXWBS+u28Gi@`U@XTqQbea4g@Rh1?}`tldYw&^jCj;61!+Kqi_30G<D
z>U!J<oOpN{d@&RUW=2MEHtevFlv+i30FPih)b6M+;Hilg$bA;ET7b~dL-#A-SU0_V
znE^ul=x75yy$>eBs4`%`j<U3nsT~%&Z-Ef)-K&pVhQakPK+)EASX!&;cJ~u=c<o6l
zXOQ>uNtXz}>+hg_Vsuk_hb8$j!#s&Gg>;5|u2#MMme$!pciZkdT*A!u3&|G@@`ZGL
zy(n+J-{(Ev!;v@`O7_nY+CLr{3$B^YAEWFi89qfHX|#_nR*L~p@Ct2_x!kc3nWJ<7
zS4JMI9Ys^M%wg;^M4it>Et(NH8;>quMYg+_Y8-Vr@j%gA4n&r$qOzTGlfv7f*)M5&
z5LAcJlS|Y6US7V5fyLdW%P`~yu*emR9Q`vypRkd}0;Mssh+ev@eeLA228=F0!PNx1
zIQUd?ocz+#U4R-u<t_|+LwvNjwI#;fM?Gsr3)IrU!gS`ho0<E24F1d3O>59Rex`56
z87+?X_I8ITO0D38guTxi&hc#5WBDwS$q5Gn=uLJ%U;8s8i@rk>TV#0n@1~O+bioLp
zh_4^7ZrQrkApaV>2qVI4PkVbIicIuQp(aN5D+@vp03WULxh}$@Sl|Nak*70hF2rP0
zQQvo*5?@-qd1j^W-u*zc2HuFnYDQPCpsf<DM5R$Nwy*8LoNj~CNoedK1VjlC;>^F}
z2Kl6tGS2l1#FYRZoYytq+Z?1A&+d1M$IoGbLrarVs``^7$ZkprjkAk0FSMODb$4}1
zb8~+Ob0vb>8ICE(L(Y}fXI={b#<YQSLlyl&yok;P&>qN08kF0=d90N<0Ui%y0t#<Z
z)(Pp-FR*HWiH%lfAws67m*n~HrQf+a$1qhqr7I~pYP_9-qPiuKdZ<CO#73Fm=fO5Y
zPvpk+%$Ss7Dw*8o?}W`MDnP+~U<Fz{cq^)W>EJiYEPer!C^Io3f-74WP=k0wFg(Xh
z4S2r1haIQ{-SxsQ;4Cz|qze=-Fbz<-6rC14Ou6mwR9j%VYhK4_RKpG$)~7d@U-v$<
zQ7<Jv^+|FO`5yptm}LULRQN=I40bwAElHgRh{glN8jPOF)lCD&K(7I(cfvw8Ltj||
z;8dm=w)Om5NF$(`^dH8dLC^}Aeu3;D%sq-b0MOY;96}$4;PiBBeC5+``L<)4Ajju(
z&3S6F!p&|KUO$(s^h?j%_IlNSKKHDUXafyD9aW8+YI;$FwGWyD3;NI_BNmr0w`Axa
zK|;!P`plX80Rb-at3lQPM$oY4dX5`~6l+Lf4oW(;u!C_CU`DnwfDeKAy@DQviEul<
z94z<nmhbxhjK#F$C_89qO5cP@P%kdZzQh7VIzwbpii*B=GSXc1fz_z~&8QZIYx)vK
znY>f&BO!OQ`ZC&6l)$Ojk^T&Gjh+k36096jYx1=XY6_r<UbnNuSfVy$)C7SIGZr^N
zFNV!{p>S0i@%cMahpUbbiNvpPKtA#tCkMzd`*#H5o56)|89TlH&6{I}hO*EP;8?5`
z8wGa_=tPJOo9GpFB_)4d{#_tG05pX;N+`fK28LV&5De%$8Y4tYiOd(i)Zmt!QBm>t
z|7|I^n>JGB=4b5E*49_6X}r&wI2t<(H0(Vk!e6|wt)>c2+`m)>Z?(^&Hr<gfDNJWR
z%AM49VswQ=Al;4~jwsua@#6&)9WXJ(QP7*5iI#2$ooJ^EP9PA;2`#AN#|y`F@9at;
zrG6Tdvd)mt97TplRLM?fF$fo_5!_6R-~J*1u(~@*#@V_^gmCieIObi4xDryG=X^RD
z_!u5;+y|tNI+wqwYmd*}=zc2TTfTZ;)w3W$IndqBUvo9|z3v@@L?gx}iuKRx3Kl07
zaJM<kTj4unt{DtOVE+`tIs$Er=6-Tgk{?hH+2!$E<h>B~PGE`PUVxW~dId)Wjt86=
zrmG^Q=rCZN1HhJrN8m_n2P~}-m~eUY!N9CwK!o;OnE4a`D}et1HN&V<9qS|^!8BoY
zB$I~UyJEUxLw686|9ti9Ta7|XOI9ysBgO5B&2@#@^6I9t=qgHStqo;Ut;lf5*51$s
z(mY`QZr%0XV?ot=OX!91@7x!&<qbXvu^c4JY&l5QL%Oud;7ec7s|^+G$r-7M+|#JK
z!4WJ4Jd8#$UNCb)h(?P6lKn&Q9a1G9JVi$G3JR{HQH|_{`qpFRy!E|kgPA0aO4GU@
zQ$s`nPm*hERnd0`O&J&v(Ep@sDv)&DyVnIe18#f%)%IKAvm_EtfQJdpWGp|<4zU>T
znVh^OW)(ltX(hmickrS}qo7(u8b-9kGf~iYzOuLsgReeYdIkolOdyWMzhT~nQW6lR
z{t759a^L#@ER_GrvznjkYNF+y=+>uY<{4=^Zly=;c2VaxfnL}rWW)ZJ3%$j;g@uLr
zdE|lTCIxtTI{@b46hJ=)J9iMLJ>rxlL5&`zB33RoQrzfkBf=Ic;p~CXI`}6U7TCsF
z&Z{b*xWzt{1wARWuA#OTkc6g~8rq%?x1vI-4GWu}+Zf)SwsmML2Gmg*8ym@bSfR={
z&XwTnBwcS~_lC_h6&>hg3CQC|jUo4!1ndQR0nBcbBcMp|sx5-h8cPF~1ibVCV4WlT
z7029GG)c6kC({17buaX=PJXN7<vq*v;-Cn_$8b&Q%f6NGlYDasD!SLBlnWJrz#{zd
z-pP__I3iFod88YK8{->TAdd%s9UcuncFJJXpj^g&BRpmxxv}due74eu_aMdQq#U_Z
zcB`Q+JZoG0kTgk7<`C?4+D7%0V`D46d|AL#wd^?%%If2HVQ$p4<Vx^=;=lc}qmepo
z3Vf<aB!Q%B8+vj~mAQ&uTo*7l^}^)O^*hA~Bk?>0Q2lQP0SY<B)ps$}SV&)Z>Pw@^
zs8;C@69*fnZ7fghS6ttCZxlXK%<`U*csJfyoF=H~{fB`Ku~;smW5X_C>*Y_el})#S
zSPkUhoSe3=KzSd%RvdU_sPA}gU)33m4#$GL844SS5)t#C4|av*S?P!b;Cvy;RcqyY
z5&Ru=;+6KWz&RG+t2!o)6rsTg`1or}i{>7BD9$hsK7akuP~Fuq^x->k@}!8+T>tko
z`_;rbq;(BfN?i4xiT~cbZ7^C^NTb~H0P%HOqelge&5;~}iHvC+c-Re5M-K1aU4cN+
z-5u>u$elp<v$qd_55!nXQc{~Bsi_~6kIU*}_Li&oSSw6LB`oFqF#fEA(ND7mALXap
zAFwM&j;`tu+n<z@MsckgQ3x6*!b1Zv{t*~%6EvNvZf%+?ov(?~7=NEiP%o2WA%3f_
zsj&*mzByfCJaPV<QCty;JjPJkA>~t}n$gvZE&r_Fk%88FU9@0B;T<Ma{>KjXN{|c;
zJ|$YqO-<R*xxg`|_tGZ{l3X!P1F>`wwYPhh{w!X98c|9mnD0EQ)pv`WO1Z8>2|Rc_
zveb8@mOxVgwIK&hez_I3NE@QcTMmy!m}9HySEz913Oi-Bb8XvZbZPgwu=P!%lTz#t
zb#JR0Dsp6beRhSWlM|=rto1tgPEHo;WW>r6TogGOAF^Gf*f;2=UBY4pYB;L6tI7B|
zpTZx{&C10$J{)nE65?0sh`TNR7#Ir4iz)?F7PzKcM%%z*qD>mts}&iTJkqd3;E`^c
zAJrYN_Yp~lgFlpCHc7ugt&_3c#g$(98v{YJBB?L%>QtbYR}e9De8V<UEc<vG|BG+M
z_kWFZW)}!(bmkcUxX;4qzT(hSA0BV^5(0bTB-Ula7mPFdXOr*dt<#R<a-M_>kC$(}
z9eIfsZ+Zf_|FIMx^pvmGcD@RT3QrbSG%R<nXYTtsV8_QNU`A|aQk!-mnE_zyr$6zl
zmKr=;K1O{{zmOREBwpONn%>2gY9pD+uV3fu*7F5Rs+XSt)kv?rS6Yn_i!^T(eXG><
zjl9Q=d*Z9c->4|pAWvLCmBfQxGhZ^=U`c-b-YYAeV{%^cY8>gN1S5i!w}D)m<fUqy
z7(ad3T3Ev6(RM0_L&B`AeL1Gjn_e_=5FtNygB<Zbm%p~RE53Q-d-GS;TVsbqcFJ|n
z+j9fWKWhE^kdAxfA<3C%A;b1qd~>C7*HgJzK}93uv!dpzxK~uW)c&Ssa@OnjmHN~A
zoA=)KBc83tV;nj$gVtS1p^<6gz70F>GjrlNGqXyqt|z{yT&g-8MOqD)f`f4gNUFtH
z`Ue*jQ<H5>h?25cKUGLc4E%SN1_D{>gO%b3L@>|+IS<z(r9F0q7wgRZSl;y6e!coy
zcSXubz43bC#T|hs9A<>%=CgN6_okh%E^|C{{9a}Ci6W2E5<^+*z1o`#VAC%cdOCrq
zO~?bQ`|S!9fR0WdZ`G*ZMt4}Gh}tQhhBT?<pkL?|4;++Hw!=@DtkEo2uVdnQymZtr
zH5#!ym_?<_KPY=MHt>n({Eacc_otOEZQhhsSt&Z_rzs&-`z2iy<qiK1<<N}u)6-wn
zosw4*X=N4ItN+km7kN=^d$NPn8>IE6ZJd3??`#F2bt{*Q4e&4}QZ#QlFQNJy+aJ`R
z6N1t;eM}+tE^G$cuf7J{6~bYhuv|4TU?EZ}g>EKM+E7!ENzo-yZS{jpQdPk?y8!zm
zqs2SrQ%ei%{CN2Jp)q!dqSivC!%FH39xeK>O2(ZBYedfd9WTEq)mHJPnUjA@r@{II
z5n|MT&yb@kJiQx8quX0Ju7PycnQwXd@6#o%Kee5ADto&myO-u&0|G2B>+%ODCLY`^
zXJ(UG`j&XMu?ECCRo^_9ssBCG&=9HG>I(UW4b4I9c~94mgI06m-Cvy++UE=Q!$SWF
znnTSGWOr=Zq}0{*>fayn94}G@AL;lT)l-KIUcG;AC`;2;-1Y9|K^~rkO7ehboj3_i
z6Aa)V0V^LON~2X=esdusqL4UQB(IWx;mNq_21X}?C)ymN7~h7TS6?M1t~@>UhGzlQ
z!@2<&k;1fd9t|C$5e@nxc<^O|XFoqea{?3_@n;kL{b#=v64%t~M0{mSI%OWwKO_6Q
zKSuG-pUGmBHkf&3QAtj{ecwJ)3fRaL;0^bRQ*FA1a*uFR-Mjb1Zdx=W=b*cugs6^Y
zUnV)~vTr%2fnXHFO;mq~2;J&EabBBAN$iYM+w!l5mz-jtIXQ}I9VJv<S0182I>^yS
zoFI)&Gbokf=TD2!`-05sG5A=6C;q-e?BDJ5>h6Lc2F4Bcsz+&}<im=9%+$ZSKau6j
ztIP}hJL;&;j1vh`%I3hY8%V8t5Nu|T-Bx+2+NNHr-1W^~NS{~2Ha4Q|2m#+WH3j>l
zCxDi1s;3-&iIfzx{`9w$hA}@*)@)D=`XTM|Tu>2(pU0Ie2D)Ld?1xyS^Wb17D>*z8
z^3?$+5I{sgRfG_)ySoh83|vbB{QVz(Fc}SkrPO=_Dy^R%_5+PT<HFEw<^-x4`M4z2
z5?y}5?8f}&=Oof!UjV8V@Nj`TyuR8luK<2yeK%5f?Fk0rYc0|ewkUEA*@oqV6JL#-
ze^7oH7&ZR!`Ln!ic|WKe`m;TeU^_qZ2WEf`6~dZub_DqySU<@uI|0ElF*%{#33Lo-
z5eSRrZ{yG!BHTBFgWvqo*h7!33omG*`#>ttlJRKkCvIVEEpaon*!hk;(=qUrxRa2O
zCA`wdJBEY|2!aBJqrE6{2aupLkLKnPz=h}1&jTg_`(W1*NB*Nn2?qj@h%Tam5by#g
z>;bf<!H)ou<#l#8NGPYkcK~z<3#X+RU?w2lVq}!m%xU~tH+UGJlHecA7e{YfSz-J<
zOcRMO^@fP->Wpi!;F~;h&ZDd-%yX%#Atv*nSo+K+NW6hOqvpJQ`*uJQ${)aTnlzM{
zeZaMOvmZVjU5f+xYR|+c;CBzr^Z+^`r<taK7$I}`nqx}sSH<}pwG^28YpI!KD1Thx
zDS;jY09lPF=e;k_nXkrF8;3;yS<fK$UP{|YihZ<)zN*Sk03H`(pOe$O=jQ|WtJVkH
zrQftktK8KPQ)9Fb03HS=08lPaorev8s7?9G%Z|7+Sgs~HK(|!Fg-_(U?&mrhGc`D<
zsDK(Rq0!XJDmCrhPBemn?F}K6zm?80h$5VONJ+x?hzG-r&{aiI3j4=*2Uq(3teB>s
zx7dFGZYXHh*{{jS#?%df&87S80v>$gHDD~cTB+9+aYs%f8(?9Hc=+&POpHy+&kzMu
z^y&ZWdHbU=TOwvPH`VinmixLr!zPN)<5aT>q9B(cl!Sc6ls+hE=K*%SjE{$x;H)QR
z$3RooDswzTod>lzEV;n6b3)ib_uOG<31~`P*W~CZQ1R0W=#C;0PO(FJfedFSl{3l9
zZn)I>O=nTtZt081@71^K6_^BE`I^<C6pI`hkdA<TDBQT=@qqG+P=8>>fk%=ui7JdB
zydkB;loljde0C7hA8F_WIG|1avla|SoWUVP*ij#gS+$Lei!ZZUl3nS0cmG9^uJ8St
z?_f`(TlLr<UY6R_i_mrfemt~-=Qoc&>DLr(j3+vYei+Ir4#3N2_n#Gy{66EHN4Ngy
z2i6Y&U#z`}Z*}wmt<`_<G~R8w<LxbO&v#6+EGj%axr-RAgP_e24C{+>RGX_SG+K{J
zT<24YUhLh!A7qP=?$w_wm)Fu)SFU>g_lNL@IyLFN?b@o;_%<Cq>0J*Rhp0E*g3@~6
z9pEPhy3jfq*z^#<Qpt&z9`bKMfrgCxG$AzQq{H59@8~$UpU@bPQnpzCAd$K+FI72A
z8F+XWEH7Ba%U(RA!Lwaq&vvSi3%CIRfJ$HAn8&jV8!b{nN_#-RJ24@FUB>l^zw5$e
zdL*l1-oMLk`!tE9pwLCl#J@C$Bg;Q3Si^Bs;9eGS{-ME!%Glj5EiXG7GURb^hh&Ba
zjAa^>)-Em#_*WYPD}3oCWX}ebu|@S=#*|`Iwm_jzaM97>Dda@Vm_9n4!1xdG$QCP%
z5bNBA=UzN_PSpu*;rQ65z1VP=UPmY0RQmvuZuz)_2M(MnG=;$erso)EJHPd(!x?ZD
z7#u>Zib_4`FgQEH4=lj40JZ@jYJr%Na`MPub|fY%V1U3ZR1=U?fQEG69}OGS+UJ(W
zL+<NT&pl+K+=NQP0aGu;#9)T$k|#?%a4+9~kW#o6jF#Z`f#B)_F1M@e65_|wuL)%Z
zFmfV=P0{jJx_WgSFi=P^_BI?H;5Q7_k%A~JaGjhMFsRwg1_p58&J$!n$c`8p6I}0N
z!+~THDq&;#U0E}u0xeTfE+~7zYc?<y;=qUZZuCAdN1Em|PJGECbl#*XGsbsR|C+Tv
z^LC;RmC5W5NW!sPPu=D?gW!1-JyI++Ahm@TGtg?l)&cNAOde@BkAXKNI*_bTabLK=
z*HhVN3yzjE5HC19h)fYc5`GhP+D*7MHPiI*vKn?k0f@kkMnctq-7|FCVCToONYn}>
z7M>lsa~ikN&`22I8<_nK>_Op4OpMTuz&sbrqyy5s8b5tH2y=@J25(+K945K#RCl=e
z_>>DZ{D5N60jq{nVq4AfF+Ys=LkotBK0pCfLWiiWT6fTny>sUT#tSNNC68T<<^4}$
z8RBv>CFSf!<}mqfB#ygO^L`@2ywOKXqv!nA>rVAMNX?Iu&L$gO;jtMzm&aGn5M1Ul
zQ*gbqI&f@=3NQ~|$a7-jirw0cSKV;X({T1T9%V?==d=2D-ipxAAujDeT)I+_VVD@g
zr?)UUlDsW=Zc>19%M-_ml+rZfCrBxqupC6?Ts_wsI&<!%7s_7jd|jEKJht}lXKcY<
z`4pR_sp-<0&+6;@NadxVVs}mgnBD!?zbmO#eRY$0^z7M9fKs3=VWJ2K@`#%-=dmm7
zD*CuyFTJ87K7w|9{OSw#BU_Hk>yCE*crPFjxI>UXT3RpeNZS<Xzp#Oh>o`eFpn)8U
zX6x2TfQ;<>){C?A?=r?vTc!kRYni;V3cZt>+7vA_n&NaPSfi=LMRH+^sUa-xc-t~c
zX<@X?n)N?H!dnJ+BplkUr_K>qslNXp8hb@5ykgR>Px2l3j>za+Z}HQVc1*J!IU6;$
zZ2^==FE5hAO_U8#R3pulWc#B`RAAG`Dv*c{?_1BK1CP4jV-h%fp)^-l)Gi(#(ufsN
zsoep?pt6#A%N>`Km#>6GDevieXl)h1y^Fnx08lS2bLOQ=ld_UJ6Zc5BK1>axq)(};
zcGRa+$ha+LDH;>sG1jQ!tupD;-pJuA-5q&dd^#vAJM)<+Hu-Pf0*k%Y^Tv6cU^aGz
zLBB0vLL{T4`3-o!pJ-@uWFyVcFiQz$CjRsaGx8ld_nO*|gJN%2Z3obog-9)aFboup
zEBkj6KayncOtFhJDs)bR&Sppzmat+Aehh-T4;dH}E+93kgva8nCJ8<B8`rNloj*$a
zih4fruBCfISuujOZ?v{E%R3~-(2=-{*Ns1}EX#X(-bqe&oNb{1-JTe7lkCD)@%HTx
z_=cG0&NLDmt-_DY$%CXmHGEPrVqwf&*gVL0a+2O@p(|B~`u43K4G|V1hXq)rZovNY
z3R*~~uTFi6t3tyJD5By1b+l0~szPL%r2fG6`SV2r%r8v^m|t4F`=I*IH`2>8@QQ#@
zI?!l1_Z<Y@4H7Jvngrc3usUU|mAH)cYF4f^mWj2lv!s3fs7|xK(AS*s+^Fi+Q*D_y
z19ph4P?J3@QpKVJQb~_{oOF5om8gOGr_F8al`|=pintk3Mgu>lQE=N%89Y^rN5O5x
z87Y$J{O{?xzK@e^h#zFX#K^dFvZt{#FEAuD5kF|6CGB2h+bw-3Jp3=t{5NCwa{Mqr
z0n?4P)cLb#Ti<21KM&?cIYx8yjtlWsyiOlO&e-vSW9Nr$l)Z1?URf{1Kc_!_ZIzh!
zcRSx8BUS{(N6)o7!{T#mAK7URahzXAME?Elyj1y>zQ{<Qx~J}5e+x(w?^f*3)~AR;
zR!#B+SLpsCv!o}~#2=RsnL%d_ui!bRSv5}yL&H09asPhhPZ~r3QVgh8lwGUu%pw2x
z?JS9)lf=n5zHO+F^J3|Fv&%330tqA*GHeL-jdR@GzrDS&%xMnMJ2}1S$ZLLZwE6t{
zJ|)roL;N!FUF@&ZZK62s`m6Onf3uDF8?MswwZTloMj20o^&Bh9GGuYK@!w)q?l=)O
zdGD_b(%5v0SG_oSidUQI_Npy$??__A<!F)d9S+U63WW!F#8OrGGwPv6q3EOfMwen6
zy4HV+MV$DGhm5oPvM-m|l$)BHCg1O~3oKhYU2r40vg=}C_<CS<qa_}zOW3Nd)#u?-
z*938j|6Dwg34Zdqxhn06l$J?I^sT?&7P`gdx{lXv|6T~kZsH@vKi=L<VCptKpJ}-r
zaIKq^T<38Y&U5bw#7GySJ%es;xQZD{5*+xMWlOba0%ffGfOu`&j!)q^B_tS>Z}Ip|
zp?$o1QebeN1^30Wi$6wA@jn%}ZwW$^9q2ab%ohvy%E%B@O%!?5|NJw8ESi0AJ+N}{
z<C{#ceYLPC5xB7B0XuiNivGQl6VyPUstrBb)t9SquEX$O3i`~s4Cn|!zPVWD24K}-
z%!RlQ6@EebU;ZJ<l~&X?3}jv~pZT!gH!h9{&4q=#ckdqRF7400%u2xd_-wvhS9A1q
z(RMX>_Iv$iA5^b6t`GZ>hix~Ji445Ugkj=N4i=V_J6p!i0S~axMvjp(a~sTDFUf-V
z*ot944t{<8`;HT5zkJg-x}q!Ubj?vq3x@#J@hTRzmj*C>d@|^TM3;IK*~Fgp4Hz50
zQF9pMoIXZM&Z6K0!5RcP<dv&KvTGJdwe*L`#n3%>o9VB{@}#PUm^m)bEd>t0cwv@w
zaX~?=gW$lysg@cZfin}=WJ>#m%C`$%*k{uuA9$*5XW_+NOiYQAihJlmg2}V#F*tHy
zeT5txU8ekla|CV~C=9{YwlFdZj*4<eS^^_t=w$FLZ?o*!wJSpsK>KD3meG3GK}<^l
zqkDd4CV5PuwfUy3a%B3dW=KKtkU`<u#cxg=O1P6}34!)g)_vQl@*cuP98n>~Rfa4k
zG&J9SNHtvLGb(tn0^rzGYmx&(Av=cnqIb%Uo)<t0{UJ&*uy*A3(1V!*f^42ei=w_R
zfL&qT!1lJi|3NBfcAb;uy_;`#zp3;ldPv`jsmp&WqWTweHi)ViutT<qLO{$fZ}C!D
z)B0CfdSURO0<D!({5$gDNRB)V39dpsCLN%QG%%2WoZN0Jix*J}Uu}O>%j!ckcIarB
z56UT9v~A?y5$K}wEz5V;kuQOPBH>bPSBKTz=i7?Y&OE!daq~})wUv!&-8)f-wXl79
zd4_fH@DAcb<(`c?Y-r@)L`<(N^~`koh7r^Q23JUc8vXn>(?!YLJf~lexgtnMPAI6D
zL_Jp8BOj+8S9YdITzk*XU(XJ|3*4bsP`J~_FLi8E5Xwg2RYRHOXd{WepPWqjVdnx>
zKzK2HK2j5}-(NxV@r&wN;)6*Mtz%>cydkhL_zEi1TSqi`xgI4?Y0^m5V4#^lC_Sz2
zLRDKnqKrEtDQO|0iUb>$VkiTxK^itU{{oIO+VN%1GtU)tF;-b5hiK|GZ~je;4==@Z
zhuZrnv{5kw>w@;?A||B+LI8g(Ej=A?A_D>>TE?b7{uj^^hlUoiaL_+6m-KCCXEFRE
z@j{Kug7Flv3lHeA%+|d)$UDcV=IgIZw!x-u2epn4YW`CVE#GW@|M~r`IRB97)e^ge
zmXjRrGu2F}OGeA+;KSU31{RZ{+ndNQ;3qaVX3qItOb&$h(j;RJY&_FlKk&^#wYO}J
zCnn>hK0Jkj5d<BSxe5vdU<cm5CC1%>V=dg+1{x6nyxdl|JBI+aAYo4})+KJMSBGY1
zjJIK)rFM8HF%we(9h=YBq|^gZ6T*If72Pl_XsBL8wqA-3o1U9%wS9;-6x=*?=0^sg
zaKa1BGeYY5<`#KdV{So#Gp1G6a%uAKz>Gt@R)2Nc04|!sEh;?u@N@#c<q>2?qwfeZ
za6tij-Y6NCQBu|GXo)?%r+d|ToZ05P;;&*SmY~^7!=LCDrpy&i9P->Apa7K$Ua*A5
zP<xK>M4cxD5``%g=TCrEikFYCFM6!7fPHSsG=>3WfEDRA@C@M7t!%=xBM?LxH3i%d
z4UZQ($tq94u7p;nA5&peo{3{ds}UG+DTbRdZR08;6W9(~I-TqBtTn||zan%5cU>iP
z5R9+TZs@cN(2Z*D422ct_ZdqWNk+DhvN7qPd&56aTrK^0SU|0-t1FZU+=}-ow{8sr
zoUvsKKiv*?_J(`)n6b*ElLaJOOTS|Rqd1tMu%){$5s94q*KpJAsy@PM1k4AnV8SK}
zG#>ziMz_Efu9c=d{yVjDz)p{U2SY{cRd8>>X}a3-vH{jPXs$*e+QBdi$j7NYkKm$;
zQzW{o4`a09ld{-pQ~oW>=yrJc&Rn~r&l=2vvM$$beHjbG13-H@n9&eejUsdr16@$$
z+Q23ggHQ%^bVdbVbA2f3%Benh@L|Z4_+KsN*<poqhDTi<1&5FH_Y0%34^}N0D$!C7
zywIcoi^E=n<~t+D09|C3qizzji~*scItatyydWW-gTW^FdDnW%qUr_^sn5Ik-Ze}E
zGqAMO{p{Ih?0S^3T3jtqEyH-7D*_`1XC<``GVHTAT}87FU3@?Q5wN8MrwGKk1S1DI
z40M4lCuZFM^$k$yA0Gaipv*MTWA%we{8@TtreW@w#I4UfYrn2bu>GNR*gkvw$BeUG
zVM&I)cS;4j#m0(0+b2%(=vz3&i_IR_8V-U63#S60R8&<!B-84{aDk3|%Ga)CJk7mx
z_mXQo4k@TwG4CMocNA#80(8V=6)qscXL+`-txS1_XT^9do!YrO<h1P4FgF);tObyI
zj@|?W)fs<^fy)lm!~li*cbL*OxZ8A<3rxE3cEYB3KoN8Rv&&GUf*PpxeZNXr#klLw
z9|3mUnl&iDLnR6@A%sp+N^0=OkECltJUo}q^<r$a<~7(^m<5%XH|}cdrP_Nu{?l~1
z_eM|koiEUBF9Lab*B<mMflkzC;PLS0i!^#*+B;TxQnaX~WFf9rAwDy64HTLb{c#vI
zgo&i=@_yqcRy8yW|A5q||JW}-2eD&t5Z*%RMCc%WIyS$w(^(snP|?Xh2R3eK9S(qB
zp@trMUv?g&83Ve8%^r~S_MR1&@!-XKY?5cc!G!5}xT6(lzkpU!ijV&O!SHfbO+{r3
z{KCy1T@}YV{fgd#!wZo5XP<Z7VMK9=p^2|wuP%*cA`WUhVQzp3^*9Wcb4EXOl~3W0
z$$m(>`7D)BzgRI;uxZ($6*HjVN|-M7x(i4!FfaSUi(F@elHikYD7OoiZ6cPfJZ;Cp
zh@I@re^70Voq}(wGhrOE?P;j9Xeku9dK(D8Be$dFRN*|Q5}E;~18+!LzPf(dqay5K
zTO(p()(~C@r!Rj*SP|GWn#tv<pav7N-)o<&6@h_ufqGAgxmd4OdDpv59nyJQjh-HY
zcUiq!4Xp359iU$ZD)Ll#9|l$fC{~yQ$O1+;=2(h!-jR>{Y3>Sk6_>EMILylqtlTC}
zmW74Jk00Sz&obMl1~a8xs2rXInVVTLz(C<q-GEA#eBAEmP`Y+}{rbc?Fh>eIPHWQG
zYy>1S`Zu*FKP$%^MvmqG!>-US)Q+ljwEAiIJp;FrK3|%yY1`rjHK$OI_ZJ3-b;qZK
z%+7^1exReeQ~ztHh#8NN-iNPWoxzC(j~n)AXH->zanVpyGXz#Eh?>?uY5std*J=1u
zA2ey;h3QPDw^mC@NW6j|8dnEqJouLCimRZVsJP<9Fu%nY9dJ?5Jh%PsLd+?wA<XB_
z`=ho&>oEXzXl_=EZe7#(7|9e^P3FDcxWhPN<OYCv&WhE=%FFm2X{M;aP*NmTF3VI>
zm<4~YkQ)m`KqU5TmGXS4mnYI+EMr6pyfdEqcO2!C(RBk)+hZhwBh{sT=Ch(Gyck42
z+h%kW?5#I$3Mw1ckgkFjVD?L_bV|S+R%sw9hpW7Fqfnc_ZAxJW{jv3>OX^E4gSHaD
z3vmK~HKyfZs=l=QbgkyWkz@)9DXFJHLbK-ST}ge3ng0{gc6iz39xx~R<HMdRRkb1O
zZ<fw;Jh_!_;g2nUG(4kb5)#~X<f#PPS;>8~&vE9O?rTQ^01pOQ4Um@zmgH}8G(s%U
z2ne4+PUrq>To187{AOLLE*HV@0QCyiC6SP5JU>Bg#x1fN_T)r<wy&T+{K?mA=H`Co
z_llH$t8E%)8xWaZ<Q$disvj04_+SZBO2<yNz2BB?z%(r%H~q|}4EEAsAM^tD=PeLD
zJe+}`t*fVJ3hKO{AKwLQk+H^Ytl}zFtl{r0pz9bV!z~UcD~s4^8MdNv9d!jErNimr
z=2r6az(UHN%SJ}8te9+Swi4tF=KmsR%&cbrZie7f)CLh?yfv9&cBIsO`7>m*$IsCR
zz!zGsP~l{Ppd(H+pz7*8+dcox)L>8h1NuY5$=e@gMMByp-sn>v9%ZBBE6p;0>wp_d
zrGmXkNg`pRg98d_X3<CTSKbX~5axjGt*%Fqh?c-p)iT&9dBR2iP1wL>h)8p4+R{oD
zCWaBpLAMnsFM9}#Ed4w}Eesu^AvZ?Vz$p8qsZn;eM<~&dVr!!FxNhHF#-H!*r#?;j
z$tPl_w~8(rRhfq5@yr~iPLMAFh@hl2-GI5rre+=Kqd82BjXvQPnLce!M9CtT*fvJC
zikPeL1EaAr?EiU|E)qC>j2tq~aeNpUVP{UC&dJX=_w9i>c{|QDnkldcwV@S-p?Z-Q
zdWMIL7(X2Q)^eZ?!S;;V842XNo<z4`IrZt2`p16;9@f!tap}r)(q)TkO)1>6DHn*8
zF67cf4<S_YN%}>jpx<411bD!$c=tgev%(a|Y^Y1H)bT99M|};V@+|32V4yNi+mJ(y
zjqwT$FeJw->YOLFGmI56VwmgfueXA6pahz}cH12F85IEfNa!~UqRImdUjMim)<Y3~
zWeCmiQljJFFx1k521d&`Q-SL@(5e8XJW7)%k~a(|jelx8f_1|M0s_`NuMBH>SE4b2
zUZm>ScV~(i?z4EX>&Aj0-3KvMA)Waj9~fT782$ro%N{Q99|qeuq3^;4SptaFrmGS^
z>?CJhglE}tN$9ZRl0rV1Kq58w>zCPJW9YDy#*E7Pm9RWoI&zEXx?JmJXE}&t66rF%
zhO@IXP-zUL4_P2d(>Y9aNBN^W0WFc1D}kP(ra!5WP^V=ka6?--{NBiU9oDebS!{;{
z>hlOE+v`n1GYO;NxF&8F56B&!3Ed+tzn6b$i>Kn0d?J_##58D^%xMK|bF;x_ps3;g
zAoYWNyU@HDgUqj>Qij3<I+$<)?P6W5B-9#oO(264(|qI9-w)V<G^l-<_ZgVGSV~pu
zo+o>8oV*kD^xbwJWgspBpawk~Tbeo#MpB<2T+B*AuMI*HK0(222#+||V0f)P2|{EO
zo>#zo%<=_xU;hsK3gmTfIoQj7K<rP-cu;_d1Hnv0(Np>}P9W?$pfbebgT<yOQnf;T
z{vBx8z<H~VEn*NAQKpBI-rz6~C@lmwvZK^RvwaT$0RYa~*S{;8{fPAc+r$aqyAhJm
z570`?HWoXKVeZRDOgM&)1va{Eb#t?`(3tOg4g($SSz}x*Xk~<>y_+lgz=BPRWjW#N
zk56>Ve7=|8wsT0C(7{HT9DZ^S8Z))}<6c#La5Gm8JIJ|zzd6DPreor)$Dsq26D19e
zDOD3#%m-rLnB@||7R90QjT^6E)sJi%o1z~+*OiqR+YX*fs_ZiIa@t_I-4Za7#%Eyc
z7P1>+el9&d*N#9md9wk~;}PKiM=dgrOaMa7->RH&bm8KvgqTu=f6#qxMb=MKv$T#?
zjQ4y&)|J}r{aRyFKLJKy`s`7Y_q1@tfL8DiY9^S!D-de+^b3dg`BocSqpEuUkP`|t
z=<^f6V?);xX6Hzk@krJ0pDiVk{&E4r{I;VzM&z9P%FZ&SUd)U32UiYOH$3zKf`(-(
z34@I~oa@X0FJG{sT92Si0k@zUZxhCQes*d0G;S_Yw3YuJWX;wEbp#d9Ha@5DhzJUD
z@{llJFls*|zMGhuqVEsL0HiizjkbMBd4am4p4v=o9<n3E0!)mst$lI;1uaelOnla&
z$kpYCdI3CUvro=gS}W{q875|T+ZOz|xsn5wKYn=3vQ(w5@ZXzU)zVq9t>}LC_Fife
zfPKahIHg0+ffb?_C6%z0wjr<@9(0xH!9BUa>yeG568xW9FGOvJ@vk#TMHE`mb5fIC
zd3B>Bq;?%#jx|auZ>*^DQ~CYrCM4Ea{2FhZ&cqyDL9GkEj8)4^8nBGuoza%khO!ha
z7=!1o`=9?6l9&S{1g_IN`@3H8XiPbF1@bnXHF`L0Wm&fILgOVE+KFTH-PFxOVj3RM
zC72fEzkWSvR|p;l)}Ia}&=^T)9^)>4xkJ>RoBQ~VRBhwuMU#_Mf}wNs0ZPrqPGYG}
zT>Z8Z=uDPiP2oMlpRj<bjr)LA8ZB}Cmc{PG6p+|aZenD@m7cOvIC8+S3=tEo8Eg@Z
zBh)g^@EHt5LmCqWFjt!ATeSY~Sr$pVJtrG}+&XQ#-plwz$~yWmF9If|uiO&cj>hP1
zm4OS8f{103ADX+++YVtZ57>pUx=2UI#c|HcIizd;51wAb4i{(2jT=oJ9XpgxEx@KK
zr~T?3`CxdSeJPY<TyN4nlE0s;5!;Q+sb|^26%4cf|Hi|@OdCK%|HZ>f?+%t1;8_UL
zl5EwXLzhlYp%wQM6rZ-X_IaEER-bL7Wm>>2aAI<sz5nX+$$x2mq<JbTXRo5SE_43n
z*P>SMv$7b+>laayk?nC`w){(%1F;?OaVt%!&!+DD9+t=>iP!G9a6QVqCUCLvCtCdL
zfn?eUT)kN9@q1gkUt0rJjbC+<M_(HKzrdRN+#$Q>|Nayw@l!JYQrHQuQ=sI5#XtSY
zns+=r+6;|$F(Zk>>43co<sBFc!lI&SWB1m7H-(H`+Uwl^pho)E3H+L&6Eqae&uKO`
zqtR1U{Roi*ECz@A`;E;4F-8(A7*W!~*!XOI;X2Ss^ccO)?*mEs&WyNqv+^ko)+)0=
z#h~qYm)z_oyh|5)x!CrAWs-}slakhbEMdTcBMqJZQQpJEm+|7g^u>XUL=)~|692ox
zdta`(RZ=%WL=6>U+O+wFYbGI);?XGOgT)%;2p8)9BFI_{#D~9eS^RW$(Ka`jTkM8Y
zD3AxIK@|gC=Az;+1q4sG(EjA{3Ls7puqNfX{k`;Un!_6`nU|{CHe(MYE2-do`xj1?
zyANWH`sJ@SJwIPx6DXJMdbGFt`#ZYLp39by&zvtjgqq;|`@g&QWvmY=X+zpAvKJg4
z2Rm~8_HIuJ)CddP-Vk*`^s?`*nijKb9Q&Z5u)F+8jc6v4ZcyX<jwUO4O1Q7|9|BWP
z^%=oz=RUOlt>!oH$ljR5d#7@@bk}noq3<UGThC4LKLv>-vNInYGWAlc$LllM%6v(J
zq07}whF1iLwmQjc6susJOY}@iTj;<k{d;e}$t=`{06hh5wWDr}zdv!F!s&p840iXh
zaJNq%_}gc8$I@1pjD*22(N8OgLq9Dc5aO8T{LIfex0>;+^|Z8JA&F})2GjX9SpE}<
z>ld%ij_wcd-Ka;EorNva@97@ZwN<~jzPhRIX@KGbMb82$D46q6;-hB(%x4*AIw;v_
zQZD^?tcG{`jKaU@<N8gBZN_eR@&HyI+u5!bm~T_&KgYzK|B63Crr3E<g*-I1p-V0o
zN#w%V7YtKO&0ttxcS(Y1Ua<VLNwL^*#&ObwaeETA#<X*F%gcY5_|-A>SVV07o-W45
zdTmxe&eiARlXduNZ(?HV{bPUoVWdwx@D2$)+bM=V&V0`NqO79DT4fv{EpunX23@0d
z)(Pp=Msn$ja`Y>Ie#sb+c2D6{?4J}Q`sbtZ|2EtRn&qG4{|7yiM8bQ>omj5I_u|<l
z8~ydF_29W<I0Gpm1pqG#nKF(E1bZOmcp-BHPs)%x@qJ?>a4#`Bg!KtfxKe^If4;!c
zT1~`CuM8|vj#m_)I0U7nTp<!wTay`*MgL*%-T@-l{qs{9>zSA-gabqBH0}D0u%w`J
zUX;uBk*|vJXzt<&T6C)(Z<iD?|J=E1i#-sI@npLE`?%nu2i^(n)w(My3F{!(=_6Yt
z^X|pTixUF}C_(<yA6oD1leEDgwycL-0M%Z733QLdG@ijlXomjGwW!sPf$9{*dN`vt
z5#TV_^S?)N2!}v0Ge&dT{7YAzn-8qHr;%o0eaSrW{l^SD<GfLzg~lmPPz1q=35G#H
zno=pyI$FVdqV#~5COk!O(}3P3{TxBz3G6kTsqIW@_j0V@dT-#y33s;$CKfNKa2qzj
z<Px~%CspATb1}+o`NQ8jotwDw+K&7s{CrMF6aNH3_{pUAk?+HS9zIHRX+T2|&?sdH
zhMNXND=5YE_@656IbZB(?eOuvaW)u^ctM4kYR?umeN3O9xXHs*+o#p}Y6W^hw0a2T
zUz7<mSkr^IDZ*8vQsRn>iCx*8K<vt`ThaWd0zK=En->68c&9YeO2Wr&!$t~PVsaYF
zZ0r4)-r^%42bSdogZ5*&-QZ$6M6FZD<@UyU`LUQ?x+JH7ZUMwZER_(BJ$aLNMrhwV
zs5=k|CDo#do}g7I2TKIgj(rnddejHBAFJ#;g3dS{Oor3W=cRNlLS+gaHdHUCuu{%l
zL&0c55e}_`=8}Nt$~;r<0)C)=N49SLRhUIQw2VjI3rh}Y-wtf2Lh;fCQ*t0JR#q@l
zy4e^mvR|X&%U)z+h`Ou&fkI)<IPQ@{4-}pzdJ$+>;1~e*j#m%Vd58=E#I>?0Wg#EL
zF^#O<A7p4kN&|2Su4Go7uhWqLgV%AkUt(17=?t<Dq&S$a(FDY-4Z^X}HUt&a)JEO(
zsN;g4tgm!l>c)!WH27s}v`|o;f}V*d*m{RpjQ&WD)KeKo+A|Gj9mhY7tlF9uIZYXW
zBt?jY?zpTP=0YTgtTO$aCa<92E?^mt7Yu_z%tgh47r0gvXs6+7v<WPna4a6p{|65N
zFHBm+7-7D1Y3ejj|Bv1R+(8)Y?E$ypz0!X^zy+XXhdt+b<qEFvB${h=GL=}cgmaeZ
zo0zaLW;QlAltRGsbPf{I4Rrj#0}#8njv;{D01Lp}QL076TC{rZ5;QQ&WG{15^Vu6E
zvXK;cb2Eb{Jom;vL30f`3@Xowv9V6jD$ol>5C)2(yyp=z8ySy3VO76}44;9h1N<5+
z6D}5TG4R!Z+Y!YZsFvt5;)_Ci54YNZU^GkWw5CvHf^>SNErl_#+7(n(l%ptC2+SVH
z5m?1qdp@8S4T2fH2AUrPZOzJR8H5=W=4HstQEwnsmvLWuHOL)_!UM>@KbGWAK>v=T
zts`ibqTPj*G8|Q7$Wh7l>%R-qq8MDCtpH%-#=`^y&mv)C*!~|Fiz1e(6%!p@hW-2g
zy^>kdYkUUWY+&(o-O=&$bO!hg?hEXEvN8TX3bOQ3TzB{K?z7uHdQ7zEQ&wJPrtO<s
zuALmJx%cjE89aeyFec8B02EXW=r9y~+Jc=&K{7!uk8zIRTXF*qeevL^@$1dx<fyf>
z+Q)HB0gk%YE@AxEA8Qp8AH*>`&5y%_87<$Rhhf#_;}7F1bjD!m_EP#gAf%&+YaknU
z7hHdFi`?z{^+R-orXSNWcfR64rVV8*{0cDL%mFnXE(s-zgN+W)_WI${+qbLG>xI5Q
zYyve+BsMYz)bh@(_vT4j(I4w~Qm~_Y(2VomRaUJ`bZ~<+j5ugE7ozi6T)d1k74Sa5
zU3DH6B_#<6w}C}DPIl@2y7|M7paK^;4P{=a(V9{V=%5qWS9JWzi9NerRp%Hq_6J{K
z^n3B5?Fe0|d%2ItNXXEsfxchO@Eo!_uvh$9DYT;WXTHFV`F)8q9NcCxzi$9U+kMaR
zb$L^rd{HgJ$0V>C8WEKyLK307?{VIhCS#OnDR@&9{D>R`c9P*7U9YE=*oE^FUf(N_
zh4>SrPWwthh{Tgu*v`zHl-3>KU-KPkEEuEO(wI4b9m0zfQVf8)i?cH=F{Vd(%6+Dv
zJhwK~cOBxw912wPKu!^R+p+ghrVw9tz|K^-9oyuzG+o0%s4d!7ZKwQ~7)rUgp7$xp
zX6gDBIXsiO2047VeB4*~XOHH7Dmo#_UJBCDH4%+bSTfk%(}!JJ(c7osXa`gqa}nK$
zk=q2<7a>D|a2_|K4y|*CZWUZt7>VdftpWrHX|Tt@ssxTOoA#UmXkh(^%hy&imG+!n
z=mwdIRs-1q_RU%2P*}L2m3qG2gt$l&FS+J$+`0)EZVkxGYcSbq2j90%zkd2A)-FWG
z3-M{DeWt=YTe7j+|3KGJU)!f33bz65QaC~oAO3e7m9gTP`KcarGANBwYV_2eoO!!X
zO7X`8vK`8Yj+1}Ln_g81)<veIAAe$%e(3hZ-d|v|!)`SB?K{Fa4+nNi>66+&%<T!E
z$xVFwVSVsi>Q;Jfd;4#6&_<uWz<LFX6%iiZc*Wqq%@%7NAT0Upi!s0I%9SMfI3)mO
z3nrU$pZ~w|-aH)ZwSD`R)gaY!HJ7GEk&F$ROLe7*1__DKBqfbhlF*9gRuds}D+x)h
z(jZf5GBi-6i3UTGC?pZ@=e)Y_-|v35_j#Y^dAIHT@Ac1pZ!50rJDlfn9>;#{`+j(E
zL79D>wr<&yg7Jub&97{4j5t|YU;N`q+>ZPAPr=`x8%=WSlwL(H;MwwXbZ2_QLt?j1
z2f;_>POf^4qXho1i{s(`NpaSlGCz(DGprrhx$}c77kVYFwGPd6TD!7K9Zw}N`bFZH
z-`C^ll|)-%e)-1raQs=TCklFX?YbbgjXRZ!W|zB9(}>i0Nv<h|s(@l9bMGJiPwsYL
zSwU<z9OUfE%)|Ond!2<WGi!MLl~p;X#_f~chHnS)#x6W}I%B9e>_9WYjUJ+**}Db@
zbG!xytK!g%iqEeQh}m*wQ^VW(BI|26{1+v6%^6^r{juh_o=q3OaSO}2+G!2R)9>`0
z8~b$EJM*+<7d5+_K&Eo@Hzv50PTR2ILQv34fQDLHcu+zvE}wxf2F8i~Fzf;AFrR(3
zTL}-&PO_@`^W=ubvPxjG&dq*<pY|X=ocUlhd$!Al4NFE1PZ!Z08lYMI^p(MvI2I@L
zkFT}eb?era-Mg>uJ!YG1y0x`Wq`gbfD3>Z*o{>}G;d3X#x5aFk_$+$CV3Kt*3f2dK
zQjEbq`5v2o($xjg3yMoiEe1|W_%F2ApGLmLAkpF{F;G|iFW?x0uSPu#j)Z^O!#Cf*
zXz9h)CBwI?g?=|FdGm%*g4lOjlolcZ%2fYV&kX6cbI*(1zh@||Jn3>R^N+DLkN?^I
zcB`9Rnt$lW>@JGirc61Zs~eY_yH_LPT)pOZ_qV5}0l?i349sbaeqj%$`gi?;gxJj?
zn|G$a9cpt=txE>kK;|J>34Xg1o_AB6R(>d3x@+WL+sv3HaXtHZm!rfa97zHXHVC(g
zeS5P?=si1P)*%AfkaW%S+4ep6ii177Ly7;K$YBXP7B3E7tgDwideo>Z|76HEvKf%U
zSskZQL5mL9{%1qr`f#K1`)*fPFNLO2cxDcT=E}mE<<GNo&)YbEe%)2q?)=bE=Wm&<
zeT^TnFVIBi94;XhJ;~;}p`E808shUcJ3YcksUPwT4%xh!p=vJw3GHsj6qW2=NGj-E
ztu$tyd}p{(Cp_-jMwq~%yzozZa*G-8jS2!nL;DNe!+;&Nox>h2tS=-(322|P;jMI^
z4e((1;03ZfF|pTp6VYSXz{+3kf>ZVBRqRBwt6j$I>RmQ_)~uz?_mel=JT-3s#ce_F
z-q|M%RaG7Tc294Nik~)hPiE0Pb%*>x70fwVxy@HAv7!fDQsB0gf%Vky+qZ8|)vTL6
zWy+MBq>?!ONDU*)4vltvx_5{Ll(|LktPfHw<>k75<)|qHt5p43Q=@h9Y^a%L-G8IE
z!}<i;?G|uxQcL-W+siJdrcPRrw$QP!fts2<ewi^N-RU#X@f|UBDGYz*!tpD!d!6a|
zlT~0gI2Q#H+?vz1^)ko`P9d45&PUD{4iq4&p8fk9VCKs)tE8m#phM``F=JrO4>T~a
zbt^G?{~%~=RoSQ!gA8*F4R33A(-oSO*0BH8em^zN{6_YZjFL?&pC)FXSTc?|s%y!+
z(9I=SYKznmUrNXYd60&v?+`W6({I_{^nF;3#tk$j^x#KS*zS9<Qm|KO;1E;jnov7k
zymaXRma-i>jQ4Dw5cELT`IPHOludMWptue~$eg8@ij^Z1v>7Kb*T`rbR{KwOpgdoA
z{c<x^9aYHfWnUNz0J*+^Gcq_h7$<Q~GW?>x3Q+zCp}2GBx1T?sUz}k38?_^Vuo^Wf
z=rHXX)&O~(7<#+!1I{?TQJ?It$LMq&sX2{m>K5Q#7!6LKtQ=j!f)N9mt(7nQ8l*eR
z*9l7{Ig|I01a&|F;RS!7i*PGFo%%NG>{$)Xs#dP5LVQPM3M;@9+MHHGy~C-ihK9v}
zDmSkdov>XR8Zpj1C@>>F20{IuJ3hVLV4X~j)H_#L{q?J3&Z#>DXdbP4j|0C1hMV)`
zK)uOT1>~hR{X%cD<sBFxQAn7XJzZ$GBQjBo1a;BG^LGh`l`!haaTz3njK%lk;vH11
zOK-K#i~50?>KL{p$|PzAG=VYG$@&v-+!(REPT2f!-Rh7(^6@9oYzkT6e+ZsCKW+gJ
zpbO&HxP|)Y`eM#Ep-pVvrp=J6LhMLFF}u}z`c$7c%63|e97oBPVTW(vgFI-=7|RF~
zx*t#P+?h6Y!uXM!bAL8IG&H;%`XN8HetPFNlFmoteD0iWtj{yAGHc-Yh=oT+&iQ+%
zPCPb3^C3_z>K8^e(dy%ebM7Y|Yh6X^A#@y`&F*>PAbSVfio_j-<gTyJ*BgXAIymB)
z?L$@sF3^h^fkh{WHZ&X*!13buHJ+bdZ*!kr-+b>&m3E+$>Vex`7Jcfn|Ab-X{LALn
zW<ySy8wbV4UHfyD&a?{0`JJs^T~WAbk)vs}W=M3T)#Yy^x2PPop*9(wI_Y6@P0;v7
z(}o|PW&3E-k6uMJA<hOP7L+EHHBKxVQH<#g@N@3wR&7a<UxH|N&`Wv)?a(wDH{RVH
zd;bG|g;`TlZLTpliD@FI8aqU7tYH6mLa(5C550lVmw4^k1SapnKBuu{KC5lDkL0(u
z9Sf4TZzXpjg~A!tI>E6H=#7y}40u*hSuuPOJJa|}sCX-PmbLC*5=*G1;^2vj4J<ju
zHc<bPJvsEG<AC0)G4U~pEo<Ld+b~A*+PjyEVLO##Y2U7@R_a+f6+V(Dr4JKyKBzml
zoj1<>WlToNSASc#@<@MM&BT|SAvL3E{why@f!hI<97&ey9Qo0U6igybyI<1}znnH@
zC+21g3LO8$?Uoluu4I)n(%6=cO3xY_H5C@(!{ZMl&Qk1Me5Cp5juEW7{`h|6ey7)^
z;tgPzVJdLd$%|Cd{&pYIZ9fMu9K5-S;fWXO%GQ3XbPgFd^;>mB?L7~vc8_JpB#tU6
zo!|ADIw?7Yr^g$h_|g0Jc%Hs_AL(As6=SWJe@_-zz^6-(yd`+m4((n2tfDSgvo<%O
zkNBQ#$U!qh-kz$bqkwb7XU{7?x4vEXGGu(Ga+~3E_tr=ccE2@G-PugDvhou=;h_K|
zM2r)Il8r-G#f&n!>=3%Glz0ack?nc$C+yPGH{`nQIAF+Uzgaz0{_yZiz`yb@?Ofmb
z$U6*-rCC*x9q3b-vUu-ZsbNa3Y5u3@iETUWH*o2EDI`pMVrvRt;z2VlKv=}j9200Q
z#?JryE=1n?%9l(dSM$aTRI4H0n6F4Xu9}uE@AUhCEx{A58=l!stIcuw99)+(<ArKf
zFQov;5m`U#t3080QiAdck6e$($FFC1m}X28lgTvng4f(S=Uet@e0FHui`3M;1}{n<
zTyn@Z*;=vkN{mT)9tkK}x>UhO`bEDHbD|eKOFoDL{sv?_=2eXwKYsL@{7-%hZMVHY
zG}<ynz^Nts<&;Joy?1ZC_Tp7nJ(3@nD^?s|cy--3@AhNUckhcbUgWip*UFNOoxMfP
zKZ^h?!q4#M;)R%OYVGPfev#G5^kZsU>^ya{9_XAqT$nd<`=^3n&!(x5da3hEa^1Y;
z4sRN?y{TYb!=^oYDM##AbjzQyetlTY!|~7G-z^Ubd9X-)n4D!#`3aeOEY@6Ix+}db
zEURl5#mc|UO1-^T%rGoo^tn)&O^Lk<l@U60@#oLD@HH+?-+leY4eJ;g{HoEhRe$?O
z>Z9_%{C(KBr(Tbo8C|>8uKMN_lA8KFW_ZN2ik{k*lHR?eM<lAV!GDTPi<|bE_~ez*
zx5Nd6Pv&;iO1`}RE6jbF^C!#BJM?$%d_Ka#G1Rlr$-Fxn&rPUe8(#ADwT0WOp1*wQ
zpLfzh{H2L{@d;fyKks*6?W1!3wsEV|R86ipIqCss{mY-}9*C+3#bQ8q_bVgy^s3U8
zCYi<Nm1@gc-}|R~-FtHE-o1rEV{%4n+~3mp?beUtytqcbOlQ=nt9Fx|^DP4Co2Ptm
z_u!f#WL(OVWv+LYDNZzc<W{A$mA2lpMn;S<r>3Ce9dS96XGz}7SN`YT;mVZe7jLul
z?3D)e5Rbg&`+yeLzCF(1<{zBkXDIl}wX!<nl1ke9_9Z3HKffv;v*tw8!w_lX{gUgy
zwOhHP_Gp_brG=M?Zrxu-kut74D?{^*_O5yJrM1(t<u-|1HY-m|+*^CTy`s@1*P+hk
zxr#xBJ`#!m?JM&Da2}Y*?$cM&s=wN@n#L;KWVf+OeJyX}Q&Rad3Sz=V`vdu#wr<sn
zK#5A$)FUT?ql@J~+FEAVVWKTRVSmK6O?Ot;t}65R>3i>R&-@wBPR}aM^y~fd>YCr=
zvlm%M^C~>)=8{9SSr@k;>LK`D_Tnz$QFJ>bcUYPMrH&EkFRnY!uisPi?I8)l<*9o#
z@Be&jM3%Azt0=e6*UJv}p_<o<Px?xhBRs)lhKdi+kQ^w8kZZX<>DB0>`<pf?RxB7+
zrBp9Q9th(V&NiT~ucQ<x1ewB}93LMaSs?@6<<0!$;!UHo-3@)DDibeUya*1x(!eb`
zqyMBU4Ij1gZu$Ik@ZhZ<A1W(HF28$r@a)G-cxha#(@sp#^T}>nsOdL6FgxGTGs4ST
zDJO=K0b<7BRgoR-d^TMVR?0CV{ZsKq3|uQ`anI<{%%ulM<Yq>kGcXLajraVKZ07Ek
z9O`*#e?;E?yx2SE7O5JfJT@?j>^0Uj`=r{0i4QD$i<=A2Z2cVr{1F3k&}di<tfcHv
zK6E#GhNtPMix<B!lI>xb-l@at8Xvv{k4;?pAlt(44v+ffix-%@Stm+-k3l9_tV@h`
z-<uyxkg$pyHGcfdT%T}#U!zE>Er~SF!ol8tQDR?T1_z!ckN{#?#3s1tXmqL=H*ng{
zF#{8}+zCC(F?GUyz4^?(rGx*?1+Z7#feJFvGiGRCT})}Bo&?rBo_D*nVA#ER21Azl
zY*X<*v!n4`?-9+=(As$D*w6h@bVk+FT|>pdF!rw9)C&86#=Cn~Rce0E=(vB0j>BmU
zd56<}(*xc2nyaM^oOH)e>-U<{?iX(*Nu(^MZ|w$HzW$|PG-;5$iB)Y#-v0a1qgyJT
z^&A!)T2>x75zX3#rSDSLgo(!{r@dFsF5!M|Zr>hySFA5ePXt+84oD8N{x+#4aD?0C
zy24+oFU<k+?tT##FXL@1hkAPC)bGjKRu=MU<NI2mByAb)#Nu}Ee#GD5hI3T<xkdZO
zZ&j=u=wJ4&{P@H*Cmz0hdC*U`JB{|s*=0m;soWGiV#dZ2qs}{?CU19`6Le<Hs{$u;
z*>8k)Yn8(A?=f$`XAB;Nqpjh9wUa03AMAbdTjc`_nK4JWi$b;Ek$1Uj_dI_Z*JbH(
z_%t*+%n3X1QLc8$T-GIURc`&=$fr@}vo+a%npgaw73Jl>+j*?6e=z!=;9!=jjm@Qd
ze?4g5DevX;z)PdVgGhTaFwJ`6qPc&){#x?8ox6(hJUvc;hq{CHeSdwBSDk8_^@z_(
zG9cQ#ZL317=eO^Udl(Rrv>{@z!O8IObMp*xMy=rh6CWp6;w`s+hwtFIV;+XCb*^32
zdB-)Gt&p7A@Yb$tPF-`$Tl#tN%YNg%ZS5=v*u1~{e1J#5%d7E6GO~XiA>oTVzlqz`
z)VyO)n+~5wIZ3S6F5S`W?0M~oB%*k^xJj?f-KQ7k_<d;JpsV|Q>C#vKdQrpV-P@e8
zc)6o7E<GwNbAc%*k7-HA-p3|xcC<aDyz*Q}acxW1wf?R+D{<1RdjquA73p3+jMqru
zh|LBD|Ni&h{QEtQpN*xnOB!l;!mZob?`XPsWt;QAp6~@;cN^8z?ax+c_X?WStijeA
zI&^#K><2ZEg5osIEP<`51g<%k{l&gUIB#Ndq~C!*#CQAkY?UIPv)xtWbD@rtR{fq`
zHz5ZLV{0g?3?*H4j-&&OyvX%^rw%=YSS!5gzK)Gtb68>isPU?*p4)9h#vN82_yp-U
zXG-pP;Rwz*LHl@LtLx2Jf%ZyDVWZcFMk<QK#pJv~b$8>3M^wg5nv_1EE-%J@hT-X-
z4Gs^h-@ba??-&`BFjV$gOA=4;J3Y2SEW#!Ix&7%8_fL*9S5PsF34WXY<gWol$G{sm
z(kI8}F~_@S&Xo9Pa2sDa$;b#p`?TGx)wfP7e#N|ylQlmowr#GkO6*}a#$VS4&qtla
zN6>n*7swvHvufHx!{=|`p3rU8o%yS#vwnQ2bL*X@#@EXEfswUZ^RBqq%qfg4<+zwJ
z@0GbnPRX(>y(B*)bC)nEHGxqwt!t^FSG#kwFl}U5_d=I23dEED@ekR^Sj;f|yLAOA
z!~492b|W2RXHfPoKCSn>d5<`V3m-kwYdJ&T4V(C59o~p9GumhVp;g6`)3c5vhZl`Z
zG^E$*Z5l;7=AV0xj@%0FZF#P^SYL9SC4Rm}_8Y{{g~$%FC#OC;LbgBB6eZ#V<>W5N
zw{{i(D|b$=)vwO7)?^GF_(@Z|pg7UQT(GR-lfPLlwJw2d80A2Q#t}Ud7!bJMIbXH{
zEOE<_NcU3GkO%fRaCZf+;A{ct-JaR+?!9~B9nvY#IQ~_9FoI{Zu|oX{Dm`mC<B~|u
zJsV2=#l$Cjdu$*ZAb9U64RkQt;R_(Z1O=R2$Li{Sj?NSt>kscg(ta~HS+e8}bw`I+
z!kP0h^gG&C%7+AzfIjRn=e6RC6w|x8$+gUjG-aB1jG3m&ZORO78<C27waoBpS3L>T
zw$Dh%LEbYN-~xlxMb^V<-h~su?CaOA%|y)TlW~SlRC9H@62J_%?r33gf@%bcKl!2U
zwvz5s%g|roKw*3sG87FiZwWY*;G$&L%FgD@y>d!vPr-<MB$PfBU%RKrSuiE}8hJ;b
zdf#JBq?Y5he7eS5K4?+S&xFO0S#RD_H6|us?s<E+R=IsdW$M9~XeibUNk$Wc3}-Dv
z1AtNxqWhP=958m%3JlFmACb$$L79D0f0n|0E>)&F^2QLsIaTB*;hUDW?(Zd$@`xFR
zu$6L{<fpszGL`c#C4-@#CH?y8cw=(VOv993Ri{cERJMRupU3wA^COfwD&C8AM>j2G
zm_gVcAtMj(rcgcEZv-zAn5pL#bM@oNU+(syl5@fEM?WUdF=TF8dHGm$1Y+lJaB=yR
z8=1wEpaEz8*EF&F&HK0NpVmUcf?cQo=RpqPrqLLYn-?sghRjOU72gpvke59dl!?^I
zG52hqBlSP@uFotMZL7;4XhE^cj-R<%7WJ$GUlP%;q=KLp1I@kk>VR0DF%#q`n3TY9
zgh!f~Hiz=F4=QSr=~9D%0sjq1aMzk~SN}j}qpZ1e=MEjSfJmA2t7QrcrG>uO8OFh(
z_Es@;ayqxs3q@f$x4n5>h~(DYyJtYKJ5dH=?f`%!WMvR07oA%rZorY%`h5UUqi_E~
zYHGceVc|}|N>N43N2kSR>-x40Gj`MSsmSw^Mv66K-dQ{qv1?sim_W;ffYJKvn4hK)
zVH(XqJgZA9=NylknJ(K)ndUdO%vznztaVG4sFDypK6d2DAKukpzGQ?I*17%UL+u}m
zU-XLEC#Mv?<Z%Cq!zZkJ9KKWX4qpKdfxA1NI@WO+r{(P4j7hSH)hQb;;!Mn+tXkio
zpdiUR%DUYrP}t2^h_Hxl2Z0ZG3CdkxRvjoLye%sX6fy6XsM7Bz3#SE@yMUZbf)7On
zW=Bb41a)KmgBGc{j7)Fw!B?+dm8hb3pub^i%r(KJRr9{aD2#w%%kg)oHk|}U1vv{R
zXpfo)qfM_8p#MS{<K|@9v5HxBpPxoQ;ldvL94`A%A3oHKqAjxB-Kz&UTng!wi3h8v
zr{`ycNB2lX^IQWwfHDvZ#>H#~S4>E#`}+0R57vNC{}mP%YWl@MOobd->Z!uVPPSJ)
zZV?+(%OQ~6WW;1QI50jEUY{tbCg^AG2aE~{v2k>aiR>_C2sAd*<Ye10yz|SgOpLIc
zCckx^*4DB!R#ETXse0dk@L=R(yWc7p6hhn~TIeeuid-{g^@izI?Kez!tlU3w;Eq!}
zb*KOD<MRJO=>D%?F*5FHk_0-+Tcc3=MEO8Sb^rF<f#V+|W^EhB`0!J@Am=xC>_A!5
z%$`&6X67*hZFX0gfJ*H%yW!Mtyl_wE8$N(rC5|;|it;R6HTJ`T56opEqM>%KL0JZB
z??Hvc17$T>U*8lNX}oK*@tiq7aVZMh@lA9wN3F>}M?DR%bR6CFe^oJmntzb<<lvyt
zv52);3tq&lkScH%65SDzapJLo79~!$jAi{{A-Xx^n3ytQYRnOWtOK{x-u?Qmef~n+
zPx$2S<I<>}%@jIIBuge(r*JSMr*cbfW**TAtxgu&3~;0sLT~F)>a5s2kMD2y@I;1s
znv1_Sxvbdl4bxcEy;ngNn04c+>e@D;Pxp7%>e8i);>(z471UyDAG*oDXpm;D>K5Sb
zuF{#u%MQ)b*`4v|E=xu@;9LbTW(MDtzK+_yrVU~B7LleICEb7QITo{$#X4;(9BHRz
z%g~Eay!mFVpr$Mx($Rjd`teyUB~vfk4iB6iVN%`RYDoIKaH<~TaeL-Rn`zfMMvfU6
z*IJY__S>ghMf~Q@kt9)VajZe{&{fLtpEA!(7#Z}~o-|d`MfDH0_h_<@U3EH0xf_<z
z+c|jr`&)l(>FX;B7<K=wwN6gjHf=)pn?@JMN0?t?bGo#5wF*9JHNLrfT<gr@splT2
zrmxmlp8t63Zq~d+b@b?G@*{8lU}O2slCk`TMk!8|uPQEP#7Fb@FUyZ_S^X(FtuYOW
z$Y^bCZBjzpQ2o&LqZr?<Je<LOIpr>Fp#7y472Ao@r-Vlx9XVr+cKtHt%iPj5@m`(j
zd?LH)*#P_dg@p!sdWY?t;(TdmdH8^Qx$m_w`XLz`c+ocasD(V&d1Kk5ne0#Fl&ie9
z*RNiI$@A)0<mx6|OH3^FG=i{G<W>>aR5@hGkOof$jfK0%hB+o!#aRT^Jl1$|KrhA8
z#fw#c5j6n0ZsdXW51KrG&pZO9Xtw$jS;Z-axW(pu<y~-*ZCpd<-rKV(lI#E}CR#^*
zy)G7eM&ufKjWAua@Kf*du5G?6PV^}R@m4r`@}y&K=!p~R-Zam5w{iTluFr^-2ZMsF
zJNM&2alEojwuZblUA*eQeCaF=9=AF3^iR@j>0Z)C>u`0;Nf@iJqrlV3im;kAXsS+U
z=3qdoruah*E1%!1=`3}Z29Mcz>-(+iLSKryJkk`-!J_l}mRCU@`(uzw*JF2srlQX&
zjyl3B<mglOYs3lcC2#&=&6}`orGbF5bUGS%8Uc(emqrk-pe3E3{&zWs>0NC!<pQd|
zeQO4>Kw0csk*~uvK+N#p1pKlYld#=7sHo2l7>ey2D#?%u+j8&SE8`{{6J9)d;tY~~
zLbsFOP9$;cdX}5G{BH5mc@aIOynY9y#CplK$9SIfAW_i(K=D16ZzdLtUz9t0PV<LC
zlH?ZMr%weJPn;LWq@0{_%!h&1vi~>v-|<edwX3l((T2Xo785WtDGV-oO1*|Ix#iuS
zEv}LNH~e0H1oo`vX+-)kal(W%^TKR*+xyNXZz72^XlZgb?kW3{9xK~COIv8f#VF9M
zaxOi%@jZ`)W$R2*#ETIVYT<@}?If7&)UYES@w4gr1dM)j6U4P$_`S7jhXP3H^i=X3
z6C}4DK0HrrhF!RI^Ja8QiKJ~<groEj;!Vb%8P1qd|1mEsx9-#@A!i^6lQ`0mI&ofB
zR&M?yEeP4cPxZ*HfwM=!mtcS_3h`cB8NqCMaoHVW_U3UmFt#8XfOI=VJIeC>?yaAt
z1(_3KEG+|k3u&-4XABMtIK5TH`_iQvbcd8en#VJ+Bc6oV^nkBOL*>qzC{(Y+XaiBt
zM3vhd*6jWRAI38A0Gr_sDzG#>DqoBok+*Q2zVaM;leY0!T~zJ}n!?<<e?`W+Y9b8A
zjqca$n*4lVTF~LSQP!7M9+F^^9zCc-l3@7ZEr$4;!5!<G<CyP2hoZ9JBEnXF)|2z8
z&kJ?uP#CPan#>N4CwuvLewvu?LyZbq1t?oU#!!{<RfoIey=PC)@7-3{eHRKA2M4}k
z*2gMFp~5N=Y9tc#eII?9%E2sML7J!6f>6P=D!HCK*LKMxlKtt1hB))3@>=-Z3VeRJ
zab9@zlXAt_8b&>^g?KL!gh{TDz>nOfWBd==HC0cej~_mKI4sN#i48ne+cz7bsd6?#
z(WEs~^8Wo3aC7dqFEX-g^YFI7WtrFI15(_(P2s@s;woqIDb^p;IJLY*Ylp>4Q6&zi
zF3Az%m7G)LH9o=$gutTd-D{+yFH<cq{=_U@bXDX>GwX1$Hm=7(#iDzD-iD|B0E*Wb
zkEZX-ww}*(Y^jU))Uio?`gjwEv&hBKopFq|F(|?M9~J$G=8<oI^(_`Ng93h{?~nyw
zCn1B2q6}cp%ZcW*7O|wtVr&^1v<KhHcF?^tcNfn~$MQm2^*v}3qEg~pzkj|_b+G>L
zJV)#Yu2SH<FEa|ES@Db)!Sa7uP!RB`i3ch^2e7X89-IrR2MU;S;s&%h3}G56ts>{p
zWt=-}Rs&LE$fT?Yr=;TqYS?TDnI5=a6|1>CDvWs}^S_RN>-TO)52<*!a7Vj){w0fY
zref_a9rSJApKCm*S9&x4Q6h~t`(0UIW9U$mXfp!hJ2sVhY#zPoXzQz=X)*6F%s|+}
zGt?5<Q2(JzCKD=zPa!o%k!Q;e0+FK6t<#?HIEhw~E}{oQ5Ap^gNP``2v}bs$Bq~oU
z>R8hwx8<Xjg20(a-AGT5bR;q=YCZd#LQMHk>FcdUqEl<S=v2#ZymU7V(y`1=`uXi;
zt_R`kDjo!J7LCY)!|*uQZF#g<_!d|hRY81SpVF|Bsj^I>;C-g@7eoWHWvU|$6jdu)
z2sU};E=+3Smtp{Kpf$24N&DG4qQqoA0`Vv1+lw_k>lVn31dcO#j<l|t@~zjCR2aXQ
zihzo3%Ahu05hG)jd-Ynie*J)+J+D$RQH;geCMR-}xzoCeb`s(7&o(G-EKOJju*4Kp
zI@bAr-eG_!iA9FQ3w1l4_ae@wV@a{HRo$Ugp*xO(k?qJ0wd>}(j0?GZE`(zM1~nb{
z)89Z(c6)m#QC{NQ$qiw{W)W@XQ(4UNp?h~kb5W{CazyBb3l|`N6u4Rz{((!Tj?Nri
zmPSevOV{*0;7B2$D2;Era=MHDDXx2FxBltbqhBuZR_bW0kpbjnWXy6Ds}8Yc3$Fw+
zfh`FvYe{We9xGuDy=RtYr^a$9Y}>1nS=6P16QAnwdc!L`6;urQXF@^+MAd?H;UvMU
z<5mZZbJ6hhdRFdoci=~G5h$T(MkLbP_wJ$38aZLY;ulZhE7RYHM44bEM!!t;DuNzM
z8GBluCh4^FP)@D_o`UcQDY!UQvD(zazgg>e#%vF*A+H<z5~O?ef|h%-X^pBlw$AEb
zYg_1k+H0`4#!!36_uIPNa-F;xV^W?+4=Hh{PwtJ=bYTa$SN|#nO%3$2N#9S$-V_}D
z_jY2i346%uiqq=c7FIQzx`*^&|A{Y-XsQu2>8O&O{+%=P?el=ex~TF}R5ly$RvMr?
zQR9>1w<i9W19Hx{5oq<wZ?e}T+mUZYJ;5g8FQzArN$#N1w_iUzmsP#5;S1UP^I7b|
z^&HQ4b$0td2iyX*ct3~k_L2TfSbh1}wKTi8innR>r~J5}jSdjTc+XRIwPm@+7x*vw
zSigw;or>nP7&;KXNw(D!zY`#glaEhoX<DE#Kh7$Fuv=WD+@;IA($bY5ENbUczjx`?
zE5NgY*XWS@4@~O>D;;mw`qx`)>XkZnbSg}`E_p|-pdLg95#o99byimF-WQ797+-ta
zqhn8xShE;bd<Z)|J78GD)?>lJiwl|sgM4^k5n}_~9ut0VjUq+vtsEKQ8fT$p6R<>E
zRdpG?SG7e~E<CMrCOJ7hN34F$VP_r(9qtx)$FtqbIafsOLwccph{Nv(rS3uN({(Rd
z;|#ny(juWog@y;#HKL<YS^bBcfvbmmF4Zay?D*+G%kO^BrJmrflNR;8p+SV@mf_{<
zBa+p2S0o#sPdo7Y{?~_X;}MZ3afI>k=&u3*KfL7zX!hhZLn>z}p@}r!^a|NGRC#*`
zhZ%;3=j>7iOhWDGL@4Oob9nK>kM(0ckIF85fP7LzFVlRYVTk7h%(py{$f_SAa%c}t
z1F7n^rQ3i3VIV5qH-?Q$Kf})HSr9?wW6CVGaq$o?K*Q~e`=RN_F`=4}GtqOXZU}w3
z6G=6=Y4UrB6U8*lsMP4~?LF760=BsPyKt^z(-2FBsCWZ=g110nWp`K}Yp9I$2s0iQ
zE%F_&yrqiC4ZWscQR+L!<W^L1>XI7;joz~koGK|T?U8?TsJc_H6{A%Av*zgjG0(_I
z^NQKNji2=1N#<#`#C-Qzl<2aWd<Y%xMz&(e+-EfTXc=N!1`M;BRMy^f_EK;P5D9`7
zlm}C^7b@OXaf}bI=D-9eBHwFXR&VmL{yFtuw>IJsx}bY7EodHlkaUyTl0Xf9h=DBA
zl7@0nq_#BQn6;b;4MDvM9%8|z(|&m%N+y<jNN;G#CEMB+Xe7JDI=}BhqQkE@UM~xM
zdC6g3&PeNgigJesCuIx(R@DnQCQE{ykV!oIsyjK-RCLrilcjsf{86yG5Xi(Pr`jK1
zmE=seHl(2EU8_O|@k)E#R7h&j@XNef<YBE4k+W*$qfi^C7U<$-17`Y{Cf~brr~9=o
zq0|39ntxXB-Q3NvRZImizhY-Uf`r5Hhk7=d9vqXm?zTXv6k+_I<Ti{q`8Rg;f(t&r
zv3!$`I4Mc0j(#rEjpo{B&b-fyKXvMqw*>qMA?&oO+yB0zDPE%?I!vGS7LDN?5xbWe
z&^8_~PP(Sb+{7>7NBj5f%Sf#w-umTRDJdvuoH_^|pr}F(&OEuJo)ws_tj^7af2?BY
z==c%(sc)($83%n2&e_O5QPw6>>7SBY_wEgDNvH=S;B3`Qo!YKlJC5SSw7jZafBt#8
z9T)aLUX@nQp8oLOj4Dq3tgvhOSd^OjCHb}*zXtJ)BBzTeZ0@b7fC2<3LfhMt`7s&E
z?}WDp<CtbEysfNUd+nLqPR{v3Hs<|%^jJCks0WZ5X~d}V778t6<9sxALE14<d5#!v
zqvrLIkfLdQ&GR2$erqSe#yIO!pR;)%$^B-vN(tT_Z+&@{iuV}UkuFnRmvip;_b+ZS
zpB9^^+>-O}r+IT^b!k3VP1aDFTwk~#HV-d(LFwb14c-2EEFQWX=pooRaP`pErF;4G
zJRJB>ZeAa5rs<F!@TD|p{h7$f<R2ennate2V@Kbt64cn_i^=*W9zbc+;vMtJayYHd
zTVI9J^cl=eOzB{0QkjFeq%udF!w%HXCdoDDv7P$y2Ox8BMP|*LM}j4g&&iX$(ATin
z!4L+|d7;FF=hVw7fyv+B2@auSf*bRa&~{YU*UvLHzJWg#Ch*Sh?{;FoA>t>L<t#6R
z*uCT5HP>=3p}EjauzC%&#Q9G?g~bK4i9Fc}xnDQ4298Tp3=^+{q~w!7d?+hhNajL}
zvCL`|O`$YZh3X!XzI{KHmp9yOHz1e_JI;a&1wYz#w;#$OF$sz__I$;lMY((j3(N&>
z4Oc0oLiy*0ImM6><K#e|cpz>m(VZ-YlgSzPH9#fej7kFM30!?5>FY5R(u5hoY37_G
zJ0pjXC62QI*c(cN)UL;*Ce?k&Sq|FFvp^v?EcAiy@eJwIsbZa8tN<(J))`P3*-;6o
zzaHi-qf0FR&f&2HEPnHtybtNl`-ECLB||12@d$7D@b_WZn}g34?+|$TlFqGNv+&`1
z^)L0$r-F@8v+;FHu(FA@(DC$ap=a@};u#h@a~0;(#6q!Sbofv*EE|e0#I)PcePC*0
z8z0XRX=1_%(f}r6&QiRPkU*VW$>v2g@x8vtlZmz0u3smXuO`y-vLB`XNR|j&7a{;t
z&z`dtK(gxGb`ThOlA_urznt`QyQw<=E6nQPsSB%ga#|iV<|qqb?CbyNr>y)#WYiu_
z*CwnBHXr~IzEOS%UOyQ#_|?68QJiwD-}2{|6n@V<Oj}8Amf;k$sEmGf$gY4`YT=V|
z&!|1<bET`tx`v5!B_QlvLF4{~zk?*L^oS&pLk}k13+zI>!@0k*{T=db9!uiazWD$s
z<ewFFZpX(DytN-B%2EF$du<V9hTW#o*w=QJ!Zq)4<H%r5bss4MCs7X&akXta814FR
zFe&=W65ETGw`Fovx&Lm>{TuK2Uv%&O|MV;2mt9(Y_Y|Q8_psUOYv3-kt94|(KDlst
zm9&Kl(8)x@l!J5qRAZWltQDA`D{jW{R^{Znw_ET*rwTQz8>i->CpRST(9+TNe1SWn
zj`=TMgH$EWnvC!ss}P1)&HBu&tRR8}EU^c6zkcJC_zh~D4zsI{QMU_$=p|WGb@ClU
zudAnFX<`!5^K}_<g9K<oo$RYMwdIr4@YxS9*l^w=7XOdz1u~Q7@LTX1j!_ZbkUMJ@
z^Yy%p4P|n(^BbLL5`8eCY4sbOojDqJZ~G{grkt52;1|AXY<b<pc{~ZY|Nrq%9S-c>
ztFcT{2!@7cuk@U($!ju<&&x)y22C9EnTF9MdE)<*y^i{Yg?sJaEw_FfetjGeib_gC
zSp#K&ANp6M>ZlH{5>TxZ$B9dr?DGeJ>520<`_{TmTzE{RJ#yz3?pwFqqCI`b?k@*x
zXnf#sJ=PzQ+9?i}Y4wNq?*0BTHvaTIx{>`1+^>DHpc_UfRK*YsiGI|aoD2GXMtGwy
zIM%YwWEd_(Myao0#of<M2=}Ee=Q7R)1eOlWYK?A+@bFc<En**JWL$I^tI_HD;?Hm1
z;4!*9WUd1}8Rh?gi(_ffeN8s^lM4fj{j_*jQqqKUXz-)Ki=y{FbW8T2EZ;rJQD6D^
zg07INemX~6#O66}UcG;JByApW{8>?Tc+(+#h5V$H;|5o$!eyMn6s<@z%}BXNdsZcb
z4g;52#pSQk2f2&9*?YBGTJPutt#1cw_sLIqmew<SOmNQ+APm%tBj){**-PHcf$6KL
zqHCMvO4g>DSo!MulgBdtl-JkkUZa7zo{%T)f7kXcQd1T04~5A$0yAYerP<ont_$$$
zqg^B23qEx5f`O1-BAAv)KmM}5*)}k(=O6H~6*hvll2^-*pSi0BU8oowvTPEU(TMZx
za|%)1N60ZrP(?2u#e@25@>O6deyG_`cn&7_3c8$EypLQ5t4Ts01=iOuU&`sq0Ta+>
zwXz%rW@CHBX%Scut~wvlb&_;cS64UP5o<zD^GwV?ia)!yS5P<$iH_bJ_k(mQG<BCt
zD|fPy0|gmij4fXddkz{rxQ~_K&at71@$O(@ly;sc75M@Krvm)P^`~ZW3{dDdF>>Tk
zuVL@N;zpNyzP;r`kBI~D+2c*zFqtC|p8-}&z_s&!kv-BnDpa@O9TM#qfqirK2TYQg
z#VQ91C(fl?-xqM4lgoHZj3PGyqH`2mX!=2A_zo)~&#?|n=VN}{cuJF8KYw3eeWaCK
z&-n3TATjfwUc7us#>~S(46vOXoVMk0{e6><ojm!1>|;nJ&-@0b9lb)JHCq6;>C^8L
zK4{z$os1XC2q|e3xJCh&0Cix^$<*|^ww;uFtv|NANK}PYRi30=O^<+h!rZ^^libS7
z6VRJ0EW^{mlUzPLCgB-f=sIxh*lpw;yecZlZV8;KRh*g9z4Yto?dCi-0Mz}x-?#}L
zO|BG|cUu|!pW-L;hwPUHpMZ8O>2vfVV7zo`6E{sSZj#;YrUv@n_%bh27{(bK9{$Js
zIs9+Kn_WuY5|WET6c6Q&0#p}Z96pr~u2E)dX=^(Xzghny^l`{=;QvVXLKg(Gi11vo
z;xs*(Rr(}8ky;{Ik&%J^WKRw`6?ATzcsgKi3M#X2UAQQ#{ofWB3sGT{^;D^p_}lTd
z(Uc$dQ)2U2QTZztAa@}f0{maMiew{Rzqtj-vb4y2$&w`?Sw0e;g_g~)y;<`|T2CXz
zYGBhF8{hAan6K<X$7VIei>32?e<1a_pOZ7J#lp`>EzZW1zs>EQkNYnrgbJ;Q+jLc~
zp|xY-_S`bES;huoE#t7Idtp~tSeWF`k`Ll3hX2Xn2Pv`lVS4&N26+&K#Hi!Q>E-g@
zRd*NN@%%06%Y-dmNnr#6&}{Q&pZ==HO9UjYK1cKsfS2J<sCk2E3UdYo*c2Z&ix<z{
zzYQG+NO9-Jr>peYwb=4SYRRI4Nnx*@dbHN6Py`Q6?AXkTG_}(G(RPzta>|E^THO>?
z425D~4SFmzI5Pp{n83kS^pQ9w{Gg=)ObaqEOuRcy<m@uKHt*2%1s`RtQMr;sa(3NT
zpZ<xNWUkBB)!&mzV;@bdU)=f8F^jlVfS;xPa5r81!QxzPZLR!Eb4?+%@1bYw4x@xe
zMbav4R(nVZ0jVUHbO%gF6PZ+QJ3pG}Z`QXwQ)ZbW&Q?7m9u>$U5X#%g4sQ@~4?Lc+
z-_?TU;3J{Fe86#;Y7Snt=ILmsg2yWs?3`MkJF)E0Yf@Lz(cB%c)mUbDkT1Hr&c@8l
zS0W51m=yYA(CfaRIo^C8bijn?NjVkm`MnEeC-H?b^6v9d-GM&%NC+0-D(s$X8lUA|
zRScIb2^SwxgZc17Sz(G1WDjZU)~y4a_8&exdNmzTlKPRW8CT6S*}T*ERiDpR{!C7q
zh+5rGqRNzB)U=sJlSYhqQdE>m>=Pz4Kc>mHn)U%XyN>0X>Y@MW*(939KYK=H>sdW&
z7<-sBzJp3LAb-EHn{K;4p`Q2&nX0`CMG8r(?#TSJ^QT(Hd$2&&d!Ev>yzqwnM!nI$
zN7|(StQvdmcZZKfQ(>StULY+3g&Mym;RWFv*HM+*1jPLw^Tp<q%Uag`$1<0XPZ@cu
zGjgQ4@?5cw7TadW_a|VY92Vd-067H=iY=;|f&N(t;;wuAbX#4S2&@3X*XU=-rfsZe
z%mSQ=*<|YEScsDD-nl~!BgEM!vP$SA-EG7s=R5oMspDq&K<sCbxyrY1H&Rt`Jj}{^
zCN|Cn|M9fN;47j$z+E1R6fy56pev{)#E`{+dHR8q^z|<T>e~N&D0`rpU0S8|qB+V!
z0lPSE>XK9Os821IgIP1{dw#48KSAobbOEab$talb`d2L+Gbo%^Fxr|5iw3jK`#HR&
z0IW(k8WY)pYE!XFJ2V}Lh@a(231Uw1uwl8<`eT4JPFSeW95*l!c*$MouQ0-~lpITU
z;#iV~c(h(^hWVtp?mkes=IPlSZo7*?1qJ(fSQ#XtC~V##Kg6zrP6Y)YQi;cdfK{1$
z!z-_+wrUA<2t9h{i418)jZpn?d{uE#0||r&87}B4qITM;4+x|zFz%c>6}tz!=DK2l
znZAAcjHhKJz<dUiQsQyx6ic-JCR4^<be6RT<$4U(`gW$oZz?8>fp*bwfasV@-T_7t
z8uo<-aq<KQzrw{AW~4!0gdy1)?Ss?y@=Fd8iSbqijGQ&=A`ES{g$oxRnqC8(Y25SA
zToX+{Py^{+rZJ^|jJcpomnAQ6<iLUHD+2pl9*7z{a9W$2zcMMa#4n55^ORI$e&q)`
zzxvAI@=HJKi$^GD;Uy=_rM<et!V>>sNgIV<9b?N@^0LnL|Lj*{GIzXTO~P6jdpH1x
z7mHXQ$+!@B84$xYrjqa-e&mMBr;*ZkNu0lJd|7-%9koE#{rgAySF2bJ!3y<RMVpo*
zRyq%7YPCyIjI(+kdbGK0fW7Z^vJhw`77CrachBQ!Gxq<@Y#Oz6G9`6fTqW<3>d&OR
zKwsjn7Iu-=l9SQ&y5cf+via@aLzZlW@sSCqQT+T*_?rLE6<^xg;>;`#eXvSz3HcuF
zWHRK*=Iq=d;NHhG@TSudUu|!1&$rtuAe<}=pYogR0)vmr_gHwv;q-qIMPb2;EBIeg
zM>TY8CU@UNm;p3}?c+p5CksAX3>%nmO0sv6aP+T%Nz7UdXX>pBe{pB@A4@!FpaT_&
zA;|PU9G{(k&7!3*ehu|aqHg%tMjAJ%fKPfU-%s&74OeMlIs*ff?#^LQD)kWKn2_3(
zR#n;E3hDahfnl7*jOo+K<!R}gc>ukT)1`2ROKxETASh0xRb+qwx52@|+BW@6N~$|4
z1Qd~S2#h9VZj_lOXXs~AICd(kNg~I{^@Xo;^IG}6;y*eg)p+HC0U*trkhz1sg@}!i
z$G{!J0##>U@Zn=2BYX;HW4}+dw=dW9(p1sZ{79^(;yC=*U$!K-C^$<hD{1{)2m7{4
z)DPEu*GDq+(&fvSxld`-lTNg$!9UrqfQf=|$*Ig%l+`}1HRIS=q;lL!&ui?I<)kH?
zRGe)B_=OZeZ&=2$E8tN=f;HNtAfGh0i3iDEccF9dqB6N;>;J&L$~Bzwe`j8WJHF+j
zQaN~Ntv5C_a8Xb1-kq+n)WKmikq?MQeehs1UPxNuT|mK$C~pMwMo?7p9XKf=gA^5c
z(&^#qg?cO}Rc2b4Y@5op9_x$N;WPB%YlKa()h=`Q%T?=BO=x;6C@2gK)v=SIPoBxP
zo}UVaW%;!=8|(g?d=Gq5^}OMxj9Lc}NtBoF1vaueqaAK~gOrG37~5SQ0|=Sor-47+
z^BMnq+$Ru{$dUpp?ejbJwt_UxGByqxwtTl$+!Yrco&>B06+>S3oEJi6OAlbRmR!x7
zK9UQ2p+2`)IluV&sb>|PrXV>nj!o}&EvAQ8RITuW6z$sgWZUnFfenqT60CrZ%A@p`
z=g#h3e(U?PVe)n<9vrN6M3TQ|YH=uuJ<)}fYq0nI)XeiwRXZ8U4gD>pRV(9i`e!6B
zP-SWDb>|MJ)q^VP>*&N0jY+C6yX>6YeI%=vw~n|h;BpHzXz5<KbthH~KA7hJx9G~k
zXl1?SV(6It2mC69qn-&3G;G_eV4$jF($%Dnb#8ftz(W~ByLN-XGutUCzYwa#`oit;
z7c!FNwBNx-;wf~OCb8$#${#7XZzFrjwQ)x8ymswcFcuT@nO9@5-yOYrD{k%53$?mS
z+KLxbPZyQ>>+==Bwq9J@xV(I8B_!qwJ`XBTnkV4bdA15bDZFhO;SDoNJKW~epTp7M
zzy)r2!P#N_Ox{d|bQLI?f^XDn|GS0Pl_to(;6(R6eHuv+nVtP0U|9NwPFeJ^Pw3kG
z5cF|of~aM;m&+8?eyrjGgM*z|q}W)3(_Z(UHz;kA&T%!xabnm<=Zyj@vc7^zjc5vA
zU9xs^cP{P(WHwHu<6ENd<mKgcnbyi&p~%bYPMj=$D=RxD$={*Ta^&Pm<*r>X0Va}4
zTAd#k9<!ZLkcziGJ$MBb@F7tYvNKLJmAC#LRrkvIQExkQt+<IB*bZPs^v8=+kG<ug
z@VdK7wHIHzK(nJ%yZc#1hD&%$(CVsJVq+lPixSl`Ts>TqEqLMF!Gi}0Eea~c?Kz_o
z*1p8bM_pI<q3dQhxzbu*qmse*Kh)Hi`0YTbrN*C^*iOFb$;mMrmo-mk4gr@HppP^D
z>BvZI#D_>{ZtWmvas^jR!9k*k*!=m`^J_MVALF^N9x)xxYJP9qT00BnmacypJ<U`#
zKa1GzQfPWd3R@kx|L|exSm%{SMn<$^3U}1M&fU?c%bToelP6DZzEKL<1!srrkS__l
zS7w?|Hh&C=Ktu?^rwAnWf?GbI7ztghX*4}GZR`Ff<zHBrXd1qL`<4O<&Krc=xjLyF
zzQ0tVG*#}zKTt3L#&K=K^w!U9@HwecxG5Yq2cU+NCr*4r1U++?5<FF+O5Q4cVBd{6
z*1GIO6%o%8@1TM$n`n=L`Ut$}HF!3phvH56J8}&EVKNOG%?t$*oRXxzVn6+M*6Dx8
zZDYzN&LUCZDiB|m5`gLSIm7UpJ2$`FMbT#&Z9@vROYB?o(@)AdO|-lN=pOIv2hp{L
zW!Lk@0Qm`CnXm%K0aZ{M3;F5$ckf7*h*d*HC*6Y7#mo)16xLTlQ_SY}^Yz7k&K$6h
zlx`bRLQanTya~;>y|(9oO*NPQ%Mc>7`?LQS8PNYhlJr0Q6)!bPq`=wg$1|{?(lJkG
zAFL0YeZ{lIqx>)cpPx}~xl4~|eEp}YA4euJBhvM=_4?!zu<s*BOiEFhE!iltjF`)W
zFI>+eMd#zj)~hZ$_Dg6|^8++_$wh_tI8mf;=7g2^AQkvjEimqBI_e`-3uR?d3Dd8(
zOvG?6vlprsub2wPgz#CY4)2k*uP?XC%myUU@XrFuIs2@F)JIOv9)`T!<h%q99~D7L
z)3%wph}jb0J1+@G^uX7aTS2)rmR;ak&KRKz0WPhe@fW#f$Z`Sgw*1^Ck4gs-kKE)t
zn<)ZdGI)Bn5Ol|m9ox{@xW~uG213b&X5rD!HpNpBRyoQ1*SC8HFjevcAWv@0S3%4V
z5rU66i<baJG_`JRp&c=Xp)@27$Re2n!f|a+x6s~P^iz0TWWZ|yo@lbSydThVzX;j4
zr{8`g1`BqM2WA81V)>0&*LUa=FU+}MiS8c*WSc*O{8)QNY@SikeJDFapI;`HKsjwH
zdQF!e-EwAX?4wC72tR(VPC@0cJFu8C{O(VP+>2Ch<G?!m+a;Vgqg%-HkRS!7dXdPC
zWUANd7&abrGqQ`|1<>+L<tXeaogHRi<K}sKQVMaV!R{;1HFEI7z!CRWET9$VBLPfo
zlTwL3d!YTsRxpb#(@-fy23BhNIr2uxU-7e{g<?Y~W)+9AV-h66FHh@%JBO*NN)_J1
z^e5O7{l5ub(CBp&sHVX5PkOe{uK{A~Aw|o7{(>?BgBG4XrDY0{1UN+?dLTd0y9L({
zS2f{)_cSjBJ$vrVl-O8kfK%}nY#`_`ucVySB)D=Zgog;m*1!=ep+F<IB@H{MB@|OW
zlHq~cg9hC`{hbuGA3aTqQ~_z+Z0scGHS$8O9u^{ItK0mQwkE*}?U;%;re`FaheVBQ
zsr+FbkJ2L6Pnjk<MgovuQd3*l2%SQJCAQaoSo%G8bpM?KtMZmmDp4G(tM#7##z#_>
z(s=%@2HT#T0;KKeC}B()8T*ZFB6s#qLGev?%X_>i>WPn!Pw(G#4yDnaO7Hk5yCuN)
z1!=U3A%%EN>P6D&uF^3~h|)jVWZg>c!t}+zv7&u05pmeY?N#sxd!Djm`};ZkWiT;U
z@($RN2kxtU)#>>r5+)U}KcUQ7^I<XYC)H!dihr8+pa7WUEqOGdX7)~NHQ~5Qm%F>7
zPedf2EB@Mw#)f&KBU}bQfB(VWd-v^I4MQse;V*f_#f-d9(!JamGi4N1Qd*j~zyBuC
z8&R$d0|;)q!*`o1V2im87~z)fAFOVKAzU`><IXDH2M%O$fYhw-)TvXPMcl~<v@PTU
zouvq{1>g2G?c+o2>QK+Y^pSc>VF-7df7iQx*Mm=PEZu|npqsSmXMIs_?ask*-vGX(
zd*#;~=pD}>4Vt42g`Od`a5k}C`|`<cb#oGaXn;SVE!T4WYIZMFl$)$3WOd=U_8mCT
zRSG?ahCx}870*V7*syw5zA7pzn$)7YMXInm^!h#_{4(5`^{_X(r5*)^5mVcVlS;UX
zAGZhsDC(DZVNDebs$;41e`JWG{`;-%L(+udB8h%O*t_Cd&&t<!GoKU<xcBAn^1csO
zZyR=K_>(0!GIUK>+Vu84s&;bbk?0}4mX`W-pVo7y)y}0y&D7`XoVF+!@J~m>bGnI-
zZ?7yq+0oBF;OBBftK!t`pSr(ydS2?W$UQHkX5+|5WtI2pw+xIAsTn&X&O{@)svB^H
zsI}9Q1`hlJrWB=h#DB58V0TXJvAwVVjLdCH)lWuZ+^%JJzwOGWaoVA7U3=2`-Z0=*
z(zZ*%ciG7P17RQ)K7RjW+pM<RiSA2oikf}sylvaYO~m$BVc58Fnx1`o_x@t4cfjWU
zbNUTmQdj)}eKRG%&8um^mOkCO6(`$9o3(A%?uHUZK64c2+u2oZ7<0R-d~1SL$cYo9
zUKjQ2)2C}{(GWSg4npndwQa__%5Q|2az*eaZ^@?BiI+T<EJ<7bg>~?%9YBwdgm(6q
z+`BHF#yI5I|2E<yLD6FR*0QgydTzM^J+@%S;`65l^y&ox!6*IgHPLP1;9H*Pjs+JM
zOBx=jm{0YP6K7S;Uk7i~LmGR}Adf<E-n_#SX>49*=fFd87O}x0Tb4pkXxB0*ZL5om
zYhVBI1$3JZO&{@O9r`6YFd?HivPQKIzbB3jk7erd)ytRfuh&XC7uJFe$g7G_9=+vI
zaU^Gt|2p)_&6`tp-F)Lw$K$7$!cN}0ao)_C5@~IHJ#%fYcdM8Ao!=-jH1Kb3{<&Rm
zZ7`IEm2aZ)`I;cn<)o@ggqhw&Dge#o<k!h5|NN7?d<0RLoJP7A^ed|^MBEAMF6fOd
zI_bX$BfM%IaV(aGfZ2!|SVK>5>#XaP2Lx7$bXCF&GOe)zVPOM$^icO6v#EIRtB|?M
z{rl@`YMykfJ^6&P#am)yV}s-|;QS$EKWw+z52|5#qklUY5CF}CfBIgu+1koVb6!-|
zqelW_fBGXcGZW^?>}Q7;PZOj&<}G{Q&&=F3spR3TEikj*Z(PWOcR#wv203WjD+WG~
zVW{`?>APRAgY*R#>2Pq#y0vTJ55&9*isK%ykDvYOu#ucxYX`vvcumu2jiaL|mU?8-
zc2iVT6im4%PnPQ|<1GuOSK_vL+qsjz0RfPrNfy@)4=nb*HUm6C3^Kuz8zebn1%G*B
zg2hBVy+_-|y`Fu8Mh#|wZFk*tO!!#$125BL+x1sp((y!X2Y&<3(nQ_Hd2{A;QdGpq
z@VHXO_C=HQyjQJR(^cy9;&S=CNiIC+Ni=T*#1(v<^p$#kv97tf`L<V+>257vclYx<
zk^rN~+(>fK??nEubrgwD-LA)PpEk`WYe!5ha2A^fsHcA)7UI_Hd9!cy>)m_SoH?8I
z=x#yFdVbwGO2&TtTs8ljw)bPlgu+Rd4L_weHX-#W?%WlOL4X3H$TPKj7|)-7)FReg
z%ziYAdcs)xpa-R#)Hj|nyX?y4%TyRXd-lYdX>yzT%6f~hExG#Ahi+Q#Ht!%`KR-2Z
zChAsyfA-sNzoEeuL{biDEVO*2P$7^}5$@};0P?qfaM?Q2R16AWYjRqE<s%8>TtU(F
za>Z~Y1L;`~$B2t|u3;WmpF6Rb@Kbu2MnsZ-Q&xT1Mu%)-D7^aZ5Z6gNNSNI|a+<rk
zxYUvuxNYBFGWTnZow<1)^?{G1aJ#OSmKO5CAw!~LVou)q%+&Pt!#D%I9#P&cr7~N*
z_!S7&VqL1X*oznED_q1k-rpDy`N9Pg1*(!m&&z9yGAb)rhRL4qm*^}1{P~k%6b{i`
z$jUZCY8{4Uo5Mv*?H*a6KsLI)4^7_|e$x3%gBfQtLI=i*zkmP!QqG6e)oa&|40<5l
z%UaR1r@x!<+=&(rpzO)Ge3n6{18jrRL{>!gyc61smJJW&Zo(#R#}i8xkq#3DR@Tsa
zWscvc4P(e~=7t$mmX(!#J;0YZCM2FeuRDCWY5cQCvuawN!Au7XBfF&CgZ}TFX>`Z7
zQ#Wj=cBy;*{5iC3&Q2cdrRRg?<XT3C?rbNwM5PzrdqzedKYemZJ)P$mbn+x9*Mw$7
z^2Mye#>U3Aqk{a+ChVc5J!bQ;CUP7tP0ce}>sGB|>BXNvU&y{W=h99t?bFisDBwFx
zv9z=#-xd6Wb+3B$>&NLrs6h|RPUJ>h%54<hR#mMZ+RJpof{D9Zw%uXoV|e%*fF@f<
zr%^BIyAsbwud!(*H*`$^7bsWr0%A5LFE)bi?n!f_f-1J~&_?_V%ygF%OPQbQR{!;1
s{-uK-@NXS_yMN)T|6l(~T8rE}O|QX&+;bUAA}2R%+FZl)(j|ZXUsvvw>;M1&

diff --git "a/an\303\241lisis-exploratorio.html" "b/an\303\241lisis-exploratorio.html"
index 5450b5f..dfd1b3b 100644
--- "a/an\303\241lisis-exploratorio.html"
+++ "b/an\303\241lisis-exploratorio.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
@@ -527,23 +527,23 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
 <span id="cb1-7"><a href="análisis-exploratorio.html#cb1-7" aria-hidden="true" tabindex="-1"></a>propinas <span class="ot">&lt;-</span> <span class="fu">read_csv</span>(<span class="st">&quot;./data/propinas.csv&quot;</span>)</span></code></pre></div>
 <p>Y vemos una muestra</p>
 <div class="sourceCode" id="cb2"><pre class="sourceCode r"><code class="sourceCode r"><span id="cb2-1"><a href="análisis-exploratorio.html#cb2-1" aria-hidden="true" tabindex="-1"></a><span class="fu">slice_sample</span>(propinas, <span class="at">n =</span> <span class="dv">10</span>) <span class="sc">|&gt;</span> <span class="fu">gt</span>()</span></code></pre></div>
-<div id="xddpgkkgug" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
-<style>#xddpgkkgug table {
+<div id="qcaxfnkqcl" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
+<style>#qcaxfnkqcl table {
   font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji';
   -webkit-font-smoothing: antialiased;
   -moz-osx-font-smoothing: grayscale;
 }
 
-#xddpgkkgug thead, #xddpgkkgug tbody, #xddpgkkgug tfoot, #xddpgkkgug tr, #xddpgkkgug td, #xddpgkkgug th {
+#qcaxfnkqcl thead, #qcaxfnkqcl tbody, #qcaxfnkqcl tfoot, #qcaxfnkqcl tr, #qcaxfnkqcl td, #qcaxfnkqcl th {
   border-style: none;
 }
 
-#xddpgkkgug p {
+#qcaxfnkqcl p {
   margin: 0;
   padding: 0;
 }
 
-#xddpgkkgug .gt_table {
+#qcaxfnkqcl .gt_table {
   display: table;
   border-collapse: collapse;
   line-height: normal;
@@ -569,12 +569,12 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-left-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_caption {
+#qcaxfnkqcl .gt_caption {
   padding-top: 4px;
   padding-bottom: 4px;
 }
 
-#xddpgkkgug .gt_title {
+#qcaxfnkqcl .gt_title {
   color: #333333;
   font-size: 125%;
   font-weight: initial;
@@ -586,7 +586,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-bottom-width: 0;
 }
 
-#xddpgkkgug .gt_subtitle {
+#qcaxfnkqcl .gt_subtitle {
   color: #333333;
   font-size: 85%;
   font-weight: initial;
@@ -598,7 +598,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-top-width: 0;
 }
 
-#xddpgkkgug .gt_heading {
+#qcaxfnkqcl .gt_heading {
   background-color: #FFFFFF;
   text-align: center;
   border-bottom-color: #FFFFFF;
@@ -610,13 +610,13 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-right-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_bottom_border {
+#qcaxfnkqcl .gt_bottom_border {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_col_headings {
+#qcaxfnkqcl .gt_col_headings {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -631,7 +631,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-right-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_col_heading {
+#qcaxfnkqcl .gt_col_heading {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -651,7 +651,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   overflow-x: hidden;
 }
 
-#xddpgkkgug .gt_column_spanner_outer {
+#qcaxfnkqcl .gt_column_spanner_outer {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -663,15 +663,15 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 4px;
 }
 
-#xddpgkkgug .gt_column_spanner_outer:first-child {
+#qcaxfnkqcl .gt_column_spanner_outer:first-child {
   padding-left: 0;
 }
 
-#xddpgkkgug .gt_column_spanner_outer:last-child {
+#qcaxfnkqcl .gt_column_spanner_outer:last-child {
   padding-right: 0;
 }
 
-#xddpgkkgug .gt_column_spanner {
+#qcaxfnkqcl .gt_column_spanner {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
@@ -683,11 +683,11 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   width: 100%;
 }
 
-#xddpgkkgug .gt_spanner_row {
+#qcaxfnkqcl .gt_spanner_row {
   border-bottom-style: hidden;
 }
 
-#xddpgkkgug .gt_group_heading {
+#qcaxfnkqcl .gt_group_heading {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -713,7 +713,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   text-align: left;
 }
 
-#xddpgkkgug .gt_empty_group_heading {
+#qcaxfnkqcl .gt_empty_group_heading {
   padding: 0.5px;
   color: #333333;
   background-color: #FFFFFF;
@@ -728,15 +728,15 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   vertical-align: middle;
 }
 
-#xddpgkkgug .gt_from_md > :first-child {
+#qcaxfnkqcl .gt_from_md > :first-child {
   margin-top: 0;
 }
 
-#xddpgkkgug .gt_from_md > :last-child {
+#qcaxfnkqcl .gt_from_md > :last-child {
   margin-bottom: 0;
 }
 
-#xddpgkkgug .gt_row {
+#qcaxfnkqcl .gt_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -755,7 +755,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   overflow-x: hidden;
 }
 
-#xddpgkkgug .gt_stub {
+#qcaxfnkqcl .gt_stub {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -768,7 +768,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 5px;
 }
 
-#xddpgkkgug .gt_stub_row_group {
+#qcaxfnkqcl .gt_stub_row_group {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -782,15 +782,15 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   vertical-align: top;
 }
 
-#xddpgkkgug .gt_row_group_first td {
+#qcaxfnkqcl .gt_row_group_first td {
   border-top-width: 2px;
 }
 
-#xddpgkkgug .gt_row_group_first th {
+#qcaxfnkqcl .gt_row_group_first th {
   border-top-width: 2px;
 }
 
-#xddpgkkgug .gt_summary_row {
+#qcaxfnkqcl .gt_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -800,16 +800,16 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 5px;
 }
 
-#xddpgkkgug .gt_first_summary_row {
+#qcaxfnkqcl .gt_first_summary_row {
   border-top-style: solid;
   border-top-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_first_summary_row.thick {
+#qcaxfnkqcl .gt_first_summary_row.thick {
   border-top-width: 2px;
 }
 
-#xddpgkkgug .gt_last_summary_row {
+#qcaxfnkqcl .gt_last_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -819,7 +819,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-bottom-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_grand_summary_row {
+#qcaxfnkqcl .gt_grand_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -829,7 +829,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 5px;
 }
 
-#xddpgkkgug .gt_first_grand_summary_row {
+#qcaxfnkqcl .gt_first_grand_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -839,7 +839,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-top-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_last_grand_summary_row_top {
+#qcaxfnkqcl .gt_last_grand_summary_row_top {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -849,11 +849,11 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-bottom-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_striped {
+#qcaxfnkqcl .gt_striped {
   background-color: rgba(128, 128, 128, 0.05);
 }
 
-#xddpgkkgug .gt_table_body {
+#qcaxfnkqcl .gt_table_body {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -862,7 +862,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-bottom-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_footnotes {
+#qcaxfnkqcl .gt_footnotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -876,7 +876,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-right-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_footnote {
+#qcaxfnkqcl .gt_footnote {
   margin: 0px;
   font-size: 90%;
   padding-top: 4px;
@@ -885,7 +885,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 5px;
 }
 
-#xddpgkkgug .gt_sourcenotes {
+#qcaxfnkqcl .gt_sourcenotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -899,7 +899,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-right-color: #D3D3D3;
 }
 
-#xddpgkkgug .gt_sourcenote {
+#qcaxfnkqcl .gt_sourcenote {
   font-size: 90%;
   padding-top: 4px;
   padding-bottom: 4px;
@@ -907,72 +907,72 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 5px;
 }
 
-#xddpgkkgug .gt_left {
+#qcaxfnkqcl .gt_left {
   text-align: left;
 }
 
-#xddpgkkgug .gt_center {
+#qcaxfnkqcl .gt_center {
   text-align: center;
 }
 
-#xddpgkkgug .gt_right {
+#qcaxfnkqcl .gt_right {
   text-align: right;
   font-variant-numeric: tabular-nums;
 }
 
-#xddpgkkgug .gt_font_normal {
+#qcaxfnkqcl .gt_font_normal {
   font-weight: normal;
 }
 
-#xddpgkkgug .gt_font_bold {
+#qcaxfnkqcl .gt_font_bold {
   font-weight: bold;
 }
 
-#xddpgkkgug .gt_font_italic {
+#qcaxfnkqcl .gt_font_italic {
   font-style: italic;
 }
 
-#xddpgkkgug .gt_super {
+#qcaxfnkqcl .gt_super {
   font-size: 65%;
 }
 
-#xddpgkkgug .gt_footnote_marks {
+#qcaxfnkqcl .gt_footnote_marks {
   font-size: 75%;
   vertical-align: 0.4em;
   position: initial;
 }
 
-#xddpgkkgug .gt_asterisk {
+#qcaxfnkqcl .gt_asterisk {
   font-size: 100%;
   vertical-align: 0;
 }
 
-#xddpgkkgug .gt_indent_1 {
+#qcaxfnkqcl .gt_indent_1 {
   text-indent: 5px;
 }
 
-#xddpgkkgug .gt_indent_2 {
+#qcaxfnkqcl .gt_indent_2 {
   text-indent: 10px;
 }
 
-#xddpgkkgug .gt_indent_3 {
+#qcaxfnkqcl .gt_indent_3 {
   text-indent: 15px;
 }
 
-#xddpgkkgug .gt_indent_4 {
+#qcaxfnkqcl .gt_indent_4 {
   text-indent: 20px;
 }
 
-#xddpgkkgug .gt_indent_5 {
+#qcaxfnkqcl .gt_indent_5 {
   text-indent: 25px;
 }
 
-#xddpgkkgug .katex-display {
+#qcaxfnkqcl .katex-display {
   display: inline-flex !important;
   margin-bottom: 0.75em !important;
 }
 
-#xddpgkkgug div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
+#qcaxfnkqcl div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
   height: 0px !important;
 }
 </style>
@@ -988,66 +988,66 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
     </tr>
   </thead>
   <tbody class="gt_table_body">
-    <tr><td headers="cuenta_total" class="gt_row gt_right">9.78</td>
-<td headers="propina" class="gt_row gt_right">1.73</td>
-<td headers="fumador" class="gt_row gt_left">No</td>
-<td headers="dia" class="gt_row gt_left">Jue</td>
-<td headers="momento" class="gt_row gt_left">Comida</td>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">15.81</td>
+<td headers="propina" class="gt_row gt_right">3.16</td>
+<td headers="fumador" class="gt_row gt_left">Si</td>
+<td headers="dia" class="gt_row gt_left">Sab</td>
+<td headers="momento" class="gt_row gt_left">Cena</td>
 <td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">13.42</td>
-<td headers="propina" class="gt_row gt_right">1.68</td>
-<td headers="fumador" class="gt_row gt_left">No</td>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">16.47</td>
+<td headers="propina" class="gt_row gt_right">3.23</td>
+<td headers="fumador" class="gt_row gt_left">Si</td>
 <td headers="dia" class="gt_row gt_left">Jue</td>
 <td headers="momento" class="gt_row gt_left">Comida</td>
-<td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">11.02</td>
-<td headers="propina" class="gt_row gt_right">1.98</td>
+<td headers="num_personas" class="gt_row gt_right">3</td></tr>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">38.01</td>
+<td headers="propina" class="gt_row gt_right">3.00</td>
 <td headers="fumador" class="gt_row gt_left">Si</td>
 <td headers="dia" class="gt_row gt_left">Sab</td>
 <td headers="momento" class="gt_row gt_left">Cena</td>
-<td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">22.49</td>
-<td headers="propina" class="gt_row gt_right">3.50</td>
-<td headers="fumador" class="gt_row gt_left">No</td>
-<td headers="dia" class="gt_row gt_left">Vie</td>
-<td headers="momento" class="gt_row gt_left">Cena</td>
-<td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">11.59</td>
-<td headers="propina" class="gt_row gt_right">1.50</td>
+<td headers="num_personas" class="gt_row gt_right">4</td></tr>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">3.07</td>
+<td headers="propina" class="gt_row gt_right">1.00</td>
 <td headers="fumador" class="gt_row gt_left">Si</td>
 <td headers="dia" class="gt_row gt_left">Sab</td>
 <td headers="momento" class="gt_row gt_left">Cena</td>
+<td headers="num_personas" class="gt_row gt_right">1</td></tr>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">12.46</td>
+<td headers="propina" class="gt_row gt_right">1.50</td>
+<td headers="fumador" class="gt_row gt_left">No</td>
+<td headers="dia" class="gt_row gt_left">Vie</td>
+<td headers="momento" class="gt_row gt_left">Cena</td>
 <td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">24.71</td>
-<td headers="propina" class="gt_row gt_right">5.85</td>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">8.52</td>
+<td headers="propina" class="gt_row gt_right">1.48</td>
 <td headers="fumador" class="gt_row gt_left">No</td>
 <td headers="dia" class="gt_row gt_left">Jue</td>
 <td headers="momento" class="gt_row gt_left">Comida</td>
 <td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">9.60</td>
-<td headers="propina" class="gt_row gt_right">4.00</td>
-<td headers="fumador" class="gt_row gt_left">Si</td>
-<td headers="dia" class="gt_row gt_left">Dom</td>
-<td headers="momento" class="gt_row gt_left">Cena</td>
-<td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">19.65</td>
-<td headers="propina" class="gt_row gt_right">3.00</td>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">11.61</td>
+<td headers="propina" class="gt_row gt_right">3.39</td>
 <td headers="fumador" class="gt_row gt_left">No</td>
 <td headers="dia" class="gt_row gt_left">Sab</td>
 <td headers="momento" class="gt_row gt_left">Cena</td>
 <td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">18.04</td>
-<td headers="propina" class="gt_row gt_right">3.00</td>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">50.81</td>
+<td headers="propina" class="gt_row gt_right">10.00</td>
+<td headers="fumador" class="gt_row gt_left">Si</td>
+<td headers="dia" class="gt_row gt_left">Sab</td>
+<td headers="momento" class="gt_row gt_left">Cena</td>
+<td headers="num_personas" class="gt_row gt_right">3</td></tr>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">38.07</td>
+<td headers="propina" class="gt_row gt_right">4.00</td>
 <td headers="fumador" class="gt_row gt_left">No</td>
 <td headers="dia" class="gt_row gt_left">Dom</td>
 <td headers="momento" class="gt_row gt_left">Cena</td>
-<td headers="num_personas" class="gt_row gt_right">2</td></tr>
-    <tr><td headers="cuenta_total" class="gt_row gt_right">30.40</td>
-<td headers="propina" class="gt_row gt_right">5.60</td>
+<td headers="num_personas" class="gt_row gt_right">3</td></tr>
+    <tr><td headers="cuenta_total" class="gt_row gt_right">15.06</td>
+<td headers="propina" class="gt_row gt_right">3.00</td>
 <td headers="fumador" class="gt_row gt_left">No</td>
-<td headers="dia" class="gt_row gt_left">Dom</td>
+<td headers="dia" class="gt_row gt_left">Sab</td>
 <td headers="momento" class="gt_row gt_left">Cena</td>
-<td headers="num_personas" class="gt_row gt_right">4</td></tr>
+<td headers="num_personas" class="gt_row gt_right">2</td></tr>
   </tbody>
   
   
@@ -1086,23 +1086,23 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
 <span id="cb3-6"><a href="análisis-exploratorio.html#cb3-6" aria-hidden="true" tabindex="-1"></a>  <span class="fu">arrange</span>(f)</span>
 <span id="cb3-7"><a href="análisis-exploratorio.html#cb3-7" aria-hidden="true" tabindex="-1"></a><span class="fu">bind_rows</span>(<span class="fu">head</span>(cuenta), <span class="fu">tail</span>(cuenta)) <span class="sc">|&gt;</span> </span>
 <span id="cb3-8"><a href="análisis-exploratorio.html#cb3-8" aria-hidden="true" tabindex="-1"></a>  <span class="fu">gt</span>() <span class="sc">|&gt;</span> <span class="fu">fmt_number</span>(<span class="at">columns =</span> f, <span class="at">decimals =</span> <span class="dv">3</span>)</span></code></pre></div>
-<div id="zwocxfeiud" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
-<style>#zwocxfeiud table {
+<div id="phufidpmfh" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
+<style>#phufidpmfh table {
   font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji';
   -webkit-font-smoothing: antialiased;
   -moz-osx-font-smoothing: grayscale;
 }
 
-#zwocxfeiud thead, #zwocxfeiud tbody, #zwocxfeiud tfoot, #zwocxfeiud tr, #zwocxfeiud td, #zwocxfeiud th {
+#phufidpmfh thead, #phufidpmfh tbody, #phufidpmfh tfoot, #phufidpmfh tr, #phufidpmfh td, #phufidpmfh th {
   border-style: none;
 }
 
-#zwocxfeiud p {
+#phufidpmfh p {
   margin: 0;
   padding: 0;
 }
 
-#zwocxfeiud .gt_table {
+#phufidpmfh .gt_table {
   display: table;
   border-collapse: collapse;
   line-height: normal;
@@ -1128,12 +1128,12 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-left-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_caption {
+#phufidpmfh .gt_caption {
   padding-top: 4px;
   padding-bottom: 4px;
 }
 
-#zwocxfeiud .gt_title {
+#phufidpmfh .gt_title {
   color: #333333;
   font-size: 125%;
   font-weight: initial;
@@ -1145,7 +1145,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-bottom-width: 0;
 }
 
-#zwocxfeiud .gt_subtitle {
+#phufidpmfh .gt_subtitle {
   color: #333333;
   font-size: 85%;
   font-weight: initial;
@@ -1157,7 +1157,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-top-width: 0;
 }
 
-#zwocxfeiud .gt_heading {
+#phufidpmfh .gt_heading {
   background-color: #FFFFFF;
   text-align: center;
   border-bottom-color: #FFFFFF;
@@ -1169,13 +1169,13 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-right-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_bottom_border {
+#phufidpmfh .gt_bottom_border {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_col_headings {
+#phufidpmfh .gt_col_headings {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -1190,7 +1190,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-right-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_col_heading {
+#phufidpmfh .gt_col_heading {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -1210,7 +1210,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   overflow-x: hidden;
 }
 
-#zwocxfeiud .gt_column_spanner_outer {
+#phufidpmfh .gt_column_spanner_outer {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -1222,15 +1222,15 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 4px;
 }
 
-#zwocxfeiud .gt_column_spanner_outer:first-child {
+#phufidpmfh .gt_column_spanner_outer:first-child {
   padding-left: 0;
 }
 
-#zwocxfeiud .gt_column_spanner_outer:last-child {
+#phufidpmfh .gt_column_spanner_outer:last-child {
   padding-right: 0;
 }
 
-#zwocxfeiud .gt_column_spanner {
+#phufidpmfh .gt_column_spanner {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
@@ -1242,11 +1242,11 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   width: 100%;
 }
 
-#zwocxfeiud .gt_spanner_row {
+#phufidpmfh .gt_spanner_row {
   border-bottom-style: hidden;
 }
 
-#zwocxfeiud .gt_group_heading {
+#phufidpmfh .gt_group_heading {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -1272,7 +1272,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   text-align: left;
 }
 
-#zwocxfeiud .gt_empty_group_heading {
+#phufidpmfh .gt_empty_group_heading {
   padding: 0.5px;
   color: #333333;
   background-color: #FFFFFF;
@@ -1287,15 +1287,15 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   vertical-align: middle;
 }
 
-#zwocxfeiud .gt_from_md > :first-child {
+#phufidpmfh .gt_from_md > :first-child {
   margin-top: 0;
 }
 
-#zwocxfeiud .gt_from_md > :last-child {
+#phufidpmfh .gt_from_md > :last-child {
   margin-bottom: 0;
 }
 
-#zwocxfeiud .gt_row {
+#phufidpmfh .gt_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -1314,7 +1314,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   overflow-x: hidden;
 }
 
-#zwocxfeiud .gt_stub {
+#phufidpmfh .gt_stub {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -1327,7 +1327,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 5px;
 }
 
-#zwocxfeiud .gt_stub_row_group {
+#phufidpmfh .gt_stub_row_group {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -1341,15 +1341,15 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   vertical-align: top;
 }
 
-#zwocxfeiud .gt_row_group_first td {
+#phufidpmfh .gt_row_group_first td {
   border-top-width: 2px;
 }
 
-#zwocxfeiud .gt_row_group_first th {
+#phufidpmfh .gt_row_group_first th {
   border-top-width: 2px;
 }
 
-#zwocxfeiud .gt_summary_row {
+#phufidpmfh .gt_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -1359,16 +1359,16 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 5px;
 }
 
-#zwocxfeiud .gt_first_summary_row {
+#phufidpmfh .gt_first_summary_row {
   border-top-style: solid;
   border-top-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_first_summary_row.thick {
+#phufidpmfh .gt_first_summary_row.thick {
   border-top-width: 2px;
 }
 
-#zwocxfeiud .gt_last_summary_row {
+#phufidpmfh .gt_last_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -1378,7 +1378,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-bottom-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_grand_summary_row {
+#phufidpmfh .gt_grand_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -1388,7 +1388,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 5px;
 }
 
-#zwocxfeiud .gt_first_grand_summary_row {
+#phufidpmfh .gt_first_grand_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -1398,7 +1398,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-top-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_last_grand_summary_row_top {
+#phufidpmfh .gt_last_grand_summary_row_top {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -1408,11 +1408,11 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-bottom-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_striped {
+#phufidpmfh .gt_striped {
   background-color: rgba(128, 128, 128, 0.05);
 }
 
-#zwocxfeiud .gt_table_body {
+#phufidpmfh .gt_table_body {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -1421,7 +1421,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-bottom-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_footnotes {
+#phufidpmfh .gt_footnotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -1435,7 +1435,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-right-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_footnote {
+#phufidpmfh .gt_footnote {
   margin: 0px;
   font-size: 90%;
   padding-top: 4px;
@@ -1444,7 +1444,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 5px;
 }
 
-#zwocxfeiud .gt_sourcenotes {
+#phufidpmfh .gt_sourcenotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -1458,7 +1458,7 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   border-right-color: #D3D3D3;
 }
 
-#zwocxfeiud .gt_sourcenote {
+#phufidpmfh .gt_sourcenote {
   font-size: 90%;
   padding-top: 4px;
   padding-bottom: 4px;
@@ -1466,72 +1466,72 @@ <h2>Algunos conceptos básicos<a href="análisis-exploratorio.html#algunos-conce
   padding-right: 5px;
 }
 
-#zwocxfeiud .gt_left {
+#phufidpmfh .gt_left {
   text-align: left;
 }
 
-#zwocxfeiud .gt_center {
+#phufidpmfh .gt_center {
   text-align: center;
 }
 
-#zwocxfeiud .gt_right {
+#phufidpmfh .gt_right {
   text-align: right;
   font-variant-numeric: tabular-nums;
 }
 
-#zwocxfeiud .gt_font_normal {
+#phufidpmfh .gt_font_normal {
   font-weight: normal;
 }
 
-#zwocxfeiud .gt_font_bold {
+#phufidpmfh .gt_font_bold {
   font-weight: bold;
 }
 
-#zwocxfeiud .gt_font_italic {
+#phufidpmfh .gt_font_italic {
   font-style: italic;
 }
 
-#zwocxfeiud .gt_super {
+#phufidpmfh .gt_super {
   font-size: 65%;
 }
 
-#zwocxfeiud .gt_footnote_marks {
+#phufidpmfh .gt_footnote_marks {
   font-size: 75%;
   vertical-align: 0.4em;
   position: initial;
 }
 
-#zwocxfeiud .gt_asterisk {
+#phufidpmfh .gt_asterisk {
   font-size: 100%;
   vertical-align: 0;
 }
 
-#zwocxfeiud .gt_indent_1 {
+#phufidpmfh .gt_indent_1 {
   text-indent: 5px;
 }
 
-#zwocxfeiud .gt_indent_2 {
+#phufidpmfh .gt_indent_2 {
   text-indent: 10px;
 }
 
-#zwocxfeiud .gt_indent_3 {
+#phufidpmfh .gt_indent_3 {
   text-indent: 15px;
 }
 
-#zwocxfeiud .gt_indent_4 {
+#phufidpmfh .gt_indent_4 {
   text-indent: 20px;
 }
 
-#zwocxfeiud .gt_indent_5 {
+#phufidpmfh .gt_indent_5 {
   text-indent: 25px;
 }
 
-#zwocxfeiud .katex-display {
+#phufidpmfh .katex-display {
   display: inline-flex !important;
   margin-bottom: 0.75em !important;
 }
 
-#zwocxfeiud div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
+#phufidpmfh div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
   height: 0px !important;
 }
 </style>
diff --git "a/ap\303\251ndice-principios-de-visualizacion.html" "b/ap\303\251ndice-principios-de-visualizacion.html"
index d6ab5c8..a9e02a3 100644
--- "a/ap\303\251ndice-principios-de-visualizacion.html"
+++ "b/ap\303\251ndice-principios-de-visualizacion.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git "a/ap\303\251ndice-transformaciones.html" "b/ap\303\251ndice-transformaciones.html"
index 75cc578..968d952 100644
--- "a/ap\303\251ndice-transformaciones.html"
+++ "b/ap\303\251ndice-transformaciones.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
@@ -553,23 +553,23 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
 <span id="cb672-7"><a href="apéndice-transformaciones.html#cb672-7" aria-hidden="true" tabindex="-1"></a>  <span class="fu">arrange</span>(body) <span class="sc">|&gt;</span> </span>
 <span id="cb672-8"><a href="apéndice-transformaciones.html#cb672-8" aria-hidden="true" tabindex="-1"></a>  gt<span class="sc">::</span><span class="fu">gt</span>() <span class="sc">|&gt;</span> </span>
 <span id="cb672-9"><a href="apéndice-transformaciones.html#cb672-9" aria-hidden="true" tabindex="-1"></a>  gt<span class="sc">::</span><span class="fu">fmt_number</span>()</span></code></pre></div>
-<div id="ufmafyhycy" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
-<style>#ufmafyhycy table {
+<div id="ociqvanemt" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;">
+<style>#ociqvanemt table {
   font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji';
   -webkit-font-smoothing: antialiased;
   -moz-osx-font-smoothing: grayscale;
 }
 
-#ufmafyhycy thead, #ufmafyhycy tbody, #ufmafyhycy tfoot, #ufmafyhycy tr, #ufmafyhycy td, #ufmafyhycy th {
+#ociqvanemt thead, #ociqvanemt tbody, #ociqvanemt tfoot, #ociqvanemt tr, #ociqvanemt td, #ociqvanemt th {
   border-style: none;
 }
 
-#ufmafyhycy p {
+#ociqvanemt p {
   margin: 0;
   padding: 0;
 }
 
-#ufmafyhycy .gt_table {
+#ociqvanemt .gt_table {
   display: table;
   border-collapse: collapse;
   line-height: normal;
@@ -595,12 +595,12 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-left-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_caption {
+#ociqvanemt .gt_caption {
   padding-top: 4px;
   padding-bottom: 4px;
 }
 
-#ufmafyhycy .gt_title {
+#ociqvanemt .gt_title {
   color: #333333;
   font-size: 125%;
   font-weight: initial;
@@ -612,7 +612,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-bottom-width: 0;
 }
 
-#ufmafyhycy .gt_subtitle {
+#ociqvanemt .gt_subtitle {
   color: #333333;
   font-size: 85%;
   font-weight: initial;
@@ -624,7 +624,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-top-width: 0;
 }
 
-#ufmafyhycy .gt_heading {
+#ociqvanemt .gt_heading {
   background-color: #FFFFFF;
   text-align: center;
   border-bottom-color: #FFFFFF;
@@ -636,13 +636,13 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-right-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_bottom_border {
+#ociqvanemt .gt_bottom_border {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_col_headings {
+#ociqvanemt .gt_col_headings {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -657,7 +657,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-right-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_col_heading {
+#ociqvanemt .gt_col_heading {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -677,7 +677,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   overflow-x: hidden;
 }
 
-#ufmafyhycy .gt_column_spanner_outer {
+#ociqvanemt .gt_column_spanner_outer {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -689,15 +689,15 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   padding-right: 4px;
 }
 
-#ufmafyhycy .gt_column_spanner_outer:first-child {
+#ociqvanemt .gt_column_spanner_outer:first-child {
   padding-left: 0;
 }
 
-#ufmafyhycy .gt_column_spanner_outer:last-child {
+#ociqvanemt .gt_column_spanner_outer:last-child {
   padding-right: 0;
 }
 
-#ufmafyhycy .gt_column_spanner {
+#ociqvanemt .gt_column_spanner {
   border-bottom-style: solid;
   border-bottom-width: 2px;
   border-bottom-color: #D3D3D3;
@@ -709,11 +709,11 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   width: 100%;
 }
 
-#ufmafyhycy .gt_spanner_row {
+#ociqvanemt .gt_spanner_row {
   border-bottom-style: hidden;
 }
 
-#ufmafyhycy .gt_group_heading {
+#ociqvanemt .gt_group_heading {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -739,7 +739,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   text-align: left;
 }
 
-#ufmafyhycy .gt_empty_group_heading {
+#ociqvanemt .gt_empty_group_heading {
   padding: 0.5px;
   color: #333333;
   background-color: #FFFFFF;
@@ -754,15 +754,15 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   vertical-align: middle;
 }
 
-#ufmafyhycy .gt_from_md > :first-child {
+#ociqvanemt .gt_from_md > :first-child {
   margin-top: 0;
 }
 
-#ufmafyhycy .gt_from_md > :last-child {
+#ociqvanemt .gt_from_md > :last-child {
   margin-bottom: 0;
 }
 
-#ufmafyhycy .gt_row {
+#ociqvanemt .gt_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -781,7 +781,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   overflow-x: hidden;
 }
 
-#ufmafyhycy .gt_stub {
+#ociqvanemt .gt_stub {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -794,7 +794,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   padding-right: 5px;
 }
 
-#ufmafyhycy .gt_stub_row_group {
+#ociqvanemt .gt_stub_row_group {
   color: #333333;
   background-color: #FFFFFF;
   font-size: 100%;
@@ -808,15 +808,15 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   vertical-align: top;
 }
 
-#ufmafyhycy .gt_row_group_first td {
+#ociqvanemt .gt_row_group_first td {
   border-top-width: 2px;
 }
 
-#ufmafyhycy .gt_row_group_first th {
+#ociqvanemt .gt_row_group_first th {
   border-top-width: 2px;
 }
 
-#ufmafyhycy .gt_summary_row {
+#ociqvanemt .gt_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -826,16 +826,16 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   padding-right: 5px;
 }
 
-#ufmafyhycy .gt_first_summary_row {
+#ociqvanemt .gt_first_summary_row {
   border-top-style: solid;
   border-top-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_first_summary_row.thick {
+#ociqvanemt .gt_first_summary_row.thick {
   border-top-width: 2px;
 }
 
-#ufmafyhycy .gt_last_summary_row {
+#ociqvanemt .gt_last_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -845,7 +845,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-bottom-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_grand_summary_row {
+#ociqvanemt .gt_grand_summary_row {
   color: #333333;
   background-color: #FFFFFF;
   text-transform: inherit;
@@ -855,7 +855,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   padding-right: 5px;
 }
 
-#ufmafyhycy .gt_first_grand_summary_row {
+#ociqvanemt .gt_first_grand_summary_row {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -865,7 +865,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-top-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_last_grand_summary_row_top {
+#ociqvanemt .gt_last_grand_summary_row_top {
   padding-top: 8px;
   padding-bottom: 8px;
   padding-left: 5px;
@@ -875,11 +875,11 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-bottom-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_striped {
+#ociqvanemt .gt_striped {
   background-color: rgba(128, 128, 128, 0.05);
 }
 
-#ufmafyhycy .gt_table_body {
+#ociqvanemt .gt_table_body {
   border-top-style: solid;
   border-top-width: 2px;
   border-top-color: #D3D3D3;
@@ -888,7 +888,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-bottom-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_footnotes {
+#ociqvanemt .gt_footnotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -902,7 +902,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-right-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_footnote {
+#ociqvanemt .gt_footnote {
   margin: 0px;
   font-size: 90%;
   padding-top: 4px;
@@ -911,7 +911,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   padding-right: 5px;
 }
 
-#ufmafyhycy .gt_sourcenotes {
+#ociqvanemt .gt_sourcenotes {
   color: #333333;
   background-color: #FFFFFF;
   border-bottom-style: none;
@@ -925,7 +925,7 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   border-right-color: #D3D3D3;
 }
 
-#ufmafyhycy .gt_sourcenote {
+#ociqvanemt .gt_sourcenote {
   font-size: 90%;
   padding-top: 4px;
   padding-bottom: 4px;
@@ -933,72 +933,72 @@ <h1>Apéndice: Transformaciones<a href="apéndice-transformaciones.html#apéndic
   padding-right: 5px;
 }
 
-#ufmafyhycy .gt_left {
+#ociqvanemt .gt_left {
   text-align: left;
 }
 
-#ufmafyhycy .gt_center {
+#ociqvanemt .gt_center {
   text-align: center;
 }
 
-#ufmafyhycy .gt_right {
+#ociqvanemt .gt_right {
   text-align: right;
   font-variant-numeric: tabular-nums;
 }
 
-#ufmafyhycy .gt_font_normal {
+#ociqvanemt .gt_font_normal {
   font-weight: normal;
 }
 
-#ufmafyhycy .gt_font_bold {
+#ociqvanemt .gt_font_bold {
   font-weight: bold;
 }
 
-#ufmafyhycy .gt_font_italic {
+#ociqvanemt .gt_font_italic {
   font-style: italic;
 }
 
-#ufmafyhycy .gt_super {
+#ociqvanemt .gt_super {
   font-size: 65%;
 }
 
-#ufmafyhycy .gt_footnote_marks {
+#ociqvanemt .gt_footnote_marks {
   font-size: 75%;
   vertical-align: 0.4em;
   position: initial;
 }
 
-#ufmafyhycy .gt_asterisk {
+#ociqvanemt .gt_asterisk {
   font-size: 100%;
   vertical-align: 0;
 }
 
-#ufmafyhycy .gt_indent_1 {
+#ociqvanemt .gt_indent_1 {
   text-indent: 5px;
 }
 
-#ufmafyhycy .gt_indent_2 {
+#ociqvanemt .gt_indent_2 {
   text-indent: 10px;
 }
 
-#ufmafyhycy .gt_indent_3 {
+#ociqvanemt .gt_indent_3 {
   text-indent: 15px;
 }
 
-#ufmafyhycy .gt_indent_4 {
+#ociqvanemt .gt_indent_4 {
   text-indent: 20px;
 }
 
-#ufmafyhycy .gt_indent_5 {
+#ociqvanemt .gt_indent_5 {
   text-indent: 25px;
 }
 
-#ufmafyhycy .katex-display {
+#ociqvanemt .katex-display {
   display: inline-flex !important;
   margin-bottom: 0.75em !important;
 }
 
-#ufmafyhycy div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
+#ociqvanemt div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after {
   height: 0px !important;
 }
 </style>
diff --git "a/bootstrap-param\303\251trico.html" "b/bootstrap-param\303\251trico.html"
index d2a5a6a..c8950f3 100644
--- "a/bootstrap-param\303\251trico.html"
+++ "b/bootstrap-param\303\251trico.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git "a/calibraci\303\263n-bayesiana-y-regularizaci\303\263n.html" "b/calibraci\303\263n-bayesiana-y-regularizaci\303\263n.html"
index b206869..c2985cc 100644
--- "a/calibraci\303\263n-bayesiana-y-regularizaci\303\263n.html"
+++ "b/calibraci\303\263n-bayesiana-y-regularizaci\303\263n.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
@@ -966,7 +966,8 @@ <h2>Ejemplo: modelo normal y estaturas<a href="calibración-bayesiana-y-regulari
 </div>
 <div id="ejemplo-estimación-de-proporciones" class="section level2 unnumbered hasAnchor">
 <h2>Ejemplo: estimación de proporciones<a href="calibración-bayesiana-y-regularización.html#ejemplo-estimación-de-proporciones" class="anchor-section" aria-label="Anchor link to header"></a></h2>
-<p>Ahora repetimos el ejercicio</p>
+<p>Ahora repetimos el ejercicio de la estimación de la proporción de hogares con ingresos
+superiores a 150 mil.</p>
 <div class="sourceCode" id="cb551"><pre class="sourceCode r"><code class="sourceCode r"><span id="cb551-1"><a href="calibración-bayesiana-y-regularización.html#cb551-1" aria-hidden="true" tabindex="-1"></a><span class="co"># inicial</span></span>
 <span id="cb551-2"><a href="calibración-bayesiana-y-regularización.html#cb551-2" aria-hidden="true" tabindex="-1"></a>a <span class="ot">&lt;-</span> <span class="dv">2</span></span>
 <span id="cb551-3"><a href="calibración-bayesiana-y-regularización.html#cb551-3" aria-hidden="true" tabindex="-1"></a>b <span class="ot">&lt;-</span> <span class="dv">100</span></span>
diff --git "a/estimaci\303\263n-por-m\303\241xima-verosimilitud.html" "b/estimaci\303\263n-por-m\303\241xima-verosimilitud.html"
index 55af588..5a24cff 100644
--- "a/estimaci\303\263n-por-m\303\241xima-verosimilitud.html"
+++ "b/estimaci\303\263n-por-m\303\241xima-verosimilitud.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git "a/estimaci\303\263n-y-distribuci\303\263n-de-muestreo-1.html" "b/estimaci\303\263n-y-distribuci\303\263n-de-muestreo-1.html"
index b050cba..815d6ff 100644
--- "a/estimaci\303\263n-y-distribuci\303\263n-de-muestreo-1.html"
+++ "b/estimaci\303\263n-y-distribuci\303\263n-de-muestreo-1.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git a/index.html b/index.html
index 4190bc7..552cf9c 100644
--- a/index.html
+++ b/index.html
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git a/intervalos-de-confianza-y-remuestreo.html b/intervalos-de-confianza-y-remuestreo.html
index 9a12855..92956e3 100644
--- a/intervalos-de-confianza-y-remuestreo.html
+++ b/intervalos-de-confianza-y-remuestreo.html
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git "a/introducci\303\263n-a-inferencia-bayesiana-1.html" "b/introducci\303\263n-a-inferencia-bayesiana-1.html"
index d623e98..a200d1d 100644
--- "a/introducci\303\263n-a-inferencia-bayesiana-1.html"
+++ "b/introducci\303\263n-a-inferencia-bayesiana-1.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
@@ -780,7 +780,7 @@ <h2>Ejemplo: estimando una proporción<a href="introducción-a-inferencia-bayesi
 que <span class="math inline">\(\theta\)</span> está en el intervalo</p>
 <div class="sourceCode" id="cb447"><pre class="sourceCode r"><code class="sourceCode r"><span id="cb447-1"><a href="introducción-a-inferencia-bayesiana-1.html#cb447-1" aria-hidden="true" tabindex="-1"></a><span class="fu">quantile</span>(sim_inicial<span class="sc">$</span>theta, <span class="fu">c</span>(<span class="fl">0.025</span>, <span class="fl">0.975</span>)) <span class="sc">|&gt;</span> <span class="fu">round</span>(<span class="dv">2</span>)</span></code></pre></div>
 <pre><code>##  2.5% 97.5% 
-##  0.14  0.85</code></pre>
+##  0.15  0.85</code></pre>
 <p>Es difícil justificar en abstracto por qué escogeriamos una inicial con esta
 forma. <em>Aunque esto los detallaremos más adelante</em>, puedes pensar, por el
 momento, que alguien observó algunos casos de esta población, y quizá vio tres éxitos y tres fracasos. Esto sugeriría que es poco probable que la probablidad
@@ -815,8 +815,8 @@ <h2>Ejemplo: estimando una proporción<a href="introducción-a-inferencia-bayesi
 <pre><code>## # A tibble: 2 × 2
 ##   dist      theta_hat
 ##   &lt;chr&gt;         &lt;dbl&gt;
-## 1 inicial       0.503
-## 2 posterior     0.61</code></pre>
+## 1 inicial        0.5 
+## 2 posterior      0.61</code></pre>
 <p>Nota que el estimador de máxima verosimilitud es <span class="math inline">\(\hat{p} = 19/30 = 0.63\)</span>, que
 es ligeramente diferente de la media posterior. ¿Por qué?</p>
 <p>Y podemos construir intervalos de percentiles, que en esta situación
@@ -829,7 +829,7 @@ <h2>Ejemplo: estimando una proporción<a href="introducción-a-inferencia-bayesi
 ## # Groups:   dist [2]
 ##   dist      `0.025` `0.975`
 ##   &lt;chr&gt;       &lt;dbl&gt;   &lt;dbl&gt;
-## 1 inicial      0.14    0.85
+## 1 inicial      0.15    0.85
 ## 2 posterior    0.45    0.76</code></pre>
 <p>El segundo renglón nos da un intervalo posterior para <span class="math inline">\(\theta\)</span> de <em>credibilidad</em>
 95%. En inferencia bayesiana esto sustituye a los intervalos de confianza.</p>
diff --git "a/m\303\241s-de-pruebas-de-hip\303\263tesis-e-intervalos.html" "b/m\303\241s-de-pruebas-de-hip\303\263tesis-e-intervalos.html"
index 8885c71..f295f36 100644
--- "a/m\303\241s-de-pruebas-de-hip\303\263tesis-e-intervalos.html"
+++ "b/m\303\241s-de-pruebas-de-hip\303\263tesis-e-intervalos.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git "a/m\303\251todos-de-cadenas-de-markov-monte-carlo.html" "b/m\303\251todos-de-cadenas-de-markov-monte-carlo.html"
index ed1a165..49423ff 100644
--- "a/m\303\251todos-de-cadenas-de-markov-monte-carlo.html"
+++ "b/m\303\251todos-de-cadenas-de-markov-monte-carlo.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
@@ -620,12 +620,12 @@ <h3>Ejemplo<a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-15" c
 <span id="cb572-2"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#cb572-2" aria-hidden="true" tabindex="-1"></a>media_post <span class="ot">&lt;-</span> <span class="fu">mean</span>(theta)</span>
 <span id="cb572-3"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#cb572-3" aria-hidden="true" tabindex="-1"></a>momento_2_post <span class="ot">&lt;-</span> <span class="fu">mean</span>(theta<span class="sc">^</span><span class="dv">2</span>)</span>
 <span id="cb572-4"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#cb572-4" aria-hidden="true" tabindex="-1"></a><span class="fu">c</span>(media_post, momento_2_post)</span></code></pre></div>
-<pre><code>## [1] 0.7155559 0.5372170</code></pre>
+<pre><code>## [1] 0.7147007 0.5364443</code></pre>
 <p>Y podemos aproximar de esta manera cualquier cantidad de interés que esté basada
 en integrales, como probabilidades asociadas a <span class="math inline">\(\theta\)</span> o cuantiles asociados.
 Por ejemplo, podemos aproximar fácilmente <span class="math inline">\(P(e^{\theta}&gt; 2|x)\)</span> haciendo</p>
 <div class="sourceCode" id="cb574"><pre class="sourceCode r"><code class="sourceCode r"><span id="cb574-1"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#cb574-1" aria-hidden="true" tabindex="-1"></a><span class="fu">mean</span>(<span class="fu">exp</span>(theta) <span class="sc">&gt;</span> <span class="dv">2</span>)</span></code></pre></div>
-<pre><code>## [1] 0.5958</code></pre>
+<pre><code>## [1] 0.5959</code></pre>
 <p>y así sucesivamente.</p>
 <p>Este enfoque, sin embargo, es mucho más flexible y poderoso.</p>
 </div>
@@ -682,10 +682,10 @@ <h3>Ejemplo: varias pruebas independientes<a href="métodos-de-cadenas-de-markov
 <pre><code>## # A tibble: 4 × 2
 ##   sabor     valor_sim
 ##   &lt;chr&gt;         &lt;dbl&gt;
-## 1 fresa         0.755
-## 2 limón         0.783
-## 3 mango         0.819
-## 4 guanábana     0.569</code></pre>
+## 1 fresa         0.886
+## 2 limón         0.729
+## 3 mango         0.701
+## 4 guanábana     0.493</code></pre>
 <div class="sourceCode" id="cb582"><pre class="sourceCode r"><code class="sourceCode r"><span id="cb582-1"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#cb582-1" aria-hidden="true" tabindex="-1"></a><span class="co"># esta no es una manera muy rápida, podríamos calcular todas las</span></span>
 <span id="cb582-2"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#cb582-2" aria-hidden="true" tabindex="-1"></a><span class="co"># simulaciones de cada parámetro de manera vectorizada</span></span>
 <span id="cb582-3"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#cb582-3" aria-hidden="true" tabindex="-1"></a>sims_posterior <span class="ot">&lt;-</span> <span class="fu">tibble</span>(<span class="at">rep =</span> <span class="dv">1</span><span class="sc">:</span><span class="dv">5000</span>) <span class="sc">%&gt;%</span> </span>
@@ -695,16 +695,16 @@ <h3>Ejemplo: varias pruebas independientes<a href="métodos-de-cadenas-de-markov
 <pre><code>## # A tibble: 20,000 × 3
 ##      rep sabor     valor_sim
 ##    &lt;int&gt; &lt;chr&gt;         &lt;dbl&gt;
-##  1     1 fresa         0.732
-##  2     1 limón         0.831
-##  3     1 mango         0.850
-##  4     1 guanábana     0.397
-##  5     2 fresa         0.670
-##  6     2 limón         0.839
-##  7     2 mango         0.664
-##  8     2 guanábana     0.558
-##  9     3 fresa         0.671
-## 10     3 limón         0.758
+##  1     1 fresa         0.727
+##  2     1 limón         0.823
+##  3     1 mango         0.849
+##  4     1 guanábana     0.474
+##  5     2 fresa         0.659
+##  6     2 limón         0.785
+##  7     2 mango         0.866
+##  8     2 guanábana     0.631
+##  9     3 fresa         0.553
+## 10     3 limón         0.719
 ## # ℹ 19,990 more rows</code></pre>
 <p>Y ahora podemos aproximar fácilmente la probabilidad de interés:</p>
 <div class="sourceCode" id="cb584"><pre class="sourceCode r"><code class="sourceCode r"><span id="cb584-1"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#cb584-1" aria-hidden="true" tabindex="-1"></a>sims_posterior <span class="sc">%&gt;%</span> </span>
@@ -718,9 +718,9 @@ <h3>Ejemplo: varias pruebas independientes<a href="métodos-de-cadenas-de-markov
 ##   sabor         n   prop
 ##   &lt;chr&gt;     &lt;int&gt;  &lt;dbl&gt;
 ## 1 fresa      1264 0.0632
-## 2 guanábana     8 0.0004
-## 3 limón      5396 0.270 
-## 4 mango     13332 0.667</code></pre>
+## 2 guanábana    20 0.001 
+## 3 limón      5424 0.271 
+## 4 mango     13292 0.665</code></pre>
 <p>Y vemos que los mejores sabores son mango y limón. La probabilidad posterior de
 que mango sea el sabor preferido por la población es de 66%. La integral correspondiente
 no es trivial.</p>
@@ -753,8 +753,8 @@ <h2>Simulando de la posterior<a href="métodos-de-cadenas-de-markov-monte-carlo.
 de una distribución cualquiera <span class="math inline">\(p(\theta) = K f(\theta)\)</span>, donde sólo conocemos
 la función <span class="math inline">\(f(\theta)\)</span>.</p>
 </div>
-<div id="ejemplo-de-islas" class="section level2 hasAnchor" number="12.1">
-<h2><span class="header-section-number">12.1</span> Ejemplo de islas<a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas" class="anchor-section" aria-label="Anchor link to header"></a></h2>
+<div id="ejemplo-de-islas" class="section level2 unnumbered hasAnchor">
+<h2>Ejemplo de islas<a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas" class="anchor-section" aria-label="Anchor link to header"></a></h2>
 <p>Comenzamos revisando el ejemplo de las islas en <span class="citation">Kruschke (<a href="#ref-Kruschke" role="doc-biblioref">2015</a>)</span> (7.2) para tener más intuición de cómo funciona este algoritmo.</p>
 </div>
 <div id="por-qué-funciona-metrópolis" class="section level2 unnumbered hasAnchor">
@@ -1063,8 +1063,8 @@ <h2>Método de Metrópolis<a href="métodos-de-cadenas-de-markov-monte-carlo.htm
 <pre><code>## # A tibble: 3 × 2
 ##   metodo            media_post
 ##   &lt;chr&gt;                  &lt;dbl&gt;
-## 1 sim Metrópolis         0.605
-## 2 sim Independiente      0.602
+## 1 sim Metrópolis         0.613
+## 2 sim Independiente      0.600
 ## 3 exacto                 0.6</code></pre>
 <div class="mathblock">
 <p>
@@ -1173,7 +1173,7 @@ <h3>Ejemplo<a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-16" c
 <pre><code>## # A tibble: 4 × 2
 ##   tipo            media
 ##   &lt;chr&gt;           &lt;dbl&gt;
-## 1 salto chico     0.128
+## 1 salto chico     0.132
 ## 2 salto grande    0.190
 ## 3 salto apropiado 0.203
 ## 4 exacta          0.2</code></pre>
diff --git "a/propiedades-te\303\263ricas-de-mle.html" "b/propiedades-te\303\263ricas-de-mle.html"
index f2359d8..9e17e2e 100644
--- "a/propiedades-te\303\263ricas-de-mle.html"
+++ "b/propiedades-te\303\263ricas-de-mle.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git "a/pruebas-de-hip\303\263tesis.html" "b/pruebas-de-hip\303\263tesis.html"
index 3c26fb3..e49045c 100644
--- "a/pruebas-de-hip\303\263tesis.html"
+++ "b/pruebas-de-hip\303\263tesis.html"
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git a/reference-keys.txt b/reference-keys.txt
index b19eb9d..e813064 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -15,4 +15,3 @@ más-de-pruebas-de-hipótesis-e-intervalos
 introducción-a-inferencia-bayesiana-1
 calibración-bayesiana-y-regularización
 métodos-de-cadenas-de-markov-monte-carlo
-ejemplo-de-islas
diff --git a/referencias.html b/referencias.html
index d97290a..5ab2a48 100644
--- a/referencias.html
+++ b/referencias.html
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git a/search_index.json b/search_index.json
index d3f2b92..6640986 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Fundamentos de Estadística con Remuestreo Información del curso", " Fundamentos de Estadística con Remuestreo Teresa Ortiz, Felipe González, Alfredo Garbuno Información del curso Notas del curso Fundamentos de Estadística con Remuestreo, este curso busca explicar los principios básicos de la estadística y su papel en el análisis de datos. Nuestro punto de vista es uno de fundamentos, con menos énfasis en recetas o técnicas particulares. Ligas Notas: https://tereom.github.io/fundamentos-2024/ Repositorio con material: https://github.com/tereom/fundamentos-2024 Correo: teresa.ortiz.mancera@gmail.com Zoom clase (miércoles 4:00-7:00 pm): https://itam.zoom.us/j/92745909276 Zoom sesiones de dudas (lunes de 5:30-6:30 pm): https://itam.zoom.us/j/92518922348 Canvas: https://itam.instructure.com/courses/13328 x Este trabajo está bajo una Licencia Creative Commons Atribución 4.0 Internacional. "],["temario.html", "Temario Plan semanal Evaluación", " Temario Plan semanal Datos y análisis exploratorio Referencias: (W. S. Cleveland 1994), (Chihara and Hesterberg 2018) Visualización1 Análisis exploratorio Tipos de datos o estudios Muestras diseñadas y muestras naturales Experimentos y datos observacionales Introducción a Pruebas de Hipótesis Referencias: (Chihara and Hesterberg 2018) Introducción a pruebas de hipótesis. Pruebas de permutaciones Muestras pareadas y otros ejemplos Estimación y distribución de muestreo Referencias: (Chihara and Hesterberg 2018), (Tim C. Hesterberg 2015b) Estimadores y su distribución de muestreo Repaso de probabilidad y Teorema del límite central Introducción a estimación por intervalos Referencias: (Chihara and Hesterberg 2018), (Efron and Tibshirani 1993), (Tim C. Hesterberg 2015b) El método plugin y el boostrap Bootstrap e Intervalos de confianza. Ejemplos. Estimación Referencias: (Chihara and Hesterberg 2018), (Wasserman 2013) Estimación por máxima verosimilitud Ejemplos de estimación por máxima verosimilitud y Bootstrap paramétrico Propiedades de estimadores de máxima verosimilitud Más de pruebas de hipótesis Referencias: (Chihara and Hesterberg 2018), (Wasserman 2013) Pruebas de hipótesis para medias y proporciones: una y dos poblaciones. Introducción a inferencia bayesiana Referencias: (Kruschke 2015) Introducción a inferencia bayesiana Ejemplos de distribuciones conjugadas Introducción a métodos computacionales básicos: Muestreadores Metrópolis y Gibbs Ejemplos de inferencia bayesiana en Stan Evaluación Se evaluará mediante tareas semanales y dos exámenes: Tareas semanales (20%) Examen parcial en clase y a casa (40%) Examen final a casa (40%) Referencias "],["análisis-exploratorio.html", "Sección 1 Análisis exploratorio El papel de la exploración en el análisis de datos Preguntas y datos Algunos conceptos básicos Ejemplos Suavizamiento loess Caso de estudio: nacimientos en México", " Sección 1 Análisis exploratorio “Exploratory data analysis can never be the whole story, but nothing else can serve as the foundation stone –as the first step.” — John Tukey El papel de la exploración en el análisis de datos El estándar científico para contestar preguntas o tomar decisiones es uno que se basa en el análisis de datos. Es decir, en primer lugar se deben reunir todos los datos que puedan contener o sugerir alguna guía para entender mejor la pregunta o la decisión a la que nos enfrentamos. Esta recopilación de datos —que pueden ser cualitativos, cuantitativos, o una mezcla de los dos— debe entonces ser analizada para extraer información relevante para nuestro problema. En análisis de datos existen dos distintos tipos de trabajo: El trabajo exploratorio o de detective: ¿cuáles son los aspectos importantes de estos datos? ¿qué indicaciones generales muestran los datos? ¿qué tareas de análisis debemos empezar haciendo? ¿cuáles son los caminos generales para formular con precisión y contestar algunas preguntas que nos interesen? El trabajo inferencial, confirmatorio, o de juez: ¿cómo evaluar el peso de la evidencia de los descubrimientos del paso anterior? ¿qué tan bien soportadas están las respuestas y conclusiones por nuestro conjunto de datos? Preguntas y datos Cuando observamos un conjunto de datos, independientemente de su tamaño, el paso inicial más importante es entender bajo qué proceso se generan los datos. A grandes rasgos, cuanto más sepamos de este proceso, mejor podemos contestar preguntas de interés. En muchos casos, tendremos que hacer algunos supuestos de cómo se generan estos datos para dar respuestas (condicionales a esos supuestos). Algunos conceptos básicos Empezamos explicando algunas ideas que no serán útiles más adelante. El primer concepto se refiere a entender cómo se distribuyen los datos a los largo de su escala de medición. Comenzamos con un ejemplo: los siguientes datos fueron registrados en un restaurante durante cuatro días consecutivos. library(tidyverse) library(patchwork) # organizar gráficas library(gt) # formatear tablas source(&quot;R/funciones_auxiliares.R&quot;) # usamos los datos tips del paquete reshape2 propinas &lt;- read_csv(&quot;./data/propinas.csv&quot;) Y vemos una muestra slice_sample(propinas, n = 10) |&gt; gt() #xddpgkkgug table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #xddpgkkgug thead, #xddpgkkgug tbody, #xddpgkkgug tfoot, #xddpgkkgug tr, #xddpgkkgug td, #xddpgkkgug th { border-style: none; } #xddpgkkgug p { margin: 0; padding: 0; } #xddpgkkgug .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #xddpgkkgug .gt_caption { padding-top: 4px; padding-bottom: 4px; } #xddpgkkgug .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #xddpgkkgug .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #xddpgkkgug .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #xddpgkkgug .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #xddpgkkgug .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #xddpgkkgug .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #xddpgkkgug .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #xddpgkkgug .gt_column_spanner_outer:first-child { padding-left: 0; } #xddpgkkgug .gt_column_spanner_outer:last-child { padding-right: 0; } #xddpgkkgug .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #xddpgkkgug .gt_spanner_row { border-bottom-style: hidden; } #xddpgkkgug .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #xddpgkkgug .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #xddpgkkgug .gt_from_md > :first-child { margin-top: 0; } #xddpgkkgug .gt_from_md > :last-child { margin-bottom: 0; } #xddpgkkgug .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #xddpgkkgug .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #xddpgkkgug .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #xddpgkkgug .gt_row_group_first td { border-top-width: 2px; } #xddpgkkgug .gt_row_group_first th { border-top-width: 2px; } #xddpgkkgug .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #xddpgkkgug .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #xddpgkkgug .gt_first_summary_row.thick { border-top-width: 2px; } #xddpgkkgug .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #xddpgkkgug .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #xddpgkkgug .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #xddpgkkgug .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #xddpgkkgug .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #xddpgkkgug .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #xddpgkkgug .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #xddpgkkgug .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #xddpgkkgug .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #xddpgkkgug .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #xddpgkkgug .gt_left { text-align: left; } #xddpgkkgug .gt_center { text-align: center; } #xddpgkkgug .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #xddpgkkgug .gt_font_normal { font-weight: normal; } #xddpgkkgug .gt_font_bold { font-weight: bold; } #xddpgkkgug .gt_font_italic { font-style: italic; } #xddpgkkgug .gt_super { font-size: 65%; } #xddpgkkgug .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #xddpgkkgug .gt_asterisk { font-size: 100%; vertical-align: 0; } #xddpgkkgug .gt_indent_1 { text-indent: 5px; } #xddpgkkgug .gt_indent_2 { text-indent: 10px; } #xddpgkkgug .gt_indent_3 { text-indent: 15px; } #xddpgkkgug .gt_indent_4 { text-indent: 20px; } #xddpgkkgug .gt_indent_5 { text-indent: 25px; } #xddpgkkgug .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #xddpgkkgug div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } cuenta_total propina fumador dia momento num_personas 9.78 1.73 No Jue Comida 2 13.42 1.68 No Jue Comida 2 11.02 1.98 Si Sab Cena 2 22.49 3.50 No Vie Cena 2 11.59 1.50 Si Sab Cena 2 24.71 5.85 No Jue Comida 2 9.60 4.00 Si Dom Cena 2 19.65 3.00 No Sab Cena 2 18.04 3.00 No Dom Cena 2 30.40 5.60 No Dom Cena 4 Aquí la unidad de observación es una cuenta particular. Tenemos tres mediciones numéricas de cada cuenta: cúanto fue la cuenta total, la propina, y el número de personas asociadas a la cuenta. Los datos están separados según se fumó o no en la mesa, y temporalmente en dos partes: el día (Jueves, Viernes, Sábado o Domingo), cada uno separado por Cena y Comida. Denotamos por \\(x\\) el valor de medición de una unidad de observación. Usualmente utilizamos sub-índices para identificar entre diferentes puntos de datos (observaciones), por ejemplo, \\(x_n\\) para la \\(n-\\)ésima observación. De tal forma que una colección de \\(N\\) observaciones la escribimos como \\[\\begin{align} \\{x_1, \\ldots, x_N\\}. \\end{align}\\] El primer tipo de comparaciones que nos interesa hacer es para una medición: ¿Varían mucho o poco los datos de un tipo de medición? ¿Cuáles son valores típicos o centrales? ¿Existen valores atípicos? Supongamos entonces que consideramos simplemente la variable de cuenta_total. Podemos comenzar por ordenar los datos, y ver cuáles datos están en los extremos y cuáles están en los lugares centrales: propinas &lt;- propinas |&gt; mutate(orden_cuenta = rank(cuenta_total, ties.method = &quot;first&quot;), f = (orden_cuenta - 0.5) / n()) cuenta &lt;- propinas |&gt; select(orden_cuenta, f, cuenta_total) |&gt; arrange(f) bind_rows(head(cuenta), tail(cuenta)) |&gt; gt() |&gt; fmt_number(columns = f, decimals = 3) #zwocxfeiud table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #zwocxfeiud thead, #zwocxfeiud tbody, #zwocxfeiud tfoot, #zwocxfeiud tr, #zwocxfeiud td, #zwocxfeiud th { border-style: none; } #zwocxfeiud p { margin: 0; padding: 0; } #zwocxfeiud .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #zwocxfeiud .gt_caption { padding-top: 4px; padding-bottom: 4px; } #zwocxfeiud .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #zwocxfeiud .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #zwocxfeiud .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #zwocxfeiud .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #zwocxfeiud .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #zwocxfeiud .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #zwocxfeiud .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #zwocxfeiud .gt_column_spanner_outer:first-child { padding-left: 0; } #zwocxfeiud .gt_column_spanner_outer:last-child { padding-right: 0; } #zwocxfeiud .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #zwocxfeiud .gt_spanner_row { border-bottom-style: hidden; } #zwocxfeiud .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #zwocxfeiud .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #zwocxfeiud .gt_from_md > :first-child { margin-top: 0; } #zwocxfeiud .gt_from_md > :last-child { margin-bottom: 0; } #zwocxfeiud .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #zwocxfeiud .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #zwocxfeiud .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #zwocxfeiud .gt_row_group_first td { border-top-width: 2px; } #zwocxfeiud .gt_row_group_first th { border-top-width: 2px; } #zwocxfeiud .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #zwocxfeiud .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #zwocxfeiud .gt_first_summary_row.thick { border-top-width: 2px; } #zwocxfeiud .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #zwocxfeiud .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #zwocxfeiud .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #zwocxfeiud .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #zwocxfeiud .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #zwocxfeiud .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #zwocxfeiud .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #zwocxfeiud .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #zwocxfeiud .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #zwocxfeiud .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #zwocxfeiud .gt_left { text-align: left; } #zwocxfeiud .gt_center { text-align: center; } #zwocxfeiud .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #zwocxfeiud .gt_font_normal { font-weight: normal; } #zwocxfeiud .gt_font_bold { font-weight: bold; } #zwocxfeiud .gt_font_italic { font-style: italic; } #zwocxfeiud .gt_super { font-size: 65%; } #zwocxfeiud .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #zwocxfeiud .gt_asterisk { font-size: 100%; vertical-align: 0; } #zwocxfeiud .gt_indent_1 { text-indent: 5px; } #zwocxfeiud .gt_indent_2 { text-indent: 10px; } #zwocxfeiud .gt_indent_3 { text-indent: 15px; } #zwocxfeiud .gt_indent_4 { text-indent: 20px; } #zwocxfeiud .gt_indent_5 { text-indent: 25px; } #zwocxfeiud .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #zwocxfeiud div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } orden_cuenta f cuenta_total 1 0.002 3.07 2 0.006 5.75 3 0.010 7.25 4 0.014 7.25 5 0.018 7.51 6 0.023 7.56 239 0.977 44.30 240 0.982 45.35 241 0.986 48.17 242 0.990 48.27 243 0.994 48.33 244 0.998 50.81 También podemos graficar los datos en orden, interpolando valores consecutivos. A esta función le llamamos la función de cuantiles para la variable cuenta_total. Nos sirve para comparar directamente los distintos valores que observamos los datos según el orden que ocupan. En particular, podemos estudiar la dispersión y valores centrales de los datos observados: El rango de datos va de unos 3 dólares hasta 50 dólares Los valores centrales, por ejemplo el 50% de los valores más centrales, están entre unos 13 y 25 dólares. El valor que divide en dos mitades iguales a los datos es de alrededor de 18 dólares. El cuantil \\(f\\), que denotamos por \\(q(f)\\) es valor a lo largo de la escala de medición de los datos tal que aproximadamente una fracción \\(f\\) de los datos son menores o iguales a \\(q(f)\\). Al cuantil \\(f=0.5\\) le llamamos la mediana. A los cuantiles \\(f=0.25\\) y \\(f=0.75\\) les llamamos cuartiles inferior y superior. En nuestro ejemplo: Los valores centrales —del cuantil 0.25 al 0.75, por decir un ejemplo— están entre unos 13 y 25 dólares. Estos dos cuantiles se llaman cuartil inferior y cuartil superior respectivamente El cuantil 0.5 (o también conocido como mediana) está alrededor de 18 dólares. Éste último puede ser utilizado para dar un valor central de la distribución de valores para cuenta_total. Asimismo podemos dar resúmenes más refinados si es necesario. Por ejemplo, podemos reportar que: El cuantil 0.95 es de unos 35 dólares — sólo 5% de las cuentas son de más de 35 dólares El cuantil 0.05 es de unos 8 dólares — sólo 5% de las cuentas son de 8 dólares o menos. Finalmente, la forma de la gráfica se interpreta usando su pendiente (tasa de cambio) haciendo comparaciones en diferentes partes de la gráfica: La distribución de valores tiene asimetría: el 10% de las cuentas más altas tiene considerablemente más dispersión que el 10% de las cuentas más bajas. Entre los cuantiles 0.2 y 0.7 es donde existe mayor densidad de datos: la pendiente (tasa de cambio) es baja, lo que significa que al avanzar en los valores observados, los cuantiles (el porcentaje de casos) aumenta rápidamente. Cuando la pendiente alta, quiere decir que los datos tienen más dispersión local o están más separados. Observación: Hay varias maneras de definir los cuantiles (ver (William S. Cleveland 1993)): Supongamos que queremos definir \\(q(f)\\), y denotamos los datos ordenados como \\(x_{(1)}, x_{(2)}, \\ldots, x_{(N)}\\), de forma que \\(x_{(1)}\\) es el dato más chico y \\(x_{(N)}\\) es el dato más grande. Para cada \\(x_{(i)}\\) definimos \\[f_i = i / N\\] entonces definimos el cuantil \\(q(f_i)=x_{(i)}\\). Para cualquier \\(f\\) entre 0 y 1, podemos definir \\(q(f)\\) como sigue: si \\(f\\) está entre \\(f_i\\) y \\(f_{i+1}\\) interpolamos linealmente los valores correspondientes \\(x_{(i)}\\) y \\(x_{(i+1)}\\). En la práctica, es más conveniente usar \\(f_i= \\frac{i - 0.5}{N}\\). La gráfica de cuantiles no cambia mucho comparado con la difinición anterior, y esto nos permitirá comparar de mejor manera con distribuciones teóricas que no tienen definido su cuantil 0 y el 1, pues tienen soporte en los números reales (como la distribución normal, por ejemplo). Asociada a la función de cuantiles \\(q\\) tenemos la distribución acumulada empírica de los datos, que es aproximadamente inversa de la función de cuantiles, y se define como: \\[\\hat{F}(x) = i/N\\] si \\(x_{(i)} \\leq x &lt; x_{(i+1)}\\). Nótese que \\(\\hat{F}(q(f_i)) = i/N = f_i\\) (demuéstralo). acum_cuenta &lt;- ecdf(cuenta$cuenta_total) cuenta &lt;- cuenta |&gt; mutate(dea_cuenta_total = acum_cuenta(cuenta_total)) g_acum &lt;- ggplot(cuenta, aes(x = cuenta_total, y = dea_cuenta_total)) + geom_point() + labs(subtitle = &quot;Distribución acum empírica de cuenta total&quot;, x = &quot;&quot;) g_cuantiles + g_acum La función de distribución acumulada empírica es otra forma de graficar la dispersión de los datos. En su gráfica vemos que proporción de los datos que son iguales o están por debajo de cada valor en el eje horizontal. Nota: En análisis de datos, es más frecuente utilizar la función de cuantiles pues existen versiones más generales que son útiles, por ejemplo, para evaluar ajuste de modelos probabilísticos En la teoría, generalmente es más común utilizar la fda empírica, que tiene una única definición que veremos coincide con definiciones teóricas. Histogramas En algunos casos, es más natural hacer un histograma, donde dividimos el rango de la variable en cubetas o intervalos (en este caso de igual longitud), y graficamos por medio de barras cuántos datos caen en cada cubeta: Es una gráfica más popular, pero perdemos cierto nivel de detalle, y distintas particiones resaltan distintos aspectos de los datos. ¿Cómo se ve la gráfica de cuantiles de las propinas? ¿Cómo crees que esta gráfica se compara con distintos histogramas? g_1 &lt;- ggplot(propinas, aes(sample = propina)) + geom_qq(distribution = stats::qunif) + labs(x = &quot;f&quot;, y = &quot;propina&quot;) g_1 Finalmente, una gráfica más compacta que resume la gráfica de cuantiles o el histograma es el diagrama de caja y brazos. Mostramos dos versiones, la clásica de Tukey (T) y otra versión menos común de Spear/Tufte (ST): library(ggthemes) cuartiles &lt;- quantile(cuenta$cuenta_total) cuartiles |&gt; round(2) ## 0% 25% 50% 75% 100% ## 3.07 13.35 17.80 24.13 50.81 g_1 &lt;- ggplot(cuenta, aes(x = f, y = cuenta_total)) + labs(subtitle = &quot;Gráfica de cuantiles: Cuenta total&quot;) + geom_hline(yintercept = cuartiles[2:4], colour = &quot;gray&quot;) + geom_point(alpha = 0.5) + geom_line() g_2 &lt;- ggplot(cuenta, aes(x = factor(&quot;ST&quot;, levels = &quot;ST&quot;), y = cuenta_total)) + geom_tufteboxplot() + labs(subtitle = &quot;&quot;, x = &quot;&quot;, y = &quot;&quot;) g_3 &lt;- ggplot(cuenta, aes(x = factor(&quot;T&quot;), y = cuenta_total)) + geom_boxplot() + labs(subtitle = &quot;&quot;, x = &quot;&quot;, y = &quot;&quot;) g_4 &lt;- ggplot(cuenta, aes(x = factor(&quot;P&quot;), y = cuenta_total)) + geom_jitter(height = 0, width =0.2, alpha = 0.5) + labs(subtitle = &quot;&quot;, x = &quot;&quot;, y = &quot;&quot;) g_5 &lt;- ggplot(cuenta, aes(x = factor(&quot;V&quot;), y = cuenta_total)) + geom_violin() + labs(subtitle = &quot;&quot;, x = &quot;&quot;, y = &quot;&quot;) g_1 + g_2 + g_3 + g_4 + plot_layout(widths = c(8, 2, 2, 2)) El diagrama de abajo explica los elementos de la versión típica del diagrama de caja y brazos (boxplot). RIC se refiere al Rango Intercuantílico, definido por la diferencia entre los cuantiles 25% y 75%. Figura: Jumanbar / CC BY-SA Ventajas en el análisis inicial En un principio del análisis, estos resúmenes (cuantiles) pueden ser más útiles que utilizar medias y varianzas, por ejemplo. La razón es que los cuantiles: Son cantidades más fácilmente interpretables Los cuantiles centrales son más resistentes a valores atípicos que medias o varianzas Permiten identificar valores extremos Es fácil comparar cuantiles de distintos bonches de datos en la misma escala Nota: Existen diferentes definiciones para calcular cuantiles de una muestra de datos, puedes leer más en este artículo. Media y desviación estándar Las medidas más comunes de localización y dispersión para un conjunto de datos son la media muestral y la desviación estándar muestral. En general, no son muy apropiadas para iniciar el análisis exploratorio, pues: Son medidas más difíciles de interpretar y explicar que los cuantiles. En este sentido, son medidas especializadas. Por ejemplo, compara una explicación intuitiva de la mediana contra una explicación intuitiva de la media. No son resistentes a valores atípicos. Su falta de resistencia los vuelve poco útiles en las primeras etapas de limpieza y descripción y en resúmenes deficientes para distribuciones irregulares (con colas largas por ejemplo). La media, o promedio, se denota por \\(\\bar x\\) y se define como \\[\\begin{align} \\bar x = \\frac1N \\sum_{n = 1}^N x_n. \\end{align}\\] La desviación estándar muestral se define como \\[\\begin{align} \\text{std}(x) = \\sqrt{\\frac1{N-1} \\sum_{n = 1}^N (x_n - \\bar x)^2}. \\end{align}\\] Observación: Si \\(N\\) no es muy chica, no importa mucho si dividimos por \\(N\\) o por \\(N-1\\) en la fórmula de la desviación estándar. La razón de que típicamente se usa \\(N-1\\) la veremos más adelante, en la parte de estimación. Por otro lado, ventajas de estas medidas de centralidad y dispersión son: La media y desviación estándar son computacionalmente convenientes. Por lo tanto regresaremos a estas medidas una vez que estudiemos modelos de probabilidad básicos. En muchas ocasiones conviene usar estas medidas pues permite hacer comparaciones históricas o tradicionales —pues análisis anteriores pudieran estar basados en éstas. Considera el caso de tener \\(N\\) observaciones y asume que ya tienes calculado el promedio para dichas observaciones. Este promedio lo denotaremos por \\(\\bar x_N\\). Ahora, considera que has obtenido \\(M\\) observaciones más. Escribe una fórmula recursiva para la media del conjunto total de datos \\(\\bar x_{N+M}\\) en función de lo que ya tenías precalculado \\(\\bar x_N.\\) ¿En qué situaciones esta propiedad puede ser conveniente? Ejemplos Precios de casas En este ejemplo consideremos los datos de precios de ventas de la ciudad de Ames, Iowa. En particular nos interesa entender la variación del precio de las casas. Por este motivo calculamos los cuantiles que corresponden al 25%, 50% y 75% (cuartiles), así como el mínimo y máximo de los precios de las casas: quantile(casas |&gt; pull(precio_miles)) ## 0% 25% 50% 75% 100% ## 37.9 132.0 165.0 215.0 755.0 Comprueba que el mínimo y máximo están asociados a los cuantiles 0% y 100%, respectivamente. Una posible comparación es considerar los precios y sus variación en función de zona de la ciudad en que se encuentra una vivienda. Podemos usar diagramas de caja y brazos para hacer una comparación burda de los precios en distintas zonas de la ciudad: ggplot(casas, aes(x = nombre_zona, y = precio_miles)) + geom_boxplot() + coord_flip() La primera pregunta que nos hacemos es cómo pueden variar las características de las casas dentro de cada zona. Para esto, podemos considerar el área de las casas. En lugar de graficar el precio, graficamos el precio por metro cuadrado, por ejemplo: ggplot(casas, aes(x = nombre_zona, y = precio_m2)) + geom_boxplot() + coord_flip() Podemos cuantificar la variación que observamos de zona a zona y la variación que hay dentro de cada una de las zonas. Una primera aproximación es observar las variación del precio al calcular la mediana dentro de cada zona, y después cuantificar por medio de cuantiles cómo varía la mediana entre zonas: casas |&gt; group_by(nombre_zona) |&gt; summarise(mediana_zona = median(precio_m2), .groups = &quot;drop&quot;) |&gt; arrange(mediana_zona) |&gt; pull(mediana_zona) |&gt; quantile() |&gt; round() ## 0% 25% 50% 75% 100% ## 963 1219 1298 1420 1725 Por otro lado, las variaciones con respecto a las medianas dentro de cada zona, por grupo, se resume como: quantile(casas |&gt; group_by(nombre_zona) |&gt; mutate(residual = precio_m2 - median(precio_m2)) |&gt; pull(residual)) |&gt; round() ## 0% 25% 50% 75% 100% ## -765 -166 0 172 1314 Nótese que este último paso tiene sentido pues la variación dentro de las zonas, en términos de precio por metro cuadrado, es similar. Esto no lo podríamos haber hecho de manera efectiva si se hubiera utilizado el precio de las casas sin ajustar por su tamaño. Podemos resumir este primer análisis con un par de gráficas de cuantiles (William S. Cleveland (1993)): mediana &lt;- median(casas$precio_m2) resumen &lt;- casas |&gt; select(nombre_zona, precio_m2) |&gt; group_by(nombre_zona) |&gt; mutate(mediana_zona = median(precio_m2)) |&gt; mutate(residual = precio_m2 - mediana_zona) |&gt; ungroup() |&gt; mutate(mediana_zona = mediana_zona - mediana) |&gt; select(nombre_zona, mediana_zona, residual) |&gt; pivot_longer(mediana_zona:residual, names_to = &quot;tipo&quot;, values_to = &quot;valor&quot;) ggplot(resumen, aes(sample = valor)) + geom_qq(distribution = stats::qunif) + facet_wrap(~ tipo) + ylab(&quot;Precio por m2&quot;) + xlab(&quot;f&quot;) + labs(subtitle = &quot;Precio por m2 por zona&quot;, caption = paste0(&quot;Mediana total de &quot;, round(mediana))) Vemos que la mayor parte de la variación del precio por metro cuadrado ocurre dentro de cada zona, una vez que controlamos por el tamaño de las casas. La variación dentro de cada zona es aproximadamente simétrica, aunque la cola derecha es ligeramente más larga con algunos valores extremos. Podemos seguir con otro indicador importante: la calificación de calidad de los terminados de las casas. Como primer intento podríamos hacer: Lo que indica que las calificaciones de calidad están distribuidas de manera muy distinta a lo largo de las zonas, y que probablemente no va ser simple desentrañar qué variación del precio se debe a la zona y cuál se debe a la calidad. Prueba Enlace Consideremos la prueba Enlace (2011) de matemáticas para primarias. Una primera pregunta que alguien podría hacerse es: ¿cuáles escuelas son mejores en este rubro, las privadas o las públicas? enlace_tbl &lt;- enlace |&gt; group_by(tipo) |&gt; summarise(n_escuelas = n(), cuantiles = list(cuantil(mate_6, c(0.05, 0.25, 0.5, 0.75, 0.95)))) |&gt; unnest(cols = cuantiles) |&gt; mutate(valor = round(valor)) enlace_tbl |&gt; spread(cuantil, valor) |&gt; gt() #sykxxonvyi table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #sykxxonvyi thead, #sykxxonvyi tbody, #sykxxonvyi tfoot, #sykxxonvyi tr, #sykxxonvyi td, #sykxxonvyi th { border-style: none; } #sykxxonvyi p { margin: 0; padding: 0; } #sykxxonvyi .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #sykxxonvyi .gt_caption { padding-top: 4px; padding-bottom: 4px; } #sykxxonvyi .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #sykxxonvyi .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #sykxxonvyi .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #sykxxonvyi .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sykxxonvyi .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #sykxxonvyi .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #sykxxonvyi .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #sykxxonvyi .gt_column_spanner_outer:first-child { padding-left: 0; } #sykxxonvyi .gt_column_spanner_outer:last-child { padding-right: 0; } #sykxxonvyi .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #sykxxonvyi .gt_spanner_row { border-bottom-style: hidden; } #sykxxonvyi .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #sykxxonvyi .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #sykxxonvyi .gt_from_md > :first-child { margin-top: 0; } #sykxxonvyi .gt_from_md > :last-child { margin-bottom: 0; } #sykxxonvyi .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #sykxxonvyi .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #sykxxonvyi .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #sykxxonvyi .gt_row_group_first td { border-top-width: 2px; } #sykxxonvyi .gt_row_group_first th { border-top-width: 2px; } #sykxxonvyi .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #sykxxonvyi .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #sykxxonvyi .gt_first_summary_row.thick { border-top-width: 2px; } #sykxxonvyi .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sykxxonvyi .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #sykxxonvyi .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #sykxxonvyi .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #sykxxonvyi .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #sykxxonvyi .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sykxxonvyi .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #sykxxonvyi .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #sykxxonvyi .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #sykxxonvyi .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #sykxxonvyi .gt_left { text-align: left; } #sykxxonvyi .gt_center { text-align: center; } #sykxxonvyi .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #sykxxonvyi .gt_font_normal { font-weight: normal; } #sykxxonvyi .gt_font_bold { font-weight: bold; } #sykxxonvyi .gt_font_italic { font-style: italic; } #sykxxonvyi .gt_super { font-size: 65%; } #sykxxonvyi .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #sykxxonvyi .gt_asterisk { font-size: 100%; vertical-align: 0; } #sykxxonvyi .gt_indent_1 { text-indent: 5px; } #sykxxonvyi .gt_indent_2 { text-indent: 10px; } #sykxxonvyi .gt_indent_3 { text-indent: 15px; } #sykxxonvyi .gt_indent_4 { text-indent: 20px; } #sykxxonvyi .gt_indent_5 { text-indent: 25px; } #sykxxonvyi .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #sykxxonvyi div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } tipo n_escuelas 0.05 0.25 0.5 0.75 0.95 Indígena/Conafe 13599 304 358 412 478 588 General 60166 380 454 502 548 631 Particular 6816 479 551 593 634 703 Para un análisis exploratorio podemos utilizar distintas gráficas. Por ejemplo, podemos utilizar nuevamente las gráficas de caja y brazos, así como graficar los percentiles. Nótese que en la gráfica 1 se utilizan los cuantiles 0.05, 0.25, 0.5, 0.75 y 0.95: ## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0. ## ℹ Please use `linewidth` instead. ## This warning is displayed once every 8 hours. ## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was ## generated. Se puede discutir qué tan apropiada es cada gráfica con el objetivo de realizar comparaciones. Sin duda, graficar más cuantiles es más útil para hacer comparaciones. Por ejemplo, en la Gráfica 1 podemos ver que la mediana de las escuelas generales está cercana al cuantil 5% de las escuelas particulares. Por otro lado, el diagrama de caja y brazos muestra también valores “atípicos”. Antes de contestar prematuramente la pregunta: ¿cuáles son las mejores escuelas? busquemos mejorar la interpretabilidad de nuestras comparaciones. Podemos comenzar por agregar, por ejemplo, el nivel del marginación del municipio donde se encuentra la escuela. Para este objetivo, podemos usar páneles (pequeños múltiplos útiles para hacer comparaciones) y graficar: Esta gráfica pone en contexto la pregunta inicial, y permite evidenciar la dificultad de contestarla. En particular: Señala que la pregunta no sólo debe concentarse en el tipo de “sistema”: pública, privada, etc. Por ejemplo, las escuelas públicas en zonas de marginación baja no tienen una distribución de calificaciones muy distinta a las privadas en zonas de marginación alta. El contexto de la escuela es importante. Debemos de pensar qué factores –por ejemplo, el entorno familiar de los estudiantes– puede resultar en comparaciones que favorecen a las escuelas privadas. Un ejemplo de esto es considerar si los estudiantes tienen que trabajar o no. A su vez, esto puede o no ser reflejo de la calidad del sistema educativo. Si esto es cierto, entonces la pregunta inicial es demasiado vaga y mal planteada. Quizá deberíamos intentar entender cuánto “aporta” cada escuela a cada estudiante, como medida de qué tan buena es cada escuela. Estados y calificaciones en SAT ¿Cómo se relaciona el gasto por alumno, a nivel estatal, con sus resultados académicos? Hay trabajo considerable en definir estos términos, pero supongamos que tenemos el siguiente conjunto de datos (Guber 1999), que son datos oficiales agregados por estado de Estados Unidos. Consideremos el subconjunto de variables sat, que es la calificación promedio de los alumnos en cada estado (para 1997) y expend, que es el gasto en miles de dólares por estudiante en (1994-1995). sat &lt;- read_csv(&quot;data/sat.csv&quot;) sat_tbl &lt;- sat |&gt; select(state, expend, sat) |&gt; gather(variable, valor, expend:sat) |&gt; group_by(variable) |&gt; summarise(cuantiles = list(cuantil(valor))) |&gt; unnest(cols = c(cuantiles)) |&gt; mutate(valor = round(valor, 1)) |&gt; spread(cuantil, valor) sat_tbl |&gt; gt() #trvhlcrcph table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #trvhlcrcph thead, #trvhlcrcph tbody, #trvhlcrcph tfoot, #trvhlcrcph tr, #trvhlcrcph td, #trvhlcrcph th { border-style: none; } #trvhlcrcph p { margin: 0; padding: 0; } #trvhlcrcph .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #trvhlcrcph .gt_caption { padding-top: 4px; padding-bottom: 4px; } #trvhlcrcph .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #trvhlcrcph .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #trvhlcrcph .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #trvhlcrcph .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #trvhlcrcph .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #trvhlcrcph .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #trvhlcrcph .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #trvhlcrcph .gt_column_spanner_outer:first-child { padding-left: 0; } #trvhlcrcph .gt_column_spanner_outer:last-child { padding-right: 0; } #trvhlcrcph .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #trvhlcrcph .gt_spanner_row { border-bottom-style: hidden; } #trvhlcrcph .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #trvhlcrcph .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #trvhlcrcph .gt_from_md > :first-child { margin-top: 0; } #trvhlcrcph .gt_from_md > :last-child { margin-bottom: 0; } #trvhlcrcph .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #trvhlcrcph .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #trvhlcrcph .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #trvhlcrcph .gt_row_group_first td { border-top-width: 2px; } #trvhlcrcph .gt_row_group_first th { border-top-width: 2px; } #trvhlcrcph .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #trvhlcrcph .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #trvhlcrcph .gt_first_summary_row.thick { border-top-width: 2px; } #trvhlcrcph .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #trvhlcrcph .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #trvhlcrcph .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #trvhlcrcph .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #trvhlcrcph .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #trvhlcrcph .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #trvhlcrcph .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #trvhlcrcph .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #trvhlcrcph .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #trvhlcrcph .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #trvhlcrcph .gt_left { text-align: left; } #trvhlcrcph .gt_center { text-align: center; } #trvhlcrcph .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #trvhlcrcph .gt_font_normal { font-weight: normal; } #trvhlcrcph .gt_font_bold { font-weight: bold; } #trvhlcrcph .gt_font_italic { font-style: italic; } #trvhlcrcph .gt_super { font-size: 65%; } #trvhlcrcph .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #trvhlcrcph .gt_asterisk { font-size: 100%; vertical-align: 0; } #trvhlcrcph .gt_indent_1 { text-indent: 5px; } #trvhlcrcph .gt_indent_2 { text-indent: 10px; } #trvhlcrcph .gt_indent_3 { text-indent: 15px; } #trvhlcrcph .gt_indent_4 { text-indent: 20px; } #trvhlcrcph .gt_indent_5 { text-indent: 25px; } #trvhlcrcph .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #trvhlcrcph div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } variable 0 0.25 0.5 0.75 1 expend 3.7 4.9 5.8 6.4 9.8 sat 844.0 897.2 945.5 1032.0 1107.0 Esta variación es considerable para promedios del SAT: el percentil 75 es alrededor de 1050 puntos, mientras que el percentil 25 corresponde a alrededor de 800. Igualmente, hay diferencias considerables de gasto por alumno (miles de dólares) a lo largo de los estados. Ahora hacemos nuestro primer ejercico de comparación: ¿Cómo se ven las calificaciones para estados en distintos niveles de gasto? Podemos usar una gráfica de dispersión: library(ggrepel) ggplot(sat, aes(x = expend, y = sat, label = state)) + geom_point(colour = &quot;red&quot;, size = 2) + geom_text_repel(colour = &quot;gray50&quot;) + xlab(&quot;Gasto por alumno (miles de dólares)&quot;) + ylab(&quot;Calificación promedio en SAT&quot;) Estas comparaciones no son de alta calidad, solo estamos usando 2 variables —que son muy pocas— y no hay mucho que podamos decir en cuanto explicación. Sin duda nos hace falta una imagen más completa. Necesitaríamos entender la correlación que existe entre las demás características de nuestras unidades de estudio. Las unidades que estamos comparando pueden diferir fuertemente en otras propiedades importantes (o dimensiones), lo cual no permite interpretar la gráfica de manera sencilla. Una variable que tenemos es el porcentaje de alumnos de cada estado que toma el SAT. Podemos agregar como sigue: ggplot(sat, aes(x = expend, y = math, label=state, colour = frac)) + geom_point() + geom_text_repel() + xlab(&quot;Gasto por alumno (miles de dólares)&quot;) + ylab(&quot;Calificación promedio en SAT&quot;) Esto nos permite entender por qué nuestra comparación inicial es relativamente pobre. Los estados con mejores resultados promedio en el SAT son aquellos donde una fracción relativamente baja de los estudiantes toma el examen. La diferencia es considerable. En este punto podemos hacer varias cosas. Una primera idea es intentar comparar estados más similares en cuanto a la población de alumnos que asiste. Podríamos hacer grupos como sigue: set.seed(991) k_medias_sat &lt;- kmeans(sat |&gt; select(frac), centers = 4, nstart = 100, iter.max = 100) sat$clase &lt;- k_medias_sat$cluster sat &lt;- sat |&gt; group_by(clase) |&gt; mutate(clase_media = round(mean(frac))) |&gt; ungroup() |&gt; mutate(clase_media = factor(clase_media)) sat &lt;- sat |&gt; mutate(rank_p = rank(frac, ties= &quot;first&quot;) / length(frac)) ggplot(sat, aes(x = rank_p, y = frac, label = state, colour = clase_media)) + geom_point(size = 2) Estos resultados indican que es más probable que buenos alumnos decidan hacer el SAT. Lo interesante es que esto ocurre de manera diferente en cada estado. Por ejemplo, en algunos estados era más común otro examen: el ACT. Si hacemos clusters de estados según el % de alumnos, empezamos a ver otra historia. Para esto, ajustemos rectas de mínimos cuadrados como referencia: Esto da una imagen muy diferente a la que originalmente planteamos. Nota que dependiendo de cómo categorizamos, esta gráfica puede variar (puedes intentar con más o menos grupos, por ejemplo). Tablas de conteos Consideremos los siguientes datos de tomadores de té (del paquete FactoMineR (Lê et al. 2008)): tea &lt;- read_csv(&quot;data/tea.csv&quot;) # nombres y códigos te &lt;- tea |&gt; select(how, price, sugar) |&gt; rename(presentacion = how, precio = price, azucar = sugar) |&gt; mutate( presentacion = fct_recode(presentacion, suelto = &quot;unpackaged&quot;, bolsas = &quot;tea bag&quot;, mixto = &quot;tea bag+unpackaged&quot;), precio = fct_recode(precio, marca = &quot;p_branded&quot;, variable = &quot;p_variable&quot;, barato = &quot;p_cheap&quot;, marca_propia = &quot;p_private label&quot;, desconocido = &quot;p_unknown&quot;, fino = &quot;p_upscale&quot;), azucar = fct_recode(azucar, sin_azúcar = &quot;No.sugar&quot;, con_azúcar = &quot;sugar&quot;)) sample_n(te, 10) ## # A tibble: 10 × 3 ## presentacion precio azucar ## &lt;fct&gt; &lt;fct&gt; &lt;fct&gt; ## 1 bolsas marca sin_azúcar ## 2 bolsas variable sin_azúcar ## 3 bolsas marca con_azúcar ## 4 bolsas fino con_azúcar ## 5 mixto variable con_azúcar ## 6 mixto fino con_azúcar ## 7 bolsas marca sin_azúcar ## 8 bolsas fino sin_azúcar ## 9 mixto variable con_azúcar ## 10 mixto variable sin_azúcar Nos interesa ver qué personas compran té suelto, y de qué tipo. Empezamos por ver las proporciones que compran té según su empaque (en bolsita o suelto): precio &lt;- te |&gt; count(precio) |&gt; mutate(prop = round(100 * n / sum(n))) |&gt; select(-n) tipo &lt;- te |&gt; count(presentacion) |&gt; mutate(pct = round(100 * n / sum(n))) tipo |&gt; gt() #wiljiluquk table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #wiljiluquk thead, #wiljiluquk tbody, #wiljiluquk tfoot, #wiljiluquk tr, #wiljiluquk td, #wiljiluquk th { border-style: none; } #wiljiluquk p { margin: 0; padding: 0; } #wiljiluquk .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #wiljiluquk .gt_caption { padding-top: 4px; padding-bottom: 4px; } #wiljiluquk .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #wiljiluquk .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #wiljiluquk .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #wiljiluquk .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #wiljiluquk .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #wiljiluquk .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #wiljiluquk .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #wiljiluquk .gt_column_spanner_outer:first-child { padding-left: 0; } #wiljiluquk .gt_column_spanner_outer:last-child { padding-right: 0; } #wiljiluquk .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #wiljiluquk .gt_spanner_row { border-bottom-style: hidden; } #wiljiluquk .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #wiljiluquk .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #wiljiluquk .gt_from_md > :first-child { margin-top: 0; } #wiljiluquk .gt_from_md > :last-child { margin-bottom: 0; } #wiljiluquk .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #wiljiluquk .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #wiljiluquk .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #wiljiluquk .gt_row_group_first td { border-top-width: 2px; } #wiljiluquk .gt_row_group_first th { border-top-width: 2px; } #wiljiluquk .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #wiljiluquk .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #wiljiluquk .gt_first_summary_row.thick { border-top-width: 2px; } #wiljiluquk .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #wiljiluquk .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #wiljiluquk .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #wiljiluquk .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #wiljiluquk .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #wiljiluquk .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #wiljiluquk .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #wiljiluquk .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #wiljiluquk .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #wiljiluquk .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #wiljiluquk .gt_left { text-align: left; } #wiljiluquk .gt_center { text-align: center; } #wiljiluquk .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #wiljiluquk .gt_font_normal { font-weight: normal; } #wiljiluquk .gt_font_bold { font-weight: bold; } #wiljiluquk .gt_font_italic { font-style: italic; } #wiljiluquk .gt_super { font-size: 65%; } #wiljiluquk .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #wiljiluquk .gt_asterisk { font-size: 100%; vertical-align: 0; } #wiljiluquk .gt_indent_1 { text-indent: 5px; } #wiljiluquk .gt_indent_2 { text-indent: 10px; } #wiljiluquk .gt_indent_3 { text-indent: 15px; } #wiljiluquk .gt_indent_4 { text-indent: 20px; } #wiljiluquk .gt_indent_5 { text-indent: 25px; } #wiljiluquk .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #wiljiluquk div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } presentacion n pct bolsas 170 57 mixto 94 31 suelto 36 12 La mayor parte de las personas toma té en bolsas. Sin embargo, el tipo de té (en términos de precio o marca) que compran es muy distinto dependiendo de la presentación: tipo &lt;- tipo |&gt; select(presentacion, prop_presentacion = pct) tabla_cruzada &lt;- te |&gt; count(presentacion, precio) |&gt; # porcentajes por presentación group_by(presentacion) |&gt; mutate(prop = round(100 * n / sum(n))) |&gt; select(-n) tabla_cruzada |&gt; pivot_wider(names_from = presentacion, values_from = prop, values_fill = list(prop = 0)) |&gt; gt() #sxkvcchtyg table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #sxkvcchtyg thead, #sxkvcchtyg tbody, #sxkvcchtyg tfoot, #sxkvcchtyg tr, #sxkvcchtyg td, #sxkvcchtyg th { border-style: none; } #sxkvcchtyg p { margin: 0; padding: 0; } #sxkvcchtyg .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #sxkvcchtyg .gt_caption { padding-top: 4px; padding-bottom: 4px; } #sxkvcchtyg .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #sxkvcchtyg .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #sxkvcchtyg .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #sxkvcchtyg .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sxkvcchtyg .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #sxkvcchtyg .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #sxkvcchtyg .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #sxkvcchtyg .gt_column_spanner_outer:first-child { padding-left: 0; } #sxkvcchtyg .gt_column_spanner_outer:last-child { padding-right: 0; } #sxkvcchtyg .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #sxkvcchtyg .gt_spanner_row { border-bottom-style: hidden; } #sxkvcchtyg .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #sxkvcchtyg .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #sxkvcchtyg .gt_from_md > :first-child { margin-top: 0; } #sxkvcchtyg .gt_from_md > :last-child { margin-bottom: 0; } #sxkvcchtyg .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #sxkvcchtyg .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #sxkvcchtyg .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #sxkvcchtyg .gt_row_group_first td { border-top-width: 2px; } #sxkvcchtyg .gt_row_group_first th { border-top-width: 2px; } #sxkvcchtyg .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #sxkvcchtyg .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #sxkvcchtyg .gt_first_summary_row.thick { border-top-width: 2px; } #sxkvcchtyg .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sxkvcchtyg .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #sxkvcchtyg .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #sxkvcchtyg .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #sxkvcchtyg .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #sxkvcchtyg .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sxkvcchtyg .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #sxkvcchtyg .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #sxkvcchtyg .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #sxkvcchtyg .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #sxkvcchtyg .gt_left { text-align: left; } #sxkvcchtyg .gt_center { text-align: center; } #sxkvcchtyg .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #sxkvcchtyg .gt_font_normal { font-weight: normal; } #sxkvcchtyg .gt_font_bold { font-weight: bold; } #sxkvcchtyg .gt_font_italic { font-style: italic; } #sxkvcchtyg .gt_super { font-size: 65%; } #sxkvcchtyg .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #sxkvcchtyg .gt_asterisk { font-size: 100%; vertical-align: 0; } #sxkvcchtyg .gt_indent_1 { text-indent: 5px; } #sxkvcchtyg .gt_indent_2 { text-indent: 10px; } #sxkvcchtyg .gt_indent_3 { text-indent: 15px; } #sxkvcchtyg .gt_indent_4 { text-indent: 20px; } #sxkvcchtyg .gt_indent_5 { text-indent: 25px; } #sxkvcchtyg .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #sxkvcchtyg div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } precio bolsas mixto suelto marca 41 21 14 barato 3 1 3 marca_propia 9 4 3 desconocido 6 1 0 fino 8 20 56 variable 32 52 25 Estos datos podemos examinarlos un rato y llegar a conclusiones, pero esta tabla no necesariamente es la mejor manera de mostrar patrones en los datos. Tampoco son muy útiles gráficas como la siguiente: ggplot(tabla_cruzada |&gt; ungroup() |&gt; mutate(price = fct_reorder(precio, prop)), aes(x = precio, y = prop, group = presentacion, colour = presentacion)) + geom_point() + coord_flip() + geom_line() En lugar de eso, calcularemos perfiles columna. Esto es, comparamos cada una de las columnas con la columna marginal (en la tabla de tipo de estilo de té): num_grupos &lt;- n_distinct(te |&gt; select(presentacion)) tabla &lt;- te |&gt; count(presentacion, precio) |&gt; group_by(presentacion) |&gt; mutate(prop_precio = (100 * n / sum(n))) |&gt; group_by(precio) |&gt; mutate(prom_prop = sum(prop_precio)/num_grupos) |&gt; mutate(perfil = 100 * (prop_precio / prom_prop - 1)) tabla ## # A tibble: 17 × 6 ## # Groups: precio [6] ## presentacion precio n prop_precio prom_prop perfil ## &lt;fct&gt; &lt;fct&gt; &lt;int&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 bolsas marca 70 41.2 25.4 61.8 ## 2 bolsas barato 5 2.94 2.26 30.1 ## 3 bolsas marca_propia 16 9.41 5.48 71.7 ## 4 bolsas desconocido 11 6.47 2.51 158. ## 5 bolsas fino 14 8.24 28.0 -70.6 ## 6 bolsas variable 54 31.8 36.3 -12.5 ## 7 mixto marca 20 21.3 25.4 -16.4 ## 8 mixto barato 1 1.06 2.26 -52.9 ## 9 mixto marca_propia 4 4.26 5.48 -22.4 ## 10 mixto desconocido 1 1.06 2.51 -57.6 ## 11 mixto fino 19 20.2 28.0 -27.8 ## 12 mixto variable 49 52.1 36.3 43.6 ## 13 suelto marca 5 13.9 25.4 -45.4 ## 14 suelto barato 1 2.78 2.26 22.9 ## 15 suelto marca_propia 1 2.78 5.48 -49.3 ## 16 suelto fino 20 55.6 28.0 98.4 ## 17 suelto variable 9 25 36.3 -31.1 tabla_perfil &lt;- tabla |&gt; select(presentacion, precio, perfil, pct = prom_prop) |&gt; pivot_wider(names_from = presentacion, values_from = perfil, values_fill = list(perfil = -100.0)) if_profile &lt;- function(x){ any(x &lt; 0) &amp; any(x &gt; 0) } marcar &lt;- marcar_tabla_fun(25, &quot;red&quot;, &quot;black&quot;) tab_out &lt;- tabla_perfil |&gt; arrange(desc(bolsas)) |&gt; select(-pct, everything()) |&gt; mutate(across(where(is.numeric), \\(x) round(x, 0))) |&gt; mutate(across(where(if_profile), \\(x) marcar(x))) |&gt; knitr::kable(format_table_salida(), escape = FALSE, digits = 0, booktabs = T) |&gt; kableExtra::kable_styling(latex_options = c(&quot;striped&quot;, &quot;scale_down&quot;), bootstrap_options = c( &quot;hover&quot;, &quot;condensed&quot;), full_width = FALSE) tab_out precio bolsas mixto suelto pct desconocido 158 -58 -100 3 marca_propia 72 -22 -49 5 marca 62 -16 -45 25 barato 30 -53 23 2 variable -12 44 -31 36 fino -71 -28 98 28 Leemos esta tabla como sigue: por ejemplo, los compradores de té suelto compran té fino a una tasa casi el doble (98%) que el promedio. También podemos graficar como: tabla_graf &lt;- tabla_perfil |&gt; ungroup() |&gt; mutate(precio = fct_reorder(precio, bolsas)) |&gt; select(-pct) |&gt; pivot_longer(cols = -precio, names_to = &quot;presentacion&quot;, values_to = &quot;perfil&quot;) g_perfil &lt;- ggplot(tabla_graf, aes(x = precio, xend = precio, y = perfil, yend = 0, group = presentacion)) + geom_point() + geom_segment() + facet_wrap(~presentacion) + geom_hline(yintercept = 0 , colour = &quot;gray&quot;)+ coord_flip() g_perfil Observación: hay dos maneras de construir la columna promedio: tomando los porcentajes sobre todos los datos, o promediando los porcentajes de las columnas. Si los grupos de las columnas están desbalanceados, estos promedios son diferentes. Cuando usamos porcentajes sobre la población, perfiles columna y renglón dan el mismo resultado Sin embargo, cuando hay un grupo considerablemente más grande que otros, las comparaciones se vuelven vs este grupo particular. No siempre queremos hacer esto. Interpretación En el último ejemplo de tomadores de té utilizamos una muestra de personas, no toda la población de tomadores de té. Eso quiere decir que tenemos cierta incertidumbre de cómo se generalizan o no los resultados que obtuvimos en nuestro análisis a la población general. Nuestra respuesta depende de cómo se extrajo la muestra que estamos considerando. Si el mecanismo de extracción incluye algún proceso probabilístico, entonces es posible en principio entender qué tan bien generalizan los resultados de nuestro análisis a la población general, y entender esto depende de entender qué tanta variación hay de muestra a muestra, de todas las posibles muestras que pudimos haber extraido. En las siguientes secciones discutiremos estos aspectos, en los cuales pasamos del trabajo de “detective” al trabajo de “juez” en nuestro trabajo analítico. Suavizamiento loess Las gráficas de dispersión son la herramienta básica para describir la relación entre dos variables cuantitativas, y como vimos en ejemplo anteriores, muchas veces podemos apreciar mejor la relación entre ellas si agregamos una curva loess. Veamos un ejemplo, los siguientes datos muestran los premios ofrecidos y las ventas totales de una lotería a lo largo de 53 sorteos (las unidades son cantidades de dinero indexadas). Graficamos en escalas logarítmicas y agregamos una curva loess. # cargamos los datos load(here::here(&quot;data&quot;, &quot;ventas_sorteo.Rdata&quot;)) ggplot(ventas.sorteo, aes(x = premio, y = ventas.tot.1)) + geom_point() + geom_smooth(method = &quot;loess&quot;, span = 0.6, method.args = list(degree = 1), se = FALSE) + scale_x_log10(breaks = c(20000, 40000, 80000)) + scale_y_log10(breaks = c(10000, 15000, 22000, 33000)) El patrón no era difícil de ver en los datos originales, sin embargo, la curva lo hace más claro, el logaritmo de las ventas tiene una relación no lineal con el logaritmo del premio: para premios no muy grandes no parece haber gran diferencia, pero cuando los premios empiezan a crecer por encima de 20,000, las ventas crecen más rápidamente que para premios menores. Este efecto se conoce como bola de nieve, y es frecuente en este tipo de loterías. Antes de adentrarnos a los modelos loess comenzamos explicando cómo se ajustan familias paramétricas de curvas a conjuntos de datos dados. El modelo de regresion lineal ajusta una recta a un conjunto de datos. Por ejemplo, consideremos la familia \\[f_{a,b}(x) = a x + b,\\] para un conjunto de datos bivariados \\(\\{ (x_1, y_1), \\ldots, (x_N, y_N)\\}\\). Buscamos encontrar \\(a\\) y \\(b\\) tales que \\(f_{a,b}\\) de un ajuste óptimo a los datos. Para esto, se minimiza la suma de errores cuadráticos \\[\\frac1N \\sum_{n = 1}^N ( y_n - a x_n - b)^2.\\] En este caso, las constantes \\(a\\) y \\(b\\) se pueden encontrar diferenciando la función de mínimos cuadrados. Nótese que podemos repetir el argumento con otras familias de funciones (por ejemplo cuadráticas). ggplot(ventas.sorteo, aes(x = premio, y = ventas.tot.1)) + geom_point() + geom_smooth(method = &quot;lm&quot;, se = FALSE) + scale_x_log10(breaks = c(20000, 40000, 80000)) + scale_y_log10(breaks = c(10000, 15000, 22000, 33000)) Si observamos la gráfica notamos que este modelo lineal (en los logaritmos) no resumen adecuadamente estos datos. Podríamos experimentar con otras familias (por ejemplo, una cuadrática o cúbica, potencias, exponenciales, etc.); sin embargo, en la etapa exploratoria es mejor tomar una ruta de ajuste más flexible y robusta. Regresión local nos provee de un método con estas características: Curvas loess (regresión local): Una manera de mejorar la flexibilidad de los modelos lineales es considerar rectas de manera local. Es decir, en cada \\(x\\) posible consideramos cuál es la recta que mejor ajusta a los datos, considerando solamente valores de \\(x_n\\) que están cercanos a \\(x\\). La siguiente gráfica muestra qué recta se ajusta alrededor de cada punto, y cómo queda el suavizador completo, con distintos valores de suavizamiento. El tono de los puntos indican en cada paso que ventana de datos es considerada: Escogiendo de los parámetros. El parámetro de suavizamiento se encuentra por ensayo y error. La idea general es que debemos encontrar una curva que explique patrones importantes en los datos (que ajuste los datos) pero que no muestre variaciones a escalas más chicas difíciles de explicar (que pueden ser el resultado de influencias de otras variables, variación muestral, ruido o errores de redondeo, por ejemplo). En el proceso de prueba y error iteramos el ajuste y en cada paso hacemos análisis de residuales, con el fin de seleccionar un suavizamiento adecuado. En lugar de usar ajustes locales lineales, podemos usar ajustes locales cuadráticos que nos permiten capturar formas locales cuadráticas sin tener que suavizar demasiado poco: Opcional: cálculo del suavizador La idea es producir ajustes locales de rectas o funciones lineales o cuadráticas. Consideremos especificar dos parámetros: Parámetro de suavizamiento \\(\\alpha\\): toma valores en \\((0,1)\\), cuando \\(\\alpha\\) es más grande, la curva ajustada es más suave. Grado de los polinomios locales que ajustamos \\(\\lambda\\): generalmente se toma \\(\\lambda=1,2\\). Entonces, supongamos que los datos están dados por \\((x_1,y_1), \\ldots, (x_N, y_N)\\), y sean \\(\\alpha\\) un parámetro de suavizamiento fijo, y \\(\\lambda=1\\). Denotamos como \\(\\hat{g}(x)\\) la curva loess ajustada, y como \\(w_n(x)\\) a una función de peso (que depende de x) para la observación \\((x_n, y_n)\\). Para poder calcular \\(w_n(x)\\) debemos comenzar calculando \\(q=\\lfloor{N\\alpha}\\rfloor\\) que suponemos mayor que uno, esta \\(q\\) es el número de puntos que se utilizan en cada ajuste local. Ahora definimos la función tricubo: \\[\\begin{align} T(u)=\\begin{cases} (1-|u|^3)^3, &amp; \\text{para $|u| &lt; 1$}.\\\\ 0, &amp; \\text{en otro caso}. \\end{cases} \\end{align}\\] entonces, para el punto \\(x\\) definimos el peso correspondiente al dato \\((x_n,y_n)\\), denotado por \\(w_n(x)\\) como: \\[w_n(x)=T\\bigg(\\frac{|x-x_n|}{d_q(x)}\\bigg)\\] donde \\(d_q(x)\\) es el valor de la \\(q\\)-ésima distancia más chica (la más grande entre las \\(q\\) más chicas) entre los valores \\(|x-x_j|\\), \\(j=1,\\ldots,N\\). De esta forma, las observaciones \\(x_n\\) reciben más peso cuanto más cerca estén de \\(x\\). En palabras, de \\(x_1,...,x_N\\) tomamos los \\(q\\) datos más cercanos a \\(x\\), que denotamos \\(x_{i_1}(x) \\leq x_{i_2}(x) \\leq \\cdots \\leq x_{i_q}(x)\\). Los re-escalamos a \\([0,1]\\) haciendo corresponder \\(x\\) a \\(0\\) y el punto más alejado de \\(x\\) (que es \\(x_{i_q}\\)) a 1. Aplicamos el tricubo (gráfica de abajo), para encontrar los pesos de cada punto. Los puntos que están a una distancia mayor a \\(d_q(x)\\) reciben un peso de cero, y los más cercanos un peso que depende de que tan cercanos están a \\(x\\). Nótese que \\(x\\) es el punto ancla en dónde estamos ajustando la regresión local. tricubo &lt;- function(x) { ifelse(abs(x) &lt; 1, (1 - abs(x) ^ 3) ^ 3, 0) } curve(tricubo, from = -1.5, to = 1.5) Finalmente, para cada valor de \\(x_k\\) que está en el conjunto de datos \\(\\{x_1,...,x_n\\}\\), ajustamos una recta de mínimos cuadrados ponderados por los pesos \\(w_n(x)\\), es decir, minimizamos (en el caso lineal): \\[\\sum_{i=1}^nw_n(x_k)(y_i-ax_n-b)^2.\\] Observaciones: Cualquier función (continua y quizás diferenciable) con la forma de flan del tricubo que se desvanece fuera de \\((-1,1),\\) es creciente en \\((-1,0)\\) y decreciente en \\((0, 1)\\) es un buen candidato para usarse en lugar del tricubo. La razón por la que escogemos precisamente esta forma algebráica no tiene que ver con el análisis exploratorio, sino con las ventajas teóricas adicionales que tiene en la inferencia. El caso \\(\\lambda=2\\) es similar. La única diferencia es en el paso de ajuste, donde usamos funciones cuadráticas, y obtendríamos entonces tres familias de parámetros \\(a(x_k), b_1(x_k), b_2(x_k),\\) para cada \\(k \\in \\{1, \\ldots, N\\}\\). Caso de estudio: nacimientos en México Podemos usar el suavizamiento loess para entender y describir el comportamiento de series de tiempo, en las cuáles intentamos entender la dependencia de una serie de mediciones indexadas por el tiempo. Típicamente es necesario utilizar distintas componentes para describir exitosamente una serie de tiempo, y para esto usamos distintos tipos de suavizamientos. Veremos que distintas componentes varían en distintas escalas de tiempo (unas muy lentas, como la tendencia, otras más rapidamente, como variación quincenal, etc.). Este caso de estudio esta basado en un análisis propuesto por A. Vehtari y A. Gelman, junto con un análisis de serie de tiempo de William S. Cleveland (1993). En nuestro caso, usaremos los datos de nacimientos registrados por día en México desde 1999. Los usaremos para contestar las preguntas: ¿cuáles son los cumpleaños más frecuentes? y ¿en qué mes del año hay más nacimientos? Podríamos utilizar una gráfica popular (ver por ejemplo esta visualización) como: Sin embargo, ¿cómo criticarías este análisis desde el punto de vista de los tres primeros principios del diseño analítico? ¿Las comparaciones son útiles? ¿Hay aspectos multivariados? ¿Qué tan bien explica o sugiere estructura, mecanismos o causalidad? Datos de natalidad para México library(lubridate) library(ggthemes) theme_set(theme_minimal(base_size = 14)) natalidad &lt;- read_rds(&quot;./data/nacimientos/natalidad.rds&quot;) |&gt; mutate(dia_semana = weekdays(fecha)) |&gt; mutate(dia_año = yday(fecha)) |&gt; mutate(año = year(fecha)) |&gt; mutate(mes = month(fecha)) |&gt; ungroup() |&gt; mutate(dia_semana = recode(dia_semana, Monday = &quot;Lunes&quot;, Tuesday = &quot;Martes&quot;, Wednesday = &quot;Miércoles&quot;, Thursday = &quot;Jueves&quot;, Friday = &quot;Viernes&quot;, Saturday = &quot;Sábado&quot;, Sunday = &quot;Domingo&quot;)) |&gt; #necesario pues el LOCALE puede cambiar mutate(dia_semana = recode(dia_semana, lunes = &quot;Lunes&quot;, martes = &quot;Martes&quot;, miércoles = &quot;Miércoles&quot;, jueves = &quot;Jueves&quot;, viernes = &quot;Viernes&quot;, sábado = &quot;Sábado&quot;, domingo = &quot;Domingo&quot;)) |&gt; mutate(dia_semana = fct_relevel(dia_semana, c(&quot;Lunes&quot;, &quot;Martes&quot;, &quot;Miércoles&quot;, &quot;Jueves&quot;, &quot;Viernes&quot;, &quot;Sábado&quot;, &quot;Domingo&quot;))) Consideremos los datos agregados del número de nacimientos (registrados) por día desde 1999 hasta 2016. Un primer intento podría ser hacer una gráfica de la serie de tiempo. Sin embargo, vemos que no es muy útil: Hay varias características que notamos. Primero, parece haber una tendencia ligeramente decreciente del número de nacimientos a lo largo de los años. Segundo, la gráfica sugiere un patrón anual. Y por último, encontramos que hay dispersión producida por los días de la semana. Sólo estas características hacen que la comparación entre días sea difícil de realizar. Supongamos que comparamos el número de nacimientos de dos miércoles dados. Esa comparación será diferente dependiendo: del año donde ocurrieron, el mes donde ocurrieron, si semana santa ocurrió en algunos de los miércoles, y así sucesivamente. Como en nuestros ejemplos anteriores, la idea del siguiente análisis es aislar las componentes que observamos en la serie de tiempo: extraemos componentes ajustadas, y luego examinamos los residuales. En este caso particular, asumiremos una descomposición aditiva de la serie de tiempo (William S. Cleveland 1993). En el estudio de series de tiempo una estructura común es considerar el efecto de diversos factores como tendencia, estacionalidad, ciclicidad e irregularidades de manera aditiva. Esto es, consideramos la descomposición \\[\\begin{align} y(t) = f_{t}(t) + f_{e}(t) + f_{c}(t) + \\varepsilon. \\end{align}\\] Una estrategia de ajuste, como veremos más adelante, es proceder de manera modular. Es decir, se ajustan los componentes de manera secuencial considerando los residuales de los anteriores. Tendencia Comenzamos por extraer la tendencia, haciendo promedios loess (William S. Cleveland 1979) con vecindades relativamente grandes. Quizá preferiríamos suavizar menos para capturar más variación lenta, pero si hacemos esto en este punto empezamos a absorber parte de la componente anual: mod_1 &lt;- loess(n ~ as.numeric(fecha), data = natalidad, span = 0.2, degree = 1) datos_dia &lt;- natalidad |&gt; mutate(ajuste_1 = fitted(mod_1)) |&gt; mutate(res_1 = n - ajuste_1) Notemos que a principios de 2000 el suavizador está en niveles de alrededor de 7000 nacimientos diarios, hacia 2015 ese número es más cercano a unos 6000. Componente anual Al obtener la tendencia podemos aislar el efecto a largo plazo y proceder a realizar mejores comparaciones (por ejemplo, comparar un día de 2000 y de 2015 tendria más sentido). Ahora, ajustamos los residuales del suavizado anterior, pero con menos suavizamiento. Así evitamos capturar tendencia: mod_anual &lt;- loess(res_1 ~ as.numeric(fecha), data = datos_dia, degree = 2, span = 0.005) datos_dia &lt;- datos_dia |&gt; mutate(ajuste_2 = fitted(mod_anual)) |&gt; mutate(res_2 = res_1 - ajuste_2) Día de la semana Hasta ahora, hemos aislado los efectos por plazos largos de tiempo (tendencia) y hemos incorporado las variaciones estacionales (componente anual) de nuestra serie de tiempo. Ahora, veremos cómo capturar el efecto por día de la semana. En este caso, podemos hacer suavizamiento loess para cada serie de manera independiente datos_dia &lt;- datos_dia |&gt; group_by(dia_semana) |&gt; nest() |&gt; mutate(ajuste_mod = map(data, ~ loess(res_2 ~ as.numeric(fecha), data = .x, span = 0.1, degree = 1))) |&gt; mutate(ajuste_3 = map(ajuste_mod, fitted)) |&gt; select(-ajuste_mod) |&gt; unnest(cols = c(data, ajuste_3)) |&gt; mutate(res_3 = res_2 - ajuste_3) |&gt; ungroup() Residuales Por último, examinamos los residuales finales quitando los efectos ajustados: ## `geom_smooth()` using formula = &#39;y ~ x&#39; Observación: nótese que la distribución de estos residuales presenta irregularidades interesantes. La distribución es de colas largas, y no se debe a unos cuantos datos atípicos. Esto generalmente es indicación que hay factores importantes que hay que examinar mas a detalle en los residuales: Reestimación Cuando hacemos este proceso secuencial de llevar el ajuste a los residual, a veces conviene iterarlo. La razón es que en una segunda o tercera pasada podemos hacer mejores estimaciones de cada componente, y es posible suavizar menos sin capturar componentes de más alta frecuencia. Así que podemos regresar a la serie original para hacer mejores estimaciones, más suavizadas: # Quitamos componente anual y efecto de día de la semana datos_dia &lt;- datos_dia |&gt; mutate(n_1 = n - ajuste_2 - ajuste_3) # Reajustamos mod_1 &lt;- loess(n_1 ~ as.numeric(fecha), data = datos_dia, span = 0.02, degree = 2, family = &quot;symmetric&quot;) Y ahora repetimos con la componente de día de la semana: Análisis de componentes Ahora comparamos las componentes estimadas y los residuales en una misma gráfica. Por definición, la suma de todas estas componentes da los datos originales. Este último paso nos permite diversas comparaciones que explican la variación que vimos en los datos. Una gran parte de los residuales está entre \\(\\pm 250\\) nacimientos por día. Sin embargo, vemos que las colas tienen una dispersión mucho mayor: quantile(datos_dia$res_6, c(00, .01,0.05, 0.10, 0.90, 0.95, 0.99, 1)) |&gt; round() ## 0% 1% 5% 10% 90% 95% 99% 100% ## -2238 -1134 -315 -202 188 268 516 2521 ¿A qué se deben estas colas tan largas? Viernes 13? Podemos empezar con una curosidad. Los días Viernes o Martes 13, ¿nacen menos niños? Nótese que fue útil agregar el indicador de Semana santa por el Viernes 13 de Semana Santa que se ve como un atípico en el panel de los viernes 13. Residuales: antes y después de 2006 Veamos primero una agregación sobre los años de los residuales. Lo primero es observar un cambio que sucedió repentinamente en 2006: La razón es un cambio en la ley acerca de cuándo pueden entrar los niños a la primaria. Antes era por edad y había poco margen. Ese exceso de nacimientos son reportes falsos para que los niños no tuvieran que esperar un año completo por haber nacido unos cuantos días después de la fecha límite. Otras características que debemos investigar: Efectos de Año Nuevo, Navidad, Septiembre 16 y otros días feriados como Febrero 14. Semana santa: como la fecha cambia, vemos que los residuales negativos tienden a ocurrir dispersos alrededor del día 100 del año. Otros días especiales: más de residuales Ahora promediamos residuales (es posible agregar barras para indicar dispersión a lo largo de los años) para cada día del año. Podemos identificar ahora los residuales más grandes: se deben, por ejemplo, a días feriados, con consecuencias adicionales que tienen en días ajuntos (excesos de nacimientos): Semana santa Para Semana Santa tenemos que hacer unos cálculos. Si alineamos los datos por días antes de Domingo de Pascua, obtenemos un patrón de caída fuerte de nacimientos el Viernes de Semana Santa, y la característica forma de “valle con hombros” en días anteriores y posteriores estos Viernes. ¿Por qué ocurre este patrón? Nótese un defecto de nuestro modelo: el patrón de “hombros” alrededor del Viernes Santo no es suficientemente fuerte para equilibrar los nacimientos faltantes. ¿Cómo podríamos mejorar nuestra descomposición? Referencias "],["tipos-de-estudio-y-experimentos.html", "Sección 2 Tipos de estudio y experimentos Muestreo aleatorio Pero si no podemos hacer muestreo aleatorio? El estimador estándar Experimentos tradicionales Bloqueo Variables desconocidas Aleatorizando el tratamiento Selección de unidades y tratamiento Asignación natural del tratamiento", " Sección 2 Tipos de estudio y experimentos Motivación Pregunta de entrevista de Google (Chihara and Hesterberg 2018) Imagina que eres consultor y te preguntan lo siguiente (ver siguiente figura): Estoy haciendo una comparación de antes y después donde la hipótesis alternativa es pre.media.error &gt; post.media.error. La distribución de ambas muestras es sesgada a la derecha. ¿Qué prueba me recomiendas para ésta situación? Figure 2.1: Error CPR, gráfica de densidad. Far better an approximate answer to the right question, which is often vague, than an exact answer to the wrong question, which can always be made precise. — John Tukey La siguiente imagen Roger Peng representa una situación común a la que se enfrenta el analista de datos, y se desarrolló en el contexto de preguntas vagas. En el esquema hay tres caminos: uno es uno ideal que pocas veces sucede, otro produce respuestas poco útiles pero es fácil, y otro es tortuoso pero que caracteriza el mejor trabajo de análisis de datos: Figure 2.2: Adaptado de R. Peng: Tukey, design thinking and better questions. Ejemplos: Alguien nos pregunta cuáles son las tiendas que mas venden de una cadena. Podríamos consultar bases de datos, hacer extracciones, definir periodos, etc. y reportar el promedio de ventas en el último mes, esta respuesta probablemente es poco útil. Nos damos cuenta, por ejemplo, porque la peor tienda es una que abrió hace relativamente poco, y la mejor es una de las tiendas más grandes que está en una zona de tráfico de alto costo. Una pregunta más interesante es, ¿qué equipos de ventas tienen mejor desempeño? ¿Cuánto aporta tener una cafetería dentro de la tienda en términos de ventas?, etc. Proceso Generador de Datos Entre las preguntas que se debe hacer el analista de datos una fundamental es entender el proceso generador de datos, pues esto determinará que otras preguntas son relevantes, y que análisis son adecuados, tanto en términos prácticos como estadísticos. La inferencia estadística busca hacer afirmaciones, cuantificadas de manera probabilista, acerca de datos que no tenemos, usando regularidades y conocimiento de datos que sí tenemos disponibles y métodos cuantitativos. Para hacer afirmaciones inferenciales eficientes y bien calibradas (con garantías estadísticas de calibración) a preguntas donde queremos generalizar de muestra a población, se requiere conocer con precisión el proceso que genera los datos muestrales. Esto incluye saber con detalle cómo se seleccionaron los datos a partir de los que se quiere hacer inferencia. En este caso, eficiente quiere decir que aprovechamos toda la información que está en los datos observados de manera que nuestros rangos de incertidumbre son lo más chico posibles (además de estar correctamente calibrados). Por su parte, probabilísticamente bien calibrados se refiere a que, lo que decimos que puede ocurrir con 10% de probabilidad ocurre efectivamente 1 de cada 10 veces, si decimos 20% entonces ocurre 2 de 20, etc. Veremos que para muestras dadas naturalmente, a veces es muy difiícil entender a fondo el proceso que generó la muestra y por tanto no tenemos las garantías de eficiencia y calibración. Ejemplo: Prevalencia de anemia Supongamos que nos interesa conocer el porcentaje de menores en edad escolar, (entre 6 y 15 años), con anemia en México. La fuente de datos disponible corresponde a registros del IMSS de hospitalizaciones de menores, ya sea por anemia o por otra causa (infecciones gastrointestinales, apendicitis, tratamiento de leucemia, …), se registró si el menor tenía anemia. En nuestra muestra el 47% de los niños tiene anemia. head(paciente) #&gt; # A tibble: 6 × 4 #&gt; edad padecimiento sexo anemia #&gt; &lt;int&gt; &lt;chr&gt; &lt;chr&gt; &lt;int&gt; #&gt; 1 8 picadura alacrán mujer 0 #&gt; 2 10 infección intestinal hombre 1 #&gt; 3 7 mordedura de perro hombre 1 #&gt; 4 8 asma hombre 1 #&gt; 5 13 infección intestinal mujer 0 #&gt; 6 7 picadura alacrán hombre 0 ¿Qué nos dice esta cantidad acerca de la anemia en la población? ¿Podemos hacer inferencia estadística? ¿Cómo calculamos intervalos de confianza? # Si calculo el error estándar de la p estimada como sigue, es correcto? p &lt;- mean(paciente$anemia) sqrt(p * (1 - p) / 5000) #&gt; [1] 0.007060751 Muestreo aleatorio En la situación ideal diseñaríamos una muestra aleatoria de menores de edad, por ejemplo, utilizando el registro en educación primaria de la SEP, y mediríamos la prevalencia de anemia en la muestra, usaríamos esta muestra para estimar la prevalencia en la población y tendríamos además las herramientas para medir la incertidumbre de nuestra estimación (reportar intervalos, o errores estándar). El elemento clave, es la aleatorización en la selección de la muestra, la idea es distribuir los efecros desconcidos o no controlables que pueden introducir sesgos o variabilidad no conocida en los resultados. Pero si no podemos hacer muestreo aleatorio? En el caso de prevalencia de anemia, discutiendo con médicos e investigadores nos informan que la anemia se presenta en tasas más altas en niños más chicos. paciente |&gt; count(edad) |&gt; mutate(prop = round(100 * n / sum(n))) #&gt; # A tibble: 10 × 3 #&gt; edad n prop #&gt; &lt;int&gt; &lt;int&gt; &lt;dbl&gt; #&gt; 1 6 1001 20 #&gt; 2 7 931 19 #&gt; 3 8 980 20 #&gt; 4 9 445 9 #&gt; 5 10 484 10 #&gt; 6 11 489 10 #&gt; 7 12 246 5 #&gt; 8 13 239 5 #&gt; 9 14 90 2 #&gt; 10 15 95 2 Y consultando con las proyecciones de población notamos que los niños chicos están sobrerepresentados en la muestra. Lo que nos hace considerar que debemos buscar una manera de ponderar nuestras observaciones para que reflejen a la población. Más aún, investigamos que algunas enfermedades están asociadas a mayor prevalencia de anemia: paciente |&gt; count(padecimiento) |&gt; arrange(-n) #&gt; # A tibble: 7 × 2 #&gt; padecimiento n #&gt; &lt;chr&gt; &lt;int&gt; #&gt; 1 infección respiratoria 745 #&gt; 2 mordedura de perro 723 #&gt; 3 úlcera 723 #&gt; 4 asma 713 #&gt; 5 apendcitis 704 #&gt; 6 picadura alacrán 701 #&gt; 7 infección intestinal 691 Utilizamos esta información para modelar y corregir nuestra estimación original. Por ejemplo con modelos de regresión. Sin embargo, debemos preguntarnos: ¿Hay más variables qué nos falta considerar? Ejemplo: Policías y tráfico Supongamos que nos preguntan en cuánto reduce un policía el tráfico en un crucero grande de la ciudad. La cultura popular ha establecido que los policías en cruceros hacen más tráfico porque no saben mover los semáforos. Nosotros decidimos buscar unos datos para entender esto. Escogemos entonces un grupo de cruceros problemáticos, registramos el tráfico cuando visitamos, y si había un policía o no. Después de este esfuerzo, obtenemos los siguientes datos: #&gt; # A tibble: 10 × 2 #&gt; # Groups: policia [2] #&gt; policia tiempo_espera_min #&gt; &lt;int&gt; &lt;dbl&gt; #&gt; 1 0 2.27 #&gt; 2 0 2.65 #&gt; 3 0 3.4 #&gt; 4 0 0.39 #&gt; 5 0 1.1 #&gt; 6 1 10.8 #&gt; 7 1 4.67 #&gt; 8 1 7.77 #&gt; 9 1 6.3 #&gt; 10 1 6.99 Lo que sabemos ahora es que la presencia de un policía es indicador de tráfico alto. El análisis prosiguiría calculando medias y medidas de error (escogimos una muestra aleatoria): Si somos ingenuos, entonces podríamos concluir que los policías efectivamente empeoran la situación cuando manipulan los semáforos, y confirmaríamos la sabiduría popular. Para juzgar este argumento desde el punto de vista causal, nos preguntamos primero: ¿Cuáles son los contrafactuales (los contrafactuales explican que pasaría si hubiéramos hecho otra cosa que la que efectivamente hicimos) de las observaciones? Efectos causales y el esquema de resultados potenciales Consideramos un tratamiento binario: Se manda policía o no se manda policía. Un resultado potencial es aquél que se observaría bajo un tratamiento particular. En cada semáforo, a una hora dada, hay dos resultados potenciales, uno por cada valor del tratamiento: \\(y_1:\\) tiempo de espera si se envía policía. \\(y_0:\\) tiempo de espera si no se envía policía. Para cada semáforo, en el momento de registro, uno observa únicamente uno de los dos resultados potenciales. El resultado no observado se conoce como resultado contrafactual. El estimador estándar A la comparación anterior - la diferencia de medias de tratados y no tratados - le llamamos usualmente el estimador estándar del efecto causal. Muchas veces este es un estimador malo del efecto causal. En nuestro ejemplo, para llegar a la conclusión errónea que confirma la sabiduría popular, hicimos un supuesto importante: En nuestra muestra, los casos con policía actúan como contrafactuales de los casos sin policía. Asi que asumimos que los casos con policía y sin policía son similares, excepto por la existencia o no de policía. En nuestro ejemplo, quizá un analista más astuto nota que tienen categorías históricas de qué tan complicado es cada crucero. Junta a sus datos, y obtiene: #&gt; # A tibble: 10 × 3 #&gt; # Groups: policia [2] #&gt; policia tiempo_espera_min categoria #&gt; &lt;int&gt; &lt;dbl&gt; &lt;fct&gt; #&gt; 1 0 2.27 Fluido #&gt; 2 0 2.65 Fluido #&gt; 3 0 3.4 Típico #&gt; 4 0 0.39 Fluido #&gt; 5 0 1.1 Fluido #&gt; 6 1 10.8 Complicado #&gt; 7 1 4.67 Típico #&gt; 8 1 7.77 Complicado #&gt; 9 1 6.3 Complicado #&gt; 10 1 6.99 Típico El analista argumenta entonces que los policías se enviaron principalmente a cruceros que se consideran Complicados según datos históricos. Esto resta credibilidad a la comparación que hicimos inicialmente: La comparación del estimador estándar no es de peras con peras: estamos comparando qué efecto tienen los policías en cruceros difíciles, con cruceros no difíciles donde no hay policía. La razón de esto es que el proceso generador de los datos incluye el hecho de que no se envían policías a lugares donde no hay tráfico. ¿Cómo producir contrafactuales para hacer la comparación correcta? Experimentos tradicionales Idealmente, quisiéramos observar un mismo crucero en las dos condiciones: con y sin policías. Esto no es posible. En un experimento “tradicional”, como nos lo explicaron en la escuela, nos aproximamos a esto preparando dos condiciones idénticas, y luego alteramos cada una de ellas con nuestra intervención. Si el experimento está bien hecho, esto nos da observaciones en pares, y cada quien tiene su contrafactual. La idea del experimiento tradicional es controlar todos los factores que intervienen en los resultados, y sólo mover el tratamiento para producir los contrafactuales. Más en general, esta estrategia consiste en hacer bloques de condiciones, donde las condiciones son prácticamente idénticas dentro de cada bloque. Comparamos entonces unidades tratadas y no tratadas dentro de cada bloque. Por ejemplo, si queremos saber si el tiempo de caída libre es diferente para un objeto más pesado que otro, prepararíamos dos pesos con el mismo tamaño pero de peso distinto. Soltaríamos los dos al mismo tiempo y compararíamos el tiempo de caída de cada uno. En nuestro caso, como es usual en problemas de negocio o sociales, hacer esto es considerablemente más difícil. No podemos “preparar” cruceros con condiciones idénticas. Sin embargo, podríamos intentar \\(bloquear\\) los cruceros según información que tenemos acerca de ellos, para hacer más comparaciones de peras con peras. Bloqueo Podemos acercanos en lo posible a este ideal de experimentación usando información existente. En lugar de hacer comparaciones directas entre unidades que recibieron el tratamiento y las que no (que pueden ser diferentes en otros aspectos, como vimos arriba), podemos refinar nuestras comparaciones bloquéandolas con variables conocidas. En el ejemplo de los policías, podemos hacer lo siguiente: dentro de cada categoría de cruceros (fluido, típico o complicado), tomaremos una muestra de cruceros, algunos con policía y otros sin. Haremos comparaciones dentro de cada categoría. Obtenemos una muestra con estas características (6 casos en cada categoría de crucero, 3 con policía y 3 sin policía): categoria policia n Fluido 0 3 Fluido 1 3 Típico 0 3 Típico 1 3 Complicado 0 3 Complicado 1 3 Y ahora hacemos comparaciones dentro de cada bloque creado por categoría: #&gt; # A tibble: 3 × 3 #&gt; # Groups: categoria [3] #&gt; categoria `policia =0` `policia =1` #&gt; &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt; #&gt; 1 Fluido 2.1 0.8 #&gt; 2 Típico 5.6 4.2 #&gt; 3 Complicado 10.4 8.6 Y empezamos a ver otra imagen en estos datos: comparando tipos e cruceros similares, los que tienen policía tienen tiempos de espera ligeramente más cortos. ¿Hemos termniado? ¿Podemos concluir que el efecto de un policía es beneficiosos pero considerablemente chico? ¿Qué problemas puede haber con este análisis? Variables desconocidas El problema con el análisis anterior es que controlamos por una variable que conocemos, pero muchas otras variables pueden estar ligadas con el proceso de selección de cruceros para enviar policías. Por ejemplo, envían o policías a cruceros Típicos solo cuando reportan mucho tráfico. No envían a un polícia a un crucero Complicado si no presenta demasiado tráfico. Existen otras variables desconocidas que los tomadores de decisiones usan para enviar a los policías. En este caso, por ejemplo, los expertos hipotéticos nos señalan que hay algunos cruceros que aunque problemáticos, a veces su tráfico se resuelve rápidamente, mientras que otros tienen tráfico más persistente, y prefieren enviar policías a los de tráfico persistente. La lista de cruceros persistentes están en una hoja de excel que se comparte de manera informal. En resumen, no tenemos conocimiento detallado del proceso generador de datos en cuanto a cómo se asignan los policías a los cruceros. Igual que en la sección anterior, podemos cortar esta complejidad usando aleatorización. Nótese que los expertos no están haciendo nada malo: en su trabajo están haciendo el mejor uso de los recursos que tienen. El problema es que por esa misma razón no podemos saber el resultado de sus esfuerzos, y si hay maneras de optimizar la asignación que hacen actualmente. Aleatorizando el tratamiento Tomamos la decisión entonces de hacer un experimento que incluya aletorización. En un dia particular, escogeremos algunos cruceros. Dicidimos usar solamente cruceros de la categoría Complicada y Típica, pues esos son los más interesantes para hacer intervenciones. Usaremos un poco de código para entener el detalle: en estos datos, tenemos para cada caso los dos posibles resultados hipotéticos \\(y_0\\) y \\(y_1\\) (con policia y sin policia). En el experimento asignamos el tratamiento al azar: muestra_exp &lt;- trafico_tbl |&gt; filter(categoria != &quot;Fluido&quot;) |&gt; sample_n(200) |&gt; # asignar tratamiento al azar, esta es nuestra intervención: mutate(tratamiento_policia = rbernoulli(length(y_0), 0.5)) |&gt; # observar resultado mutate(tiempo_espera_exp = ifelse(tratamiento_policia == 1, y_1, y_0)) Nótese la diferencia si tomamos la asignación natural del tratamiento (policía o no): set.seed(134) muestra_natural &lt;- trafico_tbl |&gt; filter(categoria != &quot;Fluido&quot;) |&gt; sample_n(200) |&gt; # usamos el tratamiento que se asignó # policia indica si hubo o no policía en ese crucero # observar resultado mutate(tiempo_espera_obs = ifelse(policia == 1, y_1, y_0)) Resumimos nuestros resultados del experimento son: #&gt; # A tibble: 2 × 3 #&gt; # Groups: categoria [2] #&gt; categoria `policia=0` `policia=1` #&gt; &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt; #&gt; 1 Típico 6.24 4.97 #&gt; 2 Complicado 15.8 8.47 Sin embargo, la muestra natural da: #&gt; # A tibble: 2 × 3 #&gt; # Groups: categoria [2] #&gt; categoria `policia=0` `policia=1` #&gt; &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt; #&gt; 1 Típico 5.49 4.35 #&gt; 2 Complicado 10.8 8.93 ¿Cuál de los dos análisis da la respuesta correcta a la pregunta: ayudan o no los policías a reducir el tráfico en los cruceros problemáticos? El experimento establece que un policía en promedio reduce a la mitad el tiempo de espera en un crucero complicado Selección de unidades y tratamiento Vimos dos tipos de inferencia que requieren distintos diseños de estudio: a poblaciones (ejemplo anemia) y causal (ejemplo policías). En el escenario ideal de cada uno de estos ejemplos requerimos un mecanismo de aleatorización, sin embargo, la aleatorización requerida en cada caso es distinta y distinguir esto es fundamental para entender las inferencias que podemos hacer en distintos escenarios. Inferencia estadística de acuerdo al tipo del diseño (Ramsey and Schafer (2012)). El cuadro arriba a la izquierda es donde el análisis es más simple y los resultados son más fáciles de interpretar. En este escenario don de la aleatorización es tanto en unidades como en grupos no hacen falta supuestos adicionales para tener las garantías de métodos de inferencia. Es posible hacer análisis fuera de este cuadro, pero el proceso es más complicado, requieren más supuestos, conocimiento del dominio y habilidades de análisis. En general resultan conclusiones menos sólidas. Muchas veces no nos queda otra más que trabajar fuera del cuadro ideal. El punto crucial para entender las medidas de incertidumbre estadística es visualizar de manera hipotética, replicaciones del estudio y las condiciones que llevaron a la selección de la muestra. Esto es, entender el proceso generador de datos e imaginar replicarlo. Ubica los siguientes tipos de análisis: Pruebas clínicas para medicinas Analizar cómo afecta tener seguro médico a los ingresos, usando datos del ENIGH. Estimación de retorno sobre inversión en modelos de marketing mix. Asignación natural del tratamiento Cuando consideramos un sistema donde se “asignan” tratamientos de manera natural, generalmente los tratamientos se asignan bajo un criterio de optimización o conveniencia (por ejemplo los policías a cruceros problemáticos). La cara buena de este hecho es que de alguna forma los resultados están intentando optimizarse, y la gente está haciendo su trabajo. La cara mala de este hecho es que no podemos evaluar de manera simple la efectividad de los tratamientos. Y esto hace difícil optimizar de forma cuantificable los procesos, o entender qué funciona y qué no. Referencias "],["pruebas-de-hipótesis.html", "Sección 3 Pruebas de hipótesis Comparación con poblaciones de referencia Comparando distribuciones Prueba de permutaciones y el lineup Comparaciones usando lineup (continuación) Prueba de permutaciones para proporciones Pruebas de hipótesis tradicionales Tomadores de té (continuación) Pruebas de permutación: implementación. Ejemplo: tiempos de fusión Ejemplo: tiempos de fusión (continuación) Separación de grupos La “crisis de replicabilidad” El jardín de los senderos que se bifurcan Ejemplo: decisiones de análisis y valores p Alternativas o soluciones", " Sección 3 Pruebas de hipótesis Las primeras técnicas inferenciales que veremos intentan contestar la siguiente pregunta: Si observamos cierto patrón en los datos, ¿cómo podemos cuantificar la evidencia de que es un patrón notable y no sólo debido a fluctuaciones en los datos particulares que tenemos? ¿Cómo sabemos que no estamos sobreinterpretando esas fluctuaciones? Por ejemplo: Un sistema tiene cierto comportamiento “usual” para el cual tenemos datos históricos. El sistema presenta fluctuaciones en el tiempo. Observamos la última salida de nuestro sistema. Naturalmente, tiene fluctuaciones. ¿Esas fluctuaciones son consistentes con la operación usual del sistema? ¿Existe evidencia para pensar que algo en el sistema cambió? Comparación con poblaciones de referencia En las prueba de hipótesis, tratamos de construir distribuciones de referencia para comparar resultados que obtengamos con un “estándar” de variación, y juzgar si nuestros resultados son consistentes con la referencia o no (Box et al. 1978). En algunos casos, ese estándar de variación puede construirse con datos históricos. Ejemplo Supongamos que estamos considerando cambios rápidos en una serie de tiempo de alta frecuencia. Hemos observado la serie en su estado “normal” durante un tiempo considerable, y cuando observamos nuevos datos quisiéramos juzgar si hay indicaciones o evidencia en contra de que el sistema sigue funcionando de manera similar. Digamos que monitoreamos ventanas de tiempo de tamaño 20 y necesitamos tomar una decisión. Abajo mostramos cinco ejemplos donde el sistema opera normalmente, que muestra la variabilidad en el tiempo en ventanas cortas del sistema. Ahora suponemos que obtenemos una nueva ventana de datos. ¿Hay evidencia en contra de que el sistema sigue funcionando de manera similar? Nuestra primera inclinación debe ser comparar: en este caso, compararamos ventanas históricas con nuestra nueva serie: # usamos datos simulados para este ejemplo set.seed(8812) historicos &lt;- simular_serie(2000) ¿Vemos algo diferente en los datos nuevos (el panel de color diferente)? Indpendientemente de la respuesta, vemos que hacer este análisis de manera tan simple no es siempre útil: seguramente podemos encontrar maneras en que la nueva muestra (4) es diferente a muestras históricas. Por ejemplo, ninguna de muestras tiene un “forma de montaña” tan clara. Nos preguntamos si no estamos sobreinterpretando variaciones que son parte normal del proceso. Podemos hacer un mejor análisis si extraemos varias muestras del comportamiento usual del sistema, graficamos junto a la nueva muestra, y revolvemos las gráficas para que no sepamos cuál es cuál. Entonces la pregunta es: ¿Podemos detectar donde están los datos nuevos? Esta se llama una prueba de lineup, o una prueba de ronda de sospechosos (Hadley Wickham et al. 2010). En la siguiente gráfica, en uno de los páneles están los datos recientemente observados. ¿Hay algo en los datos que distinga al patrón nuevo? # nuevos datos obs &lt;- simular_serie(500, x_inicial = last(obs$obs)) # muestrear datos históricos prueba_tbl &lt;- muestrear_ventanas(historicos, obs[1:20, ], n_ventana = 20) # gráfica de pequeños múltiplos ggplot(prueba_tbl$lineup, aes(x = t_0, y = obs)) + geom_line() + facet_wrap(~rep, nrow = 4) + scale_y_log10() ¿Cuáles son los datos nuevos (solo hay un panel con los nuevos datos)? ¿Qué implica que la gráfica que escojamos como “más diferente” no sean los datos nuevos? ¿Qué implica que le “atinemos” a la gráfica de los datos nuevos? Ahora observamos al sistema en otro momento y repetimos la comparación. En el siguiente caso obtenemos: Aunque es imposible estar seguros de que ha ocurrido un cambio, la diferencia de una de las series es muy considerable. Si identificamos los datos correctos, la probabilidad de que hayamos señalado la nueva serie “sobreinterpretando” fluctuaciones en un proceso que sigue comportándose normalente es 0.05 - relativamente baja. Detectar los datos diferentes es evidencia en contra de que el sistema sigue funcionando de la misma manera que antes. En el ejemplo anterior se encontraban en la posición: prueba_tbl$pos ## [1] 18 Observaciones y terminología: Llamamos hipótesis nula a la hipótesis de que los nuevos datos son producidos bajo las mismas condiciones que los datos de control o de referencia. Si no escogemos la gráfica de los nuevos datos, nuestra conclusión es que la prueba no aporta evidencia en contra de la hipótesis nula. Si escogemos la gráfica correcta, nuestra conclusión es que la prueba aporta evidencia en contra de la hipótesis nula. ¿Qué tan fuerte es la evidencia, en caso de que descubrimos los datos no nulos? Cuando el número de paneles es más grande y detectamos los datos, la evidencia es más alta en contra de la nula. Decimos que el nivel de significancia de la prueba es la probabilidad de seleccionar a los datos correctos cuando la hipótesis nula es cierta (el sistema no ha cambiado). En el caso de 20 paneles, la significancia es de 1/20 = 0.05. Cuando detectamos los datos nuevos, niveles de significancia más bajos implican más evidencia en contra de la nula. Si acertamos, y la diferencia es más notoria y fue muy fácil detectar la gráfica diferente (pues sus diferencias son más extremas), esto también sugiere más evidencia en contra de la hipótesis nula. Finalmente, esta prueba rara vez (o nunca) nos da seguridad completa acerca de ninguna conclusión, aún cuando hiciéramos muchos páneles. Comparando distribuciones Ahora intentamos un ejemplo más típico. Supongamos que tenemos muestras para tres grupos a, b y c, esto es que dentro de cada grupo, el proceso de selección de los elementos se hace al azar y de manera simétrica (por ejemplo cada elemento tiene a misma probabiidad de ser seleccionado, y las extracciones se hacen de manera independiente.) Queremos comparar las distribuciones de los datos obtenidos para cada grupo. Quizá la pregunta detrás de esta comparación es: el grupo de clientes b recibió una promoción especial. ¿Están gastando más? La medición que comparamos es el gasto de los clientes. En la muestra observamos diferencias entre los grupos. Pero notamos adicionalmente que hay mucha variación dentro de cada grupo. Nos podríamos preguntar entonces si las diferencias que observamos se deben variación muestral, por ejemplo. Podemos construir ahora una hipótesis nula, que establece que las observaciones provienen de una población similar: Las tres poblaciones (a, b, c) son prácticamente indistiguibles. En este caso, la variación que observamos se debería a que tenemos información incompleta. Como en el ejemplo anterior necesitamos construir u obtener una distribución de referencia para comparar qué tan extremos o diferentes son los datos que observamos. Esa distribución de referencia debería estar basada en el supuesto de que los grupos producen datos de distribuciones similares. Si tuvieramos mediciones similares históricas de estos tres grupos, quizá podríamos extraer datos de referencia y comparar, como hicimos en el ejempo anterior. Pero esto es menos común en este tipo de ejemplos. Prueba de permutaciones y el lineup Para abordar este problema podemos pensar en usar permutaciones de los grupos de la siguiente forma ((Box et al. 1978), (Tim C. Hesterberg 2015a)): Si los grupos producen datos bajo procesos idénticos, entonces los grupos a, b, c solo son etiquetas que no contienen información. Podríamos permutar al azar las etiquetas y observar nuevamente la gráfica de caja y brazos por grupos. Si la hipótesis nula es cierta (grupos idénticos), esta es una muestra tan verosímil como la que obtuvimos. Así que podemos construir datos de referencia permutando las etiquetas de los grupos al azar, y observando la variación que ocurre. Si la hipótesis nula es cercana a ser cierta, no deberíamos de poder distinguir fácilmente los datos observados de los producidos con las permutaciones al azar. Vamos a intentar esto, por ejemplo usando una gráfica de cuantiles simplificada. Hacemos un lineup, o una rueda de sospechosos (usamos el paquete (H. Wickham, Chowdhury, and Cook 2012), ver (Hadley Wickham et al. 2010)), donde 19 de los acusados son generados mediante permutaciones al azar de la variable del grupo, y el culpable (los verdaderos datos) están en una posición escogida al azar. ¿Podemos identificar los datos verdaderos? Para evitar sesgarnos, también ocultamos la etiqueta verdadera. Usamos una gráfica que muestra los cuantiles 0.10, 0.50, 0.90: set.seed(88) reps &lt;- lineup(null_permute(&quot;grupo&quot;), muestra_tab, n = 20) ## decrypt(&quot;M7xA 2S8S Jj dUyJ8JUj ZW&quot;) reps_mezcla &lt;- reps |&gt; mutate(grupo_1 = factor(digest::digest2int(grupo) %% 177)) grafica_cuantiles(reps_mezcla, grupo_1, x) + coord_flip() + facet_wrap(~.sample, ncol = 5) + ylab(&quot;x&quot;) + labs(caption = &quot;Mediana y percentiles 10% y 90%&quot;) + geom_point(aes(colour = grupo_1)) Y la pregunta que hacemos es ¿podemos distinguir nuestra muestra entre todas las replicaciones producidas con permutaciones? ¿Dónde están los datos observados? Según tu elección, ¿qué tan diferentes son los datos observados de los datos nulos? En este ejemplo, es difícil indicar cuáles son los datos. Los grupos tienen distribuciones similares y es factible que las diferencias que observamos se deban a variación muestral. Si la persona escoge los verdaderos datos, encontramos evidencia en contra de la hipótesis nula (los tres grupos son equivalentes). En algunos contextos, se dice que los datos son significativamente diferentes al nivel 0.05. Esto es evidencia en contra de que los datos se producen de manera homogénea, independientemente del grupo. Si la persona escoge uno de los datos permutados, no encontramos evidencia en contra de que los tres grupos producen datos con distribuciones similares. Comparaciones usando lineup (continuación) Repetimos el ejemplo para otra muestra (en este ejemplo el proceso generador de datos es diferente para el grupo b): Hacemos primero la prueba del lineup: set.seed(121) reps &lt;- lineup(null_permute(&quot;grupo&quot;), muestra_tab, n = 20) grafica_cuantiles(reps |&gt; mutate(grupo_escondido = factor(digest::digest2int(grupo) %% 177)), grupo_escondido, x) + facet_wrap(~.sample) + ylab(&quot;x&quot;) + coord_flip() + geom_point(aes(colour = grupo_escondido)) Podemos distinguir más o menos claramente que está localizada en valores más altos y tiene mayor dispersión. En este caso, como en general podemos identificar los datos, obtenemos evidencia en contra de que los tres grupos tienen distribuciones iguales. Estos ejemplos siguen la idea de inferencia visual propuestas en (Hadley Wickham et al. 2010), (Hofmann et al. 2012) son pruebas muy flexibles y estadísticamente rigurosas. Prueba de permutaciones para proporciones Veremos otro ejemplo donde podemos hacer más concreta la idea de distribución nula o de referencia usando pruebas de permutaciones. Supongamos que con nuestra muestra de tomadores de té, queremos probar la siguiente hipótesis nula: Los tomadores de té en bolsas exclusivamente, usan azúcar a tasas simillares que los tomadores de té suelto (que pueden o no también tomar té en bolsita). Los datos que obtuvimos en nuestra encuesta, en conteos, son: sugar bolsa_exclusivo suelto o bolsa No.sugar 81 74 sugar 89 56 Y en proporciones tenemos que: how prop_azucar n bolsa_exclusivo 0.52 170 suelto o bolsa 0.43 130 Pero distintas muestras podrían haber dado distintos resultados. Nos preguntamos qué tan fuerte es la evidencia en contra de que en realidad los dos grupos de personas usan azúcar en proporciones similares, y la diferencia que vemos se puede atribuir a variación muestral. En este ejemplo, podemos usar una estadística de prueba numérica, por ejemplo, la diferencia entre las dos proporciones: \\[\\hat p_1 - \\hat p_2,\\] (tomadores de té en bolsa solamente vs. suelto y bolsa). El proceso sería entonces: La hipótesis nula es que los dos grupos tienen distribuciones iguales. Este caso quiere decir que en la población, tomadores de té solo en bolsa usan azúcar a las mismas tasas que tomadores de suelto o bolsas. Bajo nuestra hipótesis nula (proporciones iguales), producimos una cantidad grande (por ejemplo 10 mil o más) de muestras permutando las etiquetas de los grupos. Evaluamos nuestra estadística de prueba en cada una de las muestras permutadas. El conjunto de valores obtenidos nos da nuestra distribución de referencia (ya no estamos limitados a 20 replicaciones como en las pruebas gráficas). Y la pregunta clave es: ¿el valor de la estadística en nuestra muestra es extrema en comparación a la distribución de referencia? dif_obs &lt;- te_azucar |&gt; mutate(usa_azucar = as.numeric(sugar == &quot;sugar&quot;)) |&gt; group_by(how) |&gt; summarise(prop_azucar = mean(usa_azucar), .groups = &#39;drop&#39;) |&gt; pivot_wider(names_from = how, values_from = prop_azucar) |&gt; mutate(diferencia_prop = bolsa_exclusivo - `suelto o bolsa`) |&gt; pull(diferencia_prop) La diferencia observada es: dif_obs |&gt; round(3) ## [1] 0.093 Ahora construimos nuestra distribución nula o de referencia: reps &lt;- lineup(null_permute(&quot;how&quot;), te_azucar, n = 50000) glimpse(reps) ## Rows: 15,000,000 ## Columns: 3 ## $ how &lt;chr&gt; &quot;bolsa_exclusivo&quot;, &quot;bolsa_exclusivo&quot;, &quot;suelto o bolsa&quot;, &quot;suelt… ## $ sugar &lt;chr&gt; &quot;sugar&quot;, &quot;No.sugar&quot;, &quot;No.sugar&quot;, &quot;sugar&quot;, &quot;No.sugar&quot;, &quot;No.suga… ## $ .sample &lt;dbl&gt; 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,… valores_ref &lt;- reps |&gt; mutate(usa_azucar = as.numeric(sugar == &quot;sugar&quot;)) |&gt; group_by(.sample, how) |&gt; summarise(prop_azucar = mean(usa_azucar), .groups = &#39;drop&#39;) |&gt; pivot_wider(names_from = how, values_from = prop_azucar) |&gt; mutate(diferencia = bolsa_exclusivo - `suelto o bolsa`) Y graficamos nuestros resultados (con un histograma y una gráfica de cuantiles, por ejemplo). la estadística evaluada un cada una de nuestras muestras permutadas: g_1 &lt;- ggplot(valores_ref, aes(sample = diferencia)) + geom_qq(distribution = stats::qunif) + xlab(&quot;f&quot;) + ylab(&quot;diferencia&quot;) + labs(subtitle = &quot;Distribución nula o de referencia&quot;) g_2 &lt;- ggplot(valores_ref, aes(x = diferencia)) + geom_histogram(binwidth = 0.04) + coord_flip() + xlab(&quot;&quot;) + labs(subtitle = &quot; &quot;) g_1 + g_2 Este es el rango de fluctuación usual para nuestra estadística bajo la hipótesis de que los dos grupos de tomadores de té consumen té a la misma tasa. El valor que obtuvimos en nuestros datos es: 0.09. Mismo que no es un valor extremo en la distribución de referencia que vimos arriba. Ésta muestra no aporta mucha evidencia en contra de que los grupos tienen distribuciones similares. Podemos graficar otra vez marcando el valor observado: # Función de distribución acumulada (inverso de función de cuantiles) dist_perm &lt;- ecdf(valores_ref$diferencia) # Calculamos el percentil del valor observado percentil_obs &lt;- dist_perm(dif_obs) g_1 &lt;- ggplot(valores_ref, aes(sample = diferencia)) + geom_qq(distribution = stats::qunif) + xlab(&quot;f&quot;) + ylab(&quot;diferencia&quot;) + labs(subtitle = &quot;Distribución nula o de referencia&quot;) + geom_hline(yintercept = dif_obs, colour = &quot;red&quot;) + annotate(&quot;text&quot;, x = 0.35, y = dif_obs - 0.03, label = &quot;diferencia observada&quot;, colour = &quot;red&quot;) g_2 &lt;- ggplot(valores_ref, aes(x = diferencia)) + geom_histogram(binwidth = 0.04) + coord_flip() + xlab(&quot;&quot;) + labs(subtitle = &quot; &quot;) + geom_vline(xintercept = dif_obs, colour = &quot;red&quot;) + annotate(&quot;text&quot;, x = dif_obs, y = N_rep * .3, label = percentil_obs,vjust = -0.2, colour = &quot;red&quot;) g_1 + g_2 Y vemos que es un valor algo (pero no muy) extremo en la distribución de referencia que vimos arriba: esta muestra no aporta una gran cantidad de evidencia en contra de que los grupos tienen distribuciones similares, que en este caso significa que los dos grupos usan azúcar a tasas similares. Es decir, sobre la hipótesis \\[H_0: p_1 = p_2,\\] o bien, \\[H_0: p_1 - p_2 = 0.\\] Pruebas de hipótesis tradicionales Comencemos recordando la definición de parámetro y estadística. Definición. Un parámetro es una característica (numérica) de una población o de una distribución de probabilidad. Usualmente denotado por \\(\\theta \\in \\mathbb{R},\\) o por \\(\\theta \\in \\mathbb{R}^p.\\) Una estadística es una característica (numérica) de los datos. Usualmente denotado por \\(\\hat \\theta.\\) Cualquier función de un parámetro es también un parámetro \\(\\varphi = h(\\theta),\\) y cualquier función de una estadística es también una estadística \\(\\hat \\varphi = h(\\hat \\theta).\\) Cuando la estadística se calcula de una muestra aleatoria (\\(T(X),\\) para \\(X_i \\sim \\pi_x\\) para \\(i = 1, \\ldots, n\\)), es por consiguiente aleatoria y es por tanto una variable aleatoria (\\(T \\sim \\pi_T\\)). Por ejemplo \\(\\mu\\) y \\(\\sigma\\) son parámetros de la distribución normal con función de densidad \\(\\pi(x) = (1/\\sqrt{2\\pi}\\sigma)e^{(x-\\mu)^2/(2\\sigma^2)}\\). La varianza \\(\\sigma^2\\), y el coeficiente de variación (cociente de señal a ruido) \\(\\mu/\\sigma\\) también son parámetros. Si \\(X_1, \\ldots ,X_n\\) son una muestra aleatoria, entonces la media \\(\\bar{X}=\\frac1n\\sum_i X_i\\) es una estadística. Ahora podemos pasar a las definiciones correspondientes a pruebas de hipótesis (o pruebas de significancia). Definición. Denotamos por \\(H_0\\) a la hipótesis nula la cual usualmente tratamos como la afirmación del status quo. La hipótesis alternativa la denotamos por \\(H_1\\) y representa el supuesto que está a prueba y para el cual buscamos evidencia en los datos. Definición. La hipótesis normalmente se plantea en términos de un parámetro (\\(\\theta\\in\\mathbb{R}\\)) o conjunto de parámetros (\\(\\theta\\in\\mathbb{R}^p\\)) de la distribución de interés (por ejemplo media, moda, varianza). Para una hipótesis nula del estilo \\(H_0: \\theta = \\theta_0,\\) la hipótesis a contrastar se puede denominar como: Hipótesis alternativa de una cola \\(H_1: \\theta \\gt \\theta_0\\) Hipótesis alternativa de dos colas \\(H_1: \\theta \\neq \\theta_0\\) En el ejemplo anterior planteamos hipótesis nula (proporciones iguales) e hipótesis alternativa que la proporción de tomadores de te suelto que usan azúcar en menor proporción, esto corresponde a una hipótesis alternativa a dos colas: \\(H_0: p_1 = p_2\\), y \\(H_1:p_1 &gt; p_2\\). Definición. Una estadística de prueba es una función numérica de los datos cuyo valor determina el resultado de la prueba. La función usualmente es denotada por \\(T(\\bf X)\\) donde \\(\\bf X\\) representa los datos como variable aleatoria. Por ejemplo, \\(T = T(X_1, \\ldots, X_n)\\) si sólo tenemos una muestra, o por \\(T = T(X_1, \\ldots, X_n, Y_1, \\ldots, Y_m)\\) en el caso de tener dos muestras. Al evaluar la prueba para un conjunto de datos dado, \\(x\\), ésta se denomina estadística de prueba observada, \\(t = T(x).\\) La estadística de prueba correspondiente al ejemplo es \\(T = \\hat p_1 - \\hat p_2.\\) Definición. El valor p es la probabilidad de que bajo la hipótesis nula los datos generen un valor tan extremo como la estadística de prueba observada. Por ejemplo, si consideramos la hipótesis nula admite valores grandes, el valor p se calcula como \\(P(T \\geq t).\\) En el ejemplo de tomadores de té: el valor p lo calculamos usando el percentil donde nuestra observación cae en la distribución generada por las permutación (valor p de una cola). 1 - dist_perm(dif_obs) ## [1] 0.04344 Por otro lado, podemos calcular: Valor p de dos colas: Si la hipótesis nula es cierta, ¿cuál es la probabilidad de observar una diferencia tan extrema o más extrema de lo que observamos? Considerando este caso interpretamos extrema como qué tan lejos cae del centro de masa de la distribución. De tal forma que podemos calcular el valor p como sigue. A partir del valor observado, consideramos cuál dato es menor: la probabilidad bajo lo hipótesis nula de observar una diferencia mayor de la que observamos, o la probabilidad de observar una diferencia menor a la que observamos. Tomamos el mínimo y multiplicamos por dos (Tim C. Hesterberg 2015a): 2 * min(dist_perm(dif_obs), (1 - dist_perm(dif_obs))) ## [1] 0.08688 Este valor p se considera como evidencia moderada en contra de la hipótesis nula. Valores p chicos (observaciones más extremas en comparación con la referencia) aportan más evidencia en contra de la hipótesis nula, y valores más grandes aportan menos evidencia en contra. Definición. Un resultado es estadisticamente significativo si tiene muy baja probabilidad de suceder al azar. Entre más pequeño requiramos un valor p oara declarar un resultado estadísticamente significativo, somos más conservadores. Las pruebas de hipótesis con frecuencia inician contestando una pregunta más general que los valores p: ¿Cuál es la distribución de la estadística de prueba cuando no hay un efecto real? Definición. La distribución nula es la distribución de la estadística de prueba si la hipótesis nula es cierta. En ocasiones también nos referimos a ella como la distribución de referencia pues estamos comparando la estadística de prueba observada a su referencia para determinar que tan inusual es. En el ejemplo de tomadores de té aproximamos la distribución nula (y los valores p) con simulación; sin embargo, para algunas estadísticas hay métodos exactos. En particular, usamos el método de pruebas de permutación. Para dicha prueba el algoritmo para en el caso de dos grupos sería como sigue: Prueba de permutación para dos muestras Supongamos que tenemos m observaciones de una población y n de otra. Combina los m+n valores. Repite: Obtén un remuestra de tamaño m sin reemplazo del total. Usa las n observaciones restantes para obtener la otra muestra. Calcula la estadística de prueba (que compara las muestras). Calcula el valor p como la fracción de las veces que la estadística sobrepasó la estadística observada, multiplica por 2 para una prueba de dos lados. La distribución de la estadística a lo largo de las remuestras de permutación es la distribución de permutación. Ésta puede ser exacta, si se calcula exhaustivamente (como cuando tenemos pocas observaciones) o aproximada (cuando enlistar todas las posible combinaciones es prohibitivo). Tomadores de té (continuación) Ahora hacemos una prueba de permutaciones para otro par de proporciones utilizando el mismo método. La hipótesis nula ahora es: Los tomadores de té Earl Gray usan azúcar a una tasa similar a los tomadores de té negro. Los datos que obtuvimos en nuestra encuesta se muestran en la siguiente tabla: sugar Earl Grey black No.sugar 84 51 sugar 109 23 Y en porcentajes tenemos que: prop_azucar &lt;- te_azucar |&gt; count(Tea, sugar) |&gt; group_by(Tea) |&gt; mutate(prop = 100 * n / sum(n), n = sum(n)) |&gt; filter(sugar == &quot;sugar&quot;) |&gt; select(Tea, prop_azucar = prop, n) |&gt; mutate(&#39;% usa azúcar&#39; = round(prop_azucar)) |&gt; select(-prop_azucar) prop_azucar |&gt; formatear_tabla() Tea n % usa azúcar Earl Grey 193 56 black 74 31 Pero distintas muestras podrían haber dado distintos resultados. Nos preguntamos qué tan fuerte es la evidencia en contra de que en realidad los dos grupos de personas usan azúcar en proporciones similares considerando que la diferencia que vemos se puede atribuir a variación muestral. Escribimos la función que calcula diferencias para cada muestra: calc_diferencia_2 &lt;- function(datos){ datos |&gt; mutate(usa_azucar = as.numeric(sugar == &quot;sugar&quot;)) |&gt; group_by(Tea) |&gt; summarise(prop_azucar = mean(usa_azucar), .groups = &#39;drop&#39;) |&gt; pivot_wider(names_from = Tea, values_from = prop_azucar) |&gt; mutate(diferencia_prop = `Earl Grey` - black) |&gt; pull(diferencia_prop) } La diferencia observada es: ## [1] 0.254 Ahora construimos nuestra distribución nula o de referencia: set.seed(2) reps &lt;- lineup(null_permute(&quot;Tea&quot;), te_azucar, n = N_rep) valores_ref &lt;- reps |&gt; group_by(.sample) |&gt; nest() |&gt; mutate(diferencia = lapply(data, calc_diferencia_2)) |&gt; unnest(diferencia) Y podemos graficar la distribución de referencia otra vez marcando el valor observado En este caso, la evidencia es muy fuerte en contra de la hipótesis nula, pues el resultado que obtuvimos es muy extremo en relación a la distribución de referencia. El valor p es cercano a 0. Haz una prueba de permutaciones para diferencia de medias para comparar la propina en cena vs en comidas. Grafica la distribución de referencia. Calcula el valor p (dos colas). Pruebas de permutación: implementación. Hasta ahora nos hemos centrado en ejemplos de diferencias en medias. Podemos extender las pruebas de permutación a \\(\\bar{X}\\) (la media de la primera muestra), \\(n\\bar{X}\\) (la suma de las observaciones en la primera muestra), y más. Teorema. En pruebas de permutación, si dos estadísticas de prueba \\(T_1\\) y \\(T_2\\) están relacionadas por una función estríctamente monótona, \\(T_1(X^*)=f(T_2(X^*))\\) donde \\(X^*\\) es una remuestra de permutación de los datos originales, entonces los valores p serán los mismos en las pruebas de permutación. Muestras con reemplazo de la Distribución Nula. En la implementación de muestreo, no nos aseguramos que las remuestras sean únicas. Sería más acertado tomar muestras sin reemplazo, sin embargo, el costo computacional es demasiado alto. Por simplicidad consideramos muestras con reemplazo del total de \\[m+n\\choose n\\] posibles remuestras. Por lo tanto, al remuestrar obtenemos una muestra de la distribución nula. Entre más muestras, más exactitud. Hemos usado \\(B = 10^3\\) remuestras (N_rep en el código), en general entre más remuestras tendremos una mejor estimación del valor p. Si el verdadero valor es \\(p\\) el estimado tendrá una varianza aproximadamente de \\(p(1- p)/B\\) donde \\(B\\) es el número de remuestras generadas. Observación. Así como los \\(n\\) datos originales son una muestra de la población, también las \\(B\\) remuestras de la estadística son una muestra de una población, en este caso de la distribución nula. La pruebas de permutaciones son más útiles cuando nuestra hipótesis nula se refiere que la distribución de los grupos son muy similares, o la independencia entre observaciones y grupo. Esto también aplica cuando queremos probar por ejemplo, que una variable numérica \\(Y\\) es independiente de \\(X.\\) Hay algunas hipótesis que no se pueden probar con este método, como por ejemplo, las que se refieren a una sola muestra: ¿los datos son consistentes con que su media es igual a 5? Adicionalmente, en algunas ocasiones queremos probar aspectos más específicos de las diferencias: como ¿son iguales las medias o medianas de dos grupos de datos? ¿Tienen dispersión similar? Es común aplicar pruebas de permutaciones a este segundo problema, sin embargo, no están tan perfectamente adaptadas a el, pues prueban todos los aspectos de las distribuciones que se comparan, aún cuando escojamos una estadística particular que pretende medir. Por ejemplo, cuando trabajamos con la diferencia de medias. Eso quiere decir que podemos rechazar igualdad de medias, por ejemplo, cuando en realidad otra característica de las distribuciones es la que difiere mucho en las poblaciones. En algunas referencias (ver (Chihara and Hesterberg 2018), (Efron and Tibshirani 1993)) se argumenta que de todas formas las pruebas de permutaciones son relativamente robustas a esta desadaptación. Un caso excepcional, por ejemplo, es cuando las poblaciones que comparamos resultan tener dispersión extremadamente distinta, y adicionalmente los tamaños de muestra de los grupos son muy desiguales (otra vez, ver ejemplos en (Chihara and Hesterberg 2018)). Ejemplo: tiempos de fusión Veamos el siguiente ejemplo, que es un experimento donde se midió el tiempo que tardan distintas personas en fusionar un estereograma para ver una imagen 3D. (William S. Cleveland (1993)). Existen dos condiciones: en una se dio indicaciones de qué figura tenían que buscar (VV) y en otra no se dio esa indicación. ¿Las instrucciones verbales ayudan a fusionar más rápido el estereograma? ## ## ── Column specification ──────────────────────────────────────────────────────── ## cols( ## n = col_double(), ## time = col_double(), ## nv.vv = col_character() ## ) La situación es la siguiente: considerando que hay mucha variación en el tiempo de fusión dentro de cada tratamiento, necesitamos calificar la evidencia de nuestra conclusión (el tiempo de fusión se reduce con información verbal). Podemos usar una prueba de permutaciones, esta vez justificándola por el hecho de que los tratamientos se asignan al azar: si los tratamientos son indistinguibles, entonces las etiquetas de los grupos son sólo etiquetas, y permutarlas daría muestras igualmente verosímiles. En este caso, compararemos gráficas de cuantiles de los datos con los producidos por permutaciones (transformamos los datos pues en este caso es más apropiado una comparación multiplicativa): ¿Podemos identificar los datos? En general, muy frecuentemente las personas identifican los datos correctamente, lo que muestra evidencia considerable de que la instrucción verbal altera los tiempos de respuesta de los partipantes, y en este caso ayuda a reducir el tiempo de fusión de los estereogramas. Ejemplo: tiempos de fusión (continuación) Podemos usar las pruebas de permutaciones para distintos tipos de estadísticas: medianas, medias, comparar dispersión usando rangos intercuartiles o varianzas, etc. Regresamos a los tiempos de fusión. Podemos hacer una prueba de permutaciones para la diferencia de las medias o medianas, por ejemplo. En este ejemplo usaremos una medida de centralidad un poco diferente, como ilustración: el promedio de los cuartiles superior e inferior de las dos distribuciones. Usaremos el cociente de estas dos cantidades para medir su diferencia # esta función hace permutaciones y calcula la diferencia para cada una permutaciones_est &lt;- function(datos, variable, calc_diferencia, n = 1000){ # calcular estadística para cada grupo permutar &lt;- function(variable){ sample(variable, length(variable)) } tbl_perms &lt;- tibble(.sample = seq(1, n-1, 1)) |&gt; mutate(diferencia = map_dbl(.sample, ~ datos |&gt; mutate({{variable}}:= permutar({{variable}})) |&gt; calc_diferencia())) bind_rows(tbl_perms, tibble(.sample = n, diferencia = calc_diferencia(datos))) } stat_fusion &lt;- function(x){ (quantile(x, 0.75) + quantile(x, 0.25))/2 } calc_fusion &lt;- function(stat_fusion){ fun &lt;- function(datos){ datos |&gt; group_by(nv.vv) |&gt; summarise(est = stat_fusion(time), .groups = &#39;drop&#39;) |&gt; pivot_wider(names_from = nv.vv, values_from = est) |&gt; mutate(dif = VV / NV ) |&gt; pull(dif) } fun } calc_cociente &lt;- calc_fusion(stat_fusion) dif_obs &lt;- calc_cociente(fusion) # permutar valores_ref &lt;- permutaciones_est(fusion, nv.vv, calc_cociente, n = N_rep) dist_perm_nv &lt;- ecdf(valores_ref$diferencia) cuantil_obs &lt;- dist_perm_nv(dif_obs) Y el valor p de dos colas es dist_perm_nv &lt;- ecdf(valores_ref$diferencia) 2 * min(dist_perm_nv(dif_obs), 1 - dist_perm_nv(dif_obs)) ## [1] 0.028 Lo que muestra evidencia considerable, aunque no muy fuerte, de que la instrucción verbal ayuda a reducir el tiempo de fusión de los estereogramas: la caja del diagrama de caja y brazos para el grupo VV está encogida por un factor menor a 1. Separación de grupos Este ejemplo tomado de (Chowdhury et al. 2015) (tanto la idea como el código). La pregunta que se aborda en ese estudio es: Existen métodos de clasificación (supervisados o no supervisados) para formar grupos en términos de variables que describen a los individuos Estos métodos (análisis discriminante, o k-means, por ejemplo), pretenden formar grupos compactos, bien separados entre ellos. Cuando aplicamos el método, obtenemos clasificadores basados en las variables de entrada. La pregunta es: ¿los grupos resultantes son producto de patrones que se generalizan a la población, o capitalizaron en variación aleatoria para formarse? Especialmente cuando tenemos muchas mediciones de los individuos, y una muestra relativamente chica, Es relativamente fácil encontrar combinaciones de variables que separan los grupos, aunque estas combinaciones y diferencias están basadas en ruido y no generalizan a la población. Como muestran en (Chowdhury et al. 2015), el lineup es útil para juzgar si tenemos evidencia en contra de que los grupos en realidad son iguales, y usamos variación muestral para separarlos. Avispas (opcional) En el siguiente ejemplo, tenemos 4 grupos de avispas (50 individuos en total), y para cada individuo se miden expresiones de 42 genes distintos. La pregunta es: ¿Podemos separar a los grupos de avispas dependiendo de sus mediciones? En este se usó análisis discriminante (LDA) para buscar proyecciones de los datos en dimensión baja de forma que los grupos sean lo más compactos y separados posibles. Para probar qué tan bien funciona este método, podemos hacer una prueba de permutación, aplicamos LDA y observamos los resultados. Y vemos que incluso permutando los grupos, es generalmente posible separarlos en grupos bien definidos: la búsqueda es suficientemente agresiva para encontrar combinaciones lineales que los separan. Que no podamos distinguir los datos verdaderos de las replicaciones nulas indica que este método difícilmente puede servir para separar los grupos claramente. Otro enfoque sería separar los datos en una muestra de entrenamiento y una de prueba (que discutiremos en la última sesión). Aplicamos el procedimiento a la muestra de entrenamiento y luego vemos qué pasa con los datos de prueba: set.seed(8) wasps_1 &lt;- wasps |&gt; mutate(u = runif(nrow(wasps), 0, 1)) wasps_entrena &lt;- wasps_1 |&gt; filter(u &lt;= 0.8) wasps_prueba &lt;- wasps_1 |&gt; filter(u &gt; 0.8) wasp.lda &lt;- MASS::lda(Group ~ ., data=wasps_entrena[,-1]) wasp_ld_entrena &lt;- predict(wasp.lda, dimen=2)$x |&gt; as_tibble(.name_repair = &quot;universal&quot;) |&gt; mutate(tipo = &quot;entrenamiento&quot;) |&gt; mutate(grupo = wasps_entrena$Group) wasp_ld_prueba &lt;- predict(wasp.lda, newdata = wasps_prueba, dimen=2)$x |&gt; as_tibble(.name_repair = &quot;universal&quot;) |&gt; mutate(tipo = &quot;prueba&quot;)|&gt; mutate(grupo = wasps_prueba$Group) wasp_lda &lt;- bind_rows(wasp_ld_entrena, wasp_ld_prueba) ggplot(wasp_lda, aes(x = LD1, y = LD2, colour = grupo)) + geom_point(size = 3) + facet_wrap(~tipo) Aunque esta separación de datos es menos efectiva en este ejemplo por la muestra chica, podemos ver que la separación lograda en los datos de entrenamiento probablemente se debe a variación muestral. La “crisis de replicabilidad” Recientemente (Ioannidis 2005) se ha reconocido en campos como la psicología la crisis de replicabilidad. Varios estudios que recibieron mucha publicidad inicialmente no han podido ser replicados posteriormente por otros investigadores. Por ejemplo: Hacer poses poderosas produce cambios fisiológicos que mejoran nuestro desempeño en ciertas tareas. Mostrar palabras relacionadas con “viejo” hacen que las personas caminen más lento (efectos de priming). En todos estos casos, el argumento de la evidencia de estos efectos fue respaldada por una prueba de hipótesis nula con un valor p menor a 0.05. La razón es que ese es el estándar de publicación seguido por varias áreas y revistas arbitradas. La tasa de no replicabilidad parece ser mucho más alta (al menos la mitad o más, según algunas fuentes como la señalada arriba) que la sugerida por la tasa de falsos positivos (menos de 5%). Este problema de replicabilidad parece ser más frecuente cuando: Se trata de estudios de potencia baja: mediciones ruidosas y tamaños de muestra chicos. El plan de análisis no está claramente definido desde un principio (lo cual es difícil cuando se están investigando “fenómenos no estudiados antes”). ¿A qué se atribuye esta crisis de replicabilidad? El jardín de los senderos que se bifurcan Aunque haya algunos ejemplos de manipulaciones conscientes —e incluso, en menos casos, malintencionadas— para obtener resultados publicables o significativos (p-hacking), como vimos en ejemplos anteriores, hay varias decisiones, todas razonables, que podemos tomar cuando estamos buscando las comparaciones correctas. Algunas pueden ser: Transformar los datos (tomar o no logaritmos, u otra transformación). Editar datos atípicos (razonable si los equipos pueden fallar, o hay errores de captura, por ejemplo). Distintas maneras de interpretar los criterios de inclusión de un estudio (por ejemplo, algunos participantes mostraron tener gripa, o revelaron que durmieron muy poco la noche anterior, etc. ¿los dejamos o los quitamos?). Dado un conjunto de datos, las justificaciones de las decisiones que se toman en cada paso son razonables, pero con datos distintos las decisiones podrían ser diferentes. Este es el jardín de los senderos que se bifurcan (ver referencia en Gelman), que invalida en parte el uso valores p como criterio de evidencia contra la hipótesis nula. Esto es exacerbado por: Tamaños de muestra chicos y efectos “inestables” que se quieren medir (por ejemplo en psicología). El hecho de que el criterio de publicación es obtener un valor \\(p &lt; 0.05\\), y la presión fuerte sobre los investigadores para producir resultados publicables (\\(p &lt; 0.05\\)). El que estudios o resultados similares que no obtuvieron valores \\(p\\) por debajo del umbral no son publicados o reportados. Ver por ejemplo el comunicado de la ASA. Ojo: esas presiones de publicación no sólo ocurre para investigadores en psicología. Cuando trabajamos en problemas de análisis de datos que son de importancia, es común que existan intereses de algunas partes o personas involucradas por algunos resultados u otros (por ejemplo, nuestros clientes de consultoría o clientes internos). Eso puede dañar nuestro trabajo como analistas, y el avance de nuestro equipo. Aunque esas presiones son inevitables, se vuelven manejables cuando hay una relación de confianza entre las partes involucradas. Ejemplo: decisiones de análisis y valores p En el ejemplo de datos de fusión, decidimos probar, por ejemplo, el promedio de los cuartiles inferior y superior, lo cual no es una decisión típica pero usamos como ilustración. Ahora intentamos usar distintas mediciones de la diferencia entre los grupos, usando distintas medidas resumen y transformaciones (por ejemplo, con o sin logaritmo). Aquí hay unas 12 combinaciones distintas para hacer el análisis (multiplicadas por criterios de “aceptación de datos en la muestra”, que simulamos tomando una submuestra al azar): calc_fusion &lt;- function(stat_fusion, trans, comparacion){ fun &lt;- function(datos){ datos |&gt; group_by(nv.vv) |&gt; summarise(est = stat_fusion({{ trans }}(time)), .groups = &#39;drop&#39;) |&gt; pivot_wider(names_from = nv.vv, values_from = est) |&gt; mutate(dif = {{ comparacion }}) |&gt; pull(dif) } fun } valor_p &lt;- function(datos, variable, calc_diferencia, n = 1000){ # calcular estadística para cada grupo permutar &lt;- function(variable){ sample(variable, length(variable)) } tbl_perms &lt;- tibble(.sample = seq(1, n-1, 1)) |&gt; mutate(diferencia = map_dbl(.sample, ~ datos |&gt; mutate({{variable}} := permutar({{variable}})) |&gt; calc_diferencia())) perms &lt;- bind_rows(tbl_perms, tibble(.sample = n, diferencia = calc_diferencia(datos))) perms_ecdf &lt;- ecdf(perms$diferencia) dif &lt;- calc_diferencia(datos) 2 * min(perms_ecdf(dif), 1- perms_ecdf(dif)) } set.seed(7272) media_cuartiles &lt;- function(x){ (quantile(x, 0.75) + quantile(x, 0.25))/2 } # nota: usar n=10000 o más, esto solo es para demostración: ejemplo &lt;- list() calc_dif &lt;- calc_fusion(mean, identity, VV - NV) ejemplo$media_dif &lt;- valor_p(fusion |&gt; sample_frac(0.95), nv.vv, calc_dif, n = N_rep) calc_dif &lt;- calc_fusion(mean, log, VV - NV) ejemplo$media_dif_log &lt;- valor_p(fusion |&gt; sample_frac(0.95), nv.vv, calc_dif, n = N_rep) calc_dif &lt;- calc_fusion(median, identity, VV / NV) ejemplo$mediana_razon &lt;- valor_p(fusion |&gt; sample_frac(0.95), nv.vv, calc_dif, n = N_rep) calc_dif &lt;- calc_fusion(media_cuartiles, identity, VV / NV) ejemplo$cuartiles_razon &lt;- valor_p(fusion |&gt; sample_frac(0.95), nv.vv, calc_dif, n = N_rep) ejemplo &lt;- read_rds(&quot;cache/ejemplo_p_val.rds&quot;) ejemplo$media_dif ## [1] 0.0658 ejemplo$media_dif_log ## [1] 0.018 ejemplo$mediana_razon ## [1] 0.049 ejemplo$cuartiles_razon ## [1] 0.0464 Si existen grados de libertad —muchas veces necesarios para hacer un análisis exitoso—, entonces los valores p pueden tener poco significado. Alternativas o soluciones El primer punto importante es reconocer que la mayor parte de nuestro trabajo es exploratorio (recordemos el proceso complicado del análisis de datos de refinamiento de preguntas). En este tipo de trabajo, reportar valores p puede tener poco sentido, y mucho menos tiene sentido aceptar algo verdadero cuando pasa un umbral de significancia dado. Nuestro interés principal al hacer análisis es: expresar correctamente, y de manera útil, la incertidumbre asociada a las conclusiones o patrones que mostramos (asociada a variación muestral, por ejemplo) con el objetivo que el proceso de toma de decisiones sea informado. Un resumen de un número (valor p, o el que sea) no puede ser tomado como criterio para tomar una decisión que generalmente es compleja. En la siguiente sección veremos cómo podemos mostrar parte de esa incertidumbre de manera más útil. Por otra parte, los estudios confirmatorios (donde se reportan valores p) también tienen un lugar. En áreas como la psicología, existen ahora movimientos fuertes en favor de la repetición de estudios prometedores pero donde hay sospecha de grados de libertad del investigador. Este movimiento sugiere dar valor a los estudios exploratorios que no reportan valor p, y posteriormente, si el estudio es de interés, puede intentarse una replicación confirmatoria, con potencia más alta y con planes de análisis predefinidos. Referencias "],["estimación-y-distribución-de-muestreo-1.html", "Sección 4 Estimación y distribución de muestreo Ejemplo: precios de casas Distribución de muestreo Más ejemplos El error estándar Calculando la distribución de muestreo Teorema central del límite Normalidad y gráficas de cuantiles normales Prueba de hipótesis de normalidad Ejemplo Más del Teorema central del límite", " Sección 4 Estimación y distribución de muestreo En esta sección discutiremos cuál el objetivo general del proceso de estimación, y cómo entender y manejar la variabilidad que se produce cuando aleatorizamos la selección de las muestras que utilizamos para hacer análisis. A diferencia de las pruebas de permutación, donde evaluábamos como cambiaría una estadísitica si un tratamiento o grupo se hubiera asignado de forma distinta, en la siguiente sección nos preguntamos como varía una estadística entre muestras. Por ejemplo, pasaremos de preguntar si una vacuna reduce el riesgo de una enfermedad a evaluar en que magnitud se reduce el riesgo de contraer la enfermedad. Ejemplo: precios de casas Supongamos que queremos conocer el valor total de las casas que se vendieron recientemente en una zona particular. Supondremos que tenemos un listado de las casas que se han vendido recientemente, pero en ese listado no se encuentra el precio de venta. Decidimos entonces tomar una muestra aleatoria de 100 de esas casas. Para esas casas hacemos trabajo de campo para averiguar el precio de venta. marco_casas &lt;- read_csv(&quot;data/casas.csv&quot;) set.seed(841) muestra_casas &lt;- sample_n(marco_casas, 100) |&gt; select(id, nombre_zona, area_habitable_sup_m2, precio_miles) sprintf(&quot;Hay %0.0f casas en total, tomamos muestra de %0.0f&quot;, nrow(marco_casas), nrow(muestra_casas)) ## [1] &quot;Hay 1144 casas en total, tomamos muestra de 100&quot; head(muestra_casas) ## # A tibble: 6 × 4 ## id nombre_zona area_habitable_sup_m2 precio_miles ## &lt;dbl&gt; &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 287 NAmes 161. 159 ## 2 755 NAmes 95.3 156 ## 3 1190 Gilbert 168. 189 ## 4 36 NridgHt 228. 309 ## 5 32 Sawyer 114. 149. ## 6 538 NAmes 80.3 111. Como tomamos una muestra aleatoria, intentamos estimar el valor total de las casas que se vendieron expandiendo el total muestral, es decir nuestro estimador \\(\\hat{t} = t(X_1,\\ldots X_{100})\\) del total poblacional \\(t\\) es \\[\\hat{t} = \\frac{N}{n} \\sum_{i=1}^{100} X_i = N\\bar{x}\\] Esta función implementa el estimador: n &lt;- nrow(muestra_casas) # tamaño muestra N &lt;- nrow(marco_casas) # tamaño población estimar_total &lt;- function(muestra_casas, N){ total_muestral &lt;- sum(muestra_casas$precio_miles) n &lt;- nrow(muestra_casas) # cada unidad de la muestra representa a N/n f_exp &lt;- N / n # estimador total es la expansión del total muestral estimador_total &lt;- f_exp * total_muestral res &lt;- tibble(total_muestra = total_muestral, factor_exp = f_exp, est_total_millones = estimador_total / 1000) res } estimar_total(muestra_casas, N) |&gt; mutate(across(where(is.numeric), \\(x) round(x, 2))) ## # A tibble: 1 × 3 ## total_muestra factor_exp est_total_millones ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 18444. 11.4 211 Sin embargo, si hubiéramos obtenido otra muestra, hubiéramos obtenido otra estimación diferente. Por ejemplo: estimar_total(sample_n(marco_casas, 100), N) |&gt; mutate(across(where(is.numeric), \\(x) round(x, 2))) ## # A tibble: 1 × 3 ## total_muestra factor_exp est_total_millones ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 17916. 11.4 205. El valor poblacional que buscamos estimar (nótese que en la práctica este no lo conocemos) es: # multiplicar por 1000 para que sea en millones de dólares total_pob &lt;- sum(marco_casas |&gt; pull(precio_miles)) / 1000 total_pob ## [1] 209.7633 Así que: Para algunas muestras esta estadística puede estar muy cercana al valor poblacional, pero para otras puede estar más lejana. Para entender qué tan buena es una estimación particular, entonces, tenemos que entender cuánta variabilidad hay de muestra a muestra debida a la aleatorización. Esto depende del diseño de la muestra y de la población de precios de casas (que no conocemos). Distribución de muestreo La distribución de muestreo de una estadística enumera los posibles resultados que puede tomar esa estadística sobre todas las muestras posibles. Este es el concepto básico para poder entender qué tan bien o mal estima un parámetro poblacional dado. En nuestro ejemplo anterior de precio de casas, no podemos calcular todas las posibles estimaciones bajo todas las posibles muestras, pero podemos aproximar repitiendo una gran cantidad de veces el proceso de muestreo, como hicimos al aproximar la distribución de permutaciones de estadísticas de prueba de las secciones anteriores. Empezamos repitiendo 10 veces y examinamos cómo varía nuestra estadística: replicar_muestreo &lt;- function(marco_casas, m = 500, n){ # n es el tamaño de muestra que se saca de marco_casas # m es el número de veces que repetimos el muestro de tamaño n resultados &lt;- map_df(1:m, function(id) { sample_n(marco_casas, n) |&gt; estimar_total(N) }, .id = &quot;id_muestra&quot;) } replicar_muestreo(marco_casas, m = 10, n = 100) |&gt; mutate(across(where(is.numeric), round, 1)) |&gt; formatear_tabla() id_muestra total_muestra factor_exp est_total_millones 1 17594.8 11.4 201.3 2 17423.9 11.4 199.3 3 18444.3 11.4 211.0 4 17696.6 11.4 202.4 5 17275.8 11.4 197.6 6 17867.6 11.4 204.4 7 18450.8 11.4 211.1 8 18187.2 11.4 208.1 9 18604.2 11.4 212.8 10 19144.4 11.4 219.0 Como vemos, hay variación considerable en nuestro estimador del total, pero la estimación que haríamos con cualquiera de estas muestras no es muy mala. Ahora examinamos un número más grande de simulaciones: replicaciones_1 &lt;- replicar_muestreo(marco_casas, m = 1500, n = 100) Y el siguiente histograma nos dice qué podemos esperar de la variación de nuestras estimaciones, y donde es más probable que una estimación particular caiga: graf_1 &lt;- ggplot(replicaciones_1, aes(x = est_total_millones)) + geom_histogram() + geom_vline(xintercept = total_pob, colour = &quot;red&quot;) + xlab(&quot;Millones de dólares&quot;) + scale_x_continuous(breaks = seq(180, 240, 10), limits = c(180, 240)) graf_1 Con muy alta probabilidad el error no será de más de unos 30 millones de dólares (o no más de 20% del valor poblacional). Definición Sea \\(X_1, X_2, \\ldots X_n\\) una muestra, y \\(T = t(X_1, X_2, \\ldots, X_n)\\) una estadística. La distribución de muestreo de \\(T\\) es la función de distribución de \\(T\\). Esta distribución es sobre todas las posibles muestras que se pueden obtener. Cuando usamos \\(T\\) para estimar algún parámetro poblacional \\(\\theta\\), decimos informalmente que el estimador es preciso si su distribución de muestreo está muy concentrada alrededor del valor \\(\\theta\\) que queremos estimar. Si la distribución de muestreo está concentrada en un conjunto muy grande o muy disperso, quiere decir que con alta probabilidad cuando obtengamos nuestra muestra y calculemos nuestra estimación, el resultado estará lejano del valor poblacional que nos interesa estimar. Veamos qué pasa cuando hacemos la muestra más grande en nuestro ejemplo: replicaciones_2 &lt;- replicar_muestreo(marco_casas, m = 1500, n = 250) Graficamos las dos distribuciones de muestreo juntas, y vemos cómo con mayor muestra obtenemos un estimador más preciso, y sin considerar el costo, preferimos el estimador mejor concentrado alrededor del valor que buscamos estimar. library(patchwork) graf_2 &lt;- ggplot(replicaciones_2, aes(x = est_total_millones)) + geom_histogram() + geom_vline(xintercept = total_pob, colour = &quot;red&quot;) + xlab(&quot;Millones de dólares&quot;) + scale_x_continuous(breaks = seq(180, 240, 10), limits = c(180, 240)) graf_1 + graf_2 Observación: a veces este concepto se confunde la distribución poblacional de las \\(X_i\\). Esto es muy diferente. Por ejemplo, en nuestro caso, el histograma de la distribución de valores poblacionales es ggplot(marco_casas, aes(x = precio_miles)) + geom_histogram() que en general no tiene ver mucho en escala o forma con la distribución de muestreo de nuestro estimador del total. Más ejemplos Podemos también considerar muestrear de poblaciones sintéticas o modelos probabilísticos que usamos para modelar poblaciones reales. Por ejemplo, supongamos que tomamos una muestra de tamaño 15 de la distribución uniforme en \\([0,1]\\). Es decir, cada \\(X_i\\) es un valor uniformemente distribuido en \\([0,1]\\), y las \\(X_i\\) se extraen independientemente unas de otras. Consideramos dos estadísticas de interés: La media muestral \\(T_1(X) = \\frac{1}{15}\\sum_{i = 1}^{15} X_i\\) El cuantil 0.75 de la muestra \\(T_2(X) = q_{0.75}(X)\\) ¿Cómo crees que se vean las distribuciones muestrales de estas estadísticas? ¿Alrededor de qué valores crees que concentren? ¿Crees que tendrán mucha o poca dispersión? ¿Qué forma crees que tengan? Para el primer caso hacemos: # simular replicar_muestreo_unif &lt;- function(est = mean, m, n = 15){ valores_est &lt;- map_dbl(1:m, ~ est(runif(n))) tibble(id_muestra = 1:m, estimacion = valores_est) } sim_estimador_1 &lt;- replicar_muestreo_unif(mean, 4000, 15) # graficar aprox de distribución de muestreo ggplot(sim_estimador_1, aes(x = estimacion)) + geom_histogram(bins = 40) + xlim(c(0, 1)) # simular para el máximo cuantil_75 &lt;- function(x) quantile(x, 0.75) sim_estimador_2 &lt;- replicar_muestreo_unif(cuantil_75, 4000, 15) # graficar distribución de muestreo ggplot(sim_estimador_2, aes(x = estimacion)) + geom_histogram(breaks = seq(0, 1, 0.02)) + xlim(c(0, 1)) Supón que tenemos una muestra de 30 observaciones de una distribución uniforme \\([0,b]\\). ¿Qué tan buen estimador de \\(b/2\\) es la media muestral? ¿Cómo lo cuantificarías? ¿Qué tan buen estimador del cuantil 0.8 de la distribución uniforme es el cuantil 0.8 muestral? ¿Qué desventajas notas en este estimador? El error estándar Una primera medida útil de la dispersión de la distribución de muestreo es su desviación estándar: la razón específica tiene qué ver con un resultado importante, el teorema central del límite, que veremos más adelante. En este caso particular, a esta desviación estándar se le llama error estándar: Definición A la desviación estándar de una estadística \\(T\\) le llamamos su error estándar, y la denotamos por \\(\\text{ee}(T)\\). A cualquier estimador de este error estándar lo denotamos como \\(\\hat{\\text{ee}}(T)\\). Este error estándar mide qué tanto varía el estimador \\(T\\) de muestra a muestra. Observación: es importante no confundir el error estándar con la desviación estándar de una muestra (o de la población). En nuestro ejemplo de las uniformes, la desviación estándar de las muestras varía como: map_dbl(1:10000, ~ sd(runif(15))) |&gt; quantile() |&gt; round(2) ## 0% 25% 50% 75% 100% ## 0.11 0.26 0.29 0.31 0.41 Mientras que el error estándar de la media es aproximadamente map_dbl(1:10000, ~ mean(runif(15))) |&gt; sd() ## [1] 0.07439575 y el error estándar del máximo es aproximadamente map_dbl(1:10000, ~ max(runif(15))) |&gt; sd() ## [1] 0.05928675 Como ejercicio para contrastar estos conceptos, puedes considerar: ¿Qué pasa con la desviación estándar de una muestra muy grande de uniformes? ¿Qué pasa con el error estándar de la media muestral de una muestra muy grande de uniformes? Ejemplo: valor de casas Consideramos el error estándar del estimador del total del inventario vendido, usando una muestra de 250 con el estimador del total que describimos arriba. Como aproximamos con simulación la distribución de muestreo, podemos hacer: ee_2 &lt;- replicaciones_2 |&gt; pull(est_total_millones) |&gt; sd() round(ee_2, 1) ## [1] 5.2 que está en millones de pesos y cuantifica la dispersión de la distribución de muestreo del estimador del total. Para tamaño de muestra 100, obtenemos más dispersión: ee_1 &lt;- replicaciones_1 |&gt; pull(est_total_millones) |&gt; sd() round(ee_1, 1) ## [1] 8.9 Nótese que esto es muy diferente, por ejemplo, a la desviación estándar poblacional o de una muestra. Estas dos cantidades miden la variabilidad del estimador del total. Calculando la distribución de muestreo En los ejemplos anteriores usamos simulación para obtener aproximaciones de la distribución de muestreo de algunos estimadores. También es posible: Hacer cálculos exactos a partir de modelos probabilísticos. Hacer aproximaciones asintóticas para muestras grandes (de las cuales la más importante es la que da el teorema central del límite). En los ejemplos de arriba, cuando muestreamos de la poblaciones, extrajimos las muestras de manera aproximadamente independiente. Cada observación \\(X_i\\) tiene la misma distribución y las \\(X_i\\)’s son independientes. Este tipo de diseños aleatorizados es de los más simples, y se llama muestreo aleatorio simple. En general, en esta parte haremos siempre este supuesto: Una muestra es iid (independiente e idénticamente distribuida) si es es un conjunto de observaciones \\(X_1,X_2, \\ldots X_n\\) independientes, y cada una con la misma distribución. En términos de poblaciones, esto lo logramos obteniendo cada observación de manera aleatoria con el mismo procedimiento. En términos de modelos probabilísticos, cada \\(X_i\\) se extrae de la misma distribución fija \\(F\\) (que pensamos como la “población”) de manera independiente. Esto lo denotamos por \\(X_i \\overset{iid}{\\sim} F.\\) Ejemplo Si \\(X_1, X_2, \\ldots X_n\\) es una muestra de uniformes independientes en \\([0,1]\\), ¿cómo calcularíamos la distribución de muestreo del máximo muestra \\(T_2 = \\max\\)? En este caso, es fácil calcular su función de distribución acumulada de manera exacta: \\[F_{\\max}(x) = P(\\max\\{X_1,X_2,\\ldots X_n\\} \\leq x)\\] El máximo es menor o igual a \\(x\\) si y sólo si todas las \\(X_i\\) son menores o iguales a \\(x\\), así que \\[F_{\\max} (x) = P(X_1\\leq x, X_2\\leq x, \\cdots, X_n\\leq x)\\] como las \\(X_i\\)’s son independientes entonces \\[F_{\\max}(x) = P(X_1\\leq x)P(X_2\\leq x)\\cdots P(X_n\\leq x) = x^n\\] para \\(x\\in [0,1]\\), pues para cada \\(X_i\\) tenemos \\(P(X_i\\leq x) = x\\). Así que no es necesario usar simulación para conocer esta distribución de muestreo. Derivando esta distribución acumulada obtenemos su densidad, que es \\[f(x) = nx^{n-1}\\] para \\(x\\in [0,1]\\), y es cero en otro caso. Si comparamos con nuestra simulación: teorica &lt;- tibble(x = seq(0, 1 ,0.001)) |&gt; mutate(f_dens = 15 * x^14) sim_estimador_3 &lt;- replicar_muestreo_unif(max, 4000, 15) ggplot(sim_estimador_3) + geom_histogram(aes(x = estimacion), breaks = seq(0, 1, 0.02)) + xlim(c(0.5, 1)) + # el histograma es de ancho 0.02 y el número de simulaciones 4000 geom_line(data = teorica, aes(x = x, y = (4000 * 0.02) * f_dens), colour = &quot;red&quot;, linewidth = 1.3) Y vemos que con la simulación obtuvimos una buena aproximación Nota: ¿cómo se relaciona un histograma con la función de densidad que genera los datos? Supón que \\(f(x)\\) es una función de densidad, y obtenemos un número \\(n\\) de simulaciones independientes. Si escogemos un histograma de ancho \\(\\Delta\\), ¿cuántas observaciones esperamos que caigan en un intervalo \\(I = [a - \\Delta/2, a + \\Delta/2]\\)?. La probabilidad de que una observación caiga en \\(I\\) es igual a \\[P(X\\in I) = \\int_I f(x)\\,dx = \\int_{a - \\Delta/2}^{a + \\Delta/2} f(x)\\,dx \\approx f(a) \\text{long}(I) = f(a) \\Delta\\] para \\(\\Delta\\) chica. Si nuestra muestra es de tamaño \\(n\\), el número esperado de observaciones que caen en \\(I\\) es entonces \\(nf(a)\\Delta\\). Eso explica el ajuste que hicimos en la gráfica de arriba. Otra manera de hacer es ajustando el histograma: si en un intervalo el histograma alcanza el valor \\(y\\), \\[f(a) = \\frac{y}{n\\Delta}\\] teorica &lt;- tibble(x = seq(0, 1 ,0.001)) |&gt; mutate(f_dens = 15*x^{14}) ggplot(sim_estimador_3) + geom_histogram(aes(x = estimacion, y = after_stat(density)), breaks = seq(0, 1, 0.02)) + xlim(c(0.5, 1)) + # el histograma es de ancho 0.02 y el número de simulaciones 4000 geom_line(data = teorica, aes(x = x, y = f_dens), colour = &quot;red&quot;, size = 1.3) Ejemplo Supongamos que las \\(X_i\\)’s son independientes y exponenciales con tasa \\(\\lambda &gt; 0\\). ¿Cuál es la distribución de muestreo de la suma \\(S = X_1 + \\cdots + X_n\\)? Sabemos que la suma de exponenciales independientes es una distribución gamma con parámetros \\((n, \\lambda)\\), y esta es la distribución de muestreo de nuestra estadística \\(S\\) bajo las hipótesis que hicimos. Podemos checar este resultado con simulación, por ejemplo para una muestra de tamaño \\(n=15\\) con \\(\\lambda = 1\\): replicar_muestreo_exp &lt;- function(est = mean, m, n = 150, lambda = 1){ valores_est &lt;- map_dbl(1:m, ~ est(rexp(n, lambda))) tibble(id_muestra = 1:m, estimacion = valores_est) } sim_estimador_1 &lt;- replicar_muestreo_exp(sum, 4000, n = 15) teorica &lt;- tibble(x = seq(0, 35, 0.001)) |&gt; mutate(f_dens = dgamma(x, shape = 15, rate = 1)) # graficar aprox de distribución de muestreo ggplot(sim_estimador_1) + geom_histogram(aes(x = estimacion, y = after_stat(density)), bins = 35) + geom_line(data = teorica, aes(x = x, y = f_dens), colour = &quot;red&quot;, linewidth = 1.2) Teorema central del límite Si consideramos los ejemplos de arriba donde tratamos con estimadores basados en una suma, total o una media —y en menor medida cuantiles muestrales—, vimos que las distribución de muestreo de las estadísticas que usamos tienden a tener una forma común. Estas son manifestaciones de una regularidad estadística importante que se conoce como el teorema central del límite: las distribuciones de muestreo de sumas y promedios son aproximadamente normales cuando el tamaño de muestra es suficientemente grande. Teorema central del límite Si \\(X_1,X_2, \\ldots, X_n\\) son independientes e idénticamente distribuidas con media \\(\\mu\\) y desviación estándar \\(\\sigma\\) finitas. Si el tamaño de muestra \\(n\\) es grande, entonces la distribución de muestreo de la media \\[\\bar{X} = \\frac{X_1 + X_2 +\\cdots + X_n}{n}\\] es aproximadamente normal con media \\(\\mu\\) y desviación estándar \\(\\sigma/\\sqrt{n}\\), que escribimos como \\[\\bar{X} \\xrightarrow{} \\mathsf{N}\\left( \\mu, \\frac{\\sigma}{\\sqrt{n}} \\right)\\] Adicionalmente, la distribución de la media estandarizada converge a una distribución normal estándar cuando \\(n\\) es grande: \\[\\sqrt{n} \\, \\left( \\frac{\\bar{X}-\\mu}{\\sigma} \\right) \\xrightarrow{} \\mathsf{N}(0, 1)\\] El error estándar de \\(\\bar{X}\\) es \\(\\text{ee}(\\bar{X}) = \\frac{\\sigma}{\\sqrt{n}}\\). Si tenemos una muestra, podemos estimar \\(\\sigma\\) con de la siguiente forma: \\[\\hat{\\sigma} =\\sqrt{\\frac{1}{n}\\sum_{i=1}^n (X_i - \\bar{X})^2}\\] o el más común (que explicaremos más adelante) \\[\\hat{s} = \\sqrt{\\frac{1}{n-1}\\sum_{i=1}^n (X_i - \\bar{X})^2}\\] Este hecho junto con el teorema del límite central nos dice cuál es la dispersión, y cómo se distribuyen las posibles desviaciones de la media muestral alrededor de la verdadera media poblacional. ¿Qué tan grande debe ser \\(n\\). Depende de cómo es la población. Cuando la población tiene una distribución muy sesgada, por ejemplo, \\(n\\) típicamente necesita ser más grande que cuando la población es simétrica si queremos obtener una aproximación “buena”. En algunos textos se afirma que \\(n\\geq 30\\) es suficiente para que la aproximación del Teorema central del límite (TCL) sea buena siempre y cuando la distribución poblacional no sea muy sesgada. Esta regla es más o menos arbitraria y es mejor no confiarse, pues fácilmente puede fallar. En la práctica es importante checar este supuesto, por ejemplo usando remuestreo (que veremos más adelante) Revisa los ejemplos que hemos visto hasta ahora (precios de casas, simulaciones de uniformes y exponenciales según las distintas estadísticas que consideramos). ¿Qué distribuciones de muestreo parecen tener una distribución normal? ¿Cómo juzgamos si estas distribuciones están cerca o lejos de una distribución normal? Normalidad y gráficas de cuantiles normales Para checar si una distribución de datos dada es similar a la normal, la herramienta mas común en estádística es la gráfica de cuantiles teóricos, que es una generalización de la gráfica de cuantiles que vimos anteriormente. En primer lugar, definimos la función de cuantiles de una distribución teórica, que es análoga a la que definimos para conjuntos de datos: Supongamos que tenemos una distribución acumulada teórica \\(\\Phi\\). Podemos definir el cuantil-\\(f\\) \\(q(f)\\) de \\(\\Phi\\) como el valor \\(q(f)\\) tal que \\[q(f) = \\text{argmin}\\{x \\,| \\, \\Phi(x)\\geq f \\}\\] En el caso de que \\(\\Phi\\) tiene densidad \\(\\phi\\), y su soporte es un intervalo (que puede ser de longitud infinita), entonces podemos también escribir \\(q(f)\\) como el valor único donde acumulamos \\(f\\) de la probabilidad \\[\\int_{-\\infty}^{q(f)} \\phi(x)\\,dx= f\\] Por ejemplo, para una densidad normal, abajo mostramos los cuantiles \\(f=0.5\\) (mediana) y \\(f=0.95\\) densidad_tbl &lt;- tibble(x = seq(0, 10, 0.01)) |&gt; mutate(densidad = dnorm(x, 5, 1)) # qnorm es la función de cuantiles de una normal cuantil_50 &lt;- qnorm(0.50, 5, 1) cuantil_90 &lt;- qnorm(0.95, 5, 1) # graficamos densidad_tbl &lt;- densidad_tbl |&gt; mutate(menor_50 = x &gt;= cuantil_50) |&gt; mutate(menor_90 = x &gt;= cuantil_90) g_normal_50 &lt;- ggplot(densidad_tbl, aes(y = densidad)) + ylab(&#39;f(x)&#39;) + geom_area(aes(x = x, fill = menor_50)) + geom_line(aes(x = x), alpha = 0.1) + geom_vline(xintercept = cuantil_50) + theme(legend.position = &quot;none&quot;) + annotate(&quot;text&quot;, 4.3, 0.2, label = &quot;50%&quot;) + labs(subtitle = paste0(&quot;q(0.5)=&quot;, round(cuantil_50,1))) g_normal_90 &lt;- ggplot(densidad_tbl, aes(y = densidad)) + ylab(&#39;f(x)&#39;) + geom_area(aes(x = x, fill = menor_90)) + geom_line(aes(x = x), alpha = 0.1) + geom_vline(xintercept = cuantil_90) + theme(legend.position = &quot;none&quot;) + annotate(&quot;text&quot;, 5.0, 0.2, label = &quot;95%&quot;) + labs(subtitle = paste0(&quot;q(0.95)=&quot;, round(cuantil_90,1))) g_normal_50 + g_normal_90 Como todas las distribuciones normales tienen la misma forma, y para obtener una de otra solo basta reescalar y desplazar, para calcular los cuantiles de una variable con distribución normal \\(\\mathsf{N}(\\mu, \\sigma)\\) sólo tenemos que saber los cuantiles de la distribución normal estándar \\(\\mathsf{N}(0,1)\\) y escalarlos apropiadamente por su media y desviación estándar \\[q(f, \\mu, \\sigma) = \\mu + \\sigma q(f, 0, 1)\\] Puedes demostrar esto sin mucha dificultad empezando con \\(P(X\\leq q) = f\\) y estandarizando: \\[P(X\\leq q(f, \\mu, \\sigma)) = f \\implies P\\left (Z\\leq \\frac{q(f,\\mu,\\sigma) - \\mu}{\\sigma}\\right)=f\\] y esto implica que \\[q(f, 0, 1) = \\frac{q(f,\\mu,\\sigma) - \\mu}{\\sigma} \\implies q(f, \\mu, \\sigma) = \\mu + \\sigma q(f, 0, 1)\\] De modo que si graficáramos los cuantiles de una distribución \\(\\mathsf{N}(\\mu, \\sigma)\\) contra los cuantiles de una distribución \\(\\mathsf{N}(0,1)\\), estos cuantiles aparecen en una línea recta: comparacion_tbl &lt;- tibble(f = seq(0.01, 0.99, 0.01)) |&gt; mutate(cuantiles_normal = qnorm(f, 5, 3), cuantiles_norm_estandar = qnorm(f, 0, 1)) ggplot(comparacion_tbl, aes(cuantiles_norm_estandar, cuantiles_normal)) + geom_point() Ahora supongamos que tenemos una muestra \\(X_1, \\ldots, X_n\\). ¿Cómo podemos checar si estos datos tienen una distribución aproximadamente normal? Si la muestra tiene una distribución aproximadamente \\(\\mathsf{N}(\\mu, \\sigma)\\), entonces sus cuantiles muestrales y los cuantiles respectivos de la normal estándar están aproximadamente en una línea recta. Primero veamos un ejemplo donde los datos son generados según una normal. set.seed(21) muestra &lt;- tibble(x_1 = rnorm(60, 10, 3), x_2 = rgamma(60, 2, 5)) graf_1 &lt;- ggplot(muestra, aes(sample = x_1)) + geom_qq(distribution = stats::qnorm) + geom_qq_line(colour = &quot;red&quot;) graf_2 &lt;- ggplot(muestra, aes(sample = x_2)) + geom_qq(distribution = stats::qnorm) + geom_qq_line(colour = &quot;red&quot;) graf_1 + graf_2 ¿Cuáles son los datos aproximadamente normales? ¿Cómo interpretas las desviaciones de la segunda gráfica en términos de la forma de la distribución normal? Prueba de hipótesis de normalidad Para interpretar las gráficas de cuantiles normales se requiere práctica, pues claramente los datos, aún cuando provengan de una distribución normal, no van a caer justo sobre una línea recta y observaremos variabilidad. Esto no descarta necesariamente que los datos sean aproximadamente normales. Con la práctica, generalmente esta gráfica nos da una buena indicación si el supuesto de normalidad es apropiado. Sin embargo, podemos hacer una prueba de hipótesis formal de normalidad si quisiéramos. La hipótesis nula es la siguiente: Los datos provienen de una distribución normal, y las desviaciones que observamos de una línea recta se deben a variación muestral. Podemos generar datos nulos tomando la media y desviación estándar muestrales, y generando muestras normales \\(\\mathsf{N}(\\bar{x}, s)\\). Usamos el lineup, produciendo datos bajo la hipótesis nula y viendo si podemos distinguir los datos. Por ejemplo: library(nullabor) lineup_normal &lt;- lineup(null_dist(&quot;x_2&quot;, dist = &quot;normal&quot;), muestra) ggplot(lineup_normal, aes(sample = x_2)) + geom_qq(distribution = stats::qnorm) + geom_qq_line(colour = &quot;red&quot;) + facet_wrap(~ .sample) En esta gráfica claramente rechazaríamos la hipótesis de normalidad. Sin embargo, para la primera muestra, obtenemos: lineup_normal &lt;- lineup(null_dist(&quot;x_1&quot;, dist = &quot;normal&quot;), muestra) ggplot(lineup_normal, aes(sample = x_1)) + geom_qq(distribution = stats::qnorm) + geom_qq_line(colour = &quot;red&quot;) + facet_wrap(~ .sample) Los datos verdaderos están en attr(lineup_normal, &quot;pos&quot;) ## [1] 4 Ejemplo Consideremos el problema de estimar el total poblacional de los precios de las casas que se vendieron. El estimador que usamos fue la suma muestral expandida por un factor. Vamos a checar qué tan cerca de la normalidad está la distribución de meustreo de esta estadística (\\(n=250\\)): replicaciones_2 ## # A tibble: 1,500 × 4 ## id_muestra total_muestra factor_exp est_total_millones ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 47089. 4.58 215. ## 2 2 45654. 4.58 209. ## 3 3 43973. 4.58 201. ## 4 4 45665. 4.58 209. ## 5 5 43551. 4.58 199. ## 6 6 46066. 4.58 211. ## 7 7 46626. 4.58 213. ## 8 8 47944. 4.58 219. ## 9 9 45381. 4.58 208. ## 10 10 46519. 4.58 213. ## # ℹ 1,490 more rows ggplot(replicaciones_2, aes(sample = est_total_millones)) + geom_qq(alpha = 0.3) + geom_qq_line(colour = &quot;red&quot;) Y vemos que en efecto el TCL aplica en este ejemplo, y la aproximación es buena. Aunque la población original es sesgada, la descripción de la distribución de muestreo es sorprendemente compacta: La distribución de muestreo de nuestro estimador del total \\(\\hat{t}\\) es aproximadamente normal con media \\(\\bar{x}\\) y desviación estándar \\(s\\), donde: mu &lt;- mean(replicaciones_2$est_total_millones) s &lt;- sd(replicaciones_2$est_total_millones) c(mu = mu, s = s) |&gt; round(2) ## mu s ## 209.90 5.24 Estas cantidades están en millones de dólares. Ejemplo Supongamos que queremos calcular la probabilidad que la suma de 30 variables uniformes en \\([0,1]\\) independientes sea mayor que 18. Podríamos aproximar esta cantidad usando simulación. Otra manera de aproximar esta cantidad es con el TCL, de la siguiente forma: Si \\(S=X_1 + X_2 + X_{30}\\), entonces la media de \\(S\\) es 15 (¿cómo se calcula?) y su desviación estándar es \\(\\sqrt{\\frac{30}{12}}\\). La suma es entonces aproximadamente \\(\\mathsf{N}\\left(15, \\sqrt{\\frac{30}{12}}\\right)\\). Entonces \\[P(S &gt; 18) = P \\left (\\frac{S - 15}{\\sqrt{\\frac{30}{12}}} &gt; \\frac{18 - 15}{\\sqrt{\\frac{30}{12}}}\\right) \\approx P(Z &gt; 1.897)\\] donde \\(Z\\) es normal estándar. Esta última cantidad la calculamos usando la función de distribución de la normal estándar, y nuestra aproximación es 1 - pnorm(1.897) ## [1] 0.02891397 Podemos checar nuestro cálculo usando simulación: tibble(n_sim = 1:100000) |&gt; mutate(suma = map_dbl(n_sim, ~ sum(runif(30)))) |&gt; summarise(prob_may_18 = mean(suma &gt; 18), .groups = &quot;drop&quot;) ## # A tibble: 1 × 1 ## prob_may_18 ## &lt;dbl&gt; ## 1 0.0280 Y vemos que la aproximación normal es buena para fines prácticos. Usando simulaciones haz un histograma que aproxime la distribución de muestreo de \\(S\\). Haz una gráfica de cuantiles normales para checar la normalidad de esta distribución. Ejemplo Cuando el sesgo de la distribución poblacional es grande, puede ser necesario que \\(n\\) sea muy grande para que la aproximación normal sea aceptable para el promedio o la suma. Por ejemplo, si tomamos una gamma con parámetro de forma chico, \\(n = 30\\) no es suficientemente bueno, especialmente si quisiéramos aproximar probabilidades en las colas de la distribución: sims_gamma &lt;- map_df(1:2000, ~ tibble(suma = sum(rgamma(30, 0.1, 1))), .id = &quot;n_sim&quot;) ggplot(sims_gamma, aes(x = suma)) + geom_histogram() Más del Teorema central del límite El teorema central del límite aplica a situaciones más generales que las del enunciado del teorema básico. Por ejemplo, aplica a poblaciones finitas (como vimos en el ejemplo de las casas), en 1960 Jaroslav Hajek demostró una versión del TCL bajo muestreo sin reemplazo. Mas allá de la media muestral, el TCL se puede utilizar para más estadísticas ya que muchas pueden verse como promedios, como totales o errores estándar. El TLC se ha generalizado incluso para cuantiles muestrales. Es importante notar que la calidad de la aproximación del TCL depende de características de la población y también del tamaño de muestra \\(n\\). Para ver si el TCL aplica, podemos hacer ejercicios de simulación bajo diferentes supuestos acerca de la población. También veremos más adelante, con remuestreo, maneras de checar si es factible el TCL dependiendo del análisis de una muestra dada que tengamos. El TCL era particularmente importante en la práctica antes de que pudiéramos hacer simulación por computadora. Era la única manera de aproximar y entender la distribución muestral fuera de cálculos analíticos (como los que hicimos para el máximo de un conjunto de uniformes, por ejemplo). Hoy en día, veremos que podemos hacer simulación para obtener respuestas más exactas, particularmente en la construcción de intervalos de confianza, por ejemplo. Dependemos menos de resultados asintóticos, como el TCL. Cuando aproximamos una distribución discreta mediante la distribución normal, conviene hacer correcciones de continuidad, como se explica en (Chihara and Hesterberg 2018), 4.3.2. Referencias "],["intervalos-de-confianza-y-remuestreo.html", "Sección 5 Intervalos de confianza y remuestreo Ejemplo introductorio La idea del bootstrap El principio de plug-in Discusión: propiedades de la distribución bootstrap Error estándar bootstrap e intervalos normales Ejemplo: inventario de casas vendidas Calibración de intervalos de confianza Interpretación de intervalos de confianza Sesgo Intervalos bootstrap de percentiles Bootstrap para dos muestras Bootstrap y otras estadísticas Bootstrap y estimadores complejos: tablas de perfiles Bootstrap y muestras complejas Bootstrap en R Conclusiones y observaciones", " Sección 5 Intervalos de confianza y remuestreo En la sección anterior, vimos el concepto de distribución de muestreo de una estadística que queremos utilizar para estimar un valor poblacional, y vimos que con esta distribución podíamos evaluar qué tan preciso es nuestro estimador evaluando qué tan concentrada está esta distribución alrededor del valor poblacion que queremos estimar. Sin embargo, en los ejemplos que vimos la población era conocida: ya sea que tuviéramos toda la población finita disponible (como el ejemplo de las casas), o donde la población estaba definida por un modelo teórico de probabilidad (como los ejemplos de las distribuciones uniforme o exponencial). Ahora vemos qué hacer en el caso que realmente nos interesa: solo tenemos una muestra disponible, y la población es desconocida. Todo lo que tenemos es una muestra y una estimación basada en la muestra, y requerimos estimar la distribución de muestreo de la estadística de interés. El enfoque que presentaremos aquí es uno de los más flexibles y poderosos que están disponibles para este problema: el método bootstrap o de remuestreo. En primer lugar explicamos el concepto de intervalo de confianza, que es una manera resumida de evaluar la precisión de nuestras estimaciones. Ejemplo introductorio Regresamos a nuestro ejemplo anterior donde muestreamos 3 grupos, y nos preguntábamos acerca de la diferencia de sus medianas. En lugar de hacer pruebas de permutaciones (ya sea pruebas gráficas o alguna prueba de permutaciones para media o mediana, por ejemplo), podríamos considerar qué tan precisa es cada una de nuestras estimaciones para las medianas de los grupos. Nuestros resultados podríamos presentarlos como sigue. Este código lo explicaremos más adelante, por el momento consideramos la gŕafica resultante: set.seed(8) pob_tab &lt;- tibble(id = 1:2000, x = rgamma(2000, 4, 1), grupo = sample(c(&quot;a&quot;,&quot;b&quot;, &quot;c&quot;), 2000, prob = c(4,2,1), replace = T)) muestra_tab &lt;- pob_tab |&gt; slice_sample(n = 125) g_1 &lt;- ggplot(muestra_tab, aes(x = grupo, y = x)) + geom_boxplot(outlier.alpha = 0) + geom_jitter(alpha = 0.3) + labs(subtitle = &quot;Muestra \\n&quot;) + ylim(c(0,14)) ## Hacemos bootstrap fun_boot &lt;- function(datos){ datos |&gt; group_by(grupo) |&gt; slice_sample(prop = 1, replace = TRUE) } reps_boot &lt;- map_df(1:2000, function(i){ muestra_tab |&gt; fun_boot() |&gt; group_by(grupo) |&gt; summarise(mediana = median(x), .groups = &quot;drop&quot;)}, .id = &#39;rep&#39;) resumen_boot &lt;- reps_boot |&gt; group_by(grupo) |&gt; summarise(ymin = quantile(mediana, 0.025), ymax = quantile(mediana, 0.975), .groups = &quot;drop&quot;) |&gt; left_join(muestra_tab |&gt; group_by(grupo) |&gt; summarise(mediana = median(x))) g_2 &lt;- ggplot(resumen_boot, aes(x = grupo, y = mediana, ymin = ymin, ymax = ymax)) + geom_linerange() + geom_point(colour = &quot;red&quot;, size = 2) + ylim(c(0,14)) + labs(subtitle = &quot;Intervalos de 95% \\n para la mediana&quot;) g_1 + g_2 Donde: En rojo está nuestro = puntual de la mediana de cada grupo (la mediana muestral), y Las segmentos muestran un intervalo de confianza del 95% para nuestra estimación de la mediana: esto quiere decir que los valores poblacionales tienen probabilidad aproximada de 95% de estar dentro del intervalo. Este análisis comunica correctamente que tenemos incertidumbre alta acerca de nuestras estimaciones (especialmente grupos b y c), y que no tenemos mucha evidencia de que el grupo b tenga una mediana poblacional considerablemente más alta que a o c. En muchos casos es más útil presentar la información de esta manera que usando alguna prueba de hipótesis. La idea del bootstrap Como explicamos, el problema que tenemos ahora es que normalmente sólo tenemos una muestra, así que no es posible calcular las distribuciones de muestreo como hicimos en la sección anterior y así evaluar qué tan preciso es nuestro estimador. Sin embargo, podemos hacer lo siguiente: Supongamos que tenemos una muestra \\(X_1,X_2,\\dots, X_n\\) independientes de alguna población desconocida y un estimador \\(T=t(X_1,\\dots, X_n)\\) Mundo poblacional Si tuviéramos la distribución poblacional, simulamos muestras iid para aproximar la distribución de muestreo de nuestro estimador, y así entender su variabilidad. Pero no tenemos la distribución poblacional. Sin embargo, podemos estimar la distribución poblacional con nuestros valores muestrales. Mundo bootstrap Si usamos la estimación del inciso 3, entonces usando el inciso 1 podríamos tomar muestras de nuestros datos muestrales, como si fueran de la población, y usando el mismo tamaño de muestra. El muestreo lo hacemos con reemplazo de manera que produzcamos muestras independientes de la misma “población estimada”, que es la muestra. Evaluamos nuestra estadística en cada una de estas remuestras, a estas les llamamos replicaciones bootstrap. A la distribución de las replicaciones le llamamos distribución bootstrap o distribución de remuestreo del estimador. Usamos la distribución bootstrap para estimar la variabilidad en nuestra estimación con la muestra original. Veamos que sucede para un ejemplo concreto, donde nos interesa estimar la media de los precios de venta de una población de casas. Tenemos nuestra muestra: set.seed(2112) poblacion_casas &lt;- read_csv(&quot;data/casas.csv&quot;) muestra &lt;- slice_sample(poblacion_casas, n = 200, replace = TRUE) mean(muestra$precio_miles) ## [1] 179.963 Esta muestra nos da nuestro estimador de la distribución poblacional: bind_rows(muestra |&gt; mutate(tipo = &quot;muestra&quot;), poblacion_casas |&gt; mutate(tipo = &quot;población&quot;)) |&gt; ggplot(aes(sample = precio_miles, colour = tipo, group = tipo)) + geom_qq(distribution = stats::qunif, alpha = 0.4, size = 1) + facet_wrap(~ tipo) O con histogramas: bind_rows(muestra |&gt; mutate(tipo = &quot;muestra&quot;), poblacion_casas |&gt; mutate(tipo = &quot;población&quot;)) |&gt; ggplot(aes(x = precio_miles, group = tipo)) + geom_histogram(aes(y=..density..), binwidth = 50) + facet_wrap(~ tipo) Y vemos que la aproximación es razonable en las partes centrales de la distribución. Ahora supongamos que nos interesa cuantificar la precisión de nuestra estimación de la media poblacional de precios de casas, y usaremos la media muestral para hacer esto. Para nuestra muestra, nuestra estimación puntual es: media &lt;- mean(muestra$precio_miles) media ## [1] 179.963 Y recordamos que para aproximar la distribución de muestreo podíamos muestrear repetidamente la población y calcular el valor del estimador en cada una de estas muestras. Aquí no tenemos la población, pero tenemos una estimación de la población: la muestra obtenida. Así que para evaluar la variabilidad de nuestro estimador, entramos en el mundo bootstrap, y consideramos que la población es nuestra muestra. Podemos entonces extraer un número grande de muestras con reemplazo de tamaño 200 de la muestra: el muestreo debe ser análogo al que se tomó para nuestra muestra original. Evaluamos nuestra estadística (en este caso la media) en cada una de estas remuestras: media_muestras &lt;- map_dbl(1:5000, ~ muestra |&gt; slice_sample(n = 200, replace = TRUE) |&gt; summarise(media_precio = mean(precio_miles), .groups = &quot;drop&quot;) |&gt; pull(media_precio)) Y nuestra estimación de la distribución de muestreo para la media es entonces: bootstrap &lt;- tibble(media = media_muestras) g_cuantiles &lt;- ggplot(bootstrap, aes(sample = media)) + geom_qq(distribution = stats::qunif) g_histograma &lt;- ggplot(bootstrap, aes(x = media)) + geom_histogram(binwidth = 2) g_cuantiles + g_histograma A esta le llamamos la distribución bootstrap (o de remuestreo) de la media, que definimos más abajo. Ahora podemos calcular un intervalo de confianza del 90% simplemente calculando los cuantiles de esta distribución (no son los cuantiles de la muestra original!): limites_ic &lt;- quantile(media_muestras, c(0.05, 0.95)) |&gt; round() limites_ic ## 5% 95% ## 171 189 Presentaríamos nuestro resultado como sigue: nuestra estimación puntual de la mediana es 180, con un intervalo de confianza del 90% de (171, 189) Otra cosa que podríamos hacer para describir la dispersión de nuestro estimador es calcular el error estándar de remuestreo, que estima el error estándar de la distribución de muestreo: ee_boot &lt;- sd(media_muestras) round(ee_boot, 2) ## [1] 5.39 Definición. Sea \\(X_1,X_2,\\ldots,X_n\\) una muestra independiente y idénticamente distribuida, y \\(T=t(X_1, X_2, \\ldots, X_n)\\) una estadística. Supongamos que sus valores que obervamos son \\(x_1, x_2,\\ldots, x_n\\). La distribución bootstrap, o distribución de remuestreo, de \\(T\\) es la distribución de \\(T^*=t(X_1^*, X_2^*, \\dots X_n^*)\\), donde cada \\(X_i^*\\) se obtiene tomando al azar uno de los valores de \\(x_1,x_2,\\ldots, x_n\\). Otra manera de decir esto es que la remuestra \\(X_1^*, X_2^*, \\ldots, X_n^*\\) es una muestra con reemplazo de los valores observados \\(x_1, x_2, \\ldots, x_n\\) Ejemplo. Si observamos la muestra muestra &lt;- sample(1:20, 5) muestra ## [1] 6 10 7 3 14 Una remuestra se obtiene: sample(muestra, size = 5, replace = TRUE) ## [1] 7 3 7 10 6 Nótese que algunos valores de la muestra original pueden aparecer varias veces, y otros no aparecen del todo. La idea del bootstrap (no paramétrico). La muestra original es una aproximación de la población de donde fue extraída. Así que remuestrear la muestra aproxima lo que pasaría si tomáramos muestras de la población. La distribución de remuestreo de una estadística, que se construye tomando muchas remuestras, aproxima la distribución de muestreo de la estadística. Y el proceso que hacemos es: Remuestreo para una población. Dada una muestra de tamaño \\(n\\) de una población, Obtenemos una remuestra de tamaño \\(n\\) con reemplazo de la muestra original y calculamos la estadística de interés. Repetimos este remuestreo muchas veces (por ejemplo, 10,000). Construímos la distribución bootstrap, y examinamos sus características (dónde está centrada, dispersión y forma). El principio de plug-in La idea básica detrás del bootstrap es el principio de plug-in para estimar parámetros poblacionales: si queremos estimar una cantidad poblacional, calculamos esa cantidad poblacional con la muestra obtenida. Es un principio común en estadística. Por ejemplo, si queremos estimar la media o desviación estándar poblacional, usamos la media muestral o la desviación estándar muestral. Si queremos estimar un cuantil de la población usamos el cuantil correspondiente de la muestra, y así sucesivamente. En todos estos casos, lo que estamos haciendo es: Tenemos una fórmula para la cantidad poblacional de interés en términos de la distribución poblacional. Tenemos una muestra, la distribución que da esta muestra se llama distribución empírica (\\(\\hat{F}(x) = \\frac{1}{n}\\{\\#valores \\le x\\}\\)). Contruimos nuestro estimador, de la cantidad poblacional de interés, “enchufando” la distribución empírica de la muestra en la fórmula del estimador. En el bootstrap aplicamos este principio simple a la distribución de muestreo: Si tenemos la población, podemos calcular la distribución de muestreo de nuestro estimador tomando muchas muestras de la población. Estimamos la poblacion con la muestra y enchufamos en la frase anterior: estimamos la distribución de muestreo de nuestro estimador tomando muchas muestras de la muestra. Nótese que el proceso de muestreo en el último paso debe ser el mismo que se usó para tomar la muestra original. Estas dos imágenes simuladas con base en un ejemplo de Chihara and Hesterberg (2018) muestran lo que acabamos de describir: Figure 5.1: Mundo Real Figure 5.2: Mundo Bootstrap Observación 1. Veremos ejemplos más complejos, pero nótese que si la muestra original son observaciones independientes obtenidas de la distribución poblacional, entonces logramos esto en las remuestras tomando observaciones con reemplazo de la muestra. Igualmente, las remuestras deben ser del mismo tamaño que la muestra original. ¿Porqué no funcionaría tomar muestras sin reemplazo? Piensa si hay independencia entre las observaciones de la remuestra, y cómo serían las remuestras sin reemplazo. ¿Por qué no se puede hacer bootstrap si no conocemos cómo se obtuvo la muestra original? Observación 2. Estos argumentos se pueden escribir con fórmulas usando por ejemplo la función de distribución acumulada \\(F\\) de la población y su estimador, que es la función empírica \\(\\hat{F}\\). Si \\(\\theta = t(F)\\) es una cantidad poblacional que queremos estimar, su estimador plug-in es \\(\\hat{\\theta} = t(\\hat{F})\\). Observación 3: La distribución empírica \\(\\hat{F}\\) es un estimador “razonable” de la distribución poblacional \\(F,\\) pues por el teorema de Glivenko-Cantelli (ver Wasserman (2013), o aquí), \\(\\hat{F}\\) converge a \\(F\\) cuando el tamaño de muestra \\(n\\to\\infty\\), lo cual es intuitivamente claro. Ejemplo En el ejemplo de tomadores de té, podemos estimar la proporción de tomadores de té que prefiere el té negro usando nuestra muestra: te &lt;- read_csv(&quot;data/tea.csv&quot;) |&gt; rowid_to_column() |&gt; select(rowid, Tea, sugar) te |&gt; mutate(negro = ifelse(Tea == &quot;black&quot;, 1, 0)) |&gt; summarise(prop_negro = mean(negro), n = length(negro), .groups = &quot;drop&quot;) ## # A tibble: 1 × 2 ## prop_negro n ## &lt;dbl&gt; &lt;int&gt; ## 1 0.247 300 ¿Cómo evaluamos la precisión de este estimador? Supondremos que el estudio se hizo tomando una muestra aleatoria simple de tamaño 300 de la población de tomadores de té que nos interesa. Podemos entonces usar el bootstrap: # paso 1: define el estimador calc_estimador &lt;- function(datos){ prop_negro &lt;- datos |&gt; mutate(negro = ifelse(Tea == &quot;black&quot;, 1, 0)) |&gt; summarise(prop_negro = mean(negro), n = length(negro), .groups = &quot;drop&quot;) |&gt; pull(prop_negro) prop_negro } # paso 2: define el proceso de remuestreo muestra_boot &lt;- function(datos){ #tomar muestra con reemplazo del mismo tamaño slice_sample(datos, prop = 1, replace = TRUE) } # paso 3: remuestrea y calcula el estimador prop_negro_tbl &lt;- tibble(prop_negro = map_dbl(1:10000, ~ calc_estimador(muestra_boot(datos = te)))) # paso 4: examina la distribución bootstrap prop_negro_tbl |&gt; ggplot(aes(x = prop_negro)) + geom_histogram(bins = 15) + geom_vline(xintercept = calc_estimador(te), color = &quot;red&quot;) Y podemos evaluar varios aspectos, por ejemplo dónde está centrada y qué tan dispersa es la distribución bootstrap: prop_negro_tbl |&gt; summarise(media = mean(prop_negro), ee = sd(prop_negro), cuantil_75 = quantile(prop_negro, 0.75), cuantil_25 = quantile(prop_negro, 0.25), .groups = &quot;drop&quot;) |&gt; mutate(across(where(is.numeric), round, 3)) |&gt; pivot_longer(cols = everything()) ## # A tibble: 4 × 2 ## name value ## &lt;chr&gt; &lt;dbl&gt; ## 1 media 0.247 ## 2 ee 0.025 ## 3 cuantil_75 0.263 ## 4 cuantil_25 0.23 –&gt; –&gt; Discusión: propiedades de la distribución bootstrap Uasremos la distribución bootstrap principalmente para evaluar la variabilidad de nuestros estimadores (y también otros aspectos como sesgo) estimando la dispersión de la distribución de muestreo. Sin embargo, es importante notar que no la usamos, por ejemplo, para saber dónde está centrada la distribución de muestreo, o para “mejorar” la estimación remuestreando. Ejemplo En este ejemplo, vemos 20 muestras de tamaño 200, y evaluamos cómo se ve la aproximación a la distribución de la población (rojo): Podemos calcular las distribuciones de remuestreo (bootstrap) para cada muestra, y compararlas con la distribución de muestreo real. # paso 1: define el estimador calc_estimador &lt;- function(datos){ media_precio &lt;- datos |&gt; summarise(media = mean(precio_miles), .groups = &quot;drop&quot;) |&gt; pull(media) media_precio } # paso 2: define el proceso de remuestreo muestra_boot &lt;- function(datos, n = NULL){ #tomar muestra con reemplazo del mismo tamaño if(is.null(n)){ m &lt;- slice_sample(datos, prop = 1, replace = TRUE)} else { m &lt;- slice_sample(datos, n = n, replace = TRUE) } m } dist_boot &lt;- datos_sim |&gt; filter(tipo == &quot;muestras&quot;) |&gt; select(precio_miles, rep) |&gt; group_by(rep) |&gt; nest() |&gt; mutate(precio_miles = map(data, function(data){ tibble(precio_miles = map_dbl(1:1000, ~ calc_estimador(muestra_boot(data)))) })) |&gt; select(rep, precio_miles) |&gt; unnest() dist_muestreo &lt;- datos_sim |&gt; filter(tipo == &quot;población&quot;) |&gt; group_by(rep) |&gt; nest() |&gt; mutate(precio_miles = map(data, function(data){ tibble(precio_miles = map_dbl(1:1000, ~ calc_estimador(muestra_boot(data, n = 200)))) })) |&gt; select(rep, precio_miles) |&gt; unnest() Obsérvese que: En algunos casos la aproximación es mejor que en otros (a veces la muestra tiene valores ligeramente más altos o más bajos). La dispersión de cada una de estas distribuciones bootstrap es similar a la de la verdadera distribución de muestreo (en rojo), pero puede está desplazada dependiendo de la muestra original que utilizamos. Adicionalmente, los valores centrales de la distribución de bootstrap tiende cubrir el verdadero valor que buscamos estimar, que es: poblacion_casas |&gt; summarise(media = mean(precio_miles), .groups = &quot;drop&quot;) ## # A tibble: 1 × 1 ## media ## &lt;dbl&gt; ## 1 183. Variación en distribuciones bootstrap En el proceso de estimación bootstrap hay dos fuentes de variación pues: La muestra original se selecciona con aleatoriedad de una población. Las muestras bootstrap se seleccionan con aleatoriedad de la muestra original. Esto es: La estimación bootstrap ideal es un resultado asintótico \\(B=\\infty\\), en esta caso \\(\\hat{\\textsf{se}}_B\\) iguala la estimación plug-in \\(se_{P_n}\\). En el proceso de bootstrap podemos controlar la variación del segundo aspecto, conocida como implementación de muestreo Monte Carlo, y la variación Monte Carlo decrece conforme incrementamos el número de muestras. Podemos eliminar la variación Monte Carlo si seleccionamos todas las posibles muestras con reemplazo de tamaño \\(n\\), hay \\({2n-1}\\choose{n}\\) posibles muestras y si seleccionamos todas obtenemos \\(\\hat{\\textsf{se}}_\\infty\\) (bootstrap ideal), sin embargo, en la mayor parte de los problemas no es factible proceder así. En la siguiente gráfica mostramos 6 posibles muestras de tamaño 50 simuladas de la población, para cada una de ellas se graficó la distribución empírica y se se realizan histogramas de la distribución bootstrap con \\(B=30\\) y \\(B=1000\\), en cada caso hacemos dos repeticiones, notemos que cuando el número de muestras bootstrap es grande las distribuciones bootstrap son muy similares (para una muestra de la población dada), esto es porque disminuimos el erro Monte Carlo. También vale la pena recalcar que la distribución bootstrap está centrada en el valor observado en la muestra (línea azúl punteada) y no en el valor poblacional sin embargo la forma de la distribución es similar a lo largo de las filas. Entonces, ¿cuántas muestras bootstrap? Incluso un número chico de replicaciones bootstrap, digamos \\(B=25\\) es informativo, y \\(B=50\\) con frecuencia es suficiente para dar una buena estimación de \\(se_P(\\hat{\\theta})\\) (Efron and Tibshirani (1993)). Cuando se busca estimar error estándar Chihara and Hesterberg (2018) recomienda \\(B=1000\\) muestras, o \\(B=10,000\\) muestras dependiendo la presición que se busque. Error estándar bootstrap e intervalos normales Ahora podemos construir nuestra primera versión de intervalos de confianza basados en la distribución bootstrap. Supongamos que queremos estimar una cantidad poblacional \\(\\theta\\) con una estadística \\(\\hat{\\theta} = t(X_1,\\ldots, X_n)\\), donde \\(X_1,\\ldots, X_n\\) es una muestra independiente e idénticamente distribuida de la población. Suponemos además que la distribución muestral de \\(\\hat{\\theta}\\) es aproximadamente normal (el teorema central del límite aplica), y está centrada en el verdadero valor poblacional \\(\\theta\\). Ahora queremos construir un intervalo que tenga probabilidad 95% de cubrir al valor poblacional \\(\\theta\\). Tenemos que \\[P(-2\\mathsf{ee}(\\hat{\\theta}) &lt; \\hat{\\theta} - \\theta &lt; 2\\mathsf{ee}(\\hat{\\theta})) \\approx 0.95\\] por las propiedades de la distribución normal (\\(P(-2\\sigma &lt; X -\\mu &lt; 2\\sigma)\\approx 0.95\\) si \\(X\\) es normal con media \\(\\mu\\) y desviación estándar \\(\\sigma\\)). Entonces \\[P(\\hat{\\theta} - 2\\mathsf{ee}(\\hat{\\theta}) &lt; \\theta &lt; \\hat{\\theta} + 2\\mathsf{ee}(\\hat{\\theta})) \\approx 0.95\\] Es decir, la probabilidad de que el verdadero valor poblacional \\(\\theta\\) esté en el intervalo \\[[\\hat{\\theta} - 2\\mathsf{ee}(\\hat{\\theta}), \\hat{\\theta} + 2\\mathsf{ee}(\\hat{\\theta})]\\] es cercano a 0.95. En este intervalo no conocemos el error estándar (es la desviación estándar de la distribución de muestreo de \\(\\hat{\\theta}\\)), y aquí es donde entre la distribución bootstrap, que aproxima la distribución de muestreo. Lo estimamos con \\[\\hat{\\mathsf{ee}}_{\\textrm{boot}}(\\hat{\\theta})\\] que es la desviación estándar de la distribución bootsrap. Definición. El error estándar bootstrap \\(\\hat{\\mathsf{ee}}_{\\textrm{boot}}(\\hat{\\theta})\\) se define como la desviación estándar de la distribución bootstrap de \\(\\theta\\). El intervalo de confianza normal bootstrap al 95% está dado por \\[[\\hat{\\theta} - 2\\hat{\\mathsf{ee}}_{\\textrm{boot}}(\\hat{\\theta}), \\hat{\\theta} + 2\\hat{\\mathsf{ee}}_{\\textrm{boot}}(\\hat{\\theta})].\\] Nótese que hay varias cosas qué revisar aquí: que el teorema central del límite aplica y que la distribución de muestreo de nuestro estimador está centrado en el valor verdadero. Esto en algunos casos se puede demostrar usando la teoría, pero más abajo veremos comprobaciones empíricas. Ejemplo: tomadores de té negro Consideremos la estimación que hicimos de el procentaje de tomadores de té que toma té negro: # paso 1: define el estimador calc_estimador &lt;- function(datos){ prop_negro &lt;- datos |&gt; mutate(negro = ifelse(Tea == &quot;black&quot;, 1, 0)) |&gt; summarise(prop_negro = mean(negro), n = length(negro)) |&gt; pull(prop_negro) prop_negro } prop_hat &lt;- calc_estimador(te) prop_hat |&gt; round(2) ## [1] 0.25 Podemos graficar su distribución bootstrap —la cual simulamos arriba—. g_hist &lt;- ggplot(prop_negro_tbl, aes(x = prop_negro)) + geom_histogram(bins = 15) g_qq_normal &lt;- ggplot(prop_negro_tbl, aes(sample = prop_negro)) + geom_qq() + geom_qq_line(colour = &quot;red&quot;) g_hist + g_qq_normal Y notamos que la distribución bootstrap es aproximadamente normal. Adicionalmente, vemos que el sesgo tiene un valor estimado de: media_boot &lt;- prop_negro_tbl |&gt; pull(prop_negro) |&gt; mean() media_boot - prop_hat ## [1] 0.0004393333 De esta forma, hemos verificado que: La distribución bootstrap es aproximadamente normal (ver gráfica de cuantiles normales); La distribución bootstrap es aproximadamente insesgada. Lo cual nos lleva a construir intervalos de confianza basados en la distribución normal. Estimamos el error estándar con la desviación estándar de la distribución bootstrap ee_boot &lt;- prop_negro_tbl |&gt; pull(prop_negro) |&gt; sd() ee_boot ## [1] 0.02485138 y construimos un intervalo de confianza del 95%: intervalo_95 &lt;- c(prop_hat - 2 * ee_boot, prop_hat + 2 * ee_boot) intervalo_95 |&gt; round(2) ## [1] 0.2 0.3 Este intervalo tiene probabilidad del 95% de capturar al verdadero poblacional. Con alta probabilidad, entonces, el porcentaje de tomadores de té en la población está entre 0.2 y 0.3. Ejemplo: inventario de casas vendidas Ahora consideremos el problema de estimar el total del valor de las casas vendidas en un periodo. Tenemos una muestra de tamaño \\(n=150\\): # muestra original set.seed(121) muestra_casas &lt;- slice_sample(poblacion_casas, n = 150) # paso 1: define el estimador calc_estimador_casas &lt;- function(datos){ N &lt;- nrow(poblacion_casas) n &lt;- nrow(datos) total_muestra &lt;- sum(datos$precio_miles) estimador_total &lt;- (N / n) * total_muestra estimador_total } # paso 2: define el proceso de remuestreo muestra_boot &lt;- function(datos){ #tomar muestra con reemplazo del mismo tamaño slice_sample(datos, prop = 1, replace = TRUE) } # paso 3: remuestrea y calcula el estimador totales_boot &lt;- tibble(total_boot = map_dbl(1:5000, ~ calc_estimador_casas(muestra_boot(muestra_casas)))) # paso 4: examina la distribución bootstrap g_hist &lt;- totales_boot |&gt; ggplot(aes(x = total_boot)) + geom_histogram() g_qq &lt;- totales_boot |&gt; ggplot(aes(sample = total_boot)) + geom_qq() + geom_qq_line(colour = &quot;red&quot;) + geom_hline(yintercept = quantile(totales_boot$total_boot, 0.975), colour = &quot;gray&quot;) + geom_hline(yintercept = quantile(totales_boot$total_boot, 0.025), colour = &quot;gray&quot;) g_hist + g_qq En este caso, distribución de muestreo presenta cierta asimetría, pero la desviación no es grande. En la parte central la aproximación normal es razonable. Procedemos a revisar sesgo total_est &lt;- calc_estimador_casas(muestra_casas) sesgo &lt;- mean(totales_boot$total_boot) - total_est sesgo ## [1] 110.0851 Este número puede parecer grande, pero sí calculamos la desviación relativa con respecto al error estándar vemos que es chico en la escala de la distribución bootstrap: ee_boot &lt;- sd(totales_boot$total_boot) sesgo_relativo &lt;- sesgo / ee_boot sesgo_relativo ## [1] 0.01522088 De forma que procedemos a construir intervalos de confianza como sigue : c(total_est - 2*ee_boot, total_est + 2*ee_boot) ## [1] 203366.6 232296.6 Que está en miles de dólares. En millones de dólares, este intervalo es: intervalo_total &lt;- c(total_est - 2*ee_boot, total_est + 2*ee_boot) / 1000 intervalo_total |&gt; round(1) ## [1] 203.4 232.3 Así que con 95% de confianza el verdadero total del valor de las casas vendidas está entre 203 y 232 millones de dólares. Nota: en este ejemplo mostraremos una alternativa de intervalos de confianza que es más apropiado cuando observamos asimetría. Sin embargo, primero tendremos que hablar de dos conceptos clave con respecto a intervalos de confianza: calibración e interpretación. Calibración de intervalos de confianza ¿Cómo sabemos que nuestros intervalos de confianza del 95% nominal tienen cobertura real de 95%? Es decir, tenemos que checar: El procedimiento para construir intervalos debe dar intervalos tales que el valor poblacional está en el intervalo de confianza para 95% de las muestras. Como solo tenemos una muestra, la calibración depende de argumentos teóricos o estudios de simulación previos. Para nuestro ejemplo de casas tenemos la población, así que podemos checar qué cobertura real tienen los intervalos normales: simular_intervalos &lt;- function(rep, size = 150){ muestra_casas &lt;- slice_sample(poblacion_casas, n = size) N &lt;- nrow(poblacion_casas) n &lt;- nrow(muestra_casas) total_est &lt;- (N / n) * sum(muestra_casas$precio_miles) # paso 1: define el estimador calc_estimador_casas &lt;- function(datos){ total_muestra &lt;- sum(datos$precio_miles) estimador_total &lt;- (N / n) * total_muestra estimador_total } # paso 2: define el proceso de remuestreo muestra_boot &lt;- function(datos){ #tomar muestra con reemplazo del mismo tamaño slice_sample(datos, prop = 1, replace = TRUE) } # paso 3: remuestrea y calcula el estimador totales_boot &lt;- map_dbl(1:2000, ~ calc_estimador_casas(muestra_boot(muestra_casas))) |&gt; tibble(total_boot = .) |&gt; summarise(ee_boot = sd(total_boot)) |&gt; mutate(inf = total_est - 2*ee_boot, sup = total_est + 2*ee_boot) |&gt; mutate(rep = rep) totales_boot } # Para recrear, correr: # sims_intervalos &lt;- map(1:100, ~ simular_intervalos(rep = .x)) # write_rds(sims_intervalos, &quot;cache/sims_intervalos.rds&quot;) # Para usar resultados en cache: sims_intervalos &lt;- read_rds(&quot;cache/sims_intervalos.rds&quot;) total &lt;- sum(poblacion_casas$precio_miles) sims_tbl &lt;- sims_intervalos |&gt; bind_rows() |&gt; mutate(cubre = inf &lt; total &amp; total &lt; sup) ggplot(sims_tbl, aes(x = rep)) + geom_hline(yintercept = total, colour = &quot;red&quot;) + geom_linerange(aes(ymin = inf, ymax = sup, colour = cubre)) La cobertura para estos 100 intervalos simulados da total &lt;- sum(poblacion_casas$precio_miles) sims_tbl |&gt; summarise(cobertura = mean(cubre)) ## # A tibble: 1 × 1 ## cobertura ## &lt;dbl&gt; ## 1 0.96 que es consistente con una cobertura real del 95% (¿qué significa “consistente”? ¿Cómo puedes checarlo con el bootstrap?) Observación. En este caso teníamos la población real, y pudimos verificar la cobertura de nuestros intervalos. En general no la tenemos. Estos ejercicios de simulación se pueden hacer con poblaciones sintéticas que se generen con las características que creemos va a tener nuestra población (por ejemplo, sesgo, colas largas, etc.). En general, no importa qué tipo de estimadores o intervalos de confianza usemos, requerimos checar la calibración. Esto puede hacerse con ejercicios de simulación con poblaciones sintéticas y tanto los procedimientos de muestreo como los tamaños de muestra que nos interesa usar. Verificar la cobertura de nuestros intervalos de confianza por medio simulación está bien estudiado para algunos casos. Por ejemplo, cuando trabajamos con estimaciones para poblaciones teóricas. En general sabemos que los procedimientos funcionan bien en casos: con distribuciones simétricas que tengan colas no muy largas; estimación de proporciones donde no tratamos con casos raros o casos seguros (probabilidades cercanas a 0 o 1). Interpretación de intervalos de confianza Como hemos visto, “intervalo de confianza” (de 90% de confianza, por ejemplo) es un término frecuentista, que significa: Cada muestra produce un intervalo distinto. Para el 90% de las muestras posibles, el intervalo cubre al valor poblacional. La afirmación es sobre el intervalo y el mecanismo para construirlo. Así que con alta probabilidad, el intervalo contiene el valor poblacional. Intervalos más anchos nos dan más incertidumbre acerca de dónde está el verdadero valor poblacional (y al revés para intervalos más angostos). Existen también “intervalos de credibilidad” (de 90% de probabilidad, por ejemplo), que se interpetan de forma bayesiana: Con 90% de probabilidad (relativamente alta), creemos que el valor poblacional está dentro del intervalo de credibilidad. Esta última interpretación es más natural. Obsérvese que para hablar de intervalos de confianza frecuentista tenemos que decir: Este intervalo particular cubre o no al verdadero valor, pero nuestro procedimiento produce intervalos que contiene el verdadero valor para el 90% de las muestras. Esta es una interpretación relativamente débil, y muchos intervalos poco útiles pueden satisfacerla. La interpretación bayesiana es más natural porque expresa más claramente incertidumbre acerca del valor poblacional. Por otro lado, La interpretación frecuentista nos da maneras empíricas de probar si los intervalos de confianza están bien calibrados o no: es un mínimo que “intervalos del 90%” deberían satisfacer. Así que tomamos el punto de vista bayesiano en la intepretación, pero buscamos que nuestros intervalos cumplan o aproximen bien garantías frecuentistas (discutimos esto más adelante). Los intervalos que producimos en esta sección pueden interpretarse de las dos maneras. Sesgo Notemos que hemos revisado el sesgo en varias ocasiones, esto es porque algunos estimadores comunes (por ejemplo, cociente de dos cantidades aleatorias) pueden sufrir de sesgo grande, especialmente en el caso de muestras chicas. Esto a su vez afecta la cobertura, pues es posible que nuestros intervalos no tengan “cobertura simétrica”, por ejemplo. Para muchos estimadores, y muestras no muy chicas, esté sesgo tiende a ser poco importante y no es necesario hacer correcciones. Si el tamaño del sesgo es grande comparado con la dispersión de la distribución bootstrap generalmente consideramos que bajo el diseño actual el estimador que estamos usando no es apropiado, y podemos proponer otro estimador u otro procedimiento para construir intervalos (ver Efron and Tibshirani (1993) intervalos BC_{a}), Efron and Tibshirani (1993) sugieren más de 20% de la desviación estándar, mientras que en Chihara and Hesterberg (2018) se sugiere 2% de la desviación estándar. Dependiendo que tan crítico es que los intervalos estén bien calibrados podemos evaluar nuestro problema particular. Intervalos bootstrap de percentiles Retomemos nuestro ejemplo del valor total del precio de las casas. A través de remuestras bootstrap hemos verificado gráficamente que la distribución de remuestreo es ligeramente asimétrica (ver la figura de abajo). Anteriormente hemos calculado intervalos de confianza basados en supuestos normales por medio del error éstandar. Este intervalo está dado por ## [1] 203.4 232.3 y por construcción sabemos que es simétrico con respecto al valor estimado, pero como podemos ver la distribución de muestreo no es simétrica, lo cual podemos confirmar por ejemplo calculando el porcentaje de muestras bootstrap que caen por arriba y por debajo del intervalo construido: ## # A tibble: 1 × 2 ## prop_inf prop_sup ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.0192 0.026 los cuales se han calculado como el porcentaje de medias bootstrap por debajo (arriba) de la cota inferior (superior), y vemos que no coinciden con el nivel de confianza prestablecido (2.5% para cada extremo). Otra opción común que se usa específicamente cuando la distribución bootstrap no es muy cercana a la normal son los intervalos de percentiles bootstrap: Definición. El intervalo de percentiles bootstrap al 95% de confianza está dado por \\[[q_{0.025}, q_{0.975}]\\] donde \\(q_f\\) es el percentil \\(f\\) de la distribución bootstrap. Otros intervalos comunes son el de 80% o 90% de confianza, por ejemplo, que corresponden a \\([q_{0.10}, q_{0.90}]\\) y \\([q_{0.05}, q_{0.95}]\\). Ojo: intervalos de confianza muy alta (por ejemplo 99.5%) pueden tener mala calibración o ser muy variables en su longitud pues dependen del comportamiento en las colas de la distribución. Para el ejemplo de las casas, calcularíamos simplemente intervalo_95 &lt;- totales_boot |&gt; pull(total_boot) |&gt; quantile(probs = c(0.025, 0.975)) / 1000 (intervalo_95) |&gt; round(1) ## 2.5% 97.5% ## 204.3 232.5 que está en millones de dólares. Nótese que es similar al intervalo de error estándar. Otro punto interesante sobre los intervalos bootstrap de percentiles es que lidian naturalmente con la asímetría de la distribución bootstrap. Ilustramos esto con la distancia de las extremos del intervalo con respecto a la media: abs(intervalo_95 - total_est/1000) ## 2.5% 97.5% ## 13.53912 14.64674 Los intervalos de confianza nos permiten presentar un rango de valores posibles para el parámetro de interés. Esto es una notable diferencia con respecto a presentar sólo un candidato como estimador. Nuestra fuente de información son los datos. Es por esto que si vemos valores muy chicos (grandes) en nuestra muestra, el intervalo se tiene que extender a la izquierda (derecha) para compensar dichas observaciones. Explica por qué cuando la aproximación normal es apropiada, el intervalo de percentiles al 95% es muy similar al intervalo normal de 2 errores estándar. Ejemplo Consideramos los datos de propinas. Queremos estimar la media de cuentas totales para la comida y la cena. Podemos hacer bootstrap de cada grupo por separado: # en este ejemplo usamos rsample, pero puedes # escribir tu propio código library(rsample) propinas &lt;- read_csv(&quot;data/propinas.csv&quot;) estimador &lt;- function(split, ...){ muestra &lt;- analysis(split) |&gt; group_by(momento) muestra |&gt; summarise(estimate = mean(cuenta_total), .groups = &#39;drop&#39;) |&gt; mutate(term = momento) } intervalo_propinas_90 &lt;- bootstraps(propinas, strata = momento, 1000) |&gt; mutate(res_boot = map(splits, estimador)) |&gt; int_pctl(res_boot, alpha = 0.10) |&gt; mutate(across(where(is.numeric), round, 2)) intervalo_propinas_90 ## # A tibble: 2 × 6 ## term .lower .estimate .upper .alpha .method ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; ## 1 Cena 19.8 20.8 22.0 0.1 percentile ## 2 Comida 15.6 17.1 18.8 0.1 percentile Nota: .estimate es la media de los valores de la estadística sobre las remuestras, no es el estimador original. De la tabla anterior inferimos que la media en la cuenta en la cena es más grande que la de la comida. Podemos graficar agregando los estimadores plugin: estimadores &lt;- propinas |&gt; group_by(momento) |&gt; rename(term = momento) |&gt; summarise(media = mean(cuenta_total)) ggplot(intervalo_propinas_90, aes(x = term)) + geom_linerange(aes(ymin = .lower, ymax = .upper)) + geom_point(data = estimadores, aes(y = media), colour = &quot;red&quot;, size = 3) + xlab(&quot;Momento&quot;) + ylab(&quot;Media de cuenta total (dólares)&quot;) + labs(subtitle = &quot;Intervalos de 90% para la media&quot;) Nótese que el bootstrap lo hicimos por separado en cada momento del día (por eso el argumento strata en la llamada a bootstraps): Justifica el procedimiento de hacer el bootstrap separado para cada grupo. ¿Qué supuestos acerca del muestreo se deben satisfacer? ¿Deben ser muestras aleatorias simples de cada momento del día, por ejemplo? ¿Qué harías si no fuera así, por ejemplo, si se escogieron al azar tickets de todos los disponibles en un periodo? Bootstrap para dos muestras En el ejemplo anterior consideramos cómo hacer bootstrap cuando tenemos muestras independientes. También podemos aplicarlo a estimadores que comparen directamente las dos muestras: Bootstrap para comparar poblaciones. Dadas muestras independientes de tamaños \\(m\\) y \\(n\\) de dos poblaciones: Extraer una remuestra de tamaño \\(m\\) con reemplazo de la primera muestra y una remuestra separada de tamaño \\(n\\) de la segunda muestra. Calcula la estadística que compara los dos grupos (por ejemplo, diferencia de medias) Repetir este proceso muchas veces (por ejemplo, 1000 - 10000). Construir la distribución bootstrap de la estadística. Examinar dispersión, sesgo y forma. Ejemplo Supongamos que queremos comparar directamente la media de la cuenta total en comida y cena. Podemos hacer: estimador_dif &lt;- function(split, ...){ muestra &lt;- analysis(split) |&gt; group_by(momento) muestra |&gt; summarise(estimate = mean(cuenta_total), .groups = &quot;drop&quot;) |&gt; pivot_wider(names_from = momento, values_from = estimate) |&gt; mutate(estimate = Cena - Comida, term = &quot;diferencia&quot;) } dist_boot &lt;- bootstraps(propinas, strata = momento, 2000) |&gt; mutate(res_boot = map(splits, estimador_dif)) g_1 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(x = estimate)) + geom_histogram(bins = 20) + xlab(&quot;Diferencia Comida vs Cena&quot;) g_2 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(sample = estimate)) + geom_qq() + geom_qq_line(colour = &#39;red&#39;) g_1 + g_2 Y podemos calcular un intervalo de confianza para la diferencia de medias: dist_boot |&gt; int_pctl(res_boot, alpha = 0.01) |&gt; mutate(across(where(is.numeric), round, 2)) |&gt; select(term, .lower, .upper) ## # A tibble: 1 × 3 ## term .lower .upper ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 diferencia 0.73 6.54 Que nos indica que con alta probabilidad las cuentas son más altas que en la cena que en la comida. La diferencia puede ir de un poco menos de un dólar hasta seis dólares con 99% de confianza. Datos pareados En otros casos, las muestras no son independientes y están pareadas. Por ejemplo, este es un estudio dende a 10 personas una noche se les dio una medicina para dormir y otra noche otra medicina. Se registraron cuántas horas de sueño extra comparados con un día donde no tomaron medicina. dormir &lt;- sleep |&gt; pivot_wider(names_from = group, names_prefix = &quot;medicina_&quot;, values_from = extra) dormir ## # A tibble: 10 × 3 ## ID medicina_1 medicina_2 ## &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 0.7 1.9 ## 2 2 -1.6 0.8 ## 3 3 -0.2 1.1 ## 4 4 -1.2 0.1 ## 5 5 -0.1 -0.1 ## 6 6 3.4 4.4 ## 7 7 3.7 5.5 ## 8 8 0.8 1.6 ## 9 9 0 4.6 ## 10 10 2 3.4 En este caso, el bootstrap se hace sobre individuos, y quisiéramos comparar la medición de la medicina_1 con la medicina_2. Usaremos la media de al diferencia entre horas de sueño entre las dos medicinas. Nuestro estimador puntual es: estimador_dif &lt;- dormir |&gt; mutate(dif_2_menos_1 = medicina_2 - medicina_1) |&gt; summarise(dif_media = mean(dif_2_menos_1)) estimador_dif ## # A tibble: 1 × 1 ## dif_media ## &lt;dbl&gt; ## 1 1.58 Esto indica que en promedio duermen hora y media más con la medicina 2 que con la medicina 1. Como hay variabilildad considerable en el número de horas extra de cada medicina dependiendo del individuo, es necesario hacer una intervalo de confianza para descartar que esta diferencia haya aparecido por azar debido a la variación muestral. Nótese que aquí no tenemos estratos, pues solo hay una muestra de individuo con dos mediciones. estimador_dif &lt;- function(split, ...){ muestra &lt;- analysis(split) muestra |&gt; mutate(dif_2_menos_1 = medicina_2 - medicina_1) |&gt; summarise(estimate = mean(dif_2_menos_1), .groups = &quot;drop&quot;) |&gt; mutate(term = &quot;diferencia 2 vs 1&quot;) } dist_boot &lt;- bootstraps(dormir, 2000) |&gt; mutate(res_boot = map(splits, estimador_dif)) g_1 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(x = estimate)) + geom_histogram(bins = 20) g_2 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(sample = estimate)) + geom_qq() + geom_qq_line(colour = &#39;red&#39;) g_1 + g_2 Nuestro intervalo de percentiles al 90% es de dist_boot |&gt; int_pctl(res_boot, 0.10) ## # A tibble: 1 × 6 ## term .lower .estimate .upper .alpha .method ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; ## 1 diferencia 2 vs 1 1.04 1.57 2.22 0.1 percentile Lo que indica con alta probabilidad que la medicina 2 da entre 1 y 2 horas extras de sueño. Nota que en este ejemplo también podríamos hacer una prueba de hipótesis por permutaciones, suponiendo como hipótesis nula que las dos medicinas son equivalentes. Sin embargo, usualmente es más informativo presentar este tipo de intervalos para estimar la diferencia. Bootstrap y otras estadísticas El bootstrap es una técnica versátil. Un ejemplo son estimadores de razón, que tienen la forma \\[ \\hat{r} = \\frac{\\overline y}{\\overline x}\\] Por ejemplo, ¿cómo haríamos estimación para el procentaje de área area habitable de las casas en relación al tamaño del lote? Una manera de estimar esta cantidad es dividiendo la suma del área habitable de nuestra muestra y dividirlo entre la suma del área de los lotes de nuestra muestra, como en la fórmula anterior. Esta fórmula es más difícil pues tanto numerador como denominador tienen variabilidad, y estas dos cantidades no varían independientemente. Con el bootstrap podemos atacar estos problemas Ejemplo: estimadores de razón Nuestra muestra original es: set.seed(250) casas_muestra &lt;- slice_sample(poblacion_casas, n = 200) El estimador de interés es: estimador_razon &lt;- function(split, ...){ muestra &lt;- analysis(split) muestra |&gt; summarise(estimate = sum(area_habitable_sup_m2) / sum(area_lote_m2), .groups = &quot;drop&quot;) |&gt; mutate(term = &quot;% area del lote construida&quot;) } Y nuestra estimación puntual es estimacion &lt;- muestra_casas |&gt; summarise(estimate = sum(area_habitable_sup_m2) / sum(area_lote_m2)) estimacion ## # A tibble: 1 × 1 ## estimate ## &lt;dbl&gt; ## 1 0.148 Es decir que en promedio, un poco más de 15% del lote total es ocupado por área habitable. Ahora hacemos bootstrap para construir un intervalo: dist_boot &lt;- bootstraps(casas_muestra, 2000) |&gt; mutate(res_boot = map(splits, estimador_razon)) g_1 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(x = estimate)) + geom_histogram(bins = 20) g_2 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(sample = estimate)) + geom_qq() + geom_qq_line(colour = &#39;red&#39;) g_1 + g_2 En este caso la cola derecha parece tener menos dispersión que una distribución normal. Usamos un intervalo de percentiles para obtener: dist_boot |&gt; int_pctl(res_boot) |&gt; mutate(estimador = estimacion$estimate) |&gt; rename(media_boot = .estimate) |&gt; mutate(bias = media_boot - estimador) |&gt; pivot_longer(is_double) |&gt; mutate(value = round(value, 3)) ## # A tibble: 6 × 4 ## term .method name value ## &lt;chr&gt; &lt;chr&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 % area del lote construida percentile .lower 0.121 ## 2 % area del lote construida percentile media_boot 0.142 ## 3 % area del lote construida percentile .upper 0.159 ## 4 % area del lote construida percentile .alpha 0.05 ## 5 % area del lote construida percentile estimador 0.148 ## 6 % area del lote construida percentile bias -0.006 De modo que en esta zona, entre 12% y 16% de toda el área disponible es ocupada por área habitable: estas son casas que tienen jardines o terrenos, garage relativamente grandes. Ejemplo: suavizadores Podemos usar el bootstrap para juzgar la variabilidad de un suavizador, que consideramos como nuestra estadística: graf_casas &lt;- function(data){ ggplot(data |&gt; filter(calidad_gral &lt; 7), aes(x = area_habitable_sup_m2)) + geom_point(aes(y = precio_m2_miles), alpha = 0.75) + geom_smooth(aes(y = precio_m2_miles), method = &quot;loess&quot;, span = 0.7, se = FALSE, method.args = list(degree = 1, family = &quot;symmetric&quot;)) } graf_casas(casas_muestra) Podemos hacer bootstrap para juzgar la estabilidad del suavizador: suaviza_boot &lt;- function(x, data){ # remuestreo muestra_boot &lt;- slice_sample(data, prop = 1, replace = T) ajuste &lt;- loess(precio_m2_miles ~ area_habitable_sup_m2, data = muestra_boot, degree = 1, span = 0.7, family = &quot;symmetric&quot;) datos_grafica &lt;- tibble(area_habitable_sup_m2 = seq(25, 250, 5)) ajustados &lt;- predict(ajuste, newdata = datos_grafica) datos_grafica |&gt; mutate(ajustados = ajustados) |&gt; mutate(rep = x) } reps &lt;- map(1:10, ~ suaviza_boot(.x, casas_muestra |&gt; filter(calidad_gral &lt; 7))) |&gt; bind_rows() # ojo: la rutina loess no tienen soporte para extrapolación graf_casas(casas_muestra) + geom_line(data = reps, aes(y = ajustados, group = rep), alpha = 1, colour = &quot;red&quot;) Donde vemos que algunas cambios de pendiente del suavizador original no son muy interpretables (por ejemplo, para áreas chicas) y alta variabilidad en general en los extremos. Podemos hacer más iteraciones para calcular bandas de confianza: reps &lt;- map(1:200, ~ suaviza_boot(.x, casas_muestra |&gt; filter(calidad_gral &lt; 7))) |&gt; bind_rows() # ojo: la rutina loess no tienen soporte para extrapolación graf_casas(casas_muestra) + geom_line(data = reps, aes(y = ajustados, group = rep), alpha = 0.2, colour = &quot;red&quot;) Donde observamos cómo tenemos incertidumbre en cuanto al nivel y forma de las curvas en los extremos de los datos (casas grandes y chicas), lo cual es natural. Aunque podemos resumir para hacer bandas de confianza, mostrar remuestras de esta manera es informativo: por ejempo: vemos cómo es probable también que para casas de menos de 70 metros cuadrados el precio por metro cuadrado no cambia tanto (líneas constantes) Bootstrap y estimadores complejos: tablas de perfiles Podemos regresar al ejemplo de la primera sesión donde calculamos perfiles de los tomadores de distintos tés: en bolsa, suelto, o combinados. Caundo hacemos estos tipos de análisis no es raro que los prefiles tengan variabilidad considerable que es necesario cuantificar. price tea bag tea bag+unpackaged unpackaged promedio p_upscale -0.71 -0.28 0.98 28 p_variable -0.12 0.44 -0.31 36 p_cheap 0.3 -0.53 0.23 2 p_branded 0.62 -0.16 -0.45 25 p_private label 0.72 -0.22 -0.49 5 p_unknown 1.58 -0.58 -1 3 Hacemos bootstrap sobre toda la muestra, y repetimos exactamente el mismo proceso de construción de perfiles: boot_perfiles &lt;- map(1:1000, function(x){ te_boot &lt;- te |&gt; slice_sample(prop = 1, replace = TRUE) calcular_perfiles(te_boot) |&gt; mutate(rep = x) }) |&gt; bind_rows() Ahora resumimos y graficamos, esta vez de manera distinta: resumen_perfiles &lt;- boot_perfiles |&gt; group_by(how, price) |&gt; summarise(perfil_media = mean(perfil), ymax = quantile(perfil, 0.9), ymin = quantile(perfil, 0.10)) resumen_bolsa &lt;- resumen_perfiles |&gt; ungroup() |&gt; filter(how == &quot;tea bag&quot;) |&gt; select(price, perfil_bolsa = perfil_media) resumen_perfiles &lt;- resumen_perfiles |&gt; left_join(resumen_bolsa) |&gt; ungroup() |&gt; mutate(price = fct_reorder(price, perfil_bolsa)) ggplot(resumen_perfiles, aes(x = price, y = perfil_media, ymax = ymax, ymin = ymin)) + geom_point(colour = &quot;red&quot;) + geom_linerange() + facet_wrap(~how) + coord_flip() + geom_hline(yintercept = 0, colour = &quot;gray&quot;) + ylab(&quot;Perfil&quot;) + xlab(&quot;Precio&quot;) Nótese una deficiencia clara del bootstrap: para los que compran té suelto, en la muestra no existen personas que desconocen de dónde provienen su té (No sabe/No contestó). Esto produce un intervalo colapsado en 0 que no es razonable. Podemos remediar esto de varias maneras: quitando del análisis los que no sabe o no contestaron, agrupando en otra categoría, usando un modelo, o regularizar usando proporciones calculadas con conteos modificados: por ejemplo, agregando un caso de cada combinación (agregaría 18 personas “falsas” a una muestra de 290 personas). Bootstrap y muestras complejas La necesidad de estimaciones confiables junto con el uso eficiente de recursos conllevan a diseños de muestras complejas. Estos diseños típicamente usan las siguientes técnicas: muestreo sin reemplazo de una población finita, muestreo sistemático, estratificación, conglomerados, ajustes a no-respuesta, postestratificación. Como consecuencia, los valores de la muestra suelen no ser independientes y los análisis de los mismos dependerá del diseño de la muestra. Comenzaremos con definiciones para entender el problema. set.seed(3872999) n_escuelas &lt;- 5000 tipo &lt;- sample(c(&quot;rural&quot;, &quot;urbano&quot;, &quot;indigena&quot;), n_escuelas, replace = TRUE, prob = c(0.3, 0.5, 0.2)) escuela &lt;- tibble(ind_escuela = 1:n_escuelas, tipo, media_tipo = case_when(tipo == &quot;urbano&quot; ~ 550, tipo == &quot;rural&quot; ~ 400, TRUE ~ 350), media_escuela = rnorm(n_escuelas, media_tipo, 30), n_estudiantes = round(rnorm(n_escuelas, 30, 4))) estudiantes &lt;- uncount(escuela, n_estudiantes, .id = &quot;id_estudiante&quot;) %&gt;% rowwise() %&gt;% mutate(calif = rnorm(1, media_escuela, 70)) %&gt;% ungroup() Imaginemos que tenemos una población de 5000 escuelas, y queremos estimar la media de las calificaciones de los estudiantes en una prueba. head(estudiantes) ## # A tibble: 6 × 6 ## ind_escuela tipo media_tipo media_escuela id_estudiante calif ## &lt;int&gt; &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; &lt;dbl&gt; ## 1 1 urbano 550 561. 1 488. ## 2 1 urbano 550 561. 2 574. ## 3 1 urbano 550 561. 3 456. ## 4 1 urbano 550 561. 4 507. ## 5 1 urbano 550 561. 5 598. ## 6 1 urbano 550 561. 6 527. La primera idea sería tomar una muestra aleatoria (MAS, muestreo aleatorio simple), donde todos los estudiantes tienen igual probabilidad de ser seleccionados. Con estas condiciones el presupuesto alcanza para seleccionar a 60 estudiantes, hacemos esto y calculamos la media. muestra &lt;- slice_sample(estudiantes, n = 60) round(mean(muestra$calif), 2) ## [1] 466.73 Este número es muy cercano a la media verdadera de la población: 466.51, pero esta es una de muchas posibles muestras. medias_mas &lt;- rerun(1000, mean(sample(estudiantes$calif, 60))) %&gt;% flatten_dbl() sd(medias_mas) ## [1] 14.75242 hist_mas &lt;- ggplot(tibble(medias_mas), aes(x = medias_mas)) + geom_histogram(binwidth = 10) + geom_vline(xintercept = mean(estudiantes$calif), color = &quot;red&quot;) + xlim(410, 520) qq_mas &lt;- ggplot(tibble(medias_mas), aes(sample = medias_mas)) + geom_qq(distribution = stats::qunif) + ylim(410, 520) hist_mas + qq_mas Algunas de las muestras generan valores alejados de la verdadera media, para minimizar la probabilidad de seleccionar muestras que lleven a estimaciones alejadas del verdadero valor poblacional podríamos tomar muestras más grandes. Pero usualmente los costos limitan el tamaño de muestra. Una alternativa es estratificar, supongamos que sabemos el tipo de cada escuela (urbana, rural o indígena) y sabemos también que la calificación de los estudiantes de escuelas urbanas tiende a ser distinta a las calificaciones que los estudiantes de escuelas rurales o indígenas. En esta caso un diseño más eficiente consiste en tomar muestras independientes dentro de cada estrato. muestra_estrat &lt;- estudiantes %&gt;% group_by(tipo) %&gt;% sample_frac(0.0004) dim(muestra_estrat) ## [1] 60 6 muestrea_estrat &lt;- function(){ muestra &lt;- estudiantes %&gt;% group_by(tipo) %&gt;% sample_frac(0.0004) mean(muestra$calif) } medias_estrat &lt;- rerun(1000, muestrea_estrat()) %&gt;% flatten_dbl() Notamos que la distribución muestral está más concentrada que el caso de MAS, el error estándar se reduce de 14.75 a 10.2 hist_estrat &lt;- ggplot(tibble(medias_estrat), aes(x = medias_estrat)) + geom_histogram(binwidth = 6) + geom_vline(xintercept = mean(estudiantes$calif), color = &quot;red&quot;) + xlim(410, 520) qq_estrat &lt;- ggplot(tibble(medias_estrat), aes(sample = medias_estrat)) + geom_qq(distribution = stats::qunif) + ylim(410, 520) hist_estrat + qq_estrat Entonces, la estratificación nos sirve para reducir el error estándar de las estimaciones. Otro procedimiento común en muestreo es introducir conglomerados, a diferencia del muestreo estratificado, el propósito principal de los conglomerados es reducir costos. Veamos cuantas escuelas tendría que visitar en una muestra dada (con diseño estratificado). n_distinct(muestra_estrat$ind_escuela) ## [1] 60 Es fácil ver que visitar una escuela para aplicar solo uno o dos exámenes no es muy eficiente en cuestión de costos. Es por ello que se suelen tomar muestras considerando conglomerados naturales, en este caso la escuela. En nuestro ejemplo es razonable suponer que una parte grande del costo del muestreo sea mandar al examinador a la escuela, y que una vez en la escuela el costo de evaluar a todo sexto, en lugar de a un único alumno, es relativamente bajo. Podemos imaginar que considerando estos costos por visita de escuela nos alcance para visitar 40 escuelas y en cada una examinar a todos los estudiantes. muestra_escuelas &lt;- escuela %&gt;% group_by(tipo) %&gt;% sample_frac(size = 0.008) muestra_cgl &lt;- muestra_escuelas %&gt;% left_join(estudiantes) mean(muestra_cgl$calif) ## [1] 462.5677 muestrea_cgl &lt;- function(){ muestra_escuelas &lt;- escuela %&gt;% group_by(tipo) %&gt;% sample_frac(size = 0.008) muestra_cgl &lt;- muestra_escuelas %&gt;% left_join(estudiantes, by = c(&quot;ind_escuela&quot;, &quot;tipo&quot;)) mean(muestra_cgl$calif) } medias_cgl &lt;- rerun(1000, muestrea_cgl()) %&gt;% flatten_dbl() En este caso, el número de estudiantes examinados es mucho mayor que en MAS y muestreo estratificado, notemos que el número de estudiantes evaluados cambiará de muestra a muestra dependiendo del número de alumnos en las escuelas seleccionadas. sd(medias_cgl) ## [1] 5.337327 hist_cgl &lt;- ggplot(tibble(medias_cgl), aes(x = medias_cgl)) + geom_histogram(binwidth = 6) + geom_vline(xintercept = mean(estudiantes$calif), color = &quot;red&quot;) + xlim(410, 520) qq_cgl &lt;- ggplot(tibble(medias_cgl), aes(sample = medias_cgl)) + geom_qq(distribution = stats::qunif) + ylim(410, 520) hist_cgl + qq_cgl Ejemplo: ENIGH La complejidad de los diseños de encuestas conlleva a que el cálculo de errores estándar sea muy complicado, para atacar este problema hay dos técnicas básicas: 1) un enfoque analítico usando linearización, 2) métodos de remuestreo como bootstrap. El incremento en el poder de cómputo ha favorecido los métodos de remuestreo pues la linearización requiere del desarrollo de una fórmula para cada estimación y supuestos adicionales para simplificar. En 1988 Rao and Wu (1988) propusieron un método de bootstrap para diseños estratificados multietápicos con reemplazo de UPMs (Unidades Primarias de Muestreo) que describimos a continuación. ENIGH. Usaremos como ejemplo la Encuesta Nacional de Ingresos y Gastos de los Hogares, ENIGH 2018 (INEGI 2018), esta encuesta usa un diseño de conglomerados estratificado. Antes de proceder a bootstrap debemos entender como se seleccionaron los datos, esto es, el diseño de la muestra: Unidad primaria de muestreo (UPM). Las UPMs están constituidas por agrupaciones de viviendas. Se les denomina unidades primarias pues corresponden a la primera etapa de selección, las unidades secundarias (USMs) serían los hogares. Estratificación. Los estratos se construyen en base a estado, ámbito (urbano, complemento urbano, rural), características sociodemográficas de los habitantes de las viviendas, características físicas y equipamiento. El proceso de estratificación resulta en 888 subestratos en todo el ámbito nacional. La selección de la muestra es independiente para cada estrato, y una vez que se obtiene la muestra se calculan los factores de expansión que reflejan las distintas probabilidades de selección. Después se llevan a cabo ajustes por no respuesta y por proyección (calibración), esta última busca que distintos dominios de la muestra coincidan con la proyección de población de INEGI. concentrado_hogar &lt;- read_csv(here::here(&quot;data&quot;, &quot;conjunto_de_datos_enigh_2018_ns_csv&quot;, &quot;conjunto_de_datos_concentradohogar_enigh_2018_ns&quot;, &quot;conjunto_de_datos&quot;, &quot;conjunto_de_datos_concentradohogar_enigh_2018_ns.csv&quot;)) # seleccionar variable de ingreso corriente hogar &lt;- concentrado_hogar %&gt;% mutate( upm = as.integer(upm), edo = str_sub(ubica_geo, 1, 2) ) %&gt;% select(folioviv, foliohog, est_dis, upm, factor, ing_cor, edad_jefe, edo) %&gt;% group_by(est_dis) %&gt;% mutate(n = n_distinct(upm)) %&gt;% # número de upms por estrato ungroup() hogar ## # A tibble: 74,647 × 9 ## folioviv foliohog est_dis upm factor ing_cor edad_jefe edo n ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; &lt;int&gt; ## 1 100013601 1 2 1 175 76404. 74 10 106 ## 2 100013602 1 2 1 175 42988. 48 10 106 ## 3 100013603 1 2 1 175 580698. 39 10 106 ## 4 100013604 1 2 1 175 46253. 70 10 106 ## 5 100013606 1 2 1 175 53837. 51 10 106 ## 6 100026701 1 2 2 189 237743. 41 10 106 ## 7 100026703 1 2 2 189 32607. 57 10 106 ## 8 100026704 1 2 2 189 169918. 53 10 106 ## 9 100026706 1 2 2 189 17311. 30 10 106 ## 10 100027201 1 2 3 186 120488. 69 10 106 ## # ℹ 74,637 more rows Para el cálculo de estadísticos debemos usar los factores de expansión, por ejemplo el ingreso trimestral total sería: sum(hogar$factor * hogar$ing_cor / 1000) ## [1] 1723700566 y ingreso trimestral medio (miles pesos) sum(hogar$factor * hogar$ing_cor / 1000) / sum(hogar$factor) ## [1] 49.61029 La estimación del error estándar, por otro lado, no es sencilla y requiere usar aproximaciones, en la metodología de INEGI proponen una aproximación con series de Taylor. Figure 5.3: Extracto de estimación de errores de muestreo, ENIGH 2018. Veamos ahora como calcular el error estándar siguiendo el bootstrap de Rao y Wu: En cada estrato se seleccionan con reemplazo \\(m_h\\) UPMs de las \\(n_h\\) de la muestra original. Denotamos por \\(m_{hi}^*\\) el número de veces que se seleccionó la UPM \\(i\\) en el estrato \\(h\\) (de tal manera que \\(\\sum m_{hi}^*=m_h\\)). Creamos una replicación del ponderador correspondiente a la \\(k\\)-ésima unidad (USM) como: \\[d_k^*=d_k \\bigg[\\bigg(1-\\sqrt{\\frac{m_h}{n_h - 1}}\\bigg) + \\bigg(\\sqrt{\\frac{m_h}{n_h - 1}}\\frac{n_h}{m_h}m_{h}^*\\bigg)\\bigg]\\] donde \\(d_k\\) es el inverso de la probabilidad de selección. Si \\(m_h &lt;(n_h -1)\\) todos los pesos definidos de esta manera serán no negativos. Calculamos el peso final \\(w_k^*\\) aplicando a \\(d_k^*\\) los mismos ajustes que se hicieron a los ponderadores originales. Calculamos el estadístico de interés \\(\\hat{\\theta}\\) usando los ponderadores \\(w_k^*\\) en lugar de los originales \\(w_k\\). Repetimos los pasos 1 y 2 \\(B\\) veces para obtener \\(\\hat{\\theta}^{*1},\\hat{\\theta}^{*2},...,\\hat{\\theta}^{*B}\\). Calculamos el error estándar como: \\[\\hat{\\textsf{se}}_B = \\bigg\\{\\frac{\\sum_{b=1}^B[\\hat{\\theta}^*(b)-\\hat{\\theta}^*(\\cdot)]^2 }{B}\\bigg\\}^{1/2}\\] En principio podemos elegir cualquier valor de \\(m_h \\geq 1\\), el más común es elegir \\(m_h=n_h-1\\), en este caso: \\[d_k^*=d_k \\frac{n_h}{n_h-1}m_{hi}^*\\] en este escenario las unidades que no se incluyen en la muestra tienen un valor de cero como ponderador. Si elegimos \\(n_h \\ne n_h-1\\) las unidades que no están en la muestra tienen ponderador distinto a cero, si \\(m_h=n_h\\) el ponderador podría tomar valores negativos. Implementemos el bootstrap de Rao y Wu a la ENIGH, usaremos \\(m_h=n_h-1\\) # creamos una tabla con los estratos y upms est_upm &lt;- hogar %&gt;% distinct(est_dis, upm, n) %&gt;% arrange(est_dis, upm) hogar_factor &lt;- est_upm %&gt;% group_by(est_dis) %&gt;% # dentro de cada estrato tomamos muestra (n_h-1) sample_n(size = first(n) - 1, replace = TRUE) %&gt;% add_count(est_dis, upm, name = &quot;m_hi&quot;) %&gt;% # calculamos m_hi* left_join(hogar, by = c(&quot;est_dis&quot;, &quot;upm&quot;, &quot;n&quot;)) |&gt; mutate(factor_b = factor * m_hi * n / (n - 1)) # unimos los pasos anteriores en una función para replicar en cada muestra bootstrap svy_boot &lt;- function(est_upm, hogar){ m_hi &lt;- est_upm %&gt;% group_split(est_dis) %&gt;% map(~sample(.$upm, size = first(.$n) - 1, replace = TRUE)) %&gt;% flatten_int() %&gt;% plyr::count() %&gt;% select(upm = x, m_h = freq) m_hi %&gt;% left_join(hogar, by = c(&quot;upm&quot;)) %&gt;% mutate(factor_b = factor * m_h * n / (n - 1)) } set.seed(1038984) boot_rep &lt;- rerun(500, svy_boot(est_upm, hogar)) # Aplicación a ingreso medio wtd_mean &lt;- function(w, x, na.rm = FALSE) { sum(w * x, na.rm = na.rm) / sum(w, na.rm = na.rm) } # La media es: hogar %&gt;% summarise(media = wtd_mean(factor, ing_cor)) ## # A tibble: 1 × 1 ## media ## &lt;dbl&gt; ## 1 49610. Y el error estándar: map_dbl(boot_rep, ~wtd_mean(w = .$factor_b, x = .$ing_cor)) %&gt;% quantile(c(0.025, 0.975)) ## 2.5% 97.5% ## 48742.12 50519.02 El método bootstrap está implementado en el paquete survey y más recientemente en srvyr que es una versión tidy que utiliza las funciones en survey. Podemos comparar nuestros resultados con la implementación en survey. # 1. Definimos el diseño de la encuesta library(survey) library(srvyr) enigh_design &lt;- hogar %&gt;% as_survey_design(ids = upm, weights = factor, strata = est_dis) # 2. Elegimos bootstrap como el método para el cálculo de errores estándar set.seed(7398731) enigh_boot &lt;- enigh_design %&gt;% as_survey_rep(type = &quot;bootstrap&quot;, replicates = 500) # 3. Así calculamos la media enigh_boot %&gt;% srvyr::summarise(mean_ingcor = survey_mean(ing_cor)) ## # A tibble: 1 × 2 ## mean_ingcor mean_ingcor_se ## &lt;dbl&gt; &lt;dbl&gt; ## 1 49610. 468. enigh_boot %&gt;% srvyr::summarise(mean_ingcor = survey_mean(ing_cor, vartype = &quot;ci&quot;)) ## # A tibble: 1 × 3 ## mean_ingcor mean_ingcor_low mean_ingcor_upp ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 49610. 48690. 50530. # por estado enigh_boot %&gt;% group_by(edo) %&gt;% srvyr::summarise(mean_ingcor = survey_mean(ing_cor)) ## # A tibble: 30 × 3 ## edo mean_ingcor mean_ingcor_se ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 10 50161. 995. ## 2 11 46142. 1241. ## 3 12 29334. 1063. ## 4 13 38783. 1019. ## 5 14 60541. 1924. ## 6 15 48013. 1359. ## 7 16 42653. 1393. ## 8 17 42973. 1601. ## 9 18 48148. 1967. ## 10 19 68959. 4062. ## # ℹ 20 more rows Resumiendo: El bootstrap de Rao y Wu genera un estimador consistente y aproximadamente insesgado de la varianza de estadísticos no lineales y para la varianza de un cuantil. Este método supone que la seleccion de UPMs es con reemplazo; hay variaciones del estimador bootstrap de Rao y Wu que extienden el método que acabamos de estudiar; sin embargo, es común ignorar este aspecto, por ejemplo Mach et al estudian las propiedades del estimador de varianza bootstrap de Rao y Wu cuando la muestra se seleccionó sin reemplazo. Bootstrap en R Es común crear nuestras propias funciones cuando usamos bootstrap, sin embargo, en R también hay alternativas que pueden resultar convenientes, mencionamos 3: El paquete rsample (forma parte de la colección tidymodels) y tiene una función bootstraps() que regresa un arreglo cuadrangular (tibble, data.frame) que incluye una columna con las muestras bootstrap y un identificador del número y tipo de muestra. Veamos un ejemplo donde seleccionamos muestras del conjunto de datos muestra_computos que contiene 10,000 observaciones. library(rsample) load(&quot;data/election_2012.rda&quot;) muestra_computos &lt;- slice_sample(election_2012, n = 10000) muestra_computos ## # A tibble: 10,000 × 23 ## state_code state_name state_abbr district_loc_17 district_fed_17 polling_id ## &lt;chr&gt; &lt;chr&gt; &lt;chr&gt; &lt;int&gt; &lt;int&gt; &lt;int&gt; ## 1 18 Nayarit NAY 11 2 86709 ## 2 27 Tabasco TAB 17 5 122035 ## 3 15 México MEX 7 35 75477 ## 4 27 Tabasco TAB 19 5 122262 ## 5 17 Morelos MOR 6 2 84733 ## 6 07 Chiapas CHPS 22 5 15376 ## 7 14 Jalisco JAL 2 2 52634 ## 8 08 Chihuahua CHIH 7 4 19097 ## 9 14 Jalisco JAL 20 20 60549 ## 10 13 Hidalgo HGO 11 4 50221 ## # ℹ 9,990 more rows ## # ℹ 17 more variables: section &lt;int&gt;, region &lt;chr&gt;, polling_type &lt;chr&gt;, ## # section_type &lt;chr&gt;, pri_pvem &lt;int&gt;, pan &lt;int&gt;, panal &lt;int&gt;, ## # prd_pt_mc &lt;int&gt;, otros &lt;int&gt;, total &lt;int&gt;, nominal_list &lt;int&gt;, ## # pri_pvem_pct &lt;dbl&gt;, pan_pct &lt;dbl&gt;, panal_pct &lt;dbl&gt;, prd_pt_mc_pct &lt;dbl&gt;, ## # otros_pct &lt;dbl&gt;, winner &lt;chr&gt; Generamos 100 muestras bootstrap, y la función nos regresa un arreglo con 100 renglones, cada uno corresponde a una muestra bootstrap. set.seed(839287482) computos_boot &lt;- bootstraps(muestra_computos, times = 100) computos_boot ## # Bootstrap sampling ## # A tibble: 100 × 2 ## splits id ## &lt;list&gt; &lt;chr&gt; ## 1 &lt;split [10000/3647]&gt; Bootstrap001 ## 2 &lt;split [10000/3623]&gt; Bootstrap002 ## 3 &lt;split [10000/3724]&gt; Bootstrap003 ## 4 &lt;split [10000/3682]&gt; Bootstrap004 ## 5 &lt;split [10000/3696]&gt; Bootstrap005 ## 6 &lt;split [10000/3716]&gt; Bootstrap006 ## 7 &lt;split [10000/3679]&gt; Bootstrap007 ## 8 &lt;split [10000/3734]&gt; Bootstrap008 ## 9 &lt;split [10000/3632]&gt; Bootstrap009 ## 10 &lt;split [10000/3692]&gt; Bootstrap010 ## # ℹ 90 more rows La columna splits tiene información de las muestras seleccionadas, para la primera vemos que de 10,000 observaciones en la muestra original la primera muestra bootstrap contiene 10000-3647=6353. first_computos_boot &lt;- computos_boot$splits[[1]] first_computos_boot ## &lt;Analysis/Assess/Total&gt; ## &lt;10000/3647/10000&gt; Y podemos obtener los datos de la muestra bootstrap con la función as.data.frame() as.data.frame(first_computos_boot) ## # A tibble: 10,000 × 23 ## state_code state_name state_abbr district_loc_17 district_fed_17 polling_id ## &lt;chr&gt; &lt;chr&gt; &lt;chr&gt; &lt;int&gt; &lt;int&gt; &lt;int&gt; ## 1 07 Chiapas CHPS 13 9 13397 ## 2 14 Jalisco JAL 15 15 58404 ## 3 09 Ciudad de M… CDMX 11 13 26471 ## 4 17 Morelos MOR 7 3 84909 ## 5 25 Sinaloa SIN 11 3 114038 ## 6 11 Guanajuato GTO 16 12 41506 ## 7 29 Tlaxcala TLAX 6 3 128006 ## 8 02 Baja Califo… BC 8 5 3901 ## 9 02 Baja Califo… BC 9 5 3779 ## 10 08 Chihuahua CHIH 19 5 19536 ## # ℹ 9,990 more rows ## # ℹ 17 more variables: section &lt;int&gt;, region &lt;chr&gt;, polling_type &lt;chr&gt;, ## # section_type &lt;chr&gt;, pri_pvem &lt;int&gt;, pan &lt;int&gt;, panal &lt;int&gt;, ## # prd_pt_mc &lt;int&gt;, otros &lt;int&gt;, total &lt;int&gt;, nominal_list &lt;int&gt;, ## # pri_pvem_pct &lt;dbl&gt;, pan_pct &lt;dbl&gt;, panal_pct &lt;dbl&gt;, prd_pt_mc_pct &lt;dbl&gt;, ## # otros_pct &lt;dbl&gt;, winner &lt;chr&gt; Una de las principales ventajas de usar este paquete es que es eficiente en el uso de memoria. library(pryr) object_size(muestra_computos) ## 1.41 MB object_size(computos_boot) ## 5.49 MB # tamaño por muestra object_size(computos_boot)/nrow(computos_boot) ## 54.92 kB # el incremento en tamaño es &lt;&lt; 1000 as.numeric(object_size(computos_boot)/object_size(muestra_computos)) ## [1] 3.895024 Adicionalmente incluye funciones para el cálculo de intervalos bootstrap: intervalo_propinas_90 &lt;- bootstraps(propinas, strata = momento, 1000) |&gt; mutate(res_boot = map(splits, estimador)) |&gt; int_pctl(res_boot, alpha = 0.10) El paquete boot está asociado al libro Bootstrap Methods and Their Applications (Davison and Hinkley (1997)) y tiene, entre otras, funciones para calcular replicaciones bootstrap y para construir intervalos de confianza usando bootstrap: calculo de replicaciones bootstrap con la función boot(), intervalos normales, de percentiles y \\(BC_a\\) con la función boot.ci(), intevalos ABC con la función `abc.ci(). El paquete bootstrap contiene datos usados en Efron and Tibshirani (1993), y la implementación de funciones para calcular replicaciones y construir intervalos de confianza: calculo de replicaciones bootstrap con la función bootstrap(), intervalos \\(BC_a\\) con la función bcanon(), intevalos ABC con la función abcnon(). Conclusiones y observaciones El principio fundamental del Bootstrap no paramétrico es que podemos estimar la distribución poblacional con la distribución empírica. Por tanto para hacer inferencia tomamos muestras con reemplazo de la muestra y analizamos la variación de la estadística de interés a lo largo de las remuestras. El bootstrap nos da la posibilidad de crear intervalos de confianza cuando no contamos con fórmulas para hacerlo de manera analítica y sin supuestos distribucionales de la población. Hay muchas opciones para construir intervalos bootstrap, los que tienen mejores propiedades son los intervalos \\(BC_a\\), sin embargo, los más comunes son los intervalos normales con error estándar bootstrap y los intervalos de percentiles de la distribución bootstrap. Antes de hacer intervalos normales vale la pena graficar la distribución bootstrap y evaluar si el supuesto de normalidad es razonable. En cuanto al número de muestras bootstrap se recomienda al menos \\(1,000\\) al hacer pruebas, y \\(10,000\\) o \\(15,000\\) para los resultados finales, sobre todo cuando se hacen intervalos de confianza de percentiles. La función de distribución empírica es una mala estimación en las colas de las distribuciones, por lo que es difícil construir intervalos de confianza (usando bootstrap no paramétrico) para estadísticas que dependen mucho de las colas. O en general para estadísticas que dependen de un número chico de observaciones de una muestra grande. Referencias "],["estimación-por-máxima-verosimilitud.html", "Sección 6 Estimación por máxima verosimilitud Introducción a estimación por máxima verosimilitud Máxima verosimilitud para observaciones continuas Aspectos numéricos Máxima verosimilitud para más de un parámetro", " Sección 6 Estimación por máxima verosimilitud Los ejemplos que hemos visto han sido todos de estimadores plug-in (o por sustitución): si queremos saber una cantidad poblacional, y tenemos una muestra dada, entonces calculamos la estadística de interés como si la muestra fuera la población. Por ejemplo, para estimar la mediana poblacional usamos la mediana muestral, si queremos estimar la media poblacional usamos la media muestral, y así sucesivamente. Estos estimadores usualmente dan resultados razonables (pero hay que checar usando muestra bootstraps, por ejemplo, y pensar lo que estamos haciendo). Cuando sabemos más acerca de la población y usamos un modelo teórico es posible hacer más: dependiendo de qué cantidades se quieren estimar, podemos construir estimadores que sean óptimos en algún sentido siempre y cuando se cumplan los supuestos teóricos, como veremos ahora. Por ejemplo: ¿deberíamos estimar el centro de una distribución simétrica con la media o con la mediana, o quizá con una media recortada? En esta parte construiremos la teoría básica de estimación cuando trabajamos con modelos teóricos conocidos. El objetivo es entender las ideas básicas de estos procedimientos, y cómo evaluar sus resultados. Recordatorio: las ventajas de usar modelos teóricos para describir distribuciones de datos está en que es posible comprimir más eficientemente la información, es posible construir modelos más complejos juntando varios de estos modelos y de sus dependencias, y de que es posible hacer más teoría útil que nos guíe. La desventaja es que es necesario que esos supuestos teóricos sean razonables. Introducción a estimación por máxima verosimilitud Uno de los procedimientos más estándar en esta situación es el método de máxima verosimilitud. Los estimadores de máxima verosimilitud tienen propiedades convenientes, y dan en general resultados razonables siempre y cuando los supuestos sean razonables. Máxima verosimilitud es un proceso intuitivo, y consiste en aprender o estimar valores de parámetros desconocidos suponiendo para los datos su explicación más probable. Para esto, usando supuestos y modelos, requeriremos calcular la probabilidad de un conjunto de observaciones. Ejemplo. Adaptado de (Chihara and Hesterberg 2018). Supongamos que una máquina produce dos tipos de bolsas de 25 galletas: la mitad de las veces produce una bolsa con 5 galletas de avena y 20 de chispas de chocolate, y la otra mitad produce bolsas con 23 galletas de avena y 2 de chispas de chocolate. Tomamos una bolsa, y no sabemos qué tipo de bolsa es (parámetro desconocido). Extraemos al azar una de las galletas, y es de chispas de chocolate (observación). Por máxima verosimilitud, inferimos que la bolsa que estamos considerando tiene 5 galletas de avena. Esto es porque es más probable observar una galleta de chispas en las bolsas que contienen 5 galletas de avena que en las bolsas que contienen 23 galletas de avena. Podemos cuantificar la probabilidad que “acertemos” en nuestra inferencia. Cómo se aprecia en el ejemplo anterior, el esquema general es: Existe un proceso del que podemos obtener observaciones de algún sistema o población real. Tenemos un modelo probabilístico que dice cómo se producen esas observaciones a partir del sistema o población real. Usualmente este modelo tiene algunas cantidades que no conocemos, que rigen el proceso y cómo se relaciona el proceso con las observaciones. Nuestro propósito es: Extraemos observaciones del proceso \\[x_1, x_2, \\ldots, x_n.\\] Queremos aprender de los parámetros desconocidos del proceso para calcular cantidades de interés acerca del sistema o población real En principio, los modelos que consideramos pueden ser complicados y tener varias partes o parámetros. Veamos primero un ejemplo clásico con un solo parámetro, y cómo lo resolveríamos usando máxima verosimilitud. Nota: Cuando decimos muestra en general nos referimos a observaciones independientes obtenidas del mismo proceso (ver la sección de distribución de muestreo) para ver qué significa que sea independientes. Este esquema es un supuesto que simplifica mucho los cálculos, como discutimos antes. Muchas veces este supuesto sale del diseño de la muestra o del estudio, pero en todo caso es importante considerar si es razonable o no para nuestro problema particular. Denotemos por \\(f(x; \\theta)\\) la función de densidad para una variable aleatoria continua con párametro asociado \\(\\theta.\\) Denotamos por \\(X_1, \\ldots, X_n,\\) una muestra aleatoria de \\(n\\) observaciones de esta distribución y por \\(x_1, \\ldots, x_n\\) los valores observados de esta muestra aleatoria. Ejemplo. Supongamos que queremos saber qué proporción de registros de una base de datos tiene algún error menor de captura. No podemos revisar todos los registros, así que tomamos una muestra de 8 registros, escogiendo uno por uno al azar de manera independiente. Revisamos los 8 registros, y obtenemos los siguientes datos: \\[x_1 = 0, x_2 = 1, x_3 = 0, x_4 = 0, x_5 =1, x_6 =0, x_7 =0, x_8 =0\\] donde 1 indica un error menor. Encontramos dos errores menores. ¿Cómo estimamos el número de registros con errores leves en la base de datos? Ya sabemos una respuesta razonable para nuestro estimador puntual, que sería \\(\\hat{p}=2/8=0.25\\). Veamos cómo se obtendría por máxima verosimilitud. Según el proceso con el que se construyó la muestra, debemos dar una probabilidad de observar los 2 errores en 8 registros. Supongamos que en realidad existe una proporción \\(p\\) de que un registro tenga un error. Entonces calculamos Probabilidad de observar la muestra: \\[P(X_1 = 0, X_2 = 1, X_3 = 0, X_4 = 0, X_5 =1, X_6 =0, X_7 =0, X_8 =0)\\] es igual a \\[P(X_1 = 0)P(X_2 = 1)P(X_3 = 0)P( X_4 = 0)P(X_5 =1)P(X_6 =0)P(X_7 =0)P(X_8 =0)\\] pues la probabilidad de que cada observación sea 0 o 1 no depende de las observaciones restantes (la muestra se extrajo de manera independiente). Esta última cantidad tiene un parámetro que no conocemos: la proporcion \\(p\\) de registros con errores. Así que lo denotamos como una cantidad desconocida \\(p\\). Nótese entonces que \\(P(X_2=1) = p\\), \\(P(X_3=0) = 1-p\\) y así sucesivamente, así que la cantidad de arriba es igual a \\[(1-p)p(1-p)(1-p)p(1-p)(1-p)(1-p) \\] que se simplifica a \\[ \\mathcal{L}(p) = p^2(1-p)^6\\] Ahora la idea es encontrar la p que maximiza la probabilidad de lo que observamos. En este caso se puede hacer con cálculo, pero vamos a ver una gráfica de esta función y cómo resolverla de manera numérica. verosimilitud &lt;- function(p){ p^2 * (1-p)^6 } dat_verosim &lt;- tibble(x = seq(0,1, 0.001)) %&gt;% mutate(prob = map_dbl(x, verosimilitud)) ggplot(dat_verosim, aes(x = x, y = prob)) + geom_line() + geom_vline(xintercept = 0.25, color = &quot;red&quot;) + xlab(&quot;p&quot;) Nótese que esta gráfica: Depende de los datos, que pensamos fijos. Cuando cambiamos la \\(p\\), la probabilidad de observar la muestra cambia. Nos interesa ver las regiones donde la probabilidad es relativamente alta. El máximo está en 0.25. Así que el estimador de máxima verosimilitud es \\(\\hat{p} = 0.25\\), que es también el estimador usual de plugin en este caso. Para uniformizar la notación con el caso continuo que veremos más adelante, usaremos la notación \\[P(X=x) = f(x)\\] donde \\(f\\) es la función de densidad (en este caso, función de masa de probabilidad) de \\(X\\). Si esta función depende de un parámetro, escribimos \\[f(x ;\\theta)\\] Definición. Sean \\(X_1, \\ldots, X_n\\) una muestra de una densidad \\(f(x; \\theta)\\) y sean \\(x_1,x_2,\\ldots, x_n\\) los valores observados. La función de verosimilitud del párametro de interés \\(\\theta\\) está definida por \\[\\begin{align} \\mathcal{L}(\\theta; x_1, \\ldots, x_n) = \\prod_{i = 1}^n f(x_i; \\theta). \\end{align}\\] Esta función nos dice qué tan creible es el valor del parámetro \\(\\theta\\) dada la muestra observada. A veces también la denotamos por \\(\\mathcal{L}_n(\\theta)\\). Ahora definimos qué es un estimador de máxima verosimilitud. Definición. Un estimador de máxima verosimilitud lo denotamos por \\(\\hat \\theta_{\\textsf{MLE}}\\) y es un valor que satisface \\[\\begin{align} \\hat \\theta_{\\textsf{MLE}} = \\underset{\\theta \\, \\in \\, \\Theta}{\\arg\\max}\\, \\mathcal{L}(\\theta; x_1, \\ldots, x_n), \\end{align}\\] donde \\(\\Theta\\) denota el espacio parametral. Es decir, el espacio válido de búsqueda congruente con la definición del modelo. Considera el caso de una normal con media y varianza desconocidas. ¿Cuáles son los espacios parametrales para efectuar \\(\\mathsf{MLE}\\)? Considera el caso de una Binomial con parámetro \\(p\\) desconocidos. ¿Cuál es el espacio parametral para la búsqueda del \\(\\mathsf{MLE}\\)? Obsérvese que para construir la verosimilitud y en consecuencia buscar por estimadores de máxima verosimlitud necesitamos: Un modelo teórico de cómo es la población con parámetros e Información de cómo se extrajo la muestra, y entonces podemos resolver nuestro problema de estimación convirtiéndolo en uno de optimización. Probamos esta idea con un proceso más complejo. Ejemplo. Supongamos que una máquina puede estar funcionando correctamente o no en cada corrida. Cada corrida se producen 500 productos, y se muestrean 10 para detectar defectos. Cuando la máquina funciona correctamente, la tasa de defectos es de 3%. Cuando la máquina no está funcionando correctamente la tasa de defectos es de 20% Supongamos que escogemos al azar 11 corridas, y obervamos los siguientes número de defectuosos: \\[1, 0, 0, 3 ,0, 0, 0, 2, 1, 0, 0\\] La pregunta es: ¿qué porcentaje del tiempo la máquina está funcionando correctamente? Primero pensemos en una corrida. La probabilidad de observar una sucesión particular de \\(r\\) defectos es \\[0.03^r(0.97)^{(10-r)}\\] cuando la máquina está funcionando correctamente. Si la máquina está fallando, la misma probabilidad es \\[0.2^r(0.8)^{(10-r)}.\\] Ahora supongamos que la máquina trabaja correctamente en una proporción \\(p\\) de las corridas. Entonces la probabilidad de observar \\(r\\) fallas se calcula promediando (probabilidad total) sobre las probabilidades de que la máquina esté funcionando bien o no: \\[0.03^r(0.97)^{(10-r)}p + 0.2^r(0.8)^{(10-r)}(1-p)\\] Y esta es nuestra función de verosimilitud para una observación. Suponemos que las \\(r_1,r_2, \\ldots, r_{11}\\) observaciones son independientes (por ejemplo, después de cada corrida la máquina se prepara de una manera estándar, y es como si el proceso comenzara otra vez). Entonces tenemos que multiplicar estas probabilidades para cada observación \\(r_1\\): # creamos la función de verosimilitud con los datos observados como dados verosim &lt;- function(p) { r &lt;- c(1, 2, 0, 3, 0, 0, 0, 2, 1, 0, 3) q_func &lt;- 0.03^r*(0.97)^(10-r) q_falla &lt;- 0.2^r*(0.8)^(10-r) prod(p * q_func + (1 - p) * q_falla) } verosim(0.1) # Una alternativa que nos da más flexibilidad para generar la función de # verosimilitud, es crear una función que recibe los datos observados y nos # regresa la función de verosimilitud correspondiente # Entonces, cal_verosim es una función que regresa una función calc_verosim &lt;- function(r){ q_func &lt;- 0.03 ^ r * (0.97) ^ (10 - r) q_falla &lt;- 0.2 ^ r * (0.8) ^ (10 - r) function(p){ #nota: esta no es la mejor manera de calcularlo, hay # que usar logaritmos. prod(p * q_func + (1 - p) * q_falla) } } verosim &lt;- calc_verosim(r = c(1, 0, 0, 3, 0, 0, 0, 2, 1, 0, 0)) verosim(0.1) ## [1] 2.692087e-14 dat_verosim &lt;- tibble(x = seq(0, 1, 0.001)) %&gt;% mutate(prob = map_dbl(x, verosim)) ggplot(dat_verosim, aes(x = x, y = prob)) + geom_line() + geom_vline(xintercept = 0.773, color = &quot;red&quot;) + xlab(&quot;prop funcionado&quot;) Y nuestra estimación puntual sería de alrededor de 80%. Máxima verosimilitud para observaciones continuas Cuando las observaciones \\(x_1,\\ldots, x_n\\) provienen de una distribución continua, no tiene sentido considerar \\(P(X = x_i)\\), pues siempre es igual a cero. Sin embargo, podemos escribir para pequeños valores \\(\\epsilon \\ll 1\\) \\[\\begin{align} P(x - \\epsilon &lt; X &lt; x + \\epsilon | \\theta) = \\int_{x - \\epsilon}^{x + \\epsilon} f(t; \\theta) \\, \\text{d} t \\approx 2 \\epsilon f(x; \\theta), \\end{align}\\] donde \\(f(x; \\theta)\\) es la función de densidad de \\(X.\\) Por lo tanto, \\[\\begin{align} \\begin{split} P&amp;(x_1 - \\epsilon &lt; X_1 &lt; x_1 + \\epsilon, \\ldots, x_n - \\epsilon &lt; X_n &lt; x_n + \\epsilon | \\theta) \\\\ &amp;= \\prod_{i = 1}^n P(x_i - \\epsilon &lt; X_i &lt; x_i + \\epsilon | \\theta) \\\\ &amp;= \\prod_{i = 1}^n 2 \\epsilon f(x_i; \\theta) = (2\\epsilon)^n \\prod_{i = 1}^n f(x_i; \\theta). \\end{split} \\end{align}\\] Notemos que si \\(\\epsilon \\rightarrow 0\\) la ecuación rápidamente converge a cero. Pero para pequeños valores de \\(\\epsilon\\) la ecuación que nos interesa es proporcional a \\(\\prod_{i = 1}^n f(x_i; \\theta).\\) De esta forma, nuestra definición de máxima verosimilitud y estimadores de máxima verosimilitud es la misma para el caso continuo (verifica las definiciones de la sección anterior). Ejemplo. Supongamos que tenemos una muestra \\(x_1\\ldots, x_n\\) extraidas de una distribución exponencial con tasa \\(\\lambda&gt;0\\) donde no conocemos \\(\\lambda\\). ¿Cuál es el estimador de máxima verosimilitud de \\(\\lambda\\)? Para \\(\\lambda&gt;0\\), tenemos que \\[{\\mathcal L}(\\lambda) = \\prod_{i=1}^n \\lambda e^{-\\lambda x_i}\\] de modo que \\[{\\mathcal L}(\\lambda) = \\lambda^n e^{-\\lambda \\sum_{i=1}^nx_i} = \\lambda^n e^{-n\\lambda\\bar{x}} = e^{n(\\log\\lambda - \\lambda\\bar{x})}\\] Que podemos maximizar usando cálculo para obtener \\(\\hat{\\lambda}_{\\mathsf{ML}} = \\frac{1}{\\bar{x}}\\) (demuéstralo). Discute por qué esto es intuitivamente razonable: ¿cuál es el valor esperado de una exponencial con parámetro \\(\\lambda\\)? Aspectos numéricos Encontrar el estimador de máxima verosimilitud (\\(\\textsf{MLE}\\)) es automático en la mayoría de los casos. En teoría, podemos reutilizar la misma rutina numérica para encontrar el estimador sin ninguna ayuda de la analista. Esto contrasta con otras técnicas de estimación en donde se requieren cálculos y manipulación de ecuaciones. Sin embargo, hay situaciones que se pueden evitar de manera general. Por ejemplo, cuando calculamos la verosimilitud arriba, nótese que estamos multiplicando números que pueden ser muy chicos (por ejemplo \\(p^6\\), etc). Esto puede producir desbordes numéricos fácilmente. Por ejemplo para un tamaño de muestra de 1000, podríamos tener que calcular p &lt;- 0.1 proba &lt;- (p ^ 800)*(1-p)^200 proba ## [1] 0 En estos casos, es mejor hacer los cálculos en escala logarítmica. El logaritmo convierte productos en sumas, y baja exponentes multiplicando. Si calculamos en escala logaritmica la cantidad de arriba, no tenemos problema: log_proba &lt;- 800 * log(p) + 200 * log(1-p) log_proba ## [1] -1863.14 Ahora notemos que Maximizar la verosimilitud es lo mismo que maximizar la log-verosimilitud, pues el logaritmo es una función creciente. Si \\(x_{\\max}\\) es el máximo de \\(f\\), tenemos que \\(f(x_{\\max})&gt;f(x)\\) para cualquier \\(x\\), entonces tomando logaritmo, \\[\\log(f(x_{max}))&gt;\\log(f(x)),\\] para cualquier \\(x.\\) Pues el logaritmo respeta la desigualdad por ser creciente. Usualmente usamos la log-verosimilitud para encontrar el estimador de máxima verosimilitud. Hay razónes teóricas y de interpretación por las que también es conveniente hacer esto. Definición. La log-verosimilitud la denotamos usualmente por \\[\\ell_n(\\theta) = \\log \\left(\\mathcal{L}_n(\\theta)\\right),\\] donde hemos suprimido la dependencia en la muestra por conveniencia. Ejemplo. En nuestro primer ejemplo, log_verosimilitud &lt;- function(p){ 2*log(p) + 6*log(1-p) } dat_verosim &lt;- tibble(x = seq(0,1, 0.01)) %&gt;% mutate(log_prob = map_dbl(x, log_verosimilitud)) ggplot(dat_verosim, aes(x = x, y = log_prob)) + geom_line() + geom_vline(xintercept = 0.25, color = &quot;red&quot;) + xlab(&quot;p&quot;) Obtenemos el mismo máximo. Podemos incluso resolver numéricamente: solucion &lt;- optim(p = 0.5, log_verosimilitud, control = list(fnscale = -1), method = &quot;Brent&quot;, lower = 0, upper = 1) solucion$par ## [1] 0.25 Y en nuestro segundo ejemplo: calc_log_verosim &lt;- function(r){ q_func &lt;- 0.03^r*(0.97)^(10-r) q_falla &lt;- 0.2^r*(0.8)^(10-r) function(p){ #nota: esta no es la mejor manera de calcularlo, hay # que usar logaritmos. sum(log(p * q_func + (1 - p) * q_falla)) } } log_verosim &lt;- calc_log_verosim(c(1, 0, 0, 3, 0, 0, 0, 2, 1, 0, 0)) log_verosim(0.1) ## [1] -31.24587 solucion &lt;- optim(p = 0.2, log_verosim, control = list(fnscale = -1), method = &quot;Brent&quot;, lower = 0, upper = 1) solucion$par ## [1] 0.7733766 solucion$convergence ## [1] 0 dat_verosim &lt;- tibble(x = seq(0,1, 0.001)) %&gt;% mutate(log_verosimilitud = map_dbl(x, log_verosim)) ggplot(dat_verosim, aes(x = x, y = log_verosimilitud)) + geom_line() + geom_vline(xintercept = 0.775, color = &quot;red&quot;) + xlab(&quot;prop funcionado&quot;) Nótese que la verosimilitud la consideramos función de los parámetros, donde los datos están fijos. Podemos construir una función que genera la función de verosimilitud dependiendo de los datos. En nuestro primer ejemplo de muestras de registros erróneos, podríamos construir una función que genera la log verosimilitud dependiendo del tamaño de muestra y del número de errores encontrado: construir_log_verosim &lt;- function(n, n_err){ # n es tamaño de muestra # n_err el número de errores detectados (datos) n_corr &lt;- n - n_err log_verosim &lt;- function(p){ n_err * log(p) + n_corr * log(1-p) } } Cuando fijamos \\(n\\) y \\(n_{\\textsf{err}}\\), esta función genera otra función, la log verosimilitud, que es la que queremos optimizar. Supongamos entonces que sacamos 20 registros al azar y observamos 10 incorrectos. La función de verosimilitud es log_vero &lt;- construir_log_verosim(20, 10) tibble(x = seq(0,1,0.001)) %&gt;% mutate(log_ver = log_vero(x)) %&gt;% ggplot(aes(x = x, y = log_ver)) + geom_line() + geom_vline(xintercept = 0.5, color = &#39;red&#39;) Ejemplo. Supongamos que en una población de transacciones hay un porcentaje \\(p\\) (desconocido) que son fraudulentas. Tenemos un sistema de clasificación humana que que marca transacciones como sospechosas. Con este sistema hemos medido que la proporción de transacciones normales que son marcadas como sospechosas es de 0.1%, y que la proporción de transacciones fraudulentas que son marcadas como sospechosas es de 98%. Supongamos que extraemos una muestra de 2000 transacciones, de manera que todas ellas tiene la misma probabilidad de ser fraudulentas. El sistema de clasificación marca 4 transacciones como sospechosas ¿Cómo estimamos la proporción de transacciones fraudulentas en la población? Solución: sea \\(p\\) la proporción de transacciones fraudulentas. Entonces la probabilidad de que una transacción sea marcada como sospechosa es (proba total): \\[0.98p + 0.001(1-p)\\] Pues tenemos que contar 98% de la proporción \\(p\\) de fraudulentas (correctamente detectadas) más 0.1% de la proporción \\((1-p)\\) de fraudulentas. Escribimos entonces nuestra función de verosimilitud crear_log_verosim &lt;- function(n, n_sosp){ # devolver la función log verosimilitud log_verosimilitud_pct &lt;- function(pct){ # sup que pct es la proporcentaje de fraudes, # que es el parámetro que queremos estimar prob_sosp &lt;- 0.98 * pct / 100 + 0.001 * (1 - pct / 100) log_prob &lt;- n_sosp * log(prob_sosp) + (n - n_sosp) * log(1- prob_sosp) log_prob } log_verosimilitud_pct } La verosimilitud es una función de \\(p\\). log_verosim &lt;- crear_log_verosim(n = 2000, n_sosp = 4) A continuación la mostramos de manera gráfica. No se ve muy claro dónde ocurre el máximo, pero podemos ampliar cerca de cero la misma gráfica: Vemos que alrededor de 0.1% maximiza la probabilidad de haber observado 4 transacciones sospechosas. Notamos sin embargo que varios valores alrededor de este valor tienen probabilidad similar, así que también son consistentes con los datos (por ejemplo, valores como 0.05 o 0.15 tienen probabilidad similar). Tendremos que considerar esto para evaluar la incertidumbre en nuestra estimación. Obsérvese adicionalmente que si no tomáramos en cuenta las probabilidades de falsos negativos y falsos positivos la estimación simple daría \\(4/2000 = 0.002\\) (0.2%), que es dos veces más grande que nuestra estimación puntual por máxima verosimilitud. Ejemplo. Este es un ejemplo donde mostramos que cuando el soporte de las densidades teóricas es acotado hay que tener cuidado en la definición de la verosimilitud. En este caso, el soporte de la variable aleatoria es el párametro de interés. Supongamos que nuestros datos son generados por medio de una distribución uniforme en el intervalo \\([0,b].\\) Contamos con una muestra de \\(n\\) observaciones generadas de manera independiente \\(X_i \\sim U[0,b]\\) para \\(i= 1, \\ldots, n.\\) Sin embargo, no conocemos el valor de \\(b\\). ¿Cómo es la función de log verosimilitud \\({\\mathcal L}_n(b)\\) para este caso? Nótese que cuando el parámetro \\(b\\) es menor que alguna \\(x_i\\), tenemos que \\({\\mathcal L}_n(b) = 0\\): la verosimilitud es cero si tomamos una \\(b\\) más chica que algún dato, pues este valor es incosistente del todo con los datos observados. En otro caso, \\[{\\mathcal L}_n(b) = \\frac{1}{b^n},\\] pues la función de densidad de una uniforme en \\([0,b]\\) es igual a \\(1/b\\) en el intervalo \\([0,b]\\), y 0 en otro caso. Podemos escribir entonces: crear_verosim &lt;- function(x){ n &lt;- length(x) verosim &lt;- function(b){ indicadora &lt;- ifelse(all(x &lt;= b), 1, 0) indicadora / b^n } } Ahora podemos hacer máxima verosimilitud para un ejemplo: set.seed(234) x &lt;- runif(10, 0, 3) verosim &lt;- crear_verosim(x) res_opt &lt;- optimize(verosim, c(-1000, 1000), maximum = TRUE) res_opt$maximum ## [1] 2.788167 Y nótese que, como esperaríamos, este valor es el máximo de la muestra: max(x) ## [1] 2.788158 La gráfica de la función de verosimilitud es: tibble(b = seq(-1, 5, 0.001)) %&gt;% mutate(verosim_1 = map_dbl(b, ~ verosim(.x))) %&gt;% ggplot() + geom_line(aes(x = b, y = verosim_1)) + geom_rug(data = tibble(x = x), aes(x = x), colour = &quot;red&quot;) Podemos escribir en una fórmula como: \\[\\begin{align} \\mathcal{L}(b; x_1, \\ldots, x_n) = \\prod_{i = 1}^n 1_{[0,b]}(x_i) \\frac1b. \\end{align}\\] Y podríamos resolver analíticamente como sigue: Si consideramos \\[ \\hat b_{\\textsf{MLE}} = x_{\\max} = \\max\\{x_i\\},\\] notemos que cualquier valor observado necesariamente satisface \\[x_i \\leq \\hat b_{\\textsf{MLE}},\\] y por lo tanto todas las funciones indicadoras están encendidas. El valor de la verosimilitud es igual a \\[\\mathcal{L}(\\hat b_{\\textsf{MLE}}; x_1, \\ldots, x_n) = \\left(\\frac{1}{x_{\\max}}\\right)^n \\geq \\left (\\frac1b\\right )^n\\] para cualquier \\(b\\geq x_{\\max}\\). Como la verosimilitud para \\(b&lt;x_{\\max}\\) es igual a cero, esto demuestra que el máximo de la muestra es el estimador de máxima verosimilitud de \\(b\\). Observación. Este ejemplo también tiene dificultades numéricas, pues la verosimilitud presenta discontinuidades y regiones con derivada igual a cero, y la mayoria de los algoritmos numéricos no tienen garantías buenas de covergencia al máximo en estos casos. Si aplicamos sin cuidado descenso en gradiente, por ejemplo, podríamos comenzar incorrectamente en un valor \\(b_0 &lt; x_{\\max}\\) y el algoritmo no avanzaría al máximo. Máxima verosimilitud para más de un parámetro Si nuestro modelo contiene más de un parámetro desconocido podemos también usar máxima verosimilitud. En este caso, optimizamos sobre todos los parámetros usando cálculo o alguna rutina numérica. Ejemplo. Considera el caso de \\(n\\) muestras iid de un modelo Gaussiano. Es decir, \\(X_1, \\ldots, X_n \\sim \\mathsf{N}(\\mu, \\sigma^2).\\) Consideremos que ambos parámetros son desconocidos y nos gustaria encontrar el \\(\\textsf{MLE}\\). Para este problema denotamos \\(\\theta \\in \\mathbb{R}^2\\), donde \\(\\theta_1 = \\mu\\) y \\(\\theta_2 = \\sigma^2.\\) La función de verosimiltud se puede calcular (ignorando algunas constantes multiplicativas) como \\[\\begin{align} \\mathcal{L}_n(\\theta) &amp;= \\prod_{i = 1}^n \\frac{1}{\\sigma} \\, \\exp\\left( - \\frac{(x_i - \\mu)^2}{2\\sigma^2}\\right) \\\\ &amp;= \\theta_2^{-\\frac{n}{2}}\\exp\\left( - \\frac{1}{2 \\theta_2} \\sum_{i = 1}^n (x_i - \\theta_1)^2 \\right). \\end{align}\\] A continuación mostramos la representación gráfica de la función de verosimilitud de este ejemplo. Notamos lo mismo que para los ejemplos anteriores. Conforme más datos tenemos, más nos acercamos a los valores reales que no conocemos. Ejemplo. Como ejercicio, podemos encontrar los estimadores de máxima verosimilitud cuando tenemos una muestra \\(X_1, \\ldots, X_n \\sim \\mathsf{N}(\\mu, \\sigma^2).\\) (puedes derivar e igualar el cero para encontrar el mínimo). También podemos resolver numéricamente, por ejemplo: Supongamos que tenemos la siguiente muestra: set.seed(41852) muestra &lt;- rnorm(150, mean = 1, sd = 2) La función generadora de la log verosimilitud para una muestra es (ve la expresión del ejercicio anterior y calcula su logaritmo), y generamos la función de verosimilitud para nuestra muestra: crear_log_p &lt;- function(x){ log_p &lt;- function(pars){ media = pars[1] desv_est = pars[2] # ve la ecuación del ejercicio anterior z &lt;- (x - media) / desv_est log_verosim &lt;- -(log(desv_est) + 0.5 * mean(z ^ 2)) log_verosim } log_p } log_p &lt;- crear_log_p(muestra) Ahora optimizamos: res &lt;- optim(c(0, 0.5), log_p, control = list(fnscale = -1, maxit = 1000), method = &quot;Nelder-Mead&quot;) res$convergence ## [1] 0 est_mv &lt;- tibble(parametro = c(&quot;media&quot;, &quot;sigma&quot;), estimador = res$par) %&gt;% column_to_rownames(var = &quot;parametro&quot;) est_mv ## estimador ## media 1.136001 ## sigma 1.838421 Verifica que el estimador de la media y de la desviación estándar es el que esperábamos (y que puedes derivar analíticamente): n &lt;- length(muestra) sd_n &lt;- function(x) sqrt(mean((x - mean(x))^2)) c(media = mean(muestra), sigma = sd_n(muestra)) %&gt;% round(4) ## media sigma ## 1.1364 1.8392 Ejemplo. Supongamos que en una población de estudiantes tenemos dos tipos: unos llenaron un examen de opción múltiple al azar (1 de 5), y otros contestaron las preguntas intentando sacar una buena calificación. Suponemos que una vez que conocemos el tipo de estudiante, todas las preguntas tienen la misma probabilidad de ser contestadas correctamente, de manera independiente. El modelo teórico está representado por la siguiente simulación: sim_formas &lt;- function(p_azar, p_corr){ tipo &lt;- rbinom(1, 1, 1 - p_azar) if(tipo==0){ # al azar x &lt;- rbinom(1, 10, 1/5) } else { # no al azar x &lt;- rbinom(1, 10, p_corr) } x } Y una muestra se ve como sigue: set.seed(12) muestra &lt;- map_dbl(1:200, ~ sim_formas(0.3, 0.75)) qplot(muestra) Supongamos que no conocemos la probabildad de contestar correctamente ni la proporción de estudiantes que contestó al azar. ¿Como estimamos estas dos cantidades? Escribimos la verosimilitud: crear_log_p &lt;- function(x){ log_p &lt;- function(pars){ p_azar = pars[1] p_corr = pars[2] sum(log(p_azar * dbinom(x, 10, 1/5) + (1 - p_azar) * dbinom(x, 10, p_corr))) } log_p } Creamos la función de verosimilitud con los datos log_p &lt;- crear_log_p(muestra) y optimizamos res &lt;- optim(c(0.5, 0.5), log_p, control = list(fnscale = -1)) res$par ## [1] 0.2827061 0.7413276 En este caso, obtenemos estimaciones razonables de ambos parámetros. Nota: dependiendo de los datos, este problema puede estar mal condicionado. Por ejemplo, ¿qué pasa si la probabilidad de acertar cuando se contesta bien está cercano al azar? La siguiente pregunta qué nos interesa hacer es: ¿cómo estimamos la variabilidad de estos estimadores? Más adelante veremos una respuesta basada en teoría, pero también podemos resolver este problema usando el bootstrap. Referencias "],["bootstrap-paramétrico.html", "Sección 7 Bootstrap paramétrico Ventajas y desventajas de bootstrap paramétrico Verificando los supuestos distribucionales Modelos mal identificados", " Sección 7 Bootstrap paramétrico Cuando nuestras observaciones provienen de un modelo teórico parametrizado con algunos parámetros que queremos estimar, y utilizamos máxima verosimilitud para hacer nuestra estimación, no es adecuado aplicar directamente el bootstrap no paramétrico que vimos en las secciones anteriores. Sin embargo, suponiendo que el modelo paramétrico que estamos usando es apropiado, podemos remuestrear de tal modelo para estimar la varianza de nuestros estimadores. Este proceso se llama el bootstrap paramétrico. Antes de hacer una definición precisa, veamos cómo calcularíamos error estándar para los estimadores de máxima verosimilitud de la normal que vimos en la sección anterior. Ejemplo (sección máxima verosimilitud). Como ejercicio, podemos encontrar los estimadores de máxima verosimilitud cuando tenemos una muestra \\(X_1, \\ldots, X_n \\sim \\mathsf{N}(\\mu, \\sigma^2)\\) (puedes derivar e igualar a cero para encontrar el mínimo). También podemos resolver numéricamente. Supongamos que tenemos la siguiente muestra: set.seed(41852) muestra &lt;- rnorm(150, mean = 1, sd = 2) La función generadora de la log-verosimilitud para una muestra es (ve la expresión del ejercicio anterior y calcula su logaritmo), y generamos la función de verosimilitud para nuestra muestra: crear_log_p &lt;- function(x){ log_p &lt;- function(pars){ media = pars[1] desv_est = pars[2] # ve la ecuación del ejercicio anterior z &lt;- (x - media) / desv_est log_verosim &lt;- -(log(desv_est) + 0.5 * mean(z^2)) log_verosim } log_p } log_p &lt;- crear_log_p(muestra) Ahora optimizamos (revisa que el método converge): res &lt;- optim(c(0, 0.5), log_p, control = list(fnscale = -1, maxit = 1000), method = &quot;Nelder-Mead&quot;) res$convergence ## [1] 0 est_mle &lt;- tibble(parametro = c(&quot;media&quot;, &quot;sigma&quot;), estimador = res$par) |&gt; column_to_rownames(var = &quot;parametro&quot;) Una vez que tenemos nuestros estimadores puntuales, est_mle ## estimador ## media 1.136001 ## sigma 1.838421 Sustituimos estos parámetros en la distribución normal y simulamos una muestra del mismo tamaño que la original: simular_modelo &lt;- function(n, media, sigma){ rnorm(n, media, sigma) } muestra_bootstrap &lt;- simular_modelo(length(muestra), est_mle[&quot;media&quot;, &quot;estimador&quot;], est_mle[&quot;sigma&quot;, &quot;estimador&quot;]) head(muestra_bootstrap) ## [1] 1.8583885 2.2084326 2.5852895 2.5174462 -0.7428032 0.5995989 Una vez que tenemos esta muestra bootstrap recalculamos los estimadores de máxima verosimlitud. Esto se hace optimizando: # creamos nueva verosimilitud para muestra bootstrap log_p_boot &lt;- crear_log_p(muestra_bootstrap) # optimizamos res_boot &lt;- optim(c(0, 0.5), log_p_boot, control = list(fnscale = -1, maxit = 1000), method = &quot;Nelder-Mead&quot;) res_boot$convergence ## [1] 0 est_mle_boot &lt;- tibble(parametro = c(&quot;media&quot;, &quot;sigma&quot;), estimador = res_boot$par) |&gt; column_to_rownames(var = &quot;parametro&quot;) est_mle_boot ## estimador ## media 1.235914 ## sigma 1.710042 Y esta es nuestra replicación bootstrap de los estimadores de máxima verosimilitud. La idea es la misma que el bootstrap no paramétrico, con la ventaja de que estamos simulando del modelo que suponemos es el correcto, es decir, estamos usando información adicional que no teníamos en el bootstrap no paramétrico. Ahora es necesario repetir un número grande de veces. Nótese que esta función envuelve el proceso de remuestreo, cálculo de la función de verosimilitud y optimización: rep_boot &lt;- function(rep, crear_log_p, est_mle, n){ muestra_bootstrap &lt;- simular_modelo(length(muestra), est_mle[&quot;media&quot;, &quot;estimador&quot;], est_mle[&quot;sigma&quot;, &quot;estimador&quot;]) log_p_boot &lt;- crear_log_p(muestra_bootstrap) # optimizamos res_boot &lt;- optim(c(0, 0.5), log_p_boot, control = list(fnscale = -1, maxit = 1000), method = &quot;Nelder-Mead&quot;) try(if(res_boot$convergence != 0) stop(&quot;No se alcanzó convergencia.&quot;)) tibble(parametro = c(&quot;media&quot;, &quot;sigma&quot;), estimador_boot = res_boot$par) } reps_boot &lt;- map_dfr(1:5000, ~ rep_boot(.x, crear_log_p, est_mle, n = length(muestra)), rep = &quot;.id&quot;) reps_boot ## # A tibble: 10,000 × 2 ## parametro estimador_boot ## &lt;chr&gt; &lt;dbl&gt; ## 1 media 0.797 ## 2 sigma 1.90 ## 3 media 1.23 ## 4 sigma 1.96 ## 5 media 1.14 ## 6 sigma 1.89 ## 7 media 1.33 ## 8 sigma 1.73 ## 9 media 1.19 ## 10 sigma 1.73 ## # ℹ 9,990 more rows Ya ahora podemos estimar error estándar: error_est &lt;- reps_boot |&gt; group_by(parametro) |&gt; summarise(ee_boot = sd(estimador_boot)) error_est ## # A tibble: 2 × 2 ## parametro ee_boot ## &lt;chr&gt; &lt;dbl&gt; ## 1 media 0.150 ## 2 sigma 0.106 Así que nuestra estimación final sería: bind_cols(est_mle, error_est) |&gt; mutate(across(where(is.numeric), round, 3)) |&gt; select(parametro, estimador, ee_boot) ## parametro estimador ee_boot ## media media 1.136 0.150 ## sigma sigma 1.838 0.106 Si usamos la rutina estándar de R (dejaremos para después explicar cómo calcula los errores estándar esta rutina —no es con bootstrap): broom::tidy(MASS::fitdistr(muestra, &quot;normal&quot;)) ## # A tibble: 2 × 3 ## term estimate std.error ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 mean 1.14 0.150 ## 2 sd 1.84 0.106 Podemos revisar también la normalidad aproximada de las distribuciones bootstrap para construir nuestros intervalos: ggplot(reps_boot, aes(sample = estimador_boot)) + geom_qq() + geom_qq_line(colour = &quot;red&quot;) + facet_wrap(~parametro, scales = &quot;free_y&quot;) La distribuciones son aproximadamente normales. Nótese que esto no siempre sucede, especialmente con parámetros de dispersión como \\(\\sigma\\). Ejemplo. Supongamos que tenemos una muestra más chica. Repasa los pasos para asegurarte que entiendes el procedimiento: set.seed(4182) muestra &lt;- rnorm(6, mean = 1, sd = 2) # función de verosimilitud log_p &lt;- crear_log_p(muestra) # máxima verosimilitud res &lt;- optim(c(0, 0.5), log_p, control = list(fnscale = -1, maxit = 1000), method = &quot;Nelder-Mead&quot;) res$convergence ## [1] 0 est_mle &lt;- tibble(parametro = c(&quot;media&quot;, &quot;sigma&quot;), estimador = res$par) |&gt; column_to_rownames(var = &quot;parametro&quot;) est_mle ## estimador ## media 0.3982829 ## sigma 2.3988969 Hacemos bootstrap paramétrico reps_boot &lt;- map_dfr(1:5000, ~ rep_boot(.x, crear_log_p, est_mle, n = length(muestra)), .id = &quot;rep&quot;) reps_boot ## # A tibble: 10,000 × 3 ## rep parametro estimador_boot ## &lt;chr&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 1 media 0.789 ## 2 1 sigma 0.945 ## 3 2 media -0.103 ## 4 2 sigma 1.37 ## 5 3 media 1.96 ## 6 3 sigma 1.70 ## 7 4 media 1.55 ## 8 4 sigma 2.28 ## 9 5 media -0.228 ## 10 5 sigma 1.73 ## # ℹ 9,990 more rows ggplot(reps_boot, aes(sample = estimador_boot)) + geom_qq() + geom_qq_line(colour = &quot;red&quot;) + facet_wrap(~parametro, scales = &quot;free_y&quot;) ggplot(reps_boot, aes(x = estimador_boot)) + geom_histogram() +facet_wrap(~parametro) Donde vemos que la distribución de \\(\\sigma\\) es asimétrica pues en algunos casos obtenemos estimaciones muy cercanas a cero. Podemos usar intervalos de percentiles. Ejercicio (extra). Con más de un parámetro, podemos preguntarnos cómo dependen las estimaciones individuales - en algunos casos pueden estar correlacionadas. Podemos examinar este comportamiendo visualizando las replicaciones bootstrap ggplot(reps_boot |&gt; pivot_wider(names_from = parametro, values_from = estimador_boot), aes(x = media, y = sigma)) + geom_point(alpha = 0.5) + coord_equal() Esta es nuestra aproximación a la distribución de remuestreo de nuestro par de estadísticas \\((\\mu_{\\mathsf{MLE}}, \\sigma_{\\mathsf{MLE}})\\). En este caso, parecen ser independientes (lo cual es posible demostrar). Bootstrap paramétrico. Supongamos que tenemos una muestra iid \\(X_1,X_2,\\ldots, X_n \\sim f(x;\\theta)\\) de un modelo paramétrico, y un estimador de máxima verosimilitud \\(\\hat{\\theta}_{\\mathsf{MLE}}\\) para \\(\\theta\\). El error estándar estimado para \\(\\hat{\\theta}_{\\mathsf{MLE}}\\) por medio del bootstrap paramétrico se calcula como sigue: Se calcula \\(\\hat{\\theta}_{\\mathsf{MLE}}\\) para la muestra observada Se simula una muestra iid de tamaño \\(n\\) de \\(f(x; \\hat{\\theta}_{\\mathsf{MLE}})\\) (muestra bootstrap) Se recalcula el estimador de máxima verosimilitud para la muestra bootstrap \\(\\hat{\\theta^*}_{\\mathsf{MLE}}\\) Se repiten 2-3 una cantidad grande de veces (1000 - 10000) Se calcula la desviación estándar de los valores \\(\\hat{\\theta^*}_{\\mathsf{MLE}}\\) obtenidos. Este es el error estándar estimado para el estimador \\(\\hat{\\theta}_{\\mathsf{MLE}}\\) Ventajas y desventajas de bootstrap paramétrico Ventaja: el bootstrap paramétrico puede dar estimadores más precisos e intervalos más angostos y bien calibrados que el no paramétrico, siempre y cuando el modelo teórico sea razonable. Desventaja: Es necesario decidir el modelo teórico, que tendrá cierto grado de desajuste vs. el proceso generador real de los datos. Si el ajuste es muy malo, los resultados tienen poca utilidad. Para el no paramétrico no es necesario hacer supuestos teóricos. Ventaja: el bootstrap paramétrico puede ser más escalable que el no paramétrico, pues no es necesario cargar y remuestrear los datos originales, y tenemos mejoras adicionales cuando tenemos expresiones explícitas para los estimadores de máxima verosimilitud (como en el caso normal, donde es innecesario hacer optimización numérica). Desventaja: el bootstrap paramétrico es conceptualmente más complicado que el no paramétrico, y como vimos arriba, sus supuestos pueden ser más frágiles que los del no-paramétrico. Verificando los supuestos distribucionales Como hemos discutido antes, podemos hacer pruebas de hipótesis para revisar si una muestra dada proviene de una distribución conocida. Sin embargo, la herramienta más común es la de los qq-plots, donde podemos juzgar fácilmente el tamaño de las desviaciones y si estas tienen implicaciones prácticas importantes. El proceso es como sigue: si \\(X_1,X_,\\ldots, X_n\\) es una muestra de \\(f(x;\\theta)\\), calculamos el estimador de máxima verosimilitud \\(\\theta_{\\mathsf{MLE}}\\) con los datos observados. Enchufamos \\(\\hat{f} = f(x;\\theta_{\\mathsf{MLE}})\\), y hacemos una gráfica de los cuantiles teóricos de \\(\\hat{f}\\) contra los cuantiles muestrales. Ejemplo. Consideramos la siguiente muestra: set.seed(32) muestra &lt;- rgamma(150, 0.4, 1) qplot(muestra) Y queremos usar un modelo exponencial. Encontramos los estimadores de maxima verosimilitud est_mle &lt;- MASS::fitdistr(muestra, &quot;exponential&quot;) rate_mle &lt;- est_mle$estimate rate_mle ## rate ## 2.76054 g_exp &lt;- ggplot(tibble(muestra = muestra), aes(sample = muestra)) + geom_qq(distribution = stats::qexp, dparams = list(rate = rate_mle)) + geom_abline() + labs(subtitle = &quot;Gráfica de cuantiles exponenciales&quot;) g_exp Donde vemos que el desajuste es considerable, y los datos tienen una cola derecha considerablemente más larga que la de exponencial (datos son casi dos veces más grande de lo que esperaríamos), y la cola izquierda está más comprimida en los datos que en una exponencial. Sin embargo, si ajustamos una gamma: est_mle &lt;- MASS::fitdistr(muestra, &quot;gamma&quot;)$estimate g_gamma &lt;- ggplot(tibble(muestra = muestra), aes(sample = muestra)) + geom_qq(distribution = stats::qgamma, dparams = list(shape = est_mle[1], rate = est_mle[2])) + geom_abline() + labs(subtitle = &quot;Gráfica de cuantiles gamma&quot;) g_exp + g_gamma El ajuste es considerablemente mejor para la distribución gamma (puedes hacer el protocolo rorschach para afinar tu diagnóstico de este tipo de gráficas). Ejempĺo. Examinamos un modelo teórico para las cuentas totales del conjunto de datos de propinas. En primer lugar: Separamos comida y cena, pues sabemos que las cuentas tienden a ser más grandes en las cenas. En lugar de modelar la cuenta total, modelamos el gasto por persona, es decir, la cuenta total dividida por el numero de personas. Grupos grandes pueden producir colas largas que no tenemos necesidad de modelar de manera probabilística, pues conocemos el número de personas. En este caso intentaremos un modelo lognormal, es decir, el logaritmo de los valores observados se comporta aproximadamente normal. Puedes también intentar con una distribución gamma. Separamos por Cena y Comida, dividimos entre número de personas y probamos ajustando un modelo para cada horario: propinas &lt;- read_csv(&quot;data/propinas.csv&quot;) |&gt; mutate(cuenta_persona = cuenta_total / num_personas) propinas_mle &lt;- propinas |&gt; group_by(momento) |&gt; summarise(est_mle = list(tidy(MASS::fitdistr(cuenta_persona, &quot;lognormal&quot;)))) |&gt; unnest(est_mle) propinas_mle ## # A tibble: 4 × 4 ## momento term estimate std.error ## &lt;chr&gt; &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 Cena meanlog 2.03 0.0273 ## 2 Cena sdlog 0.362 0.0193 ## 3 Comida meanlog 1.94 0.0366 ## 4 Comida sdlog 0.302 0.0259 Ojo: estos parámetros están en escala logarítmica. Puedes revisar aquí para ver cómo calcular media y desviación estándar de las distribuciones originales. Ahora verificamos el ajuste: g_1 &lt;- ggplot(propinas |&gt; filter(momento == &quot;Cena&quot;), aes(sample = cuenta_persona)) + geom_qq(dparams = list(mean = propinas_mle$estimate[1], sd = propinas_mle$estimate[2]), distribution = stats::qlnorm) + ylim(c(0, 20)) + geom_abline() + labs(subtitle = &quot;Cena&quot;) g_2 &lt;- ggplot(propinas |&gt; filter(momento == &quot;Comida&quot;), aes(sample = cuenta_persona)) + geom_qq(dparams = list(mean = propinas_mle$estimate[3], sd = propinas_mle$estimate[4]), distribution = stats::qlnorm) + ylim(c(0, 20)) + geom_abline() + labs(subtitle = &quot;Comida&quot;) g_1 + g_2 El ajuste es bueno, aunque podríamos revisar la cola de la derecha en la Comida: ¿por qué existen esos valores relativamente grandes (alrededor de 25% más altos de lo que esperaríamos). ¿Tiene sentido ajustar dos distribuciones con parámetros separados? ¿Crees que estas dos distribuciones podrían compartir algún parámetro? Para esto puedes revisar el error estándar de los estimadores de máxima verosimilitud que mostramos arriba. ¿Qué ventajas verías en usar menos parámetros? ¿Cómo implementarías la estimación? ¿Qué pasa si intentas ajustar un modelo normal a estos datos? Modelos mal identificados Para algunos modelos y algunos parámetros, puede suceder que existan varias configuraciones muy diferentes de los parámetros que sean consistentes con los datos (en términos de verosimilitud, tienen verosimilitud alta similar), y en estos casos decimos que el modelo (con los datos observados) está mal identificado. Esto presenta problemas múltiples: optimizar es más difícil, hay incertidumbre grande en la estimación, y los parámetros se acoplan, haciendo difícil su interpretación. Ejemplo. Consideremos el ejemplo donde queríamos estimar dos proporciones: la proporción de examenes contestados al azar y la tasa de correctos. Vamos a suponer que la probabilidad de tener respuesta correcta dado que el examen no fue contestado al azar no es muy lejano a 1/5, que es la probabilidad de acertar a al azar. Aquí está la función para simular y la log verosimilitud correspondiente. Aquí vamos a ver un problema más difícil, así que usaremos la transformación logit para las proporciones, y no obtener resultados fuera del rango 0-1 al optimizar: inv_logit &lt;- function(theta){ exp(theta) / (1 + exp(theta)) } logit &lt;- function(p){ log(p / (1-p)) } # Simular datos sim_formas &lt;- function(probs){ p_azar &lt;- probs[1] p_corr &lt;- probs[2] tipo &lt;- rbinom(1, 1, 1 - p_azar) if(tipo==0){ # al azar x &lt;- rbinom(1, 10, 1/5) } else { # no al azar x &lt;- rbinom(1, 10, p_corr) } x } simular_modelo &lt;- function(n, params){ muestra &lt;- map_dbl(1:n, ~ sim_formas(probs = inv_logit(params))) muestra } # log verosimilitud crear_log_p &lt;- function(x){ log_p &lt;- function(pars){ p_azar = inv_logit(pars[1]) p_corr = inv_logit(pars[2]) sum(log(p_azar * dbinom(x, 10, 1/5) + (1 - p_azar) * dbinom(x, 10, p_corr))) } log_p } # simular datos set.seed(12) muestra &lt;- simular_modelo(2000, params = logit(c(0.3, 0.29))) qplot(muestra) log_p &lt;- crear_log_p(muestra) res &lt;- optim(c(0.0, 0.0), log_p, control = list(fnscale = -1)) res$convergence ## [1] 0 est_mle &lt;- res$par names(est_mle) &lt;- c(&quot;p_azar_logit&quot;, &quot;p_corr_logit&quot;) est_mle ## p_azar_logit p_corr_logit ## -0.9194029 -0.8896454 probs_mle &lt;- inv_logit(est_mle) names(probs_mle) &lt;- c(&quot;p_azar&quot;, &quot;p_corr&quot;) probs_mle ## p_azar p_corr ## 0.2850796 0.2911830 En primer lugar, parece ser que nuestras estimaciones son menos precias. Vamos a hacer bootstrap paramétrico: rep_boot &lt;- function(rep, simular, crear_log_p, pars, n){ muestra_bootstrap &lt;- simular(n, pars) log_p_boot &lt;- crear_log_p(muestra_bootstrap) # optimizamos res_boot &lt;- optim(c(0.0, 0.0), log_p_boot, control = list(fnscale = -1)) try(if(res_boot$convergence != 0) stop(&quot;No se alcanzó convergencia.&quot;)) est_mle_boot &lt;- res_boot$par names(est_mle_boot) &lt;- names(pars) est_mle_boot[&quot;rep&quot;] &lt;- rep est_mle_boot[&quot;convergence&quot;] &lt;- res_boot$convergence est_mle_boot } set.seed(8934) reps_boot &lt;- map(1:500, ~ rep_boot(.x, simular_modelo, crear_log_p, est_mle, n = length(muestra))) |&gt; bind_rows() reps_boot |&gt; mutate(across(everything(), round, 2)) |&gt; head() ## # A tibble: 6 × 4 ## p_azar_logit p_corr_logit rep convergence ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 -1.14 -0.92 1 0 ## 2 -1.11 -0.85 2 0 ## 3 -1.17 -0.95 3 0 ## 4 -2.74 -1.01 4 0 ## 5 -1.05 -0.93 5 0 ## 6 -0.91 -0.87 6 0 El optimizador encontró resultados que no tienen sentido: ggplot(reps_boot, aes(x = inv_logit(p_azar_logit), y = inv_logit(p_corr_logit), colour = factor(convergence))) + geom_point(show.legend = FALSE) + xlab(&quot;p_azar&quot;) + ylab(&quot;p_corr&quot;) Y notamos un problema grave: Tenemos mucha variación en nuestros estimadores, y la correlación entre las estimaciones es alta. Esto deberíamos haberlo esperado, pues como las probabilidades de contestar correctamente son muy similares a las de contestar al azar: Existen muchas combinaciones de parámetros que son consistentes con los datos. Decimos entonces que este modelo está mal identificado con estos datos. La mala identificación, como vemos, es una propiedad tanto de modelo como datos. ¿Qué conclusiones acerca del examen obtienes al ver estas simulaciones bootstrap? ¿Cómo se deberían reportar estos resultados? Qué pasa en este ejemplo si la \\(p_{corr}\\) es más grande, o el tamaño de muestra es más grande? Repite el ejercicio con los parámetros del primer ejemplo (\\(p_{azar} = 0.3, p_{corr}=0.75\\)) y el mismo tamaño de muestra. ¿Qué sucede en este caso? En el caso extremo, decimos que el modelo no está indentificado, y eso generalmente sucede por un problema en el planteamiento del modelo, independientemente de los datos. ¿Puedes imaginar un modelo así? "],["propiedades-teóricas-de-mle.html", "Sección 8 Propiedades teóricas de MLE Consistencia Equivarianza del \\(\\textsf{MLE}\\) Normalidad asintótica Optimalidad del \\(\\textsf{MLE}\\)", " Sección 8 Propiedades teóricas de MLE El método de máxima verosimiltud es uno de los métodos más utilizados en la inferencia estadística paramétrica. En esta sección estudiaremos las propiedades teóricas que cumplen los estimadores de máxima verosimilitud (\\(\\textsf{MLE}\\)) y que han ayudado en su casi adopción universal. Estas propiedades de los \\(\\textsf{MLE}\\) son válidas siempre y cuando el modelo \\(f(x; \\theta)\\) satisfaga ciertas condiciones de regularidad. En particular veremos las condiciones para que los estimadores de máxima verosimilitud sean: consistentes, asintóticamente normales, asintóticamente insesgados, asintóticamente eficientes, y equivariantes. Los estimadores \\(\\textsf{MLE}\\) en ocasiones son malinterpretados como una estimación puntual en la inferencia, y por ende, incapaces de cuantificar incertidumbre. A lo largo de estas notas hemos visto cómo extraer intervalos de confianza por medio de simulación y por lo tanto incorporar incertidumbre en la estimación. Sin embargo, hay otros maneras de reportar incertidumbre para \\(\\textsf{MLE}\\). Y hablaremos de ello en esta sección. A lo largo de esta sección asumiremos muestras de la forma \\[\\begin{align} X_1, \\ldots, X_n \\overset{\\text{iid}}{\\sim} f(x; \\theta^*), \\end{align}\\] donde \\(\\theta^*\\) es el valor verdadero —que suponemos desconocido pero fijo— del parámetro \\(\\theta \\in \\Theta\\), y sea \\(\\hat \\theta_n\\) el estimador de máxima verosimilitud de \\(\\theta.\\) Ejemplo Usaremos este ejemplo para ilustrar los diferentes puntos teóricos a lo largo de esta sección. Consideremos el caso de una muestra de variables binarias que registran el éxito o fracaso de un experimento. Es decir, \\(X_1, \\ldots, X_n \\sim \\textsf{Bernoulli}(p),\\) donde el párametro desconocido es el procentaje de éxitos. Éste último denotado por \\(p.\\) Este ejemplo lo hemos estudiado en secciones pasadas, y sabemos que el \\(\\textsf{MLE}\\) es \\[\\begin{align} \\hat p_n = \\frac{S_n}{n} = \\bar X_n, \\end{align}\\] donde \\(S_n= \\sum_i X_i\\) es el número total de éxitos en la muestra. La figura siguiente ilustra el estimador \\(\\hat p_n\\) como función del número de observaciones en nuestra muestra. Podemos apreciar cómo el promedio parece estabilizarse alrededor del verdadero valor de \\(p^* = 0.25\\) cuando tenemos una cantidad suficientemente grande de observaciones. Como es de esperarse, diferentes muestras tendrán diferentes valores de \\(n\\) dónde las trayectorias parezca que se haya estabilizado (Ver figura siguiente). Sin embargo, se puede notar que este comportamiento parece estar controlado y son raras las trayectorias que se encuentran más lejos. Los conceptos siguientes nos permitirán cuantificar el porcentaje de trayectorias que se mantienen cercanas a \\(p^*,\\) en el caso límite de un número grande de observaciones, cuando trabajemos con estimadores de máxima verosimilitud. Más aún, nos permitirán cracterizar la distribución para dicho límite y aprenderemos de otras propiedades bajo este supuesto asintótico. Consistencia Es prudente pensar que para un estimador, lo que nos interesa es que conforme más información tengamos, más cerca esté del valor desconocido. Esta propiedad la representamos por medio del concepto de consistencia. Para hablar de esta propiedad necesitamos definir un tipo de convergencia para una secuencia de variables aleatorias, convergencia en probabilidad. Definición. Una sucesión de variables aleatorias \\(X_n\\) converge en probabilidad a la variable aleatoria \\(X,\\) lo cual denotamos por \\(X_n \\overset{P}{\\rightarrow} X\\), si para toda \\(\\epsilon \\gt 0\\), \\[\\lim_{n \\rightarrow \\infty} \\mathbb{P}(|X_n - X| &gt; \\epsilon) = 0.\\] Ahora, definimos un estimador consistente como: Definición. Un estimador \\(\\tilde \\theta_n\\) es consistente si converge en probabilidad a \\(\\theta^*.\\) Donde \\(\\theta^*\\) denota el verdadero valor del parámetro, que asumimos fijo. En particular, los estimadores \\(\\textsf{MLE}\\) son consistentes. Teorema. Sea \\(X_n \\sim f(X; \\theta^*),\\) una muestra iid, tal que \\(f(X; \\theta)\\) cumple con ciertas condiciones de regularidad. Entonces, \\(\\hat \\theta_n,\\) el estimador de máxima verosimilitud, converge en probabilidad a \\(\\theta^*.\\) Es decir, \\(\\hat \\theta_n\\) es consistente. La demostración de este teorema la pueden encontrar en Wasserman (2013). Ejemplo El estimador \\(\\hat p_n\\) es consistente. Esto quiere decir que el estimador se vuelve más preciso conforme obtengamos más información. En general esta es una propiedad que los estimadores deben satisfacer para ser útiles en la práctica. La figura siguiente muestra el estimador \\(\\hat p_n\\) como función del número de observaciones utilizado. Distintas curvas corresponden a distintas realizaciones de muestras obtenidas del modelo (\\(B = 500\\)). Nota que la banda definida por \\(\\epsilon\\) se puede hacer tan pequeña como se requiera, lo único que sucederá es que necesitaremos un mayor número de observaciones para garantizar que las trayectorias de los estimadores \\(\\hat p_n\\) se mantengan dentro de las bandas con alta probabilidad. Equivarianza del \\(\\textsf{MLE}\\) Muchas veces nos interesa reparametrizar la función de verosimilitud con el motivo de simplificar el problema de optimización asociado, o simplemente por conveniencia interpretativa. Por ejemplo, si el parámetro de interés es tal que \\(\\theta \\in [a, b],\\) entonces encontrar el \\(\\textsf{MLE}\\) se traduce en optimizar la log-verosimilitud en el espacio restringido al intervalo \\([a,b].\\) En este caso, los métodos tradicionales de búsqueda local por descenso en gradiente podrían tener problemas de estabilidad cuando la búsqueda se realice cerca de las cotas. El concepto de equivarianza nos dice que si el cambio de coordenadas parametrales está definida, y si este cambio de variable se realiza por medio de una función bien comportada (derivable y cuya derivada no es cero), entonces la solución de encontrar el \\(\\textsf{MLE}\\) en las coordenadas originales y transformar, es igual a realizar la inferencia en las coordenadas fáciles. Teorema. Sea \\(\\tau = g(\\theta)\\) una función de \\(\\theta\\) bien comportada. Entonces si \\(\\hat \\theta_n\\) es el \\(\\textsf{MLE}\\) de \\(\\theta,\\) entonces \\(\\hat \\tau_n = g(\\hat \\theta_n)\\) es el \\(\\textsf{MLE}\\) de \\(\\tau.\\) Ejemplo El concepto de equivarianza lo ilustraremos para nuestro ejemplo de esta sección. En particular la parametrización la realizamos por cuestiones de interpretación como un factor de riesgo. Como hemos visto estimador \\(\\hat p_n\\) es equivariante. Es importante mencionar que esta propiedad es general para cualquier tamaño de muestra. Es decir, no descansa en supuestos de muestras grandes. Supongamos que nos interesa estimar el momio de éxitos (bastante común en casas de apuestas). El momio está definido como \\[ \\theta = \\frac{p}{1-p},\\] y podemos rescribir la función de verosimilitud en términos de este parámetro. Sustituyendo \\(p = \\frac{\\theta}{1+\\theta}\\) en \\(\\mathcal{L}_n(p)\\) obtenemos \\[\\begin{align} \\mathcal{L}_n(\\theta) = \\left( \\frac{\\theta}{1 + \\theta} \\right)^{S_n} \\left(\\frac{1}{1 + \\theta} \\right)^{n - S_n}, \\end{align}\\] cuya función encuentra su máximo en \\[\\begin{align} \\hat \\theta_n = \\frac{\\bar X_n}{ 1 - \\bar X_n}. \\end{align}\\] Comprueba que el estimador de arriba para \\(\\theta\\) es el MLE. Normalidad asintótica Está propiedad nos permite caracterizar la distribución asintótica del MLE. Es decir, nos permite caracterizar la incertidumbre asociada una muestra suficientemente grande por medio de una distribución Gaussiana. Esto es, bajo ciertas condiciones de regularidad, \\[\\hat \\theta_n \\overset{.}{\\sim} \\mathsf{N}( \\theta^*, \\mathsf{ee}^2),\\] donde \\(\\mathsf{ee}\\) denota el error estándar del \\(\\textsf{MLE},\\) \\(\\mathsf{ee} = \\mathsf{ee}(\\hat \\theta_n) = \\sqrt{\\mathbb{V}(\\hat \\theta_n)}\\). Esta distribución se puede caracterizar de manera aproximada por métodos analíticos. Para esto necesitamos las siguientes definiciones. Definición. La función de score está definida como \\[\\begin{align} s(X; \\theta) = \\frac{\\partial \\log f(X; \\theta)}{\\partial \\theta}. \\end{align}\\] La información de Fisher está definida como \\[\\begin{align} I_n(\\theta) &amp;= \\mathbb{V}\\left( \\sum_{i = 1}^ns(X_i; \\theta) \\right) \\\\ &amp;= \\sum_{i = 1}^n \\mathbb{V} \\left(s(X_i; \\theta) \\right) \\end{align}\\] Estas cantidades nos permiten evaluar qué tan fácil será identificar el mejor modelo dentro de la familia parámetrica \\(f(X; \\theta)\\). La función de score nos dice qué tanto cambia locamente la distribución cuando cambiamos el valor del parámetro. Calcular la varianza, nos habla de la dispersión de dicho cambio a lo largo del soporte de la variable aleatoria \\(X.\\) Si \\(I_n(\\theta)\\) es grande entonces el cambio de la distribución es muy importante. Esto quiere decir que la distribución es muy diferente de las distribuciones cercanas que se generen al evaluar en \\(\\theta\\)s diferentes. Por lo tanto, si \\(I_n(\\theta)\\) es grande, la distribución será fácil de identificar cuando hagamos observaciones. La información de Fisher también nos permite caracterizar de forma analítica la varianza asíntotica del \\(\\textsf{MLE}\\) pues la aproximación \\(\\mathsf{ee}^2 \\approx \\frac{1}{I_n(\\theta^*)}\\) es válida. El siguiente resultado utiliza la propiedad de la función de score: \\(\\mathbb{E}[s(X; \\theta)] = 0,\\) que implica que \\(\\mathbb{V} \\left(s(X_i; \\theta) \\right) = \\mathbb{E}[s^2(X; \\theta)],\\) y permite a su vez un cómputo más sencillo de la información de Fisher. Teorema. El cálculo de la información de Fisher para una muestra de tamaño \\(n\\) se puede calcular de manera simplificada como \\(I_n(\\theta) = n \\, I(\\theta).\\) Por otro lado, tenemos la siguiente igualdad \\[ I(\\theta) = - \\mathbb{E}\\left( \\frac{\\partial^2 \\log f(X; \\theta)}{\\partial \\theta^2} \\right).\\] Con estas herramientas podemos formular el teorema siguiente. Teorema. Bajo ciertas condiciones de regularidad se satisface que \\(\\mathsf{ee} \\approx \\sqrt{1/I_n(\\theta^*)}\\) y \\[ \\hat \\theta_n \\overset{d}{\\rightarrow} \\mathsf{N}( \\theta^*, \\mathsf{ee}^2).\\] El resultado anterior es teóricamente interesante y nos asegura un comportamiento controlado conforme tengamos más observaciones disponibles. Sin embargo, no es práctico pues no conocemos \\(\\theta^*\\) en la práctica y por consiguiente no conoceríamos la varianza. Sin embargo, también podemos aplicar el principio de plug-in y caracterizar la varianza de la distribución asintótica por medio de \\[\\hat{\\mathsf{ee}} = \\sqrt{1/I_n(\\hat \\theta_n)}.\\] Esto último nos permite constuir intervalos de confianza, por ejemplo al 95%, a través de \\[ \\hat \\theta_n \\pm 2 \\, \\hat{\\mathsf{ee}}.\\] Asimismo, el teorema de Normalidad asintótica nos permite establecer que el \\(\\textsf{MLE}\\) es asíntoticamente insesgado. Es decir, \\[\\lim_{n \\rightarrow \\infty}\\mathbb{E}[\\hat \\theta_n] = \\theta^*.\\] Definición. Sea una muestra \\(X_1, \\ldots, X_n \\overset{iid}{\\sim} f(X; \\theta^*)\\). Un estimador \\(\\tilde \\theta_n\\) es insesgado si satisface que \\[\\mathbb{E}[\\tilde \\theta_n] =\\theta^*.\\] El sesgo del estimador es precisamente la diferencia: \\(\\textsf{Sesgo} = \\mathbb{E}[\\tilde \\theta_n] - \\theta^*.\\) Ejemplo: Información de Fisher En el caso Bernoulli obtenemos \\(I_n(\\theta) = \\frac{n}{\\theta(1-\\theta)}\\), si \\(n = 20\\) podemos comparar con \\(\\theta=0.5, 0.7, 0.8\\), library(patchwork) # Verosimilitud X_1,...,X_n ~ Bernoulli(theta) L_bernoulli &lt;- function(n, S){ function(theta){ theta ^ S * (1 - theta) ^ (n - S) } } xy &lt;- data.frame(x = 0:1) l_b1 &lt;- ggplot(xy, aes(x = x)) + stat_function(fun = L_bernoulli(n = 20, S = 10)) + xlab(expression(theta)) + ylab(expression(L(theta))) + labs(title = &quot;Verosimilitud&quot;, subtitle = &quot;n=20, S = 10&quot;) + ylim(0, 5e-05) l_b2 &lt;- ggplot(xy, aes(x = x)) + stat_function(fun = L_bernoulli(n = 20, S = 14)) + xlab(expression(theta)) + ylab(expression(L(theta))) + labs(title = &quot;Verosimilitud&quot;, subtitle = &quot;n=20, S = 14&quot;) + ylim(0, 5e-05) l_b3 &lt;- ggplot(xy, aes(x = x)) + stat_function(fun = L_bernoulli(n = 20, S = 16)) + xlab(expression(theta)) + ylab(expression(L(theta))) + labs(title = &quot;Verosimilitud&quot;, subtitle = &quot;n=20, S = 19&quot;) + ylim(0, 5e-05) l_b1 + l_b2 + l_b3 Ejemplo: Normalidad Regresando a nuestro ejemplo. Veremos empiricamente que el estimador \\(\\hat \\theta_n\\) es asintóticamente normal. Esta propiedad la hemos visto anteriormente para un caso muy particular. Lo vimos en el TLC para el caso de promedios, \\(\\bar X_n,\\) que en nuestro ejemplo corresponde a \\(\\hat p_n\\). Como hemos visto, esta propiedad la satisface cualquier otro estimador que sea máximo verosímil. Por ejemplo, podemos utilizar el \\(\\mathsf{MLE}\\) de los momios. La figura que sigue muestra la distribución de \\(\\hat \\theta_n\\) para distintas remuestras \\((B = 500)\\) con distintos valores de \\(n.\\) El gráfico anterior valida empíricamente la distribución asintótica para casos de muchas observaciones. A continuación ilustraremos cómo explotar este resultado para obtener intervalos de confianza. Para el caso de \\(\\hat p_n\\) hemos visto que el error estándar se calcula analíticamente como \\[\\textsf{ee}_p^2 = \\mathbb{V}(\\hat p_n) = \\mathbb{V}\\left(\\frac1n \\sum_{i = 1}^n x_i\\right) = \\frac{p^* (1 - p^*)}{n}.\\] Éste empata con el valor del error estándar asintótico \\[\\textsf{ee}_p^2 \\approx \\sqrt{\\frac{1}{I_n(p^*)}},\\] pues la información de Fisher es igual a \\[I_n(p) = n \\, I(p) = \\frac{n}{p ( 1- p)}.\\] En este caso podemos utilizar el estimador plug-in, \\(\\hat{\\textsf{ee}}_p = \\textsf{ee}_p(\\hat p_n).\\) Para estimar el momio, \\(\\theta,\\) el cálculo no es tan fácil pues tendríamos que calcular de manera analítica la varianza de un cociente \\[\\textsf{ee}_\\theta^2 = \\mathbb{V}\\left( \\frac{\\hat p_n}{1-\\hat p_n}\\right).\\] Utilizando la distirbución asintótica, el error estándar se puede calcular mediante \\[\\textsf{ee}_\\theta \\approx \\sqrt{\\frac{1}{I_n(\\theta^*)}} = \\sqrt{\\frac{\\theta (1 + \\theta)^2 }{n}}.\\] A continuación mostramos los errores estándar para nuestro ejemplo utilizando la distribución asintótica y por medio de la distribución de bootstrap. Como es de esperarse, ambos coinciden para muestras relativamente grandes. # Genero muestra muestras &lt;- tibble(tamanos = 2**seq(4,7)) %&gt;% mutate(obs = map(tamanos, ~rbernoulli(., p = p_true))) calcula_momio &lt;- function(x){ x / (1 - x) } calcula_ee_momio &lt;- function(x){ sqrt(((1+x)**2) * x) } # Calculo MLE muestras_est &lt;- muestras %&gt;% group_by(tamanos) %&gt;% mutate(media_hat = map_dbl(obs, mean), media_ee = sqrt(media_hat * (1 - media_hat)/tamanos), momio_hat = calcula_momio(media_hat), momio_ee = calcula_ee_momio(momio_hat)/sqrt(tamanos)) # Calculo por bootstrap muestras_boot &lt;- muestras_est %&gt;% group_by(tamanos) %&gt;% mutate(sims_muestras = map(tamanos, ~rerun(1000, sample(muestras %&gt;% filter(tamanos == ..1) %&gt;% unnest(obs) %&gt;% pull(obs), size = ., replace = TRUE))), sims_medias = map(sims_muestras, ~map_dbl(., mean)), sims_momios = map(sims_medias, ~map_dbl(., calcula_momio)), media_boot = map_dbl(sims_medias, mean), momio_boot = map_dbl(sims_momios, mean), media_ee_boot = map_dbl(sims_medias, sd), momio_ee_boot = map_dbl(sims_momios, sd) ) ## # A tibble: 4 × 5 ## # Groups: tamanos [4] ## tamanos momio_hat momio_boot momio_ee momio_ee_boot ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 16 0.333 0.367 0.192 0.236 ## 2 32 0.333 0.342 0.136 0.140 ## 3 64 0.123 0.123 0.0492 0.0498 ## 4 128 0.407 0.417 0.0793 0.0800 Comprueba las fórmulas para los errores estándar tanto para la probabilidad de éxito como para los momios. El método delta El ejercicio anterior nos sugiere una pregunta natural: Cómo establecer la distribución asintótica de un estimador cuando ya se conoce la de una pre-imagen de él? Es decir, si ya conocemos la distribución de \\(\\theta,\\) podemos establecer la distribución de \\(\\tau = g(\\theta)?\\) La respuesta es afirmativa y la enunciamos por medio de un teorema. El resultado se conoce como el método delta. Teorema. Si \\(\\tau = g(\\theta)\\) es una función diferenciable y \\(g&#39;(\\theta) \\neq 0\\), entonces \\[\\hat \\tau_n \\overset{d}{\\rightarrow} \\mathsf{N}( \\tau^*, \\hat{\\mathsf{ee}}^2_\\tau),\\] donde \\(\\hat \\tau_n = g(\\hat \\theta_n)\\) y \\[\\hat{\\mathsf{ee}}_\\tau = \\bigg| g&#39;(\\hat \\theta_n) \\bigg| \\times \\hat{\\mathsf{ee}}_\\theta(\\hat \\theta_n).\\] Por ejemplo, este resultado lo podemos utilizar para nuestro experimento de Bernoullis. Pues \\(g(p) = \\frac{p}{1-p}\\) es una función diferenciable y por lo tanto \\[\\hat{\\mathsf{ee}}_\\theta = \\sqrt{\\frac1n} \\times \\left[ \\hat p_n^{1/2} (1-\\hat p_n)^{-3/2}\\right].\\] Comprueba la fórmula del método delta para el momio en función de la fracción de éxitos, y también comprueba que de el mismo resultado analítico que habías calculado en el ejercicio anterior. Optimalidad del \\(\\textsf{MLE}\\) Consideremos el caso de una muestra iid \\(X_1, \\ldots, X_n \\sim \\mathsf{N}(\\theta, \\sigma^2).\\) Y consideremos dos estimadores para \\(\\theta.\\) El primero será la media muestral \\(\\bar X_n\\) y el segundo la mediana muestral, la cual denotaremos por \\(\\tilde \\theta_n.\\) Sabemos que ambos son insesgados. Por lo tanto, en promedio emiten estimaciones correctas. Pero ¿cómo escogemos cual utilizar? Un criterio para comparar estimadores es el error cuadrático medio (\\(\\textsf{ECM}\\), por sus siglas en inglés). Definición. El error cuadrático medio de un estimador \\(\\tilde \\theta_n\\) se calcula como \\[\\textsf{ECM}[\\tilde \\theta_n] = \\mathbb{E}[(\\tilde \\theta_n - \\theta^*)^2].\\] Por lo tanto, el \\(\\textsf{ECM}\\) mide la distancia promedio entre el estimador y el valor verdadero valor del parámetro. La siguiente igualdad es bastante útil para comparar dos estimadores. \\[\\textsf{ECM}[\\tilde \\theta_n] = \\mathbb{V}\\left(\\tilde \\theta_n\\right) + \\textsf{Sesgo}\\left[\\tilde \\theta_n\\right]^2.\\] Por lo tanto si dos estimadores son insesgados, uno es más eficiente que el otro si su varianza es menor. La media sabemos que es el \\(\\textsf{MLE}\\) y por el TCL tenemos que \\[\\sqrt{n} \\left( \\bar X_n - \\theta \\right) \\overset{d}{\\rightarrow} \\mathsf{N}( 0, \\sigma^2).\\] La mediana, en contraste, tiene una distribución asintótica \\[\\sqrt{n} \\left( \\tilde X_n - \\theta \\right) \\overset{d}{\\rightarrow} \\mathsf{N}\\left( 0, \\sigma^2 \\frac{\\pi}{2}\\right),\\] es decir tiene una varianza ligeramente mayor. Por lo tanto, decimos que la mediana tiene una eficiencia relativa con respecto a la media del \\(.63 \\% (\\approx 2/\\pi)\\). Es decir, la mediana sólo utliza una fracción de los datos comparado con la media. El siguiente teorema, la desigualdad de Cramer-Rao, nos permite establecer esta resultado de manera mas general para cualquier estimador insesgado. Teorema. Sea \\(\\tilde \\theta_n\\) cualquier estimador insesgado de \\(\\theta\\) cuyo valor verdadero es \\(\\theta^*,\\) entonces \\[\\begin{align} \\mathbb{V}(\\tilde \\theta_n) \\geq \\frac{1}{n I(\\theta^*)}. \\end{align}\\] Un estimador insesgado que satisfaga esta desigualdad se dice que es eficiente. Nota que el lado derecho de la desigualdad es precisamente la varianza asintótica del \\(\\textsf{MLE}.\\) Por lo tanto, éste es asintóticamente eficiente. Es importante hacer enfásis en que la optimalidad del \\(\\textsf{MLE}\\) es un resultado asintótico. Es decir, sólo se satisface cuando tenemos un número suficiente de observaciones. Qué tan grande debe ser el tamaño de muestra varia de problema a problema. Es por esto que para muestras de tamaño finito se prefieren estimadores que minimicen el \\(\\textsf{ECM},\\) como cuando hacemos regresión ridge o utilizamos el estimador James–Stein para un vector de medias. Referencias "],["más-de-pruebas-de-hipótesis-e-intervalos.html", "Sección 9 Más de pruebas de hipótesis e intervalos Prueba de Wald Observación: pruebas \\(t\\) y práctica estadística Prueba de Wald para dos medias o proporciones Datos pareados Pruebas de cociente de verosimilitud Otro tipo de pruebas Errores tipo I y tipo II Consideraciones prácticas Pruebas múltiples", " Sección 9 Más de pruebas de hipótesis e intervalos En esta sección veremos enfoques más clásicos para analizar una prueba de hipótesis, en particular veremos situaciones donde podemos hacer algunos supuestos teóricos acerca de la distribución de las poblaciones. Esta es una sección complementaria para entender prácticas estadísticas usuales: recuerda que discutimos antes que hacer estimación por intervalos generalmente es más útil que hacer pruebas de hipótesis, y adicionalmente, tenemos también la técnica de pruebas de permutaciones que podemos aplicar en muchos de los casos que discutiremos a continuación. El enfoque básico es el mismo que cuando vimos pruebas de permutaciones: calculamos una estadística de prueba de los datos y luego, con una distribución de referencia (asociada a la hipótesis nula), calculamos un valor-\\(p\\). Si el valor-\\(p\\) es chico, entonces los resultados observados no pueden explicarse fácilmente por variación muestral, y rechazamos la hipótesis nula. Con esta idea básica, y supuestos distribucionales acerca de las poblaciones, podemos construir pruebas que requieren menos cómputo. La desventaja es que hay que checar con cuidado los supuestos distribucionales que hagamos. Si los supuestos son incorrectos, las valores-\\(p\\) no tienen mucho sentido y son difíciles de interpretar. Para esta sección seguiremos más a Wasserman (2013) (capítulo 10), pero puedes revisar también Chihara and Hesterberg (2018) (capítulo 8). Prueba de Wald Como hemos visto, existe normalidad asintótica en varios estimadores que hemos considerado, como medias y proporciones muestrales. También vimos que estimadores de máxima verosimilitud cumplen muchas veces un teorema central del límite. Así que supongamos que tenemos una estadística \\(\\hat{\\theta}_n\\) que estima \\(\\theta\\) y es asintóticamente insesgada y normal. Denotamos por \\(\\hat{\\textsf{ee}}\\) una estimación de su error estándar —hay distintas maneras de hacerlo: por ejemplo, con simulación (bootstrap), o por medios analíticos (teoría). Recuerda que el error estándar de una estadística es la desviación estándar de su distribución de muestreo. Si nos interesa probar la hipótesis de que \\(\\theta = 125\\), por ejemplo, y \\(\\hat{\\theta}_n\\) es aproximadamente normal, entonces podemos construir una distribución de referencia aproximada como sigue: Si la nula es cierta, entonces la distribución de muestreo de \\(\\hat{\\theta}\\) es aproximadamente \\(\\mathsf{N}(125, \\hat{\\textsf{ee}})\\). Esto implica que la siguiente estadística \\(W\\) es aproximadamente normal estándar bajo la nula: \\[W = \\frac{\\hat{\\theta} - 125}{\\hat{\\textsf{ee}}} \\sim \\mathsf{N}(0,1)\\] Por lo que valores lejanos de \\([-2,2]\\), por ejemplo, dan evidencia en contra de la hipótesis nula. Como \\(W\\) no depende de ningún parámetro desconocido, podemos usarla como distribución de referencia para comparar el valor de \\(W\\) que obtuvimos en la muestra. Si observamos para nuestra muestra un valor \\(W=w\\) entonces, el valor-\\(p\\) (dos colas) de esta prueba es, aproximadamente, \\[\\mathsf{valor-}p \\approx P(|Z| &gt; |w|) = 2(1 - \\Phi(|w|))\\] donde \\(Z\\sim \\mathsf{N}(0,1)\\) y \\(\\Phi\\) es su función de distribución acumulada. Ejemplo: media muestral. La media nacional de las escuelas de enlace está alrededor de 454 (matemáticas en 6o. grado). Tomamos una muestra de 180 escuelas del Estado de México, y queremos saber si la media obtenida es consistente o no con la media nacional. Ya que estamos usando como estimador una media de una muestra iid, podemos estimar el error estándar de la media con \\[\\hat{\\textsf{ee}} = s / \\sqrt{n}\\] Obtenemos: set.seed(29) muestra_edomex &lt;- read_csv(&quot;data/enlace.csv&quot;) |&gt; filter(estado == &quot;ESTADO DE MEXICO&quot;) |&gt; sample_n(180) resumen &lt;- muestra_edomex |&gt; summarise(media = mean(mate_6), s = sd(mate_6), n = n()) |&gt; mutate(ee = s / sqrt(n)) resumen ## # A tibble: 1 × 4 ## media s n ee ## &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; &lt;dbl&gt; ## 1 456. 155. 180 11.5 La hipótesis nula es que la media poblacional del Estado de México es igual a 454. Calculamos el valor-\\(p\\) usando la prueba de Wald: dif &lt;- (resumen |&gt; pull(media)) - 454 ee &lt;- resumen |&gt; pull(ee) w &lt;- dif / ee p &lt;- 2 * (1 - pt(abs(w), 179)) p ## [1] 0.8413082 y vemos que esta muestra es consistente con la media nacional. No tenemos evidencia en contra de que la media del estado de México es muy similar a la nacional. Repite esta prueba con una muestra de Chiapas. ¿Qué resultado obtienes? Tenemos entonces: Prueba de Wald. Consideramos probar la hipótesis nula \\(H_0: \\theta = \\theta_0\\) contra la alternativa \\(H_1: \\theta \\neq \\theta_0\\). Suponemos que \\(\\hat{\\theta}_n\\) es asintóticamente normal e insesgada, de modo que bajo la hipótesis nula \\[\\frac{\\hat{\\theta}_n - \\theta_0}{\\hat{\\textsf{ee}}} \\sim \\mathsf{N}(0,1).\\] Entonces el valor-\\(p\\) de la prueba de Wald para esta hipótesis nula es \\[\\mathsf{valor-}p \\approx P(|Z| &gt; |w|) = 2(1 - \\Phi(|w|)).\\] Ejemplo. Podemos hacer la prueba de Wald para proporciones con el estimador usual \\(\\hat{p}_n\\) que estima una proporción poblacional \\(p\\). En este caso, utilizamos la estimación usual del error estándar de \\(\\hat{p}_n\\), que está dada por \\[\\hat{\\textsf{ee}} = \\sqrt{\\frac{\\hat{p}_n(1-\\hat{p}_n)}{n}}.\\] Supongamos por ejemplo que en nuestros datos observamos que en \\(n=80\\) muestras independientes, tenemos \\(x=47\\) éxitos. ¿Es esto consistente con la hipótesis nula \\(p = 0.5\\)? Calcuamos primero: p_hat &lt;- 47 / 80 ee &lt;- sqrt(p_hat * (1 - p_hat) / 80) y la estadística \\(W\\) de prueba es: w &lt;- (p_hat - 0.5) / ee w ## [1] 1.58978 Calculamos su valor p: valor_p &lt;- 2 * (1 - pnorm(abs(w))) valor_p ## [1] 0.1118843 Y vemos que en este caso tenemos evidencia baja de que la proporción poblacional es distinta de 0.5. Observación: pruebas \\(t\\) y práctica estadística Con más supuestos distribucionales podemos hacer otros tipos de pruebas donde no requerimos hacer supuestos asintóticos. Por ejemplo, si suponemos que la muestra obtenida \\(X_1,\\ldots, X_n\\) proviene de una distribución normal \\(\\mathsf{N}(\\mu, \\sigma)\\) (cosa que es necesario verificar), entonces es posible demostrar que la estadística \\[T = \\frac{\\bar{X} - \\mu}{S / \\sqrt{n}}\\] tiene una distribución exacta que es \\(t\\) de Student con \\(n-1\\) grados de libertad, y no depende de otros parámetros, de manera que podemos usarla como distribución de referencia y podemos calcular valores \\(p\\) exactos (revisa la sección 8.1 de Chihara and Hesterberg (2018)). La diferencia con usar una prueba de Wald está en que aquí consideramos también la variablidad del error estándar estimado, lo que correctamente sugiere que esperamos variaciones proporcionalmente más grandes en \\(T\\) comparado con lo que sucede si no consideramos esta variación (como en la prueba de Wald). Sin embargo: Si la muestra \\(n\\) es grande, la distribución \\(t\\) de Student con \\(n-1\\) grados de libertad es muy similar a la normal estándar, de manera que la aproximación de Wald es apropiada. Cuando la muestra \\(n\\) es chica, es difícil validar el supuesto de normalidad, a menos que tengamos alguna información adicional acerca de la distribución poblacional. La prueba tiene cierta robustez a desviaciones de normalidad de las observaciones, pero si el sesgo es muy grande, por ejemplo, el supuesto es incorrecto y da valores \\(p\\) distorsionados. Puedes ver aquí, o el apéndice B.11 de Chihara and Hesterberg (2018) para ver descripciones de la distribución \\(t\\) y cómo se compara con una normal estándar dependiendo de los grados de libertad. En muchas ocasiones, en la práctica es común no checar supuestos y saltar directamente a hacer pruebas \\(t\\), lo cual no es muy seguro. Si tenemos duda de esos supuestos, podemos hacer pruebas gráficas o de permutaciones, si son apropiadas. Prueba de Wald para dos medias o proporciones Cuando tenemos dos muestras extraidas de manera independiente de dos poblaciones distintas, y queremos ver si la hipótesis de medias poblacionales iguales es consistente con los datos, podemos usar también una prueba de Wald. Sea \\(\\bar{X}_1\\) y \\(\\bar{X}_2\\) las medias muestrales correspondientes. Si la hipótesis de normalidad aplica para ambas distribuciones muestrales (normalidad asintótica), la variable \\[\\hat{\\delta} = \\bar{X}_1 - \\bar{X}_2\\] es aproximadamente normal con media \\(\\mathsf{N}(\\mu_1 - \\mu_2, \\textsf{ee})\\), donde \\(\\mu_1\\) y \\(\\mu_2\\) son las medias poblacionales correspondientes, y donde el error estándar de \\(\\hat{\\delta}\\) es la raíz de la suma de los cuadrados de los errores estándar de \\(\\bar{X}\\) y \\(\\bar{Y}\\): \\[ \\textsf{ee} = \\sqrt{\\textsf{ee}_1^2 + \\textsf{ee}_{2}^2}.\\] Se sigue entonces que: \\[\\textsf{ee} =\\sqrt{\\frac{\\sigma_1^2}{n_1}+\\frac{\\sigma_2^2}{n_2} }\\] (Nota: usa probabilidad para explicar por qué es cierto esto). De esto se deduce que bajo la hipótesis nula de igualdad de medias \\(\\mu_1 = \\mu_2\\), tenemos que la estadística de Wald \\[W = \\frac{\\hat{\\delta} - 0}{\\sqrt{\\frac{s_1^2}{n_1}+\\frac{s_2^2}{n_2}} } \\sim \\mathsf{N}(0,1)\\] es aproximamente normal estándar. Procedemos entonces a calcular el valor \\(p\\) usando la función de distribución acumulada de la normal estándar. En el caso particular de proporciones, podemos simplificar, como hicimos arriba, a \\[W = \\frac{\\hat{p}_1 - \\hat{p}_2}{\\sqrt{\\frac{\\hat{p}_1(1-\\hat{p}_1)}{n_1}+\\frac{\\hat{p}_2(1-\\hat{p}_2)}{n_2}} } \\sim \\mathsf{N}(0,1)\\] Haz una prueba comparando las medias en enlace de la Ciudad de México vs Estado de México. ¿Hay evidencia de que tienen distintas medias? Ejemplo (Wasserman (2013)). Supongamos tenemos dos conjuntos de prueba para evaluar algoritmos de predicción, de tamaños \\(n_1=100\\) y \\(n_2=250\\) respectivamente, tenemos dos algoritmos para generar predicciones de clase (digamos positivo y negativo). Usaremos el primer conjunto para evaluar el algoritmo 1 y el segundo para evaluar el algoritmo 2. El algoritmo 1 corre en 1 hora, y el algoritmo 2 tarda 24 horas. Supón que obtenemos que la tasa de clasificación correcta del primer algoritmo es \\(\\hat{p}_1 = 0.85\\), y la tasa del segundo es de \\(\\hat{p}_2 = 0.91\\). ¿Estos datos son consistentes con la hipótesis de que los algoritmos tienen desempeño muy similar? Es decir, queremos probar la hipótesis \\(p_1 = p_2\\). Calculamos la estadística de Wald: n_1 &lt;- 100 n_2 &lt;- 250 p_hat_1 &lt;- 0.86 p_hat_2 &lt;- 0.90 ee &lt;- sqrt(p_hat_1 * (1 - p_hat_1) / n_1 + p_hat_2 * (1 - p_hat_2) / n_2) delta = p_hat_1 - p_hat_2 w &lt;- delta / ee w ## [1] -1.011443 que da un valor p de: 2 * (1 - pnorm(abs(w))) ## [1] 0.3118042 Y vemos que valor-\\(p\\) es grande, de forma que los datos son consistentes con la hipótesis de que los algoritmos tienen desempeño similar. ¿Cómo tomaríamos nuestra decisión final? Si la diferencia entre 1 hora y 24 horas no es muy importante, entonces preferíamos usar el algoritmo 2. Sin embargo, si el costo de 24 horas es más alto que 1 hora de corrida, los datos no tienen indicios fuertes de que vayamos a perder en desempeño, y podriamos seleccionar el algoritmo 1. Datos pareados Las pruebas que acabamos de ver para comparar medias requieren poblaciones independientes. Si las dos muestras están pareadas (es decir, son dos mediciones en una misma muestra), podemos tomar considerar las diferencias \\(D_i = X_i - Y_i\\) y utilizar la prueba para una sola muestra con la media \\(\\bar{D}\\). Esta es una prueba de Wald pareada. Ejemplo (Wasserman (2013)). Ahora supongamos que utilizamos la misma muestra de tamaño \\(n=300\\) para probar los dos algoritmos. En este caso, no debemos hacer la prueba para medias de muestras independientes. Sin embargo, esto podemos ponerlo en términos de una prueba para una sola muestra. Tenemos las observaciones \\(X_1,\\ldots, X_n\\) y \\(Y_1,\\dots, Y_n\\), donde \\(X_i=1\\) si el algoritmo 1 clasifica correctamente, y 0 en otro caso. Igualmente, \\(Y_i=1\\) si el algoritmo 2 clasifica correctamente, y 0 en otro caso. Definimos \\[D_i= X_i - Y_i\\] Y \\(D_1,\\ldots, D_n\\) es una muestra iid. Ahora observemos que la media \\(\\bar{D}\\) tiene valor esperado \\(p_1 - p_2\\), donde \\(p_1\\) y \\(p_2\\) son las tasas de correctos del algoritmo 1 y del algoritmo 2 respectivamente. Podemos hacer una prueba de Wald como al principio de la sección: \\[W = \\frac{\\bar{D} - 0}{{\\textsf{ee}}}\\] Y notemos que el error estándar no se calcula como en el ejemplo anterior. Podríamos usar bootstrap para estimarlo, pero en este caso podemos usar el estimador usual \\[\\hat{\\textsf{ee}} = S / \\sqrt{n}\\] donde \\[S = \\frac{1}{n}\\sum_{i=1}^n (D_i - \\bar{D})^2\\] y nótese que necesitamos las decisiones indiviudales de cada algoritmo para cada caso, en contraste al ejemplo anterior de muestras independientes donde los errores estándar se calculaban de manera independiente. Esto tiene sentido, pues la variablidad de \\(\\bar{D}\\) depende de cómo están correlacionados los aciertos de los dos algoritmos. Supongamos por ejemplo que los datos que obtenemos son: datos_clasif |&gt; head() ## # A tibble: 6 × 3 ## caso x y ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 1 1 ## 2 2 0 1 ## 3 3 0 1 ## 4 4 0 1 ## 5 5 0 1 ## 6 6 1 0 Como explicamos arriba, nos interesa la diferencia. Calculamos \\(d\\): datos_clasif &lt;- datos_clasif |&gt; mutate(d = x - y) datos_clasif |&gt; head() ## # A tibble: 6 × 4 ## caso x y d ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 1 1 0 ## 2 2 0 1 -1 ## 3 3 0 1 -1 ## 4 4 0 1 -1 ## 5 5 0 1 -1 ## 6 6 1 0 1 datos_clasif |&gt; summarise(sd_x = sd(x), sd_y = sd(y), sd_d = sd(d)) ## # A tibble: 1 × 3 ## sd_x sd_y sd_d ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 0.393 0.309 0.539 Y ahora calculamos la media de \\(d\\) (y tasa de correctos de cada clasificador:) medias_tbl &lt;- datos_clasif |&gt; summarise(across(where(is.numeric), mean, .names = &quot;{col}_hat&quot;)) d_hat &lt;- pull(medias_tbl, d_hat) medias_tbl ## # A tibble: 1 × 3 ## x_hat y_hat d_hat ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 0.81 0.893 -0.0833 Ahora necesitamos calcular el error estándar. Como explicamos arriba, hacemos ee &lt;- datos_clasif |&gt; mutate(d_hat = mean(d)) |&gt; mutate(dif_2 = (d - d_hat)) |&gt; summarise(ee = sd(dif_2) / sqrt(n())) |&gt; pull(ee) ee ## [1] 0.03112829 Y ahora podemos calcular la estadística \\(W\\) y el valor p correspondiente: w &lt;- d_hat / ee valor_p &lt;- 2 * (1 - pnorm(abs(w))) c(w = w, valor_p = valor_p) |&gt; round(3) ## w valor_p ## -2.677 0.007 Y vemos que tenemos evidencia considerable de que el desempeño no es el mismo: el algoritmo 2 parece ser mejor. ¿Qué pasaría si incorrectamente usaras la prueba de dos muestras para este ejemplo? ¿Qué cosa cambia en la fórmula de la estadística de Wald? Pruebas de cociente de verosimilitud Otra técnica clásica para hacer pruebas de hipótesis es el de cociente de verosimilitudes. Con esta técnica podemos hacer pruebas que involucren varios parámetros, y podemos contrastar hipótesis nulas contra alternativas especificas. Para aplicar este tipo de pruebas es necesario hacer supuestos distribucionales (modelos probabilísticos), pues estas pruebas se basan en la función de verosimilitud \\(\\mathcal{L}(\\theta; x_1,\\ldots, x_n)\\). Ejemplo. Supongamos que tenemos la hipótesis nula de que una moneda es justa (\\(p =0.5\\) de sol). En 120 tiros de la moneda (que suponemos independientes), observamos 75 soles. Recordemos la función de log-verosimilitud para el modelo binomial (ignorando constantes que no dependen de \\(p\\)) es \\[\\ell(p) = 75 \\log(p) + (120 - 75)\\log(1-p) \\] Primero calculamos el estimador de máxima verosimilitud de \\(p\\), que es \\(\\hat{p} = 75/120 = 0.625\\). Evaluamos la verosimilitud \\[\\ell(\\hat{p}) = \\ell(0.625) = 75\\log(0.625) + 45\\log(0.375) = -79.388\\] - Ahora evaluamos la verosimlitud según la hipótesis nula, donde asumimos que \\(p = 0.5\\): \\[\\ell(p_0) = \\ell(0.5) = 75\\log(0.5) + 45\\log(0.5) = -83.177\\] - Finalmente, contrastamos estos dos números con una estadística que denotamos con \\(\\lambda\\): \\[\\lambda = 2\\left[\\ell(\\hat{p}) - \\ell(p_0)\\right] = 2[\\ell(0.625)- \\ell(0.5)] = 2(3.79)=7.58\\] A \\(\\lambda\\) se le llama la estadística de cociente de verosimilitud. Tomamos la diferencia de log verosimilitudes, que es los mismo que tomar el logaritmo del cociente de verosimilitudes, y de ahí el nombre de la prueba. Nótese que cuando este número \\(\\lambda\\) es muy grande, esto implica que la hipótesis nula es menos creíble, o menos consistente con los datos, pues la nula tiene mucho menos verosimilitud de lo que los datos indican. Por otro lado, cuando este valor es cercano a 0, entonces tenemos menos evidencia en contra de la hipótesis nula. Esto se explica en la siguiente gráfica: log_verosim &lt;- function(p){ 75 * log(p) + (120 - 75) * log(1 - p) } verosim_tbl &lt;- tibble(p = seq(0.4, 0.7, 0.01)) |&gt; mutate(log_verosim = log_verosim(p)) ggplot(verosim_tbl, aes(x = p, y = log_verosim)) + geom_line() + geom_segment(x = 75/120, xend = 75/120, y = -130, yend = log_verosim(75/120), colour = &quot;red&quot;) + geom_segment(x = 0.5, xend = 0.5, y = -130, yend = log_verosim(0.5), colour = &quot;gray&quot;) + geom_errorbar(aes(x = 0.5, ymin = log_verosim(0.5), ymax = log_verosim(75/120)), colour = &quot;orange&quot;, width = 0.05) + annotate(&quot;text&quot;, x = 0.48, y = -81.5, label = &quot;3.79&quot;) + annotate(&quot;text&quot;, x = 0.515, y = -91, label =&quot;nula&quot;, colour = &quot;gray20&quot;) + annotate(&quot;text&quot;, x = 0.665, y = -91, label =&quot;max verosímil&quot;, colour = &quot;red&quot;) + labs(subtitle = expression(paste(lambda, &quot;=2(3.79)=7.58&quot;))) Este método puede generalizarse para que no solo aplique a hipótesis nulas donde \\(\\theta = \\theta_0\\), sino en general, \\(\\theta \\in \\Theta_0\\). Por ejemplo, podemos construir pruebas para \\(\\theta &lt; 0.4\\). Definición. Consideramos la hipótesis nula \\(\\theta= \\theta_0\\). La estadística del cociente de verosimilitudes está dada por: \\[\\lambda = 2\\log\\left( \\frac{\\max_{\\theta}\\mathcal{L}(\\theta)}{\\max_{\\theta=\\theta_0}\\mathcal{L}(\\theta)} \\right ) = 2\\log\\left( \\frac{\\mathcal{L}(\\hat{\\theta})}{\\mathcal{L}(\\theta_0)} \\right)\\] donde \\(\\hat{\\theta}\\) es el estimador de máxima verosimilitud. Para construir una prueba asociada, como siempre, necesitamos una distribución de referencia. Esto podemos hacerlo con simulación, o usando resultados asintóticos. Distribución de referencia para pruebas de cocientes Para nuestro ejemplo anterior, podemos simular datos bajo la hipótesis nula, y ver cómo se distribuye la estadística \\(\\lambda\\): Ejemplo. Simulamos bajo la hipótesis nula como sigue: n_volados &lt;- 120 # número de volados simulados_nula &lt;- rbinom(4000, n_volados, p = 0.5) lambda &lt;- function(n, x, p_0 = 0.5){ # estimador de max verosim p_mv &lt;- x / n # log verosimilitud bajo mv log_p_mv &lt;- x * log(p_mv) + (n - x) * log(1 - p_mv) # log verosimllitud bajo nula log_p_nula &lt;- x * log(p_0) + (n - x) * log(1 - p_0) lambda &lt;- 2*(log_p_mv - log_p_nula) lambda } lambda_obs &lt;- lambda(n_volados, 75, 0.5) sims_tbl &lt;- tibble(sim_x = simulados_nula) |&gt; mutate(lambda = map_dbl(sim_x, ~ lambda(n_volados, .x, p_0 = 0.5))) ggplot(sims_tbl, aes(x = lambda)) + geom_histogram(binwidth = 0.7) + geom_vline(xintercept = lambda_obs, color = &quot;red&quot;) Con esta aproximación a la distribución de referencia podemos calcular el valor p en nuestro ejemplo anterior: valor_p &lt;- mean(sims_tbl$lambda &gt;= lambda_obs) valor_p ## [1] 0.00675 y observamos que tenemos evidencia fuerte en contra de la hipótesis nula: la moneda no está balanceada. Ejemplo. Este ejemplo es un poco artificial, pero lo usamos para entender mejor las pruebas de cocientes de verosimlitud. Supongamos que tenemos una muestra de \\(\\mathsf{N}(\\mu, 1)\\), y queremos probar si \\(\\mu = 8\\). Asumimos que el supuesto de normalidad y desviación estándar iugal a 1 se cumplen. set.seed(3341) n_muestra &lt;- 100 muestra_1 &lt;- rnorm(n_muestra, 7.9, 1) crear_log_p &lt;- function(x){ # crear log verosim para dos muestras normales independientes. log_p &lt;- function(params){ mu &lt;- params[1] log_vero &lt;- dnorm(x, mean = mu, sd = 1, log = TRUE) |&gt; sum() log_vero } } lambda_calc &lt;- function(muestra, crear_log_p){ log_p &lt;- crear_log_p(muestra) res &lt;- optim(c(0), log_p, control = list(fnscale = -1)) lambda_mv &lt;- log_p(res$par) lambda_nula &lt;- log_p(8.0) lambda &lt;- 2 * (lambda_mv - lambda_nula) lambda } lambda &lt;- lambda_calc(muestra_1, crear_log_p) lambda ## [1] 2.101775 Ahora construimos con simulación la distribución de referencia usando simulaciones bajo la nula sims_nula &lt;- map(1:10000, ~ rnorm(n_muestra, 8, 1)) lambda_nula_sim &lt;- map_dbl(sims_nula, ~ lambda_calc(.x, crear_log_p)) tibble(lambda = lambda_nula_sim) |&gt; ggplot(aes(x = lambda)) + geom_histogram() + geom_vline(xintercept = lambda, colour = &quot;red&quot;) valor_p &lt;- mean(lambda_nula_sim &gt;= lambda) valor_p ## [1] 0.1537 Estos datos muestran consistencia con la hipótesis \\(\\mu = 8\\). Discusión: Nota en los dos ejemplos anteriores la similitud entre las distribuciones de referencia. En ambos casos, estas distribuciones resultan ser aproximadamente \\(\\chi\\)-cuadrada con 1 grado de libertad (ji-cuadrada). Podemos checar para el último ejemplo: teorica &lt;- tibble(x = seq(0.1, 10, 0.01)) |&gt; mutate(f_chi_1 = dchisq(x, df = 1)) tibble(lambda = lambda_nula_sim) |&gt; ggplot() + geom_histogram(aes(x = lambda, y = ..density..), binwidth = 0.1) + geom_line(data = teorica, aes(x = x, y = f_chi_1), colour = &quot;red&quot;) O mejor, con una gráfica de cuantiles de las simulaciones vs la téorica: tibble(lambda = lambda_nula_sim) |&gt; ggplot(aes(sample = lambda)) + geom_qq(distribution = stats::qchisq, dparams = list(df = 1)) + geom_qq_line(distribution = stats::qchisq, dparams = list(df = 1)) Este resultado asintótico no es trivial, y se usa comúnmente para calcular valores \\(p\\). Discutiremos más este punto más adelante. Otro tipo de pruebas Con cocientes de verosimlitud podemos diseñar pruebas para contrastar condiciones que sólo un subconjunto de parámetros cumple. Ejemplo. Supongamos que queremos hacer una prueba de igualdad de medias \\(\\mu_1 = \\mu_2\\) para dos poblaciones normales \\(\\mathsf{N}(\\mu_1, \\sigma_1)\\) y \\(\\mathsf{N}(\\mu_2, \\sigma_2)\\), donde extraemos las muestras de manera independiente, y no conocemos las desviaciones estándar. Obtenemos dos muestras (que supondremos provienen de distribuciones normales, pues ese es nuestro supuesto) set.seed(223) muestra_1 &lt;- rnorm(80, 0.8, 0.2) muestra_2 &lt;- rnorm(120, 0.8, 0.4) Necesitamos: 1) calcular el valor de la estadística \\(\\lambda\\) de cociente de verosimilitudes, 2) Calcular la distribución de referencia para \\(\\lambda\\) bajo la hipótesis nula y finalmente 3) Ver qué tan extremo es el valor obtenido de \\(\\lambda\\) en relación a la distribución de referencia. crear_log_p &lt;- function(x, y){ # crear log verosim para dos muestras normales independientes. log_p &lt;- function(params){ mu_1 &lt;- params[1] mu_2 &lt;- params[2] sigma_1 &lt;- params[3] sigma_2 &lt;- params[4] log_vero_1 &lt;- dnorm(x, mean = mu_1, sd = sigma_1, log = TRUE) |&gt; sum() log_vero_2 &lt;- dnorm(y, mean = mu_2, sd = sigma_2, log = TRUE) |&gt; sum() log_vero &lt;- log_vero_1 + log_vero_2 #se suman por independiencia log_vero } } log_p &lt;- crear_log_p(muestra_1, muestra_2) crear_log_p_nula &lt;- function(x, y){ log_p &lt;- function(params){ # misma media mu &lt;- params[1] sigma_1 &lt;- params[2] sigma_2 &lt;- params[3] log_vero_1 &lt;- dnorm(x, mean = mu, sd = sigma_1, log = TRUE) |&gt; sum() log_vero_2 &lt;- dnorm(y, mean = mu, sd = sigma_2, log = TRUE) |&gt; sum() log_vero &lt;- log_vero_1 + log_vero_2 #se suman por independiencia log_vero } } log_p_nula &lt;- crear_log_p_nula(muestra_1, muestra_2) Ahora tenemos el problema de que no conocemos las sigma. Estas deben ser estimadas para después calcular el cociente de verosimilitud: res &lt;- optim(c(0, 0, 1, 1), log_p, method = &quot;Nelder-Mead&quot;, control = list(fnscale = -1)) res$convergence ## [1] 0 est_mv &lt;- res$par names(est_mv) &lt;- c(&quot;mu_1&quot;, &quot;mu_2&quot;, &quot;sigma_1&quot;, &quot;sigma_2&quot;) est_mv ## mu_1 mu_2 sigma_1 sigma_2 ## 0.8153471 0.7819913 0.1987545 0.3940484 Y tenemos lambda_1 &lt;- log_p(est_mv) lambda_1 ## [1] -42.76723 Ahora calculamos el máximo bajo el supuesto de la hipótesis nula: res &lt;- optim(c(0, 1, 1), log_p_nula, method = &quot;Nelder-Mead&quot;, control = list(fnscale = -1)) res$convergence ## [1] 0 est_mv_nula &lt;- res$par names(est_mv) &lt;- c(&quot;mu&quot;, &quot;sigma_1&quot;, &quot;sigma_2&quot;) est_mv_nula ## [1] 0.8062091 0.1989438 0.3948603 y evaluamos lambda_2 &lt;- log_p_nula(est_mv_nula) lambda_2 ## [1] -43.07902 Finalmente, nuestra estadística \\(\\lambda\\) es lambda &lt;- 2 * (lambda_1 - lambda_2) lambda ## [1] 0.6235661 Y ahora necesitamos calcular un valor-\\(p\\). El problema que tenemos en este punto es que bajo la hipótesis nula no están determinados todos los parámetros, así que no podemos simular de manera simple muestras para obtener la distribución de referencia. Podemos sin embargo usar bootstrap paramétrico usando los estimadores de máxima verosimilitud bajo la nula simular_boot &lt;- function(n_1, n_2, est_mv_nula){ x &lt;- rnorm(n_1, est_mv_nula[1], est_mv_nula[2]) y &lt;- rnorm(n_2, est_mv_nula[1], est_mv_nula[3]) list(x = x, y = y) } lambda_nula_sim &lt;- function(est_mv_nula){ muestras &lt;- simular_boot(80, 120, est_mv_nula) x &lt;- muestras$x y &lt;- muestras$y log_p &lt;- crear_log_p(x, y) log_p_nula &lt;- crear_log_p_nula(x, y) est_1 &lt;- optim(c(0,0,1,1), log_p, control = list(fnscale = -1)) est_2 &lt;- optim(c(0,1,1), log_p_nula, control = list(fnscale = -1)) lambda &lt;- 2*(log_p(est_1$par) - log_p_nula(est_2$par)) lambda } lambda_sim &lt;- map_dbl(1:2000, ~ lambda_nula_sim(est_mv_nula = est_mv_nula)) Y graficamos la distribución de referencia junto con el valor de \\(\\lambda\\) que obtuvimos: tibble(lambda = lambda_sim) |&gt; ggplot(aes(x = lambda)) + geom_histogram() + geom_vline(xintercept = lambda, colour = &quot;red&quot;) Y claramente los datos son consistentes con medias iguales. El valor-\\(p\\) es mean(lambda_sim &gt; lambda) ## [1] 0.4275 Verificamos una vez más que la distribución de referencia es cercana a una \\(\\chi\\)-cuadrada con un grado de libertad. tibble(lambda = lambda_sim) |&gt; ggplot(aes(sample = lambda)) + geom_qq(distribution = stats::qchisq, dparams = list(df = 1)) + geom_qq_line(distribution = stats::qchisq, dparams = list(df = 1)) Esta es la definición generalizada de las pruebas de cociente de verosimilitudes Definición. Consideramos la hipótesis nula \\(\\theta \\in \\Theta_0\\). La estadística del cociente de verosimilitudes está dada por: \\[\\lambda = 2\\log\\left( \\frac{\\max_{\\theta}\\mathcal{L}(\\theta)}{\\max_{\\theta\\in\\Theta_0}\\mathcal{L}(\\theta)} \\right ) = 2\\log\\left( \\frac{ \\mathcal{L}(\\hat{\\theta})}{\\mathcal{L}(\\hat{\\theta}_0)} \\right)\\] donde \\(\\hat{\\theta}\\) es el estimador de máxima verosimilitud de \\(\\theta\\) y \\(\\hat{\\theta}_0\\) es el estimador de máxima verosimilitud de \\(\\theta\\) cuando restringimos a que \\(\\theta \\in \\Theta_0\\). En nuestro ejemplo anterior, el espacio \\(\\Theta_0\\) era \\(\\{ (\\mu,\\mu,\\sigma_1, \\sigma_2)\\}\\), que es un subconjunto de \\(\\{ (\\mu_1,\\mu_2,\\sigma_1, \\sigma_2)\\}\\). Nótese que el espacio \\(\\Theta_0\\) tiene tres parámetros libres, mientras que el espacio total tiene 4. Aunque podemos usar el bootstrap paramétrico para construir distribuciones de referencia para estas pruebas y calcular un valor-\\(p\\), el siguiente teorema, cuya demostración no es trivial, explica las observaciones que hicimos arriba. Este teorema enuncia la estrategia del enfoque clásico, que utiliza una aproximación asintótica. Valores p para pruebas de cocientes de verosimilitud. Supongamos que \\(\\theta = (\\theta_1,\\theta_2, \\ldots, \\theta_p)\\). Sea \\[\\Theta_0 = \\{\\theta : \\theta_1 = a_1, \\theta_2 = a_2, \\dots, \\theta_q = a_q \\},\\] es decir la hipótesis \\(\\theta \\in \\Theta_0\\) es que los primeros \\(q\\) parámetros de \\(\\theta\\) estan fijos en algún valor. Los otros parámetros no se consideran en esta prueba. Si \\(\\lambda\\) es la estadística de cociente de verosimilitudes de esta prueba, entonces, bajo la nula \\(\\theta \\in \\Theta_0\\) tenemos que la distribución de \\(\\lambda\\) es asintóticamente \\(\\chi\\)-cuadrada con \\(q\\) grados de libertad, denotada por \\(\\chi^2_q\\). El valor-\\(p\\) para esta prueba es \\[P(\\chi^2_{q} &gt; \\lambda)\\] Observaciones: Para hacer cálculos con la distribución \\(\\chi\\)-cuadrada usamos rutinas numéricas (por ejemplo la función pchisq en R). Nótese que \\(p\\) es la dimensión del espacio \\(\\Theta\\) (\\(p\\) parámetros), y que \\(p-q\\) es la dimensión del espacio \\(\\Theta_0\\) (pues \\(q\\) parámetros están fijos), de modo que los grados de libertad son la dimensión de \\(\\Theta\\) menos la dimensión de \\(\\Theta_0\\). En nuestro primer ejemplo (proporción de éxitos) solo teníamos un parámetro. El espacio \\(\\Theta_0\\) es de dimensión 0, así que los grados de libertad son \\(1 = 1 - 0\\) En este último ejemplo donde probamos igualdad de medias, el espacio \\(\\Theta\\) tiene dimensión 4, y el espacio \\(\\Theta_0\\) es de dimensión 3 (tres parámetros libres), por lo tanto los grados de libertad son \\(1 = 4 -3\\). Ejemplo En nuestro ejemplo de prueba de igualdad de medias, usaríamos pchisq(lambda, df =1, lower.tail = FALSE) ## [1] 0.4297252 que es similar al que obtuvimos con la estrategia del bootstrap paramétrico. Errores tipo I y tipo II En algunas ocasiones, en lugar de solamente calcular un valor-\\(p\\) queremos tomar una decisión asociada a distintas hipótesis que consideramos posibles. Por ejemplo, nuestra hipótesis nula podría ser Hipótesis nula \\(H_0\\): Una medicina nueva que estamos probando no es efectiva en reducir el colesterol en pacientes. Y queremos contrastar con una alternativa: Hipótesis alternativa \\(H_A\\): la medicina nueva reduce los niveles de colesterol en los pacientes. La decisión que está detrás de estas pruebas es: si no podemos rechazar la nula, la medicina no sale al mercado. Si rechazamos la nula, entonces la medicina es aprobada para salir al mercado. Para diseñar esta prueba, procedemos como sigue: Definimos cómo recolectar datos \\(X\\) de interés Definimos una estádistica \\(T(X)\\) de los datos. Definimos una región de rechazo \\(R\\) de valores tales que si \\(T(X)\\in R\\), entonces rechazaremos la hipótesis nula (e implícitamente tomaríamos la decisión asociada a la alternativa). Ejecutamos la prueba observando datos \\(X=x\\), calculando \\(T(x)\\), y checando si \\(T(x) \\in R\\). Si esto sucede entonces decimos que rechazamos la hipótesis nula, y tomamos la decisión asociada a la alternativa. Ejemplo. Si tenemos la hipótesis nula \\(p_1=0.5\\) para una proporción, y al alternativa es \\(p_1\\neq 0.5\\), podemos usar la estadística de Wald \\(T(x) = \\frac{\\hat{p_1} - 0.5}{\\hat{\\textsf{ee}}}\\). Podríamos definir la región de rechazo como \\(R = \\{T(x) : |T(x)| &gt; 3 \\}\\) (rechazamos si en valor absoluto la estadística de Wald es mayor que 3). Cuando diseñamos una prueba de este tipo, quisiéramos minimizar dos tipos de errores: Rechazar la hipótesis nula \\(H_0\\) cuando es cierta: Error tipo I No rechazar la hipótesis nula \\(H_0\\) cuando \\(H_0\\) es falsa: Error tipo II La gravedad de cada error depende del problema. En nuestro ejemplo de la medicina, por ejemplo: Un error tipo II resultaría en una medicina efectiva que no sale al mercado, lo que tiene consecuencias financieras (para la farmaceútica) y costos de oportunidad en salud (para la población). Por otra parte, Un error tipo I resultaría en salir al mercado con una medicina que no es efectiva. Esto tiene costos de oportunidad financieros que pueden ser grandes para la sociedad. Todos estos costos dependen, por ejempĺo, de qué tan grave es la enfermedad, qué tan costosa es la medicina, y así sucesivamente. En el enfoque más clásico, los errores tipo I y tipo II generalmente no se balancean según su severidad o probabilidad. En lugar de eso, generalmente se establece un límite para la probabilidad de cometer un error del tipo I (usualmente 5%, por una tradición que no tiene mucho fundamento) En vista de este ejemplo simple, y las observaciones de arriba: Reducir una decisión compleja a una prueba de hipótesis con resultados binarios (rechazar o no) es generalmente erróneo. Las pruebas de hipótesis se usan muchas veces incorrectamente cuando lo más apropiado es usar estimación por intervalos o algo similar que cuantifique la incertidumbre de las estimaciones. Consulta por ejemplo el comunicado de la ASA acerca de p-values y pruebas de hipótesis En el caso de la medicina, por ejemplo, realmente no nos interesa que la medicina sea mejor que un placebo. Nos importa que tenga un efecto considerable en los pacientes. Si estimamos este efecto, incluyendo incertidumbre, tenemos una mejor herramienta para hacer análisis costo-beneficio y tomar la decisión más apropiada. Como dijimos, típicamente se selecciona la región de rechazo de forma que bajo la hipótesis nula la probabilidad de cometer un error tipo I está acotada. Definición. Supongamos que los datos \\(X_1,X_2,\\ldots, X_n\\) provienen de una distribución \\(F_\\theta\\), donde no conocemos \\(\\theta\\). Supongamos que la hipótesis nula es que \\(\\theta = \\theta_0\\) (que llamamos una hipótesis simple). La función de potencia de una prueba con región de rechazo \\(R\\) se define como la probabilidad de rechazar para cada posible valor del parámetro \\(\\theta\\) \\[\\beta(\\theta) = P_\\theta (X\\in R).\\] El tamaño de una prueba se define como el valor \\[\\alpha = \\beta(\\theta_0),\\] es decir, la probabilidad de rechazar la nula (\\(\\theta = \\theta_0\\)) erróneamente. Observación. Esto se generaliza para hipótesis compuestas, donde la nula es que el parámetro \\(\\theta\\) está en un cierto conjunto \\(\\Theta_0\\). Por ejemplo, una hipótesis nula puede ser \\(\\theta &lt; 0.5\\). En este caso, \\(\\alpha\\) se define como el valor más grande que \\(\\beta(\\theta)\\) toma cuando \\(\\theta\\) está en \\(\\Theta_0\\), es decir, la probabilidad de rechazo más grande cuando la hipótesis nula se cumple. Decimos que una prueba tiene nivel de significancia de \\(\\alpha\\) si su tamaño es menor o igual a \\(\\alpha\\). Decimos que la potencia de una prueba es la probabilidad de, correctamente, rechazar la hipótesis nula cuando la alterna es verdadera: \\[\\beta(\\theta_a) = P_{\\theta_a} (X \\in R).\\] Observación: Sería deseable encontrar la prueba con mayor potencia bajo \\(H_a\\), entre todas las pruebas con tamaño \\(\\alpha\\). Esto no es trivial y no siempre existe. Observación: El valor \\(p\\) es el menor tamaño con el que podemos rechazar \\(H_0\\). Ejemplo (Chihara and Hesterberg (2018)) Supongamos que las calificaciones de Enlace de alumnos en México se distribuye aproximadamente como una normal con media 515 y desviación estándar de 120. En una ciudad particular, se quiere decidir si es neceario pedir fondos porque la media de la ciudad es más baja que la nacional. Nuestra hipótesis nula es \\(H_0: \\mu \\geq 515\\) y la alternativa es \\(\\mu &lt; 515\\), así que si rechazamos la nula se pedirían los fondos. Supondremos que la distribución de calificaciones en la ciudad es también aproximadamente normal con desviación estándar de 130. Se plantea tomar una muestra de 100 alumnos, y rechazar si la media muestral \\(\\bar{X}\\) es menor que 505. ¿Cuál es la probabilidad \\(\\alpha\\) de tener un error de tipo I? La función de potencia es \\[\\beta(\\mu) = P_\\mu(\\bar{X} &lt; 505)\\] Restando la media \\(\\mu\\) y estandarizando obtenemos \\[\\beta(\\mu) = P \\left (\\frac{\\bar{X} - \\mu}{130/\\sqrt{100}} &lt; \\frac{505 -\\mu}{130/\\sqrt{100}} \\right )\\] así que \\[\\beta(\\mu) = \\Phi \\left (\\frac{505 -\\mu}{130/\\sqrt{100}}\\right ),\\] donde \\(\\Phi\\) es la función acumulada de la normal estándar. La gráfica de la función potencia es entonces potencia_tbl &lt;- tibble(mu = seq(450, 550, 0.5)) |&gt; mutate(beta = pnorm((505 - mu)/13)) |&gt; # probabilidad de rechazar mutate(nula_verdadera = factor(mu &gt;= 515)) # nula verdadera ggplot(potencia_tbl, aes(x = mu, y = beta, colour = nula_verdadera)) + geom_line() Es decir, si la media \\(\\mu\\) de la ciudad es muy baja, con mucha seguridad rechazamos. Si es relativamente alta entonces no rechazamos. El tamaño de la prueba es el mayor valor de probabilidad de rechazo que se obtiene sobre los valores \\(\\mu\\geq 515\\) (la nula). Podemos calcularlo analíticamente como sigue: Si \\(\\mu \\geq 515\\), entonces \\[\\beta(\\mu) \\leq \\beta(515) = \\Phi\\left (\\frac{505 -515}{130/\\sqrt{100}}\\right ) = \\Phi( - 10 / 13) = \\Phi(-0.7692)\\] que es igual a pnorm(-0.7692) ## [1] 0.2208873 Y este es el tamaño de la prueba. En otras palabras: si la ciudad no está por debajo de la media nacional, hay una probabilidad de 22% de que erróneamente se pidan fondos (al rechazar \\(H_0\\)). Ejemplo Supongamos que los que programan el presupuesto deciden que se requiere tener una probabilidad de a lo más 5% de rechazar erróneamente la hipótesis nula (es decir, pedir fondos cuando en realidad su media no está debajo de la nacional) para poder recibir fondos. ¿Cuál es la región de rechazo que podríamos escoger? En el caso anterior usamos la región \\(\\bar{X}&lt;505\\). Si el tamaño de muestra está fijo en \\(n=100\\) (por presupuesto), entonces tenemos que escoger un punto de corte más extremo. Si la región de rechazo es \\(\\bar{X} &lt; C)\\) entonces tenemos, siguiendo los cálculos anteriores, que \\[0.05 = \\alpha = \\Phi \\left ( \\frac{C -515}{130/\\sqrt{100}}\\right) = \\Phi \\left( \\frac{C- 515}{13} \\right)\\] Buscamos el cuantil 0.05 de la normal estándar, que es z_alpha &lt;- qnorm(0.05) z_alpha ## [1] -1.644854 Y entonces requerimos que \\[\\frac{C- 515}{13} = -1.6448.\\] Despejando obtenemos C &lt;- 13*z_alpha + 515 C ## [1] 493.6169 Así que podemos usar la región \\(\\bar{X} &lt; 493.5\\), que es más estricta que la anterior de \\(\\bar{X} &lt; 505\\). Considera la potencia de la prueba \\(\\beta(\\mu)\\) que vimos arriba. Discute y corre algunos ejemplos para contestar las siguientes preguntas: Recuerda la definición: ¿qué significa \\(\\beta(\\mu)\\)? ¿Qué pasa con la potencia cuando \\(\\mu\\) está más lejos de los valores de la hipótesis nula? ¿Qué pasa con la potencia cuando hay menos variabilidad en la población? ¿Y cuando la muestra es más grande? ¿Qué pasa si hacemos más chico el nivel de significancia? Consideraciones prácticas Algunos recordatorios de lo que hemos visto: Rechazar la nula no quiere decir que la nula es falsa, ni que encotramos un “efecto”. Un valor-\\(p\\) chico tampoco quiere decir que la nula es falsa. Lo que quiere decir es que la nula es poco consistente con los datos que observamos, o que es muy poco probable que la nula produzca los datos que observamos. Rechazar la nula (encontrar un efecto “significativo”) no quiere decir que el efecto tiene importancia práctica. Si la potencia es alta (por ejemplo cuando el tamaño de muestra es grande), puede ser que la discrepancia de los datos con la nula es despreciable, entonces para fines prácticos podríamos trabajar bajo el supuesto de la nula. Por eso en general preferimos hacer estimación que pruebas de hipótesis para entender o resumir los datos y tamaños de las discrepancias. Adicionalmente, muchas de las hipótesis nulas que generalmente se utilizan se pueden rechazar sin datos (por ejemplo, igualdad de proporciones en dos poblaciones reales). Lo que importa es qué tan diferentes son, y qué tan bien podemos estimar sus diferencias. En la literatura, muchas veces parece que “encontrar una cosa interesante” es rechazar una hipótesis nulas con nivel 5% de significancia. Es más importante entender cómo se diseñó el estudio, cómo se recogieron los datos, cuáles fueron las decisiones de análisis que pasar el mítico nivel de 5% Cuando la potencia es baja (por ejemplo porque el tamaño de muestra es muy chico), tenemos que observar diferencias muy grandes para rechazar. Si probamos algo poco factible (por ejemplo, que la vitamina \\(C\\) aumenta la estatura a los adultos), entonces los rechazos generalmente se deben a variabilidad en la muestra (error tipo II). Cuando diseñamos y presentamos resultados de un estudio o análisis, es mejor pensar en describir los datos y su variabilidad, y mostrar estimaciones incluyendo fuentes de incertidumbre, en lugar de intentar resumir con un valor-\\(p\\) o con el resultado de una prueba de hipótesis. Pruebas múltiples En algunas ocasiones se hacen muchas pruebas para “filtrar” las cosas que son interesantes y las que no. Por ejemplo, cuando comparamos miles de genes entre dos muestras (la nula es que son similares). Si cada prueba se conduce a un nivel \\(\\alpha\\), la probablilidad de tener al menos un rechazo falso (un error tipo I) es considerablemente más alta que \\(\\alpha\\). Por ejemplo, si repetimos una prueba de hipótesis con nivel \\(\\alpha\\) con muestras independientes, la probabilidad de tener al menos un rechazo falso es \\(1-(1-\\alpha)^n\\), que es muy cercano a uno si \\(n\\) es grande (¿cómo derivas esta fórmula?). Por ejemplo, si \\(\\alpha = 0.05\\) y \\(n = 100\\), con más de 99% probabilidad tendremos al menos un rechazo falso, o un “hallazgo” falso. Sin \\(n\\) es muy grande, varios de los hallazgos que encontremos serán debidos a variabilidad muestral. Puedes ver en (Wasserman 2013), sección 10.7 métodos conservadores como corrección de Bonferroni (sólo rechazar cuando el valor-\\(p\\) es menor a \\(0.05/n\\)), o la técnica más moderna de control de tasa de descubrimientos falsos (FDR). Cuando estamos en una situación como esta (que es más retadora en cuanto a análisis), sin embargo, sugerimos usar estimaciones que tomen cuenta todos los datos con regularización apropiada: por ejemplo, en lugar de trabajar con cada muestra por separado, intentamos construir un modelo para el proceso completo de muestreo. Una posibilidad son modelos jerárquicos bayesianos. Ver por ejemplo (Gelman, Hill, and Yajima 2012). Referencias "],["introducción-a-inferencia-bayesiana-1.html", "Sección 10 Introducción a inferencia bayesiana Un primer ejemplo completo de inferencia bayesiana Ejemplo: estimando una proporción Ejemplo: observaciones uniformes Probabilidad a priori Análisis conjugado Pasos de un análisis de datos bayesiano Verificación predictiva posterior Predicción", " Sección 10 Introducción a inferencia bayesiana Para esta sección seguiremos principalmente Kruschke (2015). Adicionalmente puedes ver la sección correspondiente de Chihara and Hesterberg (2018). En las secciones anteriores estudiamos el método de máxima verosimilitud y métodos de remuestreo. Esto lo hemos hecho para estimar parámetros, y cuantificar la incertidumbre qué tenemos acerca de valores poblacionales. La inferencia bayesiana tiene objetivos similares. Igual que en máxima verosimilitud, la inferencia bayesiana comienza con modelos probabilísticos y observaciones. En contraste con máxima verosimilitud, la inferencia bayesiana está diseñada para incorporar información previa o de expertos que tengamos acerca de los parámetros de interés. La inferencia bayesiana cubre como caso particular métodos basados en máxima verosimilitud. El concepto probabilístico básico que utilizamos para construir estos modelos y la inferencia es el de probabilidad condicional: la probabilidad de que ocurran ciertos eventos dada la información disponible del fenómeno que nos interesa. Un primer ejemplo completo de inferencia bayesiana Consideremos el siguiente problema: Nos dan una moneda, y solo sabemos que la moneda puede tener probabilidad \\(3/5\\) de tirar sol (está cargada a sol) o puede ser una moneda cargada a águila, con probabilidad \\(2/5\\) de tirar sol. Vamos a lanzar la moneda dos veces y observamos su resultado (águila o sol). Queremos decir algo acerca de qué tan probable es que hayamos tirado la moneda cargada a sol o la moneda cargada a águila. En este caso, tenemos dos variables: \\(X\\), que cuenta el número de soles obtenidos en el experimento aleatorio, y \\(\\theta\\), que da la probabilidad de que un volado resulte en sol (por ejemplo, si la moneda es justa entonces \\(\\theta = 0.5\\)). ¿Qué cantidades podríamos usar para evaluar qué moneda es la que estamos usando? Si hacemos el experimento, y tiramos la moneda 2 veces, podríamos considerar la probabilidad \\[P(\\theta = 0.4 | X = x)\\] donde \\(x\\) es el número de soles que obtuvimos en el experimento. Esta es la probabilidad condicional de que estemos tirando la moneda con probabilidad de sol 2/5 dado que observamos \\(x\\) soles. Por ejemplo, si tiramos 2 soles, deberíamos calcular \\[P(\\theta=0.4|X=2).\\] ¿Cómo calculamos esta probabilidad? ¿Qué sentido tiene? Usando reglas de probabildad (regla de Bayes en particular), podríamos calcular \\[P(\\theta=0.4|X=2) = \\frac{P(X=2 | \\theta = 0.4) P(\\theta =0.4)}{P(X=2)}\\] Nota que en el numerador uno de los factores, \\(P(X=2 | \\theta = 0.4),\\) es la verosimilitud. Así que primero necesitamos la verosimilitud: \\[P(X=2|\\theta = 0.4) = (0.4)^2 = 0.16.\\] La novedad es que ahora tenemos que considerar la probabilidad \\(P(\\theta = 0.4)\\). Esta cantidad no la habíamos encontrado antes. Tenemos que pensar entonces que este parámetro es una cantidad aleatoria, y puede tomar dos valores \\(\\theta=0.4\\) ó \\(\\theta = 0.6\\). Considerar esta cantidad como aleatoria requiere pensar, en este caso, en cómo se escogió la moneda, o qué sabemos acerca de las monedas que se usan para este experimento. Supongamos que en este caso, nos dicen que la moneda se escoge al azar de una bolsa donde hay una proporción similar de los dos tipos de moneda (0.4 ó 0.6). Es decir el espacio parametral es \\(\\Theta = \\{0.4, 0.6\\},\\) y las probabilidades asociadas a cada posibilidad son las mismas. Es decir, tenemos \\[P(\\theta = 0.4) = P(\\theta = 0.6) =0.5,\\] que representa la probabilidad de escoger de manera aleatoria la moneda con una carga en particular. Ahora queremos calcular \\(P(X=2)\\), pero con el trabajo que hicimos esto es fácil. Pues requiere usar reglas de probabilidad usuales para hacerlo. Podemos utilizar probabilidad total \\[\\begin{align} P(X) &amp;= \\sum_{\\theta \\in \\Theta} P(X, \\theta)\\\\ &amp;= \\sum_{\\theta \\in \\Theta} P(X\\, |\\, \\theta) P(\\theta), \\end{align}\\] lo cual en nuestro ejemplo se traduce en escribir \\[ P(X=2) = P(X=2|\\theta = 0.4)P(\\theta = 0.4) + P(X=2|\\theta=0.6)P(\\theta =0.6),\\] por lo que obtenemos \\[P(X=2) = 0.16(0.5) + 0.36(0.5) = 0.26.\\] Finalmente la probabilidad de haber escogido la moneda con carga \\(2/5\\) dado que observamos dos soles en el lanzamiento es \\[P(\\theta=0.4|X=2) = \\frac{0.16(0.5)}{0.26} \\approx 0.31.\\] Es decir, la probabilidad posterior de que estemos tirando la moneda \\(2/5\\) baja de 0.5 (nuestra información inicial) a 0.31. Este es un ejemplo completo, aunque muy simple, de inferencia bayesiana. La estrategia de inferencia bayesiana implica tomar decisiones basadas en las probabilidades posteriores. Finalmente, podríamos hacer predicciones usando la posterior predictiva. Si \\({X}_{nv}\\) es una nueva tirada adicional de la moneda que estamos usando, nos interesaría saber: \\[P({X}_{nv}=\\mathsf{sol}\\, | \\, X=2)\\] Notemos que un volado adicional es un resultado binario. Por lo que podemos calcular observando que \\(P({X}_{nv}|X=2, \\theta)\\) es una variable Bernoulli con probabilidad \\(\\theta\\), que puede valer 0.4 ó 0.6. Como tenemos las probabilidades posteriores \\(P(\\theta|X=2)\\) podemos usar probabilidad total, condicionado en \\(X=2\\): \\[\\begin{align*} P({X}_{nv}=\\mathsf{sol}\\, | \\, X=2) &amp; = \\sum_{\\theta \\in \\Theta} P({X}_{nv}=\\mathsf{sol}, \\theta \\, | \\, X=2) &amp; \\text{(probabilidad total)}\\\\ &amp;= \\sum_{\\theta \\in \\Theta} P({X}_{nv}=\\mathsf{sol}\\, | \\theta , X=2) P(\\theta \\, | \\, X=2) &amp; \\text{(probabilidad condicional)}\\\\ &amp;= \\sum_{\\theta \\in \\Theta} P({X}_{nv}=\\mathsf{sol}\\, | \\theta ) P(\\theta \\, | \\, X=2), &amp; \\text{(independencia condicional)} \\end{align*}\\] lo que nos da el siguiente cálculo \\[P(X_{nv}=\\mathsf{sol}\\, |\\, \\theta=0.4) \\, P(\\theta=0.4|X=2) \\, +\\, P(X_{nv}=\\mathsf{sol}|\\theta = 0.6) \\, P(\\theta =0.6|X=2)\\] Es decir, promediamos ponderando con las probabilidades posteriores. Por lo tanto obtenemos \\[P(X_{nv} = \\mathsf{sol}|X=2) = 0.4 ( 0.31) + 0.6 (0.69) = 0.538.\\] Observación 0 Nótese que en contraste con máxima verosimilitud, en este ejemplo cuantificamos con probabilidad condicional la incertidumbre de los parámetros que no conocemos. En máxima verosimilitud esta probabilidad no tiene mucho sentido, pues nunca consideramos el parámetro desconocido como una cantidad aleatoria. Observación 1 Nótese el factor \\(P(X=2)\\) en la probabilidad posterior puede entenderse como un factor de normalización. Notemos que los denominadores en la distribución posterior son \\[P(X=2 | \\theta = 0.4) P(\\theta =0.4) = 0.16(0.5) = 0.08,\\] y \\[P(X=2 | \\theta = 0.6) P(\\theta =0.6) = 0.36(0.5) = 0.18.\\] Las probabilidades posteriores son proporcionales a estas dos cantidades, y como deben sumar uno, entonces normalizando estos dos números (dividiendo entre su suma) obtenemos las probabilidades. Observación 2 La nomenclatura que usamos es la siguiente: Como \\(X\\) son los datos observados, llamamos a \\(P(X|\\theta)\\) la verosimilitud, o modelo de los datos. A \\(P(\\theta)\\) le llamamos la distribución inicial o previa. La distribución que usamos para hacer inferencia \\(P(\\theta|X)\\) es la distribución final o posterior. Para utilizar inferencia bayesiana, hay que hacer supuestos para definir las primeras dos partes del modelo. La parte de iniciales o previas está ausente de enfoques como máxima verosimlitud usual. Observación 3 ¿Cómo decidimos las probabilidades iniciales, por ejemplo \\(P(\\theta=0.4)\\) ? Quizá es un supuesto y no tenemos razón para pensar que se hace de otra manera. O quizá conocemos el mecanismo concreto con el que se selecciona la moneda. Discutiremos esto más adelante. Observación 4 ¿Cómo decidimos el modelo de los datos? Aquí típicamente también tenemos que hacer algunos supuestos, aunque algunos de estos pueden estar basados en el diseño del estudio, por ejemplo. Igual que cuando usamos máxima verosimilitud, es necesario checar que nuestro modelo ajusta razonablemente a los datos. Ejercicio Cambia distintos parámetros del número de soles observados, las probabilidades de sol de las monedas, y las probabilidades iniciales de selección de las monedas. n_volados &lt;- 2 # posible valores del parámetro desconocido theta = c(0.4, 0.6) # probabilidades iniciales probs_inicial &lt;- tibble(moneda = c(1, 2), theta = theta, prob_inicial = c(0.5, 0.5)) probs_inicial ## # A tibble: 2 × 3 ## moneda theta prob_inicial ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 0.4 0.5 ## 2 2 0.6 0.5 # verosimilitud crear_verosim &lt;- function(no_soles){ verosim &lt;- function(theta){ # prob de observar no_soles en 2 volados con probabilidad de sol theta dbinom(no_soles, n_volados, theta) } verosim } # evaluar verosimilitud verosim &lt;- crear_verosim(2) # ahora usamos regla de bayes para hacer tabla de probabilidades tabla_inferencia &lt;- probs_inicial |&gt; mutate(verosimilitud = map_dbl(theta, verosim)) |&gt; mutate(inicial_x_verosim = prob_inicial * verosimilitud) |&gt; # normalizar mutate(prob_posterior = inicial_x_verosim / sum(inicial_x_verosim)) tabla_inferencia |&gt; mutate(moneda_obs = moneda) |&gt; select(moneda_obs, theta, prob_inicial, verosimilitud, prob_posterior) ## # A tibble: 2 × 5 ## moneda_obs theta prob_inicial verosimilitud prob_posterior ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 0.4 0.5 0.16 0.308 ## 2 2 0.6 0.5 0.36 0.692 ¿Qué pasa cuando el número de soles es 0? ¿Cómo cambian las probabilidades posteriores de cada moneda? Incrementa el número de volados, por ejemplo a 10. ¿Qué pasa si observaste 8 soles, por ejemplo? ¿Y si observaste 0? ¿Qué pasa si cambias las probabilidades iniciales (por ejemplo incrementas la probabilidad inicial de la moneda 1 a 0.9)? Justifica las siguientes aseveraciones (para este ejemplo): Las probabilidades posteriores o finales son una especie de punto intermedio entre verosimilitud y probablidades iniciales. Si tenemos pocas observaciones, las probabilidades posteriores son similares a las iniciales. Cuando tenemos muchos datos, las probabilidades posteriores están más concentradas, y no es tan importante la inicial. Si la inicial está muy concentrada en algún valor, la posterior requiere de muchas observaciones para que se pueda concentrar en otros valores diferentes a los de la inicial. Ahora resumimos los elementos básicos de la inferencia bayesiana, que son relativamente simples: Inferencia bayesiana. Con la notación de arriba: Como \\(X\\) son los datos observados, llamamos a \\(P(X|\\theta)\\) la verosimilitud, proceso generador de datos o modelo de los datos. El factor \\(P(\\theta)\\) le llamamos la distribución inicial o previa. La distribución que usamos para hacer inferencia \\(P(\\theta|X)\\) es la distribución final o posterior Hacemos inferencia usando la ecuación \\[P(\\theta | X) = \\frac{P(X | \\theta) P(\\theta)}{P(X)}\\] que también escribimos: \\[P(\\theta | X) \\propto P(X | \\theta) P(\\theta)\\] donde \\(\\propto\\) significa “proporcional a”. No ponemos \\(P(X)\\) pues como vimos arriba, es una constante de normalización. En estadística Bayesiana, las probablidades posteriores \\(P(\\theta|X)\\) dan toda la información que necesitamos para hacer inferencia. ¿Cuándo damos probablidad alta a un parámetro particular \\(\\theta\\)? Cuando su verosimilitud es alta y/o cuando su probabilidad inicial es alta. De este modo, la posterior combina la información inicial que tenemos acerca de los parámetros con la información en la muestra acerca de los parámetros (verosimilitud). Podemos ilustrar como sigue: Ejemplo: estimando una proporción Consideremos ahora el problema de estimar una proporción \\(\\theta\\) de una población dada usando una muestra iid \\(X_1,X_2,\\ldots, X_n\\) de variables Bernoulli. Ya sabemos calcular la verosimilitud (el modelo de los datos): \\[P(X_1=x_1,X_2 =x_2,\\ldots, X_n=x_n|\\theta) = \\theta^k(1-\\theta)^{n-k},\\] donde \\(k = x_1 + x_2 +\\cdots + x_k\\) es el número de éxitos que observamos. Ahora necesitamos una distribución inicial o previa \\(P(\\theta)\\). Aunque esta distribución puede tener cualquier forma, supongamos que nuestro conocimiento actual podemos resumirlo con una distribución \\(\\mathsf{Beta}(3, 3)\\): \\[P(\\theta) \\propto \\theta^2(1-\\theta)^2.\\] La constante de normalización es 1/30, pero no la requerimos. Podemos simular para examinar su forma: sim_inicial &lt;- tibble(theta = rbeta(10000, 3, 3)) ggplot(sim_inicial) + geom_histogram(aes(x = theta, y = ..density..), bins = 15) De modo que nuestra información inicial es que la proporción puede tomar cualquier valor entre 0 y 1, pero es probable que tome un valor no tan lejano de 0.5. Por ejemplo, con probabilidad 0.95 creemos que \\(\\theta\\) está en el intervalo quantile(sim_inicial$theta, c(0.025, 0.975)) |&gt; round(2) ## 2.5% 97.5% ## 0.14 0.85 Es difícil justificar en abstracto por qué escogeriamos una inicial con esta forma. Aunque esto los detallaremos más adelante, puedes pensar, por el momento, que alguien observó algunos casos de esta población, y quizá vio tres éxitos y tres fracasos. Esto sugeriría que es poco probable que la probablidad \\(\\theta\\) sea muy cercana a 0 o muy cercana a 1. Ahora podemos construir nuestra posterior. Tenemos que \\[P(\\theta| X_1=x_1, \\ldots, X_n=x_n) \\propto P(X_1 = x_1,\\ldots X_n=x_n | \\theta)P(\\theta) = \\theta^{k+2}(1-\\theta)^{n-k + 2}\\] donde la constante de normalización no depende de \\(\\theta\\). Como \\(\\theta\\) es un parámetro continuo, la expresión de la derecha nos debe dar una densidad posterior. Supongamos entonces que hicimos la prueba con \\(n = 30\\) (número de prueba) y observamos 19 éxitos. Tendríamos entonces \\[P(\\theta | S_n = 19) \\propto \\theta^{19 + 2} (1-\\theta)^{30-19 +2} = \\theta^{21}(1-\\theta)^{13}\\] La cantidad de la derecha, una vez que normalizemos por el número \\(P(X=19)\\), nos dará una densidad posterior (tal cual, esta expresion no integra a 1). Podemos obtenerla usando cálculo, pero recordamos que una distribución \\(\\mathsf{\\mathsf{Beta}}(a,b)\\) tiene como fórmula \\[\\frac{1}{B(a,b)} \\theta^{a-1}(1-\\theta)^{b-1}\\] Concluimos entonces que la posterior tiene una distribución \\(\\mathsf{Beta}(22, 14)\\). Podemos simular de la posterior usando código estándar para ver cómo luce: sim_inicial &lt;- sim_inicial |&gt; mutate(dist = &quot;inicial&quot;) sim_posterior &lt;- tibble(theta = rbeta(10000, 22, 14)) |&gt; mutate(dist = &quot;posterior&quot;) sims &lt;- bind_rows(sim_inicial, sim_posterior) ggplot(sims, aes(x = theta, fill = dist)) + geom_histogram(aes(x = theta), bins = 30, alpha = 0.5, position = &quot;identity&quot;) La posterior nos dice cuáles son las posibilidades de dónde puede estar el parámetro \\(\\theta\\). Nótese que ahora excluye prácticamente valores más chicos que 0.25 o mayores que 0.9. Esta distribución posterior es el objeto con el que hacemos inferencia: nos dice dónde es creíble que esté el parámetro \\(\\theta\\). Podemos resumir de varias maneras. Por ejemplo, si queremos un estimador puntual usamos la media posterior: sims |&gt; group_by(dist) |&gt; summarise(theta_hat = mean(theta) |&gt; round(3)) ## # A tibble: 2 × 2 ## dist theta_hat ## &lt;chr&gt; &lt;dbl&gt; ## 1 inicial 0.503 ## 2 posterior 0.61 Nota que el estimador de máxima verosimilitud es \\(\\hat{p} = 19/30 = 0.63\\), que es ligeramente diferente de la media posterior. ¿Por qué? Y podemos construir intervalos de percentiles, que en esta situación suelen llamarse intervalos de credibilidad, por ejemplo: f &lt;- c(0.025, 0.975) sims |&gt; group_by(dist) |&gt; summarise(cuantiles = quantile(theta, f) |&gt; round(2), f = f) |&gt; pivot_wider(names_from = f, values_from = cuantiles) ## # A tibble: 2 × 3 ## # Groups: dist [2] ## dist `0.025` `0.975` ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 inicial 0.14 0.85 ## 2 posterior 0.45 0.76 El segundo renglón nos da un intervalo posterior para \\(\\theta\\) de credibilidad 95%. En inferencia bayesiana esto sustituye a los intervalos de confianza. El intervalo de la inicial expresa nuestras creencias a priori acerca de \\(\\theta\\). Este intervalo es muy amplio (va de 0.15 a 0.85) El intervalo de la posterior actualiza nuestras creencias acerca de \\(\\theta\\) una vez que observamos los datos, y es considerablemente más angosto y por lo tanto informativo. Puedes experimentar en esta shiny app con diferentes iniciales, número de volados y observación de éxitos. Observaciones: Nótese que escogimos una forma analítica fácil para la inicial, pues resultó así que la posterior es una distribución beta. No siempre es así, y veremos qué hacer cuando nuestra inicial no es de un tipo “conveniente”. Como tenemos la forma analítica de la posterior, es posible hacer los cálculos de la media posterior, por ejemplo, integrando la densidad posterior a mano. Esto generalmente no es factible, y en este ejemplo preferimos hacer una aproximación numérica. En este caso particular es posible usando cálculo, y sabemos que la media de una \\(\\mathsf{\\mathsf{Beta}}(a,b)\\) es \\(a/(a+b)\\), de modo que nuestra media posterior es \\[\\hat{\\mu} = (19 + 2)/(30 + 4) = 21/34 = 0.617 \\] que podemos interpretar como sigue: para calcular la media posterior, a nuestras \\(n\\) pruebas iniciales agregamos 4 pruebas adicionales fijas, con 2 éxitos y 2 fracasos, y calculamos la proporción usual de éxitos. Repite el análisis considerando en general \\(n\\) pruebas, con \\(k\\) éxitos. Utiliza la misma distribución inicial. Lo mismo aplica para el intervalo de 95% (¿cómo se calcularía integrando?). También puedes usar la aproximación de R, por ejemplo: qbeta(0.025, shape1 = 22, shape2 = 14) |&gt; round(2) ## [1] 0.45 qbeta(0.975, shape1 = 22, shape2 = 14) |&gt; round(2) ## [1] 0.76 Ejemplo: observaciones uniformes Ahora regresamos al problema de estimación del máximo de una distribución uniforme. En este caso, consideraremos un problema más concreto. Supongamos que hay una lotería (tipo tradicional) en México y no sabemos cuántos números hay. Obtendremos una muestra iid de \\(n\\) números, y haremos una aproximación continua, suponiendo que \\[X_i \\sim U[0,\\theta]\\] La verosimilitud es entonces \\[P(X_1,\\ldots, X_n|\\theta) = \\theta^{-n},\\] cuando \\(\\theta\\) es mayor que todas las \\(X_i\\), y cero en otro caso. Necesitaremos una inicial \\(P(\\theta)\\). Por la forma que tiene la verosimilitud, podemos intentar una distribución Pareto, que tiene la forma \\[P(\\theta) = \\frac{\\alpha \\theta_0^\\alpha}{\\theta^{\\alpha + 1}}\\] con soporte en \\([\\theta_0,\\infty]\\). Tenemos que escoger entonces el mínimo \\(\\theta_0\\) y el parámetro \\(\\alpha\\). En primer lugar, como sabemos que es una lotería nacional, creemos que no puede haber menos de unos 300 mil números, así que \\(\\theta_0 = 300\\). La función acumulada de la pareto es \\(1- (300/\\theta)^\\alpha\\), así que si \\(\\alpha = 1.1\\) el cuantil 99% es: alpha &lt;- 1.1 (300/(0.01)^(1/alpha)) ## [1] 19738 es decir, alrededor de 20 millones de números. Creemos que es poco probable que el número de boletos sea mayor a esta cota. Nótese ahora que la posterior cumple (multiplicando verosimilitud por inicial): \\[P(\\theta|X_1,\\ldots, X_n |\\theta) \\propto \\theta^{-(n + 2.1)}\\] para \\(\\theta\\) mayor que el máximo de las \\(X_n\\)’s y 300, y cero en otro caso. Esta distribución es pareto con \\(\\theta_0&#39; = \\max\\{300, X_1,\\ldots, X_n\\}\\) y \\(\\alpha&#39; = n + 1.1\\) Una vez planteado nuestro modelo, veamos los datos. Obtuvimos la siguiente muestra de números: loteria_tbl &lt;- read_csv(&quot;data/nums_loteria_avion.csv&quot;, col_names = c(&quot;id&quot;, &quot;numero&quot;)) |&gt; mutate(numero = as.integer(numero)) set.seed(334) muestra_loteria &lt;- sample_n(loteria_tbl, 25) |&gt; mutate(numero = numero/1000) muestra_loteria |&gt; as.data.frame() |&gt; head() ## id numero ## 1 87 348.341 ## 2 5 5851.982 ## 3 40 1891.786 ## 4 51 1815.455 ## 5 14 5732.907 ## 6 48 3158.414 Podemos simular de una Pareto como sigue: rpareto &lt;- function(n, theta_0, alpha){ # usar el método de inverso de distribución acumulada u &lt;- runif(n, 0, 1) theta_0 / (1 - u)^(1/alpha) } Simulamos de la inicial: sims_pareto_inicial &lt;- tibble( theta = rpareto(20000, 300, 1.1 ), dist = &quot;inicial&quot;) Y con los datos de la muestra, simulamos de la posterior: sims_pareto_posterior &lt;- tibble( theta = rpareto(20000, max(c(300, muestra_loteria$numero)), nrow(muestra_loteria) + 1.1), dist = &quot;posterior&quot;) sims_theta &lt;- bind_rows(sims_pareto_inicial, sims_pareto_posterior) ggplot(sims_theta) + geom_histogram(aes(x = theta, fill = dist), bins = 70, alpha = 0.5, position = &quot;identity&quot;, boundary = max(muestra_loteria$numero)) + xlim(0, 15000) + scale_y_sqrt() + geom_rug(data = muestra_loteria, aes(x = numero)) Nótese que cortamos algunos valores de la inicial en la cola derecha: un defecto de esta distribución inicial, con una cola tan larga a la derecha, es que pone cierto peso en valores que son poco creíbles y la vuelve poco apropiada para este problema. Regresaremos más adelante a este problema. Si obtenemos percentiles, obtenemos el intervalo f &lt;- c(0.025, 0.5, 0.975) sims_theta |&gt; group_by(dist) |&gt; summarise(cuantiles = quantile(theta, f) |&gt; round(2), f = f) |&gt; pivot_wider(names_from = f, values_from = cuantiles) ## # A tibble: 2 × 4 ## # Groups: dist [2] ## dist `0.025` `0.5` `0.975` ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 inicial 307. 569. 8449. ## 2 posterior 5858. 6010. 6732. Estimamos entre 5.8 millones y 6.7 millones de boletos. El máximo en la muestra es de max(muestra_loteria$numero) ## [1] 5851.982 Escoger la distribución pareto como inicial es conveniente y nos permitió resolver el problema sin dificultad, pero por su forma vemos que no necesariamente es apropiada para el problema por lo que señalamos arriba. Nos gustaría, por ejemplo, poner una inicial como la siguiente qplot(rgamma(2000, 5, 0.001), geom=&quot;histogram&quot;, bins = 20) + scale_x_continuous(breaks = seq(1000, 15000, by = 2000)) Sin embargo, los cálculos no son tan simples en este caso, pues la posterior no tiene un forma reconocible. Tendremos que usar otras estrategias de simulación para ejemplos como este (Monte Carlo por medio de Cadenas de Markov, que veremos más adelante). Probabilidad a priori La inferencia bayesiana es conceptualmente simple: siempre hay que calcular la posterior a partir de verosimilitud (modelo de datos) y distribución inicial o a priori. Sin embargo, una crítica usual que se hace de la inferencia bayesiana es precisamente que hay que tener esa información inicial, y que distintos analistas llegan a distintos resultados si tienen información inicial distinta. Eso realmente no es un defecto, es una ventaja de la inferencia bayesiana. Los datos y los problemas que queremos resolver no viven en un vacío donde podemos creer que la estatura de las personas, por ejemplo, puede variar de 0 a mil kilómetros, el número de boletos de una lotería puede ir de 2 o 3 boletos o también quizá 500 millones de boletos, o la proporción de personas infectadas de una enfermedad puede ser de unos cuantos hasta miles de millones. En todos estos casos tenemos cierta información inicial que podemos usar para informar nuestras estimaciones. Esta información debe usarse. Antes de tener datos, las probabilidades iniciales deben ser examinadas en términos del conocimiento de expertos. Las probabilidades iniciales son supuestos que hacemos acerca del problema de interés, y también están sujetas a críticas y confrontación con datos. Análisis conjugado Los dos ejemplos que hemos visto arriba son ejemplos de análisis conjugado: (Beta-bernoulli) Si las observaciones \\(X_i\\) son \\(\\mathsf{Bernoulli}(p)\\) (\\(n\\) fija) queremos estimar \\(p\\), y tomamos como distribución inicial para \\(p\\) una \\(\\mathsf{Beta}(a,b)\\), entonces la posterior para \\(p\\) cuando \\(S_n=k\\) es \\(\\mathsf{Beta}(k + a, n - k + b)\\), donde \\(S_n = X_1 + X_2 +\\cdots +X_n\\). Y más en general: (Beta-binomial) Si las observaciones \\(X_i, i=1,2,\\ldots, m\\) son \\(\\mathsf{Binomial}(n_i, p)\\) (\\(n_i\\)’s fijas) independientes, queremos estimar \\(p\\), y tomamos como distribución inicial para \\(p\\) una \\(\\mathsf{Beta}(a,b)\\), entonces la posterior para \\(p\\) cuando \\(S_m=k\\) es \\(\\mathsf{Beta}(k + a, n - k + b)\\), donde \\(S_m = X_1 + X_2 +\\cdots +X_m\\) y \\(n= n_1+n_2+\\cdots+n_m\\) También aplicamos: (Uniforme-Pareto) Si el modelo de datos \\(X_i\\) es uniforme \\(\\mathsf{U}[0,\\theta]\\) (\\(n\\) fija), queremos estimar \\(\\theta\\), y tomamos como distribución inicial para \\(\\theta\\) una Pareto \\((\\theta_0, \\alpha)\\), entonces la posterior para \\(p\\) si el máximo de las \\(X_i\\)’s es igual a \\(M\\) es Pareto con parámetros \\((\\max\\{\\theta_0, M\\}, \\alpha + n)\\). Nótese que en estos casos, dada una forma de la verosimilitud, tenemos una familia conocida de iniciales tales que las posteriores están en la misma familia. Estos modelos son convenientes porque podemos hacer simulaciones de la posterior de manera fácil, o usar sus propiedades teóricas. Otro ejemplo típico es el modelo normal-normal: (Normal-normal) Si \\(X_i\\sim \\mathsf{N}(\\mu,\\sigma)\\), con \\(\\sigma\\) conocida, y tomamos como distribución inicial para \\(\\mu \\sim \\mathsf{N}(\\mu_0,\\sigma_0)\\), y definimos la precisión \\(\\tau\\) como el inverso de la varianza \\(\\sigma^2\\), entonces la posterior de \\(\\mu\\) es Normal con media \\((1-\\lambda) \\mu_0 + \\lambda\\overline{x}\\), y precisión \\(\\tau_0 + n\\tau\\), donde \\(\\lambda = \\frac{n\\tau}{\\tau_0 + n\\tau}\\) Completa cuadrados para mostrar las fórmulas del modelo normal-normal con varianza conocida. Más útil es el siguiente modelo: (Normal-Gamma inverso) Sean \\(X_i\\sim \\mathsf{N}(\\mu, \\sigma)\\). Queremos estimar \\(\\mu\\) y \\(\\sigma\\). Tomamos como distribuciones iniciales (dadas por 4 parámetros: \\(\\mu_0, n_0, \\alpha,\\beta\\)): \\(\\tau = \\frac{1}{\\sigma^2} \\sim \\mathsf{Gamma}(\\alpha,\\beta)\\) \\(\\mu|\\sigma\\) es normal con media \\(\\mu_0\\) y varianza \\(\\sigma^2 / {n_0}\\) , y \\(p(\\mu, \\sigma) = p(\\mu|\\sigma)p(\\sigma)\\) Entonces la posterior es: \\(\\tau|x\\) es \\(\\mathsf{Gamma}(\\alpha&#39;, \\beta&#39;)\\), con \\(\\alpha&#39; = \\alpha + n/2\\), \\(\\beta&#39; = \\beta + \\frac{1}{2}\\sum_{i=1}^{n}(x_{i} - \\bar{x})^2 + \\frac{nn_0}{n+n_0}\\frac{({\\bar{x}}-\\mu_{0})^2}{2}\\) \\(\\mu|\\sigma,x\\) es normal con media \\(\\mu&#39; = \\frac{n_0\\mu_{0}+n{\\bar{x}}}{n_0 +n}\\) y varianza \\(\\sigma^2/({n_0 +n})\\). \\(p(\\mu,\\sigma|x) = p(\\mu|x,\\sigma)p(\\sigma|x)\\) Observaciones Nótese que este último ejemplo tiene más de un parámetro. En estos casos, el objeto de interés es la posterior conjunta de los parámetros \\(p(\\theta_1,\\theta_2,\\cdots, \\theta_p|x)\\). Este último ejemplo es relativamente simple pues por la selección de iniciales, para simular de la conjunta de \\(\\mu\\) y \\(\\tau\\) podemos simular primero \\(\\tau\\) (o \\(\\sigma\\)), y después usar este valor para simular de \\(\\mu\\): el par de valores resultantes son una simulación de la conjunta. Los parámetros \\(\\alpha,\\beta\\) para la inicial de \\(\\tau\\) pueden interpretarse como sigue: \\(\\sqrt{\\beta/\\alpha}\\) es un valor “típico” a priori para la varianza poblacional, y \\(a\\) indica qué tan seguros estamos de este valor típico. Nótese que para que funcionen las fórmulas de la manera más simple, escogimos una dependencia a priori entre la media y la precisión: \\(\\tau = \\sigma^{-2}\\) indica la escala de variabilidad que hay en la población, la incial de la media tiene varianza \\(\\sigma^2/n_0\\). Si la escala de variabilidad de la población es más grande, tenemos más incertidumbre acerca de la localización de la media. Aunque esto tiene sentido en algunas aplicaciones, y por convenviencia usamos esta familia conjugada, muchas veces es preferible otro tipo de especificaciones para las iniciales: por ejemplo, la media normal y la desviación estándar uniforme, o media normal, con iniciales independientes. Sin embargo, estos casos no son tratables con análisis conjugado (veremos más adelante cómo tratarlos con MCMC). Ejemplo Supongamos que queremos estimar la estatura de los cantantes de tesitura tenor con una muestra iid de tenores de Estados Unidos. Usaremos el modelo normal de forma que \\(X_i\\sim \\mathsf{N}(\\mu, \\sigma^2)\\). Una vez decidido el modelo, tenemos que poner distribución inicial para los parámetros \\((\\mu, \\sigma^2)\\). Comenzamos con \\(\\sigma^2\\). Como está el modelo, esta inicial debe estar dada para la precisión \\(\\tau\\), pero podemos simular para ver cómo se ve nuestra inicial para la desviación estándar. En la población general la desviación estándar es alrededor de 7 centímetros # Comenzamos seleccionando un valor que creemos típico para la desviación estándar sigma_0 &lt;- 7 # seleccionamos un valor para a, por ejemplo: si es más chico sigma tendrá más # disperisón a &lt;- 3 # ponemos 7 = sqrt(b/a) -&gt; b = a * 64 b &lt;- a * sigma_0 ^ 2 c(a = a, b = b) ## a b ## 3 147 Ahora simulamos para calcular cuantiles tau &lt;- rgamma(1000, a, b) quantile(tau, c(0.05, 0.95)) ## 5% 95% ## 0.005781607 0.042170161 sigma &lt;- 1 / sqrt(tau) mean(sigma) ## [1] 8.002706 quantile(sigma, c(0.05, 0.95)) ## 5% 95% ## 4.869653 13.151520 Que es dispersión considerable: con poca probabilidad la desviación estándar es menor a 4 centímetros, y también creemos que es poco creíble la desviación estándar sea de más de 13 centímetros. Comenzamos con \\(\\mu\\). Sabemos, por ejemplo, que con alta probabilidad la media debe ser algún número entre 1.60 y 1.80. Podemos investigar: la media nacional en estados unidos está alrededor de 1.75, y el percentil 90% es 1.82. Esto es variabilidad en la población: debe ser muy poco probable, por ejemplo, que la media de tenores sea 1.82 Quizá los cantantes tienden a ser un poco más altos o bajos que la población general, así que podríamos agregar algo de dispersión. Podemos establecer parámetros y simular de la marginal a partir de las fórmulas de arriba para entender cómo se ve la inicial de \\(\\mu\\): mu_0 &lt;- 175 # valor medio de inicial n_0 &lt;- 5 # cuánta concentración en la inicial tau &lt;- rgamma(1000, a,b) sigma &lt;- 1/sqrt(tau) mu &lt;- map_dbl(sigma, ~ rnorm(1, mu_0, .x / sqrt(n_0))) quantile(mu, c(0.05, 0.5, 0.95)) ## 5% 50% 95% ## 168.7275 174.8412 180.7905 Que consideramos un rango en el que con alta probabilidad debe estar la media poblacional de los cantantes. Podemos checar nuestros supuestos simulando posibles muestras usando sólo nuestra información previa: simular_normal_invgamma &lt;- function(n, pars){ mu_0 &lt;- pars[1] n_0 &lt;- pars[2] a &lt;- pars[3] b &lt;- pars[4] # simular media tau &lt;- rgamma(1, a, b) sigma &lt;- 1 / sqrt(tau) mu &lt;- rnorm(1, mu_0, sigma/sqrt(n_0)) # simular sigma rnorm(n, mu, sigma) } set.seed(3461) sims_tbl &lt;- tibble(rep = 1:20) |&gt; mutate(estatura = map(rep, ~ simular_normal_invgamma(500, c(mu_0, n_0, a, b)))) |&gt; unnest(cols = c(estatura)) ggplot(sims_tbl, aes(x = estatura)) + geom_histogram() + facet_wrap(~ rep) + geom_vline(xintercept = c(150, 180), colour = &quot;red&quot;) Pusimos líneas de referencia en 150 y 180. Vemos que nuestras iniciales no producen simulaciones totalmente fuera del contexto, y parecen cubrir apropiadamente el espacio de posiblidades para estaturas de los tenores. Quizá hay algunas realizaciones poco creíbles, pero no extremadamente. En este punto, podemos regresar y ajustar la inicial para \\(\\sigma\\), que parece tomar valores demasiado grandes (produciendo por ejemplo una simulación con estatura de 220 y 140, que deberían ser menos probables). Ahora podemos usar los datos para calcular nuestras posteriores. set.seed(3413) cantantes &lt;- lattice::singer |&gt; mutate(estatura_cm = round(2.54 * height)) |&gt; filter(str_detect(voice.part, &quot;Tenor&quot;)) |&gt; sample_n(20) cantantes ## height voice.part estatura_cm ## 139 70 Tenor 1 178 ## 150 68 Tenor 2 173 ## 140 65 Tenor 1 165 ## 132 66 Tenor 1 168 ## 152 69 Tenor 2 175 ## 141 72 Tenor 1 183 ## 161 71 Tenor 2 180 ## 156 71 Tenor 2 180 ## 158 71 Tenor 2 180 ## 164 69 Tenor 2 175 ## 147 68 Tenor 1 173 ## 130 72 Tenor 1 183 ## 162 71 Tenor 2 180 ## 134 74 Tenor 1 188 ## 170 69 Tenor 2 175 ## 167 68 Tenor 2 173 ## 149 64 Tenor 1 163 ## 143 68 Tenor 1 173 ## 157 69 Tenor 2 175 ## 153 71 Tenor 2 180 Los cálculos son un poco tediosos, pero podemos construir una función apropiada: calcular_pars_posterior &lt;- function(x, pars_inicial){ # iniciales mu_0 &lt;- pars_inicial[1] n_0 &lt;- pars_inicial[2] a_0 &lt;- pars_inicial[3] b_0 &lt;- pars_inicial[4] # muestra n &lt;- length(x) media &lt;- mean(x) S2 &lt;- sum((x - media)^2) # sigma post a_1 &lt;- a_0 + 0.5 * n b_1 &lt;- b_0 + 0.5 * S2 + 0.5 * (n * n_0) / (n + n_0) * (media - mu_0)^2 # posterior mu mu_1 &lt;- (n_0 * mu_0 + n * media) / (n + n_0) n_1 &lt;- n + n_0 c(mu_1, n_1, a_1, b_1) } pars_posterior &lt;- calcular_pars_posterior(cantantes$estatura_cm, c(mu_0, n_0, a, b)) pars_posterior ## [1] 175.8 25.0 13.0 509.0 ¿Cómo se ve nuestra posterior comparada con la inicial? Podemos hacer simulaciones: sim_params &lt;- function(m, pars){ mu_0 &lt;- pars[1] n_0 &lt;- pars[2] a &lt;- pars[3] b &lt;- pars[4] # simular sigmas sims &lt;- tibble(tau = rgamma(m, a, b)) |&gt; mutate(sigma = 1 / sqrt(tau)) # simular mu sims &lt;- sims |&gt; mutate(mu = rnorm(m, mu_0, sigma / sqrt(n_0))) sims } sims_inicial &lt;- sim_params(5000, c(mu_0, n_0, a, b)) |&gt; mutate(dist = &quot;inicial&quot;) sims_posterior &lt;- sim_params(5000, pars_posterior) |&gt; mutate(dist = &quot;posterior&quot;) sims &lt;- bind_rows(sims_inicial, sims_posterior) ggplot(sims, aes(x = mu, y = sigma, colour = dist)) + geom_point(alpha = 0.4) Y vemos que nuestra posterior es consistente con la información inicial que usamos, hemos aprendido considerablemente de la muestra. La posterior se ve como sigue. Hemos marcado también las medias posteriores de cada parámetro: media y desviación estándar. medias_post &lt;- sims |&gt; filter(dist == &quot;posterior&quot;) |&gt; select(-dist) |&gt; summarise(across(everything(), mean)) ggplot(sims |&gt; filter(dist == &quot;posterior&quot;), aes(x = mu, y = sigma)) + geom_point(colour = &quot;#00BFC4&quot;) + geom_point(data = medias_post, size = 5, colour = &quot;black&quot;) + coord_equal() Podemos construir intervalos creíbles del 90% para estos dos parámetros, por ejemplo haciendo intervalos de percentiles: f &lt;- c(0.05, 0.5, 0.95) sims |&gt; pivot_longer(cols = mu:sigma, names_to = &quot;parametro&quot;) |&gt; group_by(dist, parametro) |&gt; reframe(cuantil = quantile(value, f) |&gt; round(1), f = f) |&gt; pivot_wider(names_from = f, values_from = cuantil) ## # A tibble: 4 × 5 ## dist parametro `0.05` `0.5` `0.95` ## &lt;chr&gt; &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 inicial mu 169. 175. 181. ## 2 inicial sigma 4.8 7.4 13.3 ## 3 posterior mu 174. 176. 178. ## 4 posterior sigma 5.1 6.3 8.2 Como comparación, los estimadores de máxima verosimlitud son media_mv &lt;- mean(cantantes$estatura_cm) sigma_mv &lt;- mean((cantantes$estatura_cm - media_mv)^2) |&gt; sqrt() c(media_mv, sigma_mv) ## [1] 176 6 Ahora solo resta checar que el modelo es razonable. Veremos más adelante cómo hacer esto, usando la distribución predictiva posterior. Pasos de un análisis de datos bayesiano Como vimos en los ejemplos, en general un análisis de datos bayesiano sigue los siguientes pasos: Identificar los datos releventes a nuestra pregunta de investigación, el tipo de datos que vamos a describir, que variables queremos estimar. Definir el modelo descriptivo para los datos. La forma matemática y los parámetros deben ser apropiados para los objetivos del análisis. Especificar la distribución inicial de los parámetros. Utilizar inferencia bayesiana para reubicar la credibilidad a lo largo de los posibles valores de los parámetros. Verificar que la distribución posterior replique los datos de manera razonable, de no ser el caso considerar otros modelos descriptivos para los datos. Elicitando probablidades subjetivas (opcional) No siempre es fácil elicitar probabilidades subjetivas de manera que capturemos el verdadero conocimiento de dominio que tenemos. Una manera clásica de hacerlo es con apuestas Considera una pregunta sencilla que puede afectar a un viajero: ¿Qué tanto crees que habrá una tormenta que ocasionará el cierre de la autopista México-Acapulco en el puente del \\(20\\) de noviembre? Como respuesta debes dar un número entre \\(0\\) y \\(1\\) que refleje tus creencias. Una manera de seleccionar dicho número es calibrar las creencias en relación a otros eventos cuyas probabilidades son claras. Como evento de comparación considera una experimento donde hay canicas en una urna: \\(5\\) rojas y \\(5\\) blancas. Seleccionamos una canica al azar. Usaremos esta urna como comparación para considerar la tormenta en la autopista. Ahora, considera el siguiente par de apuestas de las cuales puedes elegir una: A. Obtienes \\(\\$1000\\) si hay una tormenta que ocasiona el cierre de la autopista el próximo \\(20\\) de noviembre. B. Obtienes \\(\\$1000\\) si seleccionas una canica roja de la urna que contiene \\(5\\) canicas rojas y \\(5\\) blancas. Si prefieres la apuesta B, quiere decir que consideras que la probabilidad de tormenta es menor a \\(0.5\\), por lo que al menos sabes que tu creencia subjetiva de una la probabilidad de tormenta es menor a \\(0.5\\). Podemos continuar con el proceso para tener una mejor estimación de la creencia subjetiva. A. Obtienes \\(\\$1000\\) si hay una tormenta que ocasiona el cierre de la autopista el próximo \\(20\\) de noviembre. C. Obtienes \\(\\$1000\\) si seleccionas una canica roja de la urna que contiene \\(1\\) canica roja y \\(9\\) blancas. Si ahora seleccionas la apuesta \\(A\\), esto querría decir que consideras que la probabilidad de que ocurra una tormenta es mayor a \\(0.10\\). Si consideramos ambas comparaciones tenemos que tu probabilidad subjetiva se ubica entre \\(0.1\\) y \\(0.5\\). Verificación predictiva posterior Una vez que ajustamos un modelo bayesiano, podemos simular nuevas observaciones a partir del modelo. Esto tiene dos utilidades: Hacer predicciones acerca de datos no observados. Confirmar que nuevas observaciones, producidas simulando con el modelo son similares a las que de hecho observamos. Esto nos permite confirmar la calidad del ajuste del modelo, y se llama verificación predictiva posterior. Supongamos que tenemos la posterior \\(p(\\theta | x)\\). Podemos generar una nueva replicación de los datos como sigue: La distribución predictiva posterior genera nuevas observaciones a partir de la información observada. La denotamos como \\(p(\\tilde{x}|x)\\). Para simular de ella: Muestreamos un valor \\(\\tilde{\\theta}\\) de la posterior \\(p(\\theta|x)\\). Simulamos del modelo de las observaciones \\(\\tilde{x} \\sim p(\\tilde{x}|\\tilde{\\theta})\\). Repetimos el proceso hasta obtener una muestra grande. Usamos este método para producir, por ejemplo, intervalos de predicción para nuevos datos. Si queremos una replicación de las observaciones de la predictiva posterior, Muestreamos un valor \\(\\tilde{\\theta}\\) de la posterior \\(p(\\theta|x)\\). Simulamos del modelo de las observaciones \\(\\tilde{x}_1, \\tilde{x}_2,\\ldots, \\tilde{x}_n \\sim p(\\tilde{x}|\\tilde{\\theta})\\), done \\(n\\) es el tamaño de muestra de la muestra original \\(x\\). Usamos este método para producir conjuntos de datos simulados que comparamos con los observados para verificar nuestro modelo. Ejemplo: estaturas de tenores En este ejemplo, usaremos la posterior predictiva para checar nuestro modelo. Vamos a crear varias muestras, del mismo tamaño que la original, según nuestra predictiva posterior, y compararemos estas muestras con la observada. Y ahora simulamos otra muestra muestra_sim &lt;- simular_normal_invgamma(20, pars_posterior) muestra_sim |&gt; round(0) ## [1] 167 181 184 181 167 167 172 170 177 172 169 174 182 184 176 171 175 176 168 ## [20] 181 Podemos simular varias muestras y hacer una prueba de lineup: library(nullabor) set.seed(9921) sims_obs &lt;- tibble(.n = 1:19) |&gt; mutate(estatura_cm = map(.n, ~ simular_normal_invgamma(20, pars_posterior))) |&gt; unnest(estatura_cm) pos &lt;- sample(1:20, 1) lineup_tbl &lt;- lineup(true = cantantes |&gt; select(estatura_cm), samples = sims_obs, pos = pos) ggplot(lineup_tbl, aes(x = estatura_cm)) + geom_histogram(binwidth = 2.5) + facet_wrap(~.sample) Con este tipo de gráficas podemos checar desajustes potenciales de nuestro modelo. ¿Puedes encontrar los datos verdaderos? ¿Cuántos seleccionaron los datos correctos? Ejemplo: modelo Poisson Supongamos que pensamos el modelo para las observaciones es Poisson con parámetro \\(\\lambda\\). Pondremos como inicial para \\(\\lambda\\) una exponencial con media 10. Nótese que la posterior está dada por \\[p(\\lambda|x_1,\\ldots, x_n) \\propto e^{-n\\lambda}\\lambda^{\\sum_i x_i} e^{-0.1\\lambda} = \\lambda^{n\\overline{x}}e^{-\\lambda(n + 0.1)}\\] que es una distribución gamma con parámetros \\((n\\overline{x} + 1, n+0.1)\\) Ahora supongamos que observamos la siguiente muestra, ajustamos nuestro modelo y hacemos replicaciones posteriores de los datos observados: x &lt;- rnbinom(250, mu = 20, size = 3) crear_sim_rep &lt;- function(x){ n &lt;- length(x) suma &lt;- sum(x) sim_rep &lt;- function(rep){ lambda &lt;- rgamma(1, sum(x) + 1, n + 0.1) x_rep &lt;- rpois(n, lambda) tibble(rep = rep, x_rep = x_rep) } } sim_rep &lt;- crear_sim_rep(x) lineup_tbl &lt;- map(1:5, ~ sim_rep(.x)) |&gt; bind_rows() |&gt; bind_rows(tibble(rep = 6, x_rep = x)) ggplot(lineup_tbl, aes(x = x_rep)) + geom_histogram(bins = 15) + facet_wrap(~rep) Y vemos claramente que nuestro modelo no explica apropiadamente la variación de los datos observados. Contrasta con: set.seed(223) x &lt;- rpois(250, 15) crear_sim_rep &lt;- function(x){ n &lt;- length(x) suma &lt;- sum(x) sim_rep &lt;- function(rep){ lambda &lt;- rgamma(1, sum(x) + 1, n + 0.1) x_rep &lt;- rpois(n, lambda) tibble(rep = rep, x_rep = x_rep) } } sim_rep &lt;- crear_sim_rep(x) lineup_tbl &lt;- map(1:5, ~ sim_rep(.x)) |&gt; bind_rows() |&gt; bind_rows(tibble(rep = 6, x_rep = x)) ggplot(lineup_tbl, aes(x = x_rep)) + geom_histogram(bins = 15) + facet_wrap(~rep) Y verificamos que en este caso el ajuste del modelo es apropiado. Predicción Cuando queremos hacer predicciones particulares acerca de datos que observemos en el futuro, también podemos usar la posterior predictiva. En este caso, tenemos que considerar La variabilidad que produce la incertidumbre en la estimación de los parámetros La variabilidad de las observaciones dados los parámetros. Es decir, tenemos que simular sobre todos las combinaciones factibles de los parámetros. Ejemplo: cantantes Si un nuevo tenor llega a un coro, ¿cómo hacemos una predicción de su estatura? Como siempre, quisiéramos obtener un intervalo que exprese nuestra incertidumbre acerca del valor que vamos a observar. Entonces haríamos: sims_posterior &lt;- sim_params(50000, pars_posterior) |&gt; mutate(y_pred = rnorm(n(), mu, sigma)) sims_posterior |&gt; head() ## # A tibble: 6 × 4 ## tau sigma mu y_pred ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 0.0286 5.91 175. 181. ## 2 0.0200 7.07 177. 178. ## 3 0.0257 6.23 176. 170. ## 4 0.0344 5.39 176. 174. ## 5 0.0297 5.80 175. 169. ## 6 0.0282 5.96 177. 170. f &lt;- c(0.025, 0.5, 0.975) sims_posterior |&gt; summarise(f = f, y_pred = quantile(y_pred, f)) ## # A tibble: 3 × 2 ## f y_pred ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.025 163. ## 2 0.5 176. ## 3 0.975 189. Y con esto obtenemos el intervalo (163, 189), al 95%, para una nueva observación. Nótese que este intervalo no puede construirse con una simulación particular de la posterior de los parámetros, pues sería demasiado corto. Es posible demostrar que en este caso, la posterior predictiva tiene una forma conocida: La posterior predictiva para el modelo normal-gamma inverso es una distribución \\(t\\) con \\(2\\alpha&#39;\\) grados de libertad, centrada en \\(\\mu&#39;\\), y con escala \\(s^2 = \\frac{\\beta&#39;}{\\alpha&#39;}\\frac{n + n_0 + 1}{n +n_0}\\) mu_post &lt;- pars_posterior[1] n_post &lt;- pars_posterior[2] alpha_post &lt;- pars_posterior[3] beta_post &lt;- pars_posterior[4] s &lt;- sqrt(beta_post/alpha_post) * sqrt((n_post + 1)/n_post) qt(c(0.025, 0.5, 0.975), 2 * alpha_post) * s + mu_post ## [1] 162.6832 175.8000 188.9168 Calcula la posterior predictiva del modelo Beta-Bernoulli y Beta-Binomial. (Más difícil) Calcula la posterior predictiva del modelo Poisson-Gamma. Ejemplo: posterior predictiva de Pareto-Uniforme. La posterior predictiva del modelo Pareto-Uniforme no tiene un nombre estándar, pero podemos aproximarla usando simulación. Usando los mismos datos del ejercicio de la lotería, haríamos: rpareto &lt;- function(n, theta_0, alpha){ # usar el método de inverso de distribución acumulada u &lt;- runif(n, 0, 1) theta_0 / (1 - u)^(1/alpha) } # Simulamos de la posterior de los parámetros lim_inf_post &lt;- max(c(300, muestra_loteria$numero)) k_posterior &lt;- nrow(muestra_loteria) + 1.1 sims_pareto_posterior &lt;- tibble( theta = rpareto(100000, lim_inf_post, k_posterior)) # Simulamos una observación para cada una de las anteriores: sims_post_pred &lt;- sims_pareto_posterior |&gt; mutate(x_pred = map_dbl(theta, ~ runif(1, 0, .x))) # Graficamos ggplot(sims_post_pred, aes(x = x_pred)) + geom_histogram(binwidth = 50) + geom_vline(xintercept = lim_inf_post, colour = &quot;red&quot;) Que es una mezcla de una uniforme con una Pareto. Referencias "],["calibración-bayesiana-y-regularización.html", "Sección 11 Calibración bayesiana y Regularización Enfoque bayesiano y frecuentista Ejemplo: estimación de una proporción Intervalos de Agresti-Coull Incorporando información inicial Inferencia bayesiana y regularización Ejemplo: modelo normal y estaturas Ejemplo: estimación de proporciones Teoría de decisión Riesgo de Bayes", " Sección 11 Calibración bayesiana y Regularización El enfoque bayesiano se puede formalizar coherentemente en términos de probabilidades subjetivas, y como vimos, esta es una fortaleza del enfoque bayesiano. En la práctica, sin embargo, muchas veces puede ser difícil argumentar en términos exclusivos de probabilidad subjetiva, aunque hagamos los esfuerzos apropiados para incorporar la totalidad de información que distintos actores involucrados pueden tener. Consideremos, por ejemplo, que INEGI produjera un intervalo creíble del 95% para el ingreso mediano de los hogares de México. Aún cuando nuestra metodología sea transparente y correctamente informada, algunos investigadores interesados puede ser que tengan recelo en usar esta información, y quizá preferirían hacer estimaciones propias. Esto restaría valor al trabajo cuidadoso que pusimos en nuestras estimaciones oficiales. Por otra parte, el enfoque frecuentista provee de ciertas garantías mínimas para la utilización de las estimaciones, que no dependen de la interpretación subjetiva de la probabilidad, sino de las propiedades del muestreo. Consideremos la cobertura de los intervalos de confianza: Bajo ciertos supuestos de nuestros modelos, la probabilidad de que un intervalo de confianza del 95% cubra al verdadero valor poblacional es del 95%. Esta probabilidad es sobre las distintas muestras que se pueden obtener según el diseño del muestreo. Los intervalos creíbles en principio no tienen por qué cumplir esta propiedad, pero consideramos que en la práctica es una garantía mínima que deberían cumplir. El enfoque resultante se llama bayesiano calibrado, Little (2011) . La idea es seguir el enfoque bayesiano usual para construir nuestras estimaciones, pero verificar hasta donde sea posible que los intervalos resultantes satisfacen alguna garantía frecuentista básica. Observación. checar que la cobertura real es similar a la nominal es importante en los dos enfoques: frecuentista y bayesiano. Los intervalos frecuentistas, como hemos visto, generalmente son aproximados, y por lo tanto no cumplen automáticamente esta propiedad de calibración. Enfoque bayesiano y frecuentista Los métodos estadísticos clásicos toman el punto de vista frecuentista y se basa en los siguientes puntos (Wasserman (2013)): La probabilidad se interpreta como un límite de frecuencias relativas, donde las probabilidades son propiedades objetivas en el mundo real. En un modelo, los parámetros son constantes fijas (desconocidas). Como consecuencia, no se pueden realizar afirmaciones probabilísticas útiles en relación a éstos. Los procedimientos estadísticos deben diseñarse con el objetivo de tener propiedades frecuentistas bien definidas. Por ejemplo, un intervalo de confianza del \\(95\\%\\) debe contener el verdadero valor del parámetro con frecuencia límite de al menos el \\(95\\%\\). En contraste, el acercamiento Bayesiano muchas veces se describe por los siguientes postulados: La probabilidad describe grados de creencia, no frecuencias limite. Como tal uno puede hacer afirmaciones probabilísticas acerca de muchas cosas y no solo datos sujetos a variabilidad aleatoria. Por ejemplo, puedo decir: “La probabilidad de que Einstein tomara una taza de té el primero de agosto de \\(1948\\)” es \\(0.35\\), esto no hace referencia a ninguna frecuencia relativa sino que refleja la certeza que yo tengo de que la proposición sea verdadera. Podemos hacer afirmaciones probabilísticas de parámetros. Podemos hacer inferencia de un parámetro \\(\\theta\\) por medio de distribuciones de probabilidad. Las inferencias como estimaciones puntuales y estimaciones de intervalos se pueden extraer de dicha distribución. Finalmente, en el enfoque bayesiano calibrado (Little (2011)): Usamos el enfoque bayesiano para modelar y hacer afirmaciones probabilísticas de los parámetros. Buscamos cumplir las garantías frecuentistas del inciso 3). Ejemplo: estimación de una proporción Recordamos nuestro problema de estimación de una proporcion \\(\\theta\\). Usando la distribución inicial \\(p(\\theta)\\sim \\mathsf{Beta}(2,2)\\), y la verosimilitud estándar binomial, vimos que la posterior cuando observamos \\(k\\) éxitos es \\[p(\\theta|k) \\sim \\mathsf{Beta}(k + 2, n - k + 2)\\]. La media posterior es \\[\\frac{k + 2}{n + 4} \\] que podemos interpretar como: agrega 2 éxitos y 2 fracasos a los datos observados y calcula la proporción de éxitos. Un intervalo posterior de credibilidad del 95% se calcula encontrando los cuantiles 0.025 y 0.975 de una \\(\\mathsf{Beta}(k + 2, n - k + 2)\\) \\[I_a = \\left [q_{0.025}(k+2, n+4), q_{0.975}(k+2, n+4)\\right ]\\] Que compararemos con el intervalo usual de Wald: si \\(\\hat{\\theta} = \\frac{k}{n}\\), entonces \\[I_w = \\left [\\hat{\\theta} - 2 \\sqrt{\\frac{\\hat{\\theta}(1-\\hat{\\theta})}{n}}, \\hat{\\theta} + 2 \\sqrt{\\frac{\\hat{\\theta}(1-\\hat{\\theta})}{n}}\\right]\\] ¿Cómo podemos comparar la calibración de estos dos intervalos? Nominalmente, deben tener cobertura de 95%. Hagamos un ejercicio de simulación para distintos tamaños de muestra \\(n\\) y posibles valores \\(\\theta\\in (0,1)\\): set.seed(332) simular_muestras &lt;- function(M, n, p){ k = rbinom(M, n, p) tibble(rep = 1:M, n = n, p = p, k = k) } intervalo_wald &lt;- function(n, k){ p_hat &lt;- k / n ee_hat &lt;- sqrt(p_hat * (1 - p_hat) / n) tibble(inf = p_hat - 2 * ee_hat, sup = p_hat + 2 * ee_hat) } intervalo_bayes &lt;- function(n, k, a = 2, b = 2){ a &lt;- k + a b &lt;- n - k + b tibble(inf = qbeta(0.025, a, b), sup = qbeta(0.975, a, b)) } set.seed(812) ejemplo &lt;- simular_muestras(5, 20, 0.4) ejemplo |&gt; mutate(intervalo = intervalo_wald(n, k)) |&gt; pull(intervalo) |&gt; bind_cols(ejemplo) |&gt; select(-rep) ## # A tibble: 5 × 5 ## inf sup n p k ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; ## 1 0.0211 0.379 20 0.4 4 ## 2 0.228 0.672 20 0.4 9 ## 3 0.276 0.724 20 0.4 10 ## 4 0.228 0.672 20 0.4 9 ## 5 0.137 0.563 20 0.4 7 ejemplo |&gt; mutate(intervalo = intervalo_bayes(n, k)) |&gt; pull(intervalo) |&gt; bind_cols(ejemplo) |&gt; select(-rep) ## # A tibble: 5 × 5 ## inf sup n p k ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; ## 1 0.102 0.437 20 0.4 4 ## 2 0.268 0.655 20 0.4 9 ## 3 0.306 0.694 20 0.4 10 ## 4 0.268 0.655 20 0.4 9 ## 5 0.197 0.573 20 0.4 7 ¿Cuáles de estos intervalos cubren al verdadero valor? Nótese que no podemos descalificar a ningún método por no cubrir una vez. Es fácil producir un intervalo con 100% de cobertura: (0,1). Pero no nos informa dónde es probable que esté el parámetro. Sin embargo, podemos checar la cobertura frecuentista haciendo una cantidad grande de simulaciones: parametros &lt;- crossing(n = c(5, 10, 30, 60, 100, 400), p = c(0.01, 0.015, 0.02, 0.025, 0.03, 0.035, 0.04, 0.05, 0.07, 0.1, 0.15)) set.seed(2343) # simulaciones simulaciones &lt;- parametros |&gt; mutate(muestra = map2(n, p, ~ simular_muestras(50000, .x, .y) |&gt; select(rep, k))) |&gt; unnest(muestra) # calcular_cobertura calcular_cobertura &lt;- function(simulaciones, construir_intervalo){ # nombre de función intervalo_nombre &lt;- substitute(construir_intervalo) |&gt; as.character() cobertura_tbl &lt;- simulaciones |&gt; mutate(intervalo = construir_intervalo(n, k)) |&gt; pull(intervalo) |&gt; bind_cols(simulaciones) |&gt; mutate(cubre = p &gt;= inf &amp; p &lt;= sup) |&gt; group_by(n, p) |&gt; summarise(cobertura = mean(cubre), long_media = mean(sup - inf)) cobertura_tbl |&gt; mutate(tipo = intervalo_nombre) } cobertura_wald &lt;- calcular_cobertura(simulaciones, intervalo_wald) cobertura_wald ## # A tibble: 66 × 5 ## # Groups: n [6] ## n p cobertura long_media tipo ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; ## 1 5 0.01 0.0483 0.0347 intervalo_wald ## 2 5 0.015 0.0733 0.0527 intervalo_wald ## 3 5 0.02 0.0954 0.0689 intervalo_wald ## 4 5 0.025 0.119 0.0862 intervalo_wald ## 5 5 0.03 0.140 0.102 intervalo_wald ## 6 5 0.035 0.165 0.120 intervalo_wald ## 7 5 0.04 0.187 0.137 intervalo_wald ## 8 5 0.05 0.227 0.167 intervalo_wald ## 9 5 0.07 0.299 0.223 intervalo_wald ## 10 5 0.1 0.398 0.303 intervalo_wald ## # ℹ 56 more rows graficar_cobertura &lt;- function(cobertura_tbl){ ggplot(cobertura_tbl, aes(x = p, y = cobertura, colour = tipo)) + geom_hline(yintercept = 0.95, colour = &quot;black&quot;) + geom_line() + geom_point() + facet_wrap(~n) + ylim(0, 1) } cobertura_wald |&gt; graficar_cobertura() La cobertura real es mucho más baja que la nominal en muchos casos, especialmente cuando la \\(p\\) es baja y \\(n\\) es chica. Pero incluso para muestras relativamente grandes (100), la cobertura es mala si \\(p\\) es chica. Ahora probamos nuestro método alternativo: cobertura_bayes &lt;- calcular_cobertura(simulaciones, intervalo_bayes) bind_rows(cobertura_wald, cobertura_bayes) |&gt; mutate(tipo = factor(tipo, levels = c(&#39;intervalo_wald&#39;, &#39;intervalo_bayes&#39;))) |&gt; graficar_cobertura() Y vemos que en general el intervalo de Bayes es superior al de Wald, en sentido de que su cobertura real es más cercana a la nominal. El caso donde fallan los dos es para muestras muy chicas \\(n=5, 10\\), con probabilidades de éxito chicas \\(p\\leq 0.02\\). Sin embargo, si tenemos información previa acerca del tamaño de la proporción que estamos estimando, es posible obtener buena calibración con el método bayesiano. En este caso particular, tenemos argumentos frecuentistas para utilizar el método bayesiano. Por ejemplo, si el INEGI utilizara estos intervalos creíbles, un análisis de calibración de este tipo sostendría esa decisión. Intervalos de Agresti-Coull Un método intermedio que se usa para obtener mejores intervalos cuando estimamos proporciones es el siguiente: Agregar dos 1’s y dos 0’s a los datos. Utilizar el método de Wald con estos datos modificados. intervalo_agresti_coull &lt;- function(n, k){ p_hat &lt;- (k + 2)/ (n + 4) ee_hat &lt;- sqrt(p_hat * (1 - p_hat) / n) tibble(inf = p_hat - 2 * ee_hat, sup = p_hat + 2 * ee_hat) } cobertura_ac &lt;- calcular_cobertura(simulaciones, intervalo_agresti_coull) bind_rows(cobertura_wald, cobertura_bayes, cobertura_ac) |&gt; mutate(tipo = factor(tipo, levels = c(&#39;intervalo_wald&#39;, &#39;intervalo_bayes&#39;, &#39;intervalo_agresti_coull&#39;))) |&gt; graficar_cobertura() Que tiende a ser demasiado conservador para proporciones chicas: graficar_cobertura(cobertura_ac) + ylim(c(0.9, 1)) Conclusión 1: Los intervalos de Agresti-Coull son una buena alternativa para estimar proporciones como sustituto de los intervalos clásicos de Wald, aunque tienden a ser muy conservadores para muestras chicas Idealmente podemos utilizar un método bayesiano pues normalmente tenemos información inicial acerca de las proporciones que queremos estimar. Incorporando información inicial Nótese que generalmente tenemos información acerca de la cantidad que queremos estimar: por ejemplo, que proporción de visitantes de un sitio web compra algo (usualmente muy baja, menos de 2%), qué proporción de personas tiene diabetes tipo 1 (una proporción muy baja, menos de 1 por millar), o qué proporción de hogares tienen ingresos trimestrales mayores a 150 mil pesos (menos de %5 con alta probabilidad). En este caso, tenemos que ajustar nuestra inicial. Por ejemplo, para el problema de ingresos, podríamos usar una \\(\\mathsf{Beta}(2, 100)\\), cuyos cuantiles son: # uno de cada 100 a &lt;- 2 b &lt;- 100 beta_sims &lt;- rbeta(5000, a, b) quantile(beta_sims, c(0.01, 0.05, 0.50, 0.90, 0.99)) |&gt; round(3) ## 1% 5% 50% 90% 99% ## 0.001 0.004 0.016 0.039 0.067 qplot(beta_sims) Veamos cómo se ven los intervalos bayesianos producidos con esta inicial: crear_intervalo_bayes &lt;- function(a, b){ intervalo_fun &lt;- function(n, k){ a_post &lt;- k + a b_post &lt;- n - k + b tibble(inf = qbeta(0.025, a_post, b_post), sup = qbeta(0.975, a_post, b_post)) } intervalo_fun } intervalo_bayes_2 &lt;- crear_intervalo_bayes(a, b) cobertura_bayes &lt;- calcular_cobertura(simulaciones, intervalo_bayes_2) graficar_cobertura(bind_rows(cobertura_bayes, cobertura_ac) |&gt; filter(p &lt; 0.05)) + ylim(c(0.5, 1)) Y vemos que la calibración es similar. Notemos sin embargo que la longitud del del intervalo bayesiano es mucho menor que el de Agresti-Coull cuando la muestra es chica: ggplot(bind_rows(cobertura_bayes, cobertura_ac), aes(x = p, y = long_media, colour = tipo)) + geom_point() + facet_wrap(~n) Cuando la muestra es chica, los intervalos de bayes son similares a los iniciales, y mucho más cortos que los de Agresti-Coull. Para muestras intermedias (50-100) los intervalos bayesianos son más informativos que los de Agresti-Coull, con calibración similar, y representan aprendizaje por encima de lo que sabíamos en la inicial. Para muestras grandes, obtenemos resultados simililares. Por ejemplo: set.seed(2131) k &lt;- rbinom(1, 50, 0.03) k ## [1] 4 intervalo_agresti_coull(50, k) |&gt; round(3) ## # A tibble: 1 × 2 ## inf sup ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.022 0.2 es un intervalo muy grande que puede incluir valores negativos. En contraste, el intervalo bayesiano es: intervalo_bayes_2(50, k) |&gt; round(3) ## # A tibble: 1 × 2 ## inf sup ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.015 0.076 Aún quitando valores negativos, los intervalos de Agresti-Coull son mucho más anchos. La aproximación bayesiana, entonces, utiliza información previa para dar un resultado considerablemente más informativo, con calibración similar a Agresti-Coull. ¿Aprendimos algo? Comparemos la posterior con la inicial: beta_sims_inicial &lt;- tibble(prop = rbeta(5000, a, b), dist = &quot;inicial&quot;) beta_sims_posterior &lt;- tibble(prop = rbeta(5000, a + k, b + 50), dist = &quot;posterior&quot;) bind_rows(beta_sims_inicial, beta_sims_posterior) |&gt; ggplot(aes(x = prop, fill = dist)) + geom_histogram(alpha = 0.5, position = &quot;identity&quot;) Donde vemos que no aprendimos mucho en este caso, pero nuestras creencias sí cambiaron en comparación con la inicial. Conclusión 2: con el enfoque bayesiano podemos obtener intervalos informativos con calibración razonable, incluso con información inicial que no es muy precisa. Los intervalos de Agresti-Coull son poco informativos para muestras chicas y/o proporciones chicas. Ejemplo: porporción de hogares de ingresos grandes Usaremos los datos de ENIGH como ejemplo (ignorando el diseño, pero es posible hacer todas las estimaciones correctamente) para estimar el porcentaje de hogares que tienen ingreso corriente de más de 150 mil pesos al trimestre. Suponemos que la muestra del enigh es la población, y tomaremos una muestra iid de esta población. Usamos la misma inicial que mostramos arriba, que es una Beta con parámetros c(a,b) ## [1] 2 100 set.seed(2521) muestra_enigh &lt;- read_csv(&quot;data/conjunto_de_datos_enigh_2018_ns_csv/conjunto_de_datos_concentradohogar_enigh_2018_ns/conjunto_de_datos/conjunto_de_datos_concentradohogar_enigh_2018_ns.csv&quot;) |&gt; select(ing_cor) |&gt; sample_n(120) |&gt; mutate(mas_150mil = ing_cor &gt; 150000) Un intervalo de 95% es entonces k &lt;- sum(muestra_enigh$mas_150mil) k ## [1] 3 intervalo_bayes_2(120, sum(muestra_enigh$mas_150mil)) |&gt; round(3) ## # A tibble: 1 × 2 ## inf sup ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.007 0.046 La media posterior es prop_post &lt;- (a + k) / (120 + b) prop_post ## [1] 0.02272727 ¿Cuál es la verdadera proporción? read_csv(&quot;data/conjunto_de_datos_enigh_2018_ns_csv/conjunto_de_datos_concentradohogar_enigh_2018_ns/conjunto_de_datos/conjunto_de_datos_concentradohogar_enigh_2018_ns.csv&quot;) |&gt; select(ing_cor) |&gt; mutate(mas_150mil = ing_cor &gt; 150000) |&gt; summarise(prop_pob = mean(mas_150mil)) ## # A tibble: 1 × 1 ## prop_pob ## &lt;dbl&gt; ## 1 0.0277 En este caso, nuestro intervalo cubre a la proporción poblacional. Inferencia bayesiana y regularización Como hemos visto en análisis y modelos anteriores, la posterior que usamos para hacer inferencia combina aspectos de la inicial con la verosimilitud (los datos). Una manera de ver esta combinación y sus beneficios es pensando en término de regularización de estimaciones. En las muestras hay variación. Algunas muestras particulares nos dan estimaciones de máxima verosimilitud pobres de los parámetros de interés (estimaciones ruidosas). Cuando esas estimaciones pobres están en una zona de baja probabilidad de la inicial, la estimación posterior tiende a moverse (o encogerse) hacia las zonas de alta probabilidad de la inicial. Esto filtra ruido en las estimaciones. El mecanismo resulta en una reducción del error cuadrático medio, mediante una reducción de la varianza de los estimadores (aunque quizá el sesgo aumente). Esta es una técnica poderosa, especialmente para problemas complejos donde tenemos pocos datos para cada parámetro. En general, excluímos resultados que no concuerdan con el conocimiento previo, y esto resulta en mayor precisión en las estimaciones. Ejemplo: modelo normal y estaturas Haremos un experimento donde simularemos muestras de los datos de cantantes. Usaremos el modelo normal-gamma inverso que discutimos anteriormente, con la información inicial que elicitamos. ¿Cómo se compara la estimación de máxima verosimilitud con la media posterior? # inicial para media, ver sección anterior para discusión (normal) mu_0 &lt;- 175 n_0 &lt;- 5 # inicial para sigma^2 (gamma inversa) a &lt;- 3 b &lt;- 140 Para este ejemplo chico, usaremos muestras de tamaño 5: set.seed(3413) # ver sección anterior para explicación de esta función calcular_pars_posterior &lt;- function(x, pars_inicial){ # iniciales mu_0 &lt;- pars_inicial[1] n_0 &lt;- pars_inicial[2] a_0 &lt;- pars_inicial[3] b_0 &lt;- pars_inicial[4] # muestra n &lt;- length(x) media &lt;- mean(x) S2 &lt;- sum((x - media)^2) # sigma post a_1 &lt;- a_0 + 0.5 * n b_1 &lt;- b_0 + 0.5 * S2 + 0.5 * (n * n_0) / (n + n_0) * (media - mu_0)^2 # posterior mu mu_1 &lt;- (n_0 * mu_0 + n * media) / (n + n_0) n_1 &lt;- n + n_0 c(mu_1, n_1, a_1, b_1) } Y también de la sección anterior: sim_params &lt;- function(m, pars){ mu_0 &lt;- pars[1] n_0 &lt;- pars[2] a &lt;- pars[3] b &lt;- pars[4] # simular sigmas sims &lt;- tibble(tau = rgamma(m, a, b)) |&gt; mutate(sigma = 1 / sqrt(tau)) # simular mu sims &lt;- sims |&gt; mutate(mu = rnorm(m, mu_0, sigma / sqrt(n_0))) sims } # simular muestras y calcular medias posteriores simular_muestra &lt;- function(rep, mu_0, n_0, a_0, b_0){ cantantes &lt;- lattice::singer |&gt; mutate(estatura_cm = 2.54 * height) |&gt; filter(str_detect(voice.part, &quot;Tenor&quot;)) |&gt; sample_n(5, replace = FALSE) pars_posterior &lt;- calcular_pars_posterior(cantantes$estatura_cm, c(mu_0, n_0, a_0, b_0)) medias_post &lt;- sim_params(1000, pars_posterior) |&gt; summarise(across(everything(), mean)) |&gt; select(mu, sigma) media &lt;- mean(cantantes$estatura_cm) est_mv &lt;- c(&quot;mu&quot; = media, &quot;sigma&quot; = sqrt(mean((cantantes$estatura_cm - media)^2))) bind_rows(medias_post, est_mv) |&gt; mutate(rep = rep, tipo = c(&quot;media_post&quot;, &quot;max_verosim&quot;)) |&gt; pivot_longer(mu:sigma, names_to = &quot;parametro&quot;, values_to = &quot;estimador&quot;) } poblacion &lt;- lattice::singer |&gt; mutate(estatura_cm = 2.54 * height) |&gt; filter(str_detect(voice.part, &quot;Tenor&quot;)) |&gt; summarise(mu = mean(estatura_cm), sigma = sd(estatura_cm)) |&gt; pivot_longer(mu:sigma, names_to = &quot;parametro&quot;, values_to = &quot;valor_pob&quot;) errores &lt;- map(1:2000, ~ simular_muestra(.x, mu_0, n_0, a, b)) |&gt; bind_rows() |&gt; left_join(poblacion) |&gt; mutate(error = (estimador - valor_pob)) ggplot(errores, aes(x = error, fill = tipo)) + geom_histogram(bins = 20, position = &quot;identity&quot;, alpha = 0.5) + facet_wrap(~parametro) Vemos claramente que la estimación de la desviación estándar de nuestro modelo es claramente superior a la de máxima verosimilitud. En resumen: errores |&gt; group_by(tipo, parametro) |&gt; summarise(recm = sqrt(mean(error^2)) |&gt; round(2)) |&gt; arrange(parametro) ## # A tibble: 4 × 3 ## # Groups: tipo [2] ## tipo parametro recm ## &lt;chr&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 max_verosim mu 2.85 ## 2 media_post mu 1.55 ## 3 max_verosim sigma 2.45 ## 4 media_post sigma 1.04 Obtenemos una ganancia considerable en cuanto a la estimación de la desviación estandar de esta población. Los estimadores de la media posterior son superiores a los de máxima verosimilitud en términos de error cuadrático medio. Podemos graficar las dos estimaciones, muestra a muestra, para entender cómo sucede esto: errores |&gt; select(-error) |&gt; pivot_wider(names_from = tipo, values_from = estimador) |&gt; filter(parametro == &quot;sigma&quot;) |&gt; ggplot(aes(x = max_verosim, y = media_post)) + geom_abline(colour = &quot;red&quot;) + geom_hline(yintercept = sqrt(b/(a - 1)), lty = 2, color = &#39;black&#39;) + geom_point() + labs(subtitle = &quot;Estimación de sigma&quot;) + xlab(&quot;Estimador MV de sigma&quot;) + ylab(&quot;Media posterior de sigma&quot;) + coord_fixed() + geom_segment(aes(x = 13, y = 11, xend = 13, yend = sqrt(b/(a - 1))), colour=&#39;red&#39;, size=1, arrow =arrow(length = unit(0.5, &quot;cm&quot;))) + geom_segment(aes(x = .5, y = 6, xend = .5, yend = sqrt(b/(a - 1))), colour=&#39;red&#39;, size=1, arrow =arrow(length = unit(0.5, &quot;cm&quot;))) Nótese como estimaciones demasiado bajas o demasiada altas son contraídas hacia valores más consistentes con la inicial, lo cual resulta en menor error. El valor esperado de \\(\\sigma\\) bajo la distribución inicial se muestra como una horizontal punteada. Ejemplo: estimación de proporciones Ahora repetimos el ejercicio # inicial a &lt;- 2 b &lt;- 100 qbeta(c(0.01, 0.99), a, b) ## [1] 0.001477084 0.063921446 # datos datos &lt;- read_csv(&quot;data/conjunto_de_datos_enigh_2018_ns_csv/conjunto_de_datos_concentradohogar_enigh_2018_ns/conjunto_de_datos/conjunto_de_datos_concentradohogar_enigh_2018_ns.csv&quot;) |&gt; select(ing_cor) # estimaciones obtener_estimados &lt;- function(datos){ muestra_enigh &lt;- datos |&gt; sample_n(120) |&gt; mutate(mas_150mil = ing_cor &gt; 150000) k &lt;- sum(muestra_enigh$mas_150mil) tibble(k = k, est_mv = k/120, media_post = (a + k) / (120 + b), pob = 0.02769) } estimadores_sim &lt;- map(1:200, ~obtener_estimados(datos)) |&gt; bind_rows() # calculo de errores error_cm &lt;- estimadores_sim |&gt; summarise(error_mv = sqrt(mean((est_mv - pob)^2)), error_post = sqrt(mean((media_post - pob)^2))) error_cm ## # A tibble: 1 × 2 ## error_mv error_post ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.0147 0.00928 Podemos ver claramente que las medias posteriores están encogidas hacia valores más chicos (donde la inicial tiene densidad alta) comparadas con las estimaciones de máxima verosimilitud: estimadores_sim_ag &lt;- estimadores_sim |&gt; group_by(k, est_mv, media_post) |&gt; summarise(n = n()) ggplot(estimadores_sim_ag, aes(x = est_mv, media_post, size = n)) + geom_point() + geom_abline() Teoría de decisión En esta parte (que sigue a Wasserman (2013) a grandes rasgos), discutimos brevemente teoría general que nos sirve para seleccionar estimadores puntuales, y que esperemos ponga en contexto la parte anterior que acabamos de discutir. Usaremos algunos conceptos que vimos en la parte de propiedades de estimadores de máxima verosimilitud. Definimos una función de pérdida \\(L(\\theta, \\hat{\\theta}_n)\\), que mide el costo de la discrepancia entre nuestro estimador \\[\\hat{\\theta}_n = t(X_1,\\ldots, X_n) = t(X)\\] y el verdadero valor \\(\\theta\\) poblacional. Es posible considerar distintas funciones de pérdida, pero como en secciones anteriores, usaremos la pérdida cuadrática, definida por: \\[L(\\theta, \\hat{\\theta}_n) = (\\theta - \\hat{\\theta}_n)^2\\] Esta función toma distintos valores dependiendo de la muestra y del parámetro \\(\\theta\\), y necesitamos resumirla para dar una evaluación de qué tan bueno es el estimador \\(\\hat{\\theta}_n\\). Ahora que hemos considerado tanto estadística bayesiana como frecuentista, podemos pensar en resumir esta función de distintas maneras. Comenzamos pensando de manera frecuentista. En este caso, consideramos a \\(\\theta\\) como un valor fijo, y nos preguntamos qué pasaría con la pérdida con distintas muestras potenciales que podríamos observar. Definimos como antes el riesgo (frecuentista) como: \\[R(\\theta, t) = \\mathbb{E}_X\\left[ (\\theta - \\hat{\\theta}_n)^2 \\, \\big| \\, \\theta\\right]\\] donde promediamos a lo largo de las muestras posibles, con \\(\\theta\\) fijo. Esta cantidad no nos dice necesariamente cómo escoger un buen estimador para \\(\\theta\\), pues dependiendo de dónde está \\(\\theta\\) puede tomar valores distintos. Ahora vamos a pensar de manera bayesiana: en este caso, los datos serán fijos una vez que los obervemos de manera que \\(\\hat{\\theta}_n\\) está fijo, y el parámetro \\(\\theta\\) es una cantidad aleatoria con distribución inicial \\(p(\\theta)\\). Entonces consideraríamos el promedio sobre la posterior dado por: \\[\\rho(t, X) = \\mathbb{E}_{p(\\theta|X)}\\left[(\\theta - \\hat{\\theta})^2 \\, \\big | \\, X \\right]\\] que llamamos riesgo posterior. Esta cantidad se calcula con la posterior de los parámetros dados los datos, y nos dice, una vez que vimos los datos, cómo es el error de nuestro estimador. Nótese que esta cantidad no es útil para escoger un estimador bueno \\(t\\) antes de ver los datos, pero nos sirve para evaluar a un estimador dados los datos. En el primer caso, promediamos sobre posibles muestras, y en el segundo por valores posibles de \\(\\theta\\) para una muestra dada. Ejemplo: riesgo frecuentista Para observaciones bernoulli, el estimador de máxima verosimilitud es \\(\\hat{p}_1 = k /n\\), donde \\(n\\) es el tamaño de muestra y \\(k\\) el número de éxitos observados. Podemos usar también como estimador la media posterior de un modelo Beta-Bernoulli con inicial \\(a=2, b=2\\), que nos daría \\(\\hat{p}_2 = \\frac{k + 2}{n + 4}\\). Aunque podemos hacer los cálculos analíticos, aproximaremos el riesgo bajo el error cuadrático usando simulación perdida_cuad &lt;- function(p, p_hat){ (p - p_hat)^2 } # dos estimadores t_1 &lt;- function(n, x) x / n t_2 &lt;- function(n, x) (x + 2) / (n + 4) estimar_riesgo &lt;- function(n = 20, theta, perdida, reps = 10000){ x &lt;- rbinom(reps, n, theta) # calcular estimaciones theta_1 &lt;- t_1(n, x) theta_2 &lt;- t_2(n, x) # calcular pérdida media_perdida &lt;- tibble( n = n, theta = theta, estimador = c(&quot;MLE&quot;, &quot;Posterior&quot;), riesgo = c(mean(perdida(theta, theta_1)), mean(perdida(theta, theta_2)))) media_perdida } estimar_riesgo(n = 20, theta = 0.1, perdida = perdida_cuad) ## # A tibble: 2 × 4 ## n theta estimador riesgo ## &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 20 0.1 MLE 0.00449 ## 2 20 0.1 Posterior 0.00755 Como dijimos, esta cantidad depende de \\(\\theta\\) que no conocemos. Así que calculamos para cada valor de \\(\\theta:\\) Las funciones de riesgo \\(R(\\theta, t_1)\\) y \\(R(\\theta, t_2)\\) (dependen de \\(\\theta\\)) se ven aproximadamente como sigue: p_seq &lt;- seq(0, 1, 0.001) riesgo_tbl &lt;- map(p_seq, ~ estimar_riesgo(n = 20, theta = .x, perdida = perdida_cuad)) |&gt; bind_rows() ggplot(riesgo_tbl, aes(x = theta, y = riesgo, colour = estimador)) + geom_line() Y vemos que el riesgo depende del verdadero valor del parametro: en los extremos, el estimador de máxima verosimilitud tiene menos riesgo, pero en el centro tiene más (esto es independiente del tipo de intervalos que construyamos y su cobertura). La razón es que las estimaciones de tipo Agresti-Coull (\\(\\theta_2\\)) están contraídas hacia 0.5 (agregamos dos éxitos y dos fracasos). Esto produce sesgo en la estimación para valores extremos de \\(\\theta\\). Sin embargo, para valores centrales de \\(\\theta\\) tiene menos variabilidad (por regularización) que el estimador de máxima verosimilitud, y sufre poco de sesgo. Ejemplo: riesgo posterior Supongamos que la inicial es \\(\\theta \\sim \\mathsf{Beta}(5,3)\\) estimar_riesgo_post &lt;- function(n = 20, x, perdida, reps = 20000){ # calcular estimaciones theta_1 &lt;- t_1(n, x) theta_2 &lt;- t_2(n, x) # simular de posterior theta_post &lt;- rbeta(reps, x + 5, n - x + 3) # calcular pérdida media_perdida &lt;- tibble( n = n, x = x, estimador = c(&quot;MLE&quot;, &quot;Posterior&quot;), riesgo_post= c(mean(perdida(theta_post, theta_1)), mean(perdida(theta_post, theta_2)))) media_perdida } estimar_riesgo_post(n = 20, x = 8, perdida = perdida_cuad) ## # A tibble: 2 × 4 ## n x estimador riesgo_post ## &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 20 8 MLE 0.0127 ## 2 20 8 Posterior 0.0109 Como dijimos, esta cantidad depende de los datos \\(x\\) que no hemos observado. Así que calculamos para cada valor de \\(x\\): Las funciones de pérdida promedio \\(\\rho(x, t_1)\\) y \\(\\rho(x, t_2)\\) (dependen de \\(x\\)) se ven aproximadamente como sigue: x_seq &lt;- seq(0, 20, 1) riesgo_post_tbl &lt;- map(x_seq, ~ estimar_riesgo_post(n = 20, x = .x, perdida = perdida_cuad)) |&gt; bind_rows() ggplot(riesgo_post_tbl, aes(x = x, y = riesgo_post, colour = estimador)) + geom_line() + geom_point() Donde vemos que la pérdida del estimador bayesiano es mejor para valores extremos de número de éxitos observado \\(x\\), pero tiene más riesgo posterior para valores chicos de \\(x\\). En general es mejor el estimador \\(\\theta_2\\). El estimador de máxima verosimilitud tiene más riesgo en los extremos, lo que esperaríamos porque no tenemos la regularización que aporta la posterior. Igualmente, vemos más riesgo para valores chicos de \\(x\\) que para valores grandes: esto es porque la inicial está concentrada en valores reslativamente grandes de \\(\\theta\\). Riesgo de Bayes Finalmente, podemos crear un resumen unificado considerando: Si no conocemos el valor del parámetro \\(\\theta\\), podemos promediar el riesgo frecuentista con la inicial \\(p(\\theta)\\) Si no conocemos los datos observados, podemos promediar usando datos generados por la marginal \\(p(x)\\) de \\(x\\) bajo el modelo de datos \\(p(x|\\theta)\\) y la inicial \\(p(\\theta)\\). Por la ley de la esperanza iterada, estos dos resultados son iguales. La cantidad resultante \\[r(t) = \\int R(\\theta,t) p(\\theta)\\, d\\theta = \\int r(x, t)p(x|\\theta)p(\\theta)\\, d\\theta\\, dx\\] Se llama riesgo de Bayes para el estimador \\(t\\). Ejemplo Podemos calcular marginal_tbl &lt;- function(n = 20, m = 5000){ theta &lt;- rbeta(m, 5, 3) x &lt;- rbinom(m, size = n, p = theta) tibble(x = x) |&gt; group_by(x) |&gt; summarise(n_x = n()) } riesgo_post_tbl |&gt; left_join(marginal_tbl()) |&gt; group_by(estimador) |&gt; summarise(riesgo_bayes = sum(riesgo_post * n_x) / sum(n_x)) ## # A tibble: 2 × 2 ## estimador riesgo_bayes ## &lt;chr&gt; &lt;dbl&gt; ## 1 MLE 0.0104 ## 2 Posterior 0.00833 o también theta_tbl &lt;- tibble(theta = rbeta(50000, 5, 3) |&gt; round(3)) |&gt; group_by(theta) |&gt; summarise(n_x = n()) riesgo_tbl |&gt; left_join(theta_tbl) |&gt; mutate(n_x = ifelse(is.na(n_x), 0, n_x)) |&gt; group_by(estimador) |&gt; summarise(riesgo_bayes = sum(riesgo * n_x) / sum(n_x)) ## # A tibble: 2 × 2 ## estimador riesgo_bayes ## &lt;chr&gt; &lt;dbl&gt; ## 1 MLE 0.0104 ## 2 Posterior 0.00839 Ahora consideremos cómo decidiríamos, desde el punto de vista Bayesiano, qué estimador usar: (Estimador de Bayes) Si tenemos los datos \\(X\\), escogeríamos una función \\(t_X\\) que minimice el riesgo posterior \\(\\rho(t, X)\\), y nuestro estimador es \\(\\hat{\\theta}_n = t_X (X)\\). (Regla de Bayes) Si no tenemos los datos, escogeríamos el estimador una función \\(t\\) que minimice el riesgo de Bayes \\(r(t)\\), y estimaríamos usando \\(\\hat{\\theta}_n = t(X)\\) Pero como el riesgo de Bayes es el promedio del riesgo posterior, la solución de 1 nos da la solución de 2. Es decir, el estimador que escogemos condicional a los datos \\(X\\) es el mismo que escogeríamos antes de escoger los datos, dada una distribución inicial \\(p(\\theta).\\) Por ejemplo, es posible demostrar que bajo la pérdida cuadrática, la regla de Bayes es utilizar la media posterior, bajo la pérdida absoluta, la mediana posterior, etc. Este estimador de Bayes tiene sentido desde el punto de vista frecuentista, también, porque minimiza el riesgo frecuentista promedio, suponiendo la inicial \\(p(\\theta)\\). Por ejemplo, para la pérdida cuadrática podemos usar la descomposición de sesgo y varianza y obtenemos: \\[r(t) = \\mathbb{E}[R(\\theta,t)] = \\mathbb{E}[ \\mathsf{Sesgo}_\\theta^2(t)] +\\mathbb{E}[\\mathsf{Var}_\\theta(t)] \\] Podemos ver entonces que el estimador de Bayes, en este caso la media posterior, es resultado de minimizar la suma de estas dos cantidades: por eso puede incurrir en sesgo, si ese costo se subsana con una reducción considerable de la varianza. Los estimadores insesgados que vimos en esta sección fueron subóptimos en muchos casos justamente porque son insesgados, e incurren en varianza grande. Regresa a los ejemplos anteriores donde evaluamos el desempeño de la media posterior en varios ejemplos. Muestra en las gráficas dónde ves el balance entre sesgo y varianza que cumplen cuando los comparamos con estimadores insesgados. Desde el punto de vista frecuentista, la cuestión es más complicada y hay varias maneras de proceder. En primer lugar, comenzaríamos con el riesgo frecuentista \\(R(\\theta, t)\\). Una idea es,por ejemplo, calcular el riesgo máximo: \\[R_{\\max} (t) = \\underset{\\theta}{\\max} R(\\theta, t).\\] En nuestro ejemplo de arriba el máximo se alcanza en 0.5, y tomaríamos eso evaluación de los estimadores \\(\\theta_1\\) o \\(\\theta_2\\). Buscaríamos entonces estimadores que minimicen este máximo, es decir, estimadores minimax. Pero también es posible enfocar este problema considerando sólo estimadores insesgados, lo que nos lleva por ejemplo a buscar estimadores con mínima varianza. También podemos enfocarnos en buscar estimador admisibles, que son aquellos cuyo riesgo no está dominado para toda \\(\\theta\\) por otro estimador, y así sucesivamente. Finalmente, es posible demostrar (ver Wasserman (2013)) que típicamente, para muestras grandes, el estimador de máxima verosimilitud es cercano a ser minimax y además es una regla de Bayes. Estas son buenas propiedades, pero debemos contar con que el régimen asintótico se cumpla aproximadamente. Referencias "],["métodos-de-cadenas-de-markov-monte-carlo.html", "Sección 12 Métodos de Cadenas de Markov Monte Carlo Integrales mediante subdivisiones Métodos Monte Carlo Simulando de la posterior 12.1 Ejemplo de islas ¿Por qué funciona Metrópolis? Método de Metrópolis Ajustando el tamaño de salto Metrópolis con varios parámetros Muestreador de Gibbs Conclusiones y observaciones Metrópolis y Gibbs HMC y Stan Diagnósticos generales para MCMC", " Sección 12 Métodos de Cadenas de Markov Monte Carlo Hasta ahora, hemos considerado modelos bayesianos conjugados, donde la posterior tiene una forma conocida. Esto nos permitió simular directamente de la posterior usando las rutinas estándar de R, o utilizar cálculos teóricos o funciones estándar de R para calcular resúmenes de interés, como medias o medianas posteriores o intervalos de credibilidad. Sin embargo, en aplicaciones rara vez es factible este tipo de análisis tan simple, pues: Los modelos que estamos considerando son más complejos y la distribución posterior conjunta de los parámetros no tiene una forma simple conocida. Queremos usar distribuciones iniciales que no son conjugadas para utilizar correctamente nuestra información inicial. Recordamos que tenemos expresiones explícitas para la inicial \\(p(\\theta)\\) y la verosimilitud \\(p(x|\\theta)\\), así que conocemos explícitamente la posterior, módulo la constante de normalización, \\[p(\\theta|x) \\propto p(x|\\theta) \\, p(\\theta).\\] Supongamos por ejemplo que quisiéramos calcular las medias posteriores de los parámetros \\(\\theta\\). En teoría, tendríamos que calcular \\[\\hat \\theta = \\mathbb{E}[{\\theta}\\, |\\, x] = \\int \\theta \\, p(\\theta|x) \\, d\\theta\\] Entonces es necesario calcular también \\(p(x)\\), que resulta de la integral \\[p(x) = \\int p(x|\\theta) \\, p(\\theta)\\, d\\theta\\] Si no tenemos expresiones analíticas simples, tendremos que aproximar numéricamente estas integrales de alguna forma. Si la posterior tiene una forma conocida, podemos calcular cantidades de interés usando fórmulas o rutinas de simulación de distribuciones conocidas que producen muestras independientes. Cuando la posterior no tiene una forma conocida, sin embargo: Podemos intentar usar integración numérica usual. Como veremos, este enfoque no es muy escalable. Podemos usar simulaciones bajo cadenas de Markov (Markov Chain Monte Carlo, MCMC), que es un enfoque más escalable. Mucho del uso generalizado actual de la estadística bayesiana se debe a que gracias al poder de cómputo disponible y los métodos MCMC, no estamos restringidos al uso de 1 y 2, que tienen desventajas grandes. Primero mostraremos cómo el método de integración por subdivisión no es escalable. Integrales mediante subdivisiones Como tenemos una expresión analítica para el integrando, podemos intentar una rutina numérica de integración. Una vez calculada, podríamos entonces usar otra rutina numérica para calcular las medias posteriores \\(\\hat{\\theta}\\). Las rutinas usuales de integración pueden sernos útiles cuando el número de parámetros es chico. Consideremos primero el caso de 1 dimensión, y supongamos que \\(a\\leq\\theta\\leq b\\). Si dividimos el rango de \\(\\theta\\) en intervalos determinados por \\(a = \\theta^1&lt;\\theta^2&lt;\\cdots \\theta^M =b\\), tales que \\(\\Delta\\theta = \\theta^{i+1} -\\theta^{i}\\), podríamos aproximar con \\[p(x) \\approx \\sum_{i=1}^M p(x|\\theta^i)p(\\theta^i) \\Delta\\theta\\] Lo que requiere \\(M\\) evaluaciones del factor \\(p(x|\\theta)p(\\theta)\\). Podríamos usar por ejemplo \\(M=100\\) para tener precisión razonable. Ejemplo: estimación de una proporción Teníamos que \\(p(S_n = k|\\theta) \\propto \\theta^k(1-\\theta)^{n-k}\\) cuando observamos \\(k\\) éxitos en \\(n\\) pruebas independientes. Supongamos que nuestra inicial es \\(p(\\theta) = 2\\theta\\) (checa que es una densidad), es decir, creemos que es más probable a priori observar proporciones altas. Podemos integrar numéricamente crear_log_post &lt;- function(n, k){ function(theta){ verosim &lt;- k * log(theta) + (n - k) * log(1 - theta) inicial &lt;- log(theta) log_p_factor &lt;- verosim + inicial log_p_factor } } # observamos 3 éxitos en 4 pruebas: log_post &lt;- crear_log_post(4, 3) prob_post &lt;- function(x) { exp(log_post(x))} # integramos numéricamente p_x &lt;- integrate(prob_post, lower = 0, upper = 1, subdivisions = 100L) p_x ## 0.03333333 with absolute error &lt; 3.7e-16 Y ahora podemos calcular la media posterior: media_funcion &lt;- function(theta){ theta * prob_post(theta) / p_x$value } integral_media &lt;- integrate(media_funcion, lower = 0, upper = 1, subdivisions = 100L) media_post &lt;- integral_media$value media_post ## [1] 0.7142857 Podemos verificar nuestro trabajo pues sabemos que la posterior es \\(\\mathsf{Beta}(5, 2)\\) cuya media es 5/(2+5) ## [1] 0.7142857 Y podríamos intentar una estrategia similar, por ejemplo, para calcular intervalos de credibilidad. Sin embargo, veremos abajo que este método no escala con el número de parámetros. Más de un parámetro Ahora supongamos que tenemos \\(2\\) parámetros. Dividiríamos cada parámetro en 100 intervalos, y luego tendríamos que calcular \\[p(x) \\approx \\sum_{i=1}^M \\sum_{j=1}^M p(x|\\theta_1^i, \\theta_2^j)p(\\theta_1^i, \\theta_2^j) \\Delta\\theta_1\\Delta\\theta_2\\] Y esto requeriría \\(M^2 = 10,000\\) evaluaciones de \\(p(x|\\theta)p(\\theta)\\). Si tenemos \\(p\\) parámetros, entonces tendríamos que hacer \\(M^p\\) evaluaciones de la posterior. Incluso cuando \\(p=10\\), esta estrategia es infactible, pues tendríamos que hacer más de millones de millones de millones de evaluaciones de la posterior. Si sólo tenemos esta técnica disponible, el análisis bayesiano está considerablemente restringido. Regresión bayesiana con unas 10 covariables por ejemplo, no podría hacerse. De modo que tenemos que replantearnos cómo atacar el problema de calcular o aproximar estas integrales. Métodos Monte Carlo En varias ocasiones anteriormente hemos usado el método Monte Carlo para aproximar integrales: por ejemplo, para calcular medias posteriores. Supongamos que tenemos una densidad \\(p(\\theta)\\). Integración Monte Carlo. Supongamos que queremos calcular el valor esperado de \\(g(X)\\), donde \\(X\\sim p(X\\,|\\,\\theta).\\) Es decir, la variable aleatoria \\(X\\) se distribuye de acuerdo al modelo probabilistico \\(p(X \\, | \\, \\theta),\\) de tal forma que lo que nos interesa calcular es \\[\\mathbb{E}[g(X)] = \\int g(x) p(x|\\theta)\\, dx.\\] Si tomamos una muestra \\(x^{(1)},x^{(2)}, \\ldots x^{(N)} \\overset{iid}{\\sim} p(x|\\theta)\\), entonces \\[\\mathbb{E}[g(X)] \\approx \\, \\frac1N \\, \\sum_{n = 1}^N g(x^{(n)})\\] cuando \\(N\\) es grande. Esto es simplemente una manera de escribir la ley de los grandes números, y hemos aplicado este teorema en varias ocasiones. Nos ha sido útil cuando sabemos cómo simular de distribución \\(p(\\theta | x)\\) (usando alguna rutina de R, por ejemplo, o usando un método estándar como inversión de la función de distribución acumulada). Ejemplo En este ejemplo repetimos cosas que ya hemos visto. En el caso de estimación de una proporción \\(\\theta\\), tenemos como inicial \\(p(\\theta) \\propto \\theta\\), que es \\(\\mathsf{Beta}(2,1)\\). Si observamos 3 éxitos en 4 pruebas, entonces sabemos que la posterior es \\(p(\\theta|x)\\propto \\theta^4(1-\\theta)\\), que es \\(\\mathsf{Beta}(5, 2)\\). Si queremos calcular media y segundo momento posterior para \\(\\theta\\), en teoría necesitamos calcular \\[\\mu = \\int_0^1 \\theta p(\\theta|X = 3)\\, d\\theta,\\,\\, \\mu_2=\\int_0^1 \\theta^2 p(\\theta|X = 3)\\, d\\theta\\] integramos con Monte Carlo theta &lt;- rbeta(10000, 5, 2) media_post &lt;- mean(theta) momento_2_post &lt;- mean(theta^2) c(media_post, momento_2_post) ## [1] 0.7155559 0.5372170 Y podemos aproximar de esta manera cualquier cantidad de interés que esté basada en integrales, como probabilidades asociadas a \\(\\theta\\) o cuantiles asociados. Por ejemplo, podemos aproximar fácilmente \\(P(e^{\\theta}&gt; 2|x)\\) haciendo mean(exp(theta) &gt; 2) ## [1] 0.5958 y así sucesivamente. Este enfoque, sin embargo, es mucho más flexible y poderoso. Ejemplo: varias pruebas independientes Supongamos que probamos el nivel de gusto para 4 sabores distintos de una paleta. Usamos 4 muestras de aproximadamente 50 personas diferentes para cada sabor, y cada uno evalúa si le gustó mucho o no. Obtenemos los siguientes resultados: datos &lt;- tibble( sabor = c(&quot;fresa&quot;, &quot;limón&quot;, &quot;mango&quot;, &quot;guanábana&quot;), n = c(50, 45, 51, 50), gusto = c(36, 35, 42, 29)) %&gt;% mutate(prop_gust = gusto / n) datos ## # A tibble: 4 × 4 ## sabor n gusto prop_gust ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 fresa 50 36 0.72 ## 2 limón 45 35 0.778 ## 3 mango 51 42 0.824 ## 4 guanábana 50 29 0.58 Usaremos como inicial \\(\\mathsf{Beta}(2, 1)\\) (pues hemos obervado cierto sesgo de cortesía en la calificación de sabores, y no es tan probable tener valores muy bajos) para todos los sabores, es decir \\(p(\\theta_i)\\) es la funcion de densidad de una \\(\\mathsf{Beta}(2, 1)\\). La inicial conjunta la definimos entonces, usando idependiencia inicial, como \\[p(\\theta_1,\\theta_2, \\theta_3,\\theta_4) = p(\\theta_1)p(\\theta_2)p(\\theta_3)p(\\theta_4).\\] Pues inicialmente establecemos que ningún parámetro da información sobre otro: saber que mango es muy gustado no nos dice nada acerca del gusto por fresa. Bajo este supuesto, y el supuesto adicional de que las muestras de cada sabor son independientes, podemos mostrar que las posteriores son independientes: \\[p(\\theta_1,\\theta_2,\\theta_3, \\theta_4|k_1,k_2,k_3,k_4) = p(\\theta_4|k_1)p(\\theta_4|k_2)p(\\theta_4|k_3)p(\\theta_4|k_4)\\] De forma que podemos trabajar individualmente con cada muestra. Calculamos los parámetros de las posteriores individuales: datos &lt;- datos %&gt;% mutate(a_post = gusto + 2, b_post = n - gusto + 1) datos ## # A tibble: 4 × 6 ## sabor n gusto prop_gust a_post b_post ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 fresa 50 36 0.72 38 15 ## 2 limón 45 35 0.778 37 11 ## 3 mango 51 42 0.824 44 10 ## 4 guanábana 50 29 0.58 31 22 Ahora nos preguntamos, ¿cuál es la probabilidad posterior de que mango sea el sabor más preferido de la población? Conocemos la posterior para cada parámetro, y sabemos que los parámetros son independientes para la posterior. Eso quiere decir que podemos simular de cada parámetro independientemente para obtener simulaciones de la conjunta posterior. simular_conjunta &lt;- function(rep, datos){ datos %&gt;% mutate(valor_sim = map2_dbl(a_post, b_post, ~ rbeta(1, .x, .y))) %&gt;% select(sabor, valor_sim) } simular_conjunta(1, datos) ## # A tibble: 4 × 2 ## sabor valor_sim ## &lt;chr&gt; &lt;dbl&gt; ## 1 fresa 0.755 ## 2 limón 0.783 ## 3 mango 0.819 ## 4 guanábana 0.569 # esta no es una manera muy rápida, podríamos calcular todas las # simulaciones de cada parámetro de manera vectorizada sims_posterior &lt;- tibble(rep = 1:5000) %&gt;% mutate(sims = map(rep, ~ simular_conjunta(.x, datos))) %&gt;% unnest(cols = sims) sims_posterior ## # A tibble: 20,000 × 3 ## rep sabor valor_sim ## &lt;int&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 1 fresa 0.732 ## 2 1 limón 0.831 ## 3 1 mango 0.850 ## 4 1 guanábana 0.397 ## 5 2 fresa 0.670 ## 6 2 limón 0.839 ## 7 2 mango 0.664 ## 8 2 guanábana 0.558 ## 9 3 fresa 0.671 ## 10 3 limón 0.758 ## # ℹ 19,990 more rows Y ahora podemos aproximar fácilmente la probabilidad de interés: sims_posterior %&gt;% group_by(rep) %&gt;% mutate(sabor = sabor[which.max(valor_sim)]) %&gt;% group_by(sabor) %&gt;% count() %&gt;% ungroup() %&gt;% mutate(prop = n / sum(n)) ## # A tibble: 4 × 3 ## sabor n prop ## &lt;chr&gt; &lt;int&gt; &lt;dbl&gt; ## 1 fresa 1264 0.0632 ## 2 guanábana 8 0.0004 ## 3 limón 5396 0.270 ## 4 mango 13332 0.667 Y vemos que los mejores sabores son mango y limón. La probabilidad posterior de que mango sea el sabor preferido por la población es de 66%. La integral correspondiente no es trivial. ¿Cuáles son las probabilidades a priori de que cada sabor sea el preferido por la población? ¿Cuál es la integral correspondiente a las probabilidades que acabamos de calcular? ¿Qué tan fácil es hacer esta integral de manera analítica? Calcula la probabilidad de que mango sea preferido a limón? ¿Qué conclusión práctica sacas de estos resultados? Simulando de la posterior Hemos establecido que podemos contestar varias preguntas de inferencia usando simulación Monte Carlo, y que este enfoque es potencialmente escalable (en contraste con métodos de integración numérica por cuadrícula). Ahora el problema que necesitamos resolver es el siguiente: Conocemos \\(p(\\theta |x)\\) módulo una constante de integración. En general, \\(p(\\theta|x)\\) no tiene una forma reconocible que corresponda a un simulador estándar. ¿Cómo simulamos de esta posterior cuando sólo sabemos calcular \\(p(x|\\theta)p(\\theta)\\)? Hay varias maneras de hacer esto. Presentaremos los algoritmos en términos de una distribución cualquiera \\(p(\\theta) = K f(\\theta)\\), donde sólo conocemos la función \\(f(\\theta)\\). 12.1 Ejemplo de islas Comenzamos revisando el ejemplo de las islas en Kruschke (2015) (7.2) para tener más intuición de cómo funciona este algoritmo. ¿Por qué funciona Metrópolis? Supongamos que un vendedor de Yakult trabaja a lo largo de una cadena de islas: Constantemente viaja entre las islas ofreciendo sus productos; Al final de un día de trabajo decide si permanece en la misma isla o se transporta a una de las \\(2\\) islas vecinas; El vendedor ignora la distribución de la población en las islas y el número total de islas; sin embargo, una vez que se encuentra en una isla puede investigar la población de la misma y también de la isla a la que se propone viajar después. El objetivo del vendedor es visitar las islas de manera proporcional a la población de cada una. Con esto en mente el vendedor utiliza el siguiente proceso: Lanza un volado, si el resultado es águila se propone ir a la isla del lado izquierdo de su ubicación actual y si es sol a la del lado derecho. Si la isla propuesta en el paso anterior tiene población mayor a la población de la isla actual, el vendedor decide viajar a ella. Si la isla vecina tiene población menor, entonces visita la isla propuesta con una probabilidad que depende de la población de las islas. Sea \\(P^*\\) la población de la isla propuesta y \\(P_{t}\\) la población de la isla actual. Entonces el vendedor cambia de isla con probabilidad \\[q_{mover}=P^*/P_{t}\\] A la larga, si el vendedor sigue la heurística anterior la probabilidad de que el vendedor este en alguna de las islas coincide con la población relativa de la isla. islas &lt;- tibble(islas = 1:10, pob = 1:10) camina_isla &lt;- function(i){ # i: isla actual u &lt;- runif(1) # volado v &lt;- ifelse(u &lt; 0.5, i - 1, i + 1) # isla vecina (índice) if (v &lt; 1 | v &gt; 10) { # si estás en los extremos y el volado indica salir return(i) } p_move = ifelse(islas$pob[v] &gt; islas$pob[i], 1, islas$pob[v] / islas$pob[i]) u2 &lt;- runif(1) if (p_move &gt; u2) { return(v) # isla destino } else { return(i) # me quedo en la misma isla } } pasos &lt;- 100000 iteraciones &lt;- numeric(pasos) iteraciones[1] &lt;- sample(1:10, 1) # isla inicial for (j in 2:pasos) { iteraciones[j] &lt;- camina_isla(iteraciones[j - 1]) } caminata &lt;- tibble(pasos = 1:pasos, isla = iteraciones) plot_caminata &lt;- ggplot(caminata[1:1000, ], aes(x = pasos, y = isla)) + geom_point(size = 0.8) + geom_path(alpha = 0.5) + coord_flip() + labs(title = &quot;Caminata aleatoria&quot;) + scale_y_continuous(expression(theta), breaks = 1:10) + scale_x_continuous(&quot;Tiempo&quot;) plot_dist &lt;- ggplot(caminata, aes(x = isla)) + geom_histogram() + scale_x_continuous(expression(theta), breaks = 1:10) + labs(title = &quot;Distribución objetivo&quot;, y = expression(P(theta))) plot_caminata / plot_dist Entonces: Para aproximar la distribución objetivo debemos permitir que el vendedor recorra las islas durante una sucesión larga de pasos y registramos sus visitas. Nuestra aproximación de la distribución es justamente el registro de sus visitas. Más aún, debemos tener cuidado y excluir la porción de las visitas que se encuentran bajo la influencia de la posición inicial. Esto es, debemos excluir el periodo de calentamiento. Una vez que tenemos un registro largo de los viajes del vendedor (excluyendo el calentamiento) podemos aproximar la distribución objetivo simplemente contando el número relativo de veces que el vendedor visitó dicha isla. t &lt;- c(1:10, 20, 50, 100, 200, 1000, 5000) plots_list &lt;- map(t, function(i){ ggplot(caminata[1:i, ], aes(x = isla)) + geom_histogram() + labs(y = &quot;&quot;, x = &quot;&quot;, title = paste(&quot;t = &quot;, i, sep = &quot;&quot;)) + scale_x_continuous(expression(theta), breaks = 1:10, limits = c(0, 11)) }) wrap_plots(plots_list) Escribamos el algoritmo, para esto indexamos las islas por el valor \\(\\theta\\), es así que la isla del extremo oeste corresponde a \\(\\theta=1\\) y la población relativa de cada isla es \\(P(\\theta)\\): El vendedor se ubica en \\(\\theta^{(i)}\\) y propone moverse a la izquierda o derecha con probabilidad \\(0.5\\). El rango de los posibles valores para moverse, y la probabilidad de proponer cada uno se conoce como distribución propuesta, en nuestro ejemplo sólo toma dos valores cada uno con probabilidad \\(0.5\\). Una vez que se propone un movimiento, decidimos si aceptarlo. La decisión de aceptar se basa en el valor de la distribución objetivo en la posición propuesta, relativo al valor de la distribución objetivo en la posición actual: \\[\\alpha=\\min\\bigg\\{\\frac{P(\\theta^*)}{P(\\theta^{(i)})},1\\bigg\\},\\] donde \\(\\alpha\\) denota la probabilidad de hacer el cambio de isla. Notemos que la distribución objetivo \\(P(\\theta)\\) no necesita estar normalizada, esto es porque lo que nos interesa es el cociente \\(P(\\theta^*)/P(\\theta^{(i)})\\). Una vez que propusimos un movimiento y calculamos la probabilidad de aceptar el movimiento aceptamos o rechazamos el movimiento generando un valor de una distribución uniforme, si dicho valor es menor a la probabilidad de cambio, \\(\\alpha,\\) entonces hacemos el movimiento. Entonces, para utilizar el algoritmo necesitamos ser capaces de: Generar un valor de la distribución propuesta, que hemos denotado por \\(q,\\) (para crear \\(\\theta^*\\)). Evaluar la distribución objetivo en cualquier valor propuesto (para calcular \\(P(\\theta^*)/P(\\theta^{(i)})\\)). Generar un valor uniforme (para movernos con probabilidad \\(\\alpha\\)). Las \\(3\\) puntos anteriores nos permiten generar muestras aleatorias de la distribución objetivo, sin importar si esta está normalizada. Esta técnica es particularmente útil cuando cuando la distribución objetivo es una posterior proporcional a \\(p(x|\\theta)p(\\theta)\\). Para entender porque funciona el algoritmo de Metrópolis hace falta entender \\(2\\) puntos, primero que la distribución objetivo es estable: si la probabilidad actual de ubicarse en una posición coincide con la probabilidad en la distribución objetivo, entonces el algoritmo preserva las probabilidades. library(expm) transMat &lt;- function(P){ # recibe vector de probabilidades (o población) T &lt;- matrix(0, 10, 10) n &lt;- length(P - 1) # número de estados for (j in 2:n - 1) { # llenamos por fila T[j, j - 1] &lt;- 0.5 * min(P[j - 1] / P[j], 1) T[j, j] &lt;- 0.5 * (1 - min(P[j - 1] / P[j], 1)) + 0.5 * (1 - min(P[j + 1] / P[j], 1)) T[j, j + 1] &lt;- 0.5 * min(P[j + 1] / P[j], 1) } # faltan los casos j = 1 y j = n T[1, 1] &lt;- 0.5 + 0.5 * (1 - min(P[2] / P[1], 1)) T[1, 2] &lt;- 0.5 * min(P[2] / P[1], 1) T[n, n] &lt;- 0.5 + 0.5 * (1 - min(P[n - 1] / P[n], 1)) T[n, n - 1] &lt;- 0.5 * min(P[n - 1] / P[n], 1) T } T &lt;- transMat(islas$pob) w &lt;- c(0, 1, rep(0, 8)) t &lt;- c(1:10, 20, 50, 100, 200, 1000, 5000) expT &lt;- map_df(t, ~data.frame(t = ., w %*% (T %^% .))) expT_long &lt;- expT %&gt;% gather(theta, P, -t) %&gt;% mutate(theta = parse_number(theta)) ggplot(expT_long, aes(x = theta, y = P)) + geom_bar(stat = &quot;identity&quot;, fill = &quot;darkgray&quot;) + facet_wrap(~ t) + scale_x_continuous(expression(theta), breaks = 1:10, limits = c(0, 11)) El segundo punto es que el proceso converge a la distribución objetivo. Podemos ver, (en nuestro ejemplo sencillo) que sin importar el punto de inicio se alcanza la distribución objetivo. inicio_p &lt;- function(i){ w &lt;- rep(0, 10) w[i] &lt;- 1 t &lt;- c(1, 10, 50, 100) exp_t &lt;- map_df(t, ~ data.frame(t = .x, inicio = i, w %*% (T %^% .))) %&gt;% gather(theta, P, -t, -inicio) %&gt;% mutate(theta = parse_number(theta)) exp_t } exp_t &lt;- map_df(c(1, 3, 5, 9), inicio_p) ggplot(exp_t, aes(x = as.numeric(theta), y = P)) + geom_bar(stat = &quot;identity&quot;, fill = &quot;darkgray&quot;) + facet_grid(inicio ~ t) + scale_x_continuous(expression(theta), breaks = 1:10, limits = c(0, 11)) Método de Metrópolis En el método de Metrópolis, uno de los más antiguos, comenzamos con un valor inicial de los parámetros \\(\\theta^{(0)}\\) en el soporte de \\(p(\\theta)\\), es decir \\(p(\\theta^{(0)})&gt;0.\\) Para \\(i=1, \\ldots, M\\), hacemos: Partiendo de \\(\\theta^{(i)}\\), hacemos un salto corto en una dirección al azar para obtener una propuesta \\(\\theta^* \\sim q(\\theta \\, |\\, \\theta^{(i)}).\\) Aceptamos or rechazamos el salto: Si \\(\\alpha = \\frac{f(\\theta^*)}{f(\\theta^{(i)})} \\geq 1\\), aceptamos el salto y ponemos \\(\\theta^{(i+1)}=\\theta^*\\). Regresamos a 1 para la siguiente iteración \\(i\\leftarrow i + 1.\\) Si \\(\\alpha = \\frac{f(\\theta^*)}{f(\\theta^{(i)})} &lt; 1\\), entonces aceptamos con probabilidad \\(\\alpha\\) el salto, ponemos \\(\\theta^{(i+1)}=\\theta^*\\) y regresamos a 1 para la siguiente iteración \\(i\\leftarrow i + 1\\). Si rechazamos el salto, ponemos entonces \\(\\theta^{(i+1)}=\\theta^{(i)}\\) y regresamos a 1 para la siguiente iteración \\(i\\leftarrow i + 1.\\) Requerimos también que la función que propone los saltos sea simétrica: es decir, \\(q(\\theta^*|\\theta^{(i)})\\) debe ser igual a \\(q(\\theta^{(i)}|\\theta^*)\\). Se puede modificar el algoritmo para tratar con una propuesta que no sea simétrica. Una elección común es escoger \\(q(\\theta^* |\\theta^{(i)})\\) como una \\(\\mathsf{N}(\\theta^{(i)}, \\sigma_{salto})\\). En este curso, escribiremos varios métodos de cadenas de Markov para estimación Monte Carlo (Markov Chain Monte Carlo, MCMC) desde cero para entender los básicos de cómo funciona. Sin embargo, en la práctica no hacemos esto, sino que usamos software estándar (Stan, JAGS, BUGS, etc.) para hacer este trabajo. Expertos en MCMC, métodos numéricos, y estadística a veces escriben partes de sus rutinas de simulación, y pueden lograr mejoras de desempeño considerables. Excepto para modelos simples, esto no es trivial de hacer garantizando resultados correctos. En resumen, todo el código de esta sección es de carácter ilustrativo. Utiliza implementaciones establecidas en las aplicaciones. Abajo implementamos el algoritmo con un salto de tipo normal: crear_metropolis &lt;- function(fun_log, sigma_salto = 0.1){ # la entrada es la log posterior iterar_metropolis &lt;- function(theta_inicial, n){ p &lt;- length(theta_inicial) nombres &lt;- names(theta_inicial) iteraciones &lt;- matrix(0, nrow = n, ncol = p) colnames(iteraciones) &lt;- nombres iteraciones[1,] &lt;- theta_inicial for(i in 2:n){ theta &lt;- iteraciones[i - 1, ] theta_prop &lt;- theta + rnorm(p, 0, sigma_salto) # exp(log(p) - log(q)) = p/q cociente &lt;- exp(fun_log(theta_prop) - fun_log(theta)) if(cociente &gt;= 1 || runif(1,0,1) &lt; cociente){ iteraciones[i, ] &lt;- theta_prop } else { iteraciones[i, ] &lt;- theta } } iteraciones_tbl &lt;- iteraciones %&gt;% as_tibble() %&gt;% mutate(iter_num = row_number()) %&gt;% select(iter_num, everything()) iteraciones_tbl } iterar_metropolis } E intentamos simular de una exponencial no normalizada: exp_no_norm &lt;- function(x) { z &lt;- ifelse(x &gt; 0, exp(-0.5 * x), 0) log(z) } iterador_metro &lt;- crear_metropolis(exp_no_norm, sigma_salto = 0.25) sims_tbl &lt;- iterador_metro(c(theta = 0.5), 50000) ggplot(sims_tbl, aes(x = theta)) + geom_histogram() Ahora probemos con una \\(\\mathsf{Beta}(3, 2):\\) beta_no_norm &lt;- function(x) { z &lt;- ifelse(x &gt; 0 &amp;&amp; x &lt; 1, (x^2)*(1-x), 0) log(z) } iterador_metro &lt;- crear_metropolis(beta_no_norm, sigma_salto = 0.04) sims_metro_tbl &lt;- iterador_metro(c(theta = 0.5), 50000) sims_indep_tbl &lt;- tibble(iter_num = 1:30000, theta = rbeta(30000, 3, 2)) g_1 &lt;- ggplot(sims_metro_tbl, aes(x = theta)) + geom_histogram() + labs(subtitle = &quot;Metrópolis&quot;) g_2 &lt;- ggplot(sims_indep_tbl, aes(x = theta)) + geom_histogram() + labs(subtitle = &quot;rbeta&quot;) g_1 + g_2 Y vemos que esto funciona. Nótese sin embargo un aspecto de estas simulaciones que no habíamos encontrado en el curso. Aunque la distribución final de las simulaciones es muy cercana a la de la distribución que queremos simular, lo cual era nuestro propósito, las simulaciones no son extracciones independientes de esa distribución. La construcción del algoritmo muestra eso, pero podemos también graficar las simulaciones: g_metropolis &lt;- sims_metro_tbl %&gt;% filter(iter_num &lt; 500) %&gt;% ggplot(aes(x = iter_num, y = theta)) + geom_line() + labs(subtitle = &quot;Metrópolis&quot;) g_indep &lt;- sims_indep_tbl %&gt;% filter(iter_num &lt; 500) %&gt;% ggplot(aes(x = iter_num, y = theta)) + geom_line() + labs(subtitle = &quot;Independientes&quot;) g_metropolis + g_indep Donde vemos claramente que las simulaciones de metropolis están autocorrelacionadas: la siguiente simulación depende de la anterior. Esto define una cadena de Markov. En cualquiera de los dos casos, como vimos en los histogramas de arriba, las simulaciones “visitan” cada parte [0,1] de manera proporcional a la densidad, de manera que podemos usar ambos tipos de simulaciones para aproximar la integral o cantidad que nos interesa. Por ejemplo, la media posterior es: media_1 &lt;- sims_metro_tbl %&gt;% summarise(media_post = mean(theta)) %&gt;% pull(media_post) media_2 &lt;- sims_indep_tbl %&gt;% summarise(media_post = mean(theta)) %&gt;% pull(media_post) media_exacta &lt;- 3/(3 + 2) tibble(metodo = c(&quot;sim Metrópolis&quot;, &quot;sim Independiente&quot;, &quot;exacto&quot;), media_post = c(media_1, media_2, media_exacta)) ## # A tibble: 3 × 2 ## metodo media_post ## &lt;chr&gt; &lt;dbl&gt; ## 1 sim Metrópolis 0.605 ## 2 sim Independiente 0.602 ## 3 exacto 0.6 Supongamos que queremos simular de una distribución \\(p(\\theta)\\), pero sólo conocemos \\(p(\\theta)\\) módulo una constante. Bajo ciertas condiciones de regularidad: El algoritmo Metrópolis para la distribución \\(p(\\theta)\\) define una cadena de Markov cuya distribución a largo plazo es \\(p(\\theta)\\). Esto implica que si \\(\\theta^{(1)},\\theta^{(2)}, \\ldots, \\theta^{(M)}\\) es una simulación de esta cadena, y \\(M\\) es suficientemente grande La distribución de las \\(\\theta^{(i)}\\) es aproximadamente \\(p(\\theta)\\), Tenemos que \\[ \\frac1M \\sum_{m = 1}^M h(\\theta^{(m)}) \\to \\int h(\\theta)p(\\theta)\\, d\\theta\\] cuando \\(M\\to \\infty\\) Observaciones: Aunque hay distintas condiciones de regularidad que pueden funcionar, generalmente el supuesto es que la cadena de Markov construída es ergódica, y hay varias condiciones que garantizan esta propiedad. Una condición simple, por ejemplo, es que el soporte de la distribución \\(p(\\theta)\\) es un conjunto conexo del espacio de parámetros. Más crucialmente, este resultado no dice qué tan grande debe ser \\(M\\) para que la aproximación sea buena. Esto depende de cómo es \\(p(\\theta)\\), y de la distribución que se utiliza para obtener los saltos propuestos. Dependiendo de estos dos factores, la convergencia puede ser rápida (exponencial) o tan lenta que es infactible usarla. Más adelante veremos diagnósticos para descartar los peores casos de falta de convergencia. Ajustando el tamaño de salto En el algoritmo Metrópolis, generalmente es importante escoger la dispersión de la distribución que genera propuestas con cuidado. Si la dispersión de la propuesta es demasiado grande, tenderemos a rechazar mucho, y la convergencia será lenta. Si la dispersión de la propuesta es demasiado chica, tardaremos mucho tiempo en explorar las distintas partes de la distribución objetivo. Ejemplo Supongamos que queremos simular usando metróplis de una distribución \\(\\textsf{Gamma}(20, 100)\\). Abajo vemos la forma de esta distribución: sim_indep &lt;- tibble(theta = rgamma(10000, 20, 100)) ggplot(sim_indep, aes(x = theta)) + geom_histogram() # logaritmo de densidad no normalizada log_f_dist &lt;- function(x) 210 + dgamma(x, 20, 100, log = TRUE) # iterar iterador_metro_chico &lt;- crear_metropolis(log_f_dist, sigma_salto = 0.001) sims_chico_tbl &lt;- iterador_metro_chico(c(theta = 0.02), 50000) g_sim &lt;- ggplot(sims_chico_tbl %&gt;% filter(iter_num &lt; 3000), aes(x = iter_num, y = theta)) + geom_line() + ylim(c(0, 0.5)) dist_bplot &lt;- ggplot(tibble(x = rgamma(10000, 20, 100)), aes(y = x, x = &quot;a&quot;)) + geom_violin() + ylab(&quot;&quot;) + ylim(0, 0.5) g_sim + dist_bplot + plot_layout(widths = c(5, 1)) Nótese que después de 5 mil iteraciones estamos muy lejos de tener una muestra que se aproxime a la distribución objetivo. Empezamos en un lugar bajo, y la cadena sólo ha ido lentamente hacia las zonas de alta densidad. Cualquier resumen con esta cadena estaría fuertemente sesgado al valor donde iniciamos la iteración. Decimos que la cadena todavía no mezcla en las primeras 5 mil iteraciones. Ahora vemos qué pasa si ponemos el tamaño de salto demasiado grande: set.seed(831) iterador_metro_grande &lt;- crear_metropolis(log_f_dist, sigma_salto = 20) sims_grande_tbl &lt;- iterador_metro_grande(c(theta = 0.02), 50000) g_sim &lt;- ggplot(sims_grande_tbl %&gt;% filter(iter_num &lt; 3000), aes(x = iter_num, y = theta)) + geom_line() + ylim(c(0, 0.5)) g_sim + dist_bplot + plot_layout(widths = c(5, 1)) En este caso, la cadena se atora muchas veces, pues las propuestas tienen probabilidad muy baja, y tendemos a tener una tasa de rechazos muy alta. Esto quiere decir que la información que tenemos acerca de la posterior es relativamente poca, pues muchos datos son repeticiones del mismo valor. Cualquier resumen con esta cadena podría estar muy lejos del verdadero valor, pues su varianza es alta - otra corrida se “atoraría” en otros valores distintos. Nótese que cualquiera de estas cadenas, si la corremos suficientemente tiempo, nos daría resultados buenos. Sin embargo, el número de simulaciones puede ser infactible. Un valor intermedio nos dará mucho mejores resultados: set.seed(831) iterador_metro_apropiada &lt;- crear_metropolis(log_f_dist, sigma_salto = 0.1) sims_tbl &lt;-iterador_metro_apropiada(c(theta = 0.02), 50000) g_sim &lt;- ggplot(sims_tbl %&gt;% filter(iter_num &lt; 3000), aes(x = iter_num, y = theta)) + geom_line() + ylim(c(0, 0.5)) g_sim + dist_bplot + plot_layout(widths = c(5, 1)) Donde vemos que esta cadena parece mezclar bien (está explorando la totalidad de la distribución objetivo), y también parece estar en un estado estable. Comparemos cómo saldría por ejemplo la media posterior aproximada según los tres métodos: estimaciones_media &lt;- map_dfr( list(sims_chico_tbl, sims_grande_tbl, sims_tbl), ~ filter(.x, iter_num &lt; 3000) %&gt;% summarise(media = mean(theta))) %&gt;% mutate(tipo = c(&quot;salto chico&quot;, &quot;salto grande&quot;, &quot;salto apropiado&quot;)) estimaciones_media %&gt;% bind_rows(tibble(tipo = &quot;exacta&quot;, media = 20/100)) %&gt;% select(tipo, media) ## # A tibble: 4 × 2 ## tipo media ## &lt;chr&gt; &lt;dbl&gt; ## 1 salto chico 0.128 ## 2 salto grande 0.190 ## 3 salto apropiado 0.203 ## 4 exacta 0.2 Veamos otra corrida: set.seed(6222131) sims_chica_tbl &lt;- iterador_metro_chico(c(theta = 0.02), 5000) sims_grande_tbl &lt;- iterador_metro_grande(c(theta = 0.02), 5000) estimaciones_media &lt;- map_dfr( list(sims_chica_tbl, sims_grande_tbl, sims_tbl), ~ filter(.x, iter_num &lt; 3000) %&gt;% summarise(media = mean(theta))) %&gt;% mutate(tipo = c(&quot;salto chico&quot;, &quot;salto grande&quot;, &quot;salto apropiado&quot;)) estimaciones_media %&gt;% bind_rows(tibble(tipo = &quot;exacta&quot;, media = 20/100)) %&gt;% select(tipo, media) ## # A tibble: 4 × 2 ## tipo media ## &lt;chr&gt; &lt;dbl&gt; ## 1 salto chico 0.124 ## 2 salto grande 0.229 ## 3 salto apropiado 0.203 ## 4 exacta 0.2 Repite este proceso varias veces. Verifica que: Si el tamaño de paso es muy chico, las estimaciones de la media tienen sesgo alto. Si el tamaño de paso es muy grande, las estimaciones tienen varianza alta. Si el tamaño de paso es adecuado, obtenemos buena precisión en la estimación de la media posterior. Explica estos tres casos en términos de la convergencia de las realizaciones de la cadena de Markov. Explica cómo afecta a cada caso el valor inicial de las simulaciones de Metrópolis. Repite para otra estadística, como la desviación estándar o el rangon intercuartil. Metrópolis con varios parámetros Ahora aplicaremos el algoritmo Metrópolis cuando tenemos varios parámetros. La idea es la misma, pero nuestra distribución de salto debe ser multivariada. Una selección usual es usando saltos normales independientes para cada parámetro, es decir, la normal multivariada con matriz de varianza y covarianza diagonal. Ejemplo: el modelo normal Veremos cómo simular con Metrópolis para el problema de los cantantes. Sabemos como calcular la posterior: crear_log_posterior_norm &lt;- function(x = datos, m_0, n_0, a, b){ # calcula log_posterior log_posterior &lt;- function(mu, sigma){ log_verosim &lt;- sum(dnorm(x, mu, sigma, log = TRUE)) tau &lt;- 1 / sigma^2 log_inicial &lt;- dgamma(tau, a, b, log = TRUE) + dnorm(mu, mu_0, sigma/sqrt(n_0), log = TRUE) log_p &lt;- log_verosim + log_inicial log_p } log_posterior } # parametros de inicial y datos a &lt;- 3 b &lt;- 140 mu_0 &lt;- 175 n_0 &lt;- 5 set.seed(3413) cantantes &lt;- lattice::singer %&gt;% mutate(estatura_cm = round(2.54 * height)) %&gt;% filter(str_detect(voice.part, &quot;Tenor&quot;)) %&gt;% sample_n(20) Vemos cómo se ven las primeras iteraciones de nuestra cadena de Markov: log_p &lt;- crear_log_posterior_norm(cantantes$estatura_cm, mu_0, n_0, a, b) log_post &lt;- function(pars) { log_p(pars[1], pars[2]) } set.seed(823) metro_normal &lt;- crear_metropolis(log_post, sigma_salto = 0.5) sim_tbl &lt;- metro_normal(c(mu = 172, sigma = 3), 50000) ggplot(sim_tbl %&gt;% filter(iter_num &lt; 100), aes(x = mu, y = sigma)) + geom_path() + geom_point() Y ahora vemos todas las simulaciones: g_normal &lt;- ggplot(sim_tbl, aes(x = mu, y = sigma)) + geom_point(alpha = 0.05)+ coord_equal() + ylim(c(0, 14)) g_normal Y las medias posteriores son: sim_tbl %&gt;% summarise(across(is_double, mean)) ## # A tibble: 1 × 2 ## mu sigma ## &lt;dbl&gt; &lt;dbl&gt; ## 1 176. 6.80 Ejemplo: observaciones normales, no conjugado Arriba repetimos el análisis conjugado usando Metrópolis. Aunque ya no es necesario usar el modelo conjugado, y podemos poner iniciales que sean más intuitivas y acorde con nuestro conocimiento existente. Por ejemplo, podemos poner \\(p(\\mu, \\sigma) = p(\\mu)p(\\sigma)\\), donde la densidad de \\(\\mu \\sim \\mathsf{N}(175, 2)\\) y \\(\\sigma \\sim \\mathsf{U}[2, 20].\\) Igual que antes, la verosimilitud \\(p(x|\\mu, \\sigma)\\) es normal con media \\(\\mu\\) y desviación estándar \\(\\sigma.\\) Escribimos la posterior: crear_log_posterior &lt;- function(x, m_0, sigma_0, inf, sup){ # calcula log_posterior log_posterior &lt;- function(mu, sigma){ log_verosim &lt;- sum(dnorm(x, mu, sigma, log = TRUE)) log_inicial &lt;- dunif(sigma, inf, sup, log = TRUE) + dnorm(mu, mu_0, sigma_0, log = TRUE) log_p &lt;- log_verosim + log_inicial log_p } log_posterior } log_p &lt;- crear_log_posterior(cantantes$estatura_cm, 175, 3, 2, 20) log_post &lt;- function(pars) { log_p(pars[1], pars[2]) } set.seed(8231) metro_normal &lt;- crear_metropolis(log_post, sigma_salto = 0.5) sim_tbl &lt;- metro_normal(c(mu = 172, sigma = 5), 50000) g_normal_2 &lt;- ggplot(sim_tbl, aes(x = mu, y = sigma)) + geom_point(alpha = 0.05) + coord_equal() + ylim(c(0, 14)) g_normal + g_normal_2 Los resultados son similares, pero en nuestras estimaciones bajo el segundo modelo, la \\(\\sigma\\) está concentrada en valores un poco más bajos que el modelo normal-gamma inversa. Las medias posteriores son: sim_tbl %&gt;% summarise(across(is.numeric, mean)) ## # A tibble: 1 × 3 ## iter_num mu sigma ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 25000. 176. 6.54 Nótese que la inicial para el modelo normal-gamma inversa pone muy poca probabilidad para valores bajos de \\(\\sigma\\), mientras que el segundo modelo hay un 10% de probabilidad de que la \\(\\sigma\\) sea menor que 4. tau &lt;- rgamma(5000, 3, 150) sigma &lt;- 1/sqrt(tau) quantile(sigma, c(0.01,0.1, 0.9, 0.99)) ## 1% 10% 90% 99% ## 4.219278 5.276228 11.579358 19.038529 quantile(runif(5000, 2, 25), c(0.01,0.1, 0.9, 0.99)) ## 1% 10% 90% 99% ## 2.261297 4.254128 22.691760 24.719630 Ejemplo: exámenes Recordamos un ejemplo que vimos en la sección de máxima verosimilitud. Supongamos que en una población de estudiantes tenemos dos tipos: unos llenaron un examen de opción múltiple al azar (1 de 5), y otros contestaron las preguntas intentando sacar una buena calificación. Suponemos que una vez que conocemos el tipo de estudiante, todas las preguntas tienen la misma probabilidad de ser contestadas correctamente, de manera independiente. El modelo teórico está representado por la siguiente simulación: sim_formas &lt;- function(p_azar, p_corr){ tipo &lt;- rbinom(1, 1, 1 - p_azar) if(tipo==0){ # al azar x &lt;- rbinom(1, 10, 1/5) } else { # no al azar x &lt;- rbinom(1, 10, p_corr) } x } Y una muestra se ve como sigue: set.seed(12) muestra &lt;- map_dbl(1:200, ~ sim_formas(0.35, 0.5)) qplot(muestra) Supongamos que no conocemos la probabildad de contestar correctamente ni la proporción de estudiantes que contestó al azar. ¿Como estimamos estas dos cantidades? La verosimilitud la escribimos en el ejercicio anterior en la sección de máxima verosimilitud, está dada, para las repuestas de un estudiante, por: \\[p(X = k|\\theta_{azar}, \\theta_{corr}) \\propto \\theta_{azar}(1/5)^k(4/5)^{10-k} + (1-\\theta_{azar})\\theta_{corr}^k(1-\\theta_{corr})^{10-k}\\] Suponiendo que todas las preguntas tienen la misma dificultad, y que los estudiantes que estudiaron son homogéneos (podemos discutir qué haríamos para introducir heterogeneidad que típicamente observaríamos). Creemos que la mayoría de los estudiantes no contesta al azar, así que pondremos como inicial \\[\\theta_{azar} \\sim \\mathsf{Beta}(1, 5)\\] qbeta(c(0.1, 0.9), 1, 5) %&gt;% round(2) ## [1] 0.02 0.37 Ahora tenemos que pensar en la probabilidad \\(\\theta_{corr}\\) para los estudiantes que sí estudiaron. Imaginemos que lo probamos con un estudiante que sabemos que sí estudió, y obtuvo un porcentaje de correctos de 7/10, Podríamos poner entonces (vimos 10 intentos, con 3 fracasos y 7 éxitos): \\[\\theta_{corr} \\sim \\mathsf{Beta}(7, 3)\\] Finalmente, necesitamos la conjunta inicial. Pondremos \\[p(\\theta_{azar},\\theta_{corr}) = p(\\theta_{azar})p(\\theta_{corr})\\] con lo que expresamos que inicialmente no creemos que estos dos parámetros estén relacionados. Si pensáramos, por ejemplo, que cuando hacemos exámenes difíciles menos estudiantes estudian, entonces deberíamos intentar otra conjunta. Escribimos el producto de la verosimilitud con la inicial: crear_log_posterior &lt;- function(x){ log_posterior &lt;- function(theta_azar, theta_corr){ log_verosim &lt;- sum(log(theta_azar * dbinom(x, 10, 1/5) + (1 - theta_azar) * dbinom(x, 10, theta_corr))) log_inicial &lt;- dbeta(theta_azar, 1, 5, log = TRUE) + dbeta(theta_corr, 7, 3, log = TRUE) log_post &lt;- log_verosim + log_inicial log_post } log_posterior } Creamos la función de verosimilitud con los datos log_p &lt;- crear_log_posterior(muestra) log_post &lt;- function(pars) { log_p(pars[1], pars[2]) } set.seed(8231) metro_examenes &lt;- crear_metropolis(log_post, sigma_salto = 0.02) sim_tbl &lt;- metro_examenes(c(theta_azar = 0.5, theta_corr = 0.5), 20000) g_1 &lt;- ggplot(sim_tbl, aes(x = theta_azar, y = theta_corr)) + geom_point(alpha = 0.05) + coord_equal() g_1 Nótese que hay cierta correlación entre las dos proporciones, y esto produce intervalos posteriores relativamente amplios. Esto es de esperarse, pues los datos son consistentes con una proporción relativamente chica de estudiantes que contestan al azar, y tasas de correctos más altas entre los que sí estudian, y una proporción más grande de respuestas al azar con tasas de correctos más altas. f &lt;- c(0.05, 0.5, 0.95) sim_tbl %&gt;% pivot_longer(-iter_num, names_to = &quot;parametro&quot;, values_to = &quot;valor&quot;) %&gt;% group_by(parametro) %&gt;% summarise(cuantil = quantile(valor, f), f = f) %&gt;% mutate(cuantil = round(cuantil, 2)) %&gt;% pivot_wider(names_from = f, values_from = cuantil) ## # A tibble: 2 × 4 ## # Groups: parametro [2] ## parametro `0.05` `0.5` `0.95` ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 theta_azar 0.3 0.38 0.45 ## 2 theta_corr 0.5 0.52 0.56 Muestreador de Gibbs El algoritmo de Metrópolis es muy general y se puede aplicar a una gran variedad de problemas. Sin embargo, afinar los parámetros de la distribución propuesta para que el algoritmo funcione correctamente puede ser complicado. El muestredor de Gibbs no necesita de una distribución propuesta y por lo tanto no requiere afinar estos parámetros. Para implementar un muestreador de Gibbs se necesita ser capaz de generar muestras de la distribución posterior condicional a cada uno de los parámetros individuales. Esto es, el muestreador de Gibbs permite generar muestras de la posterior: \\[p(\\theta_1,...,\\theta_p|x)\\] siempre y cuando podamos generar valores de todas las distribuciones condicionales: \\[\\theta_k \\sim p(\\theta_k|\\theta_1,...,\\theta_{k-1},\\theta_{k+1},...,\\theta_p,x).\\] El proceso del muestreador de Gibbs es una caminata aleatoria a lo largo del espacio de parámetros. La caminata inicia en un punto arbitrario y en cada tiempo el siguiente paso depende únicamente de la posición actual. Por tanto el muestredor de Gibbs es un proceso cadena de Markov vía Monte Carlo. La diferencia entre Gibbs y Metrópolis radica en como se deciden los pasos. Muestreador Gibbs En cada punto de la caminata se selecciona uno de los componentes del vector de parámetros (típicamente se cicla en orden): Supongamos que se selecciona el parámetro \\(k\\)-ésimo después de haber modificado los \\(k-1\\) anteriores, entonces obtenemos un nuevo valor para este parámetro generando una simulación de la distribución condicional \\[\\theta_k^{(i+1)} \\sim p(\\theta_k|\\theta_1^{(i+1)},\\ldots,\\theta_{k-1}^{(i+1)},\\theta_{k+1}^{(i)},\\ldots,\\theta_p^{(i)},x)\\] El nuevo valor \\(\\theta_k^{(i+1)}\\) junto con los valores \\(\\theta_1^{(i+1)},\\ldots,\\theta_{k-1}^{(i+1)},\\theta_{k+1}^{(i)},\\ldots,\\theta_p^{(i)}\\) constituyen la nueva posición en la caminata aleatoria. Seleccionamos una nueva componente \\(\\theta_{k+1}^{(i+1)}\\) y repetimos el proceso. El muestreador de Gibbs es útil cuando no podemos determinar de manera analítica la distribución conjunta y no se puede simular directamente de ella, pero sí podemos determinar todas las distribuciones condicionales y simular de ellas. Ejemplo: dos proporciones Supongamos que queremos evaluar el balanceo de dos dados de 20 lados que produce una fábrica. En particular, evaluar la probabilidad de tirar un 20, y quizá escoger el dado que nos de mayor probabilidad de tirar un 20. Tiramos cada dado \\(n\\) veces, y denotamos por \\(X_1\\) y \\(X_2\\) el número de 20’s que tiramos en cada ocasión. El modelo de datos está dado por \\[p(x_1, x_2|\\theta_1, \\theta_2)\\propto \\theta_1^{x_1}(1-\\theta_1)^{n - x_1}\\theta_2^{x_2}(1-\\theta_2)^{n - x_2},\\] que es el producto de dos densidades binomiales, pues suponemos que las tiradas son independientes cuando conocemos los parámetros \\(\\theta_1\\) y \\(\\theta_2\\). Ahora ponemos una inicial \\[p(\\theta_i)\\sim \\mathsf{Beta}(100, 1900)\\] y aquí están las razones de nuestra elección: media &lt;- 1/20 k &lt;- 2000 a &lt;- media * k b &lt;- (1 - media) * k c(a,b) ## [1] 100 1900 qbeta(c(0.05, 0.95), a, b) %&gt;% round(3) ## [1] 0.042 0.058 y suponemos que \\[p(\\theta_1,\\theta_2) = p (\\theta_1)p(\\theta_2)\\] es decir, apriori saber el desempeño de un dado no nos da información adicional del otro (esto podría no ser cierto, por ejemplo, si el defecto es provocado por la impresión del número 20). Por lo tanto, la posterior es \\[p(\\theta_1,\\theta_2|x_1, x_2)\\propto \\theta_1^{x_1+100-1}(1-\\theta_1)^{n - x_1 + 1900-1}\\theta_2^{x_2+100 -1}(1-\\theta_2)^{n - x_2 + 1900-1}\\] Ahora consideramoso qué pasa cuando conocemos \\(\\theta_2\\) y los datos. Pensamos en todo lo que no sea \\(\\theta_1\\) como constante de modo que nos queda: \\[p(\\theta_1 | \\theta_2, x) \\propto \\theta_1^{x_1+100 -1}(1-\\theta_1)^{n - x_1 + 1900 -1}\\] que es \\(\\mathsf{Beta}(x_1 + 100, n - x_1 + 1900)\\), y por la misma razón, \\[p(\\theta_2 | \\theta_1, x) \\propto \\theta_2^{x_2+100-1}(1-\\theta_2)^{n - x_2 + 1900-1}\\] que también es es \\(\\mathsf{Beta}(x_1 + 100, n - x_1 + 1900)\\) De hecho, estas condicionales son fáciles de deducir de otra manera: en realidad estamos haciendo dos experimentos separados (pues suponemos que las iniciales son independientes y las pruebas también), así que podriamos usar el análisis Beta-Binomial para cada uno de ellos. En realidad no es necesario usar MCMC para este ejemplo. Usaremos esta función para hacer nuestras iteraciones de Gibbs: iterar_gibbs &lt;- function(pasos, n, x_1, x_2){ iteraciones &lt;- matrix(0, nrow = pasos + 1, ncol = 2) # vector guardará las simulaciones iteraciones[1, 1] &lt;- 0.5 # valor inicial media colnames(iteraciones) &lt;- c(&quot;theta_1&quot;, &quot;theta_2&quot;) # Generamos la caminata aleatoria for (j in seq(2, pasos, 2)) { # theta_1 a &lt;- x_2 + 100 - 1 b &lt;- n - x_2 + 1900 - 1 iteraciones[j, &quot;theta_2&quot;] &lt;- rbeta(1, a, b) # Actualizar theta_1 iteraciones[j, &quot;theta_1&quot;] &lt;- iteraciones[j-1, &quot;theta_1&quot;] # theta_2 a &lt;- x_1 + 100 - 1 b &lt;- n - x_1 + 1900 - 1 iteraciones[j + 1, &quot;theta_1&quot;] &lt;- rbeta(1, a, b) # Actualizar theta_1 iteraciones[j + 1, &quot;theta_2&quot;] &lt;- iteraciones[j, &quot;theta_2&quot;] } iteraciones } Y supongamos que estamos comparando los dados de dos compañías: Chessex y GameScience. Tiramos cada dado 10 mil veces, y obtenemos: # Datos de https://www.awesomedice.com/blogs/news/d20-dice-randomness-test-chessex-vs-gamescience n &lt;- 10000 x_1 &lt;- 408 # Chessex, alrededor de 0.85 dólares por dado x_2 &lt;- 474 # GameScience, alrededor 1.60 dólares por dado E iteramos: iteraciones &lt;- iterar_gibbs(20000, n, x_1, x_2) %&gt;% as_tibble() %&gt;% mutate(iter_num = row_number()) head(iteraciones) ## # A tibble: 6 × 3 ## theta_1 theta_2 iter_num ## &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; ## 1 0.5 0 1 ## 2 0.5 0.0479 2 ## 3 0.0442 0.0479 3 ## 4 0.0442 0.0452 4 ## 5 0.0411 0.0452 5 ## 6 0.0411 0.0505 6 ggplot(filter(iteraciones, iter_num &gt; 1000, iter_num&lt; 1050), aes(x = theta_1, y = theta_2)) + geom_path(alpha = 0.3) + geom_point() g_1 &lt;- ggplot(iteraciones, aes(x = theta_1, y = theta_2)) + geom_path(alpha = 0.3) + geom_point() g_2 &lt;- ggplot(iteraciones %&gt;% filter(iter_num &gt; 10), aes(x = theta_1, y = theta_2)) + geom_path(alpha = 0.3) + geom_point() + geom_abline(colour = &quot;red&quot;) + geom_point(data= tibble(theta_1=1/20, theta_2=1/20), colour = &quot;red&quot;, size = 5) g_1 + g_2 Notamos el dado de Cheesex no es consistente con 1/20 de tiros de 20s, pero el dado de GameScience sí lo es. De este gráfica vemos que Cheesex está sesgado hacia abajo, así que deberíamos escoger el dado de GameScience Podemos ver directamente cómo se distribuye la diferencia \\(\\theta_1 - \\theta_2\\). Cualquier estadística es fácil de evaluar, pues simplemente la calculamos para cada simulación y después resumimos: iteraciones &lt;- iteraciones %&gt;% mutate(dif = theta_1 - theta_2) ggplot(iteraciones %&gt;% filter(iter_num &gt; 10), aes(x = dif)) + geom_histogram(bins = 100) + geom_vline(xintercept = 0, colour = &quot;red&quot;) Y vemos que es altamente probable que el dado de Cheesex produce más 20’s que el dado de GameScience. iteraciones %&gt;% mutate(theta_1_mayor = dif &gt; 0) %&gt;% summarise(prob_theta_1_mayor = mean(theta_1_mayor)) ## # A tibble: 1 × 1 ## prob_theta_1_mayor ## &lt;dbl&gt; ## 1 0.0215 Finalmente, verificamos nuestro modelo y cuánto aprendimos. Podemos hacerlo simulando de la inicial y comparando con la posterior: inicial_tbl &lt;- tibble(theta_1 = rbeta(20000, 100, 1900), theta_2 = rbeta(20000, 100, 1900), dist = &quot;inicial&quot;) posterior_tbl &lt;- iteraciones %&gt;% filter(iter_num &gt; 10) %&gt;% mutate(dist = &quot;posterior&quot;) sims_tbl &lt;- bind_rows(inicial_tbl, posterior_tbl) ggplot(sims_tbl, aes(x = theta_1, y = theta_2, colour = dist)) + geom_point(alpha = 0.2) donde vemos que el resultado que obtuvimos es razonablemente consistente con nuestra información inicial, y las 10 mil tiradas de dado fueron altamente informativas. ¿Qué crees que pasaría si sólo hubieramos tirado 40 veces cada dado? ¿Qué tanto habríamos aprendido? Puedes usar datos simulados y repetir este ejercicio. Puedes examinar los resultados para cada cara con los datos originales. Un modelo apropiado es el Dirichlet-Multinomial. Ejemplo: Modelo normal no conjugado Retomemos el caso de observaciones normales, supongamos que tenemos una muestra \\(X_1,...,X_n\\) de observaciones independientes e identicamente distribuidas, con \\(X_i \\sim \\mathsf{N}(\\mu, \\sigma^2)\\). Usaremos iniciales distintas al modelo anterior: \\[p(\\mu, \\sigma^2) = p(\\sigma^2)p(\\mu)\\] con \\(\\mu\\) \\(\\mathsf{N}(\\mu_0, \\sigma_0)\\) y \\(\\tau = 1/\\sigma^2\\) con distribución \\(\\mathsf{Gamma}(a,b)\\). Esto no nos da el modelo conjugado que vimos antes (nota la diferencia de la especificación de la inicial conjunta). Comenzamos por escribir \\[p(\\mu, \\sigma^2|x) \\propto \\frac{1}{{\\sigma^{n/2}}} \\exp(-\\sum\\frac{(x_i-\\mu)²}{2\\sigma^2}) \\exp(- \\frac{(\\mu - \\mu_0)^2}{2\\sigma_0^2}) \\frac{1}{(\\sigma^2)^{a + 1}}\\exp (-\\beta/\\sigma^2 )\\] Comenzamos analizando \\(p(\\mu|\\sigma^2, x)\\). Por la ecuación de arriba, e ignorando los términos que no dependen de \\(\\mu\\): \\[p(\\mu|\\sigma^2, x) \\propto \\exp [ - \\sum_i (\\frac{(\\mu - x_i)^2}{2\\sigma^2} - \\frac{(\\mu - \\mu_0)^2}{2n\\sigma_0^2})]\\] que es una distribución normal (completa cuadrados): \\[\\mu|\\sigma^2,x \\sim \\mathsf{N}\\bigg(\\frac{\\sigma^2}{\\sigma^2 + n\\sigma_0^2}\\mu_0 + \\frac{n\\sigma_0^2}{\\sigma^2 + n \\sigma_0^2}\\bar{x}, \\frac{\\sigma \\sigma_0}{\\sqrt{\\sigma^2 + n\\sigma_0^2}}\\bigg)\\] Ahora consideramos \\(p(\\sigma^2|mu,x)\\). Ignoramos en \\(p(\\mu,\\sigma^2|x)\\) los términos que *no** dependen de \\(\\sigma^2\\): \\[p(\\sigma^2|\\mu, x) \\propto \\frac{1}{\\sigma^{n/2}} \\exp(-\\sum\\frac{(x_i-\\mu)²}{2\\sigma^2}) \\frac{1}{(\\sigma^2)^{a + 1}}\\exp (-\\beta/\\sigma^2)\\] que simplificando da \\[ = \\frac{1}{\\sigma^{n/2 + a + 1}}\\exp( -\\frac{\\beta +\\frac{1}{2}\\sum(x_i - \\mu)^2}{\\sigma^2} )\\] de modo que \\[\\sigma^2|\\mu, x \\sim \\mathsf{GI}(a +n/2, b + \\frac{1}{2}\\sum(x_i -\\mu)^2)\\] Ejemplo Usaremos este muestreador para el problema de la estaturas de los tenores. Comenzamos definiendo las distribuciones iniciales: \\(\\mu \\sim \\mathsf{N}(175, 3)\\) \\(\\tau = 1/\\sigma^2 \\sim \\mathsf{GI}(3, 150)\\), esto es \\(a = 3\\) y \\(b = 150\\). Escribimos el muestreador de Gibbs. n &lt;- 20 x &lt;- cantantes$estatura_cm m &lt;- 175; sigma_0 &lt;- 3; alpha &lt;- 3; beta &lt;- 150 # parámetros de iniciales pasos &lt;- 20000 iteraciones &lt;- matrix(0, nrow = pasos + 1, ncol = 2) # vector guardará las simulaciones iteraciones[1, 1] &lt;- 0 # valor inicial media colnames(iteraciones) &lt;- c(&quot;mu&quot;, &quot;sigma&quot;) # Generamos la caminata aleatoria for (j in seq(2, pasos, 2)) { # sigma^2 mu &lt;- iteraciones[j - 1, &quot;mu&quot;] a &lt;- n / 2 + alpha b &lt;- sum((x - mu) ^ 2) / 2 + beta iteraciones[j, &quot;sigma&quot;] &lt;- sqrt(1/rgamma(1, a, b)) # Actualizar sigma iteraciones[j, &quot;mu&quot;] &lt;- iteraciones[j-1, &quot;mu&quot;] # mu sigma &lt;- iteraciones[j, &quot;sigma&quot;] media &lt;- (n * sigma_0^2 * mean(x) + sigma^2 * m) / (n * sigma_0^2 + sigma^2) varianza &lt;- sigma^2 * sigma_0^2 / (n * sigma_0^2 + sigma^2) iteraciones[j+1, &quot;mu&quot;] &lt;- rnorm(1, media, sd = sqrt(varianza)) # actualizar mu iteraciones[j+1, &quot;sigma&quot;] &lt;- iteraciones[j, &quot;sigma&quot;] } caminata &lt;- data.frame(pasos = 1:pasos, mu = iteraciones[1:pasos, &quot;mu&quot;], sigma = iteraciones[1:pasos, &quot;sigma&quot;]) caminata_g &lt;- caminata %&gt;% gather(parametro, val, mu, sigma) %&gt;% arrange(pasos) Veamos primero algunos pasos: ggplot(filter(caminata, pasos &gt; 1000, pasos&lt; 1010), aes(x = mu, y = sigma)) + geom_path(alpha = 0.3) + geom_point() Donde vemos cómo en cada iteración se actualiza un solo parámetro. Una alternativa es conservar únicamente ciclos completos de la caminata u esto es lo que hacen varios programas que implementan Gibbs, sin embargo ambas cadenas (cadenas completas y conservando únicamente ciclos completos) convergen a la misma distribución posterior. Si tomamos iteraciones completas: ggplot(filter(caminata, pasos &gt; 1000, pasos&lt; 1020, pasos %% 2 == 0), aes(x = mu, y = sigma)) + geom_path(alpha = 0.3) + geom_point() Y ahora vemos cómo se ven las simulaciones: ggplot(filter(caminata, pasos &gt; 1000, pasos&lt; 10000, pasos %% 2 == 0), aes(x = mu, y = sigma)) + geom_point(alpha = 0.1) Y el diagnóstico de cada cadena: ggplot(filter(caminata_g, pasos &gt; 15000), aes(x = pasos, y = val)) + geom_path(alpha = 0.3) + facet_wrap(~parametro, ncol = 1, scales = &quot;free&quot;) + scale_y_continuous(&quot;&quot;) Estas cadenas parecen estar mezclando bien. Podemos resumirlas: ggplot(filter(caminata_g, pasos &gt; 5000), aes(x = val)) + geom_histogram(fill = &quot;gray&quot;) + facet_wrap(~parametro, ncol = 1, scales = &quot;free&quot;) caminata_g %&gt;% filter(pasos &gt; 1000) %&gt;% # eliminamos la etapa de calentamiento group_by(parametro) %&gt;% summarise( mean(val), sd(val), median(val) ) %&gt;% mutate(across(is_double, round, 2)) ## # A tibble: 2 × 4 ## parametro `mean(val)` `sd(val)` `median(val)` ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 mu 176. 1.32 176. ## 2 sigma 6.54 0.95 6.44 Y obtenemos un resultado similar a los anteriores. Conclusiones y observaciones Metrópolis y Gibbs Una generalización del algoritmo de Metrópolis es Metrópolis-Hastings. El algoritmo de Metrópolis es como sigue: Generamos un punto inicial tal que \\(p(\\theta)&gt;0\\). Para \\(i = 1,2,...\\) Se propone un nuevo valor \\(\\theta^*\\) con una distribución propuesta \\(g(\\theta^*|\\theta^{(i)})\\) es común que \\(g(\\theta^*|\\theta^{(i)})\\) sea una normal centrada en \\(\\theta^{(i)}\\). Calculamos la probabilidad de aceptación \\[\\alpha=\\min\\bigg\\{\\frac{p(\\theta^*)}{p(\\theta^{(i)})},1\\bigg\\},\\] y aceptamos \\(\\theta^*\\) con probabilidad \\(p_{mover}\\). Es así que el algorito requiere que podamos calcular el cociente en \\(p_{mover}\\) para todo \\(\\theta^{(i)}\\) y \\(\\theta^*\\), así como simular de la distribución propuesta \\(g(\\theta^*|\\theta^{(i)})\\), adicionalmente debemos poder generar valores uniformes para decidir si aceptar/rechazar. En el caso de Metrópolis un requerimiento adicional es que la distribución propuesta \\(g(\\theta_{a}|\\theta_b)\\) debe ser simétrica, es decir \\(g(\\theta_{a}|\\theta_b) = g(\\theta_{b}|\\theta_a)\\) para todo \\(\\theta_{a}\\), \\(\\theta_{b}\\). Metrópolis-Hastings generaliza Metrópolis, eliminando la restricción de simetría en la distribución propuesta \\(g(\\theta_{a}|\\theta_b)\\), sin embargo para corregir por esta asimetría debemos calcular \\(\\alpha\\) como sigue: \\[\\alpha=\\min\\bigg\\{ \\frac{p(\\theta^*)}{g(\\theta^*|\\theta^{(i)})} \\cdot \\frac{g(\\theta^{(i)}|\\theta^*)}{p(\\theta^{(i)})},1\\bigg\\}\\] La generalización de Metrópolis-Hastings puede resultar en algoritmos más veloces. Se puede ver Gibbs como una generalización de Metrópolis-Hastings, cuando estamos actualizando un componente de los parámetros, la distribución propuesta es la distribución posterior para ese parámetro, por tanto siempre es aceptado. Comparado con Metrópolis, Gibbs tiene la ventaja de que no se necesita afinar los parámetros de una distribución propuesta (o seleccionar siquiera una distribución propuesta). Además que no hay pérdida de simulaciones debido a rechazo. Por su parte, la desventaja debemos conocer las distribuciones condicionales y poder simular de ellas. En el caso de modelos complicados se utilizan combinaciones de Gibbs y Metrópolis. Cuando se consideran estos dos algoritmos Gibbs es un método más simple y es la primera opción para modelos condicionalmente conjugados. Sí solo podemos simular de un subconjunto de las distribuciones condicionales posteriores, entonces podemos usar Gibbs siempre que se pueda y Metrópolis unidimensional para el resto, o de manera más general separamos en bloques, un bloque se actualiza con Gibbs y otro con Metrópolis. El algoritmo de Gibbs puede atorarse cuando hay correlación alta entre los parámetros, reparametrizar puede ayudar, o se pueden usar otros algoritmos. JAGS (Just Another Gibbs Sampler), WinBUGS y OpenBUGS son programas que implementan métodos MCMC para generar simulaciones de distribuciones posteriores. Los paquetes rjags y R2jags permiten ajustar modelos en JAGS desde R. Es muy fácil utilizar estos programas pues uno simplemente debe especificar las distribuciones iniciales, la verosimilitud y los datos observados. Para aprender a usar JAGS se puede revisar la sección correspondiente en las notas de 2018, ahora nos concentraremos en el uso de Stan. HMC y Stan It appears to be quite a general principle that, whenever there is a randomized way of doing something, then there is a nonrandomized way that delivers better performance but requires more thought. -E.T. Jaynes Stan es un programa para generar muestras de una distribución posterior de los parámetros de un modelo, el nombre del programa hace referencia a Stanislaw Ulam (1904-1984) que fue pionero en los métodos de Monte Carlo. A diferencia de JAGS y BUGS, los pasos de la cadena de Markov se generan con un método llamado Monte Carlo Hamiltoniano (HMC). HMC es computacionalmente más costoso que Metrópolis o Gibbs, sin embargo, sus propuestas suelen ser más eficientes, y por consiguiente no necesita muestras tan grandes. En particular cuando se ajustan modelos grandes y complejos (por ejemplo, con variables con correlación alta) HMC supera a otros. Diagnósticos generales para MCMC Cuando generamos una muestra de la distribución posterior usando MCMC, sin importar el método (Metrópolis, Gibbs, HMC), buscamos que: Los valores simulados sean representativos de la distribución posterior. Esto implica que no deben estar influenciados por el valor inicial (arbitrario) y deben explorar todo el rango de la posterior, con suficientes retornos para evaluar cuánta masa hay en cada región. Debemos tener suficientes simulaciones de tal manera que las estimaciones sean precisas y estables. Queremos tener un método eficiente para generar las simulaciones. En la práctica intentamos cumplir lo más posible estos objetivos, pues aunque en principio los métodos MCMC garantizan que una cadena infinitamente larga logrará una representación perfecta, siempre debemos tener un criterio para cortar la cadena y evaluar la calidad de las simulaciones. Representatividad Burn-in e iteraciones iniciales- En primer lugar, en muchas ocasiones las condiciones iniciales de las cadenas están en partes del espacio de parámetros que son “atípicos” en términos de la posterior. Así que es común quitar algunas observaciones iniciales (iteraciones de burn-in) para minimizar su efecto en resúmenes posteriores. Por ejemplo, para el ejemplo de los cantantes, podemos ver que las iteraciones iniciales tienen como función principal llegar a las regiones de probabilidad posterior alta: log_p &lt;- crear_log_posterior_norm(cantantes$estatura_cm, mu_0, n_0, a, b) log_post &lt;- function(pars) { log_p(pars[1], pars[2]) } set.seed(823) metro_normal &lt;- crear_metropolis(log_post, sigma_salto = 0.5) sim_tbl &lt;- metro_normal(c(mu = 162, sigma = 1), 5000) ggplot(sim_tbl %&gt;% filter(iter_num &lt; 500), aes(x = mu, y = sigma)) + geom_path(alpha = 0.5) + geom_point(aes(colour = iter_num)) De modo que puede ser buena idea eliminar las primeras iteraciones. En teoría, no es necesario hacer esto si hacemos suficientes iteraciones, pues la cadena va a terminar en su estado estable explorando la posterior. En la práctica, y con pocas iteraciones, puede ayudar un poco a mejorar la precisión numérica de las cantidades que queramos calcular. sim_g &lt;- sim_tbl %&gt;% pivot_longer(-iter_num, names_to = &quot;parametro&quot;, values_to = &quot;valor&quot;) todas &lt;- ggplot(sim_g, aes(x = iter_num, y = valor)) + geom_line(alpha = 0.5) + facet_wrap(~ parametro, ncol = 1, scales = &quot;free_y&quot;) + labs(subtitle = &quot;Todas las simulaciones&quot;) sin_burnin &lt;- sim_g %&gt;% filter(iter_num &gt; 200) %&gt;% ggplot(aes(x = iter_num, y = valor)) + geom_line(alpha = 0.5) + facet_wrap(~ parametro, ncol = 1, scales = &quot;free_y&quot;) + labs(subtitle = &quot;Quitando 200 de burn-in&quot;) todas + sin_burnin Convergencia a estado límite. Para determinar la convergencia es conveniente realizar más de una cadena: buscamos ver si realmente se ha olvidado el estado inicial, si las distribuciones de cada cadena son consistentes unas con otras, y revisar que algunas cadenas no hayan quedado atoradas en regiones inusuales del espacio de parámetros. Inicializamos las cadenas con valores al azar en rangos razonables (por ejemplo simulando de la inicial): set.seed(8513) valores_iniciales &lt;- tibble(mu_0 = rnorm(4, 160, 20), sigma_0 = runif(4, 0, 20), cadena = 1:4) sims_tbl &lt;- valores_iniciales %&gt;% mutate(sims = map2(mu_0, sigma_0, ~ metro_normal(c(mu = .x, sigma = .y), 300) )) %&gt;% unnest(sims) ggplot(sims_tbl, aes(x = iter_num, y = sigma, colour = factor(cadena))) + geom_line() Y este es un ejemplo donde claramente las cadenas no han alcanzado un estado estable: tienen muy distintas medias y varianzas. Por ejemplo: set.seed(83243) sims_tbl &lt;- valores_iniciales %&gt;% mutate(sims = map2(mu_0, sigma_0, ~ metro_normal(c(mu = .x, sigma = .y), 20000) )) %&gt;% unnest(sims) ggplot(sims_tbl, aes(x = iter_num, y = sigma, colour = factor(cadena))) + geom_line() Y este resultado se ve mejor. La parte transición hacia las zonas de alta probabilidad pasa antes de unas 1000 iteraciones. Podemos hacer más simulaciones, o eliminar como burn-in las primiras iteraciones: media_g &lt;- ggplot(sims_tbl %&gt;% filter(iter_num &gt; 2000), aes(x = iter_num, y = mu, colour = factor(cadena))) + geom_line() sigma_g &lt;- ggplot(sims_tbl %&gt;% filter(iter_num &gt; 2000), aes(x = iter_num, y = sigma, colour = factor(cadena))) + geom_line() media_g / sigma_g Las gráficas anteriores nos ayudan a determinar si elegimos un periodo de calentamiento adecuado o si alguna cadena está alejada del resto. Una vez que las cadenas están en estado estable, podemos usar todas las simulaciones juntas para resumir: head(sims_tbl) ## # A tibble: 6 × 6 ## mu_0 sigma_0 cadena iter_num mu sigma ## &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; &lt;int&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 155. 3.16 1 1 155. 3.16 ## 2 155. 3.16 1 2 155. 3.16 ## 3 155. 3.16 1 3 155. 3.16 ## 4 155. 3.16 1 4 155. 3.16 ## 5 155. 3.16 1 5 155. 3.50 ## 6 155. 3.16 1 6 155. 3.81 # medias posteriores sims_tbl %&gt;% summarise(mu = mean(mu), sigma = mean(sigma)) ## # A tibble: 1 × 2 ## mu sigma ## &lt;dbl&gt; &lt;dbl&gt; ## 1 176. 6.77 Además de realizar gráficas podemos usar la medida de convergencia \\(\\hat{R}\\). La medida \\(\\hat{R}\\) se conoce como el factor de reducción potencial de escala o diagnóstico de convergencia de Gelman-Rubin, esta es una estimación de la posible reducción en la longitud de un intervalo de confianza si las simulaciones continuaran infinitamente. \\(\\hat{R}\\) es aproximadamente la raíz cuadrada de la varianza de todas las cadenas juntas dividida entre la varianza dentro de cada cadena. Si \\(\\hat{R}\\) es mucho mayor a 1 esto indica que las cadenas no se han mezclado bien. Una regla usual es iterar hasta alcanzar un valor \\(\\hat{R} \\leq 1.1\\) para todos los parámetros. \\[\\hat{R} \\approx \\sqrt{\\frac{\\hat{V}}{W}}\\] donde \\(B\\) es la varianza entre las cadenas, \\(W\\) es la varianza dentro de las cadenas \\[B = \\frac{N}{M-1}\\sum_m (\\hat{\\theta}_m - \\hat{\\theta})^2\\] \\[W = \\frac{1}{M}\\sum_m \\hat{\\sigma}_m^2\\] Y \\(\\hat{V}\\) es una estimación del varianza de posterior de \\(\\theta\\): \\[\\hat{V} = \\frac{N-1}{N}W + \\frac{M+1}{MN}B\\] #### Ejemplo {-} En nuestro ejemplo anterior, tenemos sims_tbl %&gt;% pivot_longer(mu:sigma, names_to = &quot;parametro&quot;, values_to = &quot;valor&quot;) %&gt;% group_by(parametro, cadena) %&gt;% summarise(media = mean(valor), num = n(), sigma2 = var(valor)) %&gt;% summarise(N = first(num), M = n_distinct(cadena), B = N * var(media), W = mean(sigma2), V_hat = ((N - 1) / N) * W + (M + 1)/(M * N) * B, R_hat = sqrt(V_hat / W)) ## # A tibble: 2 × 7 ## parametro N M B W V_hat R_hat ## &lt;chr&gt; &lt;int&gt; &lt;int&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 mu 20000 4 1281. 4.29 4.37 1.01 ## 2 sigma 20000 4 121. 1.31 1.32 1.00 Y verificamos que los valores de \\(\\hat{R}\\) son cercanos a uno, lo cual indica que este diagnóstico es aceptable. Si hubiéramos trabajado con las primeras 300 iteraciones sims_tbl %&gt;% filter(iter_num &lt; 300) %&gt;% pivot_longer(mu:sigma, names_to = &quot;parametro&quot;, values_to = &quot;valor&quot;) %&gt;% group_by(parametro, cadena) %&gt;% summarise(media = mean(valor), num = n(), sigma2 = var(valor)) %&gt;% summarise(N = first(num), M = n_distinct(cadena), B = N * var(media), W = mean(sigma2), V_hat = ((N - 1) / N) * W + (M + 1)/(M * N) * B, R_hat = sqrt(V_hat / W)) ## # A tibble: 2 × 7 ## parametro N M B W V_hat R_hat ## &lt;chr&gt; &lt;int&gt; &lt;int&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 mu 299 4 32334. 40.4 175. 2.08 ## 2 sigma 299 4 7394. 11.9 42.8 1.89 Y estos valores indican problemas en la convergencia de las cadenas. Es necesario diagnosticar el problema, que en este caso resolvemos incrementando el número de iteraciones. Precisión Una vez que tenemos una muestra representativa de la distribución posterior, nuestro objetivo es asegurarnos de que la muestra es lo suficientemente grande para producir estimaciones estables y precisas de la distribución. Para ello usaremos el tamaño efectivo de muestra, Si las simulaciones fueran independientes \\(N_{eff}\\) sería el número total de simulaciones; sin embargo, las simulaciones de MCMC suelen estar correlacionadas, de modo que cada iteración de MCMC es menos informativa que si fueran independientes. Ejemplo: Si graficaramos simulaciones independientes, esperaríamos valores de autocorrelación chicos: acf(rgamma(1000,1,1)) Sin embargo, los valores que simulamos tienen el siguiente perfil de autocorrelación: sigma_metro_sims &lt;- sims_tbl %&gt;% filter(cadena==4) %&gt;% pull(mu) acf(sigma_metro_sims) El tamaño efectivo de muestra nos dice qué tamaño de muestra de observaciones independientes nos daría la misma información que las simulaciones de la cadena. Una manera de manera relativamente simple de estimarlo es: \\[N_{eff} = \\frac{N}{1+2\\sum_{k=1}^\\infty ACF(k)} \\] Usualmente nos gustaría obtener un tamaño efectivo de al menos \\(100\\) (para cálculo de medias y varianzas posteriores). Esta cantidad usualmente se reporta en el software (con mejores estimaciones que la de la fórmula de arriba), y es necesario checarlo. En nuestro ejemplo hacemos una aproximación como sigue: calc_acf &lt;- function(x){ valores_acf &lt;- acf(x, lag.max = 1000, plot = FALSE)$acf %&gt;% as.numeric() valores_acf[-1] } acf_tbl &lt;- sims_tbl %&gt;% pivot_longer(mu:sigma, names_to = &quot;parametro&quot;, values_to = &quot;valor&quot;) %&gt;% group_by(parametro, cadena) %&gt;% summarise(N = n_distinct(iter_num), k = 1:1000, acf = calc_acf(valor)) %&gt;% summarise(N = first(N), N_eff = N / (1 + 2 * sum(acf))) acf_tbl ## # A tibble: 8 × 4 ## # Groups: parametro [2] ## parametro cadena N N_eff ## &lt;chr&gt; &lt;int&gt; &lt;int&gt; &lt;dbl&gt; ## 1 mu 1 20000 251. ## 2 mu 2 20000 700. ## 3 mu 3 20000 104. ## 4 mu 4 20000 394. ## 5 sigma 1 20000 421. ## 6 sigma 2 20000 411. ## 7 sigma 3 20000 93.9 ## 8 sigma 4 20000 724. Nótese que algunas cadenas tienen un tamaño efectivo de muestra relativamente bajo para el número de iteraciones que hicimos. De cualquier forma, el agregado sobre todas las cadenas es suficientemente grande para calcular resúmenes básicos: acf_tbl %&gt;% group_by(parametro) %&gt;% summarise(N = sum(N), N_eff = sum(N_eff)) ## # A tibble: 2 × 3 ## parametro N N_eff ## &lt;chr&gt; &lt;int&gt; &lt;dbl&gt; ## 1 mu 80000 1450. ## 2 sigma 80000 1650. Sin embargo, podemos hacer más simulaciones si es necesario, por ejemplo para aproximar de manera apropiada percentiles en las colas. Eficiencia Hay varias maneras para mejorar la eficiencia de un proceso MCMC: Paralelizar, no disminuimos el número de pasos en las simulaciones pero podemos disminuir el tiempo que tarda en correr. Cambiar la parametrización del modelo o transformar los datos. Adelgazar la muestra cuando tenemos problemas de uso de memoria, consiste en guardar únicamente los \\(k\\)-ésimos pasos de la cadena y resulta en cadenas con menos autocorrelación . Recomendaciones generales Gelman and Hill (2006) recomienda los siguientes pasos cuando uno esta simulando de la posterior: Cuando definimos un modelo por primera vez establecemos un valor bajo para el número de iteraciones. La razón es que la mayor parte de las veces los modelos no funcionan a la primera por lo que sería pérdida de tiempo dejarlo correr mucho tiempo antes de descubrir el problema. Si las simulaciones no han alcanzado convergencia aumentamos las iteraciones a \\(500\\) ó \\(1000\\) de tal forma que las corridas tarden segundos o unos cuantos minutos. Si tarda más que unos cuantos minutos (para problemas del tamaño que veremos en la clase) y aún así no alcanza convergencia entonces juega un poco con el modelo (por ejemplo intenta transformaciones lineales), para JAGS Gelman sugiere más técnicas para acelerar la convergencia en el capitulo \\(19\\) del libro Data Analysis Using Regression and Multilevel/Hierarchical models. En el caso de Stan veremos ejemplos de reparametrización, y se puede leer más en la guía. Otra técnica conveniente cuando se trabaja con bases de datos grandes (sobre todo en la parte exploratoria) es trabajar con un subconjunto de los datos, quizá la mitad o una quinta parte. Referencias "],["apéndice-principios-de-visualizacion.html", "Apéndice: Principios de visualizacion Introducción Visualización popular de datos Teoría de visualización de datos Ejemplo: gráfica de Minard", " Apéndice: Principios de visualizacion “The simple graph has brought more information to the data analyst’s mind than any other device.” — John Tukey El cuarteto de Anscombe En 1971 un estadístico llamado Frank Anscombe (fundador del departamento de Estadística de la Universidad de Yale) publicó cuatro conjuntos de dato. Cada uno consiste de 11 observaciones. La peculariedad de estos conjuntos es que tienen las mismas propiedades estadísticas. Sin embargo, cuando analizamos los datos de manera gráfica en un histograma encontramos rápidamente que los conjuntos de datos son muy distintos. Media de \\(x\\): 9 Varianza muestral de \\(x\\): 11 Media de \\(y\\): 7.50 Varianza muestral de \\(y\\): 4.12 Correlación entre \\(x\\) y \\(y\\): 0.816 Línea de regresión lineal: \\(y = 3.00 + 0.500x\\) En la gráfica del primer conjunto de datos, se ve clara una relación lineal simple con un modelo que cumple los supuestos de normalidad. La segunda gráfica (arriba a la derecha) muestra unos datos que tienen una asociación pero definitivamente no es lineal. En la tercera gráfica (abajo a la izquierda) están puntos alineados perfectamente en una línea recta, excepto por uno de ellos. En la última gráfica podemos ver un ejemplo en el cual basta tener una observación atípica para que se produzca un coeficiente de correlación alto aún cuando en realidad no existe una asociación lineal entre las dos variables. El cuarteto de Ascombe inspiró una técnica reciente para crear datos que comparten las mismas propiedades estadísticas al igual que en el cuarteto, pero que producen gráficas muy distintas (Matejka, Fitzmaurice). Introducción La visualización de datos no trata de hacer gráficas “bonitas” o “divertidas”, ni de simplificar lo complejo o ayudar a una persona “que no entiende mucho” a entender ideas complejas. Más bien, trata de aprovechar nuestra gran capacidad de procesamiento visual para exhibir de manera clara aspectos importantes de los datos. El siguiente ejemplo de (Tufte 2006), ilustra claramente la diferencia entre estos dos enfoques. A la izquierda están gráficas (más o menos típicas de Powerpoint) basadas en la filosofía de simplificar, de intentar no “ahogar” al lector con datos. El resultado es una colección incoherente, de bajo contenido, que no tiene mucho qué decir y que es, “indeferente al contenido y la evidencia”. A la derecha está una variación del rediseño de Tufte en forma de tabla, que en este caso particular es una manera eficiente de mostrar claramente los patrones que hay en este conjunto simple de datos. ¿Qué principios son los que soportan la efectividad de esta tabla sobre la gráfica de la derecha? Veremos que hay dos conjuntos de principios importantes: unos relacionados con el diseño y otros con la naturaleza del análisis de datos, independientemente del método de visualización. Visualización popular de datos Publicaciones populares (periódicos, revistas, sitios internet) muchas veces incluyen visualización de datos como parte de sus artículos o reportajes. En general siguen el mismo patrón que en la visión tradicionalista de la estadística: sirven más para divertir que para explicar, tienden a explicar ideas simples y conjuntos chicos de datos, y se consideran como una “ayuda” para los “lectores menos sofisticados”. Casi siempre se trata de gráficas triviales (muchas veces con errores graves) que no aportan mucho a artículos que tienen un nivel de complejidad mucho mayor (es la filosofía: lo escrito para el adulto, lo graficado para el niño). Teoría de visualización de datos Existe teoría fundamentada acerca de la visualización. Después del trabajo pionero de Tukey, los principios e indicadores de Tufte se basan en un estudio de la historia de la graficación y ejercicios de muestreo de la práctica gráfica a lo largo de varias disciplinas (¿cuáles son las mejores gráficas? ¿por qué?) El trabajo de Cleveland es orientado a la práctica del análisis de datos (¿cuáles gráficas nos han ayudado a mostrar claramente los resultados del análisis?), por una parte, y a algunos estudios de percepción visual. En resumen, hablaremos de las siguientes guías: Principios generales del diseño analítico Aplicables a una presentación o análisis completos, y como guía para construir nuevas visualizaciones (Tufte 2006). Principio 1. Muestra comparaciones, contrastes, diferencias. Principio 2. Muestra causalidad, mecanismo, explicación, estructura sistemática. Principio 3. Muestra datos multivariados, es decir, más de una o dos variables. Principio 4. Integra palabras, números, imágenes y diagramas. Principio 5. Describe la totalidad de la evidencia. Muestra fuentes usadas y problemas relevantes. Principio 6. Las presentaciones analíticas, a fin de cuentas, se sostienen o caen dependiendo de la calidad, relevancia e integridad de su contenido. Técnicas de visualización Esta categoría incluye técnicas específicas que dependen de la forma de nuestros datos y el tipo de pregunta que queremos investigar (Tukey (1977), William S. Cleveland (1993), W. S. Cleveland (1994), Tufte (2006)). Tipos de gráficas: cuantiles, histogramas, caja y brazos, gráficas de dispersión, puntos/barras/ líneas, series de tiempo. Técnicas para mejorar gráficas: Transformación de datos, transparencia, vibración, banking 45, suavizamiento y bandas de confianza. Pequeños múltiplos Indicadores de calidad gráfica Aplicables a cualquier gráfica en particular. Estas son guías concretas y relativamente objetivas para evaluar la calidad de una gráfica (Tufte 1986). Integridad Gráfica. El factor de engaño, es decir, la distorsión gráfica de las cantidades representadas, debe ser mínimo. Chartjunk. Minimizar el uso de decoración gráfica que interfiera con la interpretación de los datos: 3D, rejillas, rellenos con patrones. Tinta de datos. Maximizar la proporción de tinta de datos vs. tinta total de la gráfica. For non-data- ink, less is more. For data-ink, less is a bore. Densidad de datos. Las mejores gráficas tienen mayor densidad de datos, que es la razón entre el tamaño del conjunto de datos y el área de la gráfica. Las gráficas se pueden encoger mucho. Percepción visual. Algunas tareas son más fáciles para el ojo humano que otras (W. S. Cleveland 1994). Factor de engaño y Chartjunk El factor de engaño es el cociente entre el efecto mostrado en una gráfica y el efecto correspondiente en los datos. Idealmente, el factor de engaño debe ser 1 (ninguna distorsión). El chartjunk son aquellos elementos gráficos que no corresponden a variación de datos, o que entorpecen la interpretación de una gráfica. Estos son los indicadores de calidad más fáciles de entender y aplicar, y afortunadamente cada vez son menos comunes. Un diseño popular que califica como chartjunk y además introduce factores de engaño es el pie de 3D. En la gráfica de la derecha, podemos ver como la rebanada C se ve más grande que la rebanada A, aunque claramente ese no es el caso (factor de engaño). La razón es la variación en la perspectiva que no corresponde a variación en los datos (chartjunk). Crítica gráfica: Gráfica de pie Todavía elementos que pueden mejorar la comprensión de nuestra gráfica de pie: se trata de la decodificiación que hay que hacer categoría - color - cuantificación. Podemos agregar las etiquetas como se muestra en la serie de la derecha, pero entonces: ¿por qué no mostrar simplemente la tabla de datos? ¿qué agrega el pie a la interpretación? La deficiencias en el pie se pueden ver claramente al intentar graficar más categorías (13) . En el primer pie no podemos distinguir realmente cuáles son las categorías grandes y cuáles las chicas, y es muy difícil tener una imagen mental clara de estos datos. Agregar los porcentajes ayuda, pero entonces, otra vez, preguntamos cuál es el propósito del pie. La tabla de la izquierda hace todo el trabajo (una vez que ordenamos las categrías de la más grande a la más chica). Es posible hacer una gráfica de barras como la de abajo a la izquierda. Hay otros tipos de chartjunk comunes: uno es la textura de barras, por ejemplo. El efecto es la producción de un efecto moiré que es desagradable y quita la atención de los datos, como en la gráfica de barras de abajo. Otro común son las rejillas, como mostramos en las gráficas de la izquierda. Nótese como en estos casos hay efectos ópticos no planeados que degradan la percepción de los patrones en los datos. Pequeños múltiplos y densidad gráfica La densidad de una gráfica es el tamaño del conjunto de datos que se grafica comparado con el área total de la gráfica. En el siguiente ejemplo, graficamos en logaritmo-10 de millones de cabezas de ganado en Francia (cerdos, res, ovejas y caballos). La gráfica de la izquierda es pobre en densidad pues sólo representa 4 datos. La manera más fácil de mejorar la densidad es hacer más chica la gráfica: La razón de este encogimiento es una que tiene qué ver con las oportunidades perdidas de una gráfica grande. Si repetimos este mismo patrón (misma escala, mismos tipos de ganado) para distintos países obtenemos la siguiente gráfica: Esta es una gráfica de puntos. Es útil como sustituto de una gráfica de barras, y es superior en el sentido de que una mayor proporción de la tinta que se usa es tinta de datos. Otra vez, mayor proporción de tinta de datos representa más oportunidades que se pueden capitalizar. Más pequeños múltiplos Los pequeños múltiplos presentan oportunidades para mostrar más acerca de nuestro problema de interés. Consideramos por ejemplo la relación de radiación solar y niveles de ozono: En el ejemplo anterior incluyendo una variable adicional (velocidad del viento) podemos entender más acerca de la relación de radiación solar y niveles de ozono: Tinta de datos Maximizar la proporción de tinta de datos en nuestras gráficas tiene beneficios inmediatos. La regla es: si hay tinta que no representa variación en los datos, o la eliminación de esa tinta no representa pérdidas de significado, esa tinta debe ser eliminada. El ejemplo más claro es el de las rejillas en gráficas y tablas: ¿Por qué usar grises en lugar de negros? La respuesta tiene qué ver con el principio de tinta de datos: si marcamos las diferencias sutil pero claramente, tenemos más oportunidades abiertas para hacer énfasis en lo que nos interesa: a una gráfica o tabla saturada no se le puede hacer más - es difícil agregar elementos adicionales que ayuden a la comprensión. Si comenzamos marcando con sutileza, entonces se puede hacer más. Los mapas geográficos son un buen ejemplo de este principio. El espacio en blanco es suficientemente bueno para indicar las fronteras en una tabla, y facilita la lectura: Para un ejemplo del proceso de rediseño de una tabla, ver aquí. Finalmente, podemos ver un ejemplo que intenta incorporar los elementos del diseño analítico, incluyendo pequeños múltiplos: Decoración Percepción de escala Entre la percepción visual y la interpretación de una gráfica están implícitas tareas visuales específicas que las personas debemos realizar para ver correctamente la gráfica. En la década de los ochenta, William S. Cleveland y Robert McGill realizaron algunos experimentos identificando y clasificando estas tareas para diferentes tipos de gráficos (William S. Cleveland and McGill 1984). En estos, se le pregunta a la persona que compare dos valores dentro de una gráfica, por ejemplo, en dos barras en una gráfica de barras, o dos rebanadas de una gráfica de pie. Los resultados de Cleveland y McGill fueron replicados por Heer y Bostock en 2010 y los resultados se muestran en las gráficas de la derecha: Ejemplo: gráfica de Minard Concluimos esta sección con una gráfica que, aunque poco común, ejemplifica los principios de una buena gráfica, y es reconocida como una de las mejores visualizaciones de la historia. Una gráfica excelente, presenta datos interesantes de forma bien diseñada: es una cuestión de fondo, de diseño, y estadística… [Se] compone de ideas complejas comunicadas con claridad, precisión y eficiencia. … [Es] lo que da al espectador la mayor cantidad de ideas, en el menor tiempo, con la menor cantidad de tinta, y en el espacio más pequeño. … Es casi siempre multivariado. … Una excelente gráfica debe decir la verdad acerca de los datos. (Tufte, 1983) La famosa visualización de Charles Joseph Minard de la marcha de Napoleón sobre Moscú, ilustra los principios de una buena gráfica. Tufte señala que esta imagen “bien podría ser el mejor gráfico estadístico jamás dibujado”, y sostiene que “cuenta una historia rica y coherente con sus datos multivariados, mucho más esclarecedora que un solo número que rebota en el tiempo”. Se representan seis variables: el tamaño del ejército, su ubicación en una superficie bidimensional, la dirección del movimiento del ejército y la temperatura en varias fechas durante la retirada de Moscú”. Hoy en día Minard es reconocido como uno de los principales contribuyentes a la teoría de análisis de datos y creación de infografías con un fundamento estadístico. Se grafican 6 variables: el número de tropas de Napoleón, la distancia, la temperatura, la latitud y la longitud, la dirección en que viajaban las tropas y la localización relativa a fechas específicas. La gráfica de Minard, como la describe E.J. Marey, parece “desafiar la pluma del historiador con su brutal elocuencia”, la combinación de datos del mapa, y la serie de tiempo, dibujados en 1869, “retratan una secuencia de pérdidas devastadoras que sufrieron las tropas de Napoleón en 1812”. Comienza en la izquierda, en la frontera de Polonia y Rusia, cerca del río Niemen. La línea gruesa dorada muestra el tamaño de la Gran Armada (422,000) en el momento en que invadía Rusia en junio de 1812. El ancho de esta banda indica el tamaño de la armada en cada punto del mapa. En septiembre, la armada llegó a Moscú, que ya había sido saqueada y dejada desértica, con sólo 100,000 hombres. El camino del retiro de Napoleón desde Moscú está representado por la línea oscura (gris) que está en la parte inferior, que está relacionada a su vez con la temperatura y las fechas en el diagrama de abajo. Fue un invierno muy frío, y muchos se congelaron en su salida de Rusia. Como se muestra en el mapa, cruzar el río Berezina fue un desastre, y el ejército de Napoleón logró regresar a Polonia con tan sólo 10,000 hombres. También se muestran los movimientos de las tropas auxiliaries, que buscaban proteger por atrás y por la delantera mientras la armada avanzaba hacia Moscú. La gráfica de Minard cuenta una historia rica y cohesiva, coherente con datos multivariados y con los hechos históricos, y que puede ser más ilustrativa que tan sólo representar un número rebotando a lo largo del tiempo. Referencias "],["apéndice-transformaciones.html", "Apéndice: Transformaciones", " Apéndice: Transformaciones En ocasiones es conveniente transformar los datos para el análisis, el objetivo de los ajustes es simplificar la interpretación y el análisis al eliminar fuentes de variación conocidas, también es común realizan transformaciones para simplificar los patrones. Algunos ejemplos donde eliminamos efectos conocidos: Cuando analizamos el precio de venta de las casas podemos eliminar la variación debida al tamaño de las casas al pasar de precio de venta a precio de venta por metro cuadrado. De manera similar al analizar las propinas puede convenir considerar la propina como porcentaje de la cuenta. En series de tiempo cuando los datos están relacionados con el tamaño de la población podemos ajustar a mediciones per capita (en series de tiempo PIB). También es común ajustar por inflación, o poner cantidades monetarias en valor presente. mex_dat &lt;- global_economy |&gt; filter(Code == &quot;MEX&quot;) pib &lt;- ggplot(mex_dat, aes(x = Year, y = GDP / 1e6)) + geom_line() pib_pc &lt;- ggplot(mex_dat, aes(x = Year, y = GDP / Population)) + geom_line() pib / pib_pc Adicionalmente podemos recurrir a otras transformaciones matemáticas (e.g. logaritmo, raíz cuadrada) que simplifiquen el patrón en los datos y la interpretación. Veamos un ejemplo donde es apropiado la transformación logaritmo. Usamos los datos Animals con información de peso corporal promedio y peso cerebral promedio para 28 especies. Buscamos entender la relación entre estas dos variables, e inspeccionar que especies se desvían (residuales) del esperado. Comenzamos con un diagrama de dispersión usando las unidades originales animals_tbl &lt;- as_tibble(Animals, rownames = &quot;animal&quot;) p1 &lt;- ggplot(animals_tbl, aes(x = body, y = brain, label = animal)) + geom_point() + labs(subtitle = &quot;Unidades originales&quot;) p2 &lt;- ggplot(animals_tbl, aes(x = body, y = brain, label = animal)) + geom_point() + xlim(0, 500) + ylim(0, 1500) + geom_text_repel() + labs(subtitle = &quot;Unidades originales, eliminando &#39;grandes&#39;&quot;) (p1 + p2) Incluso cuando nos limitamos a especies de menos de 500 kg de masa corporal, la relación no es fácil de descrubir.En la suguiente gráfica hacemos la transformación logaritmo y obtenemos una gráfica más fácil de leer, además los datos se modelarán con más facilidad. p3 &lt;- ggplot(animals_tbl, aes(x = log(body), y = log(brain), label = animal)) + geom_smooth(method = &quot;lm&quot;, se = FALSE, color = &quot;red&quot;) + geom_point() + geom_text_repel() + stat_poly_eq(use_label(c(&quot;eq&quot;))) p3 ## `geom_smooth()` using formula = &#39;y ~ x&#39; La transformación logaritmo tiene también ventajas en interpretación, para diferencias chicas en escala log, las diferencias corresponden a diferencias porcentuales en la escala original, por ejempo consideremos la diferencia entre el peso en escala log de humano y borrego: 4.13 - 4.02 = 0.11. Confirmamos que el humano es aproximadamente 11% más pesado que el borrego en la escala original: 62/55.5 - 1 = 0.12 animals_tbl &lt;- animals_tbl |&gt; mutate(log_body = log(body), log_brain = log(brain)) animals_tbl |&gt; filter(animal == &quot;Human&quot; | animal == &quot;Sheep&quot;) |&gt; arrange(body) |&gt; gt::gt() |&gt; gt::fmt_number() #ufmafyhycy table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #ufmafyhycy thead, #ufmafyhycy tbody, #ufmafyhycy tfoot, #ufmafyhycy tr, #ufmafyhycy td, #ufmafyhycy th { border-style: none; } #ufmafyhycy p { margin: 0; padding: 0; } #ufmafyhycy .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #ufmafyhycy .gt_caption { padding-top: 4px; padding-bottom: 4px; } #ufmafyhycy .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #ufmafyhycy .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #ufmafyhycy .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #ufmafyhycy .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #ufmafyhycy .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #ufmafyhycy .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #ufmafyhycy .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #ufmafyhycy .gt_column_spanner_outer:first-child { padding-left: 0; } #ufmafyhycy .gt_column_spanner_outer:last-child { padding-right: 0; } #ufmafyhycy .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #ufmafyhycy .gt_spanner_row { border-bottom-style: hidden; } #ufmafyhycy .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #ufmafyhycy .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #ufmafyhycy .gt_from_md > :first-child { margin-top: 0; } #ufmafyhycy .gt_from_md > :last-child { margin-bottom: 0; } #ufmafyhycy .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #ufmafyhycy .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #ufmafyhycy .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #ufmafyhycy .gt_row_group_first td { border-top-width: 2px; } #ufmafyhycy .gt_row_group_first th { border-top-width: 2px; } #ufmafyhycy .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #ufmafyhycy .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #ufmafyhycy .gt_first_summary_row.thick { border-top-width: 2px; } #ufmafyhycy .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #ufmafyhycy .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #ufmafyhycy .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #ufmafyhycy .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #ufmafyhycy .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #ufmafyhycy .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #ufmafyhycy .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #ufmafyhycy .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #ufmafyhycy .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #ufmafyhycy .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #ufmafyhycy .gt_left { text-align: left; } #ufmafyhycy .gt_center { text-align: center; } #ufmafyhycy .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #ufmafyhycy .gt_font_normal { font-weight: normal; } #ufmafyhycy .gt_font_bold { font-weight: bold; } #ufmafyhycy .gt_font_italic { font-style: italic; } #ufmafyhycy .gt_super { font-size: 65%; } #ufmafyhycy .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #ufmafyhycy .gt_asterisk { font-size: 100%; vertical-align: 0; } #ufmafyhycy .gt_indent_1 { text-indent: 5px; } #ufmafyhycy .gt_indent_2 { text-indent: 10px; } #ufmafyhycy .gt_indent_3 { text-indent: 15px; } #ufmafyhycy .gt_indent_4 { text-indent: 20px; } #ufmafyhycy .gt_indent_5 { text-indent: 25px; } #ufmafyhycy .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #ufmafyhycy div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } animal body brain log_body log_brain Sheep 55.50 175.00 4.02 5.16 Human 62.00 1,320.00 4.13 7.19 Y podemos usarlo también para interpretar la recta de referencia \\(y = 2.55 + 0.5 x\\) , para cambios chicos: Un incremento de 10% en masa total corresponde en un incremento de 5% en masa cerebral. El coeficiente de la regresión log-log, en nuestro ejemplo 0.5, es la elasticidad y es un concepto común en economía. Justificación Para entender la interpretación como cambio porcentual recordemos primero que la representación con series de Taylor de la función exponencial es: \\[e^x = \\sum_{n=0}^\\infty \\frac{x^n}{n!}\\] Más aún podemos tener una aproximación usando polinomios de Taylor, en el caso de la exponencial el \\(k\\)-ésimo polinomio de Taylor está dado por: \\[e^\\delta \\approx 1 + \\delta + \\frac{1}{2!}\\delta^2 + \\dots + \\frac{1}{k!}\\delta^k\\] y si \\(\\delta\\) es chica (digamos menor a 0.15), entonces la aproximación de primer grado es razonable y tenemos: \\[Ae^{\\delta} \\approx A(1+\\delta)\\] dat &lt;- tibble(delta = seq(0, 1, 0.01), exp_delta = exp(delta), uno_mas_delta = 1 + delta) ggplot(dat, aes(x = uno_mas_delta, y = exp_delta)) + geom_line() + geom_abline(color = &quot;red&quot;) + annotate(&quot;text&quot;, x = 1.20, y = 1.18, label = &quot;y = x&quot;, color = &quot;red&quot;, size = 6) "],["referencias.html", "Referencias", " Referencias "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Fundamentos de Estadística con Remuestreo Información del curso", " Fundamentos de Estadística con Remuestreo Teresa Ortiz, Felipe González, Alfredo Garbuno Información del curso Notas del curso Fundamentos de Estadística con Remuestreo, este curso busca explicar los principios básicos de la estadística y su papel en el análisis de datos. Nuestro punto de vista es uno de fundamentos, con menos énfasis en recetas o técnicas particulares. Ligas Notas: https://tereom.github.io/fundamentos-2024/ Repositorio con material: https://github.com/tereom/fundamentos-2024 Correo: teresa.ortiz.mancera@gmail.com Zoom clase (miércoles 4:00-7:00 pm): https://itam.zoom.us/j/92745909276 Zoom sesiones de dudas (lunes de 5:30-6:30 pm): https://itam.zoom.us/j/92518922348 Canvas: https://itam.instructure.com/courses/13328 x Este trabajo está bajo una Licencia Creative Commons Atribución 4.0 Internacional. "],["temario.html", "Temario Plan semanal Evaluación", " Temario Plan semanal Datos y análisis exploratorio Referencias: (W. S. Cleveland 1994), (Chihara and Hesterberg 2018) Visualización1 Análisis exploratorio Tipos de datos o estudios Muestras diseñadas y muestras naturales Experimentos y datos observacionales Introducción a Pruebas de Hipótesis Referencias: (Chihara and Hesterberg 2018) Introducción a pruebas de hipótesis. Pruebas de permutaciones Muestras pareadas y otros ejemplos Estimación y distribución de muestreo Referencias: (Chihara and Hesterberg 2018), (Tim C. Hesterberg 2015b) Estimadores y su distribución de muestreo Repaso de probabilidad y Teorema del límite central Introducción a estimación por intervalos Referencias: (Chihara and Hesterberg 2018), (Efron and Tibshirani 1993), (Tim C. Hesterberg 2015b) El método plugin y el boostrap Bootstrap e Intervalos de confianza. Ejemplos. Estimación Referencias: (Chihara and Hesterberg 2018), (Wasserman 2013) Estimación por máxima verosimilitud Ejemplos de estimación por máxima verosimilitud y Bootstrap paramétrico Propiedades de estimadores de máxima verosimilitud Más de pruebas de hipótesis Referencias: (Chihara and Hesterberg 2018), (Wasserman 2013) Pruebas de hipótesis para medias y proporciones: una y dos poblaciones. Introducción a inferencia bayesiana Referencias: (Kruschke 2015) Introducción a inferencia bayesiana Ejemplos de distribuciones conjugadas Introducción a métodos computacionales básicos: Muestreadores Metrópolis y Gibbs Ejemplos de inferencia bayesiana en Stan Evaluación Se evaluará mediante tareas semanales y dos exámenes: Tareas semanales (20%) Examen parcial en clase y a casa (40%) Examen final a casa (40%) Referencias "],["análisis-exploratorio.html", "Sección 1 Análisis exploratorio El papel de la exploración en el análisis de datos Preguntas y datos Algunos conceptos básicos Ejemplos Suavizamiento loess Caso de estudio: nacimientos en México", " Sección 1 Análisis exploratorio “Exploratory data analysis can never be the whole story, but nothing else can serve as the foundation stone –as the first step.” — John Tukey El papel de la exploración en el análisis de datos El estándar científico para contestar preguntas o tomar decisiones es uno que se basa en el análisis de datos. Es decir, en primer lugar se deben reunir todos los datos que puedan contener o sugerir alguna guía para entender mejor la pregunta o la decisión a la que nos enfrentamos. Esta recopilación de datos —que pueden ser cualitativos, cuantitativos, o una mezcla de los dos— debe entonces ser analizada para extraer información relevante para nuestro problema. En análisis de datos existen dos distintos tipos de trabajo: El trabajo exploratorio o de detective: ¿cuáles son los aspectos importantes de estos datos? ¿qué indicaciones generales muestran los datos? ¿qué tareas de análisis debemos empezar haciendo? ¿cuáles son los caminos generales para formular con precisión y contestar algunas preguntas que nos interesen? El trabajo inferencial, confirmatorio, o de juez: ¿cómo evaluar el peso de la evidencia de los descubrimientos del paso anterior? ¿qué tan bien soportadas están las respuestas y conclusiones por nuestro conjunto de datos? Preguntas y datos Cuando observamos un conjunto de datos, independientemente de su tamaño, el paso inicial más importante es entender bajo qué proceso se generan los datos. A grandes rasgos, cuanto más sepamos de este proceso, mejor podemos contestar preguntas de interés. En muchos casos, tendremos que hacer algunos supuestos de cómo se generan estos datos para dar respuestas (condicionales a esos supuestos). Algunos conceptos básicos Empezamos explicando algunas ideas que no serán útiles más adelante. El primer concepto se refiere a entender cómo se distribuyen los datos a los largo de su escala de medición. Comenzamos con un ejemplo: los siguientes datos fueron registrados en un restaurante durante cuatro días consecutivos. library(tidyverse) library(patchwork) # organizar gráficas library(gt) # formatear tablas source(&quot;R/funciones_auxiliares.R&quot;) # usamos los datos tips del paquete reshape2 propinas &lt;- read_csv(&quot;./data/propinas.csv&quot;) Y vemos una muestra slice_sample(propinas, n = 10) |&gt; gt() #qcaxfnkqcl table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #qcaxfnkqcl thead, #qcaxfnkqcl tbody, #qcaxfnkqcl tfoot, #qcaxfnkqcl tr, #qcaxfnkqcl td, #qcaxfnkqcl th { border-style: none; } #qcaxfnkqcl p { margin: 0; padding: 0; } #qcaxfnkqcl .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #qcaxfnkqcl .gt_caption { padding-top: 4px; padding-bottom: 4px; } #qcaxfnkqcl .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #qcaxfnkqcl .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #qcaxfnkqcl .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #qcaxfnkqcl .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #qcaxfnkqcl .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #qcaxfnkqcl .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #qcaxfnkqcl .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #qcaxfnkqcl .gt_column_spanner_outer:first-child { padding-left: 0; } #qcaxfnkqcl .gt_column_spanner_outer:last-child { padding-right: 0; } #qcaxfnkqcl .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #qcaxfnkqcl .gt_spanner_row { border-bottom-style: hidden; } #qcaxfnkqcl .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #qcaxfnkqcl .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #qcaxfnkqcl .gt_from_md > :first-child { margin-top: 0; } #qcaxfnkqcl .gt_from_md > :last-child { margin-bottom: 0; } #qcaxfnkqcl .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #qcaxfnkqcl .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #qcaxfnkqcl .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #qcaxfnkqcl .gt_row_group_first td { border-top-width: 2px; } #qcaxfnkqcl .gt_row_group_first th { border-top-width: 2px; } #qcaxfnkqcl .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #qcaxfnkqcl .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #qcaxfnkqcl .gt_first_summary_row.thick { border-top-width: 2px; } #qcaxfnkqcl .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #qcaxfnkqcl .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #qcaxfnkqcl .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #qcaxfnkqcl .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #qcaxfnkqcl .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #qcaxfnkqcl .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #qcaxfnkqcl .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #qcaxfnkqcl .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #qcaxfnkqcl .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #qcaxfnkqcl .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #qcaxfnkqcl .gt_left { text-align: left; } #qcaxfnkqcl .gt_center { text-align: center; } #qcaxfnkqcl .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #qcaxfnkqcl .gt_font_normal { font-weight: normal; } #qcaxfnkqcl .gt_font_bold { font-weight: bold; } #qcaxfnkqcl .gt_font_italic { font-style: italic; } #qcaxfnkqcl .gt_super { font-size: 65%; } #qcaxfnkqcl .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #qcaxfnkqcl .gt_asterisk { font-size: 100%; vertical-align: 0; } #qcaxfnkqcl .gt_indent_1 { text-indent: 5px; } #qcaxfnkqcl .gt_indent_2 { text-indent: 10px; } #qcaxfnkqcl .gt_indent_3 { text-indent: 15px; } #qcaxfnkqcl .gt_indent_4 { text-indent: 20px; } #qcaxfnkqcl .gt_indent_5 { text-indent: 25px; } #qcaxfnkqcl .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #qcaxfnkqcl div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } cuenta_total propina fumador dia momento num_personas 15.81 3.16 Si Sab Cena 2 16.47 3.23 Si Jue Comida 3 38.01 3.00 Si Sab Cena 4 3.07 1.00 Si Sab Cena 1 12.46 1.50 No Vie Cena 2 8.52 1.48 No Jue Comida 2 11.61 3.39 No Sab Cena 2 50.81 10.00 Si Sab Cena 3 38.07 4.00 No Dom Cena 3 15.06 3.00 No Sab Cena 2 Aquí la unidad de observación es una cuenta particular. Tenemos tres mediciones numéricas de cada cuenta: cúanto fue la cuenta total, la propina, y el número de personas asociadas a la cuenta. Los datos están separados según se fumó o no en la mesa, y temporalmente en dos partes: el día (Jueves, Viernes, Sábado o Domingo), cada uno separado por Cena y Comida. Denotamos por \\(x\\) el valor de medición de una unidad de observación. Usualmente utilizamos sub-índices para identificar entre diferentes puntos de datos (observaciones), por ejemplo, \\(x_n\\) para la \\(n-\\)ésima observación. De tal forma que una colección de \\(N\\) observaciones la escribimos como \\[\\begin{align} \\{x_1, \\ldots, x_N\\}. \\end{align}\\] El primer tipo de comparaciones que nos interesa hacer es para una medición: ¿Varían mucho o poco los datos de un tipo de medición? ¿Cuáles son valores típicos o centrales? ¿Existen valores atípicos? Supongamos entonces que consideramos simplemente la variable de cuenta_total. Podemos comenzar por ordenar los datos, y ver cuáles datos están en los extremos y cuáles están en los lugares centrales: propinas &lt;- propinas |&gt; mutate(orden_cuenta = rank(cuenta_total, ties.method = &quot;first&quot;), f = (orden_cuenta - 0.5) / n()) cuenta &lt;- propinas |&gt; select(orden_cuenta, f, cuenta_total) |&gt; arrange(f) bind_rows(head(cuenta), tail(cuenta)) |&gt; gt() |&gt; fmt_number(columns = f, decimals = 3) #phufidpmfh table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #phufidpmfh thead, #phufidpmfh tbody, #phufidpmfh tfoot, #phufidpmfh tr, #phufidpmfh td, #phufidpmfh th { border-style: none; } #phufidpmfh p { margin: 0; padding: 0; } #phufidpmfh .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #phufidpmfh .gt_caption { padding-top: 4px; padding-bottom: 4px; } #phufidpmfh .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #phufidpmfh .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #phufidpmfh .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #phufidpmfh .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #phufidpmfh .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #phufidpmfh .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #phufidpmfh .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #phufidpmfh .gt_column_spanner_outer:first-child { padding-left: 0; } #phufidpmfh .gt_column_spanner_outer:last-child { padding-right: 0; } #phufidpmfh .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #phufidpmfh .gt_spanner_row { border-bottom-style: hidden; } #phufidpmfh .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #phufidpmfh .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #phufidpmfh .gt_from_md > :first-child { margin-top: 0; } #phufidpmfh .gt_from_md > :last-child { margin-bottom: 0; } #phufidpmfh .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #phufidpmfh .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #phufidpmfh .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #phufidpmfh .gt_row_group_first td { border-top-width: 2px; } #phufidpmfh .gt_row_group_first th { border-top-width: 2px; } #phufidpmfh .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #phufidpmfh .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #phufidpmfh .gt_first_summary_row.thick { border-top-width: 2px; } #phufidpmfh .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #phufidpmfh .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #phufidpmfh .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #phufidpmfh .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #phufidpmfh .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #phufidpmfh .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #phufidpmfh .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #phufidpmfh .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #phufidpmfh .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #phufidpmfh .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #phufidpmfh .gt_left { text-align: left; } #phufidpmfh .gt_center { text-align: center; } #phufidpmfh .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #phufidpmfh .gt_font_normal { font-weight: normal; } #phufidpmfh .gt_font_bold { font-weight: bold; } #phufidpmfh .gt_font_italic { font-style: italic; } #phufidpmfh .gt_super { font-size: 65%; } #phufidpmfh .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #phufidpmfh .gt_asterisk { font-size: 100%; vertical-align: 0; } #phufidpmfh .gt_indent_1 { text-indent: 5px; } #phufidpmfh .gt_indent_2 { text-indent: 10px; } #phufidpmfh .gt_indent_3 { text-indent: 15px; } #phufidpmfh .gt_indent_4 { text-indent: 20px; } #phufidpmfh .gt_indent_5 { text-indent: 25px; } #phufidpmfh .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #phufidpmfh div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } orden_cuenta f cuenta_total 1 0.002 3.07 2 0.006 5.75 3 0.010 7.25 4 0.014 7.25 5 0.018 7.51 6 0.023 7.56 239 0.977 44.30 240 0.982 45.35 241 0.986 48.17 242 0.990 48.27 243 0.994 48.33 244 0.998 50.81 También podemos graficar los datos en orden, interpolando valores consecutivos. A esta función le llamamos la función de cuantiles para la variable cuenta_total. Nos sirve para comparar directamente los distintos valores que observamos los datos según el orden que ocupan. En particular, podemos estudiar la dispersión y valores centrales de los datos observados: El rango de datos va de unos 3 dólares hasta 50 dólares Los valores centrales, por ejemplo el 50% de los valores más centrales, están entre unos 13 y 25 dólares. El valor que divide en dos mitades iguales a los datos es de alrededor de 18 dólares. El cuantil \\(f\\), que denotamos por \\(q(f)\\) es valor a lo largo de la escala de medición de los datos tal que aproximadamente una fracción \\(f\\) de los datos son menores o iguales a \\(q(f)\\). Al cuantil \\(f=0.5\\) le llamamos la mediana. A los cuantiles \\(f=0.25\\) y \\(f=0.75\\) les llamamos cuartiles inferior y superior. En nuestro ejemplo: Los valores centrales —del cuantil 0.25 al 0.75, por decir un ejemplo— están entre unos 13 y 25 dólares. Estos dos cuantiles se llaman cuartil inferior y cuartil superior respectivamente El cuantil 0.5 (o también conocido como mediana) está alrededor de 18 dólares. Éste último puede ser utilizado para dar un valor central de la distribución de valores para cuenta_total. Asimismo podemos dar resúmenes más refinados si es necesario. Por ejemplo, podemos reportar que: El cuantil 0.95 es de unos 35 dólares — sólo 5% de las cuentas son de más de 35 dólares El cuantil 0.05 es de unos 8 dólares — sólo 5% de las cuentas son de 8 dólares o menos. Finalmente, la forma de la gráfica se interpreta usando su pendiente (tasa de cambio) haciendo comparaciones en diferentes partes de la gráfica: La distribución de valores tiene asimetría: el 10% de las cuentas más altas tiene considerablemente más dispersión que el 10% de las cuentas más bajas. Entre los cuantiles 0.2 y 0.7 es donde existe mayor densidad de datos: la pendiente (tasa de cambio) es baja, lo que significa que al avanzar en los valores observados, los cuantiles (el porcentaje de casos) aumenta rápidamente. Cuando la pendiente alta, quiere decir que los datos tienen más dispersión local o están más separados. Observación: Hay varias maneras de definir los cuantiles (ver (William S. Cleveland 1993)): Supongamos que queremos definir \\(q(f)\\), y denotamos los datos ordenados como \\(x_{(1)}, x_{(2)}, \\ldots, x_{(N)}\\), de forma que \\(x_{(1)}\\) es el dato más chico y \\(x_{(N)}\\) es el dato más grande. Para cada \\(x_{(i)}\\) definimos \\[f_i = i / N\\] entonces definimos el cuantil \\(q(f_i)=x_{(i)}\\). Para cualquier \\(f\\) entre 0 y 1, podemos definir \\(q(f)\\) como sigue: si \\(f\\) está entre \\(f_i\\) y \\(f_{i+1}\\) interpolamos linealmente los valores correspondientes \\(x_{(i)}\\) y \\(x_{(i+1)}\\). En la práctica, es más conveniente usar \\(f_i= \\frac{i - 0.5}{N}\\). La gráfica de cuantiles no cambia mucho comparado con la difinición anterior, y esto nos permitirá comparar de mejor manera con distribuciones teóricas que no tienen definido su cuantil 0 y el 1, pues tienen soporte en los números reales (como la distribución normal, por ejemplo). Asociada a la función de cuantiles \\(q\\) tenemos la distribución acumulada empírica de los datos, que es aproximadamente inversa de la función de cuantiles, y se define como: \\[\\hat{F}(x) = i/N\\] si \\(x_{(i)} \\leq x &lt; x_{(i+1)}\\). Nótese que \\(\\hat{F}(q(f_i)) = i/N = f_i\\) (demuéstralo). acum_cuenta &lt;- ecdf(cuenta$cuenta_total) cuenta &lt;- cuenta |&gt; mutate(dea_cuenta_total = acum_cuenta(cuenta_total)) g_acum &lt;- ggplot(cuenta, aes(x = cuenta_total, y = dea_cuenta_total)) + geom_point() + labs(subtitle = &quot;Distribución acum empírica de cuenta total&quot;, x = &quot;&quot;) g_cuantiles + g_acum La función de distribución acumulada empírica es otra forma de graficar la dispersión de los datos. En su gráfica vemos que proporción de los datos que son iguales o están por debajo de cada valor en el eje horizontal. Nota: En análisis de datos, es más frecuente utilizar la función de cuantiles pues existen versiones más generales que son útiles, por ejemplo, para evaluar ajuste de modelos probabilísticos En la teoría, generalmente es más común utilizar la fda empírica, que tiene una única definición que veremos coincide con definiciones teóricas. Histogramas En algunos casos, es más natural hacer un histograma, donde dividimos el rango de la variable en cubetas o intervalos (en este caso de igual longitud), y graficamos por medio de barras cuántos datos caen en cada cubeta: Es una gráfica más popular, pero perdemos cierto nivel de detalle, y distintas particiones resaltan distintos aspectos de los datos. ¿Cómo se ve la gráfica de cuantiles de las propinas? ¿Cómo crees que esta gráfica se compara con distintos histogramas? g_1 &lt;- ggplot(propinas, aes(sample = propina)) + geom_qq(distribution = stats::qunif) + labs(x = &quot;f&quot;, y = &quot;propina&quot;) g_1 Finalmente, una gráfica más compacta que resume la gráfica de cuantiles o el histograma es el diagrama de caja y brazos. Mostramos dos versiones, la clásica de Tukey (T) y otra versión menos común de Spear/Tufte (ST): library(ggthemes) cuartiles &lt;- quantile(cuenta$cuenta_total) cuartiles |&gt; round(2) ## 0% 25% 50% 75% 100% ## 3.07 13.35 17.80 24.13 50.81 g_1 &lt;- ggplot(cuenta, aes(x = f, y = cuenta_total)) + labs(subtitle = &quot;Gráfica de cuantiles: Cuenta total&quot;) + geom_hline(yintercept = cuartiles[2:4], colour = &quot;gray&quot;) + geom_point(alpha = 0.5) + geom_line() g_2 &lt;- ggplot(cuenta, aes(x = factor(&quot;ST&quot;, levels = &quot;ST&quot;), y = cuenta_total)) + geom_tufteboxplot() + labs(subtitle = &quot;&quot;, x = &quot;&quot;, y = &quot;&quot;) g_3 &lt;- ggplot(cuenta, aes(x = factor(&quot;T&quot;), y = cuenta_total)) + geom_boxplot() + labs(subtitle = &quot;&quot;, x = &quot;&quot;, y = &quot;&quot;) g_4 &lt;- ggplot(cuenta, aes(x = factor(&quot;P&quot;), y = cuenta_total)) + geom_jitter(height = 0, width =0.2, alpha = 0.5) + labs(subtitle = &quot;&quot;, x = &quot;&quot;, y = &quot;&quot;) g_5 &lt;- ggplot(cuenta, aes(x = factor(&quot;V&quot;), y = cuenta_total)) + geom_violin() + labs(subtitle = &quot;&quot;, x = &quot;&quot;, y = &quot;&quot;) g_1 + g_2 + g_3 + g_4 + plot_layout(widths = c(8, 2, 2, 2)) El diagrama de abajo explica los elementos de la versión típica del diagrama de caja y brazos (boxplot). RIC se refiere al Rango Intercuantílico, definido por la diferencia entre los cuantiles 25% y 75%. Figura: Jumanbar / CC BY-SA Ventajas en el análisis inicial En un principio del análisis, estos resúmenes (cuantiles) pueden ser más útiles que utilizar medias y varianzas, por ejemplo. La razón es que los cuantiles: Son cantidades más fácilmente interpretables Los cuantiles centrales son más resistentes a valores atípicos que medias o varianzas Permiten identificar valores extremos Es fácil comparar cuantiles de distintos bonches de datos en la misma escala Nota: Existen diferentes definiciones para calcular cuantiles de una muestra de datos, puedes leer más en este artículo. Media y desviación estándar Las medidas más comunes de localización y dispersión para un conjunto de datos son la media muestral y la desviación estándar muestral. En general, no son muy apropiadas para iniciar el análisis exploratorio, pues: Son medidas más difíciles de interpretar y explicar que los cuantiles. En este sentido, son medidas especializadas. Por ejemplo, compara una explicación intuitiva de la mediana contra una explicación intuitiva de la media. No son resistentes a valores atípicos. Su falta de resistencia los vuelve poco útiles en las primeras etapas de limpieza y descripción y en resúmenes deficientes para distribuciones irregulares (con colas largas por ejemplo). La media, o promedio, se denota por \\(\\bar x\\) y se define como \\[\\begin{align} \\bar x = \\frac1N \\sum_{n = 1}^N x_n. \\end{align}\\] La desviación estándar muestral se define como \\[\\begin{align} \\text{std}(x) = \\sqrt{\\frac1{N-1} \\sum_{n = 1}^N (x_n - \\bar x)^2}. \\end{align}\\] Observación: Si \\(N\\) no es muy chica, no importa mucho si dividimos por \\(N\\) o por \\(N-1\\) en la fórmula de la desviación estándar. La razón de que típicamente se usa \\(N-1\\) la veremos más adelante, en la parte de estimación. Por otro lado, ventajas de estas medidas de centralidad y dispersión son: La media y desviación estándar son computacionalmente convenientes. Por lo tanto regresaremos a estas medidas una vez que estudiemos modelos de probabilidad básicos. En muchas ocasiones conviene usar estas medidas pues permite hacer comparaciones históricas o tradicionales —pues análisis anteriores pudieran estar basados en éstas. Considera el caso de tener \\(N\\) observaciones y asume que ya tienes calculado el promedio para dichas observaciones. Este promedio lo denotaremos por \\(\\bar x_N\\). Ahora, considera que has obtenido \\(M\\) observaciones más. Escribe una fórmula recursiva para la media del conjunto total de datos \\(\\bar x_{N+M}\\) en función de lo que ya tenías precalculado \\(\\bar x_N.\\) ¿En qué situaciones esta propiedad puede ser conveniente? Ejemplos Precios de casas En este ejemplo consideremos los datos de precios de ventas de la ciudad de Ames, Iowa. En particular nos interesa entender la variación del precio de las casas. Por este motivo calculamos los cuantiles que corresponden al 25%, 50% y 75% (cuartiles), así como el mínimo y máximo de los precios de las casas: quantile(casas |&gt; pull(precio_miles)) ## 0% 25% 50% 75% 100% ## 37.9 132.0 165.0 215.0 755.0 Comprueba que el mínimo y máximo están asociados a los cuantiles 0% y 100%, respectivamente. Una posible comparación es considerar los precios y sus variación en función de zona de la ciudad en que se encuentra una vivienda. Podemos usar diagramas de caja y brazos para hacer una comparación burda de los precios en distintas zonas de la ciudad: ggplot(casas, aes(x = nombre_zona, y = precio_miles)) + geom_boxplot() + coord_flip() La primera pregunta que nos hacemos es cómo pueden variar las características de las casas dentro de cada zona. Para esto, podemos considerar el área de las casas. En lugar de graficar el precio, graficamos el precio por metro cuadrado, por ejemplo: ggplot(casas, aes(x = nombre_zona, y = precio_m2)) + geom_boxplot() + coord_flip() Podemos cuantificar la variación que observamos de zona a zona y la variación que hay dentro de cada una de las zonas. Una primera aproximación es observar las variación del precio al calcular la mediana dentro de cada zona, y después cuantificar por medio de cuantiles cómo varía la mediana entre zonas: casas |&gt; group_by(nombre_zona) |&gt; summarise(mediana_zona = median(precio_m2), .groups = &quot;drop&quot;) |&gt; arrange(mediana_zona) |&gt; pull(mediana_zona) |&gt; quantile() |&gt; round() ## 0% 25% 50% 75% 100% ## 963 1219 1298 1420 1725 Por otro lado, las variaciones con respecto a las medianas dentro de cada zona, por grupo, se resume como: quantile(casas |&gt; group_by(nombre_zona) |&gt; mutate(residual = precio_m2 - median(precio_m2)) |&gt; pull(residual)) |&gt; round() ## 0% 25% 50% 75% 100% ## -765 -166 0 172 1314 Nótese que este último paso tiene sentido pues la variación dentro de las zonas, en términos de precio por metro cuadrado, es similar. Esto no lo podríamos haber hecho de manera efectiva si se hubiera utilizado el precio de las casas sin ajustar por su tamaño. Podemos resumir este primer análisis con un par de gráficas de cuantiles (William S. Cleveland (1993)): mediana &lt;- median(casas$precio_m2) resumen &lt;- casas |&gt; select(nombre_zona, precio_m2) |&gt; group_by(nombre_zona) |&gt; mutate(mediana_zona = median(precio_m2)) |&gt; mutate(residual = precio_m2 - mediana_zona) |&gt; ungroup() |&gt; mutate(mediana_zona = mediana_zona - mediana) |&gt; select(nombre_zona, mediana_zona, residual) |&gt; pivot_longer(mediana_zona:residual, names_to = &quot;tipo&quot;, values_to = &quot;valor&quot;) ggplot(resumen, aes(sample = valor)) + geom_qq(distribution = stats::qunif) + facet_wrap(~ tipo) + ylab(&quot;Precio por m2&quot;) + xlab(&quot;f&quot;) + labs(subtitle = &quot;Precio por m2 por zona&quot;, caption = paste0(&quot;Mediana total de &quot;, round(mediana))) Vemos que la mayor parte de la variación del precio por metro cuadrado ocurre dentro de cada zona, una vez que controlamos por el tamaño de las casas. La variación dentro de cada zona es aproximadamente simétrica, aunque la cola derecha es ligeramente más larga con algunos valores extremos. Podemos seguir con otro indicador importante: la calificación de calidad de los terminados de las casas. Como primer intento podríamos hacer: Lo que indica que las calificaciones de calidad están distribuidas de manera muy distinta a lo largo de las zonas, y que probablemente no va ser simple desentrañar qué variación del precio se debe a la zona y cuál se debe a la calidad. Prueba Enlace Consideremos la prueba Enlace (2011) de matemáticas para primarias. Una primera pregunta que alguien podría hacerse es: ¿cuáles escuelas son mejores en este rubro, las privadas o las públicas? enlace_tbl &lt;- enlace |&gt; group_by(tipo) |&gt; summarise(n_escuelas = n(), cuantiles = list(cuantil(mate_6, c(0.05, 0.25, 0.5, 0.75, 0.95)))) |&gt; unnest(cols = cuantiles) |&gt; mutate(valor = round(valor)) enlace_tbl |&gt; spread(cuantil, valor) |&gt; gt() #sykxxonvyi table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #sykxxonvyi thead, #sykxxonvyi tbody, #sykxxonvyi tfoot, #sykxxonvyi tr, #sykxxonvyi td, #sykxxonvyi th { border-style: none; } #sykxxonvyi p { margin: 0; padding: 0; } #sykxxonvyi .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #sykxxonvyi .gt_caption { padding-top: 4px; padding-bottom: 4px; } #sykxxonvyi .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #sykxxonvyi .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #sykxxonvyi .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #sykxxonvyi .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sykxxonvyi .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #sykxxonvyi .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #sykxxonvyi .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #sykxxonvyi .gt_column_spanner_outer:first-child { padding-left: 0; } #sykxxonvyi .gt_column_spanner_outer:last-child { padding-right: 0; } #sykxxonvyi .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #sykxxonvyi .gt_spanner_row { border-bottom-style: hidden; } #sykxxonvyi .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #sykxxonvyi .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #sykxxonvyi .gt_from_md > :first-child { margin-top: 0; } #sykxxonvyi .gt_from_md > :last-child { margin-bottom: 0; } #sykxxonvyi .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #sykxxonvyi .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #sykxxonvyi .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #sykxxonvyi .gt_row_group_first td { border-top-width: 2px; } #sykxxonvyi .gt_row_group_first th { border-top-width: 2px; } #sykxxonvyi .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #sykxxonvyi .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #sykxxonvyi .gt_first_summary_row.thick { border-top-width: 2px; } #sykxxonvyi .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sykxxonvyi .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #sykxxonvyi .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #sykxxonvyi .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #sykxxonvyi .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #sykxxonvyi .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sykxxonvyi .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #sykxxonvyi .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #sykxxonvyi .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #sykxxonvyi .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #sykxxonvyi .gt_left { text-align: left; } #sykxxonvyi .gt_center { text-align: center; } #sykxxonvyi .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #sykxxonvyi .gt_font_normal { font-weight: normal; } #sykxxonvyi .gt_font_bold { font-weight: bold; } #sykxxonvyi .gt_font_italic { font-style: italic; } #sykxxonvyi .gt_super { font-size: 65%; } #sykxxonvyi .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #sykxxonvyi .gt_asterisk { font-size: 100%; vertical-align: 0; } #sykxxonvyi .gt_indent_1 { text-indent: 5px; } #sykxxonvyi .gt_indent_2 { text-indent: 10px; } #sykxxonvyi .gt_indent_3 { text-indent: 15px; } #sykxxonvyi .gt_indent_4 { text-indent: 20px; } #sykxxonvyi .gt_indent_5 { text-indent: 25px; } #sykxxonvyi .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #sykxxonvyi div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } tipo n_escuelas 0.05 0.25 0.5 0.75 0.95 Indígena/Conafe 13599 304 358 412 478 588 General 60166 380 454 502 548 631 Particular 6816 479 551 593 634 703 Para un análisis exploratorio podemos utilizar distintas gráficas. Por ejemplo, podemos utilizar nuevamente las gráficas de caja y brazos, así como graficar los percentiles. Nótese que en la gráfica 1 se utilizan los cuantiles 0.05, 0.25, 0.5, 0.75 y 0.95: ## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0. ## ℹ Please use `linewidth` instead. ## This warning is displayed once every 8 hours. ## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was ## generated. Se puede discutir qué tan apropiada es cada gráfica con el objetivo de realizar comparaciones. Sin duda, graficar más cuantiles es más útil para hacer comparaciones. Por ejemplo, en la Gráfica 1 podemos ver que la mediana de las escuelas generales está cercana al cuantil 5% de las escuelas particulares. Por otro lado, el diagrama de caja y brazos muestra también valores “atípicos”. Antes de contestar prematuramente la pregunta: ¿cuáles son las mejores escuelas? busquemos mejorar la interpretabilidad de nuestras comparaciones. Podemos comenzar por agregar, por ejemplo, el nivel del marginación del municipio donde se encuentra la escuela. Para este objetivo, podemos usar páneles (pequeños múltiplos útiles para hacer comparaciones) y graficar: Esta gráfica pone en contexto la pregunta inicial, y permite evidenciar la dificultad de contestarla. En particular: Señala que la pregunta no sólo debe concentarse en el tipo de “sistema”: pública, privada, etc. Por ejemplo, las escuelas públicas en zonas de marginación baja no tienen una distribución de calificaciones muy distinta a las privadas en zonas de marginación alta. El contexto de la escuela es importante. Debemos de pensar qué factores –por ejemplo, el entorno familiar de los estudiantes– puede resultar en comparaciones que favorecen a las escuelas privadas. Un ejemplo de esto es considerar si los estudiantes tienen que trabajar o no. A su vez, esto puede o no ser reflejo de la calidad del sistema educativo. Si esto es cierto, entonces la pregunta inicial es demasiado vaga y mal planteada. Quizá deberíamos intentar entender cuánto “aporta” cada escuela a cada estudiante, como medida de qué tan buena es cada escuela. Estados y calificaciones en SAT ¿Cómo se relaciona el gasto por alumno, a nivel estatal, con sus resultados académicos? Hay trabajo considerable en definir estos términos, pero supongamos que tenemos el siguiente conjunto de datos (Guber 1999), que son datos oficiales agregados por estado de Estados Unidos. Consideremos el subconjunto de variables sat, que es la calificación promedio de los alumnos en cada estado (para 1997) y expend, que es el gasto en miles de dólares por estudiante en (1994-1995). sat &lt;- read_csv(&quot;data/sat.csv&quot;) sat_tbl &lt;- sat |&gt; select(state, expend, sat) |&gt; gather(variable, valor, expend:sat) |&gt; group_by(variable) |&gt; summarise(cuantiles = list(cuantil(valor))) |&gt; unnest(cols = c(cuantiles)) |&gt; mutate(valor = round(valor, 1)) |&gt; spread(cuantil, valor) sat_tbl |&gt; gt() #trvhlcrcph table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #trvhlcrcph thead, #trvhlcrcph tbody, #trvhlcrcph tfoot, #trvhlcrcph tr, #trvhlcrcph td, #trvhlcrcph th { border-style: none; } #trvhlcrcph p { margin: 0; padding: 0; } #trvhlcrcph .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #trvhlcrcph .gt_caption { padding-top: 4px; padding-bottom: 4px; } #trvhlcrcph .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #trvhlcrcph .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #trvhlcrcph .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #trvhlcrcph .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #trvhlcrcph .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #trvhlcrcph .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #trvhlcrcph .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #trvhlcrcph .gt_column_spanner_outer:first-child { padding-left: 0; } #trvhlcrcph .gt_column_spanner_outer:last-child { padding-right: 0; } #trvhlcrcph .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #trvhlcrcph .gt_spanner_row { border-bottom-style: hidden; } #trvhlcrcph .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #trvhlcrcph .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #trvhlcrcph .gt_from_md > :first-child { margin-top: 0; } #trvhlcrcph .gt_from_md > :last-child { margin-bottom: 0; } #trvhlcrcph .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #trvhlcrcph .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #trvhlcrcph .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #trvhlcrcph .gt_row_group_first td { border-top-width: 2px; } #trvhlcrcph .gt_row_group_first th { border-top-width: 2px; } #trvhlcrcph .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #trvhlcrcph .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #trvhlcrcph .gt_first_summary_row.thick { border-top-width: 2px; } #trvhlcrcph .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #trvhlcrcph .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #trvhlcrcph .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #trvhlcrcph .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #trvhlcrcph .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #trvhlcrcph .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #trvhlcrcph .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #trvhlcrcph .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #trvhlcrcph .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #trvhlcrcph .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #trvhlcrcph .gt_left { text-align: left; } #trvhlcrcph .gt_center { text-align: center; } #trvhlcrcph .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #trvhlcrcph .gt_font_normal { font-weight: normal; } #trvhlcrcph .gt_font_bold { font-weight: bold; } #trvhlcrcph .gt_font_italic { font-style: italic; } #trvhlcrcph .gt_super { font-size: 65%; } #trvhlcrcph .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #trvhlcrcph .gt_asterisk { font-size: 100%; vertical-align: 0; } #trvhlcrcph .gt_indent_1 { text-indent: 5px; } #trvhlcrcph .gt_indent_2 { text-indent: 10px; } #trvhlcrcph .gt_indent_3 { text-indent: 15px; } #trvhlcrcph .gt_indent_4 { text-indent: 20px; } #trvhlcrcph .gt_indent_5 { text-indent: 25px; } #trvhlcrcph .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #trvhlcrcph div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } variable 0 0.25 0.5 0.75 1 expend 3.7 4.9 5.8 6.4 9.8 sat 844.0 897.2 945.5 1032.0 1107.0 Esta variación es considerable para promedios del SAT: el percentil 75 es alrededor de 1050 puntos, mientras que el percentil 25 corresponde a alrededor de 800. Igualmente, hay diferencias considerables de gasto por alumno (miles de dólares) a lo largo de los estados. Ahora hacemos nuestro primer ejercico de comparación: ¿Cómo se ven las calificaciones para estados en distintos niveles de gasto? Podemos usar una gráfica de dispersión: library(ggrepel) ggplot(sat, aes(x = expend, y = sat, label = state)) + geom_point(colour = &quot;red&quot;, size = 2) + geom_text_repel(colour = &quot;gray50&quot;) + xlab(&quot;Gasto por alumno (miles de dólares)&quot;) + ylab(&quot;Calificación promedio en SAT&quot;) Estas comparaciones no son de alta calidad, solo estamos usando 2 variables —que son muy pocas— y no hay mucho que podamos decir en cuanto explicación. Sin duda nos hace falta una imagen más completa. Necesitaríamos entender la correlación que existe entre las demás características de nuestras unidades de estudio. Las unidades que estamos comparando pueden diferir fuertemente en otras propiedades importantes (o dimensiones), lo cual no permite interpretar la gráfica de manera sencilla. Una variable que tenemos es el porcentaje de alumnos de cada estado que toma el SAT. Podemos agregar como sigue: ggplot(sat, aes(x = expend, y = math, label=state, colour = frac)) + geom_point() + geom_text_repel() + xlab(&quot;Gasto por alumno (miles de dólares)&quot;) + ylab(&quot;Calificación promedio en SAT&quot;) Esto nos permite entender por qué nuestra comparación inicial es relativamente pobre. Los estados con mejores resultados promedio en el SAT son aquellos donde una fracción relativamente baja de los estudiantes toma el examen. La diferencia es considerable. En este punto podemos hacer varias cosas. Una primera idea es intentar comparar estados más similares en cuanto a la población de alumnos que asiste. Podríamos hacer grupos como sigue: set.seed(991) k_medias_sat &lt;- kmeans(sat |&gt; select(frac), centers = 4, nstart = 100, iter.max = 100) sat$clase &lt;- k_medias_sat$cluster sat &lt;- sat |&gt; group_by(clase) |&gt; mutate(clase_media = round(mean(frac))) |&gt; ungroup() |&gt; mutate(clase_media = factor(clase_media)) sat &lt;- sat |&gt; mutate(rank_p = rank(frac, ties= &quot;first&quot;) / length(frac)) ggplot(sat, aes(x = rank_p, y = frac, label = state, colour = clase_media)) + geom_point(size = 2) Estos resultados indican que es más probable que buenos alumnos decidan hacer el SAT. Lo interesante es que esto ocurre de manera diferente en cada estado. Por ejemplo, en algunos estados era más común otro examen: el ACT. Si hacemos clusters de estados según el % de alumnos, empezamos a ver otra historia. Para esto, ajustemos rectas de mínimos cuadrados como referencia: Esto da una imagen muy diferente a la que originalmente planteamos. Nota que dependiendo de cómo categorizamos, esta gráfica puede variar (puedes intentar con más o menos grupos, por ejemplo). Tablas de conteos Consideremos los siguientes datos de tomadores de té (del paquete FactoMineR (Lê et al. 2008)): tea &lt;- read_csv(&quot;data/tea.csv&quot;) # nombres y códigos te &lt;- tea |&gt; select(how, price, sugar) |&gt; rename(presentacion = how, precio = price, azucar = sugar) |&gt; mutate( presentacion = fct_recode(presentacion, suelto = &quot;unpackaged&quot;, bolsas = &quot;tea bag&quot;, mixto = &quot;tea bag+unpackaged&quot;), precio = fct_recode(precio, marca = &quot;p_branded&quot;, variable = &quot;p_variable&quot;, barato = &quot;p_cheap&quot;, marca_propia = &quot;p_private label&quot;, desconocido = &quot;p_unknown&quot;, fino = &quot;p_upscale&quot;), azucar = fct_recode(azucar, sin_azúcar = &quot;No.sugar&quot;, con_azúcar = &quot;sugar&quot;)) sample_n(te, 10) ## # A tibble: 10 × 3 ## presentacion precio azucar ## &lt;fct&gt; &lt;fct&gt; &lt;fct&gt; ## 1 bolsas marca sin_azúcar ## 2 bolsas variable sin_azúcar ## 3 bolsas marca con_azúcar ## 4 bolsas fino con_azúcar ## 5 mixto variable con_azúcar ## 6 mixto fino con_azúcar ## 7 bolsas marca sin_azúcar ## 8 bolsas fino sin_azúcar ## 9 mixto variable con_azúcar ## 10 mixto variable sin_azúcar Nos interesa ver qué personas compran té suelto, y de qué tipo. Empezamos por ver las proporciones que compran té según su empaque (en bolsita o suelto): precio &lt;- te |&gt; count(precio) |&gt; mutate(prop = round(100 * n / sum(n))) |&gt; select(-n) tipo &lt;- te |&gt; count(presentacion) |&gt; mutate(pct = round(100 * n / sum(n))) tipo |&gt; gt() #wiljiluquk table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #wiljiluquk thead, #wiljiluquk tbody, #wiljiluquk tfoot, #wiljiluquk tr, #wiljiluquk td, #wiljiluquk th { border-style: none; } #wiljiluquk p { margin: 0; padding: 0; } #wiljiluquk .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #wiljiluquk .gt_caption { padding-top: 4px; padding-bottom: 4px; } #wiljiluquk .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #wiljiluquk .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #wiljiluquk .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #wiljiluquk .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #wiljiluquk .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #wiljiluquk .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #wiljiluquk .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #wiljiluquk .gt_column_spanner_outer:first-child { padding-left: 0; } #wiljiluquk .gt_column_spanner_outer:last-child { padding-right: 0; } #wiljiluquk .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #wiljiluquk .gt_spanner_row { border-bottom-style: hidden; } #wiljiluquk .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #wiljiluquk .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #wiljiluquk .gt_from_md > :first-child { margin-top: 0; } #wiljiluquk .gt_from_md > :last-child { margin-bottom: 0; } #wiljiluquk .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #wiljiluquk .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #wiljiluquk .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #wiljiluquk .gt_row_group_first td { border-top-width: 2px; } #wiljiluquk .gt_row_group_first th { border-top-width: 2px; } #wiljiluquk .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #wiljiluquk .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #wiljiluquk .gt_first_summary_row.thick { border-top-width: 2px; } #wiljiluquk .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #wiljiluquk .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #wiljiluquk .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #wiljiluquk .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #wiljiluquk .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #wiljiluquk .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #wiljiluquk .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #wiljiluquk .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #wiljiluquk .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #wiljiluquk .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #wiljiluquk .gt_left { text-align: left; } #wiljiluquk .gt_center { text-align: center; } #wiljiluquk .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #wiljiluquk .gt_font_normal { font-weight: normal; } #wiljiluquk .gt_font_bold { font-weight: bold; } #wiljiluquk .gt_font_italic { font-style: italic; } #wiljiluquk .gt_super { font-size: 65%; } #wiljiluquk .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #wiljiluquk .gt_asterisk { font-size: 100%; vertical-align: 0; } #wiljiluquk .gt_indent_1 { text-indent: 5px; } #wiljiluquk .gt_indent_2 { text-indent: 10px; } #wiljiluquk .gt_indent_3 { text-indent: 15px; } #wiljiluquk .gt_indent_4 { text-indent: 20px; } #wiljiluquk .gt_indent_5 { text-indent: 25px; } #wiljiluquk .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #wiljiluquk div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } presentacion n pct bolsas 170 57 mixto 94 31 suelto 36 12 La mayor parte de las personas toma té en bolsas. Sin embargo, el tipo de té (en términos de precio o marca) que compran es muy distinto dependiendo de la presentación: tipo &lt;- tipo |&gt; select(presentacion, prop_presentacion = pct) tabla_cruzada &lt;- te |&gt; count(presentacion, precio) |&gt; # porcentajes por presentación group_by(presentacion) |&gt; mutate(prop = round(100 * n / sum(n))) |&gt; select(-n) tabla_cruzada |&gt; pivot_wider(names_from = presentacion, values_from = prop, values_fill = list(prop = 0)) |&gt; gt() #sxkvcchtyg table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #sxkvcchtyg thead, #sxkvcchtyg tbody, #sxkvcchtyg tfoot, #sxkvcchtyg tr, #sxkvcchtyg td, #sxkvcchtyg th { border-style: none; } #sxkvcchtyg p { margin: 0; padding: 0; } #sxkvcchtyg .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #sxkvcchtyg .gt_caption { padding-top: 4px; padding-bottom: 4px; } #sxkvcchtyg .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #sxkvcchtyg .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #sxkvcchtyg .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #sxkvcchtyg .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sxkvcchtyg .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #sxkvcchtyg .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #sxkvcchtyg .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #sxkvcchtyg .gt_column_spanner_outer:first-child { padding-left: 0; } #sxkvcchtyg .gt_column_spanner_outer:last-child { padding-right: 0; } #sxkvcchtyg .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #sxkvcchtyg .gt_spanner_row { border-bottom-style: hidden; } #sxkvcchtyg .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #sxkvcchtyg .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #sxkvcchtyg .gt_from_md > :first-child { margin-top: 0; } #sxkvcchtyg .gt_from_md > :last-child { margin-bottom: 0; } #sxkvcchtyg .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #sxkvcchtyg .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #sxkvcchtyg .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #sxkvcchtyg .gt_row_group_first td { border-top-width: 2px; } #sxkvcchtyg .gt_row_group_first th { border-top-width: 2px; } #sxkvcchtyg .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #sxkvcchtyg .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #sxkvcchtyg .gt_first_summary_row.thick { border-top-width: 2px; } #sxkvcchtyg .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sxkvcchtyg .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #sxkvcchtyg .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #sxkvcchtyg .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #sxkvcchtyg .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #sxkvcchtyg .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #sxkvcchtyg .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #sxkvcchtyg .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #sxkvcchtyg .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #sxkvcchtyg .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #sxkvcchtyg .gt_left { text-align: left; } #sxkvcchtyg .gt_center { text-align: center; } #sxkvcchtyg .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #sxkvcchtyg .gt_font_normal { font-weight: normal; } #sxkvcchtyg .gt_font_bold { font-weight: bold; } #sxkvcchtyg .gt_font_italic { font-style: italic; } #sxkvcchtyg .gt_super { font-size: 65%; } #sxkvcchtyg .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #sxkvcchtyg .gt_asterisk { font-size: 100%; vertical-align: 0; } #sxkvcchtyg .gt_indent_1 { text-indent: 5px; } #sxkvcchtyg .gt_indent_2 { text-indent: 10px; } #sxkvcchtyg .gt_indent_3 { text-indent: 15px; } #sxkvcchtyg .gt_indent_4 { text-indent: 20px; } #sxkvcchtyg .gt_indent_5 { text-indent: 25px; } #sxkvcchtyg .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #sxkvcchtyg div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } precio bolsas mixto suelto marca 41 21 14 barato 3 1 3 marca_propia 9 4 3 desconocido 6 1 0 fino 8 20 56 variable 32 52 25 Estos datos podemos examinarlos un rato y llegar a conclusiones, pero esta tabla no necesariamente es la mejor manera de mostrar patrones en los datos. Tampoco son muy útiles gráficas como la siguiente: ggplot(tabla_cruzada |&gt; ungroup() |&gt; mutate(price = fct_reorder(precio, prop)), aes(x = precio, y = prop, group = presentacion, colour = presentacion)) + geom_point() + coord_flip() + geom_line() En lugar de eso, calcularemos perfiles columna. Esto es, comparamos cada una de las columnas con la columna marginal (en la tabla de tipo de estilo de té): num_grupos &lt;- n_distinct(te |&gt; select(presentacion)) tabla &lt;- te |&gt; count(presentacion, precio) |&gt; group_by(presentacion) |&gt; mutate(prop_precio = (100 * n / sum(n))) |&gt; group_by(precio) |&gt; mutate(prom_prop = sum(prop_precio)/num_grupos) |&gt; mutate(perfil = 100 * (prop_precio / prom_prop - 1)) tabla ## # A tibble: 17 × 6 ## # Groups: precio [6] ## presentacion precio n prop_precio prom_prop perfil ## &lt;fct&gt; &lt;fct&gt; &lt;int&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 bolsas marca 70 41.2 25.4 61.8 ## 2 bolsas barato 5 2.94 2.26 30.1 ## 3 bolsas marca_propia 16 9.41 5.48 71.7 ## 4 bolsas desconocido 11 6.47 2.51 158. ## 5 bolsas fino 14 8.24 28.0 -70.6 ## 6 bolsas variable 54 31.8 36.3 -12.5 ## 7 mixto marca 20 21.3 25.4 -16.4 ## 8 mixto barato 1 1.06 2.26 -52.9 ## 9 mixto marca_propia 4 4.26 5.48 -22.4 ## 10 mixto desconocido 1 1.06 2.51 -57.6 ## 11 mixto fino 19 20.2 28.0 -27.8 ## 12 mixto variable 49 52.1 36.3 43.6 ## 13 suelto marca 5 13.9 25.4 -45.4 ## 14 suelto barato 1 2.78 2.26 22.9 ## 15 suelto marca_propia 1 2.78 5.48 -49.3 ## 16 suelto fino 20 55.6 28.0 98.4 ## 17 suelto variable 9 25 36.3 -31.1 tabla_perfil &lt;- tabla |&gt; select(presentacion, precio, perfil, pct = prom_prop) |&gt; pivot_wider(names_from = presentacion, values_from = perfil, values_fill = list(perfil = -100.0)) if_profile &lt;- function(x){ any(x &lt; 0) &amp; any(x &gt; 0) } marcar &lt;- marcar_tabla_fun(25, &quot;red&quot;, &quot;black&quot;) tab_out &lt;- tabla_perfil |&gt; arrange(desc(bolsas)) |&gt; select(-pct, everything()) |&gt; mutate(across(where(is.numeric), \\(x) round(x, 0))) |&gt; mutate(across(where(if_profile), \\(x) marcar(x))) |&gt; knitr::kable(format_table_salida(), escape = FALSE, digits = 0, booktabs = T) |&gt; kableExtra::kable_styling(latex_options = c(&quot;striped&quot;, &quot;scale_down&quot;), bootstrap_options = c( &quot;hover&quot;, &quot;condensed&quot;), full_width = FALSE) tab_out precio bolsas mixto suelto pct desconocido 158 -58 -100 3 marca_propia 72 -22 -49 5 marca 62 -16 -45 25 barato 30 -53 23 2 variable -12 44 -31 36 fino -71 -28 98 28 Leemos esta tabla como sigue: por ejemplo, los compradores de té suelto compran té fino a una tasa casi el doble (98%) que el promedio. También podemos graficar como: tabla_graf &lt;- tabla_perfil |&gt; ungroup() |&gt; mutate(precio = fct_reorder(precio, bolsas)) |&gt; select(-pct) |&gt; pivot_longer(cols = -precio, names_to = &quot;presentacion&quot;, values_to = &quot;perfil&quot;) g_perfil &lt;- ggplot(tabla_graf, aes(x = precio, xend = precio, y = perfil, yend = 0, group = presentacion)) + geom_point() + geom_segment() + facet_wrap(~presentacion) + geom_hline(yintercept = 0 , colour = &quot;gray&quot;)+ coord_flip() g_perfil Observación: hay dos maneras de construir la columna promedio: tomando los porcentajes sobre todos los datos, o promediando los porcentajes de las columnas. Si los grupos de las columnas están desbalanceados, estos promedios son diferentes. Cuando usamos porcentajes sobre la población, perfiles columna y renglón dan el mismo resultado Sin embargo, cuando hay un grupo considerablemente más grande que otros, las comparaciones se vuelven vs este grupo particular. No siempre queremos hacer esto. Interpretación En el último ejemplo de tomadores de té utilizamos una muestra de personas, no toda la población de tomadores de té. Eso quiere decir que tenemos cierta incertidumbre de cómo se generalizan o no los resultados que obtuvimos en nuestro análisis a la población general. Nuestra respuesta depende de cómo se extrajo la muestra que estamos considerando. Si el mecanismo de extracción incluye algún proceso probabilístico, entonces es posible en principio entender qué tan bien generalizan los resultados de nuestro análisis a la población general, y entender esto depende de entender qué tanta variación hay de muestra a muestra, de todas las posibles muestras que pudimos haber extraido. En las siguientes secciones discutiremos estos aspectos, en los cuales pasamos del trabajo de “detective” al trabajo de “juez” en nuestro trabajo analítico. Suavizamiento loess Las gráficas de dispersión son la herramienta básica para describir la relación entre dos variables cuantitativas, y como vimos en ejemplo anteriores, muchas veces podemos apreciar mejor la relación entre ellas si agregamos una curva loess. Veamos un ejemplo, los siguientes datos muestran los premios ofrecidos y las ventas totales de una lotería a lo largo de 53 sorteos (las unidades son cantidades de dinero indexadas). Graficamos en escalas logarítmicas y agregamos una curva loess. # cargamos los datos load(here::here(&quot;data&quot;, &quot;ventas_sorteo.Rdata&quot;)) ggplot(ventas.sorteo, aes(x = premio, y = ventas.tot.1)) + geom_point() + geom_smooth(method = &quot;loess&quot;, span = 0.6, method.args = list(degree = 1), se = FALSE) + scale_x_log10(breaks = c(20000, 40000, 80000)) + scale_y_log10(breaks = c(10000, 15000, 22000, 33000)) El patrón no era difícil de ver en los datos originales, sin embargo, la curva lo hace más claro, el logaritmo de las ventas tiene una relación no lineal con el logaritmo del premio: para premios no muy grandes no parece haber gran diferencia, pero cuando los premios empiezan a crecer por encima de 20,000, las ventas crecen más rápidamente que para premios menores. Este efecto se conoce como bola de nieve, y es frecuente en este tipo de loterías. Antes de adentrarnos a los modelos loess comenzamos explicando cómo se ajustan familias paramétricas de curvas a conjuntos de datos dados. El modelo de regresion lineal ajusta una recta a un conjunto de datos. Por ejemplo, consideremos la familia \\[f_{a,b}(x) = a x + b,\\] para un conjunto de datos bivariados \\(\\{ (x_1, y_1), \\ldots, (x_N, y_N)\\}\\). Buscamos encontrar \\(a\\) y \\(b\\) tales que \\(f_{a,b}\\) de un ajuste óptimo a los datos. Para esto, se minimiza la suma de errores cuadráticos \\[\\frac1N \\sum_{n = 1}^N ( y_n - a x_n - b)^2.\\] En este caso, las constantes \\(a\\) y \\(b\\) se pueden encontrar diferenciando la función de mínimos cuadrados. Nótese que podemos repetir el argumento con otras familias de funciones (por ejemplo cuadráticas). ggplot(ventas.sorteo, aes(x = premio, y = ventas.tot.1)) + geom_point() + geom_smooth(method = &quot;lm&quot;, se = FALSE) + scale_x_log10(breaks = c(20000, 40000, 80000)) + scale_y_log10(breaks = c(10000, 15000, 22000, 33000)) Si observamos la gráfica notamos que este modelo lineal (en los logaritmos) no resumen adecuadamente estos datos. Podríamos experimentar con otras familias (por ejemplo, una cuadrática o cúbica, potencias, exponenciales, etc.); sin embargo, en la etapa exploratoria es mejor tomar una ruta de ajuste más flexible y robusta. Regresión local nos provee de un método con estas características: Curvas loess (regresión local): Una manera de mejorar la flexibilidad de los modelos lineales es considerar rectas de manera local. Es decir, en cada \\(x\\) posible consideramos cuál es la recta que mejor ajusta a los datos, considerando solamente valores de \\(x_n\\) que están cercanos a \\(x\\). La siguiente gráfica muestra qué recta se ajusta alrededor de cada punto, y cómo queda el suavizador completo, con distintos valores de suavizamiento. El tono de los puntos indican en cada paso que ventana de datos es considerada: Escogiendo de los parámetros. El parámetro de suavizamiento se encuentra por ensayo y error. La idea general es que debemos encontrar una curva que explique patrones importantes en los datos (que ajuste los datos) pero que no muestre variaciones a escalas más chicas difíciles de explicar (que pueden ser el resultado de influencias de otras variables, variación muestral, ruido o errores de redondeo, por ejemplo). En el proceso de prueba y error iteramos el ajuste y en cada paso hacemos análisis de residuales, con el fin de seleccionar un suavizamiento adecuado. En lugar de usar ajustes locales lineales, podemos usar ajustes locales cuadráticos que nos permiten capturar formas locales cuadráticas sin tener que suavizar demasiado poco: Opcional: cálculo del suavizador La idea es producir ajustes locales de rectas o funciones lineales o cuadráticas. Consideremos especificar dos parámetros: Parámetro de suavizamiento \\(\\alpha\\): toma valores en \\((0,1)\\), cuando \\(\\alpha\\) es más grande, la curva ajustada es más suave. Grado de los polinomios locales que ajustamos \\(\\lambda\\): generalmente se toma \\(\\lambda=1,2\\). Entonces, supongamos que los datos están dados por \\((x_1,y_1), \\ldots, (x_N, y_N)\\), y sean \\(\\alpha\\) un parámetro de suavizamiento fijo, y \\(\\lambda=1\\). Denotamos como \\(\\hat{g}(x)\\) la curva loess ajustada, y como \\(w_n(x)\\) a una función de peso (que depende de x) para la observación \\((x_n, y_n)\\). Para poder calcular \\(w_n(x)\\) debemos comenzar calculando \\(q=\\lfloor{N\\alpha}\\rfloor\\) que suponemos mayor que uno, esta \\(q\\) es el número de puntos que se utilizan en cada ajuste local. Ahora definimos la función tricubo: \\[\\begin{align} T(u)=\\begin{cases} (1-|u|^3)^3, &amp; \\text{para $|u| &lt; 1$}.\\\\ 0, &amp; \\text{en otro caso}. \\end{cases} \\end{align}\\] entonces, para el punto \\(x\\) definimos el peso correspondiente al dato \\((x_n,y_n)\\), denotado por \\(w_n(x)\\) como: \\[w_n(x)=T\\bigg(\\frac{|x-x_n|}{d_q(x)}\\bigg)\\] donde \\(d_q(x)\\) es el valor de la \\(q\\)-ésima distancia más chica (la más grande entre las \\(q\\) más chicas) entre los valores \\(|x-x_j|\\), \\(j=1,\\ldots,N\\). De esta forma, las observaciones \\(x_n\\) reciben más peso cuanto más cerca estén de \\(x\\). En palabras, de \\(x_1,...,x_N\\) tomamos los \\(q\\) datos más cercanos a \\(x\\), que denotamos \\(x_{i_1}(x) \\leq x_{i_2}(x) \\leq \\cdots \\leq x_{i_q}(x)\\). Los re-escalamos a \\([0,1]\\) haciendo corresponder \\(x\\) a \\(0\\) y el punto más alejado de \\(x\\) (que es \\(x_{i_q}\\)) a 1. Aplicamos el tricubo (gráfica de abajo), para encontrar los pesos de cada punto. Los puntos que están a una distancia mayor a \\(d_q(x)\\) reciben un peso de cero, y los más cercanos un peso que depende de que tan cercanos están a \\(x\\). Nótese que \\(x\\) es el punto ancla en dónde estamos ajustando la regresión local. tricubo &lt;- function(x) { ifelse(abs(x) &lt; 1, (1 - abs(x) ^ 3) ^ 3, 0) } curve(tricubo, from = -1.5, to = 1.5) Finalmente, para cada valor de \\(x_k\\) que está en el conjunto de datos \\(\\{x_1,...,x_n\\}\\), ajustamos una recta de mínimos cuadrados ponderados por los pesos \\(w_n(x)\\), es decir, minimizamos (en el caso lineal): \\[\\sum_{i=1}^nw_n(x_k)(y_i-ax_n-b)^2.\\] Observaciones: Cualquier función (continua y quizás diferenciable) con la forma de flan del tricubo que se desvanece fuera de \\((-1,1),\\) es creciente en \\((-1,0)\\) y decreciente en \\((0, 1)\\) es un buen candidato para usarse en lugar del tricubo. La razón por la que escogemos precisamente esta forma algebráica no tiene que ver con el análisis exploratorio, sino con las ventajas teóricas adicionales que tiene en la inferencia. El caso \\(\\lambda=2\\) es similar. La única diferencia es en el paso de ajuste, donde usamos funciones cuadráticas, y obtendríamos entonces tres familias de parámetros \\(a(x_k), b_1(x_k), b_2(x_k),\\) para cada \\(k \\in \\{1, \\ldots, N\\}\\). Caso de estudio: nacimientos en México Podemos usar el suavizamiento loess para entender y describir el comportamiento de series de tiempo, en las cuáles intentamos entender la dependencia de una serie de mediciones indexadas por el tiempo. Típicamente es necesario utilizar distintas componentes para describir exitosamente una serie de tiempo, y para esto usamos distintos tipos de suavizamientos. Veremos que distintas componentes varían en distintas escalas de tiempo (unas muy lentas, como la tendencia, otras más rapidamente, como variación quincenal, etc.). Este caso de estudio esta basado en un análisis propuesto por A. Vehtari y A. Gelman, junto con un análisis de serie de tiempo de William S. Cleveland (1993). En nuestro caso, usaremos los datos de nacimientos registrados por día en México desde 1999. Los usaremos para contestar las preguntas: ¿cuáles son los cumpleaños más frecuentes? y ¿en qué mes del año hay más nacimientos? Podríamos utilizar una gráfica popular (ver por ejemplo esta visualización) como: Sin embargo, ¿cómo criticarías este análisis desde el punto de vista de los tres primeros principios del diseño analítico? ¿Las comparaciones son útiles? ¿Hay aspectos multivariados? ¿Qué tan bien explica o sugiere estructura, mecanismos o causalidad? Datos de natalidad para México library(lubridate) library(ggthemes) theme_set(theme_minimal(base_size = 14)) natalidad &lt;- read_rds(&quot;./data/nacimientos/natalidad.rds&quot;) |&gt; mutate(dia_semana = weekdays(fecha)) |&gt; mutate(dia_año = yday(fecha)) |&gt; mutate(año = year(fecha)) |&gt; mutate(mes = month(fecha)) |&gt; ungroup() |&gt; mutate(dia_semana = recode(dia_semana, Monday = &quot;Lunes&quot;, Tuesday = &quot;Martes&quot;, Wednesday = &quot;Miércoles&quot;, Thursday = &quot;Jueves&quot;, Friday = &quot;Viernes&quot;, Saturday = &quot;Sábado&quot;, Sunday = &quot;Domingo&quot;)) |&gt; #necesario pues el LOCALE puede cambiar mutate(dia_semana = recode(dia_semana, lunes = &quot;Lunes&quot;, martes = &quot;Martes&quot;, miércoles = &quot;Miércoles&quot;, jueves = &quot;Jueves&quot;, viernes = &quot;Viernes&quot;, sábado = &quot;Sábado&quot;, domingo = &quot;Domingo&quot;)) |&gt; mutate(dia_semana = fct_relevel(dia_semana, c(&quot;Lunes&quot;, &quot;Martes&quot;, &quot;Miércoles&quot;, &quot;Jueves&quot;, &quot;Viernes&quot;, &quot;Sábado&quot;, &quot;Domingo&quot;))) Consideremos los datos agregados del número de nacimientos (registrados) por día desde 1999 hasta 2016. Un primer intento podría ser hacer una gráfica de la serie de tiempo. Sin embargo, vemos que no es muy útil: Hay varias características que notamos. Primero, parece haber una tendencia ligeramente decreciente del número de nacimientos a lo largo de los años. Segundo, la gráfica sugiere un patrón anual. Y por último, encontramos que hay dispersión producida por los días de la semana. Sólo estas características hacen que la comparación entre días sea difícil de realizar. Supongamos que comparamos el número de nacimientos de dos miércoles dados. Esa comparación será diferente dependiendo: del año donde ocurrieron, el mes donde ocurrieron, si semana santa ocurrió en algunos de los miércoles, y así sucesivamente. Como en nuestros ejemplos anteriores, la idea del siguiente análisis es aislar las componentes que observamos en la serie de tiempo: extraemos componentes ajustadas, y luego examinamos los residuales. En este caso particular, asumiremos una descomposición aditiva de la serie de tiempo (William S. Cleveland 1993). En el estudio de series de tiempo una estructura común es considerar el efecto de diversos factores como tendencia, estacionalidad, ciclicidad e irregularidades de manera aditiva. Esto es, consideramos la descomposición \\[\\begin{align} y(t) = f_{t}(t) + f_{e}(t) + f_{c}(t) + \\varepsilon. \\end{align}\\] Una estrategia de ajuste, como veremos más adelante, es proceder de manera modular. Es decir, se ajustan los componentes de manera secuencial considerando los residuales de los anteriores. Tendencia Comenzamos por extraer la tendencia, haciendo promedios loess (William S. Cleveland 1979) con vecindades relativamente grandes. Quizá preferiríamos suavizar menos para capturar más variación lenta, pero si hacemos esto en este punto empezamos a absorber parte de la componente anual: mod_1 &lt;- loess(n ~ as.numeric(fecha), data = natalidad, span = 0.2, degree = 1) datos_dia &lt;- natalidad |&gt; mutate(ajuste_1 = fitted(mod_1)) |&gt; mutate(res_1 = n - ajuste_1) Notemos que a principios de 2000 el suavizador está en niveles de alrededor de 7000 nacimientos diarios, hacia 2015 ese número es más cercano a unos 6000. Componente anual Al obtener la tendencia podemos aislar el efecto a largo plazo y proceder a realizar mejores comparaciones (por ejemplo, comparar un día de 2000 y de 2015 tendria más sentido). Ahora, ajustamos los residuales del suavizado anterior, pero con menos suavizamiento. Así evitamos capturar tendencia: mod_anual &lt;- loess(res_1 ~ as.numeric(fecha), data = datos_dia, degree = 2, span = 0.005) datos_dia &lt;- datos_dia |&gt; mutate(ajuste_2 = fitted(mod_anual)) |&gt; mutate(res_2 = res_1 - ajuste_2) Día de la semana Hasta ahora, hemos aislado los efectos por plazos largos de tiempo (tendencia) y hemos incorporado las variaciones estacionales (componente anual) de nuestra serie de tiempo. Ahora, veremos cómo capturar el efecto por día de la semana. En este caso, podemos hacer suavizamiento loess para cada serie de manera independiente datos_dia &lt;- datos_dia |&gt; group_by(dia_semana) |&gt; nest() |&gt; mutate(ajuste_mod = map(data, ~ loess(res_2 ~ as.numeric(fecha), data = .x, span = 0.1, degree = 1))) |&gt; mutate(ajuste_3 = map(ajuste_mod, fitted)) |&gt; select(-ajuste_mod) |&gt; unnest(cols = c(data, ajuste_3)) |&gt; mutate(res_3 = res_2 - ajuste_3) |&gt; ungroup() Residuales Por último, examinamos los residuales finales quitando los efectos ajustados: ## `geom_smooth()` using formula = &#39;y ~ x&#39; Observación: nótese que la distribución de estos residuales presenta irregularidades interesantes. La distribución es de colas largas, y no se debe a unos cuantos datos atípicos. Esto generalmente es indicación que hay factores importantes que hay que examinar mas a detalle en los residuales: Reestimación Cuando hacemos este proceso secuencial de llevar el ajuste a los residual, a veces conviene iterarlo. La razón es que en una segunda o tercera pasada podemos hacer mejores estimaciones de cada componente, y es posible suavizar menos sin capturar componentes de más alta frecuencia. Así que podemos regresar a la serie original para hacer mejores estimaciones, más suavizadas: # Quitamos componente anual y efecto de día de la semana datos_dia &lt;- datos_dia |&gt; mutate(n_1 = n - ajuste_2 - ajuste_3) # Reajustamos mod_1 &lt;- loess(n_1 ~ as.numeric(fecha), data = datos_dia, span = 0.02, degree = 2, family = &quot;symmetric&quot;) Y ahora repetimos con la componente de día de la semana: Análisis de componentes Ahora comparamos las componentes estimadas y los residuales en una misma gráfica. Por definición, la suma de todas estas componentes da los datos originales. Este último paso nos permite diversas comparaciones que explican la variación que vimos en los datos. Una gran parte de los residuales está entre \\(\\pm 250\\) nacimientos por día. Sin embargo, vemos que las colas tienen una dispersión mucho mayor: quantile(datos_dia$res_6, c(00, .01,0.05, 0.10, 0.90, 0.95, 0.99, 1)) |&gt; round() ## 0% 1% 5% 10% 90% 95% 99% 100% ## -2238 -1134 -315 -202 188 268 516 2521 ¿A qué se deben estas colas tan largas? Viernes 13? Podemos empezar con una curosidad. Los días Viernes o Martes 13, ¿nacen menos niños? Nótese que fue útil agregar el indicador de Semana santa por el Viernes 13 de Semana Santa que se ve como un atípico en el panel de los viernes 13. Residuales: antes y después de 2006 Veamos primero una agregación sobre los años de los residuales. Lo primero es observar un cambio que sucedió repentinamente en 2006: La razón es un cambio en la ley acerca de cuándo pueden entrar los niños a la primaria. Antes era por edad y había poco margen. Ese exceso de nacimientos son reportes falsos para que los niños no tuvieran que esperar un año completo por haber nacido unos cuantos días después de la fecha límite. Otras características que debemos investigar: Efectos de Año Nuevo, Navidad, Septiembre 16 y otros días feriados como Febrero 14. Semana santa: como la fecha cambia, vemos que los residuales negativos tienden a ocurrir dispersos alrededor del día 100 del año. Otros días especiales: más de residuales Ahora promediamos residuales (es posible agregar barras para indicar dispersión a lo largo de los años) para cada día del año. Podemos identificar ahora los residuales más grandes: se deben, por ejemplo, a días feriados, con consecuencias adicionales que tienen en días ajuntos (excesos de nacimientos): Semana santa Para Semana Santa tenemos que hacer unos cálculos. Si alineamos los datos por días antes de Domingo de Pascua, obtenemos un patrón de caída fuerte de nacimientos el Viernes de Semana Santa, y la característica forma de “valle con hombros” en días anteriores y posteriores estos Viernes. ¿Por qué ocurre este patrón? Nótese un defecto de nuestro modelo: el patrón de “hombros” alrededor del Viernes Santo no es suficientemente fuerte para equilibrar los nacimientos faltantes. ¿Cómo podríamos mejorar nuestra descomposición? Referencias "],["tipos-de-estudio-y-experimentos.html", "Sección 2 Tipos de estudio y experimentos Muestreo aleatorio Pero si no podemos hacer muestreo aleatorio? El estimador estándar Experimentos tradicionales Bloqueo Variables desconocidas Aleatorizando el tratamiento Selección de unidades y tratamiento Asignación natural del tratamiento", " Sección 2 Tipos de estudio y experimentos Motivación Pregunta de entrevista de Google (Chihara and Hesterberg 2018) Imagina que eres consultor y te preguntan lo siguiente (ver siguiente figura): Estoy haciendo una comparación de antes y después donde la hipótesis alternativa es pre.media.error &gt; post.media.error. La distribución de ambas muestras es sesgada a la derecha. ¿Qué prueba me recomiendas para ésta situación? Figure 2.1: Error CPR, gráfica de densidad. Far better an approximate answer to the right question, which is often vague, than an exact answer to the wrong question, which can always be made precise. — John Tukey La siguiente imagen Roger Peng representa una situación común a la que se enfrenta el analista de datos, y se desarrolló en el contexto de preguntas vagas. En el esquema hay tres caminos: uno es uno ideal que pocas veces sucede, otro produce respuestas poco útiles pero es fácil, y otro es tortuoso pero que caracteriza el mejor trabajo de análisis de datos: Figure 2.2: Adaptado de R. Peng: Tukey, design thinking and better questions. Ejemplos: Alguien nos pregunta cuáles son las tiendas que mas venden de una cadena. Podríamos consultar bases de datos, hacer extracciones, definir periodos, etc. y reportar el promedio de ventas en el último mes, esta respuesta probablemente es poco útil. Nos damos cuenta, por ejemplo, porque la peor tienda es una que abrió hace relativamente poco, y la mejor es una de las tiendas más grandes que está en una zona de tráfico de alto costo. Una pregunta más interesante es, ¿qué equipos de ventas tienen mejor desempeño? ¿Cuánto aporta tener una cafetería dentro de la tienda en términos de ventas?, etc. Proceso Generador de Datos Entre las preguntas que se debe hacer el analista de datos una fundamental es entender el proceso generador de datos, pues esto determinará que otras preguntas son relevantes, y que análisis son adecuados, tanto en términos prácticos como estadísticos. La inferencia estadística busca hacer afirmaciones, cuantificadas de manera probabilista, acerca de datos que no tenemos, usando regularidades y conocimiento de datos que sí tenemos disponibles y métodos cuantitativos. Para hacer afirmaciones inferenciales eficientes y bien calibradas (con garantías estadísticas de calibración) a preguntas donde queremos generalizar de muestra a población, se requiere conocer con precisión el proceso que genera los datos muestrales. Esto incluye saber con detalle cómo se seleccionaron los datos a partir de los que se quiere hacer inferencia. En este caso, eficiente quiere decir que aprovechamos toda la información que está en los datos observados de manera que nuestros rangos de incertidumbre son lo más chico posibles (además de estar correctamente calibrados). Por su parte, probabilísticamente bien calibrados se refiere a que, lo que decimos que puede ocurrir con 10% de probabilidad ocurre efectivamente 1 de cada 10 veces, si decimos 20% entonces ocurre 2 de 20, etc. Veremos que para muestras dadas naturalmente, a veces es muy difiícil entender a fondo el proceso que generó la muestra y por tanto no tenemos las garantías de eficiencia y calibración. Ejemplo: Prevalencia de anemia Supongamos que nos interesa conocer el porcentaje de menores en edad escolar, (entre 6 y 15 años), con anemia en México. La fuente de datos disponible corresponde a registros del IMSS de hospitalizaciones de menores, ya sea por anemia o por otra causa (infecciones gastrointestinales, apendicitis, tratamiento de leucemia, …), se registró si el menor tenía anemia. En nuestra muestra el 47% de los niños tiene anemia. head(paciente) #&gt; # A tibble: 6 × 4 #&gt; edad padecimiento sexo anemia #&gt; &lt;int&gt; &lt;chr&gt; &lt;chr&gt; &lt;int&gt; #&gt; 1 8 picadura alacrán mujer 0 #&gt; 2 10 infección intestinal hombre 1 #&gt; 3 7 mordedura de perro hombre 1 #&gt; 4 8 asma hombre 1 #&gt; 5 13 infección intestinal mujer 0 #&gt; 6 7 picadura alacrán hombre 0 ¿Qué nos dice esta cantidad acerca de la anemia en la población? ¿Podemos hacer inferencia estadística? ¿Cómo calculamos intervalos de confianza? # Si calculo el error estándar de la p estimada como sigue, es correcto? p &lt;- mean(paciente$anemia) sqrt(p * (1 - p) / 5000) #&gt; [1] 0.007060751 Muestreo aleatorio En la situación ideal diseñaríamos una muestra aleatoria de menores de edad, por ejemplo, utilizando el registro en educación primaria de la SEP, y mediríamos la prevalencia de anemia en la muestra, usaríamos esta muestra para estimar la prevalencia en la población y tendríamos además las herramientas para medir la incertidumbre de nuestra estimación (reportar intervalos, o errores estándar). El elemento clave, es la aleatorización en la selección de la muestra, la idea es distribuir los efecros desconcidos o no controlables que pueden introducir sesgos o variabilidad no conocida en los resultados. Pero si no podemos hacer muestreo aleatorio? En el caso de prevalencia de anemia, discutiendo con médicos e investigadores nos informan que la anemia se presenta en tasas más altas en niños más chicos. paciente |&gt; count(edad) |&gt; mutate(prop = round(100 * n / sum(n))) #&gt; # A tibble: 10 × 3 #&gt; edad n prop #&gt; &lt;int&gt; &lt;int&gt; &lt;dbl&gt; #&gt; 1 6 1001 20 #&gt; 2 7 931 19 #&gt; 3 8 980 20 #&gt; 4 9 445 9 #&gt; 5 10 484 10 #&gt; 6 11 489 10 #&gt; 7 12 246 5 #&gt; 8 13 239 5 #&gt; 9 14 90 2 #&gt; 10 15 95 2 Y consultando con las proyecciones de población notamos que los niños chicos están sobrerepresentados en la muestra. Lo que nos hace considerar que debemos buscar una manera de ponderar nuestras observaciones para que reflejen a la población. Más aún, investigamos que algunas enfermedades están asociadas a mayor prevalencia de anemia: paciente |&gt; count(padecimiento) |&gt; arrange(-n) #&gt; # A tibble: 7 × 2 #&gt; padecimiento n #&gt; &lt;chr&gt; &lt;int&gt; #&gt; 1 infección respiratoria 745 #&gt; 2 mordedura de perro 723 #&gt; 3 úlcera 723 #&gt; 4 asma 713 #&gt; 5 apendcitis 704 #&gt; 6 picadura alacrán 701 #&gt; 7 infección intestinal 691 Utilizamos esta información para modelar y corregir nuestra estimación original. Por ejemplo con modelos de regresión. Sin embargo, debemos preguntarnos: ¿Hay más variables qué nos falta considerar? Ejemplo: Policías y tráfico Supongamos que nos preguntan en cuánto reduce un policía el tráfico en un crucero grande de la ciudad. La cultura popular ha establecido que los policías en cruceros hacen más tráfico porque no saben mover los semáforos. Nosotros decidimos buscar unos datos para entender esto. Escogemos entonces un grupo de cruceros problemáticos, registramos el tráfico cuando visitamos, y si había un policía o no. Después de este esfuerzo, obtenemos los siguientes datos: #&gt; # A tibble: 10 × 2 #&gt; # Groups: policia [2] #&gt; policia tiempo_espera_min #&gt; &lt;int&gt; &lt;dbl&gt; #&gt; 1 0 2.27 #&gt; 2 0 2.65 #&gt; 3 0 3.4 #&gt; 4 0 0.39 #&gt; 5 0 1.1 #&gt; 6 1 10.8 #&gt; 7 1 4.67 #&gt; 8 1 7.77 #&gt; 9 1 6.3 #&gt; 10 1 6.99 Lo que sabemos ahora es que la presencia de un policía es indicador de tráfico alto. El análisis prosiguiría calculando medias y medidas de error (escogimos una muestra aleatoria): Si somos ingenuos, entonces podríamos concluir que los policías efectivamente empeoran la situación cuando manipulan los semáforos, y confirmaríamos la sabiduría popular. Para juzgar este argumento desde el punto de vista causal, nos preguntamos primero: ¿Cuáles son los contrafactuales (los contrafactuales explican que pasaría si hubiéramos hecho otra cosa que la que efectivamente hicimos) de las observaciones? Efectos causales y el esquema de resultados potenciales Consideramos un tratamiento binario: Se manda policía o no se manda policía. Un resultado potencial es aquél que se observaría bajo un tratamiento particular. En cada semáforo, a una hora dada, hay dos resultados potenciales, uno por cada valor del tratamiento: \\(y_1:\\) tiempo de espera si se envía policía. \\(y_0:\\) tiempo de espera si no se envía policía. Para cada semáforo, en el momento de registro, uno observa únicamente uno de los dos resultados potenciales. El resultado no observado se conoce como resultado contrafactual. El estimador estándar A la comparación anterior - la diferencia de medias de tratados y no tratados - le llamamos usualmente el estimador estándar del efecto causal. Muchas veces este es un estimador malo del efecto causal. En nuestro ejemplo, para llegar a la conclusión errónea que confirma la sabiduría popular, hicimos un supuesto importante: En nuestra muestra, los casos con policía actúan como contrafactuales de los casos sin policía. Asi que asumimos que los casos con policía y sin policía son similares, excepto por la existencia o no de policía. En nuestro ejemplo, quizá un analista más astuto nota que tienen categorías históricas de qué tan complicado es cada crucero. Junta a sus datos, y obtiene: #&gt; # A tibble: 10 × 3 #&gt; # Groups: policia [2] #&gt; policia tiempo_espera_min categoria #&gt; &lt;int&gt; &lt;dbl&gt; &lt;fct&gt; #&gt; 1 0 2.27 Fluido #&gt; 2 0 2.65 Fluido #&gt; 3 0 3.4 Típico #&gt; 4 0 0.39 Fluido #&gt; 5 0 1.1 Fluido #&gt; 6 1 10.8 Complicado #&gt; 7 1 4.67 Típico #&gt; 8 1 7.77 Complicado #&gt; 9 1 6.3 Complicado #&gt; 10 1 6.99 Típico El analista argumenta entonces que los policías se enviaron principalmente a cruceros que se consideran Complicados según datos históricos. Esto resta credibilidad a la comparación que hicimos inicialmente: La comparación del estimador estándar no es de peras con peras: estamos comparando qué efecto tienen los policías en cruceros difíciles, con cruceros no difíciles donde no hay policía. La razón de esto es que el proceso generador de los datos incluye el hecho de que no se envían policías a lugares donde no hay tráfico. ¿Cómo producir contrafactuales para hacer la comparación correcta? Experimentos tradicionales Idealmente, quisiéramos observar un mismo crucero en las dos condiciones: con y sin policías. Esto no es posible. En un experimento “tradicional”, como nos lo explicaron en la escuela, nos aproximamos a esto preparando dos condiciones idénticas, y luego alteramos cada una de ellas con nuestra intervención. Si el experimento está bien hecho, esto nos da observaciones en pares, y cada quien tiene su contrafactual. La idea del experimiento tradicional es controlar todos los factores que intervienen en los resultados, y sólo mover el tratamiento para producir los contrafactuales. Más en general, esta estrategia consiste en hacer bloques de condiciones, donde las condiciones son prácticamente idénticas dentro de cada bloque. Comparamos entonces unidades tratadas y no tratadas dentro de cada bloque. Por ejemplo, si queremos saber si el tiempo de caída libre es diferente para un objeto más pesado que otro, prepararíamos dos pesos con el mismo tamaño pero de peso distinto. Soltaríamos los dos al mismo tiempo y compararíamos el tiempo de caída de cada uno. En nuestro caso, como es usual en problemas de negocio o sociales, hacer esto es considerablemente más difícil. No podemos “preparar” cruceros con condiciones idénticas. Sin embargo, podríamos intentar \\(bloquear\\) los cruceros según información que tenemos acerca de ellos, para hacer más comparaciones de peras con peras. Bloqueo Podemos acercanos en lo posible a este ideal de experimentación usando información existente. En lugar de hacer comparaciones directas entre unidades que recibieron el tratamiento y las que no (que pueden ser diferentes en otros aspectos, como vimos arriba), podemos refinar nuestras comparaciones bloquéandolas con variables conocidas. En el ejemplo de los policías, podemos hacer lo siguiente: dentro de cada categoría de cruceros (fluido, típico o complicado), tomaremos una muestra de cruceros, algunos con policía y otros sin. Haremos comparaciones dentro de cada categoría. Obtenemos una muestra con estas características (6 casos en cada categoría de crucero, 3 con policía y 3 sin policía): categoria policia n Fluido 0 3 Fluido 1 3 Típico 0 3 Típico 1 3 Complicado 0 3 Complicado 1 3 Y ahora hacemos comparaciones dentro de cada bloque creado por categoría: #&gt; # A tibble: 3 × 3 #&gt; # Groups: categoria [3] #&gt; categoria `policia =0` `policia =1` #&gt; &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt; #&gt; 1 Fluido 2.1 0.8 #&gt; 2 Típico 5.6 4.2 #&gt; 3 Complicado 10.4 8.6 Y empezamos a ver otra imagen en estos datos: comparando tipos e cruceros similares, los que tienen policía tienen tiempos de espera ligeramente más cortos. ¿Hemos termniado? ¿Podemos concluir que el efecto de un policía es beneficiosos pero considerablemente chico? ¿Qué problemas puede haber con este análisis? Variables desconocidas El problema con el análisis anterior es que controlamos por una variable que conocemos, pero muchas otras variables pueden estar ligadas con el proceso de selección de cruceros para enviar policías. Por ejemplo, envían o policías a cruceros Típicos solo cuando reportan mucho tráfico. No envían a un polícia a un crucero Complicado si no presenta demasiado tráfico. Existen otras variables desconocidas que los tomadores de decisiones usan para enviar a los policías. En este caso, por ejemplo, los expertos hipotéticos nos señalan que hay algunos cruceros que aunque problemáticos, a veces su tráfico se resuelve rápidamente, mientras que otros tienen tráfico más persistente, y prefieren enviar policías a los de tráfico persistente. La lista de cruceros persistentes están en una hoja de excel que se comparte de manera informal. En resumen, no tenemos conocimiento detallado del proceso generador de datos en cuanto a cómo se asignan los policías a los cruceros. Igual que en la sección anterior, podemos cortar esta complejidad usando aleatorización. Nótese que los expertos no están haciendo nada malo: en su trabajo están haciendo el mejor uso de los recursos que tienen. El problema es que por esa misma razón no podemos saber el resultado de sus esfuerzos, y si hay maneras de optimizar la asignación que hacen actualmente. Aleatorizando el tratamiento Tomamos la decisión entonces de hacer un experimento que incluya aletorización. En un dia particular, escogeremos algunos cruceros. Dicidimos usar solamente cruceros de la categoría Complicada y Típica, pues esos son los más interesantes para hacer intervenciones. Usaremos un poco de código para entener el detalle: en estos datos, tenemos para cada caso los dos posibles resultados hipotéticos \\(y_0\\) y \\(y_1\\) (con policia y sin policia). En el experimento asignamos el tratamiento al azar: muestra_exp &lt;- trafico_tbl |&gt; filter(categoria != &quot;Fluido&quot;) |&gt; sample_n(200) |&gt; # asignar tratamiento al azar, esta es nuestra intervención: mutate(tratamiento_policia = rbernoulli(length(y_0), 0.5)) |&gt; # observar resultado mutate(tiempo_espera_exp = ifelse(tratamiento_policia == 1, y_1, y_0)) Nótese la diferencia si tomamos la asignación natural del tratamiento (policía o no): set.seed(134) muestra_natural &lt;- trafico_tbl |&gt; filter(categoria != &quot;Fluido&quot;) |&gt; sample_n(200) |&gt; # usamos el tratamiento que se asignó # policia indica si hubo o no policía en ese crucero # observar resultado mutate(tiempo_espera_obs = ifelse(policia == 1, y_1, y_0)) Resumimos nuestros resultados del experimento son: #&gt; # A tibble: 2 × 3 #&gt; # Groups: categoria [2] #&gt; categoria `policia=0` `policia=1` #&gt; &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt; #&gt; 1 Típico 6.24 4.97 #&gt; 2 Complicado 15.8 8.47 Sin embargo, la muestra natural da: #&gt; # A tibble: 2 × 3 #&gt; # Groups: categoria [2] #&gt; categoria `policia=0` `policia=1` #&gt; &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt; #&gt; 1 Típico 5.49 4.35 #&gt; 2 Complicado 10.8 8.93 ¿Cuál de los dos análisis da la respuesta correcta a la pregunta: ayudan o no los policías a reducir el tráfico en los cruceros problemáticos? El experimento establece que un policía en promedio reduce a la mitad el tiempo de espera en un crucero complicado Selección de unidades y tratamiento Vimos dos tipos de inferencia que requieren distintos diseños de estudio: a poblaciones (ejemplo anemia) y causal (ejemplo policías). En el escenario ideal de cada uno de estos ejemplos requerimos un mecanismo de aleatorización, sin embargo, la aleatorización requerida en cada caso es distinta y distinguir esto es fundamental para entender las inferencias que podemos hacer en distintos escenarios. Inferencia estadística de acuerdo al tipo del diseño (Ramsey and Schafer (2012)). El cuadro arriba a la izquierda es donde el análisis es más simple y los resultados son más fáciles de interpretar. En este escenario don de la aleatorización es tanto en unidades como en grupos no hacen falta supuestos adicionales para tener las garantías de métodos de inferencia. Es posible hacer análisis fuera de este cuadro, pero el proceso es más complicado, requieren más supuestos, conocimiento del dominio y habilidades de análisis. En general resultan conclusiones menos sólidas. Muchas veces no nos queda otra más que trabajar fuera del cuadro ideal. El punto crucial para entender las medidas de incertidumbre estadística es visualizar de manera hipotética, replicaciones del estudio y las condiciones que llevaron a la selección de la muestra. Esto es, entender el proceso generador de datos e imaginar replicarlo. Ubica los siguientes tipos de análisis: Pruebas clínicas para medicinas Analizar cómo afecta tener seguro médico a los ingresos, usando datos del ENIGH. Estimación de retorno sobre inversión en modelos de marketing mix. Asignación natural del tratamiento Cuando consideramos un sistema donde se “asignan” tratamientos de manera natural, generalmente los tratamientos se asignan bajo un criterio de optimización o conveniencia (por ejemplo los policías a cruceros problemáticos). La cara buena de este hecho es que de alguna forma los resultados están intentando optimizarse, y la gente está haciendo su trabajo. La cara mala de este hecho es que no podemos evaluar de manera simple la efectividad de los tratamientos. Y esto hace difícil optimizar de forma cuantificable los procesos, o entender qué funciona y qué no. Referencias "],["pruebas-de-hipótesis.html", "Sección 3 Pruebas de hipótesis Comparación con poblaciones de referencia Comparando distribuciones Prueba de permutaciones y el lineup Comparaciones usando lineup (continuación) Prueba de permutaciones para proporciones Pruebas de hipótesis tradicionales Tomadores de té (continuación) Pruebas de permutación: implementación. Ejemplo: tiempos de fusión Ejemplo: tiempos de fusión (continuación) Separación de grupos La “crisis de replicabilidad” El jardín de los senderos que se bifurcan Ejemplo: decisiones de análisis y valores p Alternativas o soluciones", " Sección 3 Pruebas de hipótesis Las primeras técnicas inferenciales que veremos intentan contestar la siguiente pregunta: Si observamos cierto patrón en los datos, ¿cómo podemos cuantificar la evidencia de que es un patrón notable y no sólo debido a fluctuaciones en los datos particulares que tenemos? ¿Cómo sabemos que no estamos sobreinterpretando esas fluctuaciones? Por ejemplo: Un sistema tiene cierto comportamiento “usual” para el cual tenemos datos históricos. El sistema presenta fluctuaciones en el tiempo. Observamos la última salida de nuestro sistema. Naturalmente, tiene fluctuaciones. ¿Esas fluctuaciones son consistentes con la operación usual del sistema? ¿Existe evidencia para pensar que algo en el sistema cambió? Comparación con poblaciones de referencia En las prueba de hipótesis, tratamos de construir distribuciones de referencia para comparar resultados que obtengamos con un “estándar” de variación, y juzgar si nuestros resultados son consistentes con la referencia o no (Box et al. 1978). En algunos casos, ese estándar de variación puede construirse con datos históricos. Ejemplo Supongamos que estamos considerando cambios rápidos en una serie de tiempo de alta frecuencia. Hemos observado la serie en su estado “normal” durante un tiempo considerable, y cuando observamos nuevos datos quisiéramos juzgar si hay indicaciones o evidencia en contra de que el sistema sigue funcionando de manera similar. Digamos que monitoreamos ventanas de tiempo de tamaño 20 y necesitamos tomar una decisión. Abajo mostramos cinco ejemplos donde el sistema opera normalmente, que muestra la variabilidad en el tiempo en ventanas cortas del sistema. Ahora suponemos que obtenemos una nueva ventana de datos. ¿Hay evidencia en contra de que el sistema sigue funcionando de manera similar? Nuestra primera inclinación debe ser comparar: en este caso, compararamos ventanas históricas con nuestra nueva serie: # usamos datos simulados para este ejemplo set.seed(8812) historicos &lt;- simular_serie(2000) ¿Vemos algo diferente en los datos nuevos (el panel de color diferente)? Indpendientemente de la respuesta, vemos que hacer este análisis de manera tan simple no es siempre útil: seguramente podemos encontrar maneras en que la nueva muestra (4) es diferente a muestras históricas. Por ejemplo, ninguna de muestras tiene un “forma de montaña” tan clara. Nos preguntamos si no estamos sobreinterpretando variaciones que son parte normal del proceso. Podemos hacer un mejor análisis si extraemos varias muestras del comportamiento usual del sistema, graficamos junto a la nueva muestra, y revolvemos las gráficas para que no sepamos cuál es cuál. Entonces la pregunta es: ¿Podemos detectar donde están los datos nuevos? Esta se llama una prueba de lineup, o una prueba de ronda de sospechosos (Hadley Wickham et al. 2010). En la siguiente gráfica, en uno de los páneles están los datos recientemente observados. ¿Hay algo en los datos que distinga al patrón nuevo? # nuevos datos obs &lt;- simular_serie(500, x_inicial = last(obs$obs)) # muestrear datos históricos prueba_tbl &lt;- muestrear_ventanas(historicos, obs[1:20, ], n_ventana = 20) # gráfica de pequeños múltiplos ggplot(prueba_tbl$lineup, aes(x = t_0, y = obs)) + geom_line() + facet_wrap(~rep, nrow = 4) + scale_y_log10() ¿Cuáles son los datos nuevos (solo hay un panel con los nuevos datos)? ¿Qué implica que la gráfica que escojamos como “más diferente” no sean los datos nuevos? ¿Qué implica que le “atinemos” a la gráfica de los datos nuevos? Ahora observamos al sistema en otro momento y repetimos la comparación. En el siguiente caso obtenemos: Aunque es imposible estar seguros de que ha ocurrido un cambio, la diferencia de una de las series es muy considerable. Si identificamos los datos correctos, la probabilidad de que hayamos señalado la nueva serie “sobreinterpretando” fluctuaciones en un proceso que sigue comportándose normalente es 0.05 - relativamente baja. Detectar los datos diferentes es evidencia en contra de que el sistema sigue funcionando de la misma manera que antes. En el ejemplo anterior se encontraban en la posición: prueba_tbl$pos ## [1] 18 Observaciones y terminología: Llamamos hipótesis nula a la hipótesis de que los nuevos datos son producidos bajo las mismas condiciones que los datos de control o de referencia. Si no escogemos la gráfica de los nuevos datos, nuestra conclusión es que la prueba no aporta evidencia en contra de la hipótesis nula. Si escogemos la gráfica correcta, nuestra conclusión es que la prueba aporta evidencia en contra de la hipótesis nula. ¿Qué tan fuerte es la evidencia, en caso de que descubrimos los datos no nulos? Cuando el número de paneles es más grande y detectamos los datos, la evidencia es más alta en contra de la nula. Decimos que el nivel de significancia de la prueba es la probabilidad de seleccionar a los datos correctos cuando la hipótesis nula es cierta (el sistema no ha cambiado). En el caso de 20 paneles, la significancia es de 1/20 = 0.05. Cuando detectamos los datos nuevos, niveles de significancia más bajos implican más evidencia en contra de la nula. Si acertamos, y la diferencia es más notoria y fue muy fácil detectar la gráfica diferente (pues sus diferencias son más extremas), esto también sugiere más evidencia en contra de la hipótesis nula. Finalmente, esta prueba rara vez (o nunca) nos da seguridad completa acerca de ninguna conclusión, aún cuando hiciéramos muchos páneles. Comparando distribuciones Ahora intentamos un ejemplo más típico. Supongamos que tenemos muestras para tres grupos a, b y c, esto es que dentro de cada grupo, el proceso de selección de los elementos se hace al azar y de manera simétrica (por ejemplo cada elemento tiene a misma probabiidad de ser seleccionado, y las extracciones se hacen de manera independiente.) Queremos comparar las distribuciones de los datos obtenidos para cada grupo. Quizá la pregunta detrás de esta comparación es: el grupo de clientes b recibió una promoción especial. ¿Están gastando más? La medición que comparamos es el gasto de los clientes. En la muestra observamos diferencias entre los grupos. Pero notamos adicionalmente que hay mucha variación dentro de cada grupo. Nos podríamos preguntar entonces si las diferencias que observamos se deben variación muestral, por ejemplo. Podemos construir ahora una hipótesis nula, que establece que las observaciones provienen de una población similar: Las tres poblaciones (a, b, c) son prácticamente indistiguibles. En este caso, la variación que observamos se debería a que tenemos información incompleta. Como en el ejemplo anterior necesitamos construir u obtener una distribución de referencia para comparar qué tan extremos o diferentes son los datos que observamos. Esa distribución de referencia debería estar basada en el supuesto de que los grupos producen datos de distribuciones similares. Si tuvieramos mediciones similares históricas de estos tres grupos, quizá podríamos extraer datos de referencia y comparar, como hicimos en el ejempo anterior. Pero esto es menos común en este tipo de ejemplos. Prueba de permutaciones y el lineup Para abordar este problema podemos pensar en usar permutaciones de los grupos de la siguiente forma ((Box et al. 1978), (Tim C. Hesterberg 2015a)): Si los grupos producen datos bajo procesos idénticos, entonces los grupos a, b, c solo son etiquetas que no contienen información. Podríamos permutar al azar las etiquetas y observar nuevamente la gráfica de caja y brazos por grupos. Si la hipótesis nula es cierta (grupos idénticos), esta es una muestra tan verosímil como la que obtuvimos. Así que podemos construir datos de referencia permutando las etiquetas de los grupos al azar, y observando la variación que ocurre. Si la hipótesis nula es cercana a ser cierta, no deberíamos de poder distinguir fácilmente los datos observados de los producidos con las permutaciones al azar. Vamos a intentar esto, por ejemplo usando una gráfica de cuantiles simplificada. Hacemos un lineup, o una rueda de sospechosos (usamos el paquete (H. Wickham, Chowdhury, and Cook 2012), ver (Hadley Wickham et al. 2010)), donde 19 de los acusados son generados mediante permutaciones al azar de la variable del grupo, y el culpable (los verdaderos datos) están en una posición escogida al azar. ¿Podemos identificar los datos verdaderos? Para evitar sesgarnos, también ocultamos la etiqueta verdadera. Usamos una gráfica que muestra los cuantiles 0.10, 0.50, 0.90: set.seed(88) reps &lt;- lineup(null_permute(&quot;grupo&quot;), muestra_tab, n = 20) ## decrypt(&quot;M7xA 2S8S Jj dUyJ8JUj ZW&quot;) reps_mezcla &lt;- reps |&gt; mutate(grupo_1 = factor(digest::digest2int(grupo) %% 177)) grafica_cuantiles(reps_mezcla, grupo_1, x) + coord_flip() + facet_wrap(~.sample, ncol = 5) + ylab(&quot;x&quot;) + labs(caption = &quot;Mediana y percentiles 10% y 90%&quot;) + geom_point(aes(colour = grupo_1)) Y la pregunta que hacemos es ¿podemos distinguir nuestra muestra entre todas las replicaciones producidas con permutaciones? ¿Dónde están los datos observados? Según tu elección, ¿qué tan diferentes son los datos observados de los datos nulos? En este ejemplo, es difícil indicar cuáles son los datos. Los grupos tienen distribuciones similares y es factible que las diferencias que observamos se deban a variación muestral. Si la persona escoge los verdaderos datos, encontramos evidencia en contra de la hipótesis nula (los tres grupos son equivalentes). En algunos contextos, se dice que los datos son significativamente diferentes al nivel 0.05. Esto es evidencia en contra de que los datos se producen de manera homogénea, independientemente del grupo. Si la persona escoge uno de los datos permutados, no encontramos evidencia en contra de que los tres grupos producen datos con distribuciones similares. Comparaciones usando lineup (continuación) Repetimos el ejemplo para otra muestra (en este ejemplo el proceso generador de datos es diferente para el grupo b): Hacemos primero la prueba del lineup: set.seed(121) reps &lt;- lineup(null_permute(&quot;grupo&quot;), muestra_tab, n = 20) grafica_cuantiles(reps |&gt; mutate(grupo_escondido = factor(digest::digest2int(grupo) %% 177)), grupo_escondido, x) + facet_wrap(~.sample) + ylab(&quot;x&quot;) + coord_flip() + geom_point(aes(colour = grupo_escondido)) Podemos distinguir más o menos claramente que está localizada en valores más altos y tiene mayor dispersión. En este caso, como en general podemos identificar los datos, obtenemos evidencia en contra de que los tres grupos tienen distribuciones iguales. Estos ejemplos siguen la idea de inferencia visual propuestas en (Hadley Wickham et al. 2010), (Hofmann et al. 2012) son pruebas muy flexibles y estadísticamente rigurosas. Prueba de permutaciones para proporciones Veremos otro ejemplo donde podemos hacer más concreta la idea de distribución nula o de referencia usando pruebas de permutaciones. Supongamos que con nuestra muestra de tomadores de té, queremos probar la siguiente hipótesis nula: Los tomadores de té en bolsas exclusivamente, usan azúcar a tasas simillares que los tomadores de té suelto (que pueden o no también tomar té en bolsita). Los datos que obtuvimos en nuestra encuesta, en conteos, son: sugar bolsa_exclusivo suelto o bolsa No.sugar 81 74 sugar 89 56 Y en proporciones tenemos que: how prop_azucar n bolsa_exclusivo 0.52 170 suelto o bolsa 0.43 130 Pero distintas muestras podrían haber dado distintos resultados. Nos preguntamos qué tan fuerte es la evidencia en contra de que en realidad los dos grupos de personas usan azúcar en proporciones similares, y la diferencia que vemos se puede atribuir a variación muestral. En este ejemplo, podemos usar una estadística de prueba numérica, por ejemplo, la diferencia entre las dos proporciones: \\[\\hat p_1 - \\hat p_2,\\] (tomadores de té en bolsa solamente vs. suelto y bolsa). El proceso sería entonces: La hipótesis nula es que los dos grupos tienen distribuciones iguales. Este caso quiere decir que en la población, tomadores de té solo en bolsa usan azúcar a las mismas tasas que tomadores de suelto o bolsas. Bajo nuestra hipótesis nula (proporciones iguales), producimos una cantidad grande (por ejemplo 10 mil o más) de muestras permutando las etiquetas de los grupos. Evaluamos nuestra estadística de prueba en cada una de las muestras permutadas. El conjunto de valores obtenidos nos da nuestra distribución de referencia (ya no estamos limitados a 20 replicaciones como en las pruebas gráficas). Y la pregunta clave es: ¿el valor de la estadística en nuestra muestra es extrema en comparación a la distribución de referencia? dif_obs &lt;- te_azucar |&gt; mutate(usa_azucar = as.numeric(sugar == &quot;sugar&quot;)) |&gt; group_by(how) |&gt; summarise(prop_azucar = mean(usa_azucar), .groups = &#39;drop&#39;) |&gt; pivot_wider(names_from = how, values_from = prop_azucar) |&gt; mutate(diferencia_prop = bolsa_exclusivo - `suelto o bolsa`) |&gt; pull(diferencia_prop) La diferencia observada es: dif_obs |&gt; round(3) ## [1] 0.093 Ahora construimos nuestra distribución nula o de referencia: reps &lt;- lineup(null_permute(&quot;how&quot;), te_azucar, n = 50000) glimpse(reps) ## Rows: 15,000,000 ## Columns: 3 ## $ how &lt;chr&gt; &quot;bolsa_exclusivo&quot;, &quot;bolsa_exclusivo&quot;, &quot;suelto o bolsa&quot;, &quot;suelt… ## $ sugar &lt;chr&gt; &quot;sugar&quot;, &quot;No.sugar&quot;, &quot;No.sugar&quot;, &quot;sugar&quot;, &quot;No.sugar&quot;, &quot;No.suga… ## $ .sample &lt;dbl&gt; 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,… valores_ref &lt;- reps |&gt; mutate(usa_azucar = as.numeric(sugar == &quot;sugar&quot;)) |&gt; group_by(.sample, how) |&gt; summarise(prop_azucar = mean(usa_azucar), .groups = &#39;drop&#39;) |&gt; pivot_wider(names_from = how, values_from = prop_azucar) |&gt; mutate(diferencia = bolsa_exclusivo - `suelto o bolsa`) Y graficamos nuestros resultados (con un histograma y una gráfica de cuantiles, por ejemplo). la estadística evaluada un cada una de nuestras muestras permutadas: g_1 &lt;- ggplot(valores_ref, aes(sample = diferencia)) + geom_qq(distribution = stats::qunif) + xlab(&quot;f&quot;) + ylab(&quot;diferencia&quot;) + labs(subtitle = &quot;Distribución nula o de referencia&quot;) g_2 &lt;- ggplot(valores_ref, aes(x = diferencia)) + geom_histogram(binwidth = 0.04) + coord_flip() + xlab(&quot;&quot;) + labs(subtitle = &quot; &quot;) g_1 + g_2 Este es el rango de fluctuación usual para nuestra estadística bajo la hipótesis de que los dos grupos de tomadores de té consumen té a la misma tasa. El valor que obtuvimos en nuestros datos es: 0.09. Mismo que no es un valor extremo en la distribución de referencia que vimos arriba. Ésta muestra no aporta mucha evidencia en contra de que los grupos tienen distribuciones similares. Podemos graficar otra vez marcando el valor observado: # Función de distribución acumulada (inverso de función de cuantiles) dist_perm &lt;- ecdf(valores_ref$diferencia) # Calculamos el percentil del valor observado percentil_obs &lt;- dist_perm(dif_obs) g_1 &lt;- ggplot(valores_ref, aes(sample = diferencia)) + geom_qq(distribution = stats::qunif) + xlab(&quot;f&quot;) + ylab(&quot;diferencia&quot;) + labs(subtitle = &quot;Distribución nula o de referencia&quot;) + geom_hline(yintercept = dif_obs, colour = &quot;red&quot;) + annotate(&quot;text&quot;, x = 0.35, y = dif_obs - 0.03, label = &quot;diferencia observada&quot;, colour = &quot;red&quot;) g_2 &lt;- ggplot(valores_ref, aes(x = diferencia)) + geom_histogram(binwidth = 0.04) + coord_flip() + xlab(&quot;&quot;) + labs(subtitle = &quot; &quot;) + geom_vline(xintercept = dif_obs, colour = &quot;red&quot;) + annotate(&quot;text&quot;, x = dif_obs, y = N_rep * .3, label = percentil_obs,vjust = -0.2, colour = &quot;red&quot;) g_1 + g_2 Y vemos que es un valor algo (pero no muy) extremo en la distribución de referencia que vimos arriba: esta muestra no aporta una gran cantidad de evidencia en contra de que los grupos tienen distribuciones similares, que en este caso significa que los dos grupos usan azúcar a tasas similares. Es decir, sobre la hipótesis \\[H_0: p_1 = p_2,\\] o bien, \\[H_0: p_1 - p_2 = 0.\\] Pruebas de hipótesis tradicionales Comencemos recordando la definición de parámetro y estadística. Definición. Un parámetro es una característica (numérica) de una población o de una distribución de probabilidad. Usualmente denotado por \\(\\theta \\in \\mathbb{R},\\) o por \\(\\theta \\in \\mathbb{R}^p.\\) Una estadística es una característica (numérica) de los datos. Usualmente denotado por \\(\\hat \\theta.\\) Cualquier función de un parámetro es también un parámetro \\(\\varphi = h(\\theta),\\) y cualquier función de una estadística es también una estadística \\(\\hat \\varphi = h(\\hat \\theta).\\) Cuando la estadística se calcula de una muestra aleatoria (\\(T(X),\\) para \\(X_i \\sim \\pi_x\\) para \\(i = 1, \\ldots, n\\)), es por consiguiente aleatoria y es por tanto una variable aleatoria (\\(T \\sim \\pi_T\\)). Por ejemplo \\(\\mu\\) y \\(\\sigma\\) son parámetros de la distribución normal con función de densidad \\(\\pi(x) = (1/\\sqrt{2\\pi}\\sigma)e^{(x-\\mu)^2/(2\\sigma^2)}\\). La varianza \\(\\sigma^2\\), y el coeficiente de variación (cociente de señal a ruido) \\(\\mu/\\sigma\\) también son parámetros. Si \\(X_1, \\ldots ,X_n\\) son una muestra aleatoria, entonces la media \\(\\bar{X}=\\frac1n\\sum_i X_i\\) es una estadística. Ahora podemos pasar a las definiciones correspondientes a pruebas de hipótesis (o pruebas de significancia). Definición. Denotamos por \\(H_0\\) a la hipótesis nula la cual usualmente tratamos como la afirmación del status quo. La hipótesis alternativa la denotamos por \\(H_1\\) y representa el supuesto que está a prueba y para el cual buscamos evidencia en los datos. Definición. La hipótesis normalmente se plantea en términos de un parámetro (\\(\\theta\\in\\mathbb{R}\\)) o conjunto de parámetros (\\(\\theta\\in\\mathbb{R}^p\\)) de la distribución de interés (por ejemplo media, moda, varianza). Para una hipótesis nula del estilo \\(H_0: \\theta = \\theta_0,\\) la hipótesis a contrastar se puede denominar como: Hipótesis alternativa de una cola \\(H_1: \\theta \\gt \\theta_0\\) Hipótesis alternativa de dos colas \\(H_1: \\theta \\neq \\theta_0\\) En el ejemplo anterior planteamos hipótesis nula (proporciones iguales) e hipótesis alternativa que la proporción de tomadores de te suelto que usan azúcar en menor proporción, esto corresponde a una hipótesis alternativa a dos colas: \\(H_0: p_1 = p_2\\), y \\(H_1:p_1 &gt; p_2\\). Definición. Una estadística de prueba es una función numérica de los datos cuyo valor determina el resultado de la prueba. La función usualmente es denotada por \\(T(\\bf X)\\) donde \\(\\bf X\\) representa los datos como variable aleatoria. Por ejemplo, \\(T = T(X_1, \\ldots, X_n)\\) si sólo tenemos una muestra, o por \\(T = T(X_1, \\ldots, X_n, Y_1, \\ldots, Y_m)\\) en el caso de tener dos muestras. Al evaluar la prueba para un conjunto de datos dado, \\(x\\), ésta se denomina estadística de prueba observada, \\(t = T(x).\\) La estadística de prueba correspondiente al ejemplo es \\(T = \\hat p_1 - \\hat p_2.\\) Definición. El valor p es la probabilidad de que bajo la hipótesis nula los datos generen un valor tan extremo como la estadística de prueba observada. Por ejemplo, si consideramos la hipótesis nula admite valores grandes, el valor p se calcula como \\(P(T \\geq t).\\) En el ejemplo de tomadores de té: el valor p lo calculamos usando el percentil donde nuestra observación cae en la distribución generada por las permutación (valor p de una cola). 1 - dist_perm(dif_obs) ## [1] 0.04344 Por otro lado, podemos calcular: Valor p de dos colas: Si la hipótesis nula es cierta, ¿cuál es la probabilidad de observar una diferencia tan extrema o más extrema de lo que observamos? Considerando este caso interpretamos extrema como qué tan lejos cae del centro de masa de la distribución. De tal forma que podemos calcular el valor p como sigue. A partir del valor observado, consideramos cuál dato es menor: la probabilidad bajo lo hipótesis nula de observar una diferencia mayor de la que observamos, o la probabilidad de observar una diferencia menor a la que observamos. Tomamos el mínimo y multiplicamos por dos (Tim C. Hesterberg 2015a): 2 * min(dist_perm(dif_obs), (1 - dist_perm(dif_obs))) ## [1] 0.08688 Este valor p se considera como evidencia moderada en contra de la hipótesis nula. Valores p chicos (observaciones más extremas en comparación con la referencia) aportan más evidencia en contra de la hipótesis nula, y valores más grandes aportan menos evidencia en contra. Definición. Un resultado es estadisticamente significativo si tiene muy baja probabilidad de suceder al azar. Entre más pequeño requiramos un valor p oara declarar un resultado estadísticamente significativo, somos más conservadores. Las pruebas de hipótesis con frecuencia inician contestando una pregunta más general que los valores p: ¿Cuál es la distribución de la estadística de prueba cuando no hay un efecto real? Definición. La distribución nula es la distribución de la estadística de prueba si la hipótesis nula es cierta. En ocasiones también nos referimos a ella como la distribución de referencia pues estamos comparando la estadística de prueba observada a su referencia para determinar que tan inusual es. En el ejemplo de tomadores de té aproximamos la distribución nula (y los valores p) con simulación; sin embargo, para algunas estadísticas hay métodos exactos. En particular, usamos el método de pruebas de permutación. Para dicha prueba el algoritmo para en el caso de dos grupos sería como sigue: Prueba de permutación para dos muestras Supongamos que tenemos m observaciones de una población y n de otra. Combina los m+n valores. Repite: Obtén un remuestra de tamaño m sin reemplazo del total. Usa las n observaciones restantes para obtener la otra muestra. Calcula la estadística de prueba (que compara las muestras). Calcula el valor p como la fracción de las veces que la estadística sobrepasó la estadística observada, multiplica por 2 para una prueba de dos lados. La distribución de la estadística a lo largo de las remuestras de permutación es la distribución de permutación. Ésta puede ser exacta, si se calcula exhaustivamente (como cuando tenemos pocas observaciones) o aproximada (cuando enlistar todas las posible combinaciones es prohibitivo). Tomadores de té (continuación) Ahora hacemos una prueba de permutaciones para otro par de proporciones utilizando el mismo método. La hipótesis nula ahora es: Los tomadores de té Earl Gray usan azúcar a una tasa similar a los tomadores de té negro. Los datos que obtuvimos en nuestra encuesta se muestran en la siguiente tabla: sugar Earl Grey black No.sugar 84 51 sugar 109 23 Y en porcentajes tenemos que: prop_azucar &lt;- te_azucar |&gt; count(Tea, sugar) |&gt; group_by(Tea) |&gt; mutate(prop = 100 * n / sum(n), n = sum(n)) |&gt; filter(sugar == &quot;sugar&quot;) |&gt; select(Tea, prop_azucar = prop, n) |&gt; mutate(&#39;% usa azúcar&#39; = round(prop_azucar)) |&gt; select(-prop_azucar) prop_azucar |&gt; formatear_tabla() Tea n % usa azúcar Earl Grey 193 56 black 74 31 Pero distintas muestras podrían haber dado distintos resultados. Nos preguntamos qué tan fuerte es la evidencia en contra de que en realidad los dos grupos de personas usan azúcar en proporciones similares considerando que la diferencia que vemos se puede atribuir a variación muestral. Escribimos la función que calcula diferencias para cada muestra: calc_diferencia_2 &lt;- function(datos){ datos |&gt; mutate(usa_azucar = as.numeric(sugar == &quot;sugar&quot;)) |&gt; group_by(Tea) |&gt; summarise(prop_azucar = mean(usa_azucar), .groups = &#39;drop&#39;) |&gt; pivot_wider(names_from = Tea, values_from = prop_azucar) |&gt; mutate(diferencia_prop = `Earl Grey` - black) |&gt; pull(diferencia_prop) } La diferencia observada es: ## [1] 0.254 Ahora construimos nuestra distribución nula o de referencia: set.seed(2) reps &lt;- lineup(null_permute(&quot;Tea&quot;), te_azucar, n = N_rep) valores_ref &lt;- reps |&gt; group_by(.sample) |&gt; nest() |&gt; mutate(diferencia = lapply(data, calc_diferencia_2)) |&gt; unnest(diferencia) Y podemos graficar la distribución de referencia otra vez marcando el valor observado En este caso, la evidencia es muy fuerte en contra de la hipótesis nula, pues el resultado que obtuvimos es muy extremo en relación a la distribución de referencia. El valor p es cercano a 0. Haz una prueba de permutaciones para diferencia de medias para comparar la propina en cena vs en comidas. Grafica la distribución de referencia. Calcula el valor p (dos colas). Pruebas de permutación: implementación. Hasta ahora nos hemos centrado en ejemplos de diferencias en medias. Podemos extender las pruebas de permutación a \\(\\bar{X}\\) (la media de la primera muestra), \\(n\\bar{X}\\) (la suma de las observaciones en la primera muestra), y más. Teorema. En pruebas de permutación, si dos estadísticas de prueba \\(T_1\\) y \\(T_2\\) están relacionadas por una función estríctamente monótona, \\(T_1(X^*)=f(T_2(X^*))\\) donde \\(X^*\\) es una remuestra de permutación de los datos originales, entonces los valores p serán los mismos en las pruebas de permutación. Muestras con reemplazo de la Distribución Nula. En la implementación de muestreo, no nos aseguramos que las remuestras sean únicas. Sería más acertado tomar muestras sin reemplazo, sin embargo, el costo computacional es demasiado alto. Por simplicidad consideramos muestras con reemplazo del total de \\[m+n\\choose n\\] posibles remuestras. Por lo tanto, al remuestrar obtenemos una muestra de la distribución nula. Entre más muestras, más exactitud. Hemos usado \\(B = 10^3\\) remuestras (N_rep en el código), en general entre más remuestras tendremos una mejor estimación del valor p. Si el verdadero valor es \\(p\\) el estimado tendrá una varianza aproximadamente de \\(p(1- p)/B\\) donde \\(B\\) es el número de remuestras generadas. Observación. Así como los \\(n\\) datos originales son una muestra de la población, también las \\(B\\) remuestras de la estadística son una muestra de una población, en este caso de la distribución nula. La pruebas de permutaciones son más útiles cuando nuestra hipótesis nula se refiere que la distribución de los grupos son muy similares, o la independencia entre observaciones y grupo. Esto también aplica cuando queremos probar por ejemplo, que una variable numérica \\(Y\\) es independiente de \\(X.\\) Hay algunas hipótesis que no se pueden probar con este método, como por ejemplo, las que se refieren a una sola muestra: ¿los datos son consistentes con que su media es igual a 5? Adicionalmente, en algunas ocasiones queremos probar aspectos más específicos de las diferencias: como ¿son iguales las medias o medianas de dos grupos de datos? ¿Tienen dispersión similar? Es común aplicar pruebas de permutaciones a este segundo problema, sin embargo, no están tan perfectamente adaptadas a el, pues prueban todos los aspectos de las distribuciones que se comparan, aún cuando escojamos una estadística particular que pretende medir. Por ejemplo, cuando trabajamos con la diferencia de medias. Eso quiere decir que podemos rechazar igualdad de medias, por ejemplo, cuando en realidad otra característica de las distribuciones es la que difiere mucho en las poblaciones. En algunas referencias (ver (Chihara and Hesterberg 2018), (Efron and Tibshirani 1993)) se argumenta que de todas formas las pruebas de permutaciones son relativamente robustas a esta desadaptación. Un caso excepcional, por ejemplo, es cuando las poblaciones que comparamos resultan tener dispersión extremadamente distinta, y adicionalmente los tamaños de muestra de los grupos son muy desiguales (otra vez, ver ejemplos en (Chihara and Hesterberg 2018)). Ejemplo: tiempos de fusión Veamos el siguiente ejemplo, que es un experimento donde se midió el tiempo que tardan distintas personas en fusionar un estereograma para ver una imagen 3D. (William S. Cleveland (1993)). Existen dos condiciones: en una se dio indicaciones de qué figura tenían que buscar (VV) y en otra no se dio esa indicación. ¿Las instrucciones verbales ayudan a fusionar más rápido el estereograma? ## ## ── Column specification ──────────────────────────────────────────────────────── ## cols( ## n = col_double(), ## time = col_double(), ## nv.vv = col_character() ## ) La situación es la siguiente: considerando que hay mucha variación en el tiempo de fusión dentro de cada tratamiento, necesitamos calificar la evidencia de nuestra conclusión (el tiempo de fusión se reduce con información verbal). Podemos usar una prueba de permutaciones, esta vez justificándola por el hecho de que los tratamientos se asignan al azar: si los tratamientos son indistinguibles, entonces las etiquetas de los grupos son sólo etiquetas, y permutarlas daría muestras igualmente verosímiles. En este caso, compararemos gráficas de cuantiles de los datos con los producidos por permutaciones (transformamos los datos pues en este caso es más apropiado una comparación multiplicativa): ¿Podemos identificar los datos? En general, muy frecuentemente las personas identifican los datos correctamente, lo que muestra evidencia considerable de que la instrucción verbal altera los tiempos de respuesta de los partipantes, y en este caso ayuda a reducir el tiempo de fusión de los estereogramas. Ejemplo: tiempos de fusión (continuación) Podemos usar las pruebas de permutaciones para distintos tipos de estadísticas: medianas, medias, comparar dispersión usando rangos intercuartiles o varianzas, etc. Regresamos a los tiempos de fusión. Podemos hacer una prueba de permutaciones para la diferencia de las medias o medianas, por ejemplo. En este ejemplo usaremos una medida de centralidad un poco diferente, como ilustración: el promedio de los cuartiles superior e inferior de las dos distribuciones. Usaremos el cociente de estas dos cantidades para medir su diferencia # esta función hace permutaciones y calcula la diferencia para cada una permutaciones_est &lt;- function(datos, variable, calc_diferencia, n = 1000){ # calcular estadística para cada grupo permutar &lt;- function(variable){ sample(variable, length(variable)) } tbl_perms &lt;- tibble(.sample = seq(1, n-1, 1)) |&gt; mutate(diferencia = map_dbl(.sample, ~ datos |&gt; mutate({{variable}}:= permutar({{variable}})) |&gt; calc_diferencia())) bind_rows(tbl_perms, tibble(.sample = n, diferencia = calc_diferencia(datos))) } stat_fusion &lt;- function(x){ (quantile(x, 0.75) + quantile(x, 0.25))/2 } calc_fusion &lt;- function(stat_fusion){ fun &lt;- function(datos){ datos |&gt; group_by(nv.vv) |&gt; summarise(est = stat_fusion(time), .groups = &#39;drop&#39;) |&gt; pivot_wider(names_from = nv.vv, values_from = est) |&gt; mutate(dif = VV / NV ) |&gt; pull(dif) } fun } calc_cociente &lt;- calc_fusion(stat_fusion) dif_obs &lt;- calc_cociente(fusion) # permutar valores_ref &lt;- permutaciones_est(fusion, nv.vv, calc_cociente, n = N_rep) dist_perm_nv &lt;- ecdf(valores_ref$diferencia) cuantil_obs &lt;- dist_perm_nv(dif_obs) Y el valor p de dos colas es dist_perm_nv &lt;- ecdf(valores_ref$diferencia) 2 * min(dist_perm_nv(dif_obs), 1 - dist_perm_nv(dif_obs)) ## [1] 0.028 Lo que muestra evidencia considerable, aunque no muy fuerte, de que la instrucción verbal ayuda a reducir el tiempo de fusión de los estereogramas: la caja del diagrama de caja y brazos para el grupo VV está encogida por un factor menor a 1. Separación de grupos Este ejemplo tomado de (Chowdhury et al. 2015) (tanto la idea como el código). La pregunta que se aborda en ese estudio es: Existen métodos de clasificación (supervisados o no supervisados) para formar grupos en términos de variables que describen a los individuos Estos métodos (análisis discriminante, o k-means, por ejemplo), pretenden formar grupos compactos, bien separados entre ellos. Cuando aplicamos el método, obtenemos clasificadores basados en las variables de entrada. La pregunta es: ¿los grupos resultantes son producto de patrones que se generalizan a la población, o capitalizaron en variación aleatoria para formarse? Especialmente cuando tenemos muchas mediciones de los individuos, y una muestra relativamente chica, Es relativamente fácil encontrar combinaciones de variables que separan los grupos, aunque estas combinaciones y diferencias están basadas en ruido y no generalizan a la población. Como muestran en (Chowdhury et al. 2015), el lineup es útil para juzgar si tenemos evidencia en contra de que los grupos en realidad son iguales, y usamos variación muestral para separarlos. Avispas (opcional) En el siguiente ejemplo, tenemos 4 grupos de avispas (50 individuos en total), y para cada individuo se miden expresiones de 42 genes distintos. La pregunta es: ¿Podemos separar a los grupos de avispas dependiendo de sus mediciones? En este se usó análisis discriminante (LDA) para buscar proyecciones de los datos en dimensión baja de forma que los grupos sean lo más compactos y separados posibles. Para probar qué tan bien funciona este método, podemos hacer una prueba de permutación, aplicamos LDA y observamos los resultados. Y vemos que incluso permutando los grupos, es generalmente posible separarlos en grupos bien definidos: la búsqueda es suficientemente agresiva para encontrar combinaciones lineales que los separan. Que no podamos distinguir los datos verdaderos de las replicaciones nulas indica que este método difícilmente puede servir para separar los grupos claramente. Otro enfoque sería separar los datos en una muestra de entrenamiento y una de prueba (que discutiremos en la última sesión). Aplicamos el procedimiento a la muestra de entrenamiento y luego vemos qué pasa con los datos de prueba: set.seed(8) wasps_1 &lt;- wasps |&gt; mutate(u = runif(nrow(wasps), 0, 1)) wasps_entrena &lt;- wasps_1 |&gt; filter(u &lt;= 0.8) wasps_prueba &lt;- wasps_1 |&gt; filter(u &gt; 0.8) wasp.lda &lt;- MASS::lda(Group ~ ., data=wasps_entrena[,-1]) wasp_ld_entrena &lt;- predict(wasp.lda, dimen=2)$x |&gt; as_tibble(.name_repair = &quot;universal&quot;) |&gt; mutate(tipo = &quot;entrenamiento&quot;) |&gt; mutate(grupo = wasps_entrena$Group) wasp_ld_prueba &lt;- predict(wasp.lda, newdata = wasps_prueba, dimen=2)$x |&gt; as_tibble(.name_repair = &quot;universal&quot;) |&gt; mutate(tipo = &quot;prueba&quot;)|&gt; mutate(grupo = wasps_prueba$Group) wasp_lda &lt;- bind_rows(wasp_ld_entrena, wasp_ld_prueba) ggplot(wasp_lda, aes(x = LD1, y = LD2, colour = grupo)) + geom_point(size = 3) + facet_wrap(~tipo) Aunque esta separación de datos es menos efectiva en este ejemplo por la muestra chica, podemos ver que la separación lograda en los datos de entrenamiento probablemente se debe a variación muestral. La “crisis de replicabilidad” Recientemente (Ioannidis 2005) se ha reconocido en campos como la psicología la crisis de replicabilidad. Varios estudios que recibieron mucha publicidad inicialmente no han podido ser replicados posteriormente por otros investigadores. Por ejemplo: Hacer poses poderosas produce cambios fisiológicos que mejoran nuestro desempeño en ciertas tareas. Mostrar palabras relacionadas con “viejo” hacen que las personas caminen más lento (efectos de priming). En todos estos casos, el argumento de la evidencia de estos efectos fue respaldada por una prueba de hipótesis nula con un valor p menor a 0.05. La razón es que ese es el estándar de publicación seguido por varias áreas y revistas arbitradas. La tasa de no replicabilidad parece ser mucho más alta (al menos la mitad o más, según algunas fuentes como la señalada arriba) que la sugerida por la tasa de falsos positivos (menos de 5%). Este problema de replicabilidad parece ser más frecuente cuando: Se trata de estudios de potencia baja: mediciones ruidosas y tamaños de muestra chicos. El plan de análisis no está claramente definido desde un principio (lo cual es difícil cuando se están investigando “fenómenos no estudiados antes”). ¿A qué se atribuye esta crisis de replicabilidad? El jardín de los senderos que se bifurcan Aunque haya algunos ejemplos de manipulaciones conscientes —e incluso, en menos casos, malintencionadas— para obtener resultados publicables o significativos (p-hacking), como vimos en ejemplos anteriores, hay varias decisiones, todas razonables, que podemos tomar cuando estamos buscando las comparaciones correctas. Algunas pueden ser: Transformar los datos (tomar o no logaritmos, u otra transformación). Editar datos atípicos (razonable si los equipos pueden fallar, o hay errores de captura, por ejemplo). Distintas maneras de interpretar los criterios de inclusión de un estudio (por ejemplo, algunos participantes mostraron tener gripa, o revelaron que durmieron muy poco la noche anterior, etc. ¿los dejamos o los quitamos?). Dado un conjunto de datos, las justificaciones de las decisiones que se toman en cada paso son razonables, pero con datos distintos las decisiones podrían ser diferentes. Este es el jardín de los senderos que se bifurcan (ver referencia en Gelman), que invalida en parte el uso valores p como criterio de evidencia contra la hipótesis nula. Esto es exacerbado por: Tamaños de muestra chicos y efectos “inestables” que se quieren medir (por ejemplo en psicología). El hecho de que el criterio de publicación es obtener un valor \\(p &lt; 0.05\\), y la presión fuerte sobre los investigadores para producir resultados publicables (\\(p &lt; 0.05\\)). El que estudios o resultados similares que no obtuvieron valores \\(p\\) por debajo del umbral no son publicados o reportados. Ver por ejemplo el comunicado de la ASA. Ojo: esas presiones de publicación no sólo ocurre para investigadores en psicología. Cuando trabajamos en problemas de análisis de datos que son de importancia, es común que existan intereses de algunas partes o personas involucradas por algunos resultados u otros (por ejemplo, nuestros clientes de consultoría o clientes internos). Eso puede dañar nuestro trabajo como analistas, y el avance de nuestro equipo. Aunque esas presiones son inevitables, se vuelven manejables cuando hay una relación de confianza entre las partes involucradas. Ejemplo: decisiones de análisis y valores p En el ejemplo de datos de fusión, decidimos probar, por ejemplo, el promedio de los cuartiles inferior y superior, lo cual no es una decisión típica pero usamos como ilustración. Ahora intentamos usar distintas mediciones de la diferencia entre los grupos, usando distintas medidas resumen y transformaciones (por ejemplo, con o sin logaritmo). Aquí hay unas 12 combinaciones distintas para hacer el análisis (multiplicadas por criterios de “aceptación de datos en la muestra”, que simulamos tomando una submuestra al azar): calc_fusion &lt;- function(stat_fusion, trans, comparacion){ fun &lt;- function(datos){ datos |&gt; group_by(nv.vv) |&gt; summarise(est = stat_fusion({{ trans }}(time)), .groups = &#39;drop&#39;) |&gt; pivot_wider(names_from = nv.vv, values_from = est) |&gt; mutate(dif = {{ comparacion }}) |&gt; pull(dif) } fun } valor_p &lt;- function(datos, variable, calc_diferencia, n = 1000){ # calcular estadística para cada grupo permutar &lt;- function(variable){ sample(variable, length(variable)) } tbl_perms &lt;- tibble(.sample = seq(1, n-1, 1)) |&gt; mutate(diferencia = map_dbl(.sample, ~ datos |&gt; mutate({{variable}} := permutar({{variable}})) |&gt; calc_diferencia())) perms &lt;- bind_rows(tbl_perms, tibble(.sample = n, diferencia = calc_diferencia(datos))) perms_ecdf &lt;- ecdf(perms$diferencia) dif &lt;- calc_diferencia(datos) 2 * min(perms_ecdf(dif), 1- perms_ecdf(dif)) } set.seed(7272) media_cuartiles &lt;- function(x){ (quantile(x, 0.75) + quantile(x, 0.25))/2 } # nota: usar n=10000 o más, esto solo es para demostración: ejemplo &lt;- list() calc_dif &lt;- calc_fusion(mean, identity, VV - NV) ejemplo$media_dif &lt;- valor_p(fusion |&gt; sample_frac(0.95), nv.vv, calc_dif, n = N_rep) calc_dif &lt;- calc_fusion(mean, log, VV - NV) ejemplo$media_dif_log &lt;- valor_p(fusion |&gt; sample_frac(0.95), nv.vv, calc_dif, n = N_rep) calc_dif &lt;- calc_fusion(median, identity, VV / NV) ejemplo$mediana_razon &lt;- valor_p(fusion |&gt; sample_frac(0.95), nv.vv, calc_dif, n = N_rep) calc_dif &lt;- calc_fusion(media_cuartiles, identity, VV / NV) ejemplo$cuartiles_razon &lt;- valor_p(fusion |&gt; sample_frac(0.95), nv.vv, calc_dif, n = N_rep) ejemplo &lt;- read_rds(&quot;cache/ejemplo_p_val.rds&quot;) ejemplo$media_dif ## [1] 0.0658 ejemplo$media_dif_log ## [1] 0.018 ejemplo$mediana_razon ## [1] 0.049 ejemplo$cuartiles_razon ## [1] 0.0464 Si existen grados de libertad —muchas veces necesarios para hacer un análisis exitoso—, entonces los valores p pueden tener poco significado. Alternativas o soluciones El primer punto importante es reconocer que la mayor parte de nuestro trabajo es exploratorio (recordemos el proceso complicado del análisis de datos de refinamiento de preguntas). En este tipo de trabajo, reportar valores p puede tener poco sentido, y mucho menos tiene sentido aceptar algo verdadero cuando pasa un umbral de significancia dado. Nuestro interés principal al hacer análisis es: expresar correctamente, y de manera útil, la incertidumbre asociada a las conclusiones o patrones que mostramos (asociada a variación muestral, por ejemplo) con el objetivo que el proceso de toma de decisiones sea informado. Un resumen de un número (valor p, o el que sea) no puede ser tomado como criterio para tomar una decisión que generalmente es compleja. En la siguiente sección veremos cómo podemos mostrar parte de esa incertidumbre de manera más útil. Por otra parte, los estudios confirmatorios (donde se reportan valores p) también tienen un lugar. En áreas como la psicología, existen ahora movimientos fuertes en favor de la repetición de estudios prometedores pero donde hay sospecha de grados de libertad del investigador. Este movimiento sugiere dar valor a los estudios exploratorios que no reportan valor p, y posteriormente, si el estudio es de interés, puede intentarse una replicación confirmatoria, con potencia más alta y con planes de análisis predefinidos. Referencias "],["estimación-y-distribución-de-muestreo-1.html", "Sección 4 Estimación y distribución de muestreo Ejemplo: precios de casas Distribución de muestreo Más ejemplos El error estándar Calculando la distribución de muestreo Teorema central del límite Normalidad y gráficas de cuantiles normales Prueba de hipótesis de normalidad Ejemplo Más del Teorema central del límite", " Sección 4 Estimación y distribución de muestreo En esta sección discutiremos cuál el objetivo general del proceso de estimación, y cómo entender y manejar la variabilidad que se produce cuando aleatorizamos la selección de las muestras que utilizamos para hacer análisis. A diferencia de las pruebas de permutación, donde evaluábamos como cambiaría una estadísitica si un tratamiento o grupo se hubiera asignado de forma distinta, en la siguiente sección nos preguntamos como varía una estadística entre muestras. Por ejemplo, pasaremos de preguntar si una vacuna reduce el riesgo de una enfermedad a evaluar en que magnitud se reduce el riesgo de contraer la enfermedad. Ejemplo: precios de casas Supongamos que queremos conocer el valor total de las casas que se vendieron recientemente en una zona particular. Supondremos que tenemos un listado de las casas que se han vendido recientemente, pero en ese listado no se encuentra el precio de venta. Decidimos entonces tomar una muestra aleatoria de 100 de esas casas. Para esas casas hacemos trabajo de campo para averiguar el precio de venta. marco_casas &lt;- read_csv(&quot;data/casas.csv&quot;) set.seed(841) muestra_casas &lt;- sample_n(marco_casas, 100) |&gt; select(id, nombre_zona, area_habitable_sup_m2, precio_miles) sprintf(&quot;Hay %0.0f casas en total, tomamos muestra de %0.0f&quot;, nrow(marco_casas), nrow(muestra_casas)) ## [1] &quot;Hay 1144 casas en total, tomamos muestra de 100&quot; head(muestra_casas) ## # A tibble: 6 × 4 ## id nombre_zona area_habitable_sup_m2 precio_miles ## &lt;dbl&gt; &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 287 NAmes 161. 159 ## 2 755 NAmes 95.3 156 ## 3 1190 Gilbert 168. 189 ## 4 36 NridgHt 228. 309 ## 5 32 Sawyer 114. 149. ## 6 538 NAmes 80.3 111. Como tomamos una muestra aleatoria, intentamos estimar el valor total de las casas que se vendieron expandiendo el total muestral, es decir nuestro estimador \\(\\hat{t} = t(X_1,\\ldots X_{100})\\) del total poblacional \\(t\\) es \\[\\hat{t} = \\frac{N}{n} \\sum_{i=1}^{100} X_i = N\\bar{x}\\] Esta función implementa el estimador: n &lt;- nrow(muestra_casas) # tamaño muestra N &lt;- nrow(marco_casas) # tamaño población estimar_total &lt;- function(muestra_casas, N){ total_muestral &lt;- sum(muestra_casas$precio_miles) n &lt;- nrow(muestra_casas) # cada unidad de la muestra representa a N/n f_exp &lt;- N / n # estimador total es la expansión del total muestral estimador_total &lt;- f_exp * total_muestral res &lt;- tibble(total_muestra = total_muestral, factor_exp = f_exp, est_total_millones = estimador_total / 1000) res } estimar_total(muestra_casas, N) |&gt; mutate(across(where(is.numeric), \\(x) round(x, 2))) ## # A tibble: 1 × 3 ## total_muestra factor_exp est_total_millones ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 18444. 11.4 211 Sin embargo, si hubiéramos obtenido otra muestra, hubiéramos obtenido otra estimación diferente. Por ejemplo: estimar_total(sample_n(marco_casas, 100), N) |&gt; mutate(across(where(is.numeric), \\(x) round(x, 2))) ## # A tibble: 1 × 3 ## total_muestra factor_exp est_total_millones ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 17916. 11.4 205. El valor poblacional que buscamos estimar (nótese que en la práctica este no lo conocemos) es: # multiplicar por 1000 para que sea en millones de dólares total_pob &lt;- sum(marco_casas |&gt; pull(precio_miles)) / 1000 total_pob ## [1] 209.7633 Así que: Para algunas muestras esta estadística puede estar muy cercana al valor poblacional, pero para otras puede estar más lejana. Para entender qué tan buena es una estimación particular, entonces, tenemos que entender cuánta variabilidad hay de muestra a muestra debida a la aleatorización. Esto depende del diseño de la muestra y de la población de precios de casas (que no conocemos). Distribución de muestreo La distribución de muestreo de una estadística enumera los posibles resultados que puede tomar esa estadística sobre todas las muestras posibles. Este es el concepto básico para poder entender qué tan bien o mal estima un parámetro poblacional dado. En nuestro ejemplo anterior de precio de casas, no podemos calcular todas las posibles estimaciones bajo todas las posibles muestras, pero podemos aproximar repitiendo una gran cantidad de veces el proceso de muestreo, como hicimos al aproximar la distribución de permutaciones de estadísticas de prueba de las secciones anteriores. Empezamos repitiendo 10 veces y examinamos cómo varía nuestra estadística: replicar_muestreo &lt;- function(marco_casas, m = 500, n){ # n es el tamaño de muestra que se saca de marco_casas # m es el número de veces que repetimos el muestro de tamaño n resultados &lt;- map_df(1:m, function(id) { sample_n(marco_casas, n) |&gt; estimar_total(N) }, .id = &quot;id_muestra&quot;) } replicar_muestreo(marco_casas, m = 10, n = 100) |&gt; mutate(across(where(is.numeric), round, 1)) |&gt; formatear_tabla() id_muestra total_muestra factor_exp est_total_millones 1 17594.8 11.4 201.3 2 17423.9 11.4 199.3 3 18444.3 11.4 211.0 4 17696.6 11.4 202.4 5 17275.8 11.4 197.6 6 17867.6 11.4 204.4 7 18450.8 11.4 211.1 8 18187.2 11.4 208.1 9 18604.2 11.4 212.8 10 19144.4 11.4 219.0 Como vemos, hay variación considerable en nuestro estimador del total, pero la estimación que haríamos con cualquiera de estas muestras no es muy mala. Ahora examinamos un número más grande de simulaciones: replicaciones_1 &lt;- replicar_muestreo(marco_casas, m = 1500, n = 100) Y el siguiente histograma nos dice qué podemos esperar de la variación de nuestras estimaciones, y donde es más probable que una estimación particular caiga: graf_1 &lt;- ggplot(replicaciones_1, aes(x = est_total_millones)) + geom_histogram() + geom_vline(xintercept = total_pob, colour = &quot;red&quot;) + xlab(&quot;Millones de dólares&quot;) + scale_x_continuous(breaks = seq(180, 240, 10), limits = c(180, 240)) graf_1 Con muy alta probabilidad el error no será de más de unos 30 millones de dólares (o no más de 20% del valor poblacional). Definición Sea \\(X_1, X_2, \\ldots X_n\\) una muestra, y \\(T = t(X_1, X_2, \\ldots, X_n)\\) una estadística. La distribución de muestreo de \\(T\\) es la función de distribución de \\(T\\). Esta distribución es sobre todas las posibles muestras que se pueden obtener. Cuando usamos \\(T\\) para estimar algún parámetro poblacional \\(\\theta\\), decimos informalmente que el estimador es preciso si su distribución de muestreo está muy concentrada alrededor del valor \\(\\theta\\) que queremos estimar. Si la distribución de muestreo está concentrada en un conjunto muy grande o muy disperso, quiere decir que con alta probabilidad cuando obtengamos nuestra muestra y calculemos nuestra estimación, el resultado estará lejano del valor poblacional que nos interesa estimar. Veamos qué pasa cuando hacemos la muestra más grande en nuestro ejemplo: replicaciones_2 &lt;- replicar_muestreo(marco_casas, m = 1500, n = 250) Graficamos las dos distribuciones de muestreo juntas, y vemos cómo con mayor muestra obtenemos un estimador más preciso, y sin considerar el costo, preferimos el estimador mejor concentrado alrededor del valor que buscamos estimar. library(patchwork) graf_2 &lt;- ggplot(replicaciones_2, aes(x = est_total_millones)) + geom_histogram() + geom_vline(xintercept = total_pob, colour = &quot;red&quot;) + xlab(&quot;Millones de dólares&quot;) + scale_x_continuous(breaks = seq(180, 240, 10), limits = c(180, 240)) graf_1 + graf_2 Observación: a veces este concepto se confunde la distribución poblacional de las \\(X_i\\). Esto es muy diferente. Por ejemplo, en nuestro caso, el histograma de la distribución de valores poblacionales es ggplot(marco_casas, aes(x = precio_miles)) + geom_histogram() que en general no tiene ver mucho en escala o forma con la distribución de muestreo de nuestro estimador del total. Más ejemplos Podemos también considerar muestrear de poblaciones sintéticas o modelos probabilísticos que usamos para modelar poblaciones reales. Por ejemplo, supongamos que tomamos una muestra de tamaño 15 de la distribución uniforme en \\([0,1]\\). Es decir, cada \\(X_i\\) es un valor uniformemente distribuido en \\([0,1]\\), y las \\(X_i\\) se extraen independientemente unas de otras. Consideramos dos estadísticas de interés: La media muestral \\(T_1(X) = \\frac{1}{15}\\sum_{i = 1}^{15} X_i\\) El cuantil 0.75 de la muestra \\(T_2(X) = q_{0.75}(X)\\) ¿Cómo crees que se vean las distribuciones muestrales de estas estadísticas? ¿Alrededor de qué valores crees que concentren? ¿Crees que tendrán mucha o poca dispersión? ¿Qué forma crees que tengan? Para el primer caso hacemos: # simular replicar_muestreo_unif &lt;- function(est = mean, m, n = 15){ valores_est &lt;- map_dbl(1:m, ~ est(runif(n))) tibble(id_muestra = 1:m, estimacion = valores_est) } sim_estimador_1 &lt;- replicar_muestreo_unif(mean, 4000, 15) # graficar aprox de distribución de muestreo ggplot(sim_estimador_1, aes(x = estimacion)) + geom_histogram(bins = 40) + xlim(c(0, 1)) # simular para el máximo cuantil_75 &lt;- function(x) quantile(x, 0.75) sim_estimador_2 &lt;- replicar_muestreo_unif(cuantil_75, 4000, 15) # graficar distribución de muestreo ggplot(sim_estimador_2, aes(x = estimacion)) + geom_histogram(breaks = seq(0, 1, 0.02)) + xlim(c(0, 1)) Supón que tenemos una muestra de 30 observaciones de una distribución uniforme \\([0,b]\\). ¿Qué tan buen estimador de \\(b/2\\) es la media muestral? ¿Cómo lo cuantificarías? ¿Qué tan buen estimador del cuantil 0.8 de la distribución uniforme es el cuantil 0.8 muestral? ¿Qué desventajas notas en este estimador? El error estándar Una primera medida útil de la dispersión de la distribución de muestreo es su desviación estándar: la razón específica tiene qué ver con un resultado importante, el teorema central del límite, que veremos más adelante. En este caso particular, a esta desviación estándar se le llama error estándar: Definición A la desviación estándar de una estadística \\(T\\) le llamamos su error estándar, y la denotamos por \\(\\text{ee}(T)\\). A cualquier estimador de este error estándar lo denotamos como \\(\\hat{\\text{ee}}(T)\\). Este error estándar mide qué tanto varía el estimador \\(T\\) de muestra a muestra. Observación: es importante no confundir el error estándar con la desviación estándar de una muestra (o de la población). En nuestro ejemplo de las uniformes, la desviación estándar de las muestras varía como: map_dbl(1:10000, ~ sd(runif(15))) |&gt; quantile() |&gt; round(2) ## 0% 25% 50% 75% 100% ## 0.11 0.26 0.29 0.31 0.41 Mientras que el error estándar de la media es aproximadamente map_dbl(1:10000, ~ mean(runif(15))) |&gt; sd() ## [1] 0.07439575 y el error estándar del máximo es aproximadamente map_dbl(1:10000, ~ max(runif(15))) |&gt; sd() ## [1] 0.05928675 Como ejercicio para contrastar estos conceptos, puedes considerar: ¿Qué pasa con la desviación estándar de una muestra muy grande de uniformes? ¿Qué pasa con el error estándar de la media muestral de una muestra muy grande de uniformes? Ejemplo: valor de casas Consideramos el error estándar del estimador del total del inventario vendido, usando una muestra de 250 con el estimador del total que describimos arriba. Como aproximamos con simulación la distribución de muestreo, podemos hacer: ee_2 &lt;- replicaciones_2 |&gt; pull(est_total_millones) |&gt; sd() round(ee_2, 1) ## [1] 5.2 que está en millones de pesos y cuantifica la dispersión de la distribución de muestreo del estimador del total. Para tamaño de muestra 100, obtenemos más dispersión: ee_1 &lt;- replicaciones_1 |&gt; pull(est_total_millones) |&gt; sd() round(ee_1, 1) ## [1] 8.9 Nótese que esto es muy diferente, por ejemplo, a la desviación estándar poblacional o de una muestra. Estas dos cantidades miden la variabilidad del estimador del total. Calculando la distribución de muestreo En los ejemplos anteriores usamos simulación para obtener aproximaciones de la distribución de muestreo de algunos estimadores. También es posible: Hacer cálculos exactos a partir de modelos probabilísticos. Hacer aproximaciones asintóticas para muestras grandes (de las cuales la más importante es la que da el teorema central del límite). En los ejemplos de arriba, cuando muestreamos de la poblaciones, extrajimos las muestras de manera aproximadamente independiente. Cada observación \\(X_i\\) tiene la misma distribución y las \\(X_i\\)’s son independientes. Este tipo de diseños aleatorizados es de los más simples, y se llama muestreo aleatorio simple. En general, en esta parte haremos siempre este supuesto: Una muestra es iid (independiente e idénticamente distribuida) si es es un conjunto de observaciones \\(X_1,X_2, \\ldots X_n\\) independientes, y cada una con la misma distribución. En términos de poblaciones, esto lo logramos obteniendo cada observación de manera aleatoria con el mismo procedimiento. En términos de modelos probabilísticos, cada \\(X_i\\) se extrae de la misma distribución fija \\(F\\) (que pensamos como la “población”) de manera independiente. Esto lo denotamos por \\(X_i \\overset{iid}{\\sim} F.\\) Ejemplo Si \\(X_1, X_2, \\ldots X_n\\) es una muestra de uniformes independientes en \\([0,1]\\), ¿cómo calcularíamos la distribución de muestreo del máximo muestra \\(T_2 = \\max\\)? En este caso, es fácil calcular su función de distribución acumulada de manera exacta: \\[F_{\\max}(x) = P(\\max\\{X_1,X_2,\\ldots X_n\\} \\leq x)\\] El máximo es menor o igual a \\(x\\) si y sólo si todas las \\(X_i\\) son menores o iguales a \\(x\\), así que \\[F_{\\max} (x) = P(X_1\\leq x, X_2\\leq x, \\cdots, X_n\\leq x)\\] como las \\(X_i\\)’s son independientes entonces \\[F_{\\max}(x) = P(X_1\\leq x)P(X_2\\leq x)\\cdots P(X_n\\leq x) = x^n\\] para \\(x\\in [0,1]\\), pues para cada \\(X_i\\) tenemos \\(P(X_i\\leq x) = x\\). Así que no es necesario usar simulación para conocer esta distribución de muestreo. Derivando esta distribución acumulada obtenemos su densidad, que es \\[f(x) = nx^{n-1}\\] para \\(x\\in [0,1]\\), y es cero en otro caso. Si comparamos con nuestra simulación: teorica &lt;- tibble(x = seq(0, 1 ,0.001)) |&gt; mutate(f_dens = 15 * x^14) sim_estimador_3 &lt;- replicar_muestreo_unif(max, 4000, 15) ggplot(sim_estimador_3) + geom_histogram(aes(x = estimacion), breaks = seq(0, 1, 0.02)) + xlim(c(0.5, 1)) + # el histograma es de ancho 0.02 y el número de simulaciones 4000 geom_line(data = teorica, aes(x = x, y = (4000 * 0.02) * f_dens), colour = &quot;red&quot;, linewidth = 1.3) Y vemos que con la simulación obtuvimos una buena aproximación Nota: ¿cómo se relaciona un histograma con la función de densidad que genera los datos? Supón que \\(f(x)\\) es una función de densidad, y obtenemos un número \\(n\\) de simulaciones independientes. Si escogemos un histograma de ancho \\(\\Delta\\), ¿cuántas observaciones esperamos que caigan en un intervalo \\(I = [a - \\Delta/2, a + \\Delta/2]\\)?. La probabilidad de que una observación caiga en \\(I\\) es igual a \\[P(X\\in I) = \\int_I f(x)\\,dx = \\int_{a - \\Delta/2}^{a + \\Delta/2} f(x)\\,dx \\approx f(a) \\text{long}(I) = f(a) \\Delta\\] para \\(\\Delta\\) chica. Si nuestra muestra es de tamaño \\(n\\), el número esperado de observaciones que caen en \\(I\\) es entonces \\(nf(a)\\Delta\\). Eso explica el ajuste que hicimos en la gráfica de arriba. Otra manera de hacer es ajustando el histograma: si en un intervalo el histograma alcanza el valor \\(y\\), \\[f(a) = \\frac{y}{n\\Delta}\\] teorica &lt;- tibble(x = seq(0, 1 ,0.001)) |&gt; mutate(f_dens = 15*x^{14}) ggplot(sim_estimador_3) + geom_histogram(aes(x = estimacion, y = after_stat(density)), breaks = seq(0, 1, 0.02)) + xlim(c(0.5, 1)) + # el histograma es de ancho 0.02 y el número de simulaciones 4000 geom_line(data = teorica, aes(x = x, y = f_dens), colour = &quot;red&quot;, size = 1.3) Ejemplo Supongamos que las \\(X_i\\)’s son independientes y exponenciales con tasa \\(\\lambda &gt; 0\\). ¿Cuál es la distribución de muestreo de la suma \\(S = X_1 + \\cdots + X_n\\)? Sabemos que la suma de exponenciales independientes es una distribución gamma con parámetros \\((n, \\lambda)\\), y esta es la distribución de muestreo de nuestra estadística \\(S\\) bajo las hipótesis que hicimos. Podemos checar este resultado con simulación, por ejemplo para una muestra de tamaño \\(n=15\\) con \\(\\lambda = 1\\): replicar_muestreo_exp &lt;- function(est = mean, m, n = 150, lambda = 1){ valores_est &lt;- map_dbl(1:m, ~ est(rexp(n, lambda))) tibble(id_muestra = 1:m, estimacion = valores_est) } sim_estimador_1 &lt;- replicar_muestreo_exp(sum, 4000, n = 15) teorica &lt;- tibble(x = seq(0, 35, 0.001)) |&gt; mutate(f_dens = dgamma(x, shape = 15, rate = 1)) # graficar aprox de distribución de muestreo ggplot(sim_estimador_1) + geom_histogram(aes(x = estimacion, y = after_stat(density)), bins = 35) + geom_line(data = teorica, aes(x = x, y = f_dens), colour = &quot;red&quot;, linewidth = 1.2) Teorema central del límite Si consideramos los ejemplos de arriba donde tratamos con estimadores basados en una suma, total o una media —y en menor medida cuantiles muestrales—, vimos que las distribución de muestreo de las estadísticas que usamos tienden a tener una forma común. Estas son manifestaciones de una regularidad estadística importante que se conoce como el teorema central del límite: las distribuciones de muestreo de sumas y promedios son aproximadamente normales cuando el tamaño de muestra es suficientemente grande. Teorema central del límite Si \\(X_1,X_2, \\ldots, X_n\\) son independientes e idénticamente distribuidas con media \\(\\mu\\) y desviación estándar \\(\\sigma\\) finitas. Si el tamaño de muestra \\(n\\) es grande, entonces la distribución de muestreo de la media \\[\\bar{X} = \\frac{X_1 + X_2 +\\cdots + X_n}{n}\\] es aproximadamente normal con media \\(\\mu\\) y desviación estándar \\(\\sigma/\\sqrt{n}\\), que escribimos como \\[\\bar{X} \\xrightarrow{} \\mathsf{N}\\left( \\mu, \\frac{\\sigma}{\\sqrt{n}} \\right)\\] Adicionalmente, la distribución de la media estandarizada converge a una distribución normal estándar cuando \\(n\\) es grande: \\[\\sqrt{n} \\, \\left( \\frac{\\bar{X}-\\mu}{\\sigma} \\right) \\xrightarrow{} \\mathsf{N}(0, 1)\\] El error estándar de \\(\\bar{X}\\) es \\(\\text{ee}(\\bar{X}) = \\frac{\\sigma}{\\sqrt{n}}\\). Si tenemos una muestra, podemos estimar \\(\\sigma\\) con de la siguiente forma: \\[\\hat{\\sigma} =\\sqrt{\\frac{1}{n}\\sum_{i=1}^n (X_i - \\bar{X})^2}\\] o el más común (que explicaremos más adelante) \\[\\hat{s} = \\sqrt{\\frac{1}{n-1}\\sum_{i=1}^n (X_i - \\bar{X})^2}\\] Este hecho junto con el teorema del límite central nos dice cuál es la dispersión, y cómo se distribuyen las posibles desviaciones de la media muestral alrededor de la verdadera media poblacional. ¿Qué tan grande debe ser \\(n\\). Depende de cómo es la población. Cuando la población tiene una distribución muy sesgada, por ejemplo, \\(n\\) típicamente necesita ser más grande que cuando la población es simétrica si queremos obtener una aproximación “buena”. En algunos textos se afirma que \\(n\\geq 30\\) es suficiente para que la aproximación del Teorema central del límite (TCL) sea buena siempre y cuando la distribución poblacional no sea muy sesgada. Esta regla es más o menos arbitraria y es mejor no confiarse, pues fácilmente puede fallar. En la práctica es importante checar este supuesto, por ejemplo usando remuestreo (que veremos más adelante) Revisa los ejemplos que hemos visto hasta ahora (precios de casas, simulaciones de uniformes y exponenciales según las distintas estadísticas que consideramos). ¿Qué distribuciones de muestreo parecen tener una distribución normal? ¿Cómo juzgamos si estas distribuciones están cerca o lejos de una distribución normal? Normalidad y gráficas de cuantiles normales Para checar si una distribución de datos dada es similar a la normal, la herramienta mas común en estádística es la gráfica de cuantiles teóricos, que es una generalización de la gráfica de cuantiles que vimos anteriormente. En primer lugar, definimos la función de cuantiles de una distribución teórica, que es análoga a la que definimos para conjuntos de datos: Supongamos que tenemos una distribución acumulada teórica \\(\\Phi\\). Podemos definir el cuantil-\\(f\\) \\(q(f)\\) de \\(\\Phi\\) como el valor \\(q(f)\\) tal que \\[q(f) = \\text{argmin}\\{x \\,| \\, \\Phi(x)\\geq f \\}\\] En el caso de que \\(\\Phi\\) tiene densidad \\(\\phi\\), y su soporte es un intervalo (que puede ser de longitud infinita), entonces podemos también escribir \\(q(f)\\) como el valor único donde acumulamos \\(f\\) de la probabilidad \\[\\int_{-\\infty}^{q(f)} \\phi(x)\\,dx= f\\] Por ejemplo, para una densidad normal, abajo mostramos los cuantiles \\(f=0.5\\) (mediana) y \\(f=0.95\\) densidad_tbl &lt;- tibble(x = seq(0, 10, 0.01)) |&gt; mutate(densidad = dnorm(x, 5, 1)) # qnorm es la función de cuantiles de una normal cuantil_50 &lt;- qnorm(0.50, 5, 1) cuantil_90 &lt;- qnorm(0.95, 5, 1) # graficamos densidad_tbl &lt;- densidad_tbl |&gt; mutate(menor_50 = x &gt;= cuantil_50) |&gt; mutate(menor_90 = x &gt;= cuantil_90) g_normal_50 &lt;- ggplot(densidad_tbl, aes(y = densidad)) + ylab(&#39;f(x)&#39;) + geom_area(aes(x = x, fill = menor_50)) + geom_line(aes(x = x), alpha = 0.1) + geom_vline(xintercept = cuantil_50) + theme(legend.position = &quot;none&quot;) + annotate(&quot;text&quot;, 4.3, 0.2, label = &quot;50%&quot;) + labs(subtitle = paste0(&quot;q(0.5)=&quot;, round(cuantil_50,1))) g_normal_90 &lt;- ggplot(densidad_tbl, aes(y = densidad)) + ylab(&#39;f(x)&#39;) + geom_area(aes(x = x, fill = menor_90)) + geom_line(aes(x = x), alpha = 0.1) + geom_vline(xintercept = cuantil_90) + theme(legend.position = &quot;none&quot;) + annotate(&quot;text&quot;, 5.0, 0.2, label = &quot;95%&quot;) + labs(subtitle = paste0(&quot;q(0.95)=&quot;, round(cuantil_90,1))) g_normal_50 + g_normal_90 Como todas las distribuciones normales tienen la misma forma, y para obtener una de otra solo basta reescalar y desplazar, para calcular los cuantiles de una variable con distribución normal \\(\\mathsf{N}(\\mu, \\sigma)\\) sólo tenemos que saber los cuantiles de la distribución normal estándar \\(\\mathsf{N}(0,1)\\) y escalarlos apropiadamente por su media y desviación estándar \\[q(f, \\mu, \\sigma) = \\mu + \\sigma q(f, 0, 1)\\] Puedes demostrar esto sin mucha dificultad empezando con \\(P(X\\leq q) = f\\) y estandarizando: \\[P(X\\leq q(f, \\mu, \\sigma)) = f \\implies P\\left (Z\\leq \\frac{q(f,\\mu,\\sigma) - \\mu}{\\sigma}\\right)=f\\] y esto implica que \\[q(f, 0, 1) = \\frac{q(f,\\mu,\\sigma) - \\mu}{\\sigma} \\implies q(f, \\mu, \\sigma) = \\mu + \\sigma q(f, 0, 1)\\] De modo que si graficáramos los cuantiles de una distribución \\(\\mathsf{N}(\\mu, \\sigma)\\) contra los cuantiles de una distribución \\(\\mathsf{N}(0,1)\\), estos cuantiles aparecen en una línea recta: comparacion_tbl &lt;- tibble(f = seq(0.01, 0.99, 0.01)) |&gt; mutate(cuantiles_normal = qnorm(f, 5, 3), cuantiles_norm_estandar = qnorm(f, 0, 1)) ggplot(comparacion_tbl, aes(cuantiles_norm_estandar, cuantiles_normal)) + geom_point() Ahora supongamos que tenemos una muestra \\(X_1, \\ldots, X_n\\). ¿Cómo podemos checar si estos datos tienen una distribución aproximadamente normal? Si la muestra tiene una distribución aproximadamente \\(\\mathsf{N}(\\mu, \\sigma)\\), entonces sus cuantiles muestrales y los cuantiles respectivos de la normal estándar están aproximadamente en una línea recta. Primero veamos un ejemplo donde los datos son generados según una normal. set.seed(21) muestra &lt;- tibble(x_1 = rnorm(60, 10, 3), x_2 = rgamma(60, 2, 5)) graf_1 &lt;- ggplot(muestra, aes(sample = x_1)) + geom_qq(distribution = stats::qnorm) + geom_qq_line(colour = &quot;red&quot;) graf_2 &lt;- ggplot(muestra, aes(sample = x_2)) + geom_qq(distribution = stats::qnorm) + geom_qq_line(colour = &quot;red&quot;) graf_1 + graf_2 ¿Cuáles son los datos aproximadamente normales? ¿Cómo interpretas las desviaciones de la segunda gráfica en términos de la forma de la distribución normal? Prueba de hipótesis de normalidad Para interpretar las gráficas de cuantiles normales se requiere práctica, pues claramente los datos, aún cuando provengan de una distribución normal, no van a caer justo sobre una línea recta y observaremos variabilidad. Esto no descarta necesariamente que los datos sean aproximadamente normales. Con la práctica, generalmente esta gráfica nos da una buena indicación si el supuesto de normalidad es apropiado. Sin embargo, podemos hacer una prueba de hipótesis formal de normalidad si quisiéramos. La hipótesis nula es la siguiente: Los datos provienen de una distribución normal, y las desviaciones que observamos de una línea recta se deben a variación muestral. Podemos generar datos nulos tomando la media y desviación estándar muestrales, y generando muestras normales \\(\\mathsf{N}(\\bar{x}, s)\\). Usamos el lineup, produciendo datos bajo la hipótesis nula y viendo si podemos distinguir los datos. Por ejemplo: library(nullabor) lineup_normal &lt;- lineup(null_dist(&quot;x_2&quot;, dist = &quot;normal&quot;), muestra) ggplot(lineup_normal, aes(sample = x_2)) + geom_qq(distribution = stats::qnorm) + geom_qq_line(colour = &quot;red&quot;) + facet_wrap(~ .sample) En esta gráfica claramente rechazaríamos la hipótesis de normalidad. Sin embargo, para la primera muestra, obtenemos: lineup_normal &lt;- lineup(null_dist(&quot;x_1&quot;, dist = &quot;normal&quot;), muestra) ggplot(lineup_normal, aes(sample = x_1)) + geom_qq(distribution = stats::qnorm) + geom_qq_line(colour = &quot;red&quot;) + facet_wrap(~ .sample) Los datos verdaderos están en attr(lineup_normal, &quot;pos&quot;) ## [1] 4 Ejemplo Consideremos el problema de estimar el total poblacional de los precios de las casas que se vendieron. El estimador que usamos fue la suma muestral expandida por un factor. Vamos a checar qué tan cerca de la normalidad está la distribución de meustreo de esta estadística (\\(n=250\\)): replicaciones_2 ## # A tibble: 1,500 × 4 ## id_muestra total_muestra factor_exp est_total_millones ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 47089. 4.58 215. ## 2 2 45654. 4.58 209. ## 3 3 43973. 4.58 201. ## 4 4 45665. 4.58 209. ## 5 5 43551. 4.58 199. ## 6 6 46066. 4.58 211. ## 7 7 46626. 4.58 213. ## 8 8 47944. 4.58 219. ## 9 9 45381. 4.58 208. ## 10 10 46519. 4.58 213. ## # ℹ 1,490 more rows ggplot(replicaciones_2, aes(sample = est_total_millones)) + geom_qq(alpha = 0.3) + geom_qq_line(colour = &quot;red&quot;) Y vemos que en efecto el TCL aplica en este ejemplo, y la aproximación es buena. Aunque la población original es sesgada, la descripción de la distribución de muestreo es sorprendemente compacta: La distribución de muestreo de nuestro estimador del total \\(\\hat{t}\\) es aproximadamente normal con media \\(\\bar{x}\\) y desviación estándar \\(s\\), donde: mu &lt;- mean(replicaciones_2$est_total_millones) s &lt;- sd(replicaciones_2$est_total_millones) c(mu = mu, s = s) |&gt; round(2) ## mu s ## 209.90 5.24 Estas cantidades están en millones de dólares. Ejemplo Supongamos que queremos calcular la probabilidad que la suma de 30 variables uniformes en \\([0,1]\\) independientes sea mayor que 18. Podríamos aproximar esta cantidad usando simulación. Otra manera de aproximar esta cantidad es con el TCL, de la siguiente forma: Si \\(S=X_1 + X_2 + X_{30}\\), entonces la media de \\(S\\) es 15 (¿cómo se calcula?) y su desviación estándar es \\(\\sqrt{\\frac{30}{12}}\\). La suma es entonces aproximadamente \\(\\mathsf{N}\\left(15, \\sqrt{\\frac{30}{12}}\\right)\\). Entonces \\[P(S &gt; 18) = P \\left (\\frac{S - 15}{\\sqrt{\\frac{30}{12}}} &gt; \\frac{18 - 15}{\\sqrt{\\frac{30}{12}}}\\right) \\approx P(Z &gt; 1.897)\\] donde \\(Z\\) es normal estándar. Esta última cantidad la calculamos usando la función de distribución de la normal estándar, y nuestra aproximación es 1 - pnorm(1.897) ## [1] 0.02891397 Podemos checar nuestro cálculo usando simulación: tibble(n_sim = 1:100000) |&gt; mutate(suma = map_dbl(n_sim, ~ sum(runif(30)))) |&gt; summarise(prob_may_18 = mean(suma &gt; 18), .groups = &quot;drop&quot;) ## # A tibble: 1 × 1 ## prob_may_18 ## &lt;dbl&gt; ## 1 0.0280 Y vemos que la aproximación normal es buena para fines prácticos. Usando simulaciones haz un histograma que aproxime la distribución de muestreo de \\(S\\). Haz una gráfica de cuantiles normales para checar la normalidad de esta distribución. Ejemplo Cuando el sesgo de la distribución poblacional es grande, puede ser necesario que \\(n\\) sea muy grande para que la aproximación normal sea aceptable para el promedio o la suma. Por ejemplo, si tomamos una gamma con parámetro de forma chico, \\(n = 30\\) no es suficientemente bueno, especialmente si quisiéramos aproximar probabilidades en las colas de la distribución: sims_gamma &lt;- map_df(1:2000, ~ tibble(suma = sum(rgamma(30, 0.1, 1))), .id = &quot;n_sim&quot;) ggplot(sims_gamma, aes(x = suma)) + geom_histogram() Más del Teorema central del límite El teorema central del límite aplica a situaciones más generales que las del enunciado del teorema básico. Por ejemplo, aplica a poblaciones finitas (como vimos en el ejemplo de las casas), en 1960 Jaroslav Hajek demostró una versión del TCL bajo muestreo sin reemplazo. Mas allá de la media muestral, el TCL se puede utilizar para más estadísticas ya que muchas pueden verse como promedios, como totales o errores estándar. El TLC se ha generalizado incluso para cuantiles muestrales. Es importante notar que la calidad de la aproximación del TCL depende de características de la población y también del tamaño de muestra \\(n\\). Para ver si el TCL aplica, podemos hacer ejercicios de simulación bajo diferentes supuestos acerca de la población. También veremos más adelante, con remuestreo, maneras de checar si es factible el TCL dependiendo del análisis de una muestra dada que tengamos. El TCL era particularmente importante en la práctica antes de que pudiéramos hacer simulación por computadora. Era la única manera de aproximar y entender la distribución muestral fuera de cálculos analíticos (como los que hicimos para el máximo de un conjunto de uniformes, por ejemplo). Hoy en día, veremos que podemos hacer simulación para obtener respuestas más exactas, particularmente en la construcción de intervalos de confianza, por ejemplo. Dependemos menos de resultados asintóticos, como el TCL. Cuando aproximamos una distribución discreta mediante la distribución normal, conviene hacer correcciones de continuidad, como se explica en (Chihara and Hesterberg 2018), 4.3.2. Referencias "],["intervalos-de-confianza-y-remuestreo.html", "Sección 5 Intervalos de confianza y remuestreo Ejemplo introductorio La idea del bootstrap El principio de plug-in Discusión: propiedades de la distribución bootstrap Error estándar bootstrap e intervalos normales Ejemplo: inventario de casas vendidas Calibración de intervalos de confianza Interpretación de intervalos de confianza Sesgo Intervalos bootstrap de percentiles Bootstrap para dos muestras Bootstrap y otras estadísticas Bootstrap y estimadores complejos: tablas de perfiles Bootstrap y muestras complejas Bootstrap en R Conclusiones y observaciones", " Sección 5 Intervalos de confianza y remuestreo En la sección anterior, vimos el concepto de distribución de muestreo de una estadística que queremos utilizar para estimar un valor poblacional, y vimos que con esta distribución podíamos evaluar qué tan preciso es nuestro estimador evaluando qué tan concentrada está esta distribución alrededor del valor poblacion que queremos estimar. Sin embargo, en los ejemplos que vimos la población era conocida: ya sea que tuviéramos toda la población finita disponible (como el ejemplo de las casas), o donde la población estaba definida por un modelo teórico de probabilidad (como los ejemplos de las distribuciones uniforme o exponencial). Ahora vemos qué hacer en el caso que realmente nos interesa: solo tenemos una muestra disponible, y la población es desconocida. Todo lo que tenemos es una muestra y una estimación basada en la muestra, y requerimos estimar la distribución de muestreo de la estadística de interés. El enfoque que presentaremos aquí es uno de los más flexibles y poderosos que están disponibles para este problema: el método bootstrap o de remuestreo. En primer lugar explicamos el concepto de intervalo de confianza, que es una manera resumida de evaluar la precisión de nuestras estimaciones. Ejemplo introductorio Regresamos a nuestro ejemplo anterior donde muestreamos 3 grupos, y nos preguntábamos acerca de la diferencia de sus medianas. En lugar de hacer pruebas de permutaciones (ya sea pruebas gráficas o alguna prueba de permutaciones para media o mediana, por ejemplo), podríamos considerar qué tan precisa es cada una de nuestras estimaciones para las medianas de los grupos. Nuestros resultados podríamos presentarlos como sigue. Este código lo explicaremos más adelante, por el momento consideramos la gŕafica resultante: set.seed(8) pob_tab &lt;- tibble(id = 1:2000, x = rgamma(2000, 4, 1), grupo = sample(c(&quot;a&quot;,&quot;b&quot;, &quot;c&quot;), 2000, prob = c(4,2,1), replace = T)) muestra_tab &lt;- pob_tab |&gt; slice_sample(n = 125) g_1 &lt;- ggplot(muestra_tab, aes(x = grupo, y = x)) + geom_boxplot(outlier.alpha = 0) + geom_jitter(alpha = 0.3) + labs(subtitle = &quot;Muestra \\n&quot;) + ylim(c(0,14)) ## Hacemos bootstrap fun_boot &lt;- function(datos){ datos |&gt; group_by(grupo) |&gt; slice_sample(prop = 1, replace = TRUE) } reps_boot &lt;- map_df(1:2000, function(i){ muestra_tab |&gt; fun_boot() |&gt; group_by(grupo) |&gt; summarise(mediana = median(x), .groups = &quot;drop&quot;)}, .id = &#39;rep&#39;) resumen_boot &lt;- reps_boot |&gt; group_by(grupo) |&gt; summarise(ymin = quantile(mediana, 0.025), ymax = quantile(mediana, 0.975), .groups = &quot;drop&quot;) |&gt; left_join(muestra_tab |&gt; group_by(grupo) |&gt; summarise(mediana = median(x))) g_2 &lt;- ggplot(resumen_boot, aes(x = grupo, y = mediana, ymin = ymin, ymax = ymax)) + geom_linerange() + geom_point(colour = &quot;red&quot;, size = 2) + ylim(c(0,14)) + labs(subtitle = &quot;Intervalos de 95% \\n para la mediana&quot;) g_1 + g_2 Donde: En rojo está nuestro = puntual de la mediana de cada grupo (la mediana muestral), y Las segmentos muestran un intervalo de confianza del 95% para nuestra estimación de la mediana: esto quiere decir que los valores poblacionales tienen probabilidad aproximada de 95% de estar dentro del intervalo. Este análisis comunica correctamente que tenemos incertidumbre alta acerca de nuestras estimaciones (especialmente grupos b y c), y que no tenemos mucha evidencia de que el grupo b tenga una mediana poblacional considerablemente más alta que a o c. En muchos casos es más útil presentar la información de esta manera que usando alguna prueba de hipótesis. La idea del bootstrap Como explicamos, el problema que tenemos ahora es que normalmente sólo tenemos una muestra, así que no es posible calcular las distribuciones de muestreo como hicimos en la sección anterior y así evaluar qué tan preciso es nuestro estimador. Sin embargo, podemos hacer lo siguiente: Supongamos que tenemos una muestra \\(X_1,X_2,\\dots, X_n\\) independientes de alguna población desconocida y un estimador \\(T=t(X_1,\\dots, X_n)\\) Mundo poblacional Si tuviéramos la distribución poblacional, simulamos muestras iid para aproximar la distribución de muestreo de nuestro estimador, y así entender su variabilidad. Pero no tenemos la distribución poblacional. Sin embargo, podemos estimar la distribución poblacional con nuestros valores muestrales. Mundo bootstrap Si usamos la estimación del inciso 3, entonces usando el inciso 1 podríamos tomar muestras de nuestros datos muestrales, como si fueran de la población, y usando el mismo tamaño de muestra. El muestreo lo hacemos con reemplazo de manera que produzcamos muestras independientes de la misma “población estimada”, que es la muestra. Evaluamos nuestra estadística en cada una de estas remuestras, a estas les llamamos replicaciones bootstrap. A la distribución de las replicaciones le llamamos distribución bootstrap o distribución de remuestreo del estimador. Usamos la distribución bootstrap para estimar la variabilidad en nuestra estimación con la muestra original. Veamos que sucede para un ejemplo concreto, donde nos interesa estimar la media de los precios de venta de una población de casas. Tenemos nuestra muestra: set.seed(2112) poblacion_casas &lt;- read_csv(&quot;data/casas.csv&quot;) muestra &lt;- slice_sample(poblacion_casas, n = 200, replace = TRUE) mean(muestra$precio_miles) ## [1] 179.963 Esta muestra nos da nuestro estimador de la distribución poblacional: bind_rows(muestra |&gt; mutate(tipo = &quot;muestra&quot;), poblacion_casas |&gt; mutate(tipo = &quot;población&quot;)) |&gt; ggplot(aes(sample = precio_miles, colour = tipo, group = tipo)) + geom_qq(distribution = stats::qunif, alpha = 0.4, size = 1) + facet_wrap(~ tipo) O con histogramas: bind_rows(muestra |&gt; mutate(tipo = &quot;muestra&quot;), poblacion_casas |&gt; mutate(tipo = &quot;población&quot;)) |&gt; ggplot(aes(x = precio_miles, group = tipo)) + geom_histogram(aes(y=..density..), binwidth = 50) + facet_wrap(~ tipo) Y vemos que la aproximación es razonable en las partes centrales de la distribución. Ahora supongamos que nos interesa cuantificar la precisión de nuestra estimación de la media poblacional de precios de casas, y usaremos la media muestral para hacer esto. Para nuestra muestra, nuestra estimación puntual es: media &lt;- mean(muestra$precio_miles) media ## [1] 179.963 Y recordamos que para aproximar la distribución de muestreo podíamos muestrear repetidamente la población y calcular el valor del estimador en cada una de estas muestras. Aquí no tenemos la población, pero tenemos una estimación de la población: la muestra obtenida. Así que para evaluar la variabilidad de nuestro estimador, entramos en el mundo bootstrap, y consideramos que la población es nuestra muestra. Podemos entonces extraer un número grande de muestras con reemplazo de tamaño 200 de la muestra: el muestreo debe ser análogo al que se tomó para nuestra muestra original. Evaluamos nuestra estadística (en este caso la media) en cada una de estas remuestras: media_muestras &lt;- map_dbl(1:5000, ~ muestra |&gt; slice_sample(n = 200, replace = TRUE) |&gt; summarise(media_precio = mean(precio_miles), .groups = &quot;drop&quot;) |&gt; pull(media_precio)) Y nuestra estimación de la distribución de muestreo para la media es entonces: bootstrap &lt;- tibble(media = media_muestras) g_cuantiles &lt;- ggplot(bootstrap, aes(sample = media)) + geom_qq(distribution = stats::qunif) g_histograma &lt;- ggplot(bootstrap, aes(x = media)) + geom_histogram(binwidth = 2) g_cuantiles + g_histograma A esta le llamamos la distribución bootstrap (o de remuestreo) de la media, que definimos más abajo. Ahora podemos calcular un intervalo de confianza del 90% simplemente calculando los cuantiles de esta distribución (no son los cuantiles de la muestra original!): limites_ic &lt;- quantile(media_muestras, c(0.05, 0.95)) |&gt; round() limites_ic ## 5% 95% ## 171 189 Presentaríamos nuestro resultado como sigue: nuestra estimación puntual de la mediana es 180, con un intervalo de confianza del 90% de (171, 189) Otra cosa que podríamos hacer para describir la dispersión de nuestro estimador es calcular el error estándar de remuestreo, que estima el error estándar de la distribución de muestreo: ee_boot &lt;- sd(media_muestras) round(ee_boot, 2) ## [1] 5.39 Definición. Sea \\(X_1,X_2,\\ldots,X_n\\) una muestra independiente y idénticamente distribuida, y \\(T=t(X_1, X_2, \\ldots, X_n)\\) una estadística. Supongamos que sus valores que obervamos son \\(x_1, x_2,\\ldots, x_n\\). La distribución bootstrap, o distribución de remuestreo, de \\(T\\) es la distribución de \\(T^*=t(X_1^*, X_2^*, \\dots X_n^*)\\), donde cada \\(X_i^*\\) se obtiene tomando al azar uno de los valores de \\(x_1,x_2,\\ldots, x_n\\). Otra manera de decir esto es que la remuestra \\(X_1^*, X_2^*, \\ldots, X_n^*\\) es una muestra con reemplazo de los valores observados \\(x_1, x_2, \\ldots, x_n\\) Ejemplo. Si observamos la muestra muestra &lt;- sample(1:20, 5) muestra ## [1] 6 10 7 3 14 Una remuestra se obtiene: sample(muestra, size = 5, replace = TRUE) ## [1] 7 3 7 10 6 Nótese que algunos valores de la muestra original pueden aparecer varias veces, y otros no aparecen del todo. La idea del bootstrap (no paramétrico). La muestra original es una aproximación de la población de donde fue extraída. Así que remuestrear la muestra aproxima lo que pasaría si tomáramos muestras de la población. La distribución de remuestreo de una estadística, que se construye tomando muchas remuestras, aproxima la distribución de muestreo de la estadística. Y el proceso que hacemos es: Remuestreo para una población. Dada una muestra de tamaño \\(n\\) de una población, Obtenemos una remuestra de tamaño \\(n\\) con reemplazo de la muestra original y calculamos la estadística de interés. Repetimos este remuestreo muchas veces (por ejemplo, 10,000). Construímos la distribución bootstrap, y examinamos sus características (dónde está centrada, dispersión y forma). El principio de plug-in La idea básica detrás del bootstrap es el principio de plug-in para estimar parámetros poblacionales: si queremos estimar una cantidad poblacional, calculamos esa cantidad poblacional con la muestra obtenida. Es un principio común en estadística. Por ejemplo, si queremos estimar la media o desviación estándar poblacional, usamos la media muestral o la desviación estándar muestral. Si queremos estimar un cuantil de la población usamos el cuantil correspondiente de la muestra, y así sucesivamente. En todos estos casos, lo que estamos haciendo es: Tenemos una fórmula para la cantidad poblacional de interés en términos de la distribución poblacional. Tenemos una muestra, la distribución que da esta muestra se llama distribución empírica (\\(\\hat{F}(x) = \\frac{1}{n}\\{\\#valores \\le x\\}\\)). Contruimos nuestro estimador, de la cantidad poblacional de interés, “enchufando” la distribución empírica de la muestra en la fórmula del estimador. En el bootstrap aplicamos este principio simple a la distribución de muestreo: Si tenemos la población, podemos calcular la distribución de muestreo de nuestro estimador tomando muchas muestras de la población. Estimamos la poblacion con la muestra y enchufamos en la frase anterior: estimamos la distribución de muestreo de nuestro estimador tomando muchas muestras de la muestra. Nótese que el proceso de muestreo en el último paso debe ser el mismo que se usó para tomar la muestra original. Estas dos imágenes simuladas con base en un ejemplo de Chihara and Hesterberg (2018) muestran lo que acabamos de describir: Figure 5.1: Mundo Real Figure 5.2: Mundo Bootstrap Observación 1. Veremos ejemplos más complejos, pero nótese que si la muestra original son observaciones independientes obtenidas de la distribución poblacional, entonces logramos esto en las remuestras tomando observaciones con reemplazo de la muestra. Igualmente, las remuestras deben ser del mismo tamaño que la muestra original. ¿Porqué no funcionaría tomar muestras sin reemplazo? Piensa si hay independencia entre las observaciones de la remuestra, y cómo serían las remuestras sin reemplazo. ¿Por qué no se puede hacer bootstrap si no conocemos cómo se obtuvo la muestra original? Observación 2. Estos argumentos se pueden escribir con fórmulas usando por ejemplo la función de distribución acumulada \\(F\\) de la población y su estimador, que es la función empírica \\(\\hat{F}\\). Si \\(\\theta = t(F)\\) es una cantidad poblacional que queremos estimar, su estimador plug-in es \\(\\hat{\\theta} = t(\\hat{F})\\). Observación 3: La distribución empírica \\(\\hat{F}\\) es un estimador “razonable” de la distribución poblacional \\(F,\\) pues por el teorema de Glivenko-Cantelli (ver Wasserman (2013), o aquí), \\(\\hat{F}\\) converge a \\(F\\) cuando el tamaño de muestra \\(n\\to\\infty\\), lo cual es intuitivamente claro. Ejemplo En el ejemplo de tomadores de té, podemos estimar la proporción de tomadores de té que prefiere el té negro usando nuestra muestra: te &lt;- read_csv(&quot;data/tea.csv&quot;) |&gt; rowid_to_column() |&gt; select(rowid, Tea, sugar) te |&gt; mutate(negro = ifelse(Tea == &quot;black&quot;, 1, 0)) |&gt; summarise(prop_negro = mean(negro), n = length(negro), .groups = &quot;drop&quot;) ## # A tibble: 1 × 2 ## prop_negro n ## &lt;dbl&gt; &lt;int&gt; ## 1 0.247 300 ¿Cómo evaluamos la precisión de este estimador? Supondremos que el estudio se hizo tomando una muestra aleatoria simple de tamaño 300 de la población de tomadores de té que nos interesa. Podemos entonces usar el bootstrap: # paso 1: define el estimador calc_estimador &lt;- function(datos){ prop_negro &lt;- datos |&gt; mutate(negro = ifelse(Tea == &quot;black&quot;, 1, 0)) |&gt; summarise(prop_negro = mean(negro), n = length(negro), .groups = &quot;drop&quot;) |&gt; pull(prop_negro) prop_negro } # paso 2: define el proceso de remuestreo muestra_boot &lt;- function(datos){ #tomar muestra con reemplazo del mismo tamaño slice_sample(datos, prop = 1, replace = TRUE) } # paso 3: remuestrea y calcula el estimador prop_negro_tbl &lt;- tibble(prop_negro = map_dbl(1:10000, ~ calc_estimador(muestra_boot(datos = te)))) # paso 4: examina la distribución bootstrap prop_negro_tbl |&gt; ggplot(aes(x = prop_negro)) + geom_histogram(bins = 15) + geom_vline(xintercept = calc_estimador(te), color = &quot;red&quot;) Y podemos evaluar varios aspectos, por ejemplo dónde está centrada y qué tan dispersa es la distribución bootstrap: prop_negro_tbl |&gt; summarise(media = mean(prop_negro), ee = sd(prop_negro), cuantil_75 = quantile(prop_negro, 0.75), cuantil_25 = quantile(prop_negro, 0.25), .groups = &quot;drop&quot;) |&gt; mutate(across(where(is.numeric), round, 3)) |&gt; pivot_longer(cols = everything()) ## # A tibble: 4 × 2 ## name value ## &lt;chr&gt; &lt;dbl&gt; ## 1 media 0.247 ## 2 ee 0.025 ## 3 cuantil_75 0.263 ## 4 cuantil_25 0.23 –&gt; –&gt; Discusión: propiedades de la distribución bootstrap Uasremos la distribución bootstrap principalmente para evaluar la variabilidad de nuestros estimadores (y también otros aspectos como sesgo) estimando la dispersión de la distribución de muestreo. Sin embargo, es importante notar que no la usamos, por ejemplo, para saber dónde está centrada la distribución de muestreo, o para “mejorar” la estimación remuestreando. Ejemplo En este ejemplo, vemos 20 muestras de tamaño 200, y evaluamos cómo se ve la aproximación a la distribución de la población (rojo): Podemos calcular las distribuciones de remuestreo (bootstrap) para cada muestra, y compararlas con la distribución de muestreo real. # paso 1: define el estimador calc_estimador &lt;- function(datos){ media_precio &lt;- datos |&gt; summarise(media = mean(precio_miles), .groups = &quot;drop&quot;) |&gt; pull(media) media_precio } # paso 2: define el proceso de remuestreo muestra_boot &lt;- function(datos, n = NULL){ #tomar muestra con reemplazo del mismo tamaño if(is.null(n)){ m &lt;- slice_sample(datos, prop = 1, replace = TRUE)} else { m &lt;- slice_sample(datos, n = n, replace = TRUE) } m } dist_boot &lt;- datos_sim |&gt; filter(tipo == &quot;muestras&quot;) |&gt; select(precio_miles, rep) |&gt; group_by(rep) |&gt; nest() |&gt; mutate(precio_miles = map(data, function(data){ tibble(precio_miles = map_dbl(1:1000, ~ calc_estimador(muestra_boot(data)))) })) |&gt; select(rep, precio_miles) |&gt; unnest() dist_muestreo &lt;- datos_sim |&gt; filter(tipo == &quot;población&quot;) |&gt; group_by(rep) |&gt; nest() |&gt; mutate(precio_miles = map(data, function(data){ tibble(precio_miles = map_dbl(1:1000, ~ calc_estimador(muestra_boot(data, n = 200)))) })) |&gt; select(rep, precio_miles) |&gt; unnest() Obsérvese que: En algunos casos la aproximación es mejor que en otros (a veces la muestra tiene valores ligeramente más altos o más bajos). La dispersión de cada una de estas distribuciones bootstrap es similar a la de la verdadera distribución de muestreo (en rojo), pero puede está desplazada dependiendo de la muestra original que utilizamos. Adicionalmente, los valores centrales de la distribución de bootstrap tiende cubrir el verdadero valor que buscamos estimar, que es: poblacion_casas |&gt; summarise(media = mean(precio_miles), .groups = &quot;drop&quot;) ## # A tibble: 1 × 1 ## media ## &lt;dbl&gt; ## 1 183. Variación en distribuciones bootstrap En el proceso de estimación bootstrap hay dos fuentes de variación pues: La muestra original se selecciona con aleatoriedad de una población. Las muestras bootstrap se seleccionan con aleatoriedad de la muestra original. Esto es: La estimación bootstrap ideal es un resultado asintótico \\(B=\\infty\\), en esta caso \\(\\hat{\\textsf{se}}_B\\) iguala la estimación plug-in \\(se_{P_n}\\). En el proceso de bootstrap podemos controlar la variación del segundo aspecto, conocida como implementación de muestreo Monte Carlo, y la variación Monte Carlo decrece conforme incrementamos el número de muestras. Podemos eliminar la variación Monte Carlo si seleccionamos todas las posibles muestras con reemplazo de tamaño \\(n\\), hay \\({2n-1}\\choose{n}\\) posibles muestras y si seleccionamos todas obtenemos \\(\\hat{\\textsf{se}}_\\infty\\) (bootstrap ideal), sin embargo, en la mayor parte de los problemas no es factible proceder así. En la siguiente gráfica mostramos 6 posibles muestras de tamaño 50 simuladas de la población, para cada una de ellas se graficó la distribución empírica y se se realizan histogramas de la distribución bootstrap con \\(B=30\\) y \\(B=1000\\), en cada caso hacemos dos repeticiones, notemos que cuando el número de muestras bootstrap es grande las distribuciones bootstrap son muy similares (para una muestra de la población dada), esto es porque disminuimos el erro Monte Carlo. También vale la pena recalcar que la distribución bootstrap está centrada en el valor observado en la muestra (línea azúl punteada) y no en el valor poblacional sin embargo la forma de la distribución es similar a lo largo de las filas. Entonces, ¿cuántas muestras bootstrap? Incluso un número chico de replicaciones bootstrap, digamos \\(B=25\\) es informativo, y \\(B=50\\) con frecuencia es suficiente para dar una buena estimación de \\(se_P(\\hat{\\theta})\\) (Efron and Tibshirani (1993)). Cuando se busca estimar error estándar Chihara and Hesterberg (2018) recomienda \\(B=1000\\) muestras, o \\(B=10,000\\) muestras dependiendo la presición que se busque. Error estándar bootstrap e intervalos normales Ahora podemos construir nuestra primera versión de intervalos de confianza basados en la distribución bootstrap. Supongamos que queremos estimar una cantidad poblacional \\(\\theta\\) con una estadística \\(\\hat{\\theta} = t(X_1,\\ldots, X_n)\\), donde \\(X_1,\\ldots, X_n\\) es una muestra independiente e idénticamente distribuida de la población. Suponemos además que la distribución muestral de \\(\\hat{\\theta}\\) es aproximadamente normal (el teorema central del límite aplica), y está centrada en el verdadero valor poblacional \\(\\theta\\). Ahora queremos construir un intervalo que tenga probabilidad 95% de cubrir al valor poblacional \\(\\theta\\). Tenemos que \\[P(-2\\mathsf{ee}(\\hat{\\theta}) &lt; \\hat{\\theta} - \\theta &lt; 2\\mathsf{ee}(\\hat{\\theta})) \\approx 0.95\\] por las propiedades de la distribución normal (\\(P(-2\\sigma &lt; X -\\mu &lt; 2\\sigma)\\approx 0.95\\) si \\(X\\) es normal con media \\(\\mu\\) y desviación estándar \\(\\sigma\\)). Entonces \\[P(\\hat{\\theta} - 2\\mathsf{ee}(\\hat{\\theta}) &lt; \\theta &lt; \\hat{\\theta} + 2\\mathsf{ee}(\\hat{\\theta})) \\approx 0.95\\] Es decir, la probabilidad de que el verdadero valor poblacional \\(\\theta\\) esté en el intervalo \\[[\\hat{\\theta} - 2\\mathsf{ee}(\\hat{\\theta}), \\hat{\\theta} + 2\\mathsf{ee}(\\hat{\\theta})]\\] es cercano a 0.95. En este intervalo no conocemos el error estándar (es la desviación estándar de la distribución de muestreo de \\(\\hat{\\theta}\\)), y aquí es donde entre la distribución bootstrap, que aproxima la distribución de muestreo. Lo estimamos con \\[\\hat{\\mathsf{ee}}_{\\textrm{boot}}(\\hat{\\theta})\\] que es la desviación estándar de la distribución bootsrap. Definición. El error estándar bootstrap \\(\\hat{\\mathsf{ee}}_{\\textrm{boot}}(\\hat{\\theta})\\) se define como la desviación estándar de la distribución bootstrap de \\(\\theta\\). El intervalo de confianza normal bootstrap al 95% está dado por \\[[\\hat{\\theta} - 2\\hat{\\mathsf{ee}}_{\\textrm{boot}}(\\hat{\\theta}), \\hat{\\theta} + 2\\hat{\\mathsf{ee}}_{\\textrm{boot}}(\\hat{\\theta})].\\] Nótese que hay varias cosas qué revisar aquí: que el teorema central del límite aplica y que la distribución de muestreo de nuestro estimador está centrado en el valor verdadero. Esto en algunos casos se puede demostrar usando la teoría, pero más abajo veremos comprobaciones empíricas. Ejemplo: tomadores de té negro Consideremos la estimación que hicimos de el procentaje de tomadores de té que toma té negro: # paso 1: define el estimador calc_estimador &lt;- function(datos){ prop_negro &lt;- datos |&gt; mutate(negro = ifelse(Tea == &quot;black&quot;, 1, 0)) |&gt; summarise(prop_negro = mean(negro), n = length(negro)) |&gt; pull(prop_negro) prop_negro } prop_hat &lt;- calc_estimador(te) prop_hat |&gt; round(2) ## [1] 0.25 Podemos graficar su distribución bootstrap —la cual simulamos arriba—. g_hist &lt;- ggplot(prop_negro_tbl, aes(x = prop_negro)) + geom_histogram(bins = 15) g_qq_normal &lt;- ggplot(prop_negro_tbl, aes(sample = prop_negro)) + geom_qq() + geom_qq_line(colour = &quot;red&quot;) g_hist + g_qq_normal Y notamos que la distribución bootstrap es aproximadamente normal. Adicionalmente, vemos que el sesgo tiene un valor estimado de: media_boot &lt;- prop_negro_tbl |&gt; pull(prop_negro) |&gt; mean() media_boot - prop_hat ## [1] 0.0004393333 De esta forma, hemos verificado que: La distribución bootstrap es aproximadamente normal (ver gráfica de cuantiles normales); La distribución bootstrap es aproximadamente insesgada. Lo cual nos lleva a construir intervalos de confianza basados en la distribución normal. Estimamos el error estándar con la desviación estándar de la distribución bootstrap ee_boot &lt;- prop_negro_tbl |&gt; pull(prop_negro) |&gt; sd() ee_boot ## [1] 0.02485138 y construimos un intervalo de confianza del 95%: intervalo_95 &lt;- c(prop_hat - 2 * ee_boot, prop_hat + 2 * ee_boot) intervalo_95 |&gt; round(2) ## [1] 0.2 0.3 Este intervalo tiene probabilidad del 95% de capturar al verdadero poblacional. Con alta probabilidad, entonces, el porcentaje de tomadores de té en la población está entre 0.2 y 0.3. Ejemplo: inventario de casas vendidas Ahora consideremos el problema de estimar el total del valor de las casas vendidas en un periodo. Tenemos una muestra de tamaño \\(n=150\\): # muestra original set.seed(121) muestra_casas &lt;- slice_sample(poblacion_casas, n = 150) # paso 1: define el estimador calc_estimador_casas &lt;- function(datos){ N &lt;- nrow(poblacion_casas) n &lt;- nrow(datos) total_muestra &lt;- sum(datos$precio_miles) estimador_total &lt;- (N / n) * total_muestra estimador_total } # paso 2: define el proceso de remuestreo muestra_boot &lt;- function(datos){ #tomar muestra con reemplazo del mismo tamaño slice_sample(datos, prop = 1, replace = TRUE) } # paso 3: remuestrea y calcula el estimador totales_boot &lt;- tibble(total_boot = map_dbl(1:5000, ~ calc_estimador_casas(muestra_boot(muestra_casas)))) # paso 4: examina la distribución bootstrap g_hist &lt;- totales_boot |&gt; ggplot(aes(x = total_boot)) + geom_histogram() g_qq &lt;- totales_boot |&gt; ggplot(aes(sample = total_boot)) + geom_qq() + geom_qq_line(colour = &quot;red&quot;) + geom_hline(yintercept = quantile(totales_boot$total_boot, 0.975), colour = &quot;gray&quot;) + geom_hline(yintercept = quantile(totales_boot$total_boot, 0.025), colour = &quot;gray&quot;) g_hist + g_qq En este caso, distribución de muestreo presenta cierta asimetría, pero la desviación no es grande. En la parte central la aproximación normal es razonable. Procedemos a revisar sesgo total_est &lt;- calc_estimador_casas(muestra_casas) sesgo &lt;- mean(totales_boot$total_boot) - total_est sesgo ## [1] 110.0851 Este número puede parecer grande, pero sí calculamos la desviación relativa con respecto al error estándar vemos que es chico en la escala de la distribución bootstrap: ee_boot &lt;- sd(totales_boot$total_boot) sesgo_relativo &lt;- sesgo / ee_boot sesgo_relativo ## [1] 0.01522088 De forma que procedemos a construir intervalos de confianza como sigue : c(total_est - 2*ee_boot, total_est + 2*ee_boot) ## [1] 203366.6 232296.6 Que está en miles de dólares. En millones de dólares, este intervalo es: intervalo_total &lt;- c(total_est - 2*ee_boot, total_est + 2*ee_boot) / 1000 intervalo_total |&gt; round(1) ## [1] 203.4 232.3 Así que con 95% de confianza el verdadero total del valor de las casas vendidas está entre 203 y 232 millones de dólares. Nota: en este ejemplo mostraremos una alternativa de intervalos de confianza que es más apropiado cuando observamos asimetría. Sin embargo, primero tendremos que hablar de dos conceptos clave con respecto a intervalos de confianza: calibración e interpretación. Calibración de intervalos de confianza ¿Cómo sabemos que nuestros intervalos de confianza del 95% nominal tienen cobertura real de 95%? Es decir, tenemos que checar: El procedimiento para construir intervalos debe dar intervalos tales que el valor poblacional está en el intervalo de confianza para 95% de las muestras. Como solo tenemos una muestra, la calibración depende de argumentos teóricos o estudios de simulación previos. Para nuestro ejemplo de casas tenemos la población, así que podemos checar qué cobertura real tienen los intervalos normales: simular_intervalos &lt;- function(rep, size = 150){ muestra_casas &lt;- slice_sample(poblacion_casas, n = size) N &lt;- nrow(poblacion_casas) n &lt;- nrow(muestra_casas) total_est &lt;- (N / n) * sum(muestra_casas$precio_miles) # paso 1: define el estimador calc_estimador_casas &lt;- function(datos){ total_muestra &lt;- sum(datos$precio_miles) estimador_total &lt;- (N / n) * total_muestra estimador_total } # paso 2: define el proceso de remuestreo muestra_boot &lt;- function(datos){ #tomar muestra con reemplazo del mismo tamaño slice_sample(datos, prop = 1, replace = TRUE) } # paso 3: remuestrea y calcula el estimador totales_boot &lt;- map_dbl(1:2000, ~ calc_estimador_casas(muestra_boot(muestra_casas))) |&gt; tibble(total_boot = .) |&gt; summarise(ee_boot = sd(total_boot)) |&gt; mutate(inf = total_est - 2*ee_boot, sup = total_est + 2*ee_boot) |&gt; mutate(rep = rep) totales_boot } # Para recrear, correr: # sims_intervalos &lt;- map(1:100, ~ simular_intervalos(rep = .x)) # write_rds(sims_intervalos, &quot;cache/sims_intervalos.rds&quot;) # Para usar resultados en cache: sims_intervalos &lt;- read_rds(&quot;cache/sims_intervalos.rds&quot;) total &lt;- sum(poblacion_casas$precio_miles) sims_tbl &lt;- sims_intervalos |&gt; bind_rows() |&gt; mutate(cubre = inf &lt; total &amp; total &lt; sup) ggplot(sims_tbl, aes(x = rep)) + geom_hline(yintercept = total, colour = &quot;red&quot;) + geom_linerange(aes(ymin = inf, ymax = sup, colour = cubre)) La cobertura para estos 100 intervalos simulados da total &lt;- sum(poblacion_casas$precio_miles) sims_tbl |&gt; summarise(cobertura = mean(cubre)) ## # A tibble: 1 × 1 ## cobertura ## &lt;dbl&gt; ## 1 0.96 que es consistente con una cobertura real del 95% (¿qué significa “consistente”? ¿Cómo puedes checarlo con el bootstrap?) Observación. En este caso teníamos la población real, y pudimos verificar la cobertura de nuestros intervalos. En general no la tenemos. Estos ejercicios de simulación se pueden hacer con poblaciones sintéticas que se generen con las características que creemos va a tener nuestra población (por ejemplo, sesgo, colas largas, etc.). En general, no importa qué tipo de estimadores o intervalos de confianza usemos, requerimos checar la calibración. Esto puede hacerse con ejercicios de simulación con poblaciones sintéticas y tanto los procedimientos de muestreo como los tamaños de muestra que nos interesa usar. Verificar la cobertura de nuestros intervalos de confianza por medio simulación está bien estudiado para algunos casos. Por ejemplo, cuando trabajamos con estimaciones para poblaciones teóricas. En general sabemos que los procedimientos funcionan bien en casos: con distribuciones simétricas que tengan colas no muy largas; estimación de proporciones donde no tratamos con casos raros o casos seguros (probabilidades cercanas a 0 o 1). Interpretación de intervalos de confianza Como hemos visto, “intervalo de confianza” (de 90% de confianza, por ejemplo) es un término frecuentista, que significa: Cada muestra produce un intervalo distinto. Para el 90% de las muestras posibles, el intervalo cubre al valor poblacional. La afirmación es sobre el intervalo y el mecanismo para construirlo. Así que con alta probabilidad, el intervalo contiene el valor poblacional. Intervalos más anchos nos dan más incertidumbre acerca de dónde está el verdadero valor poblacional (y al revés para intervalos más angostos). Existen también “intervalos de credibilidad” (de 90% de probabilidad, por ejemplo), que se interpetan de forma bayesiana: Con 90% de probabilidad (relativamente alta), creemos que el valor poblacional está dentro del intervalo de credibilidad. Esta última interpretación es más natural. Obsérvese que para hablar de intervalos de confianza frecuentista tenemos que decir: Este intervalo particular cubre o no al verdadero valor, pero nuestro procedimiento produce intervalos que contiene el verdadero valor para el 90% de las muestras. Esta es una interpretación relativamente débil, y muchos intervalos poco útiles pueden satisfacerla. La interpretación bayesiana es más natural porque expresa más claramente incertidumbre acerca del valor poblacional. Por otro lado, La interpretación frecuentista nos da maneras empíricas de probar si los intervalos de confianza están bien calibrados o no: es un mínimo que “intervalos del 90%” deberían satisfacer. Así que tomamos el punto de vista bayesiano en la intepretación, pero buscamos que nuestros intervalos cumplan o aproximen bien garantías frecuentistas (discutimos esto más adelante). Los intervalos que producimos en esta sección pueden interpretarse de las dos maneras. Sesgo Notemos que hemos revisado el sesgo en varias ocasiones, esto es porque algunos estimadores comunes (por ejemplo, cociente de dos cantidades aleatorias) pueden sufrir de sesgo grande, especialmente en el caso de muestras chicas. Esto a su vez afecta la cobertura, pues es posible que nuestros intervalos no tengan “cobertura simétrica”, por ejemplo. Para muchos estimadores, y muestras no muy chicas, esté sesgo tiende a ser poco importante y no es necesario hacer correcciones. Si el tamaño del sesgo es grande comparado con la dispersión de la distribución bootstrap generalmente consideramos que bajo el diseño actual el estimador que estamos usando no es apropiado, y podemos proponer otro estimador u otro procedimiento para construir intervalos (ver Efron and Tibshirani (1993) intervalos BC_{a}), Efron and Tibshirani (1993) sugieren más de 20% de la desviación estándar, mientras que en Chihara and Hesterberg (2018) se sugiere 2% de la desviación estándar. Dependiendo que tan crítico es que los intervalos estén bien calibrados podemos evaluar nuestro problema particular. Intervalos bootstrap de percentiles Retomemos nuestro ejemplo del valor total del precio de las casas. A través de remuestras bootstrap hemos verificado gráficamente que la distribución de remuestreo es ligeramente asimétrica (ver la figura de abajo). Anteriormente hemos calculado intervalos de confianza basados en supuestos normales por medio del error éstandar. Este intervalo está dado por ## [1] 203.4 232.3 y por construcción sabemos que es simétrico con respecto al valor estimado, pero como podemos ver la distribución de muestreo no es simétrica, lo cual podemos confirmar por ejemplo calculando el porcentaje de muestras bootstrap que caen por arriba y por debajo del intervalo construido: ## # A tibble: 1 × 2 ## prop_inf prop_sup ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.0192 0.026 los cuales se han calculado como el porcentaje de medias bootstrap por debajo (arriba) de la cota inferior (superior), y vemos que no coinciden con el nivel de confianza prestablecido (2.5% para cada extremo). Otra opción común que se usa específicamente cuando la distribución bootstrap no es muy cercana a la normal son los intervalos de percentiles bootstrap: Definición. El intervalo de percentiles bootstrap al 95% de confianza está dado por \\[[q_{0.025}, q_{0.975}]\\] donde \\(q_f\\) es el percentil \\(f\\) de la distribución bootstrap. Otros intervalos comunes son el de 80% o 90% de confianza, por ejemplo, que corresponden a \\([q_{0.10}, q_{0.90}]\\) y \\([q_{0.05}, q_{0.95}]\\). Ojo: intervalos de confianza muy alta (por ejemplo 99.5%) pueden tener mala calibración o ser muy variables en su longitud pues dependen del comportamiento en las colas de la distribución. Para el ejemplo de las casas, calcularíamos simplemente intervalo_95 &lt;- totales_boot |&gt; pull(total_boot) |&gt; quantile(probs = c(0.025, 0.975)) / 1000 (intervalo_95) |&gt; round(1) ## 2.5% 97.5% ## 204.3 232.5 que está en millones de dólares. Nótese que es similar al intervalo de error estándar. Otro punto interesante sobre los intervalos bootstrap de percentiles es que lidian naturalmente con la asímetría de la distribución bootstrap. Ilustramos esto con la distancia de las extremos del intervalo con respecto a la media: abs(intervalo_95 - total_est/1000) ## 2.5% 97.5% ## 13.53912 14.64674 Los intervalos de confianza nos permiten presentar un rango de valores posibles para el parámetro de interés. Esto es una notable diferencia con respecto a presentar sólo un candidato como estimador. Nuestra fuente de información son los datos. Es por esto que si vemos valores muy chicos (grandes) en nuestra muestra, el intervalo se tiene que extender a la izquierda (derecha) para compensar dichas observaciones. Explica por qué cuando la aproximación normal es apropiada, el intervalo de percentiles al 95% es muy similar al intervalo normal de 2 errores estándar. Ejemplo Consideramos los datos de propinas. Queremos estimar la media de cuentas totales para la comida y la cena. Podemos hacer bootstrap de cada grupo por separado: # en este ejemplo usamos rsample, pero puedes # escribir tu propio código library(rsample) propinas &lt;- read_csv(&quot;data/propinas.csv&quot;) estimador &lt;- function(split, ...){ muestra &lt;- analysis(split) |&gt; group_by(momento) muestra |&gt; summarise(estimate = mean(cuenta_total), .groups = &#39;drop&#39;) |&gt; mutate(term = momento) } intervalo_propinas_90 &lt;- bootstraps(propinas, strata = momento, 1000) |&gt; mutate(res_boot = map(splits, estimador)) |&gt; int_pctl(res_boot, alpha = 0.10) |&gt; mutate(across(where(is.numeric), round, 2)) intervalo_propinas_90 ## # A tibble: 2 × 6 ## term .lower .estimate .upper .alpha .method ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; ## 1 Cena 19.8 20.8 22.0 0.1 percentile ## 2 Comida 15.6 17.1 18.8 0.1 percentile Nota: .estimate es la media de los valores de la estadística sobre las remuestras, no es el estimador original. De la tabla anterior inferimos que la media en la cuenta en la cena es más grande que la de la comida. Podemos graficar agregando los estimadores plugin: estimadores &lt;- propinas |&gt; group_by(momento) |&gt; rename(term = momento) |&gt; summarise(media = mean(cuenta_total)) ggplot(intervalo_propinas_90, aes(x = term)) + geom_linerange(aes(ymin = .lower, ymax = .upper)) + geom_point(data = estimadores, aes(y = media), colour = &quot;red&quot;, size = 3) + xlab(&quot;Momento&quot;) + ylab(&quot;Media de cuenta total (dólares)&quot;) + labs(subtitle = &quot;Intervalos de 90% para la media&quot;) Nótese que el bootstrap lo hicimos por separado en cada momento del día (por eso el argumento strata en la llamada a bootstraps): Justifica el procedimiento de hacer el bootstrap separado para cada grupo. ¿Qué supuestos acerca del muestreo se deben satisfacer? ¿Deben ser muestras aleatorias simples de cada momento del día, por ejemplo? ¿Qué harías si no fuera así, por ejemplo, si se escogieron al azar tickets de todos los disponibles en un periodo? Bootstrap para dos muestras En el ejemplo anterior consideramos cómo hacer bootstrap cuando tenemos muestras independientes. También podemos aplicarlo a estimadores que comparen directamente las dos muestras: Bootstrap para comparar poblaciones. Dadas muestras independientes de tamaños \\(m\\) y \\(n\\) de dos poblaciones: Extraer una remuestra de tamaño \\(m\\) con reemplazo de la primera muestra y una remuestra separada de tamaño \\(n\\) de la segunda muestra. Calcula la estadística que compara los dos grupos (por ejemplo, diferencia de medias) Repetir este proceso muchas veces (por ejemplo, 1000 - 10000). Construir la distribución bootstrap de la estadística. Examinar dispersión, sesgo y forma. Ejemplo Supongamos que queremos comparar directamente la media de la cuenta total en comida y cena. Podemos hacer: estimador_dif &lt;- function(split, ...){ muestra &lt;- analysis(split) |&gt; group_by(momento) muestra |&gt; summarise(estimate = mean(cuenta_total), .groups = &quot;drop&quot;) |&gt; pivot_wider(names_from = momento, values_from = estimate) |&gt; mutate(estimate = Cena - Comida, term = &quot;diferencia&quot;) } dist_boot &lt;- bootstraps(propinas, strata = momento, 2000) |&gt; mutate(res_boot = map(splits, estimador_dif)) g_1 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(x = estimate)) + geom_histogram(bins = 20) + xlab(&quot;Diferencia Comida vs Cena&quot;) g_2 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(sample = estimate)) + geom_qq() + geom_qq_line(colour = &#39;red&#39;) g_1 + g_2 Y podemos calcular un intervalo de confianza para la diferencia de medias: dist_boot |&gt; int_pctl(res_boot, alpha = 0.01) |&gt; mutate(across(where(is.numeric), round, 2)) |&gt; select(term, .lower, .upper) ## # A tibble: 1 × 3 ## term .lower .upper ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 diferencia 0.73 6.54 Que nos indica que con alta probabilidad las cuentas son más altas que en la cena que en la comida. La diferencia puede ir de un poco menos de un dólar hasta seis dólares con 99% de confianza. Datos pareados En otros casos, las muestras no son independientes y están pareadas. Por ejemplo, este es un estudio dende a 10 personas una noche se les dio una medicina para dormir y otra noche otra medicina. Se registraron cuántas horas de sueño extra comparados con un día donde no tomaron medicina. dormir &lt;- sleep |&gt; pivot_wider(names_from = group, names_prefix = &quot;medicina_&quot;, values_from = extra) dormir ## # A tibble: 10 × 3 ## ID medicina_1 medicina_2 ## &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 0.7 1.9 ## 2 2 -1.6 0.8 ## 3 3 -0.2 1.1 ## 4 4 -1.2 0.1 ## 5 5 -0.1 -0.1 ## 6 6 3.4 4.4 ## 7 7 3.7 5.5 ## 8 8 0.8 1.6 ## 9 9 0 4.6 ## 10 10 2 3.4 En este caso, el bootstrap se hace sobre individuos, y quisiéramos comparar la medición de la medicina_1 con la medicina_2. Usaremos la media de al diferencia entre horas de sueño entre las dos medicinas. Nuestro estimador puntual es: estimador_dif &lt;- dormir |&gt; mutate(dif_2_menos_1 = medicina_2 - medicina_1) |&gt; summarise(dif_media = mean(dif_2_menos_1)) estimador_dif ## # A tibble: 1 × 1 ## dif_media ## &lt;dbl&gt; ## 1 1.58 Esto indica que en promedio duermen hora y media más con la medicina 2 que con la medicina 1. Como hay variabilildad considerable en el número de horas extra de cada medicina dependiendo del individuo, es necesario hacer una intervalo de confianza para descartar que esta diferencia haya aparecido por azar debido a la variación muestral. Nótese que aquí no tenemos estratos, pues solo hay una muestra de individuo con dos mediciones. estimador_dif &lt;- function(split, ...){ muestra &lt;- analysis(split) muestra |&gt; mutate(dif_2_menos_1 = medicina_2 - medicina_1) |&gt; summarise(estimate = mean(dif_2_menos_1), .groups = &quot;drop&quot;) |&gt; mutate(term = &quot;diferencia 2 vs 1&quot;) } dist_boot &lt;- bootstraps(dormir, 2000) |&gt; mutate(res_boot = map(splits, estimador_dif)) g_1 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(x = estimate)) + geom_histogram(bins = 20) g_2 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(sample = estimate)) + geom_qq() + geom_qq_line(colour = &#39;red&#39;) g_1 + g_2 Nuestro intervalo de percentiles al 90% es de dist_boot |&gt; int_pctl(res_boot, 0.10) ## # A tibble: 1 × 6 ## term .lower .estimate .upper .alpha .method ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; ## 1 diferencia 2 vs 1 1.04 1.57 2.22 0.1 percentile Lo que indica con alta probabilidad que la medicina 2 da entre 1 y 2 horas extras de sueño. Nota que en este ejemplo también podríamos hacer una prueba de hipótesis por permutaciones, suponiendo como hipótesis nula que las dos medicinas son equivalentes. Sin embargo, usualmente es más informativo presentar este tipo de intervalos para estimar la diferencia. Bootstrap y otras estadísticas El bootstrap es una técnica versátil. Un ejemplo son estimadores de razón, que tienen la forma \\[ \\hat{r} = \\frac{\\overline y}{\\overline x}\\] Por ejemplo, ¿cómo haríamos estimación para el procentaje de área area habitable de las casas en relación al tamaño del lote? Una manera de estimar esta cantidad es dividiendo la suma del área habitable de nuestra muestra y dividirlo entre la suma del área de los lotes de nuestra muestra, como en la fórmula anterior. Esta fórmula es más difícil pues tanto numerador como denominador tienen variabilidad, y estas dos cantidades no varían independientemente. Con el bootstrap podemos atacar estos problemas Ejemplo: estimadores de razón Nuestra muestra original es: set.seed(250) casas_muestra &lt;- slice_sample(poblacion_casas, n = 200) El estimador de interés es: estimador_razon &lt;- function(split, ...){ muestra &lt;- analysis(split) muestra |&gt; summarise(estimate = sum(area_habitable_sup_m2) / sum(area_lote_m2), .groups = &quot;drop&quot;) |&gt; mutate(term = &quot;% area del lote construida&quot;) } Y nuestra estimación puntual es estimacion &lt;- muestra_casas |&gt; summarise(estimate = sum(area_habitable_sup_m2) / sum(area_lote_m2)) estimacion ## # A tibble: 1 × 1 ## estimate ## &lt;dbl&gt; ## 1 0.148 Es decir que en promedio, un poco más de 15% del lote total es ocupado por área habitable. Ahora hacemos bootstrap para construir un intervalo: dist_boot &lt;- bootstraps(casas_muestra, 2000) |&gt; mutate(res_boot = map(splits, estimador_razon)) g_1 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(x = estimate)) + geom_histogram(bins = 20) g_2 &lt;- ggplot(dist_boot |&gt; unnest(res_boot), aes(sample = estimate)) + geom_qq() + geom_qq_line(colour = &#39;red&#39;) g_1 + g_2 En este caso la cola derecha parece tener menos dispersión que una distribución normal. Usamos un intervalo de percentiles para obtener: dist_boot |&gt; int_pctl(res_boot) |&gt; mutate(estimador = estimacion$estimate) |&gt; rename(media_boot = .estimate) |&gt; mutate(bias = media_boot - estimador) |&gt; pivot_longer(is_double) |&gt; mutate(value = round(value, 3)) ## # A tibble: 6 × 4 ## term .method name value ## &lt;chr&gt; &lt;chr&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 % area del lote construida percentile .lower 0.121 ## 2 % area del lote construida percentile media_boot 0.142 ## 3 % area del lote construida percentile .upper 0.159 ## 4 % area del lote construida percentile .alpha 0.05 ## 5 % area del lote construida percentile estimador 0.148 ## 6 % area del lote construida percentile bias -0.006 De modo que en esta zona, entre 12% y 16% de toda el área disponible es ocupada por área habitable: estas son casas que tienen jardines o terrenos, garage relativamente grandes. Ejemplo: suavizadores Podemos usar el bootstrap para juzgar la variabilidad de un suavizador, que consideramos como nuestra estadística: graf_casas &lt;- function(data){ ggplot(data |&gt; filter(calidad_gral &lt; 7), aes(x = area_habitable_sup_m2)) + geom_point(aes(y = precio_m2_miles), alpha = 0.75) + geom_smooth(aes(y = precio_m2_miles), method = &quot;loess&quot;, span = 0.7, se = FALSE, method.args = list(degree = 1, family = &quot;symmetric&quot;)) } graf_casas(casas_muestra) Podemos hacer bootstrap para juzgar la estabilidad del suavizador: suaviza_boot &lt;- function(x, data){ # remuestreo muestra_boot &lt;- slice_sample(data, prop = 1, replace = T) ajuste &lt;- loess(precio_m2_miles ~ area_habitable_sup_m2, data = muestra_boot, degree = 1, span = 0.7, family = &quot;symmetric&quot;) datos_grafica &lt;- tibble(area_habitable_sup_m2 = seq(25, 250, 5)) ajustados &lt;- predict(ajuste, newdata = datos_grafica) datos_grafica |&gt; mutate(ajustados = ajustados) |&gt; mutate(rep = x) } reps &lt;- map(1:10, ~ suaviza_boot(.x, casas_muestra |&gt; filter(calidad_gral &lt; 7))) |&gt; bind_rows() # ojo: la rutina loess no tienen soporte para extrapolación graf_casas(casas_muestra) + geom_line(data = reps, aes(y = ajustados, group = rep), alpha = 1, colour = &quot;red&quot;) Donde vemos que algunas cambios de pendiente del suavizador original no son muy interpretables (por ejemplo, para áreas chicas) y alta variabilidad en general en los extremos. Podemos hacer más iteraciones para calcular bandas de confianza: reps &lt;- map(1:200, ~ suaviza_boot(.x, casas_muestra |&gt; filter(calidad_gral &lt; 7))) |&gt; bind_rows() # ojo: la rutina loess no tienen soporte para extrapolación graf_casas(casas_muestra) + geom_line(data = reps, aes(y = ajustados, group = rep), alpha = 0.2, colour = &quot;red&quot;) Donde observamos cómo tenemos incertidumbre en cuanto al nivel y forma de las curvas en los extremos de los datos (casas grandes y chicas), lo cual es natural. Aunque podemos resumir para hacer bandas de confianza, mostrar remuestras de esta manera es informativo: por ejempo: vemos cómo es probable también que para casas de menos de 70 metros cuadrados el precio por metro cuadrado no cambia tanto (líneas constantes) Bootstrap y estimadores complejos: tablas de perfiles Podemos regresar al ejemplo de la primera sesión donde calculamos perfiles de los tomadores de distintos tés: en bolsa, suelto, o combinados. Caundo hacemos estos tipos de análisis no es raro que los prefiles tengan variabilidad considerable que es necesario cuantificar. price tea bag tea bag+unpackaged unpackaged promedio p_upscale -0.71 -0.28 0.98 28 p_variable -0.12 0.44 -0.31 36 p_cheap 0.3 -0.53 0.23 2 p_branded 0.62 -0.16 -0.45 25 p_private label 0.72 -0.22 -0.49 5 p_unknown 1.58 -0.58 -1 3 Hacemos bootstrap sobre toda la muestra, y repetimos exactamente el mismo proceso de construción de perfiles: boot_perfiles &lt;- map(1:1000, function(x){ te_boot &lt;- te |&gt; slice_sample(prop = 1, replace = TRUE) calcular_perfiles(te_boot) |&gt; mutate(rep = x) }) |&gt; bind_rows() Ahora resumimos y graficamos, esta vez de manera distinta: resumen_perfiles &lt;- boot_perfiles |&gt; group_by(how, price) |&gt; summarise(perfil_media = mean(perfil), ymax = quantile(perfil, 0.9), ymin = quantile(perfil, 0.10)) resumen_bolsa &lt;- resumen_perfiles |&gt; ungroup() |&gt; filter(how == &quot;tea bag&quot;) |&gt; select(price, perfil_bolsa = perfil_media) resumen_perfiles &lt;- resumen_perfiles |&gt; left_join(resumen_bolsa) |&gt; ungroup() |&gt; mutate(price = fct_reorder(price, perfil_bolsa)) ggplot(resumen_perfiles, aes(x = price, y = perfil_media, ymax = ymax, ymin = ymin)) + geom_point(colour = &quot;red&quot;) + geom_linerange() + facet_wrap(~how) + coord_flip() + geom_hline(yintercept = 0, colour = &quot;gray&quot;) + ylab(&quot;Perfil&quot;) + xlab(&quot;Precio&quot;) Nótese una deficiencia clara del bootstrap: para los que compran té suelto, en la muestra no existen personas que desconocen de dónde provienen su té (No sabe/No contestó). Esto produce un intervalo colapsado en 0 que no es razonable. Podemos remediar esto de varias maneras: quitando del análisis los que no sabe o no contestaron, agrupando en otra categoría, usando un modelo, o regularizar usando proporciones calculadas con conteos modificados: por ejemplo, agregando un caso de cada combinación (agregaría 18 personas “falsas” a una muestra de 290 personas). Bootstrap y muestras complejas La necesidad de estimaciones confiables junto con el uso eficiente de recursos conllevan a diseños de muestras complejas. Estos diseños típicamente usan las siguientes técnicas: muestreo sin reemplazo de una población finita, muestreo sistemático, estratificación, conglomerados, ajustes a no-respuesta, postestratificación. Como consecuencia, los valores de la muestra suelen no ser independientes y los análisis de los mismos dependerá del diseño de la muestra. Comenzaremos con definiciones para entender el problema. set.seed(3872999) n_escuelas &lt;- 5000 tipo &lt;- sample(c(&quot;rural&quot;, &quot;urbano&quot;, &quot;indigena&quot;), n_escuelas, replace = TRUE, prob = c(0.3, 0.5, 0.2)) escuela &lt;- tibble(ind_escuela = 1:n_escuelas, tipo, media_tipo = case_when(tipo == &quot;urbano&quot; ~ 550, tipo == &quot;rural&quot; ~ 400, TRUE ~ 350), media_escuela = rnorm(n_escuelas, media_tipo, 30), n_estudiantes = round(rnorm(n_escuelas, 30, 4))) estudiantes &lt;- uncount(escuela, n_estudiantes, .id = &quot;id_estudiante&quot;) %&gt;% rowwise() %&gt;% mutate(calif = rnorm(1, media_escuela, 70)) %&gt;% ungroup() Imaginemos que tenemos una población de 5000 escuelas, y queremos estimar la media de las calificaciones de los estudiantes en una prueba. head(estudiantes) ## # A tibble: 6 × 6 ## ind_escuela tipo media_tipo media_escuela id_estudiante calif ## &lt;int&gt; &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; &lt;dbl&gt; ## 1 1 urbano 550 561. 1 488. ## 2 1 urbano 550 561. 2 574. ## 3 1 urbano 550 561. 3 456. ## 4 1 urbano 550 561. 4 507. ## 5 1 urbano 550 561. 5 598. ## 6 1 urbano 550 561. 6 527. La primera idea sería tomar una muestra aleatoria (MAS, muestreo aleatorio simple), donde todos los estudiantes tienen igual probabilidad de ser seleccionados. Con estas condiciones el presupuesto alcanza para seleccionar a 60 estudiantes, hacemos esto y calculamos la media. muestra &lt;- slice_sample(estudiantes, n = 60) round(mean(muestra$calif), 2) ## [1] 466.73 Este número es muy cercano a la media verdadera de la población: 466.51, pero esta es una de muchas posibles muestras. medias_mas &lt;- rerun(1000, mean(sample(estudiantes$calif, 60))) %&gt;% flatten_dbl() sd(medias_mas) ## [1] 14.75242 hist_mas &lt;- ggplot(tibble(medias_mas), aes(x = medias_mas)) + geom_histogram(binwidth = 10) + geom_vline(xintercept = mean(estudiantes$calif), color = &quot;red&quot;) + xlim(410, 520) qq_mas &lt;- ggplot(tibble(medias_mas), aes(sample = medias_mas)) + geom_qq(distribution = stats::qunif) + ylim(410, 520) hist_mas + qq_mas Algunas de las muestras generan valores alejados de la verdadera media, para minimizar la probabilidad de seleccionar muestras que lleven a estimaciones alejadas del verdadero valor poblacional podríamos tomar muestras más grandes. Pero usualmente los costos limitan el tamaño de muestra. Una alternativa es estratificar, supongamos que sabemos el tipo de cada escuela (urbana, rural o indígena) y sabemos también que la calificación de los estudiantes de escuelas urbanas tiende a ser distinta a las calificaciones que los estudiantes de escuelas rurales o indígenas. En esta caso un diseño más eficiente consiste en tomar muestras independientes dentro de cada estrato. muestra_estrat &lt;- estudiantes %&gt;% group_by(tipo) %&gt;% sample_frac(0.0004) dim(muestra_estrat) ## [1] 60 6 muestrea_estrat &lt;- function(){ muestra &lt;- estudiantes %&gt;% group_by(tipo) %&gt;% sample_frac(0.0004) mean(muestra$calif) } medias_estrat &lt;- rerun(1000, muestrea_estrat()) %&gt;% flatten_dbl() Notamos que la distribución muestral está más concentrada que el caso de MAS, el error estándar se reduce de 14.75 a 10.2 hist_estrat &lt;- ggplot(tibble(medias_estrat), aes(x = medias_estrat)) + geom_histogram(binwidth = 6) + geom_vline(xintercept = mean(estudiantes$calif), color = &quot;red&quot;) + xlim(410, 520) qq_estrat &lt;- ggplot(tibble(medias_estrat), aes(sample = medias_estrat)) + geom_qq(distribution = stats::qunif) + ylim(410, 520) hist_estrat + qq_estrat Entonces, la estratificación nos sirve para reducir el error estándar de las estimaciones. Otro procedimiento común en muestreo es introducir conglomerados, a diferencia del muestreo estratificado, el propósito principal de los conglomerados es reducir costos. Veamos cuantas escuelas tendría que visitar en una muestra dada (con diseño estratificado). n_distinct(muestra_estrat$ind_escuela) ## [1] 60 Es fácil ver que visitar una escuela para aplicar solo uno o dos exámenes no es muy eficiente en cuestión de costos. Es por ello que se suelen tomar muestras considerando conglomerados naturales, en este caso la escuela. En nuestro ejemplo es razonable suponer que una parte grande del costo del muestreo sea mandar al examinador a la escuela, y que una vez en la escuela el costo de evaluar a todo sexto, en lugar de a un único alumno, es relativamente bajo. Podemos imaginar que considerando estos costos por visita de escuela nos alcance para visitar 40 escuelas y en cada una examinar a todos los estudiantes. muestra_escuelas &lt;- escuela %&gt;% group_by(tipo) %&gt;% sample_frac(size = 0.008) muestra_cgl &lt;- muestra_escuelas %&gt;% left_join(estudiantes) mean(muestra_cgl$calif) ## [1] 462.5677 muestrea_cgl &lt;- function(){ muestra_escuelas &lt;- escuela %&gt;% group_by(tipo) %&gt;% sample_frac(size = 0.008) muestra_cgl &lt;- muestra_escuelas %&gt;% left_join(estudiantes, by = c(&quot;ind_escuela&quot;, &quot;tipo&quot;)) mean(muestra_cgl$calif) } medias_cgl &lt;- rerun(1000, muestrea_cgl()) %&gt;% flatten_dbl() En este caso, el número de estudiantes examinados es mucho mayor que en MAS y muestreo estratificado, notemos que el número de estudiantes evaluados cambiará de muestra a muestra dependiendo del número de alumnos en las escuelas seleccionadas. sd(medias_cgl) ## [1] 5.337327 hist_cgl &lt;- ggplot(tibble(medias_cgl), aes(x = medias_cgl)) + geom_histogram(binwidth = 6) + geom_vline(xintercept = mean(estudiantes$calif), color = &quot;red&quot;) + xlim(410, 520) qq_cgl &lt;- ggplot(tibble(medias_cgl), aes(sample = medias_cgl)) + geom_qq(distribution = stats::qunif) + ylim(410, 520) hist_cgl + qq_cgl Ejemplo: ENIGH La complejidad de los diseños de encuestas conlleva a que el cálculo de errores estándar sea muy complicado, para atacar este problema hay dos técnicas básicas: 1) un enfoque analítico usando linearización, 2) métodos de remuestreo como bootstrap. El incremento en el poder de cómputo ha favorecido los métodos de remuestreo pues la linearización requiere del desarrollo de una fórmula para cada estimación y supuestos adicionales para simplificar. En 1988 Rao and Wu (1988) propusieron un método de bootstrap para diseños estratificados multietápicos con reemplazo de UPMs (Unidades Primarias de Muestreo) que describimos a continuación. ENIGH. Usaremos como ejemplo la Encuesta Nacional de Ingresos y Gastos de los Hogares, ENIGH 2018 (INEGI 2018), esta encuesta usa un diseño de conglomerados estratificado. Antes de proceder a bootstrap debemos entender como se seleccionaron los datos, esto es, el diseño de la muestra: Unidad primaria de muestreo (UPM). Las UPMs están constituidas por agrupaciones de viviendas. Se les denomina unidades primarias pues corresponden a la primera etapa de selección, las unidades secundarias (USMs) serían los hogares. Estratificación. Los estratos se construyen en base a estado, ámbito (urbano, complemento urbano, rural), características sociodemográficas de los habitantes de las viviendas, características físicas y equipamiento. El proceso de estratificación resulta en 888 subestratos en todo el ámbito nacional. La selección de la muestra es independiente para cada estrato, y una vez que se obtiene la muestra se calculan los factores de expansión que reflejan las distintas probabilidades de selección. Después se llevan a cabo ajustes por no respuesta y por proyección (calibración), esta última busca que distintos dominios de la muestra coincidan con la proyección de población de INEGI. concentrado_hogar &lt;- read_csv(here::here(&quot;data&quot;, &quot;conjunto_de_datos_enigh_2018_ns_csv&quot;, &quot;conjunto_de_datos_concentradohogar_enigh_2018_ns&quot;, &quot;conjunto_de_datos&quot;, &quot;conjunto_de_datos_concentradohogar_enigh_2018_ns.csv&quot;)) # seleccionar variable de ingreso corriente hogar &lt;- concentrado_hogar %&gt;% mutate( upm = as.integer(upm), edo = str_sub(ubica_geo, 1, 2) ) %&gt;% select(folioviv, foliohog, est_dis, upm, factor, ing_cor, edad_jefe, edo) %&gt;% group_by(est_dis) %&gt;% mutate(n = n_distinct(upm)) %&gt;% # número de upms por estrato ungroup() hogar ## # A tibble: 74,647 × 9 ## folioviv foliohog est_dis upm factor ing_cor edad_jefe edo n ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; &lt;int&gt; ## 1 100013601 1 2 1 175 76404. 74 10 106 ## 2 100013602 1 2 1 175 42988. 48 10 106 ## 3 100013603 1 2 1 175 580698. 39 10 106 ## 4 100013604 1 2 1 175 46253. 70 10 106 ## 5 100013606 1 2 1 175 53837. 51 10 106 ## 6 100026701 1 2 2 189 237743. 41 10 106 ## 7 100026703 1 2 2 189 32607. 57 10 106 ## 8 100026704 1 2 2 189 169918. 53 10 106 ## 9 100026706 1 2 2 189 17311. 30 10 106 ## 10 100027201 1 2 3 186 120488. 69 10 106 ## # ℹ 74,637 more rows Para el cálculo de estadísticos debemos usar los factores de expansión, por ejemplo el ingreso trimestral total sería: sum(hogar$factor * hogar$ing_cor / 1000) ## [1] 1723700566 y ingreso trimestral medio (miles pesos) sum(hogar$factor * hogar$ing_cor / 1000) / sum(hogar$factor) ## [1] 49.61029 La estimación del error estándar, por otro lado, no es sencilla y requiere usar aproximaciones, en la metodología de INEGI proponen una aproximación con series de Taylor. Figure 5.3: Extracto de estimación de errores de muestreo, ENIGH 2018. Veamos ahora como calcular el error estándar siguiendo el bootstrap de Rao y Wu: En cada estrato se seleccionan con reemplazo \\(m_h\\) UPMs de las \\(n_h\\) de la muestra original. Denotamos por \\(m_{hi}^*\\) el número de veces que se seleccionó la UPM \\(i\\) en el estrato \\(h\\) (de tal manera que \\(\\sum m_{hi}^*=m_h\\)). Creamos una replicación del ponderador correspondiente a la \\(k\\)-ésima unidad (USM) como: \\[d_k^*=d_k \\bigg[\\bigg(1-\\sqrt{\\frac{m_h}{n_h - 1}}\\bigg) + \\bigg(\\sqrt{\\frac{m_h}{n_h - 1}}\\frac{n_h}{m_h}m_{h}^*\\bigg)\\bigg]\\] donde \\(d_k\\) es el inverso de la probabilidad de selección. Si \\(m_h &lt;(n_h -1)\\) todos los pesos definidos de esta manera serán no negativos. Calculamos el peso final \\(w_k^*\\) aplicando a \\(d_k^*\\) los mismos ajustes que se hicieron a los ponderadores originales. Calculamos el estadístico de interés \\(\\hat{\\theta}\\) usando los ponderadores \\(w_k^*\\) en lugar de los originales \\(w_k\\). Repetimos los pasos 1 y 2 \\(B\\) veces para obtener \\(\\hat{\\theta}^{*1},\\hat{\\theta}^{*2},...,\\hat{\\theta}^{*B}\\). Calculamos el error estándar como: \\[\\hat{\\textsf{se}}_B = \\bigg\\{\\frac{\\sum_{b=1}^B[\\hat{\\theta}^*(b)-\\hat{\\theta}^*(\\cdot)]^2 }{B}\\bigg\\}^{1/2}\\] En principio podemos elegir cualquier valor de \\(m_h \\geq 1\\), el más común es elegir \\(m_h=n_h-1\\), en este caso: \\[d_k^*=d_k \\frac{n_h}{n_h-1}m_{hi}^*\\] en este escenario las unidades que no se incluyen en la muestra tienen un valor de cero como ponderador. Si elegimos \\(n_h \\ne n_h-1\\) las unidades que no están en la muestra tienen ponderador distinto a cero, si \\(m_h=n_h\\) el ponderador podría tomar valores negativos. Implementemos el bootstrap de Rao y Wu a la ENIGH, usaremos \\(m_h=n_h-1\\) # creamos una tabla con los estratos y upms est_upm &lt;- hogar %&gt;% distinct(est_dis, upm, n) %&gt;% arrange(est_dis, upm) hogar_factor &lt;- est_upm %&gt;% group_by(est_dis) %&gt;% # dentro de cada estrato tomamos muestra (n_h-1) sample_n(size = first(n) - 1, replace = TRUE) %&gt;% add_count(est_dis, upm, name = &quot;m_hi&quot;) %&gt;% # calculamos m_hi* left_join(hogar, by = c(&quot;est_dis&quot;, &quot;upm&quot;, &quot;n&quot;)) |&gt; mutate(factor_b = factor * m_hi * n / (n - 1)) # unimos los pasos anteriores en una función para replicar en cada muestra bootstrap svy_boot &lt;- function(est_upm, hogar){ m_hi &lt;- est_upm %&gt;% group_split(est_dis) %&gt;% map(~sample(.$upm, size = first(.$n) - 1, replace = TRUE)) %&gt;% flatten_int() %&gt;% plyr::count() %&gt;% select(upm = x, m_h = freq) m_hi %&gt;% left_join(hogar, by = c(&quot;upm&quot;)) %&gt;% mutate(factor_b = factor * m_h * n / (n - 1)) } set.seed(1038984) boot_rep &lt;- rerun(500, svy_boot(est_upm, hogar)) # Aplicación a ingreso medio wtd_mean &lt;- function(w, x, na.rm = FALSE) { sum(w * x, na.rm = na.rm) / sum(w, na.rm = na.rm) } # La media es: hogar %&gt;% summarise(media = wtd_mean(factor, ing_cor)) ## # A tibble: 1 × 1 ## media ## &lt;dbl&gt; ## 1 49610. Y el error estándar: map_dbl(boot_rep, ~wtd_mean(w = .$factor_b, x = .$ing_cor)) %&gt;% quantile(c(0.025, 0.975)) ## 2.5% 97.5% ## 48742.12 50519.02 El método bootstrap está implementado en el paquete survey y más recientemente en srvyr que es una versión tidy que utiliza las funciones en survey. Podemos comparar nuestros resultados con la implementación en survey. # 1. Definimos el diseño de la encuesta library(survey) library(srvyr) enigh_design &lt;- hogar %&gt;% as_survey_design(ids = upm, weights = factor, strata = est_dis) # 2. Elegimos bootstrap como el método para el cálculo de errores estándar set.seed(7398731) enigh_boot &lt;- enigh_design %&gt;% as_survey_rep(type = &quot;bootstrap&quot;, replicates = 500) # 3. Así calculamos la media enigh_boot %&gt;% srvyr::summarise(mean_ingcor = survey_mean(ing_cor)) ## # A tibble: 1 × 2 ## mean_ingcor mean_ingcor_se ## &lt;dbl&gt; &lt;dbl&gt; ## 1 49610. 468. enigh_boot %&gt;% srvyr::summarise(mean_ingcor = survey_mean(ing_cor, vartype = &quot;ci&quot;)) ## # A tibble: 1 × 3 ## mean_ingcor mean_ingcor_low mean_ingcor_upp ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 49610. 48690. 50530. # por estado enigh_boot %&gt;% group_by(edo) %&gt;% srvyr::summarise(mean_ingcor = survey_mean(ing_cor)) ## # A tibble: 30 × 3 ## edo mean_ingcor mean_ingcor_se ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 10 50161. 995. ## 2 11 46142. 1241. ## 3 12 29334. 1063. ## 4 13 38783. 1019. ## 5 14 60541. 1924. ## 6 15 48013. 1359. ## 7 16 42653. 1393. ## 8 17 42973. 1601. ## 9 18 48148. 1967. ## 10 19 68959. 4062. ## # ℹ 20 more rows Resumiendo: El bootstrap de Rao y Wu genera un estimador consistente y aproximadamente insesgado de la varianza de estadísticos no lineales y para la varianza de un cuantil. Este método supone que la seleccion de UPMs es con reemplazo; hay variaciones del estimador bootstrap de Rao y Wu que extienden el método que acabamos de estudiar; sin embargo, es común ignorar este aspecto, por ejemplo Mach et al estudian las propiedades del estimador de varianza bootstrap de Rao y Wu cuando la muestra se seleccionó sin reemplazo. Bootstrap en R Es común crear nuestras propias funciones cuando usamos bootstrap, sin embargo, en R también hay alternativas que pueden resultar convenientes, mencionamos 3: El paquete rsample (forma parte de la colección tidymodels) y tiene una función bootstraps() que regresa un arreglo cuadrangular (tibble, data.frame) que incluye una columna con las muestras bootstrap y un identificador del número y tipo de muestra. Veamos un ejemplo donde seleccionamos muestras del conjunto de datos muestra_computos que contiene 10,000 observaciones. library(rsample) load(&quot;data/election_2012.rda&quot;) muestra_computos &lt;- slice_sample(election_2012, n = 10000) muestra_computos ## # A tibble: 10,000 × 23 ## state_code state_name state_abbr district_loc_17 district_fed_17 polling_id ## &lt;chr&gt; &lt;chr&gt; &lt;chr&gt; &lt;int&gt; &lt;int&gt; &lt;int&gt; ## 1 18 Nayarit NAY 11 2 86709 ## 2 27 Tabasco TAB 17 5 122035 ## 3 15 México MEX 7 35 75477 ## 4 27 Tabasco TAB 19 5 122262 ## 5 17 Morelos MOR 6 2 84733 ## 6 07 Chiapas CHPS 22 5 15376 ## 7 14 Jalisco JAL 2 2 52634 ## 8 08 Chihuahua CHIH 7 4 19097 ## 9 14 Jalisco JAL 20 20 60549 ## 10 13 Hidalgo HGO 11 4 50221 ## # ℹ 9,990 more rows ## # ℹ 17 more variables: section &lt;int&gt;, region &lt;chr&gt;, polling_type &lt;chr&gt;, ## # section_type &lt;chr&gt;, pri_pvem &lt;int&gt;, pan &lt;int&gt;, panal &lt;int&gt;, ## # prd_pt_mc &lt;int&gt;, otros &lt;int&gt;, total &lt;int&gt;, nominal_list &lt;int&gt;, ## # pri_pvem_pct &lt;dbl&gt;, pan_pct &lt;dbl&gt;, panal_pct &lt;dbl&gt;, prd_pt_mc_pct &lt;dbl&gt;, ## # otros_pct &lt;dbl&gt;, winner &lt;chr&gt; Generamos 100 muestras bootstrap, y la función nos regresa un arreglo con 100 renglones, cada uno corresponde a una muestra bootstrap. set.seed(839287482) computos_boot &lt;- bootstraps(muestra_computos, times = 100) computos_boot ## # Bootstrap sampling ## # A tibble: 100 × 2 ## splits id ## &lt;list&gt; &lt;chr&gt; ## 1 &lt;split [10000/3647]&gt; Bootstrap001 ## 2 &lt;split [10000/3623]&gt; Bootstrap002 ## 3 &lt;split [10000/3724]&gt; Bootstrap003 ## 4 &lt;split [10000/3682]&gt; Bootstrap004 ## 5 &lt;split [10000/3696]&gt; Bootstrap005 ## 6 &lt;split [10000/3716]&gt; Bootstrap006 ## 7 &lt;split [10000/3679]&gt; Bootstrap007 ## 8 &lt;split [10000/3734]&gt; Bootstrap008 ## 9 &lt;split [10000/3632]&gt; Bootstrap009 ## 10 &lt;split [10000/3692]&gt; Bootstrap010 ## # ℹ 90 more rows La columna splits tiene información de las muestras seleccionadas, para la primera vemos que de 10,000 observaciones en la muestra original la primera muestra bootstrap contiene 10000-3647=6353. first_computos_boot &lt;- computos_boot$splits[[1]] first_computos_boot ## &lt;Analysis/Assess/Total&gt; ## &lt;10000/3647/10000&gt; Y podemos obtener los datos de la muestra bootstrap con la función as.data.frame() as.data.frame(first_computos_boot) ## # A tibble: 10,000 × 23 ## state_code state_name state_abbr district_loc_17 district_fed_17 polling_id ## &lt;chr&gt; &lt;chr&gt; &lt;chr&gt; &lt;int&gt; &lt;int&gt; &lt;int&gt; ## 1 07 Chiapas CHPS 13 9 13397 ## 2 14 Jalisco JAL 15 15 58404 ## 3 09 Ciudad de M… CDMX 11 13 26471 ## 4 17 Morelos MOR 7 3 84909 ## 5 25 Sinaloa SIN 11 3 114038 ## 6 11 Guanajuato GTO 16 12 41506 ## 7 29 Tlaxcala TLAX 6 3 128006 ## 8 02 Baja Califo… BC 8 5 3901 ## 9 02 Baja Califo… BC 9 5 3779 ## 10 08 Chihuahua CHIH 19 5 19536 ## # ℹ 9,990 more rows ## # ℹ 17 more variables: section &lt;int&gt;, region &lt;chr&gt;, polling_type &lt;chr&gt;, ## # section_type &lt;chr&gt;, pri_pvem &lt;int&gt;, pan &lt;int&gt;, panal &lt;int&gt;, ## # prd_pt_mc &lt;int&gt;, otros &lt;int&gt;, total &lt;int&gt;, nominal_list &lt;int&gt;, ## # pri_pvem_pct &lt;dbl&gt;, pan_pct &lt;dbl&gt;, panal_pct &lt;dbl&gt;, prd_pt_mc_pct &lt;dbl&gt;, ## # otros_pct &lt;dbl&gt;, winner &lt;chr&gt; Una de las principales ventajas de usar este paquete es que es eficiente en el uso de memoria. library(pryr) object_size(muestra_computos) ## 1.41 MB object_size(computos_boot) ## 5.49 MB # tamaño por muestra object_size(computos_boot)/nrow(computos_boot) ## 54.92 kB # el incremento en tamaño es &lt;&lt; 1000 as.numeric(object_size(computos_boot)/object_size(muestra_computos)) ## [1] 3.895024 Adicionalmente incluye funciones para el cálculo de intervalos bootstrap: intervalo_propinas_90 &lt;- bootstraps(propinas, strata = momento, 1000) |&gt; mutate(res_boot = map(splits, estimador)) |&gt; int_pctl(res_boot, alpha = 0.10) El paquete boot está asociado al libro Bootstrap Methods and Their Applications (Davison and Hinkley (1997)) y tiene, entre otras, funciones para calcular replicaciones bootstrap y para construir intervalos de confianza usando bootstrap: calculo de replicaciones bootstrap con la función boot(), intervalos normales, de percentiles y \\(BC_a\\) con la función boot.ci(), intevalos ABC con la función `abc.ci(). El paquete bootstrap contiene datos usados en Efron and Tibshirani (1993), y la implementación de funciones para calcular replicaciones y construir intervalos de confianza: calculo de replicaciones bootstrap con la función bootstrap(), intervalos \\(BC_a\\) con la función bcanon(), intevalos ABC con la función abcnon(). Conclusiones y observaciones El principio fundamental del Bootstrap no paramétrico es que podemos estimar la distribución poblacional con la distribución empírica. Por tanto para hacer inferencia tomamos muestras con reemplazo de la muestra y analizamos la variación de la estadística de interés a lo largo de las remuestras. El bootstrap nos da la posibilidad de crear intervalos de confianza cuando no contamos con fórmulas para hacerlo de manera analítica y sin supuestos distribucionales de la población. Hay muchas opciones para construir intervalos bootstrap, los que tienen mejores propiedades son los intervalos \\(BC_a\\), sin embargo, los más comunes son los intervalos normales con error estándar bootstrap y los intervalos de percentiles de la distribución bootstrap. Antes de hacer intervalos normales vale la pena graficar la distribución bootstrap y evaluar si el supuesto de normalidad es razonable. En cuanto al número de muestras bootstrap se recomienda al menos \\(1,000\\) al hacer pruebas, y \\(10,000\\) o \\(15,000\\) para los resultados finales, sobre todo cuando se hacen intervalos de confianza de percentiles. La función de distribución empírica es una mala estimación en las colas de las distribuciones, por lo que es difícil construir intervalos de confianza (usando bootstrap no paramétrico) para estadísticas que dependen mucho de las colas. O en general para estadísticas que dependen de un número chico de observaciones de una muestra grande. Referencias "],["estimación-por-máxima-verosimilitud.html", "Sección 6 Estimación por máxima verosimilitud Introducción a estimación por máxima verosimilitud Máxima verosimilitud para observaciones continuas Aspectos numéricos Máxima verosimilitud para más de un parámetro", " Sección 6 Estimación por máxima verosimilitud Los ejemplos que hemos visto han sido todos de estimadores plug-in (o por sustitución): si queremos saber una cantidad poblacional, y tenemos una muestra dada, entonces calculamos la estadística de interés como si la muestra fuera la población. Por ejemplo, para estimar la mediana poblacional usamos la mediana muestral, si queremos estimar la media poblacional usamos la media muestral, y así sucesivamente. Estos estimadores usualmente dan resultados razonables (pero hay que checar usando muestra bootstraps, por ejemplo, y pensar lo que estamos haciendo). Cuando sabemos más acerca de la población y usamos un modelo teórico es posible hacer más: dependiendo de qué cantidades se quieren estimar, podemos construir estimadores que sean óptimos en algún sentido siempre y cuando se cumplan los supuestos teóricos, como veremos ahora. Por ejemplo: ¿deberíamos estimar el centro de una distribución simétrica con la media o con la mediana, o quizá con una media recortada? En esta parte construiremos la teoría básica de estimación cuando trabajamos con modelos teóricos conocidos. El objetivo es entender las ideas básicas de estos procedimientos, y cómo evaluar sus resultados. Recordatorio: las ventajas de usar modelos teóricos para describir distribuciones de datos está en que es posible comprimir más eficientemente la información, es posible construir modelos más complejos juntando varios de estos modelos y de sus dependencias, y de que es posible hacer más teoría útil que nos guíe. La desventaja es que es necesario que esos supuestos teóricos sean razonables. Introducción a estimación por máxima verosimilitud Uno de los procedimientos más estándar en esta situación es el método de máxima verosimilitud. Los estimadores de máxima verosimilitud tienen propiedades convenientes, y dan en general resultados razonables siempre y cuando los supuestos sean razonables. Máxima verosimilitud es un proceso intuitivo, y consiste en aprender o estimar valores de parámetros desconocidos suponiendo para los datos su explicación más probable. Para esto, usando supuestos y modelos, requeriremos calcular la probabilidad de un conjunto de observaciones. Ejemplo. Adaptado de (Chihara and Hesterberg 2018). Supongamos que una máquina produce dos tipos de bolsas de 25 galletas: la mitad de las veces produce una bolsa con 5 galletas de avena y 20 de chispas de chocolate, y la otra mitad produce bolsas con 23 galletas de avena y 2 de chispas de chocolate. Tomamos una bolsa, y no sabemos qué tipo de bolsa es (parámetro desconocido). Extraemos al azar una de las galletas, y es de chispas de chocolate (observación). Por máxima verosimilitud, inferimos que la bolsa que estamos considerando tiene 5 galletas de avena. Esto es porque es más probable observar una galleta de chispas en las bolsas que contienen 5 galletas de avena que en las bolsas que contienen 23 galletas de avena. Podemos cuantificar la probabilidad que “acertemos” en nuestra inferencia. Cómo se aprecia en el ejemplo anterior, el esquema general es: Existe un proceso del que podemos obtener observaciones de algún sistema o población real. Tenemos un modelo probabilístico que dice cómo se producen esas observaciones a partir del sistema o población real. Usualmente este modelo tiene algunas cantidades que no conocemos, que rigen el proceso y cómo se relaciona el proceso con las observaciones. Nuestro propósito es: Extraemos observaciones del proceso \\[x_1, x_2, \\ldots, x_n.\\] Queremos aprender de los parámetros desconocidos del proceso para calcular cantidades de interés acerca del sistema o población real En principio, los modelos que consideramos pueden ser complicados y tener varias partes o parámetros. Veamos primero un ejemplo clásico con un solo parámetro, y cómo lo resolveríamos usando máxima verosimilitud. Nota: Cuando decimos muestra en general nos referimos a observaciones independientes obtenidas del mismo proceso (ver la sección de distribución de muestreo) para ver qué significa que sea independientes. Este esquema es un supuesto que simplifica mucho los cálculos, como discutimos antes. Muchas veces este supuesto sale del diseño de la muestra o del estudio, pero en todo caso es importante considerar si es razonable o no para nuestro problema particular. Denotemos por \\(f(x; \\theta)\\) la función de densidad para una variable aleatoria continua con párametro asociado \\(\\theta.\\) Denotamos por \\(X_1, \\ldots, X_n,\\) una muestra aleatoria de \\(n\\) observaciones de esta distribución y por \\(x_1, \\ldots, x_n\\) los valores observados de esta muestra aleatoria. Ejemplo. Supongamos que queremos saber qué proporción de registros de una base de datos tiene algún error menor de captura. No podemos revisar todos los registros, así que tomamos una muestra de 8 registros, escogiendo uno por uno al azar de manera independiente. Revisamos los 8 registros, y obtenemos los siguientes datos: \\[x_1 = 0, x_2 = 1, x_3 = 0, x_4 = 0, x_5 =1, x_6 =0, x_7 =0, x_8 =0\\] donde 1 indica un error menor. Encontramos dos errores menores. ¿Cómo estimamos el número de registros con errores leves en la base de datos? Ya sabemos una respuesta razonable para nuestro estimador puntual, que sería \\(\\hat{p}=2/8=0.25\\). Veamos cómo se obtendría por máxima verosimilitud. Según el proceso con el que se construyó la muestra, debemos dar una probabilidad de observar los 2 errores en 8 registros. Supongamos que en realidad existe una proporción \\(p\\) de que un registro tenga un error. Entonces calculamos Probabilidad de observar la muestra: \\[P(X_1 = 0, X_2 = 1, X_3 = 0, X_4 = 0, X_5 =1, X_6 =0, X_7 =0, X_8 =0)\\] es igual a \\[P(X_1 = 0)P(X_2 = 1)P(X_3 = 0)P( X_4 = 0)P(X_5 =1)P(X_6 =0)P(X_7 =0)P(X_8 =0)\\] pues la probabilidad de que cada observación sea 0 o 1 no depende de las observaciones restantes (la muestra se extrajo de manera independiente). Esta última cantidad tiene un parámetro que no conocemos: la proporcion \\(p\\) de registros con errores. Así que lo denotamos como una cantidad desconocida \\(p\\). Nótese entonces que \\(P(X_2=1) = p\\), \\(P(X_3=0) = 1-p\\) y así sucesivamente, así que la cantidad de arriba es igual a \\[(1-p)p(1-p)(1-p)p(1-p)(1-p)(1-p) \\] que se simplifica a \\[ \\mathcal{L}(p) = p^2(1-p)^6\\] Ahora la idea es encontrar la p que maximiza la probabilidad de lo que observamos. En este caso se puede hacer con cálculo, pero vamos a ver una gráfica de esta función y cómo resolverla de manera numérica. verosimilitud &lt;- function(p){ p^2 * (1-p)^6 } dat_verosim &lt;- tibble(x = seq(0,1, 0.001)) %&gt;% mutate(prob = map_dbl(x, verosimilitud)) ggplot(dat_verosim, aes(x = x, y = prob)) + geom_line() + geom_vline(xintercept = 0.25, color = &quot;red&quot;) + xlab(&quot;p&quot;) Nótese que esta gráfica: Depende de los datos, que pensamos fijos. Cuando cambiamos la \\(p\\), la probabilidad de observar la muestra cambia. Nos interesa ver las regiones donde la probabilidad es relativamente alta. El máximo está en 0.25. Así que el estimador de máxima verosimilitud es \\(\\hat{p} = 0.25\\), que es también el estimador usual de plugin en este caso. Para uniformizar la notación con el caso continuo que veremos más adelante, usaremos la notación \\[P(X=x) = f(x)\\] donde \\(f\\) es la función de densidad (en este caso, función de masa de probabilidad) de \\(X\\). Si esta función depende de un parámetro, escribimos \\[f(x ;\\theta)\\] Definición. Sean \\(X_1, \\ldots, X_n\\) una muestra de una densidad \\(f(x; \\theta)\\) y sean \\(x_1,x_2,\\ldots, x_n\\) los valores observados. La función de verosimilitud del párametro de interés \\(\\theta\\) está definida por \\[\\begin{align} \\mathcal{L}(\\theta; x_1, \\ldots, x_n) = \\prod_{i = 1}^n f(x_i; \\theta). \\end{align}\\] Esta función nos dice qué tan creible es el valor del parámetro \\(\\theta\\) dada la muestra observada. A veces también la denotamos por \\(\\mathcal{L}_n(\\theta)\\). Ahora definimos qué es un estimador de máxima verosimilitud. Definición. Un estimador de máxima verosimilitud lo denotamos por \\(\\hat \\theta_{\\textsf{MLE}}\\) y es un valor que satisface \\[\\begin{align} \\hat \\theta_{\\textsf{MLE}} = \\underset{\\theta \\, \\in \\, \\Theta}{\\arg\\max}\\, \\mathcal{L}(\\theta; x_1, \\ldots, x_n), \\end{align}\\] donde \\(\\Theta\\) denota el espacio parametral. Es decir, el espacio válido de búsqueda congruente con la definición del modelo. Considera el caso de una normal con media y varianza desconocidas. ¿Cuáles son los espacios parametrales para efectuar \\(\\mathsf{MLE}\\)? Considera el caso de una Binomial con parámetro \\(p\\) desconocidos. ¿Cuál es el espacio parametral para la búsqueda del \\(\\mathsf{MLE}\\)? Obsérvese que para construir la verosimilitud y en consecuencia buscar por estimadores de máxima verosimlitud necesitamos: Un modelo teórico de cómo es la población con parámetros e Información de cómo se extrajo la muestra, y entonces podemos resolver nuestro problema de estimación convirtiéndolo en uno de optimización. Probamos esta idea con un proceso más complejo. Ejemplo. Supongamos que una máquina puede estar funcionando correctamente o no en cada corrida. Cada corrida se producen 500 productos, y se muestrean 10 para detectar defectos. Cuando la máquina funciona correctamente, la tasa de defectos es de 3%. Cuando la máquina no está funcionando correctamente la tasa de defectos es de 20% Supongamos que escogemos al azar 11 corridas, y obervamos los siguientes número de defectuosos: \\[1, 0, 0, 3 ,0, 0, 0, 2, 1, 0, 0\\] La pregunta es: ¿qué porcentaje del tiempo la máquina está funcionando correctamente? Primero pensemos en una corrida. La probabilidad de observar una sucesión particular de \\(r\\) defectos es \\[0.03^r(0.97)^{(10-r)}\\] cuando la máquina está funcionando correctamente. Si la máquina está fallando, la misma probabilidad es \\[0.2^r(0.8)^{(10-r)}.\\] Ahora supongamos que la máquina trabaja correctamente en una proporción \\(p\\) de las corridas. Entonces la probabilidad de observar \\(r\\) fallas se calcula promediando (probabilidad total) sobre las probabilidades de que la máquina esté funcionando bien o no: \\[0.03^r(0.97)^{(10-r)}p + 0.2^r(0.8)^{(10-r)}(1-p)\\] Y esta es nuestra función de verosimilitud para una observación. Suponemos que las \\(r_1,r_2, \\ldots, r_{11}\\) observaciones son independientes (por ejemplo, después de cada corrida la máquina se prepara de una manera estándar, y es como si el proceso comenzara otra vez). Entonces tenemos que multiplicar estas probabilidades para cada observación \\(r_1\\): # creamos la función de verosimilitud con los datos observados como dados verosim &lt;- function(p) { r &lt;- c(1, 2, 0, 3, 0, 0, 0, 2, 1, 0, 3) q_func &lt;- 0.03^r*(0.97)^(10-r) q_falla &lt;- 0.2^r*(0.8)^(10-r) prod(p * q_func + (1 - p) * q_falla) } verosim(0.1) # Una alternativa que nos da más flexibilidad para generar la función de # verosimilitud, es crear una función que recibe los datos observados y nos # regresa la función de verosimilitud correspondiente # Entonces, cal_verosim es una función que regresa una función calc_verosim &lt;- function(r){ q_func &lt;- 0.03 ^ r * (0.97) ^ (10 - r) q_falla &lt;- 0.2 ^ r * (0.8) ^ (10 - r) function(p){ #nota: esta no es la mejor manera de calcularlo, hay # que usar logaritmos. prod(p * q_func + (1 - p) * q_falla) } } verosim &lt;- calc_verosim(r = c(1, 0, 0, 3, 0, 0, 0, 2, 1, 0, 0)) verosim(0.1) ## [1] 2.692087e-14 dat_verosim &lt;- tibble(x = seq(0, 1, 0.001)) %&gt;% mutate(prob = map_dbl(x, verosim)) ggplot(dat_verosim, aes(x = x, y = prob)) + geom_line() + geom_vline(xintercept = 0.773, color = &quot;red&quot;) + xlab(&quot;prop funcionado&quot;) Y nuestra estimación puntual sería de alrededor de 80%. Máxima verosimilitud para observaciones continuas Cuando las observaciones \\(x_1,\\ldots, x_n\\) provienen de una distribución continua, no tiene sentido considerar \\(P(X = x_i)\\), pues siempre es igual a cero. Sin embargo, podemos escribir para pequeños valores \\(\\epsilon \\ll 1\\) \\[\\begin{align} P(x - \\epsilon &lt; X &lt; x + \\epsilon | \\theta) = \\int_{x - \\epsilon}^{x + \\epsilon} f(t; \\theta) \\, \\text{d} t \\approx 2 \\epsilon f(x; \\theta), \\end{align}\\] donde \\(f(x; \\theta)\\) es la función de densidad de \\(X.\\) Por lo tanto, \\[\\begin{align} \\begin{split} P&amp;(x_1 - \\epsilon &lt; X_1 &lt; x_1 + \\epsilon, \\ldots, x_n - \\epsilon &lt; X_n &lt; x_n + \\epsilon | \\theta) \\\\ &amp;= \\prod_{i = 1}^n P(x_i - \\epsilon &lt; X_i &lt; x_i + \\epsilon | \\theta) \\\\ &amp;= \\prod_{i = 1}^n 2 \\epsilon f(x_i; \\theta) = (2\\epsilon)^n \\prod_{i = 1}^n f(x_i; \\theta). \\end{split} \\end{align}\\] Notemos que si \\(\\epsilon \\rightarrow 0\\) la ecuación rápidamente converge a cero. Pero para pequeños valores de \\(\\epsilon\\) la ecuación que nos interesa es proporcional a \\(\\prod_{i = 1}^n f(x_i; \\theta).\\) De esta forma, nuestra definición de máxima verosimilitud y estimadores de máxima verosimilitud es la misma para el caso continuo (verifica las definiciones de la sección anterior). Ejemplo. Supongamos que tenemos una muestra \\(x_1\\ldots, x_n\\) extraidas de una distribución exponencial con tasa \\(\\lambda&gt;0\\) donde no conocemos \\(\\lambda\\). ¿Cuál es el estimador de máxima verosimilitud de \\(\\lambda\\)? Para \\(\\lambda&gt;0\\), tenemos que \\[{\\mathcal L}(\\lambda) = \\prod_{i=1}^n \\lambda e^{-\\lambda x_i}\\] de modo que \\[{\\mathcal L}(\\lambda) = \\lambda^n e^{-\\lambda \\sum_{i=1}^nx_i} = \\lambda^n e^{-n\\lambda\\bar{x}} = e^{n(\\log\\lambda - \\lambda\\bar{x})}\\] Que podemos maximizar usando cálculo para obtener \\(\\hat{\\lambda}_{\\mathsf{ML}} = \\frac{1}{\\bar{x}}\\) (demuéstralo). Discute por qué esto es intuitivamente razonable: ¿cuál es el valor esperado de una exponencial con parámetro \\(\\lambda\\)? Aspectos numéricos Encontrar el estimador de máxima verosimilitud (\\(\\textsf{MLE}\\)) es automático en la mayoría de los casos. En teoría, podemos reutilizar la misma rutina numérica para encontrar el estimador sin ninguna ayuda de la analista. Esto contrasta con otras técnicas de estimación en donde se requieren cálculos y manipulación de ecuaciones. Sin embargo, hay situaciones que se pueden evitar de manera general. Por ejemplo, cuando calculamos la verosimilitud arriba, nótese que estamos multiplicando números que pueden ser muy chicos (por ejemplo \\(p^6\\), etc). Esto puede producir desbordes numéricos fácilmente. Por ejemplo para un tamaño de muestra de 1000, podríamos tener que calcular p &lt;- 0.1 proba &lt;- (p ^ 800)*(1-p)^200 proba ## [1] 0 En estos casos, es mejor hacer los cálculos en escala logarítmica. El logaritmo convierte productos en sumas, y baja exponentes multiplicando. Si calculamos en escala logaritmica la cantidad de arriba, no tenemos problema: log_proba &lt;- 800 * log(p) + 200 * log(1-p) log_proba ## [1] -1863.14 Ahora notemos que Maximizar la verosimilitud es lo mismo que maximizar la log-verosimilitud, pues el logaritmo es una función creciente. Si \\(x_{\\max}\\) es el máximo de \\(f\\), tenemos que \\(f(x_{\\max})&gt;f(x)\\) para cualquier \\(x\\), entonces tomando logaritmo, \\[\\log(f(x_{max}))&gt;\\log(f(x)),\\] para cualquier \\(x.\\) Pues el logaritmo respeta la desigualdad por ser creciente. Usualmente usamos la log-verosimilitud para encontrar el estimador de máxima verosimilitud. Hay razónes teóricas y de interpretación por las que también es conveniente hacer esto. Definición. La log-verosimilitud la denotamos usualmente por \\[\\ell_n(\\theta) = \\log \\left(\\mathcal{L}_n(\\theta)\\right),\\] donde hemos suprimido la dependencia en la muestra por conveniencia. Ejemplo. En nuestro primer ejemplo, log_verosimilitud &lt;- function(p){ 2*log(p) + 6*log(1-p) } dat_verosim &lt;- tibble(x = seq(0,1, 0.01)) %&gt;% mutate(log_prob = map_dbl(x, log_verosimilitud)) ggplot(dat_verosim, aes(x = x, y = log_prob)) + geom_line() + geom_vline(xintercept = 0.25, color = &quot;red&quot;) + xlab(&quot;p&quot;) Obtenemos el mismo máximo. Podemos incluso resolver numéricamente: solucion &lt;- optim(p = 0.5, log_verosimilitud, control = list(fnscale = -1), method = &quot;Brent&quot;, lower = 0, upper = 1) solucion$par ## [1] 0.25 Y en nuestro segundo ejemplo: calc_log_verosim &lt;- function(r){ q_func &lt;- 0.03^r*(0.97)^(10-r) q_falla &lt;- 0.2^r*(0.8)^(10-r) function(p){ #nota: esta no es la mejor manera de calcularlo, hay # que usar logaritmos. sum(log(p * q_func + (1 - p) * q_falla)) } } log_verosim &lt;- calc_log_verosim(c(1, 0, 0, 3, 0, 0, 0, 2, 1, 0, 0)) log_verosim(0.1) ## [1] -31.24587 solucion &lt;- optim(p = 0.2, log_verosim, control = list(fnscale = -1), method = &quot;Brent&quot;, lower = 0, upper = 1) solucion$par ## [1] 0.7733766 solucion$convergence ## [1] 0 dat_verosim &lt;- tibble(x = seq(0,1, 0.001)) %&gt;% mutate(log_verosimilitud = map_dbl(x, log_verosim)) ggplot(dat_verosim, aes(x = x, y = log_verosimilitud)) + geom_line() + geom_vline(xintercept = 0.775, color = &quot;red&quot;) + xlab(&quot;prop funcionado&quot;) Nótese que la verosimilitud la consideramos función de los parámetros, donde los datos están fijos. Podemos construir una función que genera la función de verosimilitud dependiendo de los datos. En nuestro primer ejemplo de muestras de registros erróneos, podríamos construir una función que genera la log verosimilitud dependiendo del tamaño de muestra y del número de errores encontrado: construir_log_verosim &lt;- function(n, n_err){ # n es tamaño de muestra # n_err el número de errores detectados (datos) n_corr &lt;- n - n_err log_verosim &lt;- function(p){ n_err * log(p) + n_corr * log(1-p) } } Cuando fijamos \\(n\\) y \\(n_{\\textsf{err}}\\), esta función genera otra función, la log verosimilitud, que es la que queremos optimizar. Supongamos entonces que sacamos 20 registros al azar y observamos 10 incorrectos. La función de verosimilitud es log_vero &lt;- construir_log_verosim(20, 10) tibble(x = seq(0,1,0.001)) %&gt;% mutate(log_ver = log_vero(x)) %&gt;% ggplot(aes(x = x, y = log_ver)) + geom_line() + geom_vline(xintercept = 0.5, color = &#39;red&#39;) Ejemplo. Supongamos que en una población de transacciones hay un porcentaje \\(p\\) (desconocido) que son fraudulentas. Tenemos un sistema de clasificación humana que que marca transacciones como sospechosas. Con este sistema hemos medido que la proporción de transacciones normales que son marcadas como sospechosas es de 0.1%, y que la proporción de transacciones fraudulentas que son marcadas como sospechosas es de 98%. Supongamos que extraemos una muestra de 2000 transacciones, de manera que todas ellas tiene la misma probabilidad de ser fraudulentas. El sistema de clasificación marca 4 transacciones como sospechosas ¿Cómo estimamos la proporción de transacciones fraudulentas en la población? Solución: sea \\(p\\) la proporción de transacciones fraudulentas. Entonces la probabilidad de que una transacción sea marcada como sospechosa es (proba total): \\[0.98p + 0.001(1-p)\\] Pues tenemos que contar 98% de la proporción \\(p\\) de fraudulentas (correctamente detectadas) más 0.1% de la proporción \\((1-p)\\) de fraudulentas. Escribimos entonces nuestra función de verosimilitud crear_log_verosim &lt;- function(n, n_sosp){ # devolver la función log verosimilitud log_verosimilitud_pct &lt;- function(pct){ # sup que pct es la proporcentaje de fraudes, # que es el parámetro que queremos estimar prob_sosp &lt;- 0.98 * pct / 100 + 0.001 * (1 - pct / 100) log_prob &lt;- n_sosp * log(prob_sosp) + (n - n_sosp) * log(1- prob_sosp) log_prob } log_verosimilitud_pct } La verosimilitud es una función de \\(p\\). log_verosim &lt;- crear_log_verosim(n = 2000, n_sosp = 4) A continuación la mostramos de manera gráfica. No se ve muy claro dónde ocurre el máximo, pero podemos ampliar cerca de cero la misma gráfica: Vemos que alrededor de 0.1% maximiza la probabilidad de haber observado 4 transacciones sospechosas. Notamos sin embargo que varios valores alrededor de este valor tienen probabilidad similar, así que también son consistentes con los datos (por ejemplo, valores como 0.05 o 0.15 tienen probabilidad similar). Tendremos que considerar esto para evaluar la incertidumbre en nuestra estimación. Obsérvese adicionalmente que si no tomáramos en cuenta las probabilidades de falsos negativos y falsos positivos la estimación simple daría \\(4/2000 = 0.002\\) (0.2%), que es dos veces más grande que nuestra estimación puntual por máxima verosimilitud. Ejemplo. Este es un ejemplo donde mostramos que cuando el soporte de las densidades teóricas es acotado hay que tener cuidado en la definición de la verosimilitud. En este caso, el soporte de la variable aleatoria es el párametro de interés. Supongamos que nuestros datos son generados por medio de una distribución uniforme en el intervalo \\([0,b].\\) Contamos con una muestra de \\(n\\) observaciones generadas de manera independiente \\(X_i \\sim U[0,b]\\) para \\(i= 1, \\ldots, n.\\) Sin embargo, no conocemos el valor de \\(b\\). ¿Cómo es la función de log verosimilitud \\({\\mathcal L}_n(b)\\) para este caso? Nótese que cuando el parámetro \\(b\\) es menor que alguna \\(x_i\\), tenemos que \\({\\mathcal L}_n(b) = 0\\): la verosimilitud es cero si tomamos una \\(b\\) más chica que algún dato, pues este valor es incosistente del todo con los datos observados. En otro caso, \\[{\\mathcal L}_n(b) = \\frac{1}{b^n},\\] pues la función de densidad de una uniforme en \\([0,b]\\) es igual a \\(1/b\\) en el intervalo \\([0,b]\\), y 0 en otro caso. Podemos escribir entonces: crear_verosim &lt;- function(x){ n &lt;- length(x) verosim &lt;- function(b){ indicadora &lt;- ifelse(all(x &lt;= b), 1, 0) indicadora / b^n } } Ahora podemos hacer máxima verosimilitud para un ejemplo: set.seed(234) x &lt;- runif(10, 0, 3) verosim &lt;- crear_verosim(x) res_opt &lt;- optimize(verosim, c(-1000, 1000), maximum = TRUE) res_opt$maximum ## [1] 2.788167 Y nótese que, como esperaríamos, este valor es el máximo de la muestra: max(x) ## [1] 2.788158 La gráfica de la función de verosimilitud es: tibble(b = seq(-1, 5, 0.001)) %&gt;% mutate(verosim_1 = map_dbl(b, ~ verosim(.x))) %&gt;% ggplot() + geom_line(aes(x = b, y = verosim_1)) + geom_rug(data = tibble(x = x), aes(x = x), colour = &quot;red&quot;) Podemos escribir en una fórmula como: \\[\\begin{align} \\mathcal{L}(b; x_1, \\ldots, x_n) = \\prod_{i = 1}^n 1_{[0,b]}(x_i) \\frac1b. \\end{align}\\] Y podríamos resolver analíticamente como sigue: Si consideramos \\[ \\hat b_{\\textsf{MLE}} = x_{\\max} = \\max\\{x_i\\},\\] notemos que cualquier valor observado necesariamente satisface \\[x_i \\leq \\hat b_{\\textsf{MLE}},\\] y por lo tanto todas las funciones indicadoras están encendidas. El valor de la verosimilitud es igual a \\[\\mathcal{L}(\\hat b_{\\textsf{MLE}}; x_1, \\ldots, x_n) = \\left(\\frac{1}{x_{\\max}}\\right)^n \\geq \\left (\\frac1b\\right )^n\\] para cualquier \\(b\\geq x_{\\max}\\). Como la verosimilitud para \\(b&lt;x_{\\max}\\) es igual a cero, esto demuestra que el máximo de la muestra es el estimador de máxima verosimilitud de \\(b\\). Observación. Este ejemplo también tiene dificultades numéricas, pues la verosimilitud presenta discontinuidades y regiones con derivada igual a cero, y la mayoria de los algoritmos numéricos no tienen garantías buenas de covergencia al máximo en estos casos. Si aplicamos sin cuidado descenso en gradiente, por ejemplo, podríamos comenzar incorrectamente en un valor \\(b_0 &lt; x_{\\max}\\) y el algoritmo no avanzaría al máximo. Máxima verosimilitud para más de un parámetro Si nuestro modelo contiene más de un parámetro desconocido podemos también usar máxima verosimilitud. En este caso, optimizamos sobre todos los parámetros usando cálculo o alguna rutina numérica. Ejemplo. Considera el caso de \\(n\\) muestras iid de un modelo Gaussiano. Es decir, \\(X_1, \\ldots, X_n \\sim \\mathsf{N}(\\mu, \\sigma^2).\\) Consideremos que ambos parámetros son desconocidos y nos gustaria encontrar el \\(\\textsf{MLE}\\). Para este problema denotamos \\(\\theta \\in \\mathbb{R}^2\\), donde \\(\\theta_1 = \\mu\\) y \\(\\theta_2 = \\sigma^2.\\) La función de verosimiltud se puede calcular (ignorando algunas constantes multiplicativas) como \\[\\begin{align} \\mathcal{L}_n(\\theta) &amp;= \\prod_{i = 1}^n \\frac{1}{\\sigma} \\, \\exp\\left( - \\frac{(x_i - \\mu)^2}{2\\sigma^2}\\right) \\\\ &amp;= \\theta_2^{-\\frac{n}{2}}\\exp\\left( - \\frac{1}{2 \\theta_2} \\sum_{i = 1}^n (x_i - \\theta_1)^2 \\right). \\end{align}\\] A continuación mostramos la representación gráfica de la función de verosimilitud de este ejemplo. Notamos lo mismo que para los ejemplos anteriores. Conforme más datos tenemos, más nos acercamos a los valores reales que no conocemos. Ejemplo. Como ejercicio, podemos encontrar los estimadores de máxima verosimilitud cuando tenemos una muestra \\(X_1, \\ldots, X_n \\sim \\mathsf{N}(\\mu, \\sigma^2).\\) (puedes derivar e igualar el cero para encontrar el mínimo). También podemos resolver numéricamente, por ejemplo: Supongamos que tenemos la siguiente muestra: set.seed(41852) muestra &lt;- rnorm(150, mean = 1, sd = 2) La función generadora de la log verosimilitud para una muestra es (ve la expresión del ejercicio anterior y calcula su logaritmo), y generamos la función de verosimilitud para nuestra muestra: crear_log_p &lt;- function(x){ log_p &lt;- function(pars){ media = pars[1] desv_est = pars[2] # ve la ecuación del ejercicio anterior z &lt;- (x - media) / desv_est log_verosim &lt;- -(log(desv_est) + 0.5 * mean(z ^ 2)) log_verosim } log_p } log_p &lt;- crear_log_p(muestra) Ahora optimizamos: res &lt;- optim(c(0, 0.5), log_p, control = list(fnscale = -1, maxit = 1000), method = &quot;Nelder-Mead&quot;) res$convergence ## [1] 0 est_mv &lt;- tibble(parametro = c(&quot;media&quot;, &quot;sigma&quot;), estimador = res$par) %&gt;% column_to_rownames(var = &quot;parametro&quot;) est_mv ## estimador ## media 1.136001 ## sigma 1.838421 Verifica que el estimador de la media y de la desviación estándar es el que esperábamos (y que puedes derivar analíticamente): n &lt;- length(muestra) sd_n &lt;- function(x) sqrt(mean((x - mean(x))^2)) c(media = mean(muestra), sigma = sd_n(muestra)) %&gt;% round(4) ## media sigma ## 1.1364 1.8392 Ejemplo. Supongamos que en una población de estudiantes tenemos dos tipos: unos llenaron un examen de opción múltiple al azar (1 de 5), y otros contestaron las preguntas intentando sacar una buena calificación. Suponemos que una vez que conocemos el tipo de estudiante, todas las preguntas tienen la misma probabilidad de ser contestadas correctamente, de manera independiente. El modelo teórico está representado por la siguiente simulación: sim_formas &lt;- function(p_azar, p_corr){ tipo &lt;- rbinom(1, 1, 1 - p_azar) if(tipo==0){ # al azar x &lt;- rbinom(1, 10, 1/5) } else { # no al azar x &lt;- rbinom(1, 10, p_corr) } x } Y una muestra se ve como sigue: set.seed(12) muestra &lt;- map_dbl(1:200, ~ sim_formas(0.3, 0.75)) qplot(muestra) Supongamos que no conocemos la probabildad de contestar correctamente ni la proporción de estudiantes que contestó al azar. ¿Como estimamos estas dos cantidades? Escribimos la verosimilitud: crear_log_p &lt;- function(x){ log_p &lt;- function(pars){ p_azar = pars[1] p_corr = pars[2] sum(log(p_azar * dbinom(x, 10, 1/5) + (1 - p_azar) * dbinom(x, 10, p_corr))) } log_p } Creamos la función de verosimilitud con los datos log_p &lt;- crear_log_p(muestra) y optimizamos res &lt;- optim(c(0.5, 0.5), log_p, control = list(fnscale = -1)) res$par ## [1] 0.2827061 0.7413276 En este caso, obtenemos estimaciones razonables de ambos parámetros. Nota: dependiendo de los datos, este problema puede estar mal condicionado. Por ejemplo, ¿qué pasa si la probabilidad de acertar cuando se contesta bien está cercano al azar? La siguiente pregunta qué nos interesa hacer es: ¿cómo estimamos la variabilidad de estos estimadores? Más adelante veremos una respuesta basada en teoría, pero también podemos resolver este problema usando el bootstrap. Referencias "],["bootstrap-paramétrico.html", "Sección 7 Bootstrap paramétrico Ventajas y desventajas de bootstrap paramétrico Verificando los supuestos distribucionales Modelos mal identificados", " Sección 7 Bootstrap paramétrico Cuando nuestras observaciones provienen de un modelo teórico parametrizado con algunos parámetros que queremos estimar, y utilizamos máxima verosimilitud para hacer nuestra estimación, no es adecuado aplicar directamente el bootstrap no paramétrico que vimos en las secciones anteriores. Sin embargo, suponiendo que el modelo paramétrico que estamos usando es apropiado, podemos remuestrear de tal modelo para estimar la varianza de nuestros estimadores. Este proceso se llama el bootstrap paramétrico. Antes de hacer una definición precisa, veamos cómo calcularíamos error estándar para los estimadores de máxima verosimilitud de la normal que vimos en la sección anterior. Ejemplo (sección máxima verosimilitud). Como ejercicio, podemos encontrar los estimadores de máxima verosimilitud cuando tenemos una muestra \\(X_1, \\ldots, X_n \\sim \\mathsf{N}(\\mu, \\sigma^2)\\) (puedes derivar e igualar a cero para encontrar el mínimo). También podemos resolver numéricamente. Supongamos que tenemos la siguiente muestra: set.seed(41852) muestra &lt;- rnorm(150, mean = 1, sd = 2) La función generadora de la log-verosimilitud para una muestra es (ve la expresión del ejercicio anterior y calcula su logaritmo), y generamos la función de verosimilitud para nuestra muestra: crear_log_p &lt;- function(x){ log_p &lt;- function(pars){ media = pars[1] desv_est = pars[2] # ve la ecuación del ejercicio anterior z &lt;- (x - media) / desv_est log_verosim &lt;- -(log(desv_est) + 0.5 * mean(z^2)) log_verosim } log_p } log_p &lt;- crear_log_p(muestra) Ahora optimizamos (revisa que el método converge): res &lt;- optim(c(0, 0.5), log_p, control = list(fnscale = -1, maxit = 1000), method = &quot;Nelder-Mead&quot;) res$convergence ## [1] 0 est_mle &lt;- tibble(parametro = c(&quot;media&quot;, &quot;sigma&quot;), estimador = res$par) |&gt; column_to_rownames(var = &quot;parametro&quot;) Una vez que tenemos nuestros estimadores puntuales, est_mle ## estimador ## media 1.136001 ## sigma 1.838421 Sustituimos estos parámetros en la distribución normal y simulamos una muestra del mismo tamaño que la original: simular_modelo &lt;- function(n, media, sigma){ rnorm(n, media, sigma) } muestra_bootstrap &lt;- simular_modelo(length(muestra), est_mle[&quot;media&quot;, &quot;estimador&quot;], est_mle[&quot;sigma&quot;, &quot;estimador&quot;]) head(muestra_bootstrap) ## [1] 1.8583885 2.2084326 2.5852895 2.5174462 -0.7428032 0.5995989 Una vez que tenemos esta muestra bootstrap recalculamos los estimadores de máxima verosimlitud. Esto se hace optimizando: # creamos nueva verosimilitud para muestra bootstrap log_p_boot &lt;- crear_log_p(muestra_bootstrap) # optimizamos res_boot &lt;- optim(c(0, 0.5), log_p_boot, control = list(fnscale = -1, maxit = 1000), method = &quot;Nelder-Mead&quot;) res_boot$convergence ## [1] 0 est_mle_boot &lt;- tibble(parametro = c(&quot;media&quot;, &quot;sigma&quot;), estimador = res_boot$par) |&gt; column_to_rownames(var = &quot;parametro&quot;) est_mle_boot ## estimador ## media 1.235914 ## sigma 1.710042 Y esta es nuestra replicación bootstrap de los estimadores de máxima verosimilitud. La idea es la misma que el bootstrap no paramétrico, con la ventaja de que estamos simulando del modelo que suponemos es el correcto, es decir, estamos usando información adicional que no teníamos en el bootstrap no paramétrico. Ahora es necesario repetir un número grande de veces. Nótese que esta función envuelve el proceso de remuestreo, cálculo de la función de verosimilitud y optimización: rep_boot &lt;- function(rep, crear_log_p, est_mle, n){ muestra_bootstrap &lt;- simular_modelo(length(muestra), est_mle[&quot;media&quot;, &quot;estimador&quot;], est_mle[&quot;sigma&quot;, &quot;estimador&quot;]) log_p_boot &lt;- crear_log_p(muestra_bootstrap) # optimizamos res_boot &lt;- optim(c(0, 0.5), log_p_boot, control = list(fnscale = -1, maxit = 1000), method = &quot;Nelder-Mead&quot;) try(if(res_boot$convergence != 0) stop(&quot;No se alcanzó convergencia.&quot;)) tibble(parametro = c(&quot;media&quot;, &quot;sigma&quot;), estimador_boot = res_boot$par) } reps_boot &lt;- map_dfr(1:5000, ~ rep_boot(.x, crear_log_p, est_mle, n = length(muestra)), rep = &quot;.id&quot;) reps_boot ## # A tibble: 10,000 × 2 ## parametro estimador_boot ## &lt;chr&gt; &lt;dbl&gt; ## 1 media 0.797 ## 2 sigma 1.90 ## 3 media 1.23 ## 4 sigma 1.96 ## 5 media 1.14 ## 6 sigma 1.89 ## 7 media 1.33 ## 8 sigma 1.73 ## 9 media 1.19 ## 10 sigma 1.73 ## # ℹ 9,990 more rows Ya ahora podemos estimar error estándar: error_est &lt;- reps_boot |&gt; group_by(parametro) |&gt; summarise(ee_boot = sd(estimador_boot)) error_est ## # A tibble: 2 × 2 ## parametro ee_boot ## &lt;chr&gt; &lt;dbl&gt; ## 1 media 0.150 ## 2 sigma 0.106 Así que nuestra estimación final sería: bind_cols(est_mle, error_est) |&gt; mutate(across(where(is.numeric), round, 3)) |&gt; select(parametro, estimador, ee_boot) ## parametro estimador ee_boot ## media media 1.136 0.150 ## sigma sigma 1.838 0.106 Si usamos la rutina estándar de R (dejaremos para después explicar cómo calcula los errores estándar esta rutina —no es con bootstrap): broom::tidy(MASS::fitdistr(muestra, &quot;normal&quot;)) ## # A tibble: 2 × 3 ## term estimate std.error ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 mean 1.14 0.150 ## 2 sd 1.84 0.106 Podemos revisar también la normalidad aproximada de las distribuciones bootstrap para construir nuestros intervalos: ggplot(reps_boot, aes(sample = estimador_boot)) + geom_qq() + geom_qq_line(colour = &quot;red&quot;) + facet_wrap(~parametro, scales = &quot;free_y&quot;) La distribuciones son aproximadamente normales. Nótese que esto no siempre sucede, especialmente con parámetros de dispersión como \\(\\sigma\\). Ejemplo. Supongamos que tenemos una muestra más chica. Repasa los pasos para asegurarte que entiendes el procedimiento: set.seed(4182) muestra &lt;- rnorm(6, mean = 1, sd = 2) # función de verosimilitud log_p &lt;- crear_log_p(muestra) # máxima verosimilitud res &lt;- optim(c(0, 0.5), log_p, control = list(fnscale = -1, maxit = 1000), method = &quot;Nelder-Mead&quot;) res$convergence ## [1] 0 est_mle &lt;- tibble(parametro = c(&quot;media&quot;, &quot;sigma&quot;), estimador = res$par) |&gt; column_to_rownames(var = &quot;parametro&quot;) est_mle ## estimador ## media 0.3982829 ## sigma 2.3988969 Hacemos bootstrap paramétrico reps_boot &lt;- map_dfr(1:5000, ~ rep_boot(.x, crear_log_p, est_mle, n = length(muestra)), .id = &quot;rep&quot;) reps_boot ## # A tibble: 10,000 × 3 ## rep parametro estimador_boot ## &lt;chr&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 1 media 0.789 ## 2 1 sigma 0.945 ## 3 2 media -0.103 ## 4 2 sigma 1.37 ## 5 3 media 1.96 ## 6 3 sigma 1.70 ## 7 4 media 1.55 ## 8 4 sigma 2.28 ## 9 5 media -0.228 ## 10 5 sigma 1.73 ## # ℹ 9,990 more rows ggplot(reps_boot, aes(sample = estimador_boot)) + geom_qq() + geom_qq_line(colour = &quot;red&quot;) + facet_wrap(~parametro, scales = &quot;free_y&quot;) ggplot(reps_boot, aes(x = estimador_boot)) + geom_histogram() +facet_wrap(~parametro) Donde vemos que la distribución de \\(\\sigma\\) es asimétrica pues en algunos casos obtenemos estimaciones muy cercanas a cero. Podemos usar intervalos de percentiles. Ejercicio (extra). Con más de un parámetro, podemos preguntarnos cómo dependen las estimaciones individuales - en algunos casos pueden estar correlacionadas. Podemos examinar este comportamiendo visualizando las replicaciones bootstrap ggplot(reps_boot |&gt; pivot_wider(names_from = parametro, values_from = estimador_boot), aes(x = media, y = sigma)) + geom_point(alpha = 0.5) + coord_equal() Esta es nuestra aproximación a la distribución de remuestreo de nuestro par de estadísticas \\((\\mu_{\\mathsf{MLE}}, \\sigma_{\\mathsf{MLE}})\\). En este caso, parecen ser independientes (lo cual es posible demostrar). Bootstrap paramétrico. Supongamos que tenemos una muestra iid \\(X_1,X_2,\\ldots, X_n \\sim f(x;\\theta)\\) de un modelo paramétrico, y un estimador de máxima verosimilitud \\(\\hat{\\theta}_{\\mathsf{MLE}}\\) para \\(\\theta\\). El error estándar estimado para \\(\\hat{\\theta}_{\\mathsf{MLE}}\\) por medio del bootstrap paramétrico se calcula como sigue: Se calcula \\(\\hat{\\theta}_{\\mathsf{MLE}}\\) para la muestra observada Se simula una muestra iid de tamaño \\(n\\) de \\(f(x; \\hat{\\theta}_{\\mathsf{MLE}})\\) (muestra bootstrap) Se recalcula el estimador de máxima verosimilitud para la muestra bootstrap \\(\\hat{\\theta^*}_{\\mathsf{MLE}}\\) Se repiten 2-3 una cantidad grande de veces (1000 - 10000) Se calcula la desviación estándar de los valores \\(\\hat{\\theta^*}_{\\mathsf{MLE}}\\) obtenidos. Este es el error estándar estimado para el estimador \\(\\hat{\\theta}_{\\mathsf{MLE}}\\) Ventajas y desventajas de bootstrap paramétrico Ventaja: el bootstrap paramétrico puede dar estimadores más precisos e intervalos más angostos y bien calibrados que el no paramétrico, siempre y cuando el modelo teórico sea razonable. Desventaja: Es necesario decidir el modelo teórico, que tendrá cierto grado de desajuste vs. el proceso generador real de los datos. Si el ajuste es muy malo, los resultados tienen poca utilidad. Para el no paramétrico no es necesario hacer supuestos teóricos. Ventaja: el bootstrap paramétrico puede ser más escalable que el no paramétrico, pues no es necesario cargar y remuestrear los datos originales, y tenemos mejoras adicionales cuando tenemos expresiones explícitas para los estimadores de máxima verosimilitud (como en el caso normal, donde es innecesario hacer optimización numérica). Desventaja: el bootstrap paramétrico es conceptualmente más complicado que el no paramétrico, y como vimos arriba, sus supuestos pueden ser más frágiles que los del no-paramétrico. Verificando los supuestos distribucionales Como hemos discutido antes, podemos hacer pruebas de hipótesis para revisar si una muestra dada proviene de una distribución conocida. Sin embargo, la herramienta más común es la de los qq-plots, donde podemos juzgar fácilmente el tamaño de las desviaciones y si estas tienen implicaciones prácticas importantes. El proceso es como sigue: si \\(X_1,X_,\\ldots, X_n\\) es una muestra de \\(f(x;\\theta)\\), calculamos el estimador de máxima verosimilitud \\(\\theta_{\\mathsf{MLE}}\\) con los datos observados. Enchufamos \\(\\hat{f} = f(x;\\theta_{\\mathsf{MLE}})\\), y hacemos una gráfica de los cuantiles teóricos de \\(\\hat{f}\\) contra los cuantiles muestrales. Ejemplo. Consideramos la siguiente muestra: set.seed(32) muestra &lt;- rgamma(150, 0.4, 1) qplot(muestra) Y queremos usar un modelo exponencial. Encontramos los estimadores de maxima verosimilitud est_mle &lt;- MASS::fitdistr(muestra, &quot;exponential&quot;) rate_mle &lt;- est_mle$estimate rate_mle ## rate ## 2.76054 g_exp &lt;- ggplot(tibble(muestra = muestra), aes(sample = muestra)) + geom_qq(distribution = stats::qexp, dparams = list(rate = rate_mle)) + geom_abline() + labs(subtitle = &quot;Gráfica de cuantiles exponenciales&quot;) g_exp Donde vemos que el desajuste es considerable, y los datos tienen una cola derecha considerablemente más larga que la de exponencial (datos son casi dos veces más grande de lo que esperaríamos), y la cola izquierda está más comprimida en los datos que en una exponencial. Sin embargo, si ajustamos una gamma: est_mle &lt;- MASS::fitdistr(muestra, &quot;gamma&quot;)$estimate g_gamma &lt;- ggplot(tibble(muestra = muestra), aes(sample = muestra)) + geom_qq(distribution = stats::qgamma, dparams = list(shape = est_mle[1], rate = est_mle[2])) + geom_abline() + labs(subtitle = &quot;Gráfica de cuantiles gamma&quot;) g_exp + g_gamma El ajuste es considerablemente mejor para la distribución gamma (puedes hacer el protocolo rorschach para afinar tu diagnóstico de este tipo de gráficas). Ejempĺo. Examinamos un modelo teórico para las cuentas totales del conjunto de datos de propinas. En primer lugar: Separamos comida y cena, pues sabemos que las cuentas tienden a ser más grandes en las cenas. En lugar de modelar la cuenta total, modelamos el gasto por persona, es decir, la cuenta total dividida por el numero de personas. Grupos grandes pueden producir colas largas que no tenemos necesidad de modelar de manera probabilística, pues conocemos el número de personas. En este caso intentaremos un modelo lognormal, es decir, el logaritmo de los valores observados se comporta aproximadamente normal. Puedes también intentar con una distribución gamma. Separamos por Cena y Comida, dividimos entre número de personas y probamos ajustando un modelo para cada horario: propinas &lt;- read_csv(&quot;data/propinas.csv&quot;) |&gt; mutate(cuenta_persona = cuenta_total / num_personas) propinas_mle &lt;- propinas |&gt; group_by(momento) |&gt; summarise(est_mle = list(tidy(MASS::fitdistr(cuenta_persona, &quot;lognormal&quot;)))) |&gt; unnest(est_mle) propinas_mle ## # A tibble: 4 × 4 ## momento term estimate std.error ## &lt;chr&gt; &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 Cena meanlog 2.03 0.0273 ## 2 Cena sdlog 0.362 0.0193 ## 3 Comida meanlog 1.94 0.0366 ## 4 Comida sdlog 0.302 0.0259 Ojo: estos parámetros están en escala logarítmica. Puedes revisar aquí para ver cómo calcular media y desviación estándar de las distribuciones originales. Ahora verificamos el ajuste: g_1 &lt;- ggplot(propinas |&gt; filter(momento == &quot;Cena&quot;), aes(sample = cuenta_persona)) + geom_qq(dparams = list(mean = propinas_mle$estimate[1], sd = propinas_mle$estimate[2]), distribution = stats::qlnorm) + ylim(c(0, 20)) + geom_abline() + labs(subtitle = &quot;Cena&quot;) g_2 &lt;- ggplot(propinas |&gt; filter(momento == &quot;Comida&quot;), aes(sample = cuenta_persona)) + geom_qq(dparams = list(mean = propinas_mle$estimate[3], sd = propinas_mle$estimate[4]), distribution = stats::qlnorm) + ylim(c(0, 20)) + geom_abline() + labs(subtitle = &quot;Comida&quot;) g_1 + g_2 El ajuste es bueno, aunque podríamos revisar la cola de la derecha en la Comida: ¿por qué existen esos valores relativamente grandes (alrededor de 25% más altos de lo que esperaríamos). ¿Tiene sentido ajustar dos distribuciones con parámetros separados? ¿Crees que estas dos distribuciones podrían compartir algún parámetro? Para esto puedes revisar el error estándar de los estimadores de máxima verosimilitud que mostramos arriba. ¿Qué ventajas verías en usar menos parámetros? ¿Cómo implementarías la estimación? ¿Qué pasa si intentas ajustar un modelo normal a estos datos? Modelos mal identificados Para algunos modelos y algunos parámetros, puede suceder que existan varias configuraciones muy diferentes de los parámetros que sean consistentes con los datos (en términos de verosimilitud, tienen verosimilitud alta similar), y en estos casos decimos que el modelo (con los datos observados) está mal identificado. Esto presenta problemas múltiples: optimizar es más difícil, hay incertidumbre grande en la estimación, y los parámetros se acoplan, haciendo difícil su interpretación. Ejemplo. Consideremos el ejemplo donde queríamos estimar dos proporciones: la proporción de examenes contestados al azar y la tasa de correctos. Vamos a suponer que la probabilidad de tener respuesta correcta dado que el examen no fue contestado al azar no es muy lejano a 1/5, que es la probabilidad de acertar a al azar. Aquí está la función para simular y la log verosimilitud correspondiente. Aquí vamos a ver un problema más difícil, así que usaremos la transformación logit para las proporciones, y no obtener resultados fuera del rango 0-1 al optimizar: inv_logit &lt;- function(theta){ exp(theta) / (1 + exp(theta)) } logit &lt;- function(p){ log(p / (1-p)) } # Simular datos sim_formas &lt;- function(probs){ p_azar &lt;- probs[1] p_corr &lt;- probs[2] tipo &lt;- rbinom(1, 1, 1 - p_azar) if(tipo==0){ # al azar x &lt;- rbinom(1, 10, 1/5) } else { # no al azar x &lt;- rbinom(1, 10, p_corr) } x } simular_modelo &lt;- function(n, params){ muestra &lt;- map_dbl(1:n, ~ sim_formas(probs = inv_logit(params))) muestra } # log verosimilitud crear_log_p &lt;- function(x){ log_p &lt;- function(pars){ p_azar = inv_logit(pars[1]) p_corr = inv_logit(pars[2]) sum(log(p_azar * dbinom(x, 10, 1/5) + (1 - p_azar) * dbinom(x, 10, p_corr))) } log_p } # simular datos set.seed(12) muestra &lt;- simular_modelo(2000, params = logit(c(0.3, 0.29))) qplot(muestra) log_p &lt;- crear_log_p(muestra) res &lt;- optim(c(0.0, 0.0), log_p, control = list(fnscale = -1)) res$convergence ## [1] 0 est_mle &lt;- res$par names(est_mle) &lt;- c(&quot;p_azar_logit&quot;, &quot;p_corr_logit&quot;) est_mle ## p_azar_logit p_corr_logit ## -0.9194029 -0.8896454 probs_mle &lt;- inv_logit(est_mle) names(probs_mle) &lt;- c(&quot;p_azar&quot;, &quot;p_corr&quot;) probs_mle ## p_azar p_corr ## 0.2850796 0.2911830 En primer lugar, parece ser que nuestras estimaciones son menos precias. Vamos a hacer bootstrap paramétrico: rep_boot &lt;- function(rep, simular, crear_log_p, pars, n){ muestra_bootstrap &lt;- simular(n, pars) log_p_boot &lt;- crear_log_p(muestra_bootstrap) # optimizamos res_boot &lt;- optim(c(0.0, 0.0), log_p_boot, control = list(fnscale = -1)) try(if(res_boot$convergence != 0) stop(&quot;No se alcanzó convergencia.&quot;)) est_mle_boot &lt;- res_boot$par names(est_mle_boot) &lt;- names(pars) est_mle_boot[&quot;rep&quot;] &lt;- rep est_mle_boot[&quot;convergence&quot;] &lt;- res_boot$convergence est_mle_boot } set.seed(8934) reps_boot &lt;- map(1:500, ~ rep_boot(.x, simular_modelo, crear_log_p, est_mle, n = length(muestra))) |&gt; bind_rows() reps_boot |&gt; mutate(across(everything(), round, 2)) |&gt; head() ## # A tibble: 6 × 4 ## p_azar_logit p_corr_logit rep convergence ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 -1.14 -0.92 1 0 ## 2 -1.11 -0.85 2 0 ## 3 -1.17 -0.95 3 0 ## 4 -2.74 -1.01 4 0 ## 5 -1.05 -0.93 5 0 ## 6 -0.91 -0.87 6 0 El optimizador encontró resultados que no tienen sentido: ggplot(reps_boot, aes(x = inv_logit(p_azar_logit), y = inv_logit(p_corr_logit), colour = factor(convergence))) + geom_point(show.legend = FALSE) + xlab(&quot;p_azar&quot;) + ylab(&quot;p_corr&quot;) Y notamos un problema grave: Tenemos mucha variación en nuestros estimadores, y la correlación entre las estimaciones es alta. Esto deberíamos haberlo esperado, pues como las probabilidades de contestar correctamente son muy similares a las de contestar al azar: Existen muchas combinaciones de parámetros que son consistentes con los datos. Decimos entonces que este modelo está mal identificado con estos datos. La mala identificación, como vemos, es una propiedad tanto de modelo como datos. ¿Qué conclusiones acerca del examen obtienes al ver estas simulaciones bootstrap? ¿Cómo se deberían reportar estos resultados? Qué pasa en este ejemplo si la \\(p_{corr}\\) es más grande, o el tamaño de muestra es más grande? Repite el ejercicio con los parámetros del primer ejemplo (\\(p_{azar} = 0.3, p_{corr}=0.75\\)) y el mismo tamaño de muestra. ¿Qué sucede en este caso? En el caso extremo, decimos que el modelo no está indentificado, y eso generalmente sucede por un problema en el planteamiento del modelo, independientemente de los datos. ¿Puedes imaginar un modelo así? "],["propiedades-teóricas-de-mle.html", "Sección 8 Propiedades teóricas de MLE Consistencia Equivarianza del \\(\\textsf{MLE}\\) Normalidad asintótica Optimalidad del \\(\\textsf{MLE}\\)", " Sección 8 Propiedades teóricas de MLE El método de máxima verosimiltud es uno de los métodos más utilizados en la inferencia estadística paramétrica. En esta sección estudiaremos las propiedades teóricas que cumplen los estimadores de máxima verosimilitud (\\(\\textsf{MLE}\\)) y que han ayudado en su casi adopción universal. Estas propiedades de los \\(\\textsf{MLE}\\) son válidas siempre y cuando el modelo \\(f(x; \\theta)\\) satisfaga ciertas condiciones de regularidad. En particular veremos las condiciones para que los estimadores de máxima verosimilitud sean: consistentes, asintóticamente normales, asintóticamente insesgados, asintóticamente eficientes, y equivariantes. Los estimadores \\(\\textsf{MLE}\\) en ocasiones son malinterpretados como una estimación puntual en la inferencia, y por ende, incapaces de cuantificar incertidumbre. A lo largo de estas notas hemos visto cómo extraer intervalos de confianza por medio de simulación y por lo tanto incorporar incertidumbre en la estimación. Sin embargo, hay otros maneras de reportar incertidumbre para \\(\\textsf{MLE}\\). Y hablaremos de ello en esta sección. A lo largo de esta sección asumiremos muestras de la forma \\[\\begin{align} X_1, \\ldots, X_n \\overset{\\text{iid}}{\\sim} f(x; \\theta^*), \\end{align}\\] donde \\(\\theta^*\\) es el valor verdadero —que suponemos desconocido pero fijo— del parámetro \\(\\theta \\in \\Theta\\), y sea \\(\\hat \\theta_n\\) el estimador de máxima verosimilitud de \\(\\theta.\\) Ejemplo Usaremos este ejemplo para ilustrar los diferentes puntos teóricos a lo largo de esta sección. Consideremos el caso de una muestra de variables binarias que registran el éxito o fracaso de un experimento. Es decir, \\(X_1, \\ldots, X_n \\sim \\textsf{Bernoulli}(p),\\) donde el párametro desconocido es el procentaje de éxitos. Éste último denotado por \\(p.\\) Este ejemplo lo hemos estudiado en secciones pasadas, y sabemos que el \\(\\textsf{MLE}\\) es \\[\\begin{align} \\hat p_n = \\frac{S_n}{n} = \\bar X_n, \\end{align}\\] donde \\(S_n= \\sum_i X_i\\) es el número total de éxitos en la muestra. La figura siguiente ilustra el estimador \\(\\hat p_n\\) como función del número de observaciones en nuestra muestra. Podemos apreciar cómo el promedio parece estabilizarse alrededor del verdadero valor de \\(p^* = 0.25\\) cuando tenemos una cantidad suficientemente grande de observaciones. Como es de esperarse, diferentes muestras tendrán diferentes valores de \\(n\\) dónde las trayectorias parezca que se haya estabilizado (Ver figura siguiente). Sin embargo, se puede notar que este comportamiento parece estar controlado y son raras las trayectorias que se encuentran más lejos. Los conceptos siguientes nos permitirán cuantificar el porcentaje de trayectorias que se mantienen cercanas a \\(p^*,\\) en el caso límite de un número grande de observaciones, cuando trabajemos con estimadores de máxima verosimilitud. Más aún, nos permitirán cracterizar la distribución para dicho límite y aprenderemos de otras propiedades bajo este supuesto asintótico. Consistencia Es prudente pensar que para un estimador, lo que nos interesa es que conforme más información tengamos, más cerca esté del valor desconocido. Esta propiedad la representamos por medio del concepto de consistencia. Para hablar de esta propiedad necesitamos definir un tipo de convergencia para una secuencia de variables aleatorias, convergencia en probabilidad. Definición. Una sucesión de variables aleatorias \\(X_n\\) converge en probabilidad a la variable aleatoria \\(X,\\) lo cual denotamos por \\(X_n \\overset{P}{\\rightarrow} X\\), si para toda \\(\\epsilon \\gt 0\\), \\[\\lim_{n \\rightarrow \\infty} \\mathbb{P}(|X_n - X| &gt; \\epsilon) = 0.\\] Ahora, definimos un estimador consistente como: Definición. Un estimador \\(\\tilde \\theta_n\\) es consistente si converge en probabilidad a \\(\\theta^*.\\) Donde \\(\\theta^*\\) denota el verdadero valor del parámetro, que asumimos fijo. En particular, los estimadores \\(\\textsf{MLE}\\) son consistentes. Teorema. Sea \\(X_n \\sim f(X; \\theta^*),\\) una muestra iid, tal que \\(f(X; \\theta)\\) cumple con ciertas condiciones de regularidad. Entonces, \\(\\hat \\theta_n,\\) el estimador de máxima verosimilitud, converge en probabilidad a \\(\\theta^*.\\) Es decir, \\(\\hat \\theta_n\\) es consistente. La demostración de este teorema la pueden encontrar en Wasserman (2013). Ejemplo El estimador \\(\\hat p_n\\) es consistente. Esto quiere decir que el estimador se vuelve más preciso conforme obtengamos más información. En general esta es una propiedad que los estimadores deben satisfacer para ser útiles en la práctica. La figura siguiente muestra el estimador \\(\\hat p_n\\) como función del número de observaciones utilizado. Distintas curvas corresponden a distintas realizaciones de muestras obtenidas del modelo (\\(B = 500\\)). Nota que la banda definida por \\(\\epsilon\\) se puede hacer tan pequeña como se requiera, lo único que sucederá es que necesitaremos un mayor número de observaciones para garantizar que las trayectorias de los estimadores \\(\\hat p_n\\) se mantengan dentro de las bandas con alta probabilidad. Equivarianza del \\(\\textsf{MLE}\\) Muchas veces nos interesa reparametrizar la función de verosimilitud con el motivo de simplificar el problema de optimización asociado, o simplemente por conveniencia interpretativa. Por ejemplo, si el parámetro de interés es tal que \\(\\theta \\in [a, b],\\) entonces encontrar el \\(\\textsf{MLE}\\) se traduce en optimizar la log-verosimilitud en el espacio restringido al intervalo \\([a,b].\\) En este caso, los métodos tradicionales de búsqueda local por descenso en gradiente podrían tener problemas de estabilidad cuando la búsqueda se realice cerca de las cotas. El concepto de equivarianza nos dice que si el cambio de coordenadas parametrales está definida, y si este cambio de variable se realiza por medio de una función bien comportada (derivable y cuya derivada no es cero), entonces la solución de encontrar el \\(\\textsf{MLE}\\) en las coordenadas originales y transformar, es igual a realizar la inferencia en las coordenadas fáciles. Teorema. Sea \\(\\tau = g(\\theta)\\) una función de \\(\\theta\\) bien comportada. Entonces si \\(\\hat \\theta_n\\) es el \\(\\textsf{MLE}\\) de \\(\\theta,\\) entonces \\(\\hat \\tau_n = g(\\hat \\theta_n)\\) es el \\(\\textsf{MLE}\\) de \\(\\tau.\\) Ejemplo El concepto de equivarianza lo ilustraremos para nuestro ejemplo de esta sección. En particular la parametrización la realizamos por cuestiones de interpretación como un factor de riesgo. Como hemos visto estimador \\(\\hat p_n\\) es equivariante. Es importante mencionar que esta propiedad es general para cualquier tamaño de muestra. Es decir, no descansa en supuestos de muestras grandes. Supongamos que nos interesa estimar el momio de éxitos (bastante común en casas de apuestas). El momio está definido como \\[ \\theta = \\frac{p}{1-p},\\] y podemos rescribir la función de verosimilitud en términos de este parámetro. Sustituyendo \\(p = \\frac{\\theta}{1+\\theta}\\) en \\(\\mathcal{L}_n(p)\\) obtenemos \\[\\begin{align} \\mathcal{L}_n(\\theta) = \\left( \\frac{\\theta}{1 + \\theta} \\right)^{S_n} \\left(\\frac{1}{1 + \\theta} \\right)^{n - S_n}, \\end{align}\\] cuya función encuentra su máximo en \\[\\begin{align} \\hat \\theta_n = \\frac{\\bar X_n}{ 1 - \\bar X_n}. \\end{align}\\] Comprueba que el estimador de arriba para \\(\\theta\\) es el MLE. Normalidad asintótica Está propiedad nos permite caracterizar la distribución asintótica del MLE. Es decir, nos permite caracterizar la incertidumbre asociada una muestra suficientemente grande por medio de una distribución Gaussiana. Esto es, bajo ciertas condiciones de regularidad, \\[\\hat \\theta_n \\overset{.}{\\sim} \\mathsf{N}( \\theta^*, \\mathsf{ee}^2),\\] donde \\(\\mathsf{ee}\\) denota el error estándar del \\(\\textsf{MLE},\\) \\(\\mathsf{ee} = \\mathsf{ee}(\\hat \\theta_n) = \\sqrt{\\mathbb{V}(\\hat \\theta_n)}\\). Esta distribución se puede caracterizar de manera aproximada por métodos analíticos. Para esto necesitamos las siguientes definiciones. Definición. La función de score está definida como \\[\\begin{align} s(X; \\theta) = \\frac{\\partial \\log f(X; \\theta)}{\\partial \\theta}. \\end{align}\\] La información de Fisher está definida como \\[\\begin{align} I_n(\\theta) &amp;= \\mathbb{V}\\left( \\sum_{i = 1}^ns(X_i; \\theta) \\right) \\\\ &amp;= \\sum_{i = 1}^n \\mathbb{V} \\left(s(X_i; \\theta) \\right) \\end{align}\\] Estas cantidades nos permiten evaluar qué tan fácil será identificar el mejor modelo dentro de la familia parámetrica \\(f(X; \\theta)\\). La función de score nos dice qué tanto cambia locamente la distribución cuando cambiamos el valor del parámetro. Calcular la varianza, nos habla de la dispersión de dicho cambio a lo largo del soporte de la variable aleatoria \\(X.\\) Si \\(I_n(\\theta)\\) es grande entonces el cambio de la distribución es muy importante. Esto quiere decir que la distribución es muy diferente de las distribuciones cercanas que se generen al evaluar en \\(\\theta\\)s diferentes. Por lo tanto, si \\(I_n(\\theta)\\) es grande, la distribución será fácil de identificar cuando hagamos observaciones. La información de Fisher también nos permite caracterizar de forma analítica la varianza asíntotica del \\(\\textsf{MLE}\\) pues la aproximación \\(\\mathsf{ee}^2 \\approx \\frac{1}{I_n(\\theta^*)}\\) es válida. El siguiente resultado utiliza la propiedad de la función de score: \\(\\mathbb{E}[s(X; \\theta)] = 0,\\) que implica que \\(\\mathbb{V} \\left(s(X_i; \\theta) \\right) = \\mathbb{E}[s^2(X; \\theta)],\\) y permite a su vez un cómputo más sencillo de la información de Fisher. Teorema. El cálculo de la información de Fisher para una muestra de tamaño \\(n\\) se puede calcular de manera simplificada como \\(I_n(\\theta) = n \\, I(\\theta).\\) Por otro lado, tenemos la siguiente igualdad \\[ I(\\theta) = - \\mathbb{E}\\left( \\frac{\\partial^2 \\log f(X; \\theta)}{\\partial \\theta^2} \\right).\\] Con estas herramientas podemos formular el teorema siguiente. Teorema. Bajo ciertas condiciones de regularidad se satisface que \\(\\mathsf{ee} \\approx \\sqrt{1/I_n(\\theta^*)}\\) y \\[ \\hat \\theta_n \\overset{d}{\\rightarrow} \\mathsf{N}( \\theta^*, \\mathsf{ee}^2).\\] El resultado anterior es teóricamente interesante y nos asegura un comportamiento controlado conforme tengamos más observaciones disponibles. Sin embargo, no es práctico pues no conocemos \\(\\theta^*\\) en la práctica y por consiguiente no conoceríamos la varianza. Sin embargo, también podemos aplicar el principio de plug-in y caracterizar la varianza de la distribución asintótica por medio de \\[\\hat{\\mathsf{ee}} = \\sqrt{1/I_n(\\hat \\theta_n)}.\\] Esto último nos permite constuir intervalos de confianza, por ejemplo al 95%, a través de \\[ \\hat \\theta_n \\pm 2 \\, \\hat{\\mathsf{ee}}.\\] Asimismo, el teorema de Normalidad asintótica nos permite establecer que el \\(\\textsf{MLE}\\) es asíntoticamente insesgado. Es decir, \\[\\lim_{n \\rightarrow \\infty}\\mathbb{E}[\\hat \\theta_n] = \\theta^*.\\] Definición. Sea una muestra \\(X_1, \\ldots, X_n \\overset{iid}{\\sim} f(X; \\theta^*)\\). Un estimador \\(\\tilde \\theta_n\\) es insesgado si satisface que \\[\\mathbb{E}[\\tilde \\theta_n] =\\theta^*.\\] El sesgo del estimador es precisamente la diferencia: \\(\\textsf{Sesgo} = \\mathbb{E}[\\tilde \\theta_n] - \\theta^*.\\) Ejemplo: Información de Fisher En el caso Bernoulli obtenemos \\(I_n(\\theta) = \\frac{n}{\\theta(1-\\theta)}\\), si \\(n = 20\\) podemos comparar con \\(\\theta=0.5, 0.7, 0.8\\), library(patchwork) # Verosimilitud X_1,...,X_n ~ Bernoulli(theta) L_bernoulli &lt;- function(n, S){ function(theta){ theta ^ S * (1 - theta) ^ (n - S) } } xy &lt;- data.frame(x = 0:1) l_b1 &lt;- ggplot(xy, aes(x = x)) + stat_function(fun = L_bernoulli(n = 20, S = 10)) + xlab(expression(theta)) + ylab(expression(L(theta))) + labs(title = &quot;Verosimilitud&quot;, subtitle = &quot;n=20, S = 10&quot;) + ylim(0, 5e-05) l_b2 &lt;- ggplot(xy, aes(x = x)) + stat_function(fun = L_bernoulli(n = 20, S = 14)) + xlab(expression(theta)) + ylab(expression(L(theta))) + labs(title = &quot;Verosimilitud&quot;, subtitle = &quot;n=20, S = 14&quot;) + ylim(0, 5e-05) l_b3 &lt;- ggplot(xy, aes(x = x)) + stat_function(fun = L_bernoulli(n = 20, S = 16)) + xlab(expression(theta)) + ylab(expression(L(theta))) + labs(title = &quot;Verosimilitud&quot;, subtitle = &quot;n=20, S = 19&quot;) + ylim(0, 5e-05) l_b1 + l_b2 + l_b3 Ejemplo: Normalidad Regresando a nuestro ejemplo. Veremos empiricamente que el estimador \\(\\hat \\theta_n\\) es asintóticamente normal. Esta propiedad la hemos visto anteriormente para un caso muy particular. Lo vimos en el TLC para el caso de promedios, \\(\\bar X_n,\\) que en nuestro ejemplo corresponde a \\(\\hat p_n\\). Como hemos visto, esta propiedad la satisface cualquier otro estimador que sea máximo verosímil. Por ejemplo, podemos utilizar el \\(\\mathsf{MLE}\\) de los momios. La figura que sigue muestra la distribución de \\(\\hat \\theta_n\\) para distintas remuestras \\((B = 500)\\) con distintos valores de \\(n.\\) El gráfico anterior valida empíricamente la distribución asintótica para casos de muchas observaciones. A continuación ilustraremos cómo explotar este resultado para obtener intervalos de confianza. Para el caso de \\(\\hat p_n\\) hemos visto que el error estándar se calcula analíticamente como \\[\\textsf{ee}_p^2 = \\mathbb{V}(\\hat p_n) = \\mathbb{V}\\left(\\frac1n \\sum_{i = 1}^n x_i\\right) = \\frac{p^* (1 - p^*)}{n}.\\] Éste empata con el valor del error estándar asintótico \\[\\textsf{ee}_p^2 \\approx \\sqrt{\\frac{1}{I_n(p^*)}},\\] pues la información de Fisher es igual a \\[I_n(p) = n \\, I(p) = \\frac{n}{p ( 1- p)}.\\] En este caso podemos utilizar el estimador plug-in, \\(\\hat{\\textsf{ee}}_p = \\textsf{ee}_p(\\hat p_n).\\) Para estimar el momio, \\(\\theta,\\) el cálculo no es tan fácil pues tendríamos que calcular de manera analítica la varianza de un cociente \\[\\textsf{ee}_\\theta^2 = \\mathbb{V}\\left( \\frac{\\hat p_n}{1-\\hat p_n}\\right).\\] Utilizando la distirbución asintótica, el error estándar se puede calcular mediante \\[\\textsf{ee}_\\theta \\approx \\sqrt{\\frac{1}{I_n(\\theta^*)}} = \\sqrt{\\frac{\\theta (1 + \\theta)^2 }{n}}.\\] A continuación mostramos los errores estándar para nuestro ejemplo utilizando la distribución asintótica y por medio de la distribución de bootstrap. Como es de esperarse, ambos coinciden para muestras relativamente grandes. # Genero muestra muestras &lt;- tibble(tamanos = 2**seq(4,7)) %&gt;% mutate(obs = map(tamanos, ~rbernoulli(., p = p_true))) calcula_momio &lt;- function(x){ x / (1 - x) } calcula_ee_momio &lt;- function(x){ sqrt(((1+x)**2) * x) } # Calculo MLE muestras_est &lt;- muestras %&gt;% group_by(tamanos) %&gt;% mutate(media_hat = map_dbl(obs, mean), media_ee = sqrt(media_hat * (1 - media_hat)/tamanos), momio_hat = calcula_momio(media_hat), momio_ee = calcula_ee_momio(momio_hat)/sqrt(tamanos)) # Calculo por bootstrap muestras_boot &lt;- muestras_est %&gt;% group_by(tamanos) %&gt;% mutate(sims_muestras = map(tamanos, ~rerun(1000, sample(muestras %&gt;% filter(tamanos == ..1) %&gt;% unnest(obs) %&gt;% pull(obs), size = ., replace = TRUE))), sims_medias = map(sims_muestras, ~map_dbl(., mean)), sims_momios = map(sims_medias, ~map_dbl(., calcula_momio)), media_boot = map_dbl(sims_medias, mean), momio_boot = map_dbl(sims_momios, mean), media_ee_boot = map_dbl(sims_medias, sd), momio_ee_boot = map_dbl(sims_momios, sd) ) ## # A tibble: 4 × 5 ## # Groups: tamanos [4] ## tamanos momio_hat momio_boot momio_ee momio_ee_boot ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 16 0.333 0.367 0.192 0.236 ## 2 32 0.333 0.342 0.136 0.140 ## 3 64 0.123 0.123 0.0492 0.0498 ## 4 128 0.407 0.417 0.0793 0.0800 Comprueba las fórmulas para los errores estándar tanto para la probabilidad de éxito como para los momios. El método delta El ejercicio anterior nos sugiere una pregunta natural: Cómo establecer la distribución asintótica de un estimador cuando ya se conoce la de una pre-imagen de él? Es decir, si ya conocemos la distribución de \\(\\theta,\\) podemos establecer la distribución de \\(\\tau = g(\\theta)?\\) La respuesta es afirmativa y la enunciamos por medio de un teorema. El resultado se conoce como el método delta. Teorema. Si \\(\\tau = g(\\theta)\\) es una función diferenciable y \\(g&#39;(\\theta) \\neq 0\\), entonces \\[\\hat \\tau_n \\overset{d}{\\rightarrow} \\mathsf{N}( \\tau^*, \\hat{\\mathsf{ee}}^2_\\tau),\\] donde \\(\\hat \\tau_n = g(\\hat \\theta_n)\\) y \\[\\hat{\\mathsf{ee}}_\\tau = \\bigg| g&#39;(\\hat \\theta_n) \\bigg| \\times \\hat{\\mathsf{ee}}_\\theta(\\hat \\theta_n).\\] Por ejemplo, este resultado lo podemos utilizar para nuestro experimento de Bernoullis. Pues \\(g(p) = \\frac{p}{1-p}\\) es una función diferenciable y por lo tanto \\[\\hat{\\mathsf{ee}}_\\theta = \\sqrt{\\frac1n} \\times \\left[ \\hat p_n^{1/2} (1-\\hat p_n)^{-3/2}\\right].\\] Comprueba la fórmula del método delta para el momio en función de la fracción de éxitos, y también comprueba que de el mismo resultado analítico que habías calculado en el ejercicio anterior. Optimalidad del \\(\\textsf{MLE}\\) Consideremos el caso de una muestra iid \\(X_1, \\ldots, X_n \\sim \\mathsf{N}(\\theta, \\sigma^2).\\) Y consideremos dos estimadores para \\(\\theta.\\) El primero será la media muestral \\(\\bar X_n\\) y el segundo la mediana muestral, la cual denotaremos por \\(\\tilde \\theta_n.\\) Sabemos que ambos son insesgados. Por lo tanto, en promedio emiten estimaciones correctas. Pero ¿cómo escogemos cual utilizar? Un criterio para comparar estimadores es el error cuadrático medio (\\(\\textsf{ECM}\\), por sus siglas en inglés). Definición. El error cuadrático medio de un estimador \\(\\tilde \\theta_n\\) se calcula como \\[\\textsf{ECM}[\\tilde \\theta_n] = \\mathbb{E}[(\\tilde \\theta_n - \\theta^*)^2].\\] Por lo tanto, el \\(\\textsf{ECM}\\) mide la distancia promedio entre el estimador y el valor verdadero valor del parámetro. La siguiente igualdad es bastante útil para comparar dos estimadores. \\[\\textsf{ECM}[\\tilde \\theta_n] = \\mathbb{V}\\left(\\tilde \\theta_n\\right) + \\textsf{Sesgo}\\left[\\tilde \\theta_n\\right]^2.\\] Por lo tanto si dos estimadores son insesgados, uno es más eficiente que el otro si su varianza es menor. La media sabemos que es el \\(\\textsf{MLE}\\) y por el TCL tenemos que \\[\\sqrt{n} \\left( \\bar X_n - \\theta \\right) \\overset{d}{\\rightarrow} \\mathsf{N}( 0, \\sigma^2).\\] La mediana, en contraste, tiene una distribución asintótica \\[\\sqrt{n} \\left( \\tilde X_n - \\theta \\right) \\overset{d}{\\rightarrow} \\mathsf{N}\\left( 0, \\sigma^2 \\frac{\\pi}{2}\\right),\\] es decir tiene una varianza ligeramente mayor. Por lo tanto, decimos que la mediana tiene una eficiencia relativa con respecto a la media del \\(.63 \\% (\\approx 2/\\pi)\\). Es decir, la mediana sólo utliza una fracción de los datos comparado con la media. El siguiente teorema, la desigualdad de Cramer-Rao, nos permite establecer esta resultado de manera mas general para cualquier estimador insesgado. Teorema. Sea \\(\\tilde \\theta_n\\) cualquier estimador insesgado de \\(\\theta\\) cuyo valor verdadero es \\(\\theta^*,\\) entonces \\[\\begin{align} \\mathbb{V}(\\tilde \\theta_n) \\geq \\frac{1}{n I(\\theta^*)}. \\end{align}\\] Un estimador insesgado que satisfaga esta desigualdad se dice que es eficiente. Nota que el lado derecho de la desigualdad es precisamente la varianza asintótica del \\(\\textsf{MLE}.\\) Por lo tanto, éste es asintóticamente eficiente. Es importante hacer enfásis en que la optimalidad del \\(\\textsf{MLE}\\) es un resultado asintótico. Es decir, sólo se satisface cuando tenemos un número suficiente de observaciones. Qué tan grande debe ser el tamaño de muestra varia de problema a problema. Es por esto que para muestras de tamaño finito se prefieren estimadores que minimicen el \\(\\textsf{ECM},\\) como cuando hacemos regresión ridge o utilizamos el estimador James–Stein para un vector de medias. Referencias "],["más-de-pruebas-de-hipótesis-e-intervalos.html", "Sección 9 Más de pruebas de hipótesis e intervalos Prueba de Wald Observación: pruebas \\(t\\) y práctica estadística Prueba de Wald para dos medias o proporciones Datos pareados Pruebas de cociente de verosimilitud Otro tipo de pruebas Errores tipo I y tipo II Consideraciones prácticas Pruebas múltiples", " Sección 9 Más de pruebas de hipótesis e intervalos En esta sección veremos enfoques más clásicos para analizar una prueba de hipótesis, en particular veremos situaciones donde podemos hacer algunos supuestos teóricos acerca de la distribución de las poblaciones. Esta es una sección complementaria para entender prácticas estadísticas usuales: recuerda que discutimos antes que hacer estimación por intervalos generalmente es más útil que hacer pruebas de hipótesis, y adicionalmente, tenemos también la técnica de pruebas de permutaciones que podemos aplicar en muchos de los casos que discutiremos a continuación. El enfoque básico es el mismo que cuando vimos pruebas de permutaciones: calculamos una estadística de prueba de los datos y luego, con una distribución de referencia (asociada a la hipótesis nula), calculamos un valor-\\(p\\). Si el valor-\\(p\\) es chico, entonces los resultados observados no pueden explicarse fácilmente por variación muestral, y rechazamos la hipótesis nula. Con esta idea básica, y supuestos distribucionales acerca de las poblaciones, podemos construir pruebas que requieren menos cómputo. La desventaja es que hay que checar con cuidado los supuestos distribucionales que hagamos. Si los supuestos son incorrectos, las valores-\\(p\\) no tienen mucho sentido y son difíciles de interpretar. Para esta sección seguiremos más a Wasserman (2013) (capítulo 10), pero puedes revisar también Chihara and Hesterberg (2018) (capítulo 8). Prueba de Wald Como hemos visto, existe normalidad asintótica en varios estimadores que hemos considerado, como medias y proporciones muestrales. También vimos que estimadores de máxima verosimilitud cumplen muchas veces un teorema central del límite. Así que supongamos que tenemos una estadística \\(\\hat{\\theta}_n\\) que estima \\(\\theta\\) y es asintóticamente insesgada y normal. Denotamos por \\(\\hat{\\textsf{ee}}\\) una estimación de su error estándar —hay distintas maneras de hacerlo: por ejemplo, con simulación (bootstrap), o por medios analíticos (teoría). Recuerda que el error estándar de una estadística es la desviación estándar de su distribución de muestreo. Si nos interesa probar la hipótesis de que \\(\\theta = 125\\), por ejemplo, y \\(\\hat{\\theta}_n\\) es aproximadamente normal, entonces podemos construir una distribución de referencia aproximada como sigue: Si la nula es cierta, entonces la distribución de muestreo de \\(\\hat{\\theta}\\) es aproximadamente \\(\\mathsf{N}(125, \\hat{\\textsf{ee}})\\). Esto implica que la siguiente estadística \\(W\\) es aproximadamente normal estándar bajo la nula: \\[W = \\frac{\\hat{\\theta} - 125}{\\hat{\\textsf{ee}}} \\sim \\mathsf{N}(0,1)\\] Por lo que valores lejanos de \\([-2,2]\\), por ejemplo, dan evidencia en contra de la hipótesis nula. Como \\(W\\) no depende de ningún parámetro desconocido, podemos usarla como distribución de referencia para comparar el valor de \\(W\\) que obtuvimos en la muestra. Si observamos para nuestra muestra un valor \\(W=w\\) entonces, el valor-\\(p\\) (dos colas) de esta prueba es, aproximadamente, \\[\\mathsf{valor-}p \\approx P(|Z| &gt; |w|) = 2(1 - \\Phi(|w|))\\] donde \\(Z\\sim \\mathsf{N}(0,1)\\) y \\(\\Phi\\) es su función de distribución acumulada. Ejemplo: media muestral. La media nacional de las escuelas de enlace está alrededor de 454 (matemáticas en 6o. grado). Tomamos una muestra de 180 escuelas del Estado de México, y queremos saber si la media obtenida es consistente o no con la media nacional. Ya que estamos usando como estimador una media de una muestra iid, podemos estimar el error estándar de la media con \\[\\hat{\\textsf{ee}} = s / \\sqrt{n}\\] Obtenemos: set.seed(29) muestra_edomex &lt;- read_csv(&quot;data/enlace.csv&quot;) |&gt; filter(estado == &quot;ESTADO DE MEXICO&quot;) |&gt; sample_n(180) resumen &lt;- muestra_edomex |&gt; summarise(media = mean(mate_6), s = sd(mate_6), n = n()) |&gt; mutate(ee = s / sqrt(n)) resumen ## # A tibble: 1 × 4 ## media s n ee ## &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; &lt;dbl&gt; ## 1 456. 155. 180 11.5 La hipótesis nula es que la media poblacional del Estado de México es igual a 454. Calculamos el valor-\\(p\\) usando la prueba de Wald: dif &lt;- (resumen |&gt; pull(media)) - 454 ee &lt;- resumen |&gt; pull(ee) w &lt;- dif / ee p &lt;- 2 * (1 - pt(abs(w), 179)) p ## [1] 0.8413082 y vemos que esta muestra es consistente con la media nacional. No tenemos evidencia en contra de que la media del estado de México es muy similar a la nacional. Repite esta prueba con una muestra de Chiapas. ¿Qué resultado obtienes? Tenemos entonces: Prueba de Wald. Consideramos probar la hipótesis nula \\(H_0: \\theta = \\theta_0\\) contra la alternativa \\(H_1: \\theta \\neq \\theta_0\\). Suponemos que \\(\\hat{\\theta}_n\\) es asintóticamente normal e insesgada, de modo que bajo la hipótesis nula \\[\\frac{\\hat{\\theta}_n - \\theta_0}{\\hat{\\textsf{ee}}} \\sim \\mathsf{N}(0,1).\\] Entonces el valor-\\(p\\) de la prueba de Wald para esta hipótesis nula es \\[\\mathsf{valor-}p \\approx P(|Z| &gt; |w|) = 2(1 - \\Phi(|w|)).\\] Ejemplo. Podemos hacer la prueba de Wald para proporciones con el estimador usual \\(\\hat{p}_n\\) que estima una proporción poblacional \\(p\\). En este caso, utilizamos la estimación usual del error estándar de \\(\\hat{p}_n\\), que está dada por \\[\\hat{\\textsf{ee}} = \\sqrt{\\frac{\\hat{p}_n(1-\\hat{p}_n)}{n}}.\\] Supongamos por ejemplo que en nuestros datos observamos que en \\(n=80\\) muestras independientes, tenemos \\(x=47\\) éxitos. ¿Es esto consistente con la hipótesis nula \\(p = 0.5\\)? Calcuamos primero: p_hat &lt;- 47 / 80 ee &lt;- sqrt(p_hat * (1 - p_hat) / 80) y la estadística \\(W\\) de prueba es: w &lt;- (p_hat - 0.5) / ee w ## [1] 1.58978 Calculamos su valor p: valor_p &lt;- 2 * (1 - pnorm(abs(w))) valor_p ## [1] 0.1118843 Y vemos que en este caso tenemos evidencia baja de que la proporción poblacional es distinta de 0.5. Observación: pruebas \\(t\\) y práctica estadística Con más supuestos distribucionales podemos hacer otros tipos de pruebas donde no requerimos hacer supuestos asintóticos. Por ejemplo, si suponemos que la muestra obtenida \\(X_1,\\ldots, X_n\\) proviene de una distribución normal \\(\\mathsf{N}(\\mu, \\sigma)\\) (cosa que es necesario verificar), entonces es posible demostrar que la estadística \\[T = \\frac{\\bar{X} - \\mu}{S / \\sqrt{n}}\\] tiene una distribución exacta que es \\(t\\) de Student con \\(n-1\\) grados de libertad, y no depende de otros parámetros, de manera que podemos usarla como distribución de referencia y podemos calcular valores \\(p\\) exactos (revisa la sección 8.1 de Chihara and Hesterberg (2018)). La diferencia con usar una prueba de Wald está en que aquí consideramos también la variablidad del error estándar estimado, lo que correctamente sugiere que esperamos variaciones proporcionalmente más grandes en \\(T\\) comparado con lo que sucede si no consideramos esta variación (como en la prueba de Wald). Sin embargo: Si la muestra \\(n\\) es grande, la distribución \\(t\\) de Student con \\(n-1\\) grados de libertad es muy similar a la normal estándar, de manera que la aproximación de Wald es apropiada. Cuando la muestra \\(n\\) es chica, es difícil validar el supuesto de normalidad, a menos que tengamos alguna información adicional acerca de la distribución poblacional. La prueba tiene cierta robustez a desviaciones de normalidad de las observaciones, pero si el sesgo es muy grande, por ejemplo, el supuesto es incorrecto y da valores \\(p\\) distorsionados. Puedes ver aquí, o el apéndice B.11 de Chihara and Hesterberg (2018) para ver descripciones de la distribución \\(t\\) y cómo se compara con una normal estándar dependiendo de los grados de libertad. En muchas ocasiones, en la práctica es común no checar supuestos y saltar directamente a hacer pruebas \\(t\\), lo cual no es muy seguro. Si tenemos duda de esos supuestos, podemos hacer pruebas gráficas o de permutaciones, si son apropiadas. Prueba de Wald para dos medias o proporciones Cuando tenemos dos muestras extraidas de manera independiente de dos poblaciones distintas, y queremos ver si la hipótesis de medias poblacionales iguales es consistente con los datos, podemos usar también una prueba de Wald. Sea \\(\\bar{X}_1\\) y \\(\\bar{X}_2\\) las medias muestrales correspondientes. Si la hipótesis de normalidad aplica para ambas distribuciones muestrales (normalidad asintótica), la variable \\[\\hat{\\delta} = \\bar{X}_1 - \\bar{X}_2\\] es aproximadamente normal con media \\(\\mathsf{N}(\\mu_1 - \\mu_2, \\textsf{ee})\\), donde \\(\\mu_1\\) y \\(\\mu_2\\) son las medias poblacionales correspondientes, y donde el error estándar de \\(\\hat{\\delta}\\) es la raíz de la suma de los cuadrados de los errores estándar de \\(\\bar{X}\\) y \\(\\bar{Y}\\): \\[ \\textsf{ee} = \\sqrt{\\textsf{ee}_1^2 + \\textsf{ee}_{2}^2}.\\] Se sigue entonces que: \\[\\textsf{ee} =\\sqrt{\\frac{\\sigma_1^2}{n_1}+\\frac{\\sigma_2^2}{n_2} }\\] (Nota: usa probabilidad para explicar por qué es cierto esto). De esto se deduce que bajo la hipótesis nula de igualdad de medias \\(\\mu_1 = \\mu_2\\), tenemos que la estadística de Wald \\[W = \\frac{\\hat{\\delta} - 0}{\\sqrt{\\frac{s_1^2}{n_1}+\\frac{s_2^2}{n_2}} } \\sim \\mathsf{N}(0,1)\\] es aproximamente normal estándar. Procedemos entonces a calcular el valor \\(p\\) usando la función de distribución acumulada de la normal estándar. En el caso particular de proporciones, podemos simplificar, como hicimos arriba, a \\[W = \\frac{\\hat{p}_1 - \\hat{p}_2}{\\sqrt{\\frac{\\hat{p}_1(1-\\hat{p}_1)}{n_1}+\\frac{\\hat{p}_2(1-\\hat{p}_2)}{n_2}} } \\sim \\mathsf{N}(0,1)\\] Haz una prueba comparando las medias en enlace de la Ciudad de México vs Estado de México. ¿Hay evidencia de que tienen distintas medias? Ejemplo (Wasserman (2013)). Supongamos tenemos dos conjuntos de prueba para evaluar algoritmos de predicción, de tamaños \\(n_1=100\\) y \\(n_2=250\\) respectivamente, tenemos dos algoritmos para generar predicciones de clase (digamos positivo y negativo). Usaremos el primer conjunto para evaluar el algoritmo 1 y el segundo para evaluar el algoritmo 2. El algoritmo 1 corre en 1 hora, y el algoritmo 2 tarda 24 horas. Supón que obtenemos que la tasa de clasificación correcta del primer algoritmo es \\(\\hat{p}_1 = 0.85\\), y la tasa del segundo es de \\(\\hat{p}_2 = 0.91\\). ¿Estos datos son consistentes con la hipótesis de que los algoritmos tienen desempeño muy similar? Es decir, queremos probar la hipótesis \\(p_1 = p_2\\). Calculamos la estadística de Wald: n_1 &lt;- 100 n_2 &lt;- 250 p_hat_1 &lt;- 0.86 p_hat_2 &lt;- 0.90 ee &lt;- sqrt(p_hat_1 * (1 - p_hat_1) / n_1 + p_hat_2 * (1 - p_hat_2) / n_2) delta = p_hat_1 - p_hat_2 w &lt;- delta / ee w ## [1] -1.011443 que da un valor p de: 2 * (1 - pnorm(abs(w))) ## [1] 0.3118042 Y vemos que valor-\\(p\\) es grande, de forma que los datos son consistentes con la hipótesis de que los algoritmos tienen desempeño similar. ¿Cómo tomaríamos nuestra decisión final? Si la diferencia entre 1 hora y 24 horas no es muy importante, entonces preferíamos usar el algoritmo 2. Sin embargo, si el costo de 24 horas es más alto que 1 hora de corrida, los datos no tienen indicios fuertes de que vayamos a perder en desempeño, y podriamos seleccionar el algoritmo 1. Datos pareados Las pruebas que acabamos de ver para comparar medias requieren poblaciones independientes. Si las dos muestras están pareadas (es decir, son dos mediciones en una misma muestra), podemos tomar considerar las diferencias \\(D_i = X_i - Y_i\\) y utilizar la prueba para una sola muestra con la media \\(\\bar{D}\\). Esta es una prueba de Wald pareada. Ejemplo (Wasserman (2013)). Ahora supongamos que utilizamos la misma muestra de tamaño \\(n=300\\) para probar los dos algoritmos. En este caso, no debemos hacer la prueba para medias de muestras independientes. Sin embargo, esto podemos ponerlo en términos de una prueba para una sola muestra. Tenemos las observaciones \\(X_1,\\ldots, X_n\\) y \\(Y_1,\\dots, Y_n\\), donde \\(X_i=1\\) si el algoritmo 1 clasifica correctamente, y 0 en otro caso. Igualmente, \\(Y_i=1\\) si el algoritmo 2 clasifica correctamente, y 0 en otro caso. Definimos \\[D_i= X_i - Y_i\\] Y \\(D_1,\\ldots, D_n\\) es una muestra iid. Ahora observemos que la media \\(\\bar{D}\\) tiene valor esperado \\(p_1 - p_2\\), donde \\(p_1\\) y \\(p_2\\) son las tasas de correctos del algoritmo 1 y del algoritmo 2 respectivamente. Podemos hacer una prueba de Wald como al principio de la sección: \\[W = \\frac{\\bar{D} - 0}{{\\textsf{ee}}}\\] Y notemos que el error estándar no se calcula como en el ejemplo anterior. Podríamos usar bootstrap para estimarlo, pero en este caso podemos usar el estimador usual \\[\\hat{\\textsf{ee}} = S / \\sqrt{n}\\] donde \\[S = \\frac{1}{n}\\sum_{i=1}^n (D_i - \\bar{D})^2\\] y nótese que necesitamos las decisiones indiviudales de cada algoritmo para cada caso, en contraste al ejemplo anterior de muestras independientes donde los errores estándar se calculaban de manera independiente. Esto tiene sentido, pues la variablidad de \\(\\bar{D}\\) depende de cómo están correlacionados los aciertos de los dos algoritmos. Supongamos por ejemplo que los datos que obtenemos son: datos_clasif |&gt; head() ## # A tibble: 6 × 3 ## caso x y ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 1 1 ## 2 2 0 1 ## 3 3 0 1 ## 4 4 0 1 ## 5 5 0 1 ## 6 6 1 0 Como explicamos arriba, nos interesa la diferencia. Calculamos \\(d\\): datos_clasif &lt;- datos_clasif |&gt; mutate(d = x - y) datos_clasif |&gt; head() ## # A tibble: 6 × 4 ## caso x y d ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 1 1 0 ## 2 2 0 1 -1 ## 3 3 0 1 -1 ## 4 4 0 1 -1 ## 5 5 0 1 -1 ## 6 6 1 0 1 datos_clasif |&gt; summarise(sd_x = sd(x), sd_y = sd(y), sd_d = sd(d)) ## # A tibble: 1 × 3 ## sd_x sd_y sd_d ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 0.393 0.309 0.539 Y ahora calculamos la media de \\(d\\) (y tasa de correctos de cada clasificador:) medias_tbl &lt;- datos_clasif |&gt; summarise(across(where(is.numeric), mean, .names = &quot;{col}_hat&quot;)) d_hat &lt;- pull(medias_tbl, d_hat) medias_tbl ## # A tibble: 1 × 3 ## x_hat y_hat d_hat ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 0.81 0.893 -0.0833 Ahora necesitamos calcular el error estándar. Como explicamos arriba, hacemos ee &lt;- datos_clasif |&gt; mutate(d_hat = mean(d)) |&gt; mutate(dif_2 = (d - d_hat)) |&gt; summarise(ee = sd(dif_2) / sqrt(n())) |&gt; pull(ee) ee ## [1] 0.03112829 Y ahora podemos calcular la estadística \\(W\\) y el valor p correspondiente: w &lt;- d_hat / ee valor_p &lt;- 2 * (1 - pnorm(abs(w))) c(w = w, valor_p = valor_p) |&gt; round(3) ## w valor_p ## -2.677 0.007 Y vemos que tenemos evidencia considerable de que el desempeño no es el mismo: el algoritmo 2 parece ser mejor. ¿Qué pasaría si incorrectamente usaras la prueba de dos muestras para este ejemplo? ¿Qué cosa cambia en la fórmula de la estadística de Wald? Pruebas de cociente de verosimilitud Otra técnica clásica para hacer pruebas de hipótesis es el de cociente de verosimilitudes. Con esta técnica podemos hacer pruebas que involucren varios parámetros, y podemos contrastar hipótesis nulas contra alternativas especificas. Para aplicar este tipo de pruebas es necesario hacer supuestos distribucionales (modelos probabilísticos), pues estas pruebas se basan en la función de verosimilitud \\(\\mathcal{L}(\\theta; x_1,\\ldots, x_n)\\). Ejemplo. Supongamos que tenemos la hipótesis nula de que una moneda es justa (\\(p =0.5\\) de sol). En 120 tiros de la moneda (que suponemos independientes), observamos 75 soles. Recordemos la función de log-verosimilitud para el modelo binomial (ignorando constantes que no dependen de \\(p\\)) es \\[\\ell(p) = 75 \\log(p) + (120 - 75)\\log(1-p) \\] Primero calculamos el estimador de máxima verosimilitud de \\(p\\), que es \\(\\hat{p} = 75/120 = 0.625\\). Evaluamos la verosimilitud \\[\\ell(\\hat{p}) = \\ell(0.625) = 75\\log(0.625) + 45\\log(0.375) = -79.388\\] - Ahora evaluamos la verosimlitud según la hipótesis nula, donde asumimos que \\(p = 0.5\\): \\[\\ell(p_0) = \\ell(0.5) = 75\\log(0.5) + 45\\log(0.5) = -83.177\\] - Finalmente, contrastamos estos dos números con una estadística que denotamos con \\(\\lambda\\): \\[\\lambda = 2\\left[\\ell(\\hat{p}) - \\ell(p_0)\\right] = 2[\\ell(0.625)- \\ell(0.5)] = 2(3.79)=7.58\\] A \\(\\lambda\\) se le llama la estadística de cociente de verosimilitud. Tomamos la diferencia de log verosimilitudes, que es los mismo que tomar el logaritmo del cociente de verosimilitudes, y de ahí el nombre de la prueba. Nótese que cuando este número \\(\\lambda\\) es muy grande, esto implica que la hipótesis nula es menos creíble, o menos consistente con los datos, pues la nula tiene mucho menos verosimilitud de lo que los datos indican. Por otro lado, cuando este valor es cercano a 0, entonces tenemos menos evidencia en contra de la hipótesis nula. Esto se explica en la siguiente gráfica: log_verosim &lt;- function(p){ 75 * log(p) + (120 - 75) * log(1 - p) } verosim_tbl &lt;- tibble(p = seq(0.4, 0.7, 0.01)) |&gt; mutate(log_verosim = log_verosim(p)) ggplot(verosim_tbl, aes(x = p, y = log_verosim)) + geom_line() + geom_segment(x = 75/120, xend = 75/120, y = -130, yend = log_verosim(75/120), colour = &quot;red&quot;) + geom_segment(x = 0.5, xend = 0.5, y = -130, yend = log_verosim(0.5), colour = &quot;gray&quot;) + geom_errorbar(aes(x = 0.5, ymin = log_verosim(0.5), ymax = log_verosim(75/120)), colour = &quot;orange&quot;, width = 0.05) + annotate(&quot;text&quot;, x = 0.48, y = -81.5, label = &quot;3.79&quot;) + annotate(&quot;text&quot;, x = 0.515, y = -91, label =&quot;nula&quot;, colour = &quot;gray20&quot;) + annotate(&quot;text&quot;, x = 0.665, y = -91, label =&quot;max verosímil&quot;, colour = &quot;red&quot;) + labs(subtitle = expression(paste(lambda, &quot;=2(3.79)=7.58&quot;))) Este método puede generalizarse para que no solo aplique a hipótesis nulas donde \\(\\theta = \\theta_0\\), sino en general, \\(\\theta \\in \\Theta_0\\). Por ejemplo, podemos construir pruebas para \\(\\theta &lt; 0.4\\). Definición. Consideramos la hipótesis nula \\(\\theta= \\theta_0\\). La estadística del cociente de verosimilitudes está dada por: \\[\\lambda = 2\\log\\left( \\frac{\\max_{\\theta}\\mathcal{L}(\\theta)}{\\max_{\\theta=\\theta_0}\\mathcal{L}(\\theta)} \\right ) = 2\\log\\left( \\frac{\\mathcal{L}(\\hat{\\theta})}{\\mathcal{L}(\\theta_0)} \\right)\\] donde \\(\\hat{\\theta}\\) es el estimador de máxima verosimilitud. Para construir una prueba asociada, como siempre, necesitamos una distribución de referencia. Esto podemos hacerlo con simulación, o usando resultados asintóticos. Distribución de referencia para pruebas de cocientes Para nuestro ejemplo anterior, podemos simular datos bajo la hipótesis nula, y ver cómo se distribuye la estadística \\(\\lambda\\): Ejemplo. Simulamos bajo la hipótesis nula como sigue: n_volados &lt;- 120 # número de volados simulados_nula &lt;- rbinom(4000, n_volados, p = 0.5) lambda &lt;- function(n, x, p_0 = 0.5){ # estimador de max verosim p_mv &lt;- x / n # log verosimilitud bajo mv log_p_mv &lt;- x * log(p_mv) + (n - x) * log(1 - p_mv) # log verosimllitud bajo nula log_p_nula &lt;- x * log(p_0) + (n - x) * log(1 - p_0) lambda &lt;- 2*(log_p_mv - log_p_nula) lambda } lambda_obs &lt;- lambda(n_volados, 75, 0.5) sims_tbl &lt;- tibble(sim_x = simulados_nula) |&gt; mutate(lambda = map_dbl(sim_x, ~ lambda(n_volados, .x, p_0 = 0.5))) ggplot(sims_tbl, aes(x = lambda)) + geom_histogram(binwidth = 0.7) + geom_vline(xintercept = lambda_obs, color = &quot;red&quot;) Con esta aproximación a la distribución de referencia podemos calcular el valor p en nuestro ejemplo anterior: valor_p &lt;- mean(sims_tbl$lambda &gt;= lambda_obs) valor_p ## [1] 0.00675 y observamos que tenemos evidencia fuerte en contra de la hipótesis nula: la moneda no está balanceada. Ejemplo. Este ejemplo es un poco artificial, pero lo usamos para entender mejor las pruebas de cocientes de verosimlitud. Supongamos que tenemos una muestra de \\(\\mathsf{N}(\\mu, 1)\\), y queremos probar si \\(\\mu = 8\\). Asumimos que el supuesto de normalidad y desviación estándar iugal a 1 se cumplen. set.seed(3341) n_muestra &lt;- 100 muestra_1 &lt;- rnorm(n_muestra, 7.9, 1) crear_log_p &lt;- function(x){ # crear log verosim para dos muestras normales independientes. log_p &lt;- function(params){ mu &lt;- params[1] log_vero &lt;- dnorm(x, mean = mu, sd = 1, log = TRUE) |&gt; sum() log_vero } } lambda_calc &lt;- function(muestra, crear_log_p){ log_p &lt;- crear_log_p(muestra) res &lt;- optim(c(0), log_p, control = list(fnscale = -1)) lambda_mv &lt;- log_p(res$par) lambda_nula &lt;- log_p(8.0) lambda &lt;- 2 * (lambda_mv - lambda_nula) lambda } lambda &lt;- lambda_calc(muestra_1, crear_log_p) lambda ## [1] 2.101775 Ahora construimos con simulación la distribución de referencia usando simulaciones bajo la nula sims_nula &lt;- map(1:10000, ~ rnorm(n_muestra, 8, 1)) lambda_nula_sim &lt;- map_dbl(sims_nula, ~ lambda_calc(.x, crear_log_p)) tibble(lambda = lambda_nula_sim) |&gt; ggplot(aes(x = lambda)) + geom_histogram() + geom_vline(xintercept = lambda, colour = &quot;red&quot;) valor_p &lt;- mean(lambda_nula_sim &gt;= lambda) valor_p ## [1] 0.1537 Estos datos muestran consistencia con la hipótesis \\(\\mu = 8\\). Discusión: Nota en los dos ejemplos anteriores la similitud entre las distribuciones de referencia. En ambos casos, estas distribuciones resultan ser aproximadamente \\(\\chi\\)-cuadrada con 1 grado de libertad (ji-cuadrada). Podemos checar para el último ejemplo: teorica &lt;- tibble(x = seq(0.1, 10, 0.01)) |&gt; mutate(f_chi_1 = dchisq(x, df = 1)) tibble(lambda = lambda_nula_sim) |&gt; ggplot() + geom_histogram(aes(x = lambda, y = ..density..), binwidth = 0.1) + geom_line(data = teorica, aes(x = x, y = f_chi_1), colour = &quot;red&quot;) O mejor, con una gráfica de cuantiles de las simulaciones vs la téorica: tibble(lambda = lambda_nula_sim) |&gt; ggplot(aes(sample = lambda)) + geom_qq(distribution = stats::qchisq, dparams = list(df = 1)) + geom_qq_line(distribution = stats::qchisq, dparams = list(df = 1)) Este resultado asintótico no es trivial, y se usa comúnmente para calcular valores \\(p\\). Discutiremos más este punto más adelante. Otro tipo de pruebas Con cocientes de verosimlitud podemos diseñar pruebas para contrastar condiciones que sólo un subconjunto de parámetros cumple. Ejemplo. Supongamos que queremos hacer una prueba de igualdad de medias \\(\\mu_1 = \\mu_2\\) para dos poblaciones normales \\(\\mathsf{N}(\\mu_1, \\sigma_1)\\) y \\(\\mathsf{N}(\\mu_2, \\sigma_2)\\), donde extraemos las muestras de manera independiente, y no conocemos las desviaciones estándar. Obtenemos dos muestras (que supondremos provienen de distribuciones normales, pues ese es nuestro supuesto) set.seed(223) muestra_1 &lt;- rnorm(80, 0.8, 0.2) muestra_2 &lt;- rnorm(120, 0.8, 0.4) Necesitamos: 1) calcular el valor de la estadística \\(\\lambda\\) de cociente de verosimilitudes, 2) Calcular la distribución de referencia para \\(\\lambda\\) bajo la hipótesis nula y finalmente 3) Ver qué tan extremo es el valor obtenido de \\(\\lambda\\) en relación a la distribución de referencia. crear_log_p &lt;- function(x, y){ # crear log verosim para dos muestras normales independientes. log_p &lt;- function(params){ mu_1 &lt;- params[1] mu_2 &lt;- params[2] sigma_1 &lt;- params[3] sigma_2 &lt;- params[4] log_vero_1 &lt;- dnorm(x, mean = mu_1, sd = sigma_1, log = TRUE) |&gt; sum() log_vero_2 &lt;- dnorm(y, mean = mu_2, sd = sigma_2, log = TRUE) |&gt; sum() log_vero &lt;- log_vero_1 + log_vero_2 #se suman por independiencia log_vero } } log_p &lt;- crear_log_p(muestra_1, muestra_2) crear_log_p_nula &lt;- function(x, y){ log_p &lt;- function(params){ # misma media mu &lt;- params[1] sigma_1 &lt;- params[2] sigma_2 &lt;- params[3] log_vero_1 &lt;- dnorm(x, mean = mu, sd = sigma_1, log = TRUE) |&gt; sum() log_vero_2 &lt;- dnorm(y, mean = mu, sd = sigma_2, log = TRUE) |&gt; sum() log_vero &lt;- log_vero_1 + log_vero_2 #se suman por independiencia log_vero } } log_p_nula &lt;- crear_log_p_nula(muestra_1, muestra_2) Ahora tenemos el problema de que no conocemos las sigma. Estas deben ser estimadas para después calcular el cociente de verosimilitud: res &lt;- optim(c(0, 0, 1, 1), log_p, method = &quot;Nelder-Mead&quot;, control = list(fnscale = -1)) res$convergence ## [1] 0 est_mv &lt;- res$par names(est_mv) &lt;- c(&quot;mu_1&quot;, &quot;mu_2&quot;, &quot;sigma_1&quot;, &quot;sigma_2&quot;) est_mv ## mu_1 mu_2 sigma_1 sigma_2 ## 0.8153471 0.7819913 0.1987545 0.3940484 Y tenemos lambda_1 &lt;- log_p(est_mv) lambda_1 ## [1] -42.76723 Ahora calculamos el máximo bajo el supuesto de la hipótesis nula: res &lt;- optim(c(0, 1, 1), log_p_nula, method = &quot;Nelder-Mead&quot;, control = list(fnscale = -1)) res$convergence ## [1] 0 est_mv_nula &lt;- res$par names(est_mv) &lt;- c(&quot;mu&quot;, &quot;sigma_1&quot;, &quot;sigma_2&quot;) est_mv_nula ## [1] 0.8062091 0.1989438 0.3948603 y evaluamos lambda_2 &lt;- log_p_nula(est_mv_nula) lambda_2 ## [1] -43.07902 Finalmente, nuestra estadística \\(\\lambda\\) es lambda &lt;- 2 * (lambda_1 - lambda_2) lambda ## [1] 0.6235661 Y ahora necesitamos calcular un valor-\\(p\\). El problema que tenemos en este punto es que bajo la hipótesis nula no están determinados todos los parámetros, así que no podemos simular de manera simple muestras para obtener la distribución de referencia. Podemos sin embargo usar bootstrap paramétrico usando los estimadores de máxima verosimilitud bajo la nula simular_boot &lt;- function(n_1, n_2, est_mv_nula){ x &lt;- rnorm(n_1, est_mv_nula[1], est_mv_nula[2]) y &lt;- rnorm(n_2, est_mv_nula[1], est_mv_nula[3]) list(x = x, y = y) } lambda_nula_sim &lt;- function(est_mv_nula){ muestras &lt;- simular_boot(80, 120, est_mv_nula) x &lt;- muestras$x y &lt;- muestras$y log_p &lt;- crear_log_p(x, y) log_p_nula &lt;- crear_log_p_nula(x, y) est_1 &lt;- optim(c(0,0,1,1), log_p, control = list(fnscale = -1)) est_2 &lt;- optim(c(0,1,1), log_p_nula, control = list(fnscale = -1)) lambda &lt;- 2*(log_p(est_1$par) - log_p_nula(est_2$par)) lambda } lambda_sim &lt;- map_dbl(1:2000, ~ lambda_nula_sim(est_mv_nula = est_mv_nula)) Y graficamos la distribución de referencia junto con el valor de \\(\\lambda\\) que obtuvimos: tibble(lambda = lambda_sim) |&gt; ggplot(aes(x = lambda)) + geom_histogram() + geom_vline(xintercept = lambda, colour = &quot;red&quot;) Y claramente los datos son consistentes con medias iguales. El valor-\\(p\\) es mean(lambda_sim &gt; lambda) ## [1] 0.4275 Verificamos una vez más que la distribución de referencia es cercana a una \\(\\chi\\)-cuadrada con un grado de libertad. tibble(lambda = lambda_sim) |&gt; ggplot(aes(sample = lambda)) + geom_qq(distribution = stats::qchisq, dparams = list(df = 1)) + geom_qq_line(distribution = stats::qchisq, dparams = list(df = 1)) Esta es la definición generalizada de las pruebas de cociente de verosimilitudes Definición. Consideramos la hipótesis nula \\(\\theta \\in \\Theta_0\\). La estadística del cociente de verosimilitudes está dada por: \\[\\lambda = 2\\log\\left( \\frac{\\max_{\\theta}\\mathcal{L}(\\theta)}{\\max_{\\theta\\in\\Theta_0}\\mathcal{L}(\\theta)} \\right ) = 2\\log\\left( \\frac{ \\mathcal{L}(\\hat{\\theta})}{\\mathcal{L}(\\hat{\\theta}_0)} \\right)\\] donde \\(\\hat{\\theta}\\) es el estimador de máxima verosimilitud de \\(\\theta\\) y \\(\\hat{\\theta}_0\\) es el estimador de máxima verosimilitud de \\(\\theta\\) cuando restringimos a que \\(\\theta \\in \\Theta_0\\). En nuestro ejemplo anterior, el espacio \\(\\Theta_0\\) era \\(\\{ (\\mu,\\mu,\\sigma_1, \\sigma_2)\\}\\), que es un subconjunto de \\(\\{ (\\mu_1,\\mu_2,\\sigma_1, \\sigma_2)\\}\\). Nótese que el espacio \\(\\Theta_0\\) tiene tres parámetros libres, mientras que el espacio total tiene 4. Aunque podemos usar el bootstrap paramétrico para construir distribuciones de referencia para estas pruebas y calcular un valor-\\(p\\), el siguiente teorema, cuya demostración no es trivial, explica las observaciones que hicimos arriba. Este teorema enuncia la estrategia del enfoque clásico, que utiliza una aproximación asintótica. Valores p para pruebas de cocientes de verosimilitud. Supongamos que \\(\\theta = (\\theta_1,\\theta_2, \\ldots, \\theta_p)\\). Sea \\[\\Theta_0 = \\{\\theta : \\theta_1 = a_1, \\theta_2 = a_2, \\dots, \\theta_q = a_q \\},\\] es decir la hipótesis \\(\\theta \\in \\Theta_0\\) es que los primeros \\(q\\) parámetros de \\(\\theta\\) estan fijos en algún valor. Los otros parámetros no se consideran en esta prueba. Si \\(\\lambda\\) es la estadística de cociente de verosimilitudes de esta prueba, entonces, bajo la nula \\(\\theta \\in \\Theta_0\\) tenemos que la distribución de \\(\\lambda\\) es asintóticamente \\(\\chi\\)-cuadrada con \\(q\\) grados de libertad, denotada por \\(\\chi^2_q\\). El valor-\\(p\\) para esta prueba es \\[P(\\chi^2_{q} &gt; \\lambda)\\] Observaciones: Para hacer cálculos con la distribución \\(\\chi\\)-cuadrada usamos rutinas numéricas (por ejemplo la función pchisq en R). Nótese que \\(p\\) es la dimensión del espacio \\(\\Theta\\) (\\(p\\) parámetros), y que \\(p-q\\) es la dimensión del espacio \\(\\Theta_0\\) (pues \\(q\\) parámetros están fijos), de modo que los grados de libertad son la dimensión de \\(\\Theta\\) menos la dimensión de \\(\\Theta_0\\). En nuestro primer ejemplo (proporción de éxitos) solo teníamos un parámetro. El espacio \\(\\Theta_0\\) es de dimensión 0, así que los grados de libertad son \\(1 = 1 - 0\\) En este último ejemplo donde probamos igualdad de medias, el espacio \\(\\Theta\\) tiene dimensión 4, y el espacio \\(\\Theta_0\\) es de dimensión 3 (tres parámetros libres), por lo tanto los grados de libertad son \\(1 = 4 -3\\). Ejemplo En nuestro ejemplo de prueba de igualdad de medias, usaríamos pchisq(lambda, df =1, lower.tail = FALSE) ## [1] 0.4297252 que es similar al que obtuvimos con la estrategia del bootstrap paramétrico. Errores tipo I y tipo II En algunas ocasiones, en lugar de solamente calcular un valor-\\(p\\) queremos tomar una decisión asociada a distintas hipótesis que consideramos posibles. Por ejemplo, nuestra hipótesis nula podría ser Hipótesis nula \\(H_0\\): Una medicina nueva que estamos probando no es efectiva en reducir el colesterol en pacientes. Y queremos contrastar con una alternativa: Hipótesis alternativa \\(H_A\\): la medicina nueva reduce los niveles de colesterol en los pacientes. La decisión que está detrás de estas pruebas es: si no podemos rechazar la nula, la medicina no sale al mercado. Si rechazamos la nula, entonces la medicina es aprobada para salir al mercado. Para diseñar esta prueba, procedemos como sigue: Definimos cómo recolectar datos \\(X\\) de interés Definimos una estádistica \\(T(X)\\) de los datos. Definimos una región de rechazo \\(R\\) de valores tales que si \\(T(X)\\in R\\), entonces rechazaremos la hipótesis nula (e implícitamente tomaríamos la decisión asociada a la alternativa). Ejecutamos la prueba observando datos \\(X=x\\), calculando \\(T(x)\\), y checando si \\(T(x) \\in R\\). Si esto sucede entonces decimos que rechazamos la hipótesis nula, y tomamos la decisión asociada a la alternativa. Ejemplo. Si tenemos la hipótesis nula \\(p_1=0.5\\) para una proporción, y al alternativa es \\(p_1\\neq 0.5\\), podemos usar la estadística de Wald \\(T(x) = \\frac{\\hat{p_1} - 0.5}{\\hat{\\textsf{ee}}}\\). Podríamos definir la región de rechazo como \\(R = \\{T(x) : |T(x)| &gt; 3 \\}\\) (rechazamos si en valor absoluto la estadística de Wald es mayor que 3). Cuando diseñamos una prueba de este tipo, quisiéramos minimizar dos tipos de errores: Rechazar la hipótesis nula \\(H_0\\) cuando es cierta: Error tipo I No rechazar la hipótesis nula \\(H_0\\) cuando \\(H_0\\) es falsa: Error tipo II La gravedad de cada error depende del problema. En nuestro ejemplo de la medicina, por ejemplo: Un error tipo II resultaría en una medicina efectiva que no sale al mercado, lo que tiene consecuencias financieras (para la farmaceútica) y costos de oportunidad en salud (para la población). Por otra parte, Un error tipo I resultaría en salir al mercado con una medicina que no es efectiva. Esto tiene costos de oportunidad financieros que pueden ser grandes para la sociedad. Todos estos costos dependen, por ejempĺo, de qué tan grave es la enfermedad, qué tan costosa es la medicina, y así sucesivamente. En el enfoque más clásico, los errores tipo I y tipo II generalmente no se balancean según su severidad o probabilidad. En lugar de eso, generalmente se establece un límite para la probabilidad de cometer un error del tipo I (usualmente 5%, por una tradición que no tiene mucho fundamento) En vista de este ejemplo simple, y las observaciones de arriba: Reducir una decisión compleja a una prueba de hipótesis con resultados binarios (rechazar o no) es generalmente erróneo. Las pruebas de hipótesis se usan muchas veces incorrectamente cuando lo más apropiado es usar estimación por intervalos o algo similar que cuantifique la incertidumbre de las estimaciones. Consulta por ejemplo el comunicado de la ASA acerca de p-values y pruebas de hipótesis En el caso de la medicina, por ejemplo, realmente no nos interesa que la medicina sea mejor que un placebo. Nos importa que tenga un efecto considerable en los pacientes. Si estimamos este efecto, incluyendo incertidumbre, tenemos una mejor herramienta para hacer análisis costo-beneficio y tomar la decisión más apropiada. Como dijimos, típicamente se selecciona la región de rechazo de forma que bajo la hipótesis nula la probabilidad de cometer un error tipo I está acotada. Definición. Supongamos que los datos \\(X_1,X_2,\\ldots, X_n\\) provienen de una distribución \\(F_\\theta\\), donde no conocemos \\(\\theta\\). Supongamos que la hipótesis nula es que \\(\\theta = \\theta_0\\) (que llamamos una hipótesis simple). La función de potencia de una prueba con región de rechazo \\(R\\) se define como la probabilidad de rechazar para cada posible valor del parámetro \\(\\theta\\) \\[\\beta(\\theta) = P_\\theta (X\\in R).\\] El tamaño de una prueba se define como el valor \\[\\alpha = \\beta(\\theta_0),\\] es decir, la probabilidad de rechazar la nula (\\(\\theta = \\theta_0\\)) erróneamente. Observación. Esto se generaliza para hipótesis compuestas, donde la nula es que el parámetro \\(\\theta\\) está en un cierto conjunto \\(\\Theta_0\\). Por ejemplo, una hipótesis nula puede ser \\(\\theta &lt; 0.5\\). En este caso, \\(\\alpha\\) se define como el valor más grande que \\(\\beta(\\theta)\\) toma cuando \\(\\theta\\) está en \\(\\Theta_0\\), es decir, la probabilidad de rechazo más grande cuando la hipótesis nula se cumple. Decimos que una prueba tiene nivel de significancia de \\(\\alpha\\) si su tamaño es menor o igual a \\(\\alpha\\). Decimos que la potencia de una prueba es la probabilidad de, correctamente, rechazar la hipótesis nula cuando la alterna es verdadera: \\[\\beta(\\theta_a) = P_{\\theta_a} (X \\in R).\\] Observación: Sería deseable encontrar la prueba con mayor potencia bajo \\(H_a\\), entre todas las pruebas con tamaño \\(\\alpha\\). Esto no es trivial y no siempre existe. Observación: El valor \\(p\\) es el menor tamaño con el que podemos rechazar \\(H_0\\). Ejemplo (Chihara and Hesterberg (2018)) Supongamos que las calificaciones de Enlace de alumnos en México se distribuye aproximadamente como una normal con media 515 y desviación estándar de 120. En una ciudad particular, se quiere decidir si es neceario pedir fondos porque la media de la ciudad es más baja que la nacional. Nuestra hipótesis nula es \\(H_0: \\mu \\geq 515\\) y la alternativa es \\(\\mu &lt; 515\\), así que si rechazamos la nula se pedirían los fondos. Supondremos que la distribución de calificaciones en la ciudad es también aproximadamente normal con desviación estándar de 130. Se plantea tomar una muestra de 100 alumnos, y rechazar si la media muestral \\(\\bar{X}\\) es menor que 505. ¿Cuál es la probabilidad \\(\\alpha\\) de tener un error de tipo I? La función de potencia es \\[\\beta(\\mu) = P_\\mu(\\bar{X} &lt; 505)\\] Restando la media \\(\\mu\\) y estandarizando obtenemos \\[\\beta(\\mu) = P \\left (\\frac{\\bar{X} - \\mu}{130/\\sqrt{100}} &lt; \\frac{505 -\\mu}{130/\\sqrt{100}} \\right )\\] así que \\[\\beta(\\mu) = \\Phi \\left (\\frac{505 -\\mu}{130/\\sqrt{100}}\\right ),\\] donde \\(\\Phi\\) es la función acumulada de la normal estándar. La gráfica de la función potencia es entonces potencia_tbl &lt;- tibble(mu = seq(450, 550, 0.5)) |&gt; mutate(beta = pnorm((505 - mu)/13)) |&gt; # probabilidad de rechazar mutate(nula_verdadera = factor(mu &gt;= 515)) # nula verdadera ggplot(potencia_tbl, aes(x = mu, y = beta, colour = nula_verdadera)) + geom_line() Es decir, si la media \\(\\mu\\) de la ciudad es muy baja, con mucha seguridad rechazamos. Si es relativamente alta entonces no rechazamos. El tamaño de la prueba es el mayor valor de probabilidad de rechazo que se obtiene sobre los valores \\(\\mu\\geq 515\\) (la nula). Podemos calcularlo analíticamente como sigue: Si \\(\\mu \\geq 515\\), entonces \\[\\beta(\\mu) \\leq \\beta(515) = \\Phi\\left (\\frac{505 -515}{130/\\sqrt{100}}\\right ) = \\Phi( - 10 / 13) = \\Phi(-0.7692)\\] que es igual a pnorm(-0.7692) ## [1] 0.2208873 Y este es el tamaño de la prueba. En otras palabras: si la ciudad no está por debajo de la media nacional, hay una probabilidad de 22% de que erróneamente se pidan fondos (al rechazar \\(H_0\\)). Ejemplo Supongamos que los que programan el presupuesto deciden que se requiere tener una probabilidad de a lo más 5% de rechazar erróneamente la hipótesis nula (es decir, pedir fondos cuando en realidad su media no está debajo de la nacional) para poder recibir fondos. ¿Cuál es la región de rechazo que podríamos escoger? En el caso anterior usamos la región \\(\\bar{X}&lt;505\\). Si el tamaño de muestra está fijo en \\(n=100\\) (por presupuesto), entonces tenemos que escoger un punto de corte más extremo. Si la región de rechazo es \\(\\bar{X} &lt; C)\\) entonces tenemos, siguiendo los cálculos anteriores, que \\[0.05 = \\alpha = \\Phi \\left ( \\frac{C -515}{130/\\sqrt{100}}\\right) = \\Phi \\left( \\frac{C- 515}{13} \\right)\\] Buscamos el cuantil 0.05 de la normal estándar, que es z_alpha &lt;- qnorm(0.05) z_alpha ## [1] -1.644854 Y entonces requerimos que \\[\\frac{C- 515}{13} = -1.6448.\\] Despejando obtenemos C &lt;- 13*z_alpha + 515 C ## [1] 493.6169 Así que podemos usar la región \\(\\bar{X} &lt; 493.5\\), que es más estricta que la anterior de \\(\\bar{X} &lt; 505\\). Considera la potencia de la prueba \\(\\beta(\\mu)\\) que vimos arriba. Discute y corre algunos ejemplos para contestar las siguientes preguntas: Recuerda la definición: ¿qué significa \\(\\beta(\\mu)\\)? ¿Qué pasa con la potencia cuando \\(\\mu\\) está más lejos de los valores de la hipótesis nula? ¿Qué pasa con la potencia cuando hay menos variabilidad en la población? ¿Y cuando la muestra es más grande? ¿Qué pasa si hacemos más chico el nivel de significancia? Consideraciones prácticas Algunos recordatorios de lo que hemos visto: Rechazar la nula no quiere decir que la nula es falsa, ni que encotramos un “efecto”. Un valor-\\(p\\) chico tampoco quiere decir que la nula es falsa. Lo que quiere decir es que la nula es poco consistente con los datos que observamos, o que es muy poco probable que la nula produzca los datos que observamos. Rechazar la nula (encontrar un efecto “significativo”) no quiere decir que el efecto tiene importancia práctica. Si la potencia es alta (por ejemplo cuando el tamaño de muestra es grande), puede ser que la discrepancia de los datos con la nula es despreciable, entonces para fines prácticos podríamos trabajar bajo el supuesto de la nula. Por eso en general preferimos hacer estimación que pruebas de hipótesis para entender o resumir los datos y tamaños de las discrepancias. Adicionalmente, muchas de las hipótesis nulas que generalmente se utilizan se pueden rechazar sin datos (por ejemplo, igualdad de proporciones en dos poblaciones reales). Lo que importa es qué tan diferentes son, y qué tan bien podemos estimar sus diferencias. En la literatura, muchas veces parece que “encontrar una cosa interesante” es rechazar una hipótesis nulas con nivel 5% de significancia. Es más importante entender cómo se diseñó el estudio, cómo se recogieron los datos, cuáles fueron las decisiones de análisis que pasar el mítico nivel de 5% Cuando la potencia es baja (por ejemplo porque el tamaño de muestra es muy chico), tenemos que observar diferencias muy grandes para rechazar. Si probamos algo poco factible (por ejemplo, que la vitamina \\(C\\) aumenta la estatura a los adultos), entonces los rechazos generalmente se deben a variabilidad en la muestra (error tipo II). Cuando diseñamos y presentamos resultados de un estudio o análisis, es mejor pensar en describir los datos y su variabilidad, y mostrar estimaciones incluyendo fuentes de incertidumbre, en lugar de intentar resumir con un valor-\\(p\\) o con el resultado de una prueba de hipótesis. Pruebas múltiples En algunas ocasiones se hacen muchas pruebas para “filtrar” las cosas que son interesantes y las que no. Por ejemplo, cuando comparamos miles de genes entre dos muestras (la nula es que son similares). Si cada prueba se conduce a un nivel \\(\\alpha\\), la probablilidad de tener al menos un rechazo falso (un error tipo I) es considerablemente más alta que \\(\\alpha\\). Por ejemplo, si repetimos una prueba de hipótesis con nivel \\(\\alpha\\) con muestras independientes, la probabilidad de tener al menos un rechazo falso es \\(1-(1-\\alpha)^n\\), que es muy cercano a uno si \\(n\\) es grande (¿cómo derivas esta fórmula?). Por ejemplo, si \\(\\alpha = 0.05\\) y \\(n = 100\\), con más de 99% probabilidad tendremos al menos un rechazo falso, o un “hallazgo” falso. Sin \\(n\\) es muy grande, varios de los hallazgos que encontremos serán debidos a variabilidad muestral. Puedes ver en (Wasserman 2013), sección 10.7 métodos conservadores como corrección de Bonferroni (sólo rechazar cuando el valor-\\(p\\) es menor a \\(0.05/n\\)), o la técnica más moderna de control de tasa de descubrimientos falsos (FDR). Cuando estamos en una situación como esta (que es más retadora en cuanto a análisis), sin embargo, sugerimos usar estimaciones que tomen cuenta todos los datos con regularización apropiada: por ejemplo, en lugar de trabajar con cada muestra por separado, intentamos construir un modelo para el proceso completo de muestreo. Una posibilidad son modelos jerárquicos bayesianos. Ver por ejemplo (Gelman, Hill, and Yajima 2012). Referencias "],["introducción-a-inferencia-bayesiana-1.html", "Sección 10 Introducción a inferencia bayesiana Un primer ejemplo completo de inferencia bayesiana Ejemplo: estimando una proporción Ejemplo: observaciones uniformes Probabilidad a priori Análisis conjugado Pasos de un análisis de datos bayesiano Verificación predictiva posterior Predicción", " Sección 10 Introducción a inferencia bayesiana Para esta sección seguiremos principalmente Kruschke (2015). Adicionalmente puedes ver la sección correspondiente de Chihara and Hesterberg (2018). En las secciones anteriores estudiamos el método de máxima verosimilitud y métodos de remuestreo. Esto lo hemos hecho para estimar parámetros, y cuantificar la incertidumbre qué tenemos acerca de valores poblacionales. La inferencia bayesiana tiene objetivos similares. Igual que en máxima verosimilitud, la inferencia bayesiana comienza con modelos probabilísticos y observaciones. En contraste con máxima verosimilitud, la inferencia bayesiana está diseñada para incorporar información previa o de expertos que tengamos acerca de los parámetros de interés. La inferencia bayesiana cubre como caso particular métodos basados en máxima verosimilitud. El concepto probabilístico básico que utilizamos para construir estos modelos y la inferencia es el de probabilidad condicional: la probabilidad de que ocurran ciertos eventos dada la información disponible del fenómeno que nos interesa. Un primer ejemplo completo de inferencia bayesiana Consideremos el siguiente problema: Nos dan una moneda, y solo sabemos que la moneda puede tener probabilidad \\(3/5\\) de tirar sol (está cargada a sol) o puede ser una moneda cargada a águila, con probabilidad \\(2/5\\) de tirar sol. Vamos a lanzar la moneda dos veces y observamos su resultado (águila o sol). Queremos decir algo acerca de qué tan probable es que hayamos tirado la moneda cargada a sol o la moneda cargada a águila. En este caso, tenemos dos variables: \\(X\\), que cuenta el número de soles obtenidos en el experimento aleatorio, y \\(\\theta\\), que da la probabilidad de que un volado resulte en sol (por ejemplo, si la moneda es justa entonces \\(\\theta = 0.5\\)). ¿Qué cantidades podríamos usar para evaluar qué moneda es la que estamos usando? Si hacemos el experimento, y tiramos la moneda 2 veces, podríamos considerar la probabilidad \\[P(\\theta = 0.4 | X = x)\\] donde \\(x\\) es el número de soles que obtuvimos en el experimento. Esta es la probabilidad condicional de que estemos tirando la moneda con probabilidad de sol 2/5 dado que observamos \\(x\\) soles. Por ejemplo, si tiramos 2 soles, deberíamos calcular \\[P(\\theta=0.4|X=2).\\] ¿Cómo calculamos esta probabilidad? ¿Qué sentido tiene? Usando reglas de probabildad (regla de Bayes en particular), podríamos calcular \\[P(\\theta=0.4|X=2) = \\frac{P(X=2 | \\theta = 0.4) P(\\theta =0.4)}{P(X=2)}\\] Nota que en el numerador uno de los factores, \\(P(X=2 | \\theta = 0.4),\\) es la verosimilitud. Así que primero necesitamos la verosimilitud: \\[P(X=2|\\theta = 0.4) = (0.4)^2 = 0.16.\\] La novedad es que ahora tenemos que considerar la probabilidad \\(P(\\theta = 0.4)\\). Esta cantidad no la habíamos encontrado antes. Tenemos que pensar entonces que este parámetro es una cantidad aleatoria, y puede tomar dos valores \\(\\theta=0.4\\) ó \\(\\theta = 0.6\\). Considerar esta cantidad como aleatoria requiere pensar, en este caso, en cómo se escogió la moneda, o qué sabemos acerca de las monedas que se usan para este experimento. Supongamos que en este caso, nos dicen que la moneda se escoge al azar de una bolsa donde hay una proporción similar de los dos tipos de moneda (0.4 ó 0.6). Es decir el espacio parametral es \\(\\Theta = \\{0.4, 0.6\\},\\) y las probabilidades asociadas a cada posibilidad son las mismas. Es decir, tenemos \\[P(\\theta = 0.4) = P(\\theta = 0.6) =0.5,\\] que representa la probabilidad de escoger de manera aleatoria la moneda con una carga en particular. Ahora queremos calcular \\(P(X=2)\\), pero con el trabajo que hicimos esto es fácil. Pues requiere usar reglas de probabilidad usuales para hacerlo. Podemos utilizar probabilidad total \\[\\begin{align} P(X) &amp;= \\sum_{\\theta \\in \\Theta} P(X, \\theta)\\\\ &amp;= \\sum_{\\theta \\in \\Theta} P(X\\, |\\, \\theta) P(\\theta), \\end{align}\\] lo cual en nuestro ejemplo se traduce en escribir \\[ P(X=2) = P(X=2|\\theta = 0.4)P(\\theta = 0.4) + P(X=2|\\theta=0.6)P(\\theta =0.6),\\] por lo que obtenemos \\[P(X=2) = 0.16(0.5) + 0.36(0.5) = 0.26.\\] Finalmente la probabilidad de haber escogido la moneda con carga \\(2/5\\) dado que observamos dos soles en el lanzamiento es \\[P(\\theta=0.4|X=2) = \\frac{0.16(0.5)}{0.26} \\approx 0.31.\\] Es decir, la probabilidad posterior de que estemos tirando la moneda \\(2/5\\) baja de 0.5 (nuestra información inicial) a 0.31. Este es un ejemplo completo, aunque muy simple, de inferencia bayesiana. La estrategia de inferencia bayesiana implica tomar decisiones basadas en las probabilidades posteriores. Finalmente, podríamos hacer predicciones usando la posterior predictiva. Si \\({X}_{nv}\\) es una nueva tirada adicional de la moneda que estamos usando, nos interesaría saber: \\[P({X}_{nv}=\\mathsf{sol}\\, | \\, X=2)\\] Notemos que un volado adicional es un resultado binario. Por lo que podemos calcular observando que \\(P({X}_{nv}|X=2, \\theta)\\) es una variable Bernoulli con probabilidad \\(\\theta\\), que puede valer 0.4 ó 0.6. Como tenemos las probabilidades posteriores \\(P(\\theta|X=2)\\) podemos usar probabilidad total, condicionado en \\(X=2\\): \\[\\begin{align*} P({X}_{nv}=\\mathsf{sol}\\, | \\, X=2) &amp; = \\sum_{\\theta \\in \\Theta} P({X}_{nv}=\\mathsf{sol}, \\theta \\, | \\, X=2) &amp; \\text{(probabilidad total)}\\\\ &amp;= \\sum_{\\theta \\in \\Theta} P({X}_{nv}=\\mathsf{sol}\\, | \\theta , X=2) P(\\theta \\, | \\, X=2) &amp; \\text{(probabilidad condicional)}\\\\ &amp;= \\sum_{\\theta \\in \\Theta} P({X}_{nv}=\\mathsf{sol}\\, | \\theta ) P(\\theta \\, | \\, X=2), &amp; \\text{(independencia condicional)} \\end{align*}\\] lo que nos da el siguiente cálculo \\[P(X_{nv}=\\mathsf{sol}\\, |\\, \\theta=0.4) \\, P(\\theta=0.4|X=2) \\, +\\, P(X_{nv}=\\mathsf{sol}|\\theta = 0.6) \\, P(\\theta =0.6|X=2)\\] Es decir, promediamos ponderando con las probabilidades posteriores. Por lo tanto obtenemos \\[P(X_{nv} = \\mathsf{sol}|X=2) = 0.4 ( 0.31) + 0.6 (0.69) = 0.538.\\] Observación 0 Nótese que en contraste con máxima verosimilitud, en este ejemplo cuantificamos con probabilidad condicional la incertidumbre de los parámetros que no conocemos. En máxima verosimilitud esta probabilidad no tiene mucho sentido, pues nunca consideramos el parámetro desconocido como una cantidad aleatoria. Observación 1 Nótese el factor \\(P(X=2)\\) en la probabilidad posterior puede entenderse como un factor de normalización. Notemos que los denominadores en la distribución posterior son \\[P(X=2 | \\theta = 0.4) P(\\theta =0.4) = 0.16(0.5) = 0.08,\\] y \\[P(X=2 | \\theta = 0.6) P(\\theta =0.6) = 0.36(0.5) = 0.18.\\] Las probabilidades posteriores son proporcionales a estas dos cantidades, y como deben sumar uno, entonces normalizando estos dos números (dividiendo entre su suma) obtenemos las probabilidades. Observación 2 La nomenclatura que usamos es la siguiente: Como \\(X\\) son los datos observados, llamamos a \\(P(X|\\theta)\\) la verosimilitud, o modelo de los datos. A \\(P(\\theta)\\) le llamamos la distribución inicial o previa. La distribución que usamos para hacer inferencia \\(P(\\theta|X)\\) es la distribución final o posterior. Para utilizar inferencia bayesiana, hay que hacer supuestos para definir las primeras dos partes del modelo. La parte de iniciales o previas está ausente de enfoques como máxima verosimlitud usual. Observación 3 ¿Cómo decidimos las probabilidades iniciales, por ejemplo \\(P(\\theta=0.4)\\) ? Quizá es un supuesto y no tenemos razón para pensar que se hace de otra manera. O quizá conocemos el mecanismo concreto con el que se selecciona la moneda. Discutiremos esto más adelante. Observación 4 ¿Cómo decidimos el modelo de los datos? Aquí típicamente también tenemos que hacer algunos supuestos, aunque algunos de estos pueden estar basados en el diseño del estudio, por ejemplo. Igual que cuando usamos máxima verosimilitud, es necesario checar que nuestro modelo ajusta razonablemente a los datos. Ejercicio Cambia distintos parámetros del número de soles observados, las probabilidades de sol de las monedas, y las probabilidades iniciales de selección de las monedas. n_volados &lt;- 2 # posible valores del parámetro desconocido theta = c(0.4, 0.6) # probabilidades iniciales probs_inicial &lt;- tibble(moneda = c(1, 2), theta = theta, prob_inicial = c(0.5, 0.5)) probs_inicial ## # A tibble: 2 × 3 ## moneda theta prob_inicial ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 0.4 0.5 ## 2 2 0.6 0.5 # verosimilitud crear_verosim &lt;- function(no_soles){ verosim &lt;- function(theta){ # prob de observar no_soles en 2 volados con probabilidad de sol theta dbinom(no_soles, n_volados, theta) } verosim } # evaluar verosimilitud verosim &lt;- crear_verosim(2) # ahora usamos regla de bayes para hacer tabla de probabilidades tabla_inferencia &lt;- probs_inicial |&gt; mutate(verosimilitud = map_dbl(theta, verosim)) |&gt; mutate(inicial_x_verosim = prob_inicial * verosimilitud) |&gt; # normalizar mutate(prob_posterior = inicial_x_verosim / sum(inicial_x_verosim)) tabla_inferencia |&gt; mutate(moneda_obs = moneda) |&gt; select(moneda_obs, theta, prob_inicial, verosimilitud, prob_posterior) ## # A tibble: 2 × 5 ## moneda_obs theta prob_inicial verosimilitud prob_posterior ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 1 0.4 0.5 0.16 0.308 ## 2 2 0.6 0.5 0.36 0.692 ¿Qué pasa cuando el número de soles es 0? ¿Cómo cambian las probabilidades posteriores de cada moneda? Incrementa el número de volados, por ejemplo a 10. ¿Qué pasa si observaste 8 soles, por ejemplo? ¿Y si observaste 0? ¿Qué pasa si cambias las probabilidades iniciales (por ejemplo incrementas la probabilidad inicial de la moneda 1 a 0.9)? Justifica las siguientes aseveraciones (para este ejemplo): Las probabilidades posteriores o finales son una especie de punto intermedio entre verosimilitud y probablidades iniciales. Si tenemos pocas observaciones, las probabilidades posteriores son similares a las iniciales. Cuando tenemos muchos datos, las probabilidades posteriores están más concentradas, y no es tan importante la inicial. Si la inicial está muy concentrada en algún valor, la posterior requiere de muchas observaciones para que se pueda concentrar en otros valores diferentes a los de la inicial. Ahora resumimos los elementos básicos de la inferencia bayesiana, que son relativamente simples: Inferencia bayesiana. Con la notación de arriba: Como \\(X\\) son los datos observados, llamamos a \\(P(X|\\theta)\\) la verosimilitud, proceso generador de datos o modelo de los datos. El factor \\(P(\\theta)\\) le llamamos la distribución inicial o previa. La distribución que usamos para hacer inferencia \\(P(\\theta|X)\\) es la distribución final o posterior Hacemos inferencia usando la ecuación \\[P(\\theta | X) = \\frac{P(X | \\theta) P(\\theta)}{P(X)}\\] que también escribimos: \\[P(\\theta | X) \\propto P(X | \\theta) P(\\theta)\\] donde \\(\\propto\\) significa “proporcional a”. No ponemos \\(P(X)\\) pues como vimos arriba, es una constante de normalización. En estadística Bayesiana, las probablidades posteriores \\(P(\\theta|X)\\) dan toda la información que necesitamos para hacer inferencia. ¿Cuándo damos probablidad alta a un parámetro particular \\(\\theta\\)? Cuando su verosimilitud es alta y/o cuando su probabilidad inicial es alta. De este modo, la posterior combina la información inicial que tenemos acerca de los parámetros con la información en la muestra acerca de los parámetros (verosimilitud). Podemos ilustrar como sigue: Ejemplo: estimando una proporción Consideremos ahora el problema de estimar una proporción \\(\\theta\\) de una población dada usando una muestra iid \\(X_1,X_2,\\ldots, X_n\\) de variables Bernoulli. Ya sabemos calcular la verosimilitud (el modelo de los datos): \\[P(X_1=x_1,X_2 =x_2,\\ldots, X_n=x_n|\\theta) = \\theta^k(1-\\theta)^{n-k},\\] donde \\(k = x_1 + x_2 +\\cdots + x_k\\) es el número de éxitos que observamos. Ahora necesitamos una distribución inicial o previa \\(P(\\theta)\\). Aunque esta distribución puede tener cualquier forma, supongamos que nuestro conocimiento actual podemos resumirlo con una distribución \\(\\mathsf{Beta}(3, 3)\\): \\[P(\\theta) \\propto \\theta^2(1-\\theta)^2.\\] La constante de normalización es 1/30, pero no la requerimos. Podemos simular para examinar su forma: sim_inicial &lt;- tibble(theta = rbeta(10000, 3, 3)) ggplot(sim_inicial) + geom_histogram(aes(x = theta, y = ..density..), bins = 15) De modo que nuestra información inicial es que la proporción puede tomar cualquier valor entre 0 y 1, pero es probable que tome un valor no tan lejano de 0.5. Por ejemplo, con probabilidad 0.95 creemos que \\(\\theta\\) está en el intervalo quantile(sim_inicial$theta, c(0.025, 0.975)) |&gt; round(2) ## 2.5% 97.5% ## 0.15 0.85 Es difícil justificar en abstracto por qué escogeriamos una inicial con esta forma. Aunque esto los detallaremos más adelante, puedes pensar, por el momento, que alguien observó algunos casos de esta población, y quizá vio tres éxitos y tres fracasos. Esto sugeriría que es poco probable que la probablidad \\(\\theta\\) sea muy cercana a 0 o muy cercana a 1. Ahora podemos construir nuestra posterior. Tenemos que \\[P(\\theta| X_1=x_1, \\ldots, X_n=x_n) \\propto P(X_1 = x_1,\\ldots X_n=x_n | \\theta)P(\\theta) = \\theta^{k+2}(1-\\theta)^{n-k + 2}\\] donde la constante de normalización no depende de \\(\\theta\\). Como \\(\\theta\\) es un parámetro continuo, la expresión de la derecha nos debe dar una densidad posterior. Supongamos entonces que hicimos la prueba con \\(n = 30\\) (número de prueba) y observamos 19 éxitos. Tendríamos entonces \\[P(\\theta | S_n = 19) \\propto \\theta^{19 + 2} (1-\\theta)^{30-19 +2} = \\theta^{21}(1-\\theta)^{13}\\] La cantidad de la derecha, una vez que normalizemos por el número \\(P(X=19)\\), nos dará una densidad posterior (tal cual, esta expresion no integra a 1). Podemos obtenerla usando cálculo, pero recordamos que una distribución \\(\\mathsf{\\mathsf{Beta}}(a,b)\\) tiene como fórmula \\[\\frac{1}{B(a,b)} \\theta^{a-1}(1-\\theta)^{b-1}\\] Concluimos entonces que la posterior tiene una distribución \\(\\mathsf{Beta}(22, 14)\\). Podemos simular de la posterior usando código estándar para ver cómo luce: sim_inicial &lt;- sim_inicial |&gt; mutate(dist = &quot;inicial&quot;) sim_posterior &lt;- tibble(theta = rbeta(10000, 22, 14)) |&gt; mutate(dist = &quot;posterior&quot;) sims &lt;- bind_rows(sim_inicial, sim_posterior) ggplot(sims, aes(x = theta, fill = dist)) + geom_histogram(aes(x = theta), bins = 30, alpha = 0.5, position = &quot;identity&quot;) La posterior nos dice cuáles son las posibilidades de dónde puede estar el parámetro \\(\\theta\\). Nótese que ahora excluye prácticamente valores más chicos que 0.25 o mayores que 0.9. Esta distribución posterior es el objeto con el que hacemos inferencia: nos dice dónde es creíble que esté el parámetro \\(\\theta\\). Podemos resumir de varias maneras. Por ejemplo, si queremos un estimador puntual usamos la media posterior: sims |&gt; group_by(dist) |&gt; summarise(theta_hat = mean(theta) |&gt; round(3)) ## # A tibble: 2 × 2 ## dist theta_hat ## &lt;chr&gt; &lt;dbl&gt; ## 1 inicial 0.5 ## 2 posterior 0.61 Nota que el estimador de máxima verosimilitud es \\(\\hat{p} = 19/30 = 0.63\\), que es ligeramente diferente de la media posterior. ¿Por qué? Y podemos construir intervalos de percentiles, que en esta situación suelen llamarse intervalos de credibilidad, por ejemplo: f &lt;- c(0.025, 0.975) sims |&gt; group_by(dist) |&gt; summarise(cuantiles = quantile(theta, f) |&gt; round(2), f = f) |&gt; pivot_wider(names_from = f, values_from = cuantiles) ## # A tibble: 2 × 3 ## # Groups: dist [2] ## dist `0.025` `0.975` ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 inicial 0.15 0.85 ## 2 posterior 0.45 0.76 El segundo renglón nos da un intervalo posterior para \\(\\theta\\) de credibilidad 95%. En inferencia bayesiana esto sustituye a los intervalos de confianza. El intervalo de la inicial expresa nuestras creencias a priori acerca de \\(\\theta\\). Este intervalo es muy amplio (va de 0.15 a 0.85) El intervalo de la posterior actualiza nuestras creencias acerca de \\(\\theta\\) una vez que observamos los datos, y es considerablemente más angosto y por lo tanto informativo. Puedes experimentar en esta shiny app con diferentes iniciales, número de volados y observación de éxitos. Observaciones: Nótese que escogimos una forma analítica fácil para la inicial, pues resultó así que la posterior es una distribución beta. No siempre es así, y veremos qué hacer cuando nuestra inicial no es de un tipo “conveniente”. Como tenemos la forma analítica de la posterior, es posible hacer los cálculos de la media posterior, por ejemplo, integrando la densidad posterior a mano. Esto generalmente no es factible, y en este ejemplo preferimos hacer una aproximación numérica. En este caso particular es posible usando cálculo, y sabemos que la media de una \\(\\mathsf{\\mathsf{Beta}}(a,b)\\) es \\(a/(a+b)\\), de modo que nuestra media posterior es \\[\\hat{\\mu} = (19 + 2)/(30 + 4) = 21/34 = 0.617 \\] que podemos interpretar como sigue: para calcular la media posterior, a nuestras \\(n\\) pruebas iniciales agregamos 4 pruebas adicionales fijas, con 2 éxitos y 2 fracasos, y calculamos la proporción usual de éxitos. Repite el análisis considerando en general \\(n\\) pruebas, con \\(k\\) éxitos. Utiliza la misma distribución inicial. Lo mismo aplica para el intervalo de 95% (¿cómo se calcularía integrando?). También puedes usar la aproximación de R, por ejemplo: qbeta(0.025, shape1 = 22, shape2 = 14) |&gt; round(2) ## [1] 0.45 qbeta(0.975, shape1 = 22, shape2 = 14) |&gt; round(2) ## [1] 0.76 Ejemplo: observaciones uniformes Ahora regresamos al problema de estimación del máximo de una distribución uniforme. En este caso, consideraremos un problema más concreto. Supongamos que hay una lotería (tipo tradicional) en México y no sabemos cuántos números hay. Obtendremos una muestra iid de \\(n\\) números, y haremos una aproximación continua, suponiendo que \\[X_i \\sim U[0,\\theta]\\] La verosimilitud es entonces \\[P(X_1,\\ldots, X_n|\\theta) = \\theta^{-n},\\] cuando \\(\\theta\\) es mayor que todas las \\(X_i\\), y cero en otro caso. Necesitaremos una inicial \\(P(\\theta)\\). Por la forma que tiene la verosimilitud, podemos intentar una distribución Pareto, que tiene la forma \\[P(\\theta) = \\frac{\\alpha \\theta_0^\\alpha}{\\theta^{\\alpha + 1}}\\] con soporte en \\([\\theta_0,\\infty]\\). Tenemos que escoger entonces el mínimo \\(\\theta_0\\) y el parámetro \\(\\alpha\\). En primer lugar, como sabemos que es una lotería nacional, creemos que no puede haber menos de unos 300 mil números, así que \\(\\theta_0 = 300\\). La función acumulada de la pareto es \\(1- (300/\\theta)^\\alpha\\), así que si \\(\\alpha = 1.1\\) el cuantil 99% es: alpha &lt;- 1.1 (300/(0.01)^(1/alpha)) ## [1] 19738 es decir, alrededor de 20 millones de números. Creemos que es poco probable que el número de boletos sea mayor a esta cota. Nótese ahora que la posterior cumple (multiplicando verosimilitud por inicial): \\[P(\\theta|X_1,\\ldots, X_n |\\theta) \\propto \\theta^{-(n + 2.1)}\\] para \\(\\theta\\) mayor que el máximo de las \\(X_n\\)’s y 300, y cero en otro caso. Esta distribución es pareto con \\(\\theta_0&#39; = \\max\\{300, X_1,\\ldots, X_n\\}\\) y \\(\\alpha&#39; = n + 1.1\\) Una vez planteado nuestro modelo, veamos los datos. Obtuvimos la siguiente muestra de números: loteria_tbl &lt;- read_csv(&quot;data/nums_loteria_avion.csv&quot;, col_names = c(&quot;id&quot;, &quot;numero&quot;)) |&gt; mutate(numero = as.integer(numero)) set.seed(334) muestra_loteria &lt;- sample_n(loteria_tbl, 25) |&gt; mutate(numero = numero/1000) muestra_loteria |&gt; as.data.frame() |&gt; head() ## id numero ## 1 87 348.341 ## 2 5 5851.982 ## 3 40 1891.786 ## 4 51 1815.455 ## 5 14 5732.907 ## 6 48 3158.414 Podemos simular de una Pareto como sigue: rpareto &lt;- function(n, theta_0, alpha){ # usar el método de inverso de distribución acumulada u &lt;- runif(n, 0, 1) theta_0 / (1 - u)^(1/alpha) } Simulamos de la inicial: sims_pareto_inicial &lt;- tibble( theta = rpareto(20000, 300, 1.1 ), dist = &quot;inicial&quot;) Y con los datos de la muestra, simulamos de la posterior: sims_pareto_posterior &lt;- tibble( theta = rpareto(20000, max(c(300, muestra_loteria$numero)), nrow(muestra_loteria) + 1.1), dist = &quot;posterior&quot;) sims_theta &lt;- bind_rows(sims_pareto_inicial, sims_pareto_posterior) ggplot(sims_theta) + geom_histogram(aes(x = theta, fill = dist), bins = 70, alpha = 0.5, position = &quot;identity&quot;, boundary = max(muestra_loteria$numero)) + xlim(0, 15000) + scale_y_sqrt() + geom_rug(data = muestra_loteria, aes(x = numero)) Nótese que cortamos algunos valores de la inicial en la cola derecha: un defecto de esta distribución inicial, con una cola tan larga a la derecha, es que pone cierto peso en valores que son poco creíbles y la vuelve poco apropiada para este problema. Regresaremos más adelante a este problema. Si obtenemos percentiles, obtenemos el intervalo f &lt;- c(0.025, 0.5, 0.975) sims_theta |&gt; group_by(dist) |&gt; summarise(cuantiles = quantile(theta, f) |&gt; round(2), f = f) |&gt; pivot_wider(names_from = f, values_from = cuantiles) ## # A tibble: 2 × 4 ## # Groups: dist [2] ## dist `0.025` `0.5` `0.975` ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 inicial 307. 569. 8449. ## 2 posterior 5858. 6010. 6732. Estimamos entre 5.8 millones y 6.7 millones de boletos. El máximo en la muestra es de max(muestra_loteria$numero) ## [1] 5851.982 Escoger la distribución pareto como inicial es conveniente y nos permitió resolver el problema sin dificultad, pero por su forma vemos que no necesariamente es apropiada para el problema por lo que señalamos arriba. Nos gustaría, por ejemplo, poner una inicial como la siguiente qplot(rgamma(2000, 5, 0.001), geom=&quot;histogram&quot;, bins = 20) + scale_x_continuous(breaks = seq(1000, 15000, by = 2000)) Sin embargo, los cálculos no son tan simples en este caso, pues la posterior no tiene un forma reconocible. Tendremos que usar otras estrategias de simulación para ejemplos como este (Monte Carlo por medio de Cadenas de Markov, que veremos más adelante). Probabilidad a priori La inferencia bayesiana es conceptualmente simple: siempre hay que calcular la posterior a partir de verosimilitud (modelo de datos) y distribución inicial o a priori. Sin embargo, una crítica usual que se hace de la inferencia bayesiana es precisamente que hay que tener esa información inicial, y que distintos analistas llegan a distintos resultados si tienen información inicial distinta. Eso realmente no es un defecto, es una ventaja de la inferencia bayesiana. Los datos y los problemas que queremos resolver no viven en un vacío donde podemos creer que la estatura de las personas, por ejemplo, puede variar de 0 a mil kilómetros, el número de boletos de una lotería puede ir de 2 o 3 boletos o también quizá 500 millones de boletos, o la proporción de personas infectadas de una enfermedad puede ser de unos cuantos hasta miles de millones. En todos estos casos tenemos cierta información inicial que podemos usar para informar nuestras estimaciones. Esta información debe usarse. Antes de tener datos, las probabilidades iniciales deben ser examinadas en términos del conocimiento de expertos. Las probabilidades iniciales son supuestos que hacemos acerca del problema de interés, y también están sujetas a críticas y confrontación con datos. Análisis conjugado Los dos ejemplos que hemos visto arriba son ejemplos de análisis conjugado: (Beta-bernoulli) Si las observaciones \\(X_i\\) son \\(\\mathsf{Bernoulli}(p)\\) (\\(n\\) fija) queremos estimar \\(p\\), y tomamos como distribución inicial para \\(p\\) una \\(\\mathsf{Beta}(a,b)\\), entonces la posterior para \\(p\\) cuando \\(S_n=k\\) es \\(\\mathsf{Beta}(k + a, n - k + b)\\), donde \\(S_n = X_1 + X_2 +\\cdots +X_n\\). Y más en general: (Beta-binomial) Si las observaciones \\(X_i, i=1,2,\\ldots, m\\) son \\(\\mathsf{Binomial}(n_i, p)\\) (\\(n_i\\)’s fijas) independientes, queremos estimar \\(p\\), y tomamos como distribución inicial para \\(p\\) una \\(\\mathsf{Beta}(a,b)\\), entonces la posterior para \\(p\\) cuando \\(S_m=k\\) es \\(\\mathsf{Beta}(k + a, n - k + b)\\), donde \\(S_m = X_1 + X_2 +\\cdots +X_m\\) y \\(n= n_1+n_2+\\cdots+n_m\\) También aplicamos: (Uniforme-Pareto) Si el modelo de datos \\(X_i\\) es uniforme \\(\\mathsf{U}[0,\\theta]\\) (\\(n\\) fija), queremos estimar \\(\\theta\\), y tomamos como distribución inicial para \\(\\theta\\) una Pareto \\((\\theta_0, \\alpha)\\), entonces la posterior para \\(p\\) si el máximo de las \\(X_i\\)’s es igual a \\(M\\) es Pareto con parámetros \\((\\max\\{\\theta_0, M\\}, \\alpha + n)\\). Nótese que en estos casos, dada una forma de la verosimilitud, tenemos una familia conocida de iniciales tales que las posteriores están en la misma familia. Estos modelos son convenientes porque podemos hacer simulaciones de la posterior de manera fácil, o usar sus propiedades teóricas. Otro ejemplo típico es el modelo normal-normal: (Normal-normal) Si \\(X_i\\sim \\mathsf{N}(\\mu,\\sigma)\\), con \\(\\sigma\\) conocida, y tomamos como distribución inicial para \\(\\mu \\sim \\mathsf{N}(\\mu_0,\\sigma_0)\\), y definimos la precisión \\(\\tau\\) como el inverso de la varianza \\(\\sigma^2\\), entonces la posterior de \\(\\mu\\) es Normal con media \\((1-\\lambda) \\mu_0 + \\lambda\\overline{x}\\), y precisión \\(\\tau_0 + n\\tau\\), donde \\(\\lambda = \\frac{n\\tau}{\\tau_0 + n\\tau}\\) Completa cuadrados para mostrar las fórmulas del modelo normal-normal con varianza conocida. Más útil es el siguiente modelo: (Normal-Gamma inverso) Sean \\(X_i\\sim \\mathsf{N}(\\mu, \\sigma)\\). Queremos estimar \\(\\mu\\) y \\(\\sigma\\). Tomamos como distribuciones iniciales (dadas por 4 parámetros: \\(\\mu_0, n_0, \\alpha,\\beta\\)): \\(\\tau = \\frac{1}{\\sigma^2} \\sim \\mathsf{Gamma}(\\alpha,\\beta)\\) \\(\\mu|\\sigma\\) es normal con media \\(\\mu_0\\) y varianza \\(\\sigma^2 / {n_0}\\) , y \\(p(\\mu, \\sigma) = p(\\mu|\\sigma)p(\\sigma)\\) Entonces la posterior es: \\(\\tau|x\\) es \\(\\mathsf{Gamma}(\\alpha&#39;, \\beta&#39;)\\), con \\(\\alpha&#39; = \\alpha + n/2\\), \\(\\beta&#39; = \\beta + \\frac{1}{2}\\sum_{i=1}^{n}(x_{i} - \\bar{x})^2 + \\frac{nn_0}{n+n_0}\\frac{({\\bar{x}}-\\mu_{0})^2}{2}\\) \\(\\mu|\\sigma,x\\) es normal con media \\(\\mu&#39; = \\frac{n_0\\mu_{0}+n{\\bar{x}}}{n_0 +n}\\) y varianza \\(\\sigma^2/({n_0 +n})\\). \\(p(\\mu,\\sigma|x) = p(\\mu|x,\\sigma)p(\\sigma|x)\\) Observaciones Nótese que este último ejemplo tiene más de un parámetro. En estos casos, el objeto de interés es la posterior conjunta de los parámetros \\(p(\\theta_1,\\theta_2,\\cdots, \\theta_p|x)\\). Este último ejemplo es relativamente simple pues por la selección de iniciales, para simular de la conjunta de \\(\\mu\\) y \\(\\tau\\) podemos simular primero \\(\\tau\\) (o \\(\\sigma\\)), y después usar este valor para simular de \\(\\mu\\): el par de valores resultantes son una simulación de la conjunta. Los parámetros \\(\\alpha,\\beta\\) para la inicial de \\(\\tau\\) pueden interpretarse como sigue: \\(\\sqrt{\\beta/\\alpha}\\) es un valor “típico” a priori para la varianza poblacional, y \\(a\\) indica qué tan seguros estamos de este valor típico. Nótese que para que funcionen las fórmulas de la manera más simple, escogimos una dependencia a priori entre la media y la precisión: \\(\\tau = \\sigma^{-2}\\) indica la escala de variabilidad que hay en la población, la incial de la media tiene varianza \\(\\sigma^2/n_0\\). Si la escala de variabilidad de la población es más grande, tenemos más incertidumbre acerca de la localización de la media. Aunque esto tiene sentido en algunas aplicaciones, y por convenviencia usamos esta familia conjugada, muchas veces es preferible otro tipo de especificaciones para las iniciales: por ejemplo, la media normal y la desviación estándar uniforme, o media normal, con iniciales independientes. Sin embargo, estos casos no son tratables con análisis conjugado (veremos más adelante cómo tratarlos con MCMC). Ejemplo Supongamos que queremos estimar la estatura de los cantantes de tesitura tenor con una muestra iid de tenores de Estados Unidos. Usaremos el modelo normal de forma que \\(X_i\\sim \\mathsf{N}(\\mu, \\sigma^2)\\). Una vez decidido el modelo, tenemos que poner distribución inicial para los parámetros \\((\\mu, \\sigma^2)\\). Comenzamos con \\(\\sigma^2\\). Como está el modelo, esta inicial debe estar dada para la precisión \\(\\tau\\), pero podemos simular para ver cómo se ve nuestra inicial para la desviación estándar. En la población general la desviación estándar es alrededor de 7 centímetros # Comenzamos seleccionando un valor que creemos típico para la desviación estándar sigma_0 &lt;- 7 # seleccionamos un valor para a, por ejemplo: si es más chico sigma tendrá más # disperisón a &lt;- 3 # ponemos 7 = sqrt(b/a) -&gt; b = a * 64 b &lt;- a * sigma_0 ^ 2 c(a = a, b = b) ## a b ## 3 147 Ahora simulamos para calcular cuantiles tau &lt;- rgamma(1000, a, b) quantile(tau, c(0.05, 0.95)) ## 5% 95% ## 0.005781607 0.042170161 sigma &lt;- 1 / sqrt(tau) mean(sigma) ## [1] 8.002706 quantile(sigma, c(0.05, 0.95)) ## 5% 95% ## 4.869653 13.151520 Que es dispersión considerable: con poca probabilidad la desviación estándar es menor a 4 centímetros, y también creemos que es poco creíble la desviación estándar sea de más de 13 centímetros. Comenzamos con \\(\\mu\\). Sabemos, por ejemplo, que con alta probabilidad la media debe ser algún número entre 1.60 y 1.80. Podemos investigar: la media nacional en estados unidos está alrededor de 1.75, y el percentil 90% es 1.82. Esto es variabilidad en la población: debe ser muy poco probable, por ejemplo, que la media de tenores sea 1.82 Quizá los cantantes tienden a ser un poco más altos o bajos que la población general, así que podríamos agregar algo de dispersión. Podemos establecer parámetros y simular de la marginal a partir de las fórmulas de arriba para entender cómo se ve la inicial de \\(\\mu\\): mu_0 &lt;- 175 # valor medio de inicial n_0 &lt;- 5 # cuánta concentración en la inicial tau &lt;- rgamma(1000, a,b) sigma &lt;- 1/sqrt(tau) mu &lt;- map_dbl(sigma, ~ rnorm(1, mu_0, .x / sqrt(n_0))) quantile(mu, c(0.05, 0.5, 0.95)) ## 5% 50% 95% ## 168.7275 174.8412 180.7905 Que consideramos un rango en el que con alta probabilidad debe estar la media poblacional de los cantantes. Podemos checar nuestros supuestos simulando posibles muestras usando sólo nuestra información previa: simular_normal_invgamma &lt;- function(n, pars){ mu_0 &lt;- pars[1] n_0 &lt;- pars[2] a &lt;- pars[3] b &lt;- pars[4] # simular media tau &lt;- rgamma(1, a, b) sigma &lt;- 1 / sqrt(tau) mu &lt;- rnorm(1, mu_0, sigma/sqrt(n_0)) # simular sigma rnorm(n, mu, sigma) } set.seed(3461) sims_tbl &lt;- tibble(rep = 1:20) |&gt; mutate(estatura = map(rep, ~ simular_normal_invgamma(500, c(mu_0, n_0, a, b)))) |&gt; unnest(cols = c(estatura)) ggplot(sims_tbl, aes(x = estatura)) + geom_histogram() + facet_wrap(~ rep) + geom_vline(xintercept = c(150, 180), colour = &quot;red&quot;) Pusimos líneas de referencia en 150 y 180. Vemos que nuestras iniciales no producen simulaciones totalmente fuera del contexto, y parecen cubrir apropiadamente el espacio de posiblidades para estaturas de los tenores. Quizá hay algunas realizaciones poco creíbles, pero no extremadamente. En este punto, podemos regresar y ajustar la inicial para \\(\\sigma\\), que parece tomar valores demasiado grandes (produciendo por ejemplo una simulación con estatura de 220 y 140, que deberían ser menos probables). Ahora podemos usar los datos para calcular nuestras posteriores. set.seed(3413) cantantes &lt;- lattice::singer |&gt; mutate(estatura_cm = round(2.54 * height)) |&gt; filter(str_detect(voice.part, &quot;Tenor&quot;)) |&gt; sample_n(20) cantantes ## height voice.part estatura_cm ## 139 70 Tenor 1 178 ## 150 68 Tenor 2 173 ## 140 65 Tenor 1 165 ## 132 66 Tenor 1 168 ## 152 69 Tenor 2 175 ## 141 72 Tenor 1 183 ## 161 71 Tenor 2 180 ## 156 71 Tenor 2 180 ## 158 71 Tenor 2 180 ## 164 69 Tenor 2 175 ## 147 68 Tenor 1 173 ## 130 72 Tenor 1 183 ## 162 71 Tenor 2 180 ## 134 74 Tenor 1 188 ## 170 69 Tenor 2 175 ## 167 68 Tenor 2 173 ## 149 64 Tenor 1 163 ## 143 68 Tenor 1 173 ## 157 69 Tenor 2 175 ## 153 71 Tenor 2 180 Los cálculos son un poco tediosos, pero podemos construir una función apropiada: calcular_pars_posterior &lt;- function(x, pars_inicial){ # iniciales mu_0 &lt;- pars_inicial[1] n_0 &lt;- pars_inicial[2] a_0 &lt;- pars_inicial[3] b_0 &lt;- pars_inicial[4] # muestra n &lt;- length(x) media &lt;- mean(x) S2 &lt;- sum((x - media)^2) # sigma post a_1 &lt;- a_0 + 0.5 * n b_1 &lt;- b_0 + 0.5 * S2 + 0.5 * (n * n_0) / (n + n_0) * (media - mu_0)^2 # posterior mu mu_1 &lt;- (n_0 * mu_0 + n * media) / (n + n_0) n_1 &lt;- n + n_0 c(mu_1, n_1, a_1, b_1) } pars_posterior &lt;- calcular_pars_posterior(cantantes$estatura_cm, c(mu_0, n_0, a, b)) pars_posterior ## [1] 175.8 25.0 13.0 509.0 ¿Cómo se ve nuestra posterior comparada con la inicial? Podemos hacer simulaciones: sim_params &lt;- function(m, pars){ mu_0 &lt;- pars[1] n_0 &lt;- pars[2] a &lt;- pars[3] b &lt;- pars[4] # simular sigmas sims &lt;- tibble(tau = rgamma(m, a, b)) |&gt; mutate(sigma = 1 / sqrt(tau)) # simular mu sims &lt;- sims |&gt; mutate(mu = rnorm(m, mu_0, sigma / sqrt(n_0))) sims } sims_inicial &lt;- sim_params(5000, c(mu_0, n_0, a, b)) |&gt; mutate(dist = &quot;inicial&quot;) sims_posterior &lt;- sim_params(5000, pars_posterior) |&gt; mutate(dist = &quot;posterior&quot;) sims &lt;- bind_rows(sims_inicial, sims_posterior) ggplot(sims, aes(x = mu, y = sigma, colour = dist)) + geom_point(alpha = 0.4) Y vemos que nuestra posterior es consistente con la información inicial que usamos, hemos aprendido considerablemente de la muestra. La posterior se ve como sigue. Hemos marcado también las medias posteriores de cada parámetro: media y desviación estándar. medias_post &lt;- sims |&gt; filter(dist == &quot;posterior&quot;) |&gt; select(-dist) |&gt; summarise(across(everything(), mean)) ggplot(sims |&gt; filter(dist == &quot;posterior&quot;), aes(x = mu, y = sigma)) + geom_point(colour = &quot;#00BFC4&quot;) + geom_point(data = medias_post, size = 5, colour = &quot;black&quot;) + coord_equal() Podemos construir intervalos creíbles del 90% para estos dos parámetros, por ejemplo haciendo intervalos de percentiles: f &lt;- c(0.05, 0.5, 0.95) sims |&gt; pivot_longer(cols = mu:sigma, names_to = &quot;parametro&quot;) |&gt; group_by(dist, parametro) |&gt; reframe(cuantil = quantile(value, f) |&gt; round(1), f = f) |&gt; pivot_wider(names_from = f, values_from = cuantil) ## # A tibble: 4 × 5 ## dist parametro `0.05` `0.5` `0.95` ## &lt;chr&gt; &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 inicial mu 169. 175. 181. ## 2 inicial sigma 4.8 7.4 13.3 ## 3 posterior mu 174. 176. 178. ## 4 posterior sigma 5.1 6.3 8.2 Como comparación, los estimadores de máxima verosimlitud son media_mv &lt;- mean(cantantes$estatura_cm) sigma_mv &lt;- mean((cantantes$estatura_cm - media_mv)^2) |&gt; sqrt() c(media_mv, sigma_mv) ## [1] 176 6 Ahora solo resta checar que el modelo es razonable. Veremos más adelante cómo hacer esto, usando la distribución predictiva posterior. Pasos de un análisis de datos bayesiano Como vimos en los ejemplos, en general un análisis de datos bayesiano sigue los siguientes pasos: Identificar los datos releventes a nuestra pregunta de investigación, el tipo de datos que vamos a describir, que variables queremos estimar. Definir el modelo descriptivo para los datos. La forma matemática y los parámetros deben ser apropiados para los objetivos del análisis. Especificar la distribución inicial de los parámetros. Utilizar inferencia bayesiana para reubicar la credibilidad a lo largo de los posibles valores de los parámetros. Verificar que la distribución posterior replique los datos de manera razonable, de no ser el caso considerar otros modelos descriptivos para los datos. Elicitando probablidades subjetivas (opcional) No siempre es fácil elicitar probabilidades subjetivas de manera que capturemos el verdadero conocimiento de dominio que tenemos. Una manera clásica de hacerlo es con apuestas Considera una pregunta sencilla que puede afectar a un viajero: ¿Qué tanto crees que habrá una tormenta que ocasionará el cierre de la autopista México-Acapulco en el puente del \\(20\\) de noviembre? Como respuesta debes dar un número entre \\(0\\) y \\(1\\) que refleje tus creencias. Una manera de seleccionar dicho número es calibrar las creencias en relación a otros eventos cuyas probabilidades son claras. Como evento de comparación considera una experimento donde hay canicas en una urna: \\(5\\) rojas y \\(5\\) blancas. Seleccionamos una canica al azar. Usaremos esta urna como comparación para considerar la tormenta en la autopista. Ahora, considera el siguiente par de apuestas de las cuales puedes elegir una: A. Obtienes \\(\\$1000\\) si hay una tormenta que ocasiona el cierre de la autopista el próximo \\(20\\) de noviembre. B. Obtienes \\(\\$1000\\) si seleccionas una canica roja de la urna que contiene \\(5\\) canicas rojas y \\(5\\) blancas. Si prefieres la apuesta B, quiere decir que consideras que la probabilidad de tormenta es menor a \\(0.5\\), por lo que al menos sabes que tu creencia subjetiva de una la probabilidad de tormenta es menor a \\(0.5\\). Podemos continuar con el proceso para tener una mejor estimación de la creencia subjetiva. A. Obtienes \\(\\$1000\\) si hay una tormenta que ocasiona el cierre de la autopista el próximo \\(20\\) de noviembre. C. Obtienes \\(\\$1000\\) si seleccionas una canica roja de la urna que contiene \\(1\\) canica roja y \\(9\\) blancas. Si ahora seleccionas la apuesta \\(A\\), esto querría decir que consideras que la probabilidad de que ocurra una tormenta es mayor a \\(0.10\\). Si consideramos ambas comparaciones tenemos que tu probabilidad subjetiva se ubica entre \\(0.1\\) y \\(0.5\\). Verificación predictiva posterior Una vez que ajustamos un modelo bayesiano, podemos simular nuevas observaciones a partir del modelo. Esto tiene dos utilidades: Hacer predicciones acerca de datos no observados. Confirmar que nuevas observaciones, producidas simulando con el modelo son similares a las que de hecho observamos. Esto nos permite confirmar la calidad del ajuste del modelo, y se llama verificación predictiva posterior. Supongamos que tenemos la posterior \\(p(\\theta | x)\\). Podemos generar una nueva replicación de los datos como sigue: La distribución predictiva posterior genera nuevas observaciones a partir de la información observada. La denotamos como \\(p(\\tilde{x}|x)\\). Para simular de ella: Muestreamos un valor \\(\\tilde{\\theta}\\) de la posterior \\(p(\\theta|x)\\). Simulamos del modelo de las observaciones \\(\\tilde{x} \\sim p(\\tilde{x}|\\tilde{\\theta})\\). Repetimos el proceso hasta obtener una muestra grande. Usamos este método para producir, por ejemplo, intervalos de predicción para nuevos datos. Si queremos una replicación de las observaciones de la predictiva posterior, Muestreamos un valor \\(\\tilde{\\theta}\\) de la posterior \\(p(\\theta|x)\\). Simulamos del modelo de las observaciones \\(\\tilde{x}_1, \\tilde{x}_2,\\ldots, \\tilde{x}_n \\sim p(\\tilde{x}|\\tilde{\\theta})\\), done \\(n\\) es el tamaño de muestra de la muestra original \\(x\\). Usamos este método para producir conjuntos de datos simulados que comparamos con los observados para verificar nuestro modelo. Ejemplo: estaturas de tenores En este ejemplo, usaremos la posterior predictiva para checar nuestro modelo. Vamos a crear varias muestras, del mismo tamaño que la original, según nuestra predictiva posterior, y compararemos estas muestras con la observada. Y ahora simulamos otra muestra muestra_sim &lt;- simular_normal_invgamma(20, pars_posterior) muestra_sim |&gt; round(0) ## [1] 167 181 184 181 167 167 172 170 177 172 169 174 182 184 176 171 175 176 168 ## [20] 181 Podemos simular varias muestras y hacer una prueba de lineup: library(nullabor) set.seed(9921) sims_obs &lt;- tibble(.n = 1:19) |&gt; mutate(estatura_cm = map(.n, ~ simular_normal_invgamma(20, pars_posterior))) |&gt; unnest(estatura_cm) pos &lt;- sample(1:20, 1) lineup_tbl &lt;- lineup(true = cantantes |&gt; select(estatura_cm), samples = sims_obs, pos = pos) ggplot(lineup_tbl, aes(x = estatura_cm)) + geom_histogram(binwidth = 2.5) + facet_wrap(~.sample) Con este tipo de gráficas podemos checar desajustes potenciales de nuestro modelo. ¿Puedes encontrar los datos verdaderos? ¿Cuántos seleccionaron los datos correctos? Ejemplo: modelo Poisson Supongamos que pensamos el modelo para las observaciones es Poisson con parámetro \\(\\lambda\\). Pondremos como inicial para \\(\\lambda\\) una exponencial con media 10. Nótese que la posterior está dada por \\[p(\\lambda|x_1,\\ldots, x_n) \\propto e^{-n\\lambda}\\lambda^{\\sum_i x_i} e^{-0.1\\lambda} = \\lambda^{n\\overline{x}}e^{-\\lambda(n + 0.1)}\\] que es una distribución gamma con parámetros \\((n\\overline{x} + 1, n+0.1)\\) Ahora supongamos que observamos la siguiente muestra, ajustamos nuestro modelo y hacemos replicaciones posteriores de los datos observados: x &lt;- rnbinom(250, mu = 20, size = 3) crear_sim_rep &lt;- function(x){ n &lt;- length(x) suma &lt;- sum(x) sim_rep &lt;- function(rep){ lambda &lt;- rgamma(1, sum(x) + 1, n + 0.1) x_rep &lt;- rpois(n, lambda) tibble(rep = rep, x_rep = x_rep) } } sim_rep &lt;- crear_sim_rep(x) lineup_tbl &lt;- map(1:5, ~ sim_rep(.x)) |&gt; bind_rows() |&gt; bind_rows(tibble(rep = 6, x_rep = x)) ggplot(lineup_tbl, aes(x = x_rep)) + geom_histogram(bins = 15) + facet_wrap(~rep) Y vemos claramente que nuestro modelo no explica apropiadamente la variación de los datos observados. Contrasta con: set.seed(223) x &lt;- rpois(250, 15) crear_sim_rep &lt;- function(x){ n &lt;- length(x) suma &lt;- sum(x) sim_rep &lt;- function(rep){ lambda &lt;- rgamma(1, sum(x) + 1, n + 0.1) x_rep &lt;- rpois(n, lambda) tibble(rep = rep, x_rep = x_rep) } } sim_rep &lt;- crear_sim_rep(x) lineup_tbl &lt;- map(1:5, ~ sim_rep(.x)) |&gt; bind_rows() |&gt; bind_rows(tibble(rep = 6, x_rep = x)) ggplot(lineup_tbl, aes(x = x_rep)) + geom_histogram(bins = 15) + facet_wrap(~rep) Y verificamos que en este caso el ajuste del modelo es apropiado. Predicción Cuando queremos hacer predicciones particulares acerca de datos que observemos en el futuro, también podemos usar la posterior predictiva. En este caso, tenemos que considerar La variabilidad que produce la incertidumbre en la estimación de los parámetros La variabilidad de las observaciones dados los parámetros. Es decir, tenemos que simular sobre todos las combinaciones factibles de los parámetros. Ejemplo: cantantes Si un nuevo tenor llega a un coro, ¿cómo hacemos una predicción de su estatura? Como siempre, quisiéramos obtener un intervalo que exprese nuestra incertidumbre acerca del valor que vamos a observar. Entonces haríamos: sims_posterior &lt;- sim_params(50000, pars_posterior) |&gt; mutate(y_pred = rnorm(n(), mu, sigma)) sims_posterior |&gt; head() ## # A tibble: 6 × 4 ## tau sigma mu y_pred ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 0.0286 5.91 175. 181. ## 2 0.0200 7.07 177. 178. ## 3 0.0257 6.23 176. 170. ## 4 0.0344 5.39 176. 174. ## 5 0.0297 5.80 175. 169. ## 6 0.0282 5.96 177. 170. f &lt;- c(0.025, 0.5, 0.975) sims_posterior |&gt; summarise(f = f, y_pred = quantile(y_pred, f)) ## # A tibble: 3 × 2 ## f y_pred ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.025 163. ## 2 0.5 176. ## 3 0.975 189. Y con esto obtenemos el intervalo (163, 189), al 95%, para una nueva observación. Nótese que este intervalo no puede construirse con una simulación particular de la posterior de los parámetros, pues sería demasiado corto. Es posible demostrar que en este caso, la posterior predictiva tiene una forma conocida: La posterior predictiva para el modelo normal-gamma inverso es una distribución \\(t\\) con \\(2\\alpha&#39;\\) grados de libertad, centrada en \\(\\mu&#39;\\), y con escala \\(s^2 = \\frac{\\beta&#39;}{\\alpha&#39;}\\frac{n + n_0 + 1}{n +n_0}\\) mu_post &lt;- pars_posterior[1] n_post &lt;- pars_posterior[2] alpha_post &lt;- pars_posterior[3] beta_post &lt;- pars_posterior[4] s &lt;- sqrt(beta_post/alpha_post) * sqrt((n_post + 1)/n_post) qt(c(0.025, 0.5, 0.975), 2 * alpha_post) * s + mu_post ## [1] 162.6832 175.8000 188.9168 Calcula la posterior predictiva del modelo Beta-Bernoulli y Beta-Binomial. (Más difícil) Calcula la posterior predictiva del modelo Poisson-Gamma. Ejemplo: posterior predictiva de Pareto-Uniforme. La posterior predictiva del modelo Pareto-Uniforme no tiene un nombre estándar, pero podemos aproximarla usando simulación. Usando los mismos datos del ejercicio de la lotería, haríamos: rpareto &lt;- function(n, theta_0, alpha){ # usar el método de inverso de distribución acumulada u &lt;- runif(n, 0, 1) theta_0 / (1 - u)^(1/alpha) } # Simulamos de la posterior de los parámetros lim_inf_post &lt;- max(c(300, muestra_loteria$numero)) k_posterior &lt;- nrow(muestra_loteria) + 1.1 sims_pareto_posterior &lt;- tibble( theta = rpareto(100000, lim_inf_post, k_posterior)) # Simulamos una observación para cada una de las anteriores: sims_post_pred &lt;- sims_pareto_posterior |&gt; mutate(x_pred = map_dbl(theta, ~ runif(1, 0, .x))) # Graficamos ggplot(sims_post_pred, aes(x = x_pred)) + geom_histogram(binwidth = 50) + geom_vline(xintercept = lim_inf_post, colour = &quot;red&quot;) Que es una mezcla de una uniforme con una Pareto. Referencias "],["calibración-bayesiana-y-regularización.html", "Sección 11 Calibración bayesiana y Regularización Enfoque bayesiano y frecuentista Ejemplo: estimación de una proporción Intervalos de Agresti-Coull Incorporando información inicial Inferencia bayesiana y regularización Ejemplo: modelo normal y estaturas Ejemplo: estimación de proporciones Teoría de decisión Riesgo de Bayes", " Sección 11 Calibración bayesiana y Regularización El enfoque bayesiano se puede formalizar coherentemente en términos de probabilidades subjetivas, y como vimos, esta es una fortaleza del enfoque bayesiano. En la práctica, sin embargo, muchas veces puede ser difícil argumentar en términos exclusivos de probabilidad subjetiva, aunque hagamos los esfuerzos apropiados para incorporar la totalidad de información que distintos actores involucrados pueden tener. Consideremos, por ejemplo, que INEGI produjera un intervalo creíble del 95% para el ingreso mediano de los hogares de México. Aún cuando nuestra metodología sea transparente y correctamente informada, algunos investigadores interesados puede ser que tengan recelo en usar esta información, y quizá preferirían hacer estimaciones propias. Esto restaría valor al trabajo cuidadoso que pusimos en nuestras estimaciones oficiales. Por otra parte, el enfoque frecuentista provee de ciertas garantías mínimas para la utilización de las estimaciones, que no dependen de la interpretación subjetiva de la probabilidad, sino de las propiedades del muestreo. Consideremos la cobertura de los intervalos de confianza: Bajo ciertos supuestos de nuestros modelos, la probabilidad de que un intervalo de confianza del 95% cubra al verdadero valor poblacional es del 95%. Esta probabilidad es sobre las distintas muestras que se pueden obtener según el diseño del muestreo. Los intervalos creíbles en principio no tienen por qué cumplir esta propiedad, pero consideramos que en la práctica es una garantía mínima que deberían cumplir. El enfoque resultante se llama bayesiano calibrado, Little (2011) . La idea es seguir el enfoque bayesiano usual para construir nuestras estimaciones, pero verificar hasta donde sea posible que los intervalos resultantes satisfacen alguna garantía frecuentista básica. Observación. checar que la cobertura real es similar a la nominal es importante en los dos enfoques: frecuentista y bayesiano. Los intervalos frecuentistas, como hemos visto, generalmente son aproximados, y por lo tanto no cumplen automáticamente esta propiedad de calibración. Enfoque bayesiano y frecuentista Los métodos estadísticos clásicos toman el punto de vista frecuentista y se basa en los siguientes puntos (Wasserman (2013)): La probabilidad se interpreta como un límite de frecuencias relativas, donde las probabilidades son propiedades objetivas en el mundo real. En un modelo, los parámetros son constantes fijas (desconocidas). Como consecuencia, no se pueden realizar afirmaciones probabilísticas útiles en relación a éstos. Los procedimientos estadísticos deben diseñarse con el objetivo de tener propiedades frecuentistas bien definidas. Por ejemplo, un intervalo de confianza del \\(95\\%\\) debe contener el verdadero valor del parámetro con frecuencia límite de al menos el \\(95\\%\\). En contraste, el acercamiento Bayesiano muchas veces se describe por los siguientes postulados: La probabilidad describe grados de creencia, no frecuencias limite. Como tal uno puede hacer afirmaciones probabilísticas acerca de muchas cosas y no solo datos sujetos a variabilidad aleatoria. Por ejemplo, puedo decir: “La probabilidad de que Einstein tomara una taza de té el primero de agosto de \\(1948\\)” es \\(0.35\\), esto no hace referencia a ninguna frecuencia relativa sino que refleja la certeza que yo tengo de que la proposición sea verdadera. Podemos hacer afirmaciones probabilísticas de parámetros. Podemos hacer inferencia de un parámetro \\(\\theta\\) por medio de distribuciones de probabilidad. Las inferencias como estimaciones puntuales y estimaciones de intervalos se pueden extraer de dicha distribución. Finalmente, en el enfoque bayesiano calibrado (Little (2011)): Usamos el enfoque bayesiano para modelar y hacer afirmaciones probabilísticas de los parámetros. Buscamos cumplir las garantías frecuentistas del inciso 3). Ejemplo: estimación de una proporción Recordamos nuestro problema de estimación de una proporcion \\(\\theta\\). Usando la distribución inicial \\(p(\\theta)\\sim \\mathsf{Beta}(2,2)\\), y la verosimilitud estándar binomial, vimos que la posterior cuando observamos \\(k\\) éxitos es \\[p(\\theta|k) \\sim \\mathsf{Beta}(k + 2, n - k + 2)\\]. La media posterior es \\[\\frac{k + 2}{n + 4} \\] que podemos interpretar como: agrega 2 éxitos y 2 fracasos a los datos observados y calcula la proporción de éxitos. Un intervalo posterior de credibilidad del 95% se calcula encontrando los cuantiles 0.025 y 0.975 de una \\(\\mathsf{Beta}(k + 2, n - k + 2)\\) \\[I_a = \\left [q_{0.025}(k+2, n+4), q_{0.975}(k+2, n+4)\\right ]\\] Que compararemos con el intervalo usual de Wald: si \\(\\hat{\\theta} = \\frac{k}{n}\\), entonces \\[I_w = \\left [\\hat{\\theta} - 2 \\sqrt{\\frac{\\hat{\\theta}(1-\\hat{\\theta})}{n}}, \\hat{\\theta} + 2 \\sqrt{\\frac{\\hat{\\theta}(1-\\hat{\\theta})}{n}}\\right]\\] ¿Cómo podemos comparar la calibración de estos dos intervalos? Nominalmente, deben tener cobertura de 95%. Hagamos un ejercicio de simulación para distintos tamaños de muestra \\(n\\) y posibles valores \\(\\theta\\in (0,1)\\): set.seed(332) simular_muestras &lt;- function(M, n, p){ k = rbinom(M, n, p) tibble(rep = 1:M, n = n, p = p, k = k) } intervalo_wald &lt;- function(n, k){ p_hat &lt;- k / n ee_hat &lt;- sqrt(p_hat * (1 - p_hat) / n) tibble(inf = p_hat - 2 * ee_hat, sup = p_hat + 2 * ee_hat) } intervalo_bayes &lt;- function(n, k, a = 2, b = 2){ a &lt;- k + a b &lt;- n - k + b tibble(inf = qbeta(0.025, a, b), sup = qbeta(0.975, a, b)) } set.seed(812) ejemplo &lt;- simular_muestras(5, 20, 0.4) ejemplo |&gt; mutate(intervalo = intervalo_wald(n, k)) |&gt; pull(intervalo) |&gt; bind_cols(ejemplo) |&gt; select(-rep) ## # A tibble: 5 × 5 ## inf sup n p k ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; ## 1 0.0211 0.379 20 0.4 4 ## 2 0.228 0.672 20 0.4 9 ## 3 0.276 0.724 20 0.4 10 ## 4 0.228 0.672 20 0.4 9 ## 5 0.137 0.563 20 0.4 7 ejemplo |&gt; mutate(intervalo = intervalo_bayes(n, k)) |&gt; pull(intervalo) |&gt; bind_cols(ejemplo) |&gt; select(-rep) ## # A tibble: 5 × 5 ## inf sup n p k ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; ## 1 0.102 0.437 20 0.4 4 ## 2 0.268 0.655 20 0.4 9 ## 3 0.306 0.694 20 0.4 10 ## 4 0.268 0.655 20 0.4 9 ## 5 0.197 0.573 20 0.4 7 ¿Cuáles de estos intervalos cubren al verdadero valor? Nótese que no podemos descalificar a ningún método por no cubrir una vez. Es fácil producir un intervalo con 100% de cobertura: (0,1). Pero no nos informa dónde es probable que esté el parámetro. Sin embargo, podemos checar la cobertura frecuentista haciendo una cantidad grande de simulaciones: parametros &lt;- crossing(n = c(5, 10, 30, 60, 100, 400), p = c(0.01, 0.015, 0.02, 0.025, 0.03, 0.035, 0.04, 0.05, 0.07, 0.1, 0.15)) set.seed(2343) # simulaciones simulaciones &lt;- parametros |&gt; mutate(muestra = map2(n, p, ~ simular_muestras(50000, .x, .y) |&gt; select(rep, k))) |&gt; unnest(muestra) # calcular_cobertura calcular_cobertura &lt;- function(simulaciones, construir_intervalo){ # nombre de función intervalo_nombre &lt;- substitute(construir_intervalo) |&gt; as.character() cobertura_tbl &lt;- simulaciones |&gt; mutate(intervalo = construir_intervalo(n, k)) |&gt; pull(intervalo) |&gt; bind_cols(simulaciones) |&gt; mutate(cubre = p &gt;= inf &amp; p &lt;= sup) |&gt; group_by(n, p) |&gt; summarise(cobertura = mean(cubre), long_media = mean(sup - inf)) cobertura_tbl |&gt; mutate(tipo = intervalo_nombre) } cobertura_wald &lt;- calcular_cobertura(simulaciones, intervalo_wald) cobertura_wald ## # A tibble: 66 × 5 ## # Groups: n [6] ## n p cobertura long_media tipo ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; ## 1 5 0.01 0.0483 0.0347 intervalo_wald ## 2 5 0.015 0.0733 0.0527 intervalo_wald ## 3 5 0.02 0.0954 0.0689 intervalo_wald ## 4 5 0.025 0.119 0.0862 intervalo_wald ## 5 5 0.03 0.140 0.102 intervalo_wald ## 6 5 0.035 0.165 0.120 intervalo_wald ## 7 5 0.04 0.187 0.137 intervalo_wald ## 8 5 0.05 0.227 0.167 intervalo_wald ## 9 5 0.07 0.299 0.223 intervalo_wald ## 10 5 0.1 0.398 0.303 intervalo_wald ## # ℹ 56 more rows graficar_cobertura &lt;- function(cobertura_tbl){ ggplot(cobertura_tbl, aes(x = p, y = cobertura, colour = tipo)) + geom_hline(yintercept = 0.95, colour = &quot;black&quot;) + geom_line() + geom_point() + facet_wrap(~n) + ylim(0, 1) } cobertura_wald |&gt; graficar_cobertura() La cobertura real es mucho más baja que la nominal en muchos casos, especialmente cuando la \\(p\\) es baja y \\(n\\) es chica. Pero incluso para muestras relativamente grandes (100), la cobertura es mala si \\(p\\) es chica. Ahora probamos nuestro método alternativo: cobertura_bayes &lt;- calcular_cobertura(simulaciones, intervalo_bayes) bind_rows(cobertura_wald, cobertura_bayes) |&gt; mutate(tipo = factor(tipo, levels = c(&#39;intervalo_wald&#39;, &#39;intervalo_bayes&#39;))) |&gt; graficar_cobertura() Y vemos que en general el intervalo de Bayes es superior al de Wald, en sentido de que su cobertura real es más cercana a la nominal. El caso donde fallan los dos es para muestras muy chicas \\(n=5, 10\\), con probabilidades de éxito chicas \\(p\\leq 0.02\\). Sin embargo, si tenemos información previa acerca del tamaño de la proporción que estamos estimando, es posible obtener buena calibración con el método bayesiano. En este caso particular, tenemos argumentos frecuentistas para utilizar el método bayesiano. Por ejemplo, si el INEGI utilizara estos intervalos creíbles, un análisis de calibración de este tipo sostendría esa decisión. Intervalos de Agresti-Coull Un método intermedio que se usa para obtener mejores intervalos cuando estimamos proporciones es el siguiente: Agregar dos 1’s y dos 0’s a los datos. Utilizar el método de Wald con estos datos modificados. intervalo_agresti_coull &lt;- function(n, k){ p_hat &lt;- (k + 2)/ (n + 4) ee_hat &lt;- sqrt(p_hat * (1 - p_hat) / n) tibble(inf = p_hat - 2 * ee_hat, sup = p_hat + 2 * ee_hat) } cobertura_ac &lt;- calcular_cobertura(simulaciones, intervalo_agresti_coull) bind_rows(cobertura_wald, cobertura_bayes, cobertura_ac) |&gt; mutate(tipo = factor(tipo, levels = c(&#39;intervalo_wald&#39;, &#39;intervalo_bayes&#39;, &#39;intervalo_agresti_coull&#39;))) |&gt; graficar_cobertura() Que tiende a ser demasiado conservador para proporciones chicas: graficar_cobertura(cobertura_ac) + ylim(c(0.9, 1)) Conclusión 1: Los intervalos de Agresti-Coull son una buena alternativa para estimar proporciones como sustituto de los intervalos clásicos de Wald, aunque tienden a ser muy conservadores para muestras chicas Idealmente podemos utilizar un método bayesiano pues normalmente tenemos información inicial acerca de las proporciones que queremos estimar. Incorporando información inicial Nótese que generalmente tenemos información acerca de la cantidad que queremos estimar: por ejemplo, que proporción de visitantes de un sitio web compra algo (usualmente muy baja, menos de 2%), qué proporción de personas tiene diabetes tipo 1 (una proporción muy baja, menos de 1 por millar), o qué proporción de hogares tienen ingresos trimestrales mayores a 150 mil pesos (menos de %5 con alta probabilidad). En este caso, tenemos que ajustar nuestra inicial. Por ejemplo, para el problema de ingresos, podríamos usar una \\(\\mathsf{Beta}(2, 100)\\), cuyos cuantiles son: # uno de cada 100 a &lt;- 2 b &lt;- 100 beta_sims &lt;- rbeta(5000, a, b) quantile(beta_sims, c(0.01, 0.05, 0.50, 0.90, 0.99)) |&gt; round(3) ## 1% 5% 50% 90% 99% ## 0.001 0.004 0.016 0.039 0.067 qplot(beta_sims) Veamos cómo se ven los intervalos bayesianos producidos con esta inicial: crear_intervalo_bayes &lt;- function(a, b){ intervalo_fun &lt;- function(n, k){ a_post &lt;- k + a b_post &lt;- n - k + b tibble(inf = qbeta(0.025, a_post, b_post), sup = qbeta(0.975, a_post, b_post)) } intervalo_fun } intervalo_bayes_2 &lt;- crear_intervalo_bayes(a, b) cobertura_bayes &lt;- calcular_cobertura(simulaciones, intervalo_bayes_2) graficar_cobertura(bind_rows(cobertura_bayes, cobertura_ac) |&gt; filter(p &lt; 0.05)) + ylim(c(0.5, 1)) Y vemos que la calibración es similar. Notemos sin embargo que la longitud del del intervalo bayesiano es mucho menor que el de Agresti-Coull cuando la muestra es chica: ggplot(bind_rows(cobertura_bayes, cobertura_ac), aes(x = p, y = long_media, colour = tipo)) + geom_point() + facet_wrap(~n) Cuando la muestra es chica, los intervalos de bayes son similares a los iniciales, y mucho más cortos que los de Agresti-Coull. Para muestras intermedias (50-100) los intervalos bayesianos son más informativos que los de Agresti-Coull, con calibración similar, y representan aprendizaje por encima de lo que sabíamos en la inicial. Para muestras grandes, obtenemos resultados simililares. Por ejemplo: set.seed(2131) k &lt;- rbinom(1, 50, 0.03) k ## [1] 4 intervalo_agresti_coull(50, k) |&gt; round(3) ## # A tibble: 1 × 2 ## inf sup ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.022 0.2 es un intervalo muy grande que puede incluir valores negativos. En contraste, el intervalo bayesiano es: intervalo_bayes_2(50, k) |&gt; round(3) ## # A tibble: 1 × 2 ## inf sup ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.015 0.076 Aún quitando valores negativos, los intervalos de Agresti-Coull son mucho más anchos. La aproximación bayesiana, entonces, utiliza información previa para dar un resultado considerablemente más informativo, con calibración similar a Agresti-Coull. ¿Aprendimos algo? Comparemos la posterior con la inicial: beta_sims_inicial &lt;- tibble(prop = rbeta(5000, a, b), dist = &quot;inicial&quot;) beta_sims_posterior &lt;- tibble(prop = rbeta(5000, a + k, b + 50), dist = &quot;posterior&quot;) bind_rows(beta_sims_inicial, beta_sims_posterior) |&gt; ggplot(aes(x = prop, fill = dist)) + geom_histogram(alpha = 0.5, position = &quot;identity&quot;) Donde vemos que no aprendimos mucho en este caso, pero nuestras creencias sí cambiaron en comparación con la inicial. Conclusión 2: con el enfoque bayesiano podemos obtener intervalos informativos con calibración razonable, incluso con información inicial que no es muy precisa. Los intervalos de Agresti-Coull son poco informativos para muestras chicas y/o proporciones chicas. Ejemplo: porporción de hogares de ingresos grandes Usaremos los datos de ENIGH como ejemplo (ignorando el diseño, pero es posible hacer todas las estimaciones correctamente) para estimar el porcentaje de hogares que tienen ingreso corriente de más de 150 mil pesos al trimestre. Suponemos que la muestra del enigh es la población, y tomaremos una muestra iid de esta población. Usamos la misma inicial que mostramos arriba, que es una Beta con parámetros c(a,b) ## [1] 2 100 set.seed(2521) muestra_enigh &lt;- read_csv(&quot;data/conjunto_de_datos_enigh_2018_ns_csv/conjunto_de_datos_concentradohogar_enigh_2018_ns/conjunto_de_datos/conjunto_de_datos_concentradohogar_enigh_2018_ns.csv&quot;) |&gt; select(ing_cor) |&gt; sample_n(120) |&gt; mutate(mas_150mil = ing_cor &gt; 150000) Un intervalo de 95% es entonces k &lt;- sum(muestra_enigh$mas_150mil) k ## [1] 3 intervalo_bayes_2(120, sum(muestra_enigh$mas_150mil)) |&gt; round(3) ## # A tibble: 1 × 2 ## inf sup ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.007 0.046 La media posterior es prop_post &lt;- (a + k) / (120 + b) prop_post ## [1] 0.02272727 ¿Cuál es la verdadera proporción? read_csv(&quot;data/conjunto_de_datos_enigh_2018_ns_csv/conjunto_de_datos_concentradohogar_enigh_2018_ns/conjunto_de_datos/conjunto_de_datos_concentradohogar_enigh_2018_ns.csv&quot;) |&gt; select(ing_cor) |&gt; mutate(mas_150mil = ing_cor &gt; 150000) |&gt; summarise(prop_pob = mean(mas_150mil)) ## # A tibble: 1 × 1 ## prop_pob ## &lt;dbl&gt; ## 1 0.0277 En este caso, nuestro intervalo cubre a la proporción poblacional. Inferencia bayesiana y regularización Como hemos visto en análisis y modelos anteriores, la posterior que usamos para hacer inferencia combina aspectos de la inicial con la verosimilitud (los datos). Una manera de ver esta combinación y sus beneficios es pensando en término de regularización de estimaciones. En las muestras hay variación. Algunas muestras particulares nos dan estimaciones de máxima verosimilitud pobres de los parámetros de interés (estimaciones ruidosas). Cuando esas estimaciones pobres están en una zona de baja probabilidad de la inicial, la estimación posterior tiende a moverse (o encogerse) hacia las zonas de alta probabilidad de la inicial. Esto filtra ruido en las estimaciones. El mecanismo resulta en una reducción del error cuadrático medio, mediante una reducción de la varianza de los estimadores (aunque quizá el sesgo aumente). Esta es una técnica poderosa, especialmente para problemas complejos donde tenemos pocos datos para cada parámetro. En general, excluímos resultados que no concuerdan con el conocimiento previo, y esto resulta en mayor precisión en las estimaciones. Ejemplo: modelo normal y estaturas Haremos un experimento donde simularemos muestras de los datos de cantantes. Usaremos el modelo normal-gamma inverso que discutimos anteriormente, con la información inicial que elicitamos. ¿Cómo se compara la estimación de máxima verosimilitud con la media posterior? # inicial para media, ver sección anterior para discusión (normal) mu_0 &lt;- 175 n_0 &lt;- 5 # inicial para sigma^2 (gamma inversa) a &lt;- 3 b &lt;- 140 Para este ejemplo chico, usaremos muestras de tamaño 5: set.seed(3413) # ver sección anterior para explicación de esta función calcular_pars_posterior &lt;- function(x, pars_inicial){ # iniciales mu_0 &lt;- pars_inicial[1] n_0 &lt;- pars_inicial[2] a_0 &lt;- pars_inicial[3] b_0 &lt;- pars_inicial[4] # muestra n &lt;- length(x) media &lt;- mean(x) S2 &lt;- sum((x - media)^2) # sigma post a_1 &lt;- a_0 + 0.5 * n b_1 &lt;- b_0 + 0.5 * S2 + 0.5 * (n * n_0) / (n + n_0) * (media - mu_0)^2 # posterior mu mu_1 &lt;- (n_0 * mu_0 + n * media) / (n + n_0) n_1 &lt;- n + n_0 c(mu_1, n_1, a_1, b_1) } Y también de la sección anterior: sim_params &lt;- function(m, pars){ mu_0 &lt;- pars[1] n_0 &lt;- pars[2] a &lt;- pars[3] b &lt;- pars[4] # simular sigmas sims &lt;- tibble(tau = rgamma(m, a, b)) |&gt; mutate(sigma = 1 / sqrt(tau)) # simular mu sims &lt;- sims |&gt; mutate(mu = rnorm(m, mu_0, sigma / sqrt(n_0))) sims } # simular muestras y calcular medias posteriores simular_muestra &lt;- function(rep, mu_0, n_0, a_0, b_0){ cantantes &lt;- lattice::singer |&gt; mutate(estatura_cm = 2.54 * height) |&gt; filter(str_detect(voice.part, &quot;Tenor&quot;)) |&gt; sample_n(5, replace = FALSE) pars_posterior &lt;- calcular_pars_posterior(cantantes$estatura_cm, c(mu_0, n_0, a_0, b_0)) medias_post &lt;- sim_params(1000, pars_posterior) |&gt; summarise(across(everything(), mean)) |&gt; select(mu, sigma) media &lt;- mean(cantantes$estatura_cm) est_mv &lt;- c(&quot;mu&quot; = media, &quot;sigma&quot; = sqrt(mean((cantantes$estatura_cm - media)^2))) bind_rows(medias_post, est_mv) |&gt; mutate(rep = rep, tipo = c(&quot;media_post&quot;, &quot;max_verosim&quot;)) |&gt; pivot_longer(mu:sigma, names_to = &quot;parametro&quot;, values_to = &quot;estimador&quot;) } poblacion &lt;- lattice::singer |&gt; mutate(estatura_cm = 2.54 * height) |&gt; filter(str_detect(voice.part, &quot;Tenor&quot;)) |&gt; summarise(mu = mean(estatura_cm), sigma = sd(estatura_cm)) |&gt; pivot_longer(mu:sigma, names_to = &quot;parametro&quot;, values_to = &quot;valor_pob&quot;) errores &lt;- map(1:2000, ~ simular_muestra(.x, mu_0, n_0, a, b)) |&gt; bind_rows() |&gt; left_join(poblacion) |&gt; mutate(error = (estimador - valor_pob)) ggplot(errores, aes(x = error, fill = tipo)) + geom_histogram(bins = 20, position = &quot;identity&quot;, alpha = 0.5) + facet_wrap(~parametro) Vemos claramente que la estimación de la desviación estándar de nuestro modelo es claramente superior a la de máxima verosimilitud. En resumen: errores |&gt; group_by(tipo, parametro) |&gt; summarise(recm = sqrt(mean(error^2)) |&gt; round(2)) |&gt; arrange(parametro) ## # A tibble: 4 × 3 ## # Groups: tipo [2] ## tipo parametro recm ## &lt;chr&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 max_verosim mu 2.85 ## 2 media_post mu 1.55 ## 3 max_verosim sigma 2.45 ## 4 media_post sigma 1.04 Obtenemos una ganancia considerable en cuanto a la estimación de la desviación estandar de esta población. Los estimadores de la media posterior son superiores a los de máxima verosimilitud en términos de error cuadrático medio. Podemos graficar las dos estimaciones, muestra a muestra, para entender cómo sucede esto: errores |&gt; select(-error) |&gt; pivot_wider(names_from = tipo, values_from = estimador) |&gt; filter(parametro == &quot;sigma&quot;) |&gt; ggplot(aes(x = max_verosim, y = media_post)) + geom_abline(colour = &quot;red&quot;) + geom_hline(yintercept = sqrt(b/(a - 1)), lty = 2, color = &#39;black&#39;) + geom_point() + labs(subtitle = &quot;Estimación de sigma&quot;) + xlab(&quot;Estimador MV de sigma&quot;) + ylab(&quot;Media posterior de sigma&quot;) + coord_fixed() + geom_segment(aes(x = 13, y = 11, xend = 13, yend = sqrt(b/(a - 1))), colour=&#39;red&#39;, size=1, arrow =arrow(length = unit(0.5, &quot;cm&quot;))) + geom_segment(aes(x = .5, y = 6, xend = .5, yend = sqrt(b/(a - 1))), colour=&#39;red&#39;, size=1, arrow =arrow(length = unit(0.5, &quot;cm&quot;))) Nótese como estimaciones demasiado bajas o demasiada altas son contraídas hacia valores más consistentes con la inicial, lo cual resulta en menor error. El valor esperado de \\(\\sigma\\) bajo la distribución inicial se muestra como una horizontal punteada. Ejemplo: estimación de proporciones Ahora repetimos el ejercicio de la estimación de la proporción de hogares con ingresos superiores a 150 mil. # inicial a &lt;- 2 b &lt;- 100 qbeta(c(0.01, 0.99), a, b) ## [1] 0.001477084 0.063921446 # datos datos &lt;- read_csv(&quot;data/conjunto_de_datos_enigh_2018_ns_csv/conjunto_de_datos_concentradohogar_enigh_2018_ns/conjunto_de_datos/conjunto_de_datos_concentradohogar_enigh_2018_ns.csv&quot;) |&gt; select(ing_cor) # estimaciones obtener_estimados &lt;- function(datos){ muestra_enigh &lt;- datos |&gt; sample_n(120) |&gt; mutate(mas_150mil = ing_cor &gt; 150000) k &lt;- sum(muestra_enigh$mas_150mil) tibble(k = k, est_mv = k/120, media_post = (a + k) / (120 + b), pob = 0.02769) } estimadores_sim &lt;- map(1:200, ~obtener_estimados(datos)) |&gt; bind_rows() # calculo de errores error_cm &lt;- estimadores_sim |&gt; summarise(error_mv = sqrt(mean((est_mv - pob)^2)), error_post = sqrt(mean((media_post - pob)^2))) error_cm ## # A tibble: 1 × 2 ## error_mv error_post ## &lt;dbl&gt; &lt;dbl&gt; ## 1 0.0147 0.00928 Podemos ver claramente que las medias posteriores están encogidas hacia valores más chicos (donde la inicial tiene densidad alta) comparadas con las estimaciones de máxima verosimilitud: estimadores_sim_ag &lt;- estimadores_sim |&gt; group_by(k, est_mv, media_post) |&gt; summarise(n = n()) ggplot(estimadores_sim_ag, aes(x = est_mv, media_post, size = n)) + geom_point() + geom_abline() Teoría de decisión En esta parte (que sigue a Wasserman (2013) a grandes rasgos), discutimos brevemente teoría general que nos sirve para seleccionar estimadores puntuales, y que esperemos ponga en contexto la parte anterior que acabamos de discutir. Usaremos algunos conceptos que vimos en la parte de propiedades de estimadores de máxima verosimilitud. Definimos una función de pérdida \\(L(\\theta, \\hat{\\theta}_n)\\), que mide el costo de la discrepancia entre nuestro estimador \\[\\hat{\\theta}_n = t(X_1,\\ldots, X_n) = t(X)\\] y el verdadero valor \\(\\theta\\) poblacional. Es posible considerar distintas funciones de pérdida, pero como en secciones anteriores, usaremos la pérdida cuadrática, definida por: \\[L(\\theta, \\hat{\\theta}_n) = (\\theta - \\hat{\\theta}_n)^2\\] Esta función toma distintos valores dependiendo de la muestra y del parámetro \\(\\theta\\), y necesitamos resumirla para dar una evaluación de qué tan bueno es el estimador \\(\\hat{\\theta}_n\\). Ahora que hemos considerado tanto estadística bayesiana como frecuentista, podemos pensar en resumir esta función de distintas maneras. Comenzamos pensando de manera frecuentista. En este caso, consideramos a \\(\\theta\\) como un valor fijo, y nos preguntamos qué pasaría con la pérdida con distintas muestras potenciales que podríamos observar. Definimos como antes el riesgo (frecuentista) como: \\[R(\\theta, t) = \\mathbb{E}_X\\left[ (\\theta - \\hat{\\theta}_n)^2 \\, \\big| \\, \\theta\\right]\\] donde promediamos a lo largo de las muestras posibles, con \\(\\theta\\) fijo. Esta cantidad no nos dice necesariamente cómo escoger un buen estimador para \\(\\theta\\), pues dependiendo de dónde está \\(\\theta\\) puede tomar valores distintos. Ahora vamos a pensar de manera bayesiana: en este caso, los datos serán fijos una vez que los obervemos de manera que \\(\\hat{\\theta}_n\\) está fijo, y el parámetro \\(\\theta\\) es una cantidad aleatoria con distribución inicial \\(p(\\theta)\\). Entonces consideraríamos el promedio sobre la posterior dado por: \\[\\rho(t, X) = \\mathbb{E}_{p(\\theta|X)}\\left[(\\theta - \\hat{\\theta})^2 \\, \\big | \\, X \\right]\\] que llamamos riesgo posterior. Esta cantidad se calcula con la posterior de los parámetros dados los datos, y nos dice, una vez que vimos los datos, cómo es el error de nuestro estimador. Nótese que esta cantidad no es útil para escoger un estimador bueno \\(t\\) antes de ver los datos, pero nos sirve para evaluar a un estimador dados los datos. En el primer caso, promediamos sobre posibles muestras, y en el segundo por valores posibles de \\(\\theta\\) para una muestra dada. Ejemplo: riesgo frecuentista Para observaciones bernoulli, el estimador de máxima verosimilitud es \\(\\hat{p}_1 = k /n\\), donde \\(n\\) es el tamaño de muestra y \\(k\\) el número de éxitos observados. Podemos usar también como estimador la media posterior de un modelo Beta-Bernoulli con inicial \\(a=2, b=2\\), que nos daría \\(\\hat{p}_2 = \\frac{k + 2}{n + 4}\\). Aunque podemos hacer los cálculos analíticos, aproximaremos el riesgo bajo el error cuadrático usando simulación perdida_cuad &lt;- function(p, p_hat){ (p - p_hat)^2 } # dos estimadores t_1 &lt;- function(n, x) x / n t_2 &lt;- function(n, x) (x + 2) / (n + 4) estimar_riesgo &lt;- function(n = 20, theta, perdida, reps = 10000){ x &lt;- rbinom(reps, n, theta) # calcular estimaciones theta_1 &lt;- t_1(n, x) theta_2 &lt;- t_2(n, x) # calcular pérdida media_perdida &lt;- tibble( n = n, theta = theta, estimador = c(&quot;MLE&quot;, &quot;Posterior&quot;), riesgo = c(mean(perdida(theta, theta_1)), mean(perdida(theta, theta_2)))) media_perdida } estimar_riesgo(n = 20, theta = 0.1, perdida = perdida_cuad) ## # A tibble: 2 × 4 ## n theta estimador riesgo ## &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 20 0.1 MLE 0.00449 ## 2 20 0.1 Posterior 0.00755 Como dijimos, esta cantidad depende de \\(\\theta\\) que no conocemos. Así que calculamos para cada valor de \\(\\theta:\\) Las funciones de riesgo \\(R(\\theta, t_1)\\) y \\(R(\\theta, t_2)\\) (dependen de \\(\\theta\\)) se ven aproximadamente como sigue: p_seq &lt;- seq(0, 1, 0.001) riesgo_tbl &lt;- map(p_seq, ~ estimar_riesgo(n = 20, theta = .x, perdida = perdida_cuad)) |&gt; bind_rows() ggplot(riesgo_tbl, aes(x = theta, y = riesgo, colour = estimador)) + geom_line() Y vemos que el riesgo depende del verdadero valor del parametro: en los extremos, el estimador de máxima verosimilitud tiene menos riesgo, pero en el centro tiene más (esto es independiente del tipo de intervalos que construyamos y su cobertura). La razón es que las estimaciones de tipo Agresti-Coull (\\(\\theta_2\\)) están contraídas hacia 0.5 (agregamos dos éxitos y dos fracasos). Esto produce sesgo en la estimación para valores extremos de \\(\\theta\\). Sin embargo, para valores centrales de \\(\\theta\\) tiene menos variabilidad (por regularización) que el estimador de máxima verosimilitud, y sufre poco de sesgo. Ejemplo: riesgo posterior Supongamos que la inicial es \\(\\theta \\sim \\mathsf{Beta}(5,3)\\) estimar_riesgo_post &lt;- function(n = 20, x, perdida, reps = 20000){ # calcular estimaciones theta_1 &lt;- t_1(n, x) theta_2 &lt;- t_2(n, x) # simular de posterior theta_post &lt;- rbeta(reps, x + 5, n - x + 3) # calcular pérdida media_perdida &lt;- tibble( n = n, x = x, estimador = c(&quot;MLE&quot;, &quot;Posterior&quot;), riesgo_post= c(mean(perdida(theta_post, theta_1)), mean(perdida(theta_post, theta_2)))) media_perdida } estimar_riesgo_post(n = 20, x = 8, perdida = perdida_cuad) ## # A tibble: 2 × 4 ## n x estimador riesgo_post ## &lt;dbl&gt; &lt;dbl&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 20 8 MLE 0.0127 ## 2 20 8 Posterior 0.0109 Como dijimos, esta cantidad depende de los datos \\(x\\) que no hemos observado. Así que calculamos para cada valor de \\(x\\): Las funciones de pérdida promedio \\(\\rho(x, t_1)\\) y \\(\\rho(x, t_2)\\) (dependen de \\(x\\)) se ven aproximadamente como sigue: x_seq &lt;- seq(0, 20, 1) riesgo_post_tbl &lt;- map(x_seq, ~ estimar_riesgo_post(n = 20, x = .x, perdida = perdida_cuad)) |&gt; bind_rows() ggplot(riesgo_post_tbl, aes(x = x, y = riesgo_post, colour = estimador)) + geom_line() + geom_point() Donde vemos que la pérdida del estimador bayesiano es mejor para valores extremos de número de éxitos observado \\(x\\), pero tiene más riesgo posterior para valores chicos de \\(x\\). En general es mejor el estimador \\(\\theta_2\\). El estimador de máxima verosimilitud tiene más riesgo en los extremos, lo que esperaríamos porque no tenemos la regularización que aporta la posterior. Igualmente, vemos más riesgo para valores chicos de \\(x\\) que para valores grandes: esto es porque la inicial está concentrada en valores reslativamente grandes de \\(\\theta\\). Riesgo de Bayes Finalmente, podemos crear un resumen unificado considerando: Si no conocemos el valor del parámetro \\(\\theta\\), podemos promediar el riesgo frecuentista con la inicial \\(p(\\theta)\\) Si no conocemos los datos observados, podemos promediar usando datos generados por la marginal \\(p(x)\\) de \\(x\\) bajo el modelo de datos \\(p(x|\\theta)\\) y la inicial \\(p(\\theta)\\). Por la ley de la esperanza iterada, estos dos resultados son iguales. La cantidad resultante \\[r(t) = \\int R(\\theta,t) p(\\theta)\\, d\\theta = \\int r(x, t)p(x|\\theta)p(\\theta)\\, d\\theta\\, dx\\] Se llama riesgo de Bayes para el estimador \\(t\\). Ejemplo Podemos calcular marginal_tbl &lt;- function(n = 20, m = 5000){ theta &lt;- rbeta(m, 5, 3) x &lt;- rbinom(m, size = n, p = theta) tibble(x = x) |&gt; group_by(x) |&gt; summarise(n_x = n()) } riesgo_post_tbl |&gt; left_join(marginal_tbl()) |&gt; group_by(estimador) |&gt; summarise(riesgo_bayes = sum(riesgo_post * n_x) / sum(n_x)) ## # A tibble: 2 × 2 ## estimador riesgo_bayes ## &lt;chr&gt; &lt;dbl&gt; ## 1 MLE 0.0104 ## 2 Posterior 0.00833 o también theta_tbl &lt;- tibble(theta = rbeta(50000, 5, 3) |&gt; round(3)) |&gt; group_by(theta) |&gt; summarise(n_x = n()) riesgo_tbl |&gt; left_join(theta_tbl) |&gt; mutate(n_x = ifelse(is.na(n_x), 0, n_x)) |&gt; group_by(estimador) |&gt; summarise(riesgo_bayes = sum(riesgo * n_x) / sum(n_x)) ## # A tibble: 2 × 2 ## estimador riesgo_bayes ## &lt;chr&gt; &lt;dbl&gt; ## 1 MLE 0.0104 ## 2 Posterior 0.00839 Ahora consideremos cómo decidiríamos, desde el punto de vista Bayesiano, qué estimador usar: (Estimador de Bayes) Si tenemos los datos \\(X\\), escogeríamos una función \\(t_X\\) que minimice el riesgo posterior \\(\\rho(t, X)\\), y nuestro estimador es \\(\\hat{\\theta}_n = t_X (X)\\). (Regla de Bayes) Si no tenemos los datos, escogeríamos el estimador una función \\(t\\) que minimice el riesgo de Bayes \\(r(t)\\), y estimaríamos usando \\(\\hat{\\theta}_n = t(X)\\) Pero como el riesgo de Bayes es el promedio del riesgo posterior, la solución de 1 nos da la solución de 2. Es decir, el estimador que escogemos condicional a los datos \\(X\\) es el mismo que escogeríamos antes de escoger los datos, dada una distribución inicial \\(p(\\theta).\\) Por ejemplo, es posible demostrar que bajo la pérdida cuadrática, la regla de Bayes es utilizar la media posterior, bajo la pérdida absoluta, la mediana posterior, etc. Este estimador de Bayes tiene sentido desde el punto de vista frecuentista, también, porque minimiza el riesgo frecuentista promedio, suponiendo la inicial \\(p(\\theta)\\). Por ejemplo, para la pérdida cuadrática podemos usar la descomposición de sesgo y varianza y obtenemos: \\[r(t) = \\mathbb{E}[R(\\theta,t)] = \\mathbb{E}[ \\mathsf{Sesgo}_\\theta^2(t)] +\\mathbb{E}[\\mathsf{Var}_\\theta(t)] \\] Podemos ver entonces que el estimador de Bayes, en este caso la media posterior, es resultado de minimizar la suma de estas dos cantidades: por eso puede incurrir en sesgo, si ese costo se subsana con una reducción considerable de la varianza. Los estimadores insesgados que vimos en esta sección fueron subóptimos en muchos casos justamente porque son insesgados, e incurren en varianza grande. Regresa a los ejemplos anteriores donde evaluamos el desempeño de la media posterior en varios ejemplos. Muestra en las gráficas dónde ves el balance entre sesgo y varianza que cumplen cuando los comparamos con estimadores insesgados. Desde el punto de vista frecuentista, la cuestión es más complicada y hay varias maneras de proceder. En primer lugar, comenzaríamos con el riesgo frecuentista \\(R(\\theta, t)\\). Una idea es,por ejemplo, calcular el riesgo máximo: \\[R_{\\max} (t) = \\underset{\\theta}{\\max} R(\\theta, t).\\] En nuestro ejemplo de arriba el máximo se alcanza en 0.5, y tomaríamos eso evaluación de los estimadores \\(\\theta_1\\) o \\(\\theta_2\\). Buscaríamos entonces estimadores que minimicen este máximo, es decir, estimadores minimax. Pero también es posible enfocar este problema considerando sólo estimadores insesgados, lo que nos lleva por ejemplo a buscar estimadores con mínima varianza. También podemos enfocarnos en buscar estimador admisibles, que son aquellos cuyo riesgo no está dominado para toda \\(\\theta\\) por otro estimador, y así sucesivamente. Finalmente, es posible demostrar (ver Wasserman (2013)) que típicamente, para muestras grandes, el estimador de máxima verosimilitud es cercano a ser minimax y además es una regla de Bayes. Estas son buenas propiedades, pero debemos contar con que el régimen asintótico se cumpla aproximadamente. Referencias "],["métodos-de-cadenas-de-markov-monte-carlo.html", "Sección 12 Métodos de Cadenas de Markov Monte Carlo Integrales mediante subdivisiones Métodos Monte Carlo Simulando de la posterior Ejemplo de islas ¿Por qué funciona Metrópolis? Método de Metrópolis Ajustando el tamaño de salto Metrópolis con varios parámetros Muestreador de Gibbs Conclusiones y observaciones Metrópolis y Gibbs HMC y Stan Diagnósticos generales para MCMC", " Sección 12 Métodos de Cadenas de Markov Monte Carlo Hasta ahora, hemos considerado modelos bayesianos conjugados, donde la posterior tiene una forma conocida. Esto nos permitió simular directamente de la posterior usando las rutinas estándar de R, o utilizar cálculos teóricos o funciones estándar de R para calcular resúmenes de interés, como medias o medianas posteriores o intervalos de credibilidad. Sin embargo, en aplicaciones rara vez es factible este tipo de análisis tan simple, pues: Los modelos que estamos considerando son más complejos y la distribución posterior conjunta de los parámetros no tiene una forma simple conocida. Queremos usar distribuciones iniciales que no son conjugadas para utilizar correctamente nuestra información inicial. Recordamos que tenemos expresiones explícitas para la inicial \\(p(\\theta)\\) y la verosimilitud \\(p(x|\\theta)\\), así que conocemos explícitamente la posterior, módulo la constante de normalización, \\[p(\\theta|x) \\propto p(x|\\theta) \\, p(\\theta).\\] Supongamos por ejemplo que quisiéramos calcular las medias posteriores de los parámetros \\(\\theta\\). En teoría, tendríamos que calcular \\[\\hat \\theta = \\mathbb{E}[{\\theta}\\, |\\, x] = \\int \\theta \\, p(\\theta|x) \\, d\\theta\\] Entonces es necesario calcular también \\(p(x)\\), que resulta de la integral \\[p(x) = \\int p(x|\\theta) \\, p(\\theta)\\, d\\theta\\] Si no tenemos expresiones analíticas simples, tendremos que aproximar numéricamente estas integrales de alguna forma. Si la posterior tiene una forma conocida, podemos calcular cantidades de interés usando fórmulas o rutinas de simulación de distribuciones conocidas que producen muestras independientes. Cuando la posterior no tiene una forma conocida, sin embargo: Podemos intentar usar integración numérica usual. Como veremos, este enfoque no es muy escalable. Podemos usar simulaciones bajo cadenas de Markov (Markov Chain Monte Carlo, MCMC), que es un enfoque más escalable. Mucho del uso generalizado actual de la estadística bayesiana se debe a que gracias al poder de cómputo disponible y los métodos MCMC, no estamos restringidos al uso de 1 y 2, que tienen desventajas grandes. Primero mostraremos cómo el método de integración por subdivisión no es escalable. Integrales mediante subdivisiones Como tenemos una expresión analítica para el integrando, podemos intentar una rutina numérica de integración. Una vez calculada, podríamos entonces usar otra rutina numérica para calcular las medias posteriores \\(\\hat{\\theta}\\). Las rutinas usuales de integración pueden sernos útiles cuando el número de parámetros es chico. Consideremos primero el caso de 1 dimensión, y supongamos que \\(a\\leq\\theta\\leq b\\). Si dividimos el rango de \\(\\theta\\) en intervalos determinados por \\(a = \\theta^1&lt;\\theta^2&lt;\\cdots \\theta^M =b\\), tales que \\(\\Delta\\theta = \\theta^{i+1} -\\theta^{i}\\), podríamos aproximar con \\[p(x) \\approx \\sum_{i=1}^M p(x|\\theta^i)p(\\theta^i) \\Delta\\theta\\] Lo que requiere \\(M\\) evaluaciones del factor \\(p(x|\\theta)p(\\theta)\\). Podríamos usar por ejemplo \\(M=100\\) para tener precisión razonable. Ejemplo: estimación de una proporción Teníamos que \\(p(S_n = k|\\theta) \\propto \\theta^k(1-\\theta)^{n-k}\\) cuando observamos \\(k\\) éxitos en \\(n\\) pruebas independientes. Supongamos que nuestra inicial es \\(p(\\theta) = 2\\theta\\) (checa que es una densidad), es decir, creemos que es más probable a priori observar proporciones altas. Podemos integrar numéricamente crear_log_post &lt;- function(n, k){ function(theta){ verosim &lt;- k * log(theta) + (n - k) * log(1 - theta) inicial &lt;- log(theta) log_p_factor &lt;- verosim + inicial log_p_factor } } # observamos 3 éxitos en 4 pruebas: log_post &lt;- crear_log_post(4, 3) prob_post &lt;- function(x) { exp(log_post(x))} # integramos numéricamente p_x &lt;- integrate(prob_post, lower = 0, upper = 1, subdivisions = 100L) p_x ## 0.03333333 with absolute error &lt; 3.7e-16 Y ahora podemos calcular la media posterior: media_funcion &lt;- function(theta){ theta * prob_post(theta) / p_x$value } integral_media &lt;- integrate(media_funcion, lower = 0, upper = 1, subdivisions = 100L) media_post &lt;- integral_media$value media_post ## [1] 0.7142857 Podemos verificar nuestro trabajo pues sabemos que la posterior es \\(\\mathsf{Beta}(5, 2)\\) cuya media es 5/(2+5) ## [1] 0.7142857 Y podríamos intentar una estrategia similar, por ejemplo, para calcular intervalos de credibilidad. Sin embargo, veremos abajo que este método no escala con el número de parámetros. Más de un parámetro Ahora supongamos que tenemos \\(2\\) parámetros. Dividiríamos cada parámetro en 100 intervalos, y luego tendríamos que calcular \\[p(x) \\approx \\sum_{i=1}^M \\sum_{j=1}^M p(x|\\theta_1^i, \\theta_2^j)p(\\theta_1^i, \\theta_2^j) \\Delta\\theta_1\\Delta\\theta_2\\] Y esto requeriría \\(M^2 = 10,000\\) evaluaciones de \\(p(x|\\theta)p(\\theta)\\). Si tenemos \\(p\\) parámetros, entonces tendríamos que hacer \\(M^p\\) evaluaciones de la posterior. Incluso cuando \\(p=10\\), esta estrategia es infactible, pues tendríamos que hacer más de millones de millones de millones de evaluaciones de la posterior. Si sólo tenemos esta técnica disponible, el análisis bayesiano está considerablemente restringido. Regresión bayesiana con unas 10 covariables por ejemplo, no podría hacerse. De modo que tenemos que replantearnos cómo atacar el problema de calcular o aproximar estas integrales. Métodos Monte Carlo En varias ocasiones anteriormente hemos usado el método Monte Carlo para aproximar integrales: por ejemplo, para calcular medias posteriores. Supongamos que tenemos una densidad \\(p(\\theta)\\). Integración Monte Carlo. Supongamos que queremos calcular el valor esperado de \\(g(X)\\), donde \\(X\\sim p(X\\,|\\,\\theta).\\) Es decir, la variable aleatoria \\(X\\) se distribuye de acuerdo al modelo probabilistico \\(p(X \\, | \\, \\theta),\\) de tal forma que lo que nos interesa calcular es \\[\\mathbb{E}[g(X)] = \\int g(x) p(x|\\theta)\\, dx.\\] Si tomamos una muestra \\(x^{(1)},x^{(2)}, \\ldots x^{(N)} \\overset{iid}{\\sim} p(x|\\theta)\\), entonces \\[\\mathbb{E}[g(X)] \\approx \\, \\frac1N \\, \\sum_{n = 1}^N g(x^{(n)})\\] cuando \\(N\\) es grande. Esto es simplemente una manera de escribir la ley de los grandes números, y hemos aplicado este teorema en varias ocasiones. Nos ha sido útil cuando sabemos cómo simular de distribución \\(p(\\theta | x)\\) (usando alguna rutina de R, por ejemplo, o usando un método estándar como inversión de la función de distribución acumulada). Ejemplo En este ejemplo repetimos cosas que ya hemos visto. En el caso de estimación de una proporción \\(\\theta\\), tenemos como inicial \\(p(\\theta) \\propto \\theta\\), que es \\(\\mathsf{Beta}(2,1)\\). Si observamos 3 éxitos en 4 pruebas, entonces sabemos que la posterior es \\(p(\\theta|x)\\propto \\theta^4(1-\\theta)\\), que es \\(\\mathsf{Beta}(5, 2)\\). Si queremos calcular media y segundo momento posterior para \\(\\theta\\), en teoría necesitamos calcular \\[\\mu = \\int_0^1 \\theta p(\\theta|X = 3)\\, d\\theta,\\,\\, \\mu_2=\\int_0^1 \\theta^2 p(\\theta|X = 3)\\, d\\theta\\] integramos con Monte Carlo theta &lt;- rbeta(10000, 5, 2) media_post &lt;- mean(theta) momento_2_post &lt;- mean(theta^2) c(media_post, momento_2_post) ## [1] 0.7147007 0.5364443 Y podemos aproximar de esta manera cualquier cantidad de interés que esté basada en integrales, como probabilidades asociadas a \\(\\theta\\) o cuantiles asociados. Por ejemplo, podemos aproximar fácilmente \\(P(e^{\\theta}&gt; 2|x)\\) haciendo mean(exp(theta) &gt; 2) ## [1] 0.5959 y así sucesivamente. Este enfoque, sin embargo, es mucho más flexible y poderoso. Ejemplo: varias pruebas independientes Supongamos que probamos el nivel de gusto para 4 sabores distintos de una paleta. Usamos 4 muestras de aproximadamente 50 personas diferentes para cada sabor, y cada uno evalúa si le gustó mucho o no. Obtenemos los siguientes resultados: datos &lt;- tibble( sabor = c(&quot;fresa&quot;, &quot;limón&quot;, &quot;mango&quot;, &quot;guanábana&quot;), n = c(50, 45, 51, 50), gusto = c(36, 35, 42, 29)) %&gt;% mutate(prop_gust = gusto / n) datos ## # A tibble: 4 × 4 ## sabor n gusto prop_gust ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 fresa 50 36 0.72 ## 2 limón 45 35 0.778 ## 3 mango 51 42 0.824 ## 4 guanábana 50 29 0.58 Usaremos como inicial \\(\\mathsf{Beta}(2, 1)\\) (pues hemos obervado cierto sesgo de cortesía en la calificación de sabores, y no es tan probable tener valores muy bajos) para todos los sabores, es decir \\(p(\\theta_i)\\) es la funcion de densidad de una \\(\\mathsf{Beta}(2, 1)\\). La inicial conjunta la definimos entonces, usando idependiencia inicial, como \\[p(\\theta_1,\\theta_2, \\theta_3,\\theta_4) = p(\\theta_1)p(\\theta_2)p(\\theta_3)p(\\theta_4).\\] Pues inicialmente establecemos que ningún parámetro da información sobre otro: saber que mango es muy gustado no nos dice nada acerca del gusto por fresa. Bajo este supuesto, y el supuesto adicional de que las muestras de cada sabor son independientes, podemos mostrar que las posteriores son independientes: \\[p(\\theta_1,\\theta_2,\\theta_3, \\theta_4|k_1,k_2,k_3,k_4) = p(\\theta_4|k_1)p(\\theta_4|k_2)p(\\theta_4|k_3)p(\\theta_4|k_4)\\] De forma que podemos trabajar individualmente con cada muestra. Calculamos los parámetros de las posteriores individuales: datos &lt;- datos %&gt;% mutate(a_post = gusto + 2, b_post = n - gusto + 1) datos ## # A tibble: 4 × 6 ## sabor n gusto prop_gust a_post b_post ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 fresa 50 36 0.72 38 15 ## 2 limón 45 35 0.778 37 11 ## 3 mango 51 42 0.824 44 10 ## 4 guanábana 50 29 0.58 31 22 Ahora nos preguntamos, ¿cuál es la probabilidad posterior de que mango sea el sabor más preferido de la población? Conocemos la posterior para cada parámetro, y sabemos que los parámetros son independientes para la posterior. Eso quiere decir que podemos simular de cada parámetro independientemente para obtener simulaciones de la conjunta posterior. simular_conjunta &lt;- function(rep, datos){ datos %&gt;% mutate(valor_sim = map2_dbl(a_post, b_post, ~ rbeta(1, .x, .y))) %&gt;% select(sabor, valor_sim) } simular_conjunta(1, datos) ## # A tibble: 4 × 2 ## sabor valor_sim ## &lt;chr&gt; &lt;dbl&gt; ## 1 fresa 0.886 ## 2 limón 0.729 ## 3 mango 0.701 ## 4 guanábana 0.493 # esta no es una manera muy rápida, podríamos calcular todas las # simulaciones de cada parámetro de manera vectorizada sims_posterior &lt;- tibble(rep = 1:5000) %&gt;% mutate(sims = map(rep, ~ simular_conjunta(.x, datos))) %&gt;% unnest(cols = sims) sims_posterior ## # A tibble: 20,000 × 3 ## rep sabor valor_sim ## &lt;int&gt; &lt;chr&gt; &lt;dbl&gt; ## 1 1 fresa 0.727 ## 2 1 limón 0.823 ## 3 1 mango 0.849 ## 4 1 guanábana 0.474 ## 5 2 fresa 0.659 ## 6 2 limón 0.785 ## 7 2 mango 0.866 ## 8 2 guanábana 0.631 ## 9 3 fresa 0.553 ## 10 3 limón 0.719 ## # ℹ 19,990 more rows Y ahora podemos aproximar fácilmente la probabilidad de interés: sims_posterior %&gt;% group_by(rep) %&gt;% mutate(sabor = sabor[which.max(valor_sim)]) %&gt;% group_by(sabor) %&gt;% count() %&gt;% ungroup() %&gt;% mutate(prop = n / sum(n)) ## # A tibble: 4 × 3 ## sabor n prop ## &lt;chr&gt; &lt;int&gt; &lt;dbl&gt; ## 1 fresa 1264 0.0632 ## 2 guanábana 20 0.001 ## 3 limón 5424 0.271 ## 4 mango 13292 0.665 Y vemos que los mejores sabores son mango y limón. La probabilidad posterior de que mango sea el sabor preferido por la población es de 66%. La integral correspondiente no es trivial. ¿Cuáles son las probabilidades a priori de que cada sabor sea el preferido por la población? ¿Cuál es la integral correspondiente a las probabilidades que acabamos de calcular? ¿Qué tan fácil es hacer esta integral de manera analítica? Calcula la probabilidad de que mango sea preferido a limón? ¿Qué conclusión práctica sacas de estos resultados? Simulando de la posterior Hemos establecido que podemos contestar varias preguntas de inferencia usando simulación Monte Carlo, y que este enfoque es potencialmente escalable (en contraste con métodos de integración numérica por cuadrícula). Ahora el problema que necesitamos resolver es el siguiente: Conocemos \\(p(\\theta |x)\\) módulo una constante de integración. En general, \\(p(\\theta|x)\\) no tiene una forma reconocible que corresponda a un simulador estándar. ¿Cómo simulamos de esta posterior cuando sólo sabemos calcular \\(p(x|\\theta)p(\\theta)\\)? Hay varias maneras de hacer esto. Presentaremos los algoritmos en términos de una distribución cualquiera \\(p(\\theta) = K f(\\theta)\\), donde sólo conocemos la función \\(f(\\theta)\\). Ejemplo de islas Comenzamos revisando el ejemplo de las islas en Kruschke (2015) (7.2) para tener más intuición de cómo funciona este algoritmo. ¿Por qué funciona Metrópolis? Supongamos que un vendedor de Yakult trabaja a lo largo de una cadena de islas: Constantemente viaja entre las islas ofreciendo sus productos; Al final de un día de trabajo decide si permanece en la misma isla o se transporta a una de las \\(2\\) islas vecinas; El vendedor ignora la distribución de la población en las islas y el número total de islas; sin embargo, una vez que se encuentra en una isla puede investigar la población de la misma y también de la isla a la que se propone viajar después. El objetivo del vendedor es visitar las islas de manera proporcional a la población de cada una. Con esto en mente el vendedor utiliza el siguiente proceso: Lanza un volado, si el resultado es águila se propone ir a la isla del lado izquierdo de su ubicación actual y si es sol a la del lado derecho. Si la isla propuesta en el paso anterior tiene población mayor a la población de la isla actual, el vendedor decide viajar a ella. Si la isla vecina tiene población menor, entonces visita la isla propuesta con una probabilidad que depende de la población de las islas. Sea \\(P^*\\) la población de la isla propuesta y \\(P_{t}\\) la población de la isla actual. Entonces el vendedor cambia de isla con probabilidad \\[q_{mover}=P^*/P_{t}\\] A la larga, si el vendedor sigue la heurística anterior la probabilidad de que el vendedor este en alguna de las islas coincide con la población relativa de la isla. islas &lt;- tibble(islas = 1:10, pob = 1:10) camina_isla &lt;- function(i){ # i: isla actual u &lt;- runif(1) # volado v &lt;- ifelse(u &lt; 0.5, i - 1, i + 1) # isla vecina (índice) if (v &lt; 1 | v &gt; 10) { # si estás en los extremos y el volado indica salir return(i) } p_move = ifelse(islas$pob[v] &gt; islas$pob[i], 1, islas$pob[v] / islas$pob[i]) u2 &lt;- runif(1) if (p_move &gt; u2) { return(v) # isla destino } else { return(i) # me quedo en la misma isla } } pasos &lt;- 100000 iteraciones &lt;- numeric(pasos) iteraciones[1] &lt;- sample(1:10, 1) # isla inicial for (j in 2:pasos) { iteraciones[j] &lt;- camina_isla(iteraciones[j - 1]) } caminata &lt;- tibble(pasos = 1:pasos, isla = iteraciones) plot_caminata &lt;- ggplot(caminata[1:1000, ], aes(x = pasos, y = isla)) + geom_point(size = 0.8) + geom_path(alpha = 0.5) + coord_flip() + labs(title = &quot;Caminata aleatoria&quot;) + scale_y_continuous(expression(theta), breaks = 1:10) + scale_x_continuous(&quot;Tiempo&quot;) plot_dist &lt;- ggplot(caminata, aes(x = isla)) + geom_histogram() + scale_x_continuous(expression(theta), breaks = 1:10) + labs(title = &quot;Distribución objetivo&quot;, y = expression(P(theta))) plot_caminata / plot_dist Entonces: Para aproximar la distribución objetivo debemos permitir que el vendedor recorra las islas durante una sucesión larga de pasos y registramos sus visitas. Nuestra aproximación de la distribución es justamente el registro de sus visitas. Más aún, debemos tener cuidado y excluir la porción de las visitas que se encuentran bajo la influencia de la posición inicial. Esto es, debemos excluir el periodo de calentamiento. Una vez que tenemos un registro largo de los viajes del vendedor (excluyendo el calentamiento) podemos aproximar la distribución objetivo simplemente contando el número relativo de veces que el vendedor visitó dicha isla. t &lt;- c(1:10, 20, 50, 100, 200, 1000, 5000) plots_list &lt;- map(t, function(i){ ggplot(caminata[1:i, ], aes(x = isla)) + geom_histogram() + labs(y = &quot;&quot;, x = &quot;&quot;, title = paste(&quot;t = &quot;, i, sep = &quot;&quot;)) + scale_x_continuous(expression(theta), breaks = 1:10, limits = c(0, 11)) }) wrap_plots(plots_list) Escribamos el algoritmo, para esto indexamos las islas por el valor \\(\\theta\\), es así que la isla del extremo oeste corresponde a \\(\\theta=1\\) y la población relativa de cada isla es \\(P(\\theta)\\): El vendedor se ubica en \\(\\theta^{(i)}\\) y propone moverse a la izquierda o derecha con probabilidad \\(0.5\\). El rango de los posibles valores para moverse, y la probabilidad de proponer cada uno se conoce como distribución propuesta, en nuestro ejemplo sólo toma dos valores cada uno con probabilidad \\(0.5\\). Una vez que se propone un movimiento, decidimos si aceptarlo. La decisión de aceptar se basa en el valor de la distribución objetivo en la posición propuesta, relativo al valor de la distribución objetivo en la posición actual: \\[\\alpha=\\min\\bigg\\{\\frac{P(\\theta^*)}{P(\\theta^{(i)})},1\\bigg\\},\\] donde \\(\\alpha\\) denota la probabilidad de hacer el cambio de isla. Notemos que la distribución objetivo \\(P(\\theta)\\) no necesita estar normalizada, esto es porque lo que nos interesa es el cociente \\(P(\\theta^*)/P(\\theta^{(i)})\\). Una vez que propusimos un movimiento y calculamos la probabilidad de aceptar el movimiento aceptamos o rechazamos el movimiento generando un valor de una distribución uniforme, si dicho valor es menor a la probabilidad de cambio, \\(\\alpha,\\) entonces hacemos el movimiento. Entonces, para utilizar el algoritmo necesitamos ser capaces de: Generar un valor de la distribución propuesta, que hemos denotado por \\(q,\\) (para crear \\(\\theta^*\\)). Evaluar la distribución objetivo en cualquier valor propuesto (para calcular \\(P(\\theta^*)/P(\\theta^{(i)})\\)). Generar un valor uniforme (para movernos con probabilidad \\(\\alpha\\)). Las \\(3\\) puntos anteriores nos permiten generar muestras aleatorias de la distribución objetivo, sin importar si esta está normalizada. Esta técnica es particularmente útil cuando cuando la distribución objetivo es una posterior proporcional a \\(p(x|\\theta)p(\\theta)\\). Para entender porque funciona el algoritmo de Metrópolis hace falta entender \\(2\\) puntos, primero que la distribución objetivo es estable: si la probabilidad actual de ubicarse en una posición coincide con la probabilidad en la distribución objetivo, entonces el algoritmo preserva las probabilidades. library(expm) transMat &lt;- function(P){ # recibe vector de probabilidades (o población) T &lt;- matrix(0, 10, 10) n &lt;- length(P - 1) # número de estados for (j in 2:n - 1) { # llenamos por fila T[j, j - 1] &lt;- 0.5 * min(P[j - 1] / P[j], 1) T[j, j] &lt;- 0.5 * (1 - min(P[j - 1] / P[j], 1)) + 0.5 * (1 - min(P[j + 1] / P[j], 1)) T[j, j + 1] &lt;- 0.5 * min(P[j + 1] / P[j], 1) } # faltan los casos j = 1 y j = n T[1, 1] &lt;- 0.5 + 0.5 * (1 - min(P[2] / P[1], 1)) T[1, 2] &lt;- 0.5 * min(P[2] / P[1], 1) T[n, n] &lt;- 0.5 + 0.5 * (1 - min(P[n - 1] / P[n], 1)) T[n, n - 1] &lt;- 0.5 * min(P[n - 1] / P[n], 1) T } T &lt;- transMat(islas$pob) w &lt;- c(0, 1, rep(0, 8)) t &lt;- c(1:10, 20, 50, 100, 200, 1000, 5000) expT &lt;- map_df(t, ~data.frame(t = ., w %*% (T %^% .))) expT_long &lt;- expT %&gt;% gather(theta, P, -t) %&gt;% mutate(theta = parse_number(theta)) ggplot(expT_long, aes(x = theta, y = P)) + geom_bar(stat = &quot;identity&quot;, fill = &quot;darkgray&quot;) + facet_wrap(~ t) + scale_x_continuous(expression(theta), breaks = 1:10, limits = c(0, 11)) El segundo punto es que el proceso converge a la distribución objetivo. Podemos ver, (en nuestro ejemplo sencillo) que sin importar el punto de inicio se alcanza la distribución objetivo. inicio_p &lt;- function(i){ w &lt;- rep(0, 10) w[i] &lt;- 1 t &lt;- c(1, 10, 50, 100) exp_t &lt;- map_df(t, ~ data.frame(t = .x, inicio = i, w %*% (T %^% .))) %&gt;% gather(theta, P, -t, -inicio) %&gt;% mutate(theta = parse_number(theta)) exp_t } exp_t &lt;- map_df(c(1, 3, 5, 9), inicio_p) ggplot(exp_t, aes(x = as.numeric(theta), y = P)) + geom_bar(stat = &quot;identity&quot;, fill = &quot;darkgray&quot;) + facet_grid(inicio ~ t) + scale_x_continuous(expression(theta), breaks = 1:10, limits = c(0, 11)) Método de Metrópolis En el método de Metrópolis, uno de los más antiguos, comenzamos con un valor inicial de los parámetros \\(\\theta^{(0)}\\) en el soporte de \\(p(\\theta)\\), es decir \\(p(\\theta^{(0)})&gt;0.\\) Para \\(i=1, \\ldots, M\\), hacemos: Partiendo de \\(\\theta^{(i)}\\), hacemos un salto corto en una dirección al azar para obtener una propuesta \\(\\theta^* \\sim q(\\theta \\, |\\, \\theta^{(i)}).\\) Aceptamos or rechazamos el salto: Si \\(\\alpha = \\frac{f(\\theta^*)}{f(\\theta^{(i)})} \\geq 1\\), aceptamos el salto y ponemos \\(\\theta^{(i+1)}=\\theta^*\\). Regresamos a 1 para la siguiente iteración \\(i\\leftarrow i + 1.\\) Si \\(\\alpha = \\frac{f(\\theta^*)}{f(\\theta^{(i)})} &lt; 1\\), entonces aceptamos con probabilidad \\(\\alpha\\) el salto, ponemos \\(\\theta^{(i+1)}=\\theta^*\\) y regresamos a 1 para la siguiente iteración \\(i\\leftarrow i + 1\\). Si rechazamos el salto, ponemos entonces \\(\\theta^{(i+1)}=\\theta^{(i)}\\) y regresamos a 1 para la siguiente iteración \\(i\\leftarrow i + 1.\\) Requerimos también que la función que propone los saltos sea simétrica: es decir, \\(q(\\theta^*|\\theta^{(i)})\\) debe ser igual a \\(q(\\theta^{(i)}|\\theta^*)\\). Se puede modificar el algoritmo para tratar con una propuesta que no sea simétrica. Una elección común es escoger \\(q(\\theta^* |\\theta^{(i)})\\) como una \\(\\mathsf{N}(\\theta^{(i)}, \\sigma_{salto})\\). En este curso, escribiremos varios métodos de cadenas de Markov para estimación Monte Carlo (Markov Chain Monte Carlo, MCMC) desde cero para entender los básicos de cómo funciona. Sin embargo, en la práctica no hacemos esto, sino que usamos software estándar (Stan, JAGS, BUGS, etc.) para hacer este trabajo. Expertos en MCMC, métodos numéricos, y estadística a veces escriben partes de sus rutinas de simulación, y pueden lograr mejoras de desempeño considerables. Excepto para modelos simples, esto no es trivial de hacer garantizando resultados correctos. En resumen, todo el código de esta sección es de carácter ilustrativo. Utiliza implementaciones establecidas en las aplicaciones. Abajo implementamos el algoritmo con un salto de tipo normal: crear_metropolis &lt;- function(fun_log, sigma_salto = 0.1){ # la entrada es la log posterior iterar_metropolis &lt;- function(theta_inicial, n){ p &lt;- length(theta_inicial) nombres &lt;- names(theta_inicial) iteraciones &lt;- matrix(0, nrow = n, ncol = p) colnames(iteraciones) &lt;- nombres iteraciones[1,] &lt;- theta_inicial for(i in 2:n){ theta &lt;- iteraciones[i - 1, ] theta_prop &lt;- theta + rnorm(p, 0, sigma_salto) # exp(log(p) - log(q)) = p/q cociente &lt;- exp(fun_log(theta_prop) - fun_log(theta)) if(cociente &gt;= 1 || runif(1,0,1) &lt; cociente){ iteraciones[i, ] &lt;- theta_prop } else { iteraciones[i, ] &lt;- theta } } iteraciones_tbl &lt;- iteraciones %&gt;% as_tibble() %&gt;% mutate(iter_num = row_number()) %&gt;% select(iter_num, everything()) iteraciones_tbl } iterar_metropolis } E intentamos simular de una exponencial no normalizada: exp_no_norm &lt;- function(x) { z &lt;- ifelse(x &gt; 0, exp(-0.5 * x), 0) log(z) } iterador_metro &lt;- crear_metropolis(exp_no_norm, sigma_salto = 0.25) sims_tbl &lt;- iterador_metro(c(theta = 0.5), 50000) ggplot(sims_tbl, aes(x = theta)) + geom_histogram() Ahora probemos con una \\(\\mathsf{Beta}(3, 2):\\) beta_no_norm &lt;- function(x) { z &lt;- ifelse(x &gt; 0 &amp;&amp; x &lt; 1, (x^2)*(1-x), 0) log(z) } iterador_metro &lt;- crear_metropolis(beta_no_norm, sigma_salto = 0.04) sims_metro_tbl &lt;- iterador_metro(c(theta = 0.5), 50000) sims_indep_tbl &lt;- tibble(iter_num = 1:30000, theta = rbeta(30000, 3, 2)) g_1 &lt;- ggplot(sims_metro_tbl, aes(x = theta)) + geom_histogram() + labs(subtitle = &quot;Metrópolis&quot;) g_2 &lt;- ggplot(sims_indep_tbl, aes(x = theta)) + geom_histogram() + labs(subtitle = &quot;rbeta&quot;) g_1 + g_2 Y vemos que esto funciona. Nótese sin embargo un aspecto de estas simulaciones que no habíamos encontrado en el curso. Aunque la distribución final de las simulaciones es muy cercana a la de la distribución que queremos simular, lo cual era nuestro propósito, las simulaciones no son extracciones independientes de esa distribución. La construcción del algoritmo muestra eso, pero podemos también graficar las simulaciones: g_metropolis &lt;- sims_metro_tbl %&gt;% filter(iter_num &lt; 500) %&gt;% ggplot(aes(x = iter_num, y = theta)) + geom_line() + labs(subtitle = &quot;Metrópolis&quot;) g_indep &lt;- sims_indep_tbl %&gt;% filter(iter_num &lt; 500) %&gt;% ggplot(aes(x = iter_num, y = theta)) + geom_line() + labs(subtitle = &quot;Independientes&quot;) g_metropolis + g_indep Donde vemos claramente que las simulaciones de metropolis están autocorrelacionadas: la siguiente simulación depende de la anterior. Esto define una cadena de Markov. En cualquiera de los dos casos, como vimos en los histogramas de arriba, las simulaciones “visitan” cada parte [0,1] de manera proporcional a la densidad, de manera que podemos usar ambos tipos de simulaciones para aproximar la integral o cantidad que nos interesa. Por ejemplo, la media posterior es: media_1 &lt;- sims_metro_tbl %&gt;% summarise(media_post = mean(theta)) %&gt;% pull(media_post) media_2 &lt;- sims_indep_tbl %&gt;% summarise(media_post = mean(theta)) %&gt;% pull(media_post) media_exacta &lt;- 3/(3 + 2) tibble(metodo = c(&quot;sim Metrópolis&quot;, &quot;sim Independiente&quot;, &quot;exacto&quot;), media_post = c(media_1, media_2, media_exacta)) ## # A tibble: 3 × 2 ## metodo media_post ## &lt;chr&gt; &lt;dbl&gt; ## 1 sim Metrópolis 0.613 ## 2 sim Independiente 0.600 ## 3 exacto 0.6 Supongamos que queremos simular de una distribución \\(p(\\theta)\\), pero sólo conocemos \\(p(\\theta)\\) módulo una constante. Bajo ciertas condiciones de regularidad: El algoritmo Metrópolis para la distribución \\(p(\\theta)\\) define una cadena de Markov cuya distribución a largo plazo es \\(p(\\theta)\\). Esto implica que si \\(\\theta^{(1)},\\theta^{(2)}, \\ldots, \\theta^{(M)}\\) es una simulación de esta cadena, y \\(M\\) es suficientemente grande La distribución de las \\(\\theta^{(i)}\\) es aproximadamente \\(p(\\theta)\\), Tenemos que \\[ \\frac1M \\sum_{m = 1}^M h(\\theta^{(m)}) \\to \\int h(\\theta)p(\\theta)\\, d\\theta\\] cuando \\(M\\to \\infty\\) Observaciones: Aunque hay distintas condiciones de regularidad que pueden funcionar, generalmente el supuesto es que la cadena de Markov construída es ergódica, y hay varias condiciones que garantizan esta propiedad. Una condición simple, por ejemplo, es que el soporte de la distribución \\(p(\\theta)\\) es un conjunto conexo del espacio de parámetros. Más crucialmente, este resultado no dice qué tan grande debe ser \\(M\\) para que la aproximación sea buena. Esto depende de cómo es \\(p(\\theta)\\), y de la distribución que se utiliza para obtener los saltos propuestos. Dependiendo de estos dos factores, la convergencia puede ser rápida (exponencial) o tan lenta que es infactible usarla. Más adelante veremos diagnósticos para descartar los peores casos de falta de convergencia. Ajustando el tamaño de salto En el algoritmo Metrópolis, generalmente es importante escoger la dispersión de la distribución que genera propuestas con cuidado. Si la dispersión de la propuesta es demasiado grande, tenderemos a rechazar mucho, y la convergencia será lenta. Si la dispersión de la propuesta es demasiado chica, tardaremos mucho tiempo en explorar las distintas partes de la distribución objetivo. Ejemplo Supongamos que queremos simular usando metróplis de una distribución \\(\\textsf{Gamma}(20, 100)\\). Abajo vemos la forma de esta distribución: sim_indep &lt;- tibble(theta = rgamma(10000, 20, 100)) ggplot(sim_indep, aes(x = theta)) + geom_histogram() # logaritmo de densidad no normalizada log_f_dist &lt;- function(x) 210 + dgamma(x, 20, 100, log = TRUE) # iterar iterador_metro_chico &lt;- crear_metropolis(log_f_dist, sigma_salto = 0.001) sims_chico_tbl &lt;- iterador_metro_chico(c(theta = 0.02), 50000) g_sim &lt;- ggplot(sims_chico_tbl %&gt;% filter(iter_num &lt; 3000), aes(x = iter_num, y = theta)) + geom_line() + ylim(c(0, 0.5)) dist_bplot &lt;- ggplot(tibble(x = rgamma(10000, 20, 100)), aes(y = x, x = &quot;a&quot;)) + geom_violin() + ylab(&quot;&quot;) + ylim(0, 0.5) g_sim + dist_bplot + plot_layout(widths = c(5, 1)) Nótese que después de 5 mil iteraciones estamos muy lejos de tener una muestra que se aproxime a la distribución objetivo. Empezamos en un lugar bajo, y la cadena sólo ha ido lentamente hacia las zonas de alta densidad. Cualquier resumen con esta cadena estaría fuertemente sesgado al valor donde iniciamos la iteración. Decimos que la cadena todavía no mezcla en las primeras 5 mil iteraciones. Ahora vemos qué pasa si ponemos el tamaño de salto demasiado grande: set.seed(831) iterador_metro_grande &lt;- crear_metropolis(log_f_dist, sigma_salto = 20) sims_grande_tbl &lt;- iterador_metro_grande(c(theta = 0.02), 50000) g_sim &lt;- ggplot(sims_grande_tbl %&gt;% filter(iter_num &lt; 3000), aes(x = iter_num, y = theta)) + geom_line() + ylim(c(0, 0.5)) g_sim + dist_bplot + plot_layout(widths = c(5, 1)) En este caso, la cadena se atora muchas veces, pues las propuestas tienen probabilidad muy baja, y tendemos a tener una tasa de rechazos muy alta. Esto quiere decir que la información que tenemos acerca de la posterior es relativamente poca, pues muchos datos son repeticiones del mismo valor. Cualquier resumen con esta cadena podría estar muy lejos del verdadero valor, pues su varianza es alta - otra corrida se “atoraría” en otros valores distintos. Nótese que cualquiera de estas cadenas, si la corremos suficientemente tiempo, nos daría resultados buenos. Sin embargo, el número de simulaciones puede ser infactible. Un valor intermedio nos dará mucho mejores resultados: set.seed(831) iterador_metro_apropiada &lt;- crear_metropolis(log_f_dist, sigma_salto = 0.1) sims_tbl &lt;-iterador_metro_apropiada(c(theta = 0.02), 50000) g_sim &lt;- ggplot(sims_tbl %&gt;% filter(iter_num &lt; 3000), aes(x = iter_num, y = theta)) + geom_line() + ylim(c(0, 0.5)) g_sim + dist_bplot + plot_layout(widths = c(5, 1)) Donde vemos que esta cadena parece mezclar bien (está explorando la totalidad de la distribución objetivo), y también parece estar en un estado estable. Comparemos cómo saldría por ejemplo la media posterior aproximada según los tres métodos: estimaciones_media &lt;- map_dfr( list(sims_chico_tbl, sims_grande_tbl, sims_tbl), ~ filter(.x, iter_num &lt; 3000) %&gt;% summarise(media = mean(theta))) %&gt;% mutate(tipo = c(&quot;salto chico&quot;, &quot;salto grande&quot;, &quot;salto apropiado&quot;)) estimaciones_media %&gt;% bind_rows(tibble(tipo = &quot;exacta&quot;, media = 20/100)) %&gt;% select(tipo, media) ## # A tibble: 4 × 2 ## tipo media ## &lt;chr&gt; &lt;dbl&gt; ## 1 salto chico 0.132 ## 2 salto grande 0.190 ## 3 salto apropiado 0.203 ## 4 exacta 0.2 Veamos otra corrida: set.seed(6222131) sims_chica_tbl &lt;- iterador_metro_chico(c(theta = 0.02), 5000) sims_grande_tbl &lt;- iterador_metro_grande(c(theta = 0.02), 5000) estimaciones_media &lt;- map_dfr( list(sims_chica_tbl, sims_grande_tbl, sims_tbl), ~ filter(.x, iter_num &lt; 3000) %&gt;% summarise(media = mean(theta))) %&gt;% mutate(tipo = c(&quot;salto chico&quot;, &quot;salto grande&quot;, &quot;salto apropiado&quot;)) estimaciones_media %&gt;% bind_rows(tibble(tipo = &quot;exacta&quot;, media = 20/100)) %&gt;% select(tipo, media) ## # A tibble: 4 × 2 ## tipo media ## &lt;chr&gt; &lt;dbl&gt; ## 1 salto chico 0.124 ## 2 salto grande 0.229 ## 3 salto apropiado 0.203 ## 4 exacta 0.2 Repite este proceso varias veces. Verifica que: Si el tamaño de paso es muy chico, las estimaciones de la media tienen sesgo alto. Si el tamaño de paso es muy grande, las estimaciones tienen varianza alta. Si el tamaño de paso es adecuado, obtenemos buena precisión en la estimación de la media posterior. Explica estos tres casos en términos de la convergencia de las realizaciones de la cadena de Markov. Explica cómo afecta a cada caso el valor inicial de las simulaciones de Metrópolis. Repite para otra estadística, como la desviación estándar o el rangon intercuartil. Metrópolis con varios parámetros Ahora aplicaremos el algoritmo Metrópolis cuando tenemos varios parámetros. La idea es la misma, pero nuestra distribución de salto debe ser multivariada. Una selección usual es usando saltos normales independientes para cada parámetro, es decir, la normal multivariada con matriz de varianza y covarianza diagonal. Ejemplo: el modelo normal Veremos cómo simular con Metrópolis para el problema de los cantantes. Sabemos como calcular la posterior: crear_log_posterior_norm &lt;- function(x = datos, m_0, n_0, a, b){ # calcula log_posterior log_posterior &lt;- function(mu, sigma){ log_verosim &lt;- sum(dnorm(x, mu, sigma, log = TRUE)) tau &lt;- 1 / sigma^2 log_inicial &lt;- dgamma(tau, a, b, log = TRUE) + dnorm(mu, mu_0, sigma/sqrt(n_0), log = TRUE) log_p &lt;- log_verosim + log_inicial log_p } log_posterior } # parametros de inicial y datos a &lt;- 3 b &lt;- 140 mu_0 &lt;- 175 n_0 &lt;- 5 set.seed(3413) cantantes &lt;- lattice::singer %&gt;% mutate(estatura_cm = round(2.54 * height)) %&gt;% filter(str_detect(voice.part, &quot;Tenor&quot;)) %&gt;% sample_n(20) Vemos cómo se ven las primeras iteraciones de nuestra cadena de Markov: log_p &lt;- crear_log_posterior_norm(cantantes$estatura_cm, mu_0, n_0, a, b) log_post &lt;- function(pars) { log_p(pars[1], pars[2]) } set.seed(823) metro_normal &lt;- crear_metropolis(log_post, sigma_salto = 0.5) sim_tbl &lt;- metro_normal(c(mu = 172, sigma = 3), 50000) ggplot(sim_tbl %&gt;% filter(iter_num &lt; 100), aes(x = mu, y = sigma)) + geom_path() + geom_point() Y ahora vemos todas las simulaciones: g_normal &lt;- ggplot(sim_tbl, aes(x = mu, y = sigma)) + geom_point(alpha = 0.05)+ coord_equal() + ylim(c(0, 14)) g_normal Y las medias posteriores son: sim_tbl %&gt;% summarise(across(is_double, mean)) ## # A tibble: 1 × 2 ## mu sigma ## &lt;dbl&gt; &lt;dbl&gt; ## 1 176. 6.80 Ejemplo: observaciones normales, no conjugado Arriba repetimos el análisis conjugado usando Metrópolis. Aunque ya no es necesario usar el modelo conjugado, y podemos poner iniciales que sean más intuitivas y acorde con nuestro conocimiento existente. Por ejemplo, podemos poner \\(p(\\mu, \\sigma) = p(\\mu)p(\\sigma)\\), donde la densidad de \\(\\mu \\sim \\mathsf{N}(175, 2)\\) y \\(\\sigma \\sim \\mathsf{U}[2, 20].\\) Igual que antes, la verosimilitud \\(p(x|\\mu, \\sigma)\\) es normal con media \\(\\mu\\) y desviación estándar \\(\\sigma.\\) Escribimos la posterior: crear_log_posterior &lt;- function(x, m_0, sigma_0, inf, sup){ # calcula log_posterior log_posterior &lt;- function(mu, sigma){ log_verosim &lt;- sum(dnorm(x, mu, sigma, log = TRUE)) log_inicial &lt;- dunif(sigma, inf, sup, log = TRUE) + dnorm(mu, mu_0, sigma_0, log = TRUE) log_p &lt;- log_verosim + log_inicial log_p } log_posterior } log_p &lt;- crear_log_posterior(cantantes$estatura_cm, 175, 3, 2, 20) log_post &lt;- function(pars) { log_p(pars[1], pars[2]) } set.seed(8231) metro_normal &lt;- crear_metropolis(log_post, sigma_salto = 0.5) sim_tbl &lt;- metro_normal(c(mu = 172, sigma = 5), 50000) g_normal_2 &lt;- ggplot(sim_tbl, aes(x = mu, y = sigma)) + geom_point(alpha = 0.05) + coord_equal() + ylim(c(0, 14)) g_normal + g_normal_2 Los resultados son similares, pero en nuestras estimaciones bajo el segundo modelo, la \\(\\sigma\\) está concentrada en valores un poco más bajos que el modelo normal-gamma inversa. Las medias posteriores son: sim_tbl %&gt;% summarise(across(is.numeric, mean)) ## # A tibble: 1 × 3 ## iter_num mu sigma ## &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 25000. 176. 6.54 Nótese que la inicial para el modelo normal-gamma inversa pone muy poca probabilidad para valores bajos de \\(\\sigma\\), mientras que el segundo modelo hay un 10% de probabilidad de que la \\(\\sigma\\) sea menor que 4. tau &lt;- rgamma(5000, 3, 150) sigma &lt;- 1/sqrt(tau) quantile(sigma, c(0.01,0.1, 0.9, 0.99)) ## 1% 10% 90% 99% ## 4.219278 5.276228 11.579358 19.038529 quantile(runif(5000, 2, 25), c(0.01,0.1, 0.9, 0.99)) ## 1% 10% 90% 99% ## 2.261297 4.254128 22.691760 24.719630 Ejemplo: exámenes Recordamos un ejemplo que vimos en la sección de máxima verosimilitud. Supongamos que en una población de estudiantes tenemos dos tipos: unos llenaron un examen de opción múltiple al azar (1 de 5), y otros contestaron las preguntas intentando sacar una buena calificación. Suponemos que una vez que conocemos el tipo de estudiante, todas las preguntas tienen la misma probabilidad de ser contestadas correctamente, de manera independiente. El modelo teórico está representado por la siguiente simulación: sim_formas &lt;- function(p_azar, p_corr){ tipo &lt;- rbinom(1, 1, 1 - p_azar) if(tipo==0){ # al azar x &lt;- rbinom(1, 10, 1/5) } else { # no al azar x &lt;- rbinom(1, 10, p_corr) } x } Y una muestra se ve como sigue: set.seed(12) muestra &lt;- map_dbl(1:200, ~ sim_formas(0.35, 0.5)) qplot(muestra) Supongamos que no conocemos la probabildad de contestar correctamente ni la proporción de estudiantes que contestó al azar. ¿Como estimamos estas dos cantidades? La verosimilitud la escribimos en el ejercicio anterior en la sección de máxima verosimilitud, está dada, para las repuestas de un estudiante, por: \\[p(X = k|\\theta_{azar}, \\theta_{corr}) \\propto \\theta_{azar}(1/5)^k(4/5)^{10-k} + (1-\\theta_{azar})\\theta_{corr}^k(1-\\theta_{corr})^{10-k}\\] Suponiendo que todas las preguntas tienen la misma dificultad, y que los estudiantes que estudiaron son homogéneos (podemos discutir qué haríamos para introducir heterogeneidad que típicamente observaríamos). Creemos que la mayoría de los estudiantes no contesta al azar, así que pondremos como inicial \\[\\theta_{azar} \\sim \\mathsf{Beta}(1, 5)\\] qbeta(c(0.1, 0.9), 1, 5) %&gt;% round(2) ## [1] 0.02 0.37 Ahora tenemos que pensar en la probabilidad \\(\\theta_{corr}\\) para los estudiantes que sí estudiaron. Imaginemos que lo probamos con un estudiante que sabemos que sí estudió, y obtuvo un porcentaje de correctos de 7/10, Podríamos poner entonces (vimos 10 intentos, con 3 fracasos y 7 éxitos): \\[\\theta_{corr} \\sim \\mathsf{Beta}(7, 3)\\] Finalmente, necesitamos la conjunta inicial. Pondremos \\[p(\\theta_{azar},\\theta_{corr}) = p(\\theta_{azar})p(\\theta_{corr})\\] con lo que expresamos que inicialmente no creemos que estos dos parámetros estén relacionados. Si pensáramos, por ejemplo, que cuando hacemos exámenes difíciles menos estudiantes estudian, entonces deberíamos intentar otra conjunta. Escribimos el producto de la verosimilitud con la inicial: crear_log_posterior &lt;- function(x){ log_posterior &lt;- function(theta_azar, theta_corr){ log_verosim &lt;- sum(log(theta_azar * dbinom(x, 10, 1/5) + (1 - theta_azar) * dbinom(x, 10, theta_corr))) log_inicial &lt;- dbeta(theta_azar, 1, 5, log = TRUE) + dbeta(theta_corr, 7, 3, log = TRUE) log_post &lt;- log_verosim + log_inicial log_post } log_posterior } Creamos la función de verosimilitud con los datos log_p &lt;- crear_log_posterior(muestra) log_post &lt;- function(pars) { log_p(pars[1], pars[2]) } set.seed(8231) metro_examenes &lt;- crear_metropolis(log_post, sigma_salto = 0.02) sim_tbl &lt;- metro_examenes(c(theta_azar = 0.5, theta_corr = 0.5), 20000) g_1 &lt;- ggplot(sim_tbl, aes(x = theta_azar, y = theta_corr)) + geom_point(alpha = 0.05) + coord_equal() g_1 Nótese que hay cierta correlación entre las dos proporciones, y esto produce intervalos posteriores relativamente amplios. Esto es de esperarse, pues los datos son consistentes con una proporción relativamente chica de estudiantes que contestan al azar, y tasas de correctos más altas entre los que sí estudian, y una proporción más grande de respuestas al azar con tasas de correctos más altas. f &lt;- c(0.05, 0.5, 0.95) sim_tbl %&gt;% pivot_longer(-iter_num, names_to = &quot;parametro&quot;, values_to = &quot;valor&quot;) %&gt;% group_by(parametro) %&gt;% summarise(cuantil = quantile(valor, f), f = f) %&gt;% mutate(cuantil = round(cuantil, 2)) %&gt;% pivot_wider(names_from = f, values_from = cuantil) ## # A tibble: 2 × 4 ## # Groups: parametro [2] ## parametro `0.05` `0.5` `0.95` ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 theta_azar 0.3 0.38 0.45 ## 2 theta_corr 0.5 0.52 0.56 Muestreador de Gibbs El algoritmo de Metrópolis es muy general y se puede aplicar a una gran variedad de problemas. Sin embargo, afinar los parámetros de la distribución propuesta para que el algoritmo funcione correctamente puede ser complicado. El muestredor de Gibbs no necesita de una distribución propuesta y por lo tanto no requiere afinar estos parámetros. Para implementar un muestreador de Gibbs se necesita ser capaz de generar muestras de la distribución posterior condicional a cada uno de los parámetros individuales. Esto es, el muestreador de Gibbs permite generar muestras de la posterior: \\[p(\\theta_1,...,\\theta_p|x)\\] siempre y cuando podamos generar valores de todas las distribuciones condicionales: \\[\\theta_k \\sim p(\\theta_k|\\theta_1,...,\\theta_{k-1},\\theta_{k+1},...,\\theta_p,x).\\] El proceso del muestreador de Gibbs es una caminata aleatoria a lo largo del espacio de parámetros. La caminata inicia en un punto arbitrario y en cada tiempo el siguiente paso depende únicamente de la posición actual. Por tanto el muestredor de Gibbs es un proceso cadena de Markov vía Monte Carlo. La diferencia entre Gibbs y Metrópolis radica en como se deciden los pasos. Muestreador Gibbs En cada punto de la caminata se selecciona uno de los componentes del vector de parámetros (típicamente se cicla en orden): Supongamos que se selecciona el parámetro \\(k\\)-ésimo después de haber modificado los \\(k-1\\) anteriores, entonces obtenemos un nuevo valor para este parámetro generando una simulación de la distribución condicional \\[\\theta_k^{(i+1)} \\sim p(\\theta_k|\\theta_1^{(i+1)},\\ldots,\\theta_{k-1}^{(i+1)},\\theta_{k+1}^{(i)},\\ldots,\\theta_p^{(i)},x)\\] El nuevo valor \\(\\theta_k^{(i+1)}\\) junto con los valores \\(\\theta_1^{(i+1)},\\ldots,\\theta_{k-1}^{(i+1)},\\theta_{k+1}^{(i)},\\ldots,\\theta_p^{(i)}\\) constituyen la nueva posición en la caminata aleatoria. Seleccionamos una nueva componente \\(\\theta_{k+1}^{(i+1)}\\) y repetimos el proceso. El muestreador de Gibbs es útil cuando no podemos determinar de manera analítica la distribución conjunta y no se puede simular directamente de ella, pero sí podemos determinar todas las distribuciones condicionales y simular de ellas. Ejemplo: dos proporciones Supongamos que queremos evaluar el balanceo de dos dados de 20 lados que produce una fábrica. En particular, evaluar la probabilidad de tirar un 20, y quizá escoger el dado que nos de mayor probabilidad de tirar un 20. Tiramos cada dado \\(n\\) veces, y denotamos por \\(X_1\\) y \\(X_2\\) el número de 20’s que tiramos en cada ocasión. El modelo de datos está dado por \\[p(x_1, x_2|\\theta_1, \\theta_2)\\propto \\theta_1^{x_1}(1-\\theta_1)^{n - x_1}\\theta_2^{x_2}(1-\\theta_2)^{n - x_2},\\] que es el producto de dos densidades binomiales, pues suponemos que las tiradas son independientes cuando conocemos los parámetros \\(\\theta_1\\) y \\(\\theta_2\\). Ahora ponemos una inicial \\[p(\\theta_i)\\sim \\mathsf{Beta}(100, 1900)\\] y aquí están las razones de nuestra elección: media &lt;- 1/20 k &lt;- 2000 a &lt;- media * k b &lt;- (1 - media) * k c(a,b) ## [1] 100 1900 qbeta(c(0.05, 0.95), a, b) %&gt;% round(3) ## [1] 0.042 0.058 y suponemos que \\[p(\\theta_1,\\theta_2) = p (\\theta_1)p(\\theta_2)\\] es decir, apriori saber el desempeño de un dado no nos da información adicional del otro (esto podría no ser cierto, por ejemplo, si el defecto es provocado por la impresión del número 20). Por lo tanto, la posterior es \\[p(\\theta_1,\\theta_2|x_1, x_2)\\propto \\theta_1^{x_1+100-1}(1-\\theta_1)^{n - x_1 + 1900-1}\\theta_2^{x_2+100 -1}(1-\\theta_2)^{n - x_2 + 1900-1}\\] Ahora consideramoso qué pasa cuando conocemos \\(\\theta_2\\) y los datos. Pensamos en todo lo que no sea \\(\\theta_1\\) como constante de modo que nos queda: \\[p(\\theta_1 | \\theta_2, x) \\propto \\theta_1^{x_1+100 -1}(1-\\theta_1)^{n - x_1 + 1900 -1}\\] que es \\(\\mathsf{Beta}(x_1 + 100, n - x_1 + 1900)\\), y por la misma razón, \\[p(\\theta_2 | \\theta_1, x) \\propto \\theta_2^{x_2+100-1}(1-\\theta_2)^{n - x_2 + 1900-1}\\] que también es es \\(\\mathsf{Beta}(x_1 + 100, n - x_1 + 1900)\\) De hecho, estas condicionales son fáciles de deducir de otra manera: en realidad estamos haciendo dos experimentos separados (pues suponemos que las iniciales son independientes y las pruebas también), así que podriamos usar el análisis Beta-Binomial para cada uno de ellos. En realidad no es necesario usar MCMC para este ejemplo. Usaremos esta función para hacer nuestras iteraciones de Gibbs: iterar_gibbs &lt;- function(pasos, n, x_1, x_2){ iteraciones &lt;- matrix(0, nrow = pasos + 1, ncol = 2) # vector guardará las simulaciones iteraciones[1, 1] &lt;- 0.5 # valor inicial media colnames(iteraciones) &lt;- c(&quot;theta_1&quot;, &quot;theta_2&quot;) # Generamos la caminata aleatoria for (j in seq(2, pasos, 2)) { # theta_1 a &lt;- x_2 + 100 - 1 b &lt;- n - x_2 + 1900 - 1 iteraciones[j, &quot;theta_2&quot;] &lt;- rbeta(1, a, b) # Actualizar theta_1 iteraciones[j, &quot;theta_1&quot;] &lt;- iteraciones[j-1, &quot;theta_1&quot;] # theta_2 a &lt;- x_1 + 100 - 1 b &lt;- n - x_1 + 1900 - 1 iteraciones[j + 1, &quot;theta_1&quot;] &lt;- rbeta(1, a, b) # Actualizar theta_1 iteraciones[j + 1, &quot;theta_2&quot;] &lt;- iteraciones[j, &quot;theta_2&quot;] } iteraciones } Y supongamos que estamos comparando los dados de dos compañías: Chessex y GameScience. Tiramos cada dado 10 mil veces, y obtenemos: # Datos de https://www.awesomedice.com/blogs/news/d20-dice-randomness-test-chessex-vs-gamescience n &lt;- 10000 x_1 &lt;- 408 # Chessex, alrededor de 0.85 dólares por dado x_2 &lt;- 474 # GameScience, alrededor 1.60 dólares por dado E iteramos: iteraciones &lt;- iterar_gibbs(20000, n, x_1, x_2) %&gt;% as_tibble() %&gt;% mutate(iter_num = row_number()) head(iteraciones) ## # A tibble: 6 × 3 ## theta_1 theta_2 iter_num ## &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; ## 1 0.5 0 1 ## 2 0.5 0.0479 2 ## 3 0.0442 0.0479 3 ## 4 0.0442 0.0452 4 ## 5 0.0411 0.0452 5 ## 6 0.0411 0.0505 6 ggplot(filter(iteraciones, iter_num &gt; 1000, iter_num&lt; 1050), aes(x = theta_1, y = theta_2)) + geom_path(alpha = 0.3) + geom_point() g_1 &lt;- ggplot(iteraciones, aes(x = theta_1, y = theta_2)) + geom_path(alpha = 0.3) + geom_point() g_2 &lt;- ggplot(iteraciones %&gt;% filter(iter_num &gt; 10), aes(x = theta_1, y = theta_2)) + geom_path(alpha = 0.3) + geom_point() + geom_abline(colour = &quot;red&quot;) + geom_point(data= tibble(theta_1=1/20, theta_2=1/20), colour = &quot;red&quot;, size = 5) g_1 + g_2 Notamos el dado de Cheesex no es consistente con 1/20 de tiros de 20s, pero el dado de GameScience sí lo es. De este gráfica vemos que Cheesex está sesgado hacia abajo, así que deberíamos escoger el dado de GameScience Podemos ver directamente cómo se distribuye la diferencia \\(\\theta_1 - \\theta_2\\). Cualquier estadística es fácil de evaluar, pues simplemente la calculamos para cada simulación y después resumimos: iteraciones &lt;- iteraciones %&gt;% mutate(dif = theta_1 - theta_2) ggplot(iteraciones %&gt;% filter(iter_num &gt; 10), aes(x = dif)) + geom_histogram(bins = 100) + geom_vline(xintercept = 0, colour = &quot;red&quot;) Y vemos que es altamente probable que el dado de Cheesex produce más 20’s que el dado de GameScience. iteraciones %&gt;% mutate(theta_1_mayor = dif &gt; 0) %&gt;% summarise(prob_theta_1_mayor = mean(theta_1_mayor)) ## # A tibble: 1 × 1 ## prob_theta_1_mayor ## &lt;dbl&gt; ## 1 0.0215 Finalmente, verificamos nuestro modelo y cuánto aprendimos. Podemos hacerlo simulando de la inicial y comparando con la posterior: inicial_tbl &lt;- tibble(theta_1 = rbeta(20000, 100, 1900), theta_2 = rbeta(20000, 100, 1900), dist = &quot;inicial&quot;) posterior_tbl &lt;- iteraciones %&gt;% filter(iter_num &gt; 10) %&gt;% mutate(dist = &quot;posterior&quot;) sims_tbl &lt;- bind_rows(inicial_tbl, posterior_tbl) ggplot(sims_tbl, aes(x = theta_1, y = theta_2, colour = dist)) + geom_point(alpha = 0.2) donde vemos que el resultado que obtuvimos es razonablemente consistente con nuestra información inicial, y las 10 mil tiradas de dado fueron altamente informativas. ¿Qué crees que pasaría si sólo hubieramos tirado 40 veces cada dado? ¿Qué tanto habríamos aprendido? Puedes usar datos simulados y repetir este ejercicio. Puedes examinar los resultados para cada cara con los datos originales. Un modelo apropiado es el Dirichlet-Multinomial. Ejemplo: Modelo normal no conjugado Retomemos el caso de observaciones normales, supongamos que tenemos una muestra \\(X_1,...,X_n\\) de observaciones independientes e identicamente distribuidas, con \\(X_i \\sim \\mathsf{N}(\\mu, \\sigma^2)\\). Usaremos iniciales distintas al modelo anterior: \\[p(\\mu, \\sigma^2) = p(\\sigma^2)p(\\mu)\\] con \\(\\mu\\) \\(\\mathsf{N}(\\mu_0, \\sigma_0)\\) y \\(\\tau = 1/\\sigma^2\\) con distribución \\(\\mathsf{Gamma}(a,b)\\). Esto no nos da el modelo conjugado que vimos antes (nota la diferencia de la especificación de la inicial conjunta). Comenzamos por escribir \\[p(\\mu, \\sigma^2|x) \\propto \\frac{1}{{\\sigma^{n/2}}} \\exp(-\\sum\\frac{(x_i-\\mu)²}{2\\sigma^2}) \\exp(- \\frac{(\\mu - \\mu_0)^2}{2\\sigma_0^2}) \\frac{1}{(\\sigma^2)^{a + 1}}\\exp (-\\beta/\\sigma^2 )\\] Comenzamos analizando \\(p(\\mu|\\sigma^2, x)\\). Por la ecuación de arriba, e ignorando los términos que no dependen de \\(\\mu\\): \\[p(\\mu|\\sigma^2, x) \\propto \\exp [ - \\sum_i (\\frac{(\\mu - x_i)^2}{2\\sigma^2} - \\frac{(\\mu - \\mu_0)^2}{2n\\sigma_0^2})]\\] que es una distribución normal (completa cuadrados): \\[\\mu|\\sigma^2,x \\sim \\mathsf{N}\\bigg(\\frac{\\sigma^2}{\\sigma^2 + n\\sigma_0^2}\\mu_0 + \\frac{n\\sigma_0^2}{\\sigma^2 + n \\sigma_0^2}\\bar{x}, \\frac{\\sigma \\sigma_0}{\\sqrt{\\sigma^2 + n\\sigma_0^2}}\\bigg)\\] Ahora consideramos \\(p(\\sigma^2|mu,x)\\). Ignoramos en \\(p(\\mu,\\sigma^2|x)\\) los términos que *no** dependen de \\(\\sigma^2\\): \\[p(\\sigma^2|\\mu, x) \\propto \\frac{1}{\\sigma^{n/2}} \\exp(-\\sum\\frac{(x_i-\\mu)²}{2\\sigma^2}) \\frac{1}{(\\sigma^2)^{a + 1}}\\exp (-\\beta/\\sigma^2)\\] que simplificando da \\[ = \\frac{1}{\\sigma^{n/2 + a + 1}}\\exp( -\\frac{\\beta +\\frac{1}{2}\\sum(x_i - \\mu)^2}{\\sigma^2} )\\] de modo que \\[\\sigma^2|\\mu, x \\sim \\mathsf{GI}(a +n/2, b + \\frac{1}{2}\\sum(x_i -\\mu)^2)\\] Ejemplo Usaremos este muestreador para el problema de la estaturas de los tenores. Comenzamos definiendo las distribuciones iniciales: \\(\\mu \\sim \\mathsf{N}(175, 3)\\) \\(\\tau = 1/\\sigma^2 \\sim \\mathsf{GI}(3, 150)\\), esto es \\(a = 3\\) y \\(b = 150\\). Escribimos el muestreador de Gibbs. n &lt;- 20 x &lt;- cantantes$estatura_cm m &lt;- 175; sigma_0 &lt;- 3; alpha &lt;- 3; beta &lt;- 150 # parámetros de iniciales pasos &lt;- 20000 iteraciones &lt;- matrix(0, nrow = pasos + 1, ncol = 2) # vector guardará las simulaciones iteraciones[1, 1] &lt;- 0 # valor inicial media colnames(iteraciones) &lt;- c(&quot;mu&quot;, &quot;sigma&quot;) # Generamos la caminata aleatoria for (j in seq(2, pasos, 2)) { # sigma^2 mu &lt;- iteraciones[j - 1, &quot;mu&quot;] a &lt;- n / 2 + alpha b &lt;- sum((x - mu) ^ 2) / 2 + beta iteraciones[j, &quot;sigma&quot;] &lt;- sqrt(1/rgamma(1, a, b)) # Actualizar sigma iteraciones[j, &quot;mu&quot;] &lt;- iteraciones[j-1, &quot;mu&quot;] # mu sigma &lt;- iteraciones[j, &quot;sigma&quot;] media &lt;- (n * sigma_0^2 * mean(x) + sigma^2 * m) / (n * sigma_0^2 + sigma^2) varianza &lt;- sigma^2 * sigma_0^2 / (n * sigma_0^2 + sigma^2) iteraciones[j+1, &quot;mu&quot;] &lt;- rnorm(1, media, sd = sqrt(varianza)) # actualizar mu iteraciones[j+1, &quot;sigma&quot;] &lt;- iteraciones[j, &quot;sigma&quot;] } caminata &lt;- data.frame(pasos = 1:pasos, mu = iteraciones[1:pasos, &quot;mu&quot;], sigma = iteraciones[1:pasos, &quot;sigma&quot;]) caminata_g &lt;- caminata %&gt;% gather(parametro, val, mu, sigma) %&gt;% arrange(pasos) Veamos primero algunos pasos: ggplot(filter(caminata, pasos &gt; 1000, pasos&lt; 1010), aes(x = mu, y = sigma)) + geom_path(alpha = 0.3) + geom_point() Donde vemos cómo en cada iteración se actualiza un solo parámetro. Una alternativa es conservar únicamente ciclos completos de la caminata u esto es lo que hacen varios programas que implementan Gibbs, sin embargo ambas cadenas (cadenas completas y conservando únicamente ciclos completos) convergen a la misma distribución posterior. Si tomamos iteraciones completas: ggplot(filter(caminata, pasos &gt; 1000, pasos&lt; 1020, pasos %% 2 == 0), aes(x = mu, y = sigma)) + geom_path(alpha = 0.3) + geom_point() Y ahora vemos cómo se ven las simulaciones: ggplot(filter(caminata, pasos &gt; 1000, pasos&lt; 10000, pasos %% 2 == 0), aes(x = mu, y = sigma)) + geom_point(alpha = 0.1) Y el diagnóstico de cada cadena: ggplot(filter(caminata_g, pasos &gt; 15000), aes(x = pasos, y = val)) + geom_path(alpha = 0.3) + facet_wrap(~parametro, ncol = 1, scales = &quot;free&quot;) + scale_y_continuous(&quot;&quot;) Estas cadenas parecen estar mezclando bien. Podemos resumirlas: ggplot(filter(caminata_g, pasos &gt; 5000), aes(x = val)) + geom_histogram(fill = &quot;gray&quot;) + facet_wrap(~parametro, ncol = 1, scales = &quot;free&quot;) caminata_g %&gt;% filter(pasos &gt; 1000) %&gt;% # eliminamos la etapa de calentamiento group_by(parametro) %&gt;% summarise( mean(val), sd(val), median(val) ) %&gt;% mutate(across(is_double, round, 2)) ## # A tibble: 2 × 4 ## parametro `mean(val)` `sd(val)` `median(val)` ## &lt;chr&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 mu 176. 1.32 176. ## 2 sigma 6.54 0.95 6.44 Y obtenemos un resultado similar a los anteriores. Conclusiones y observaciones Metrópolis y Gibbs Una generalización del algoritmo de Metrópolis es Metrópolis-Hastings. El algoritmo de Metrópolis es como sigue: Generamos un punto inicial tal que \\(p(\\theta)&gt;0\\). Para \\(i = 1,2,...\\) Se propone un nuevo valor \\(\\theta^*\\) con una distribución propuesta \\(g(\\theta^*|\\theta^{(i)})\\) es común que \\(g(\\theta^*|\\theta^{(i)})\\) sea una normal centrada en \\(\\theta^{(i)}\\). Calculamos la probabilidad de aceptación \\[\\alpha=\\min\\bigg\\{\\frac{p(\\theta^*)}{p(\\theta^{(i)})},1\\bigg\\},\\] y aceptamos \\(\\theta^*\\) con probabilidad \\(p_{mover}\\). Es así que el algorito requiere que podamos calcular el cociente en \\(p_{mover}\\) para todo \\(\\theta^{(i)}\\) y \\(\\theta^*\\), así como simular de la distribución propuesta \\(g(\\theta^*|\\theta^{(i)})\\), adicionalmente debemos poder generar valores uniformes para decidir si aceptar/rechazar. En el caso de Metrópolis un requerimiento adicional es que la distribución propuesta \\(g(\\theta_{a}|\\theta_b)\\) debe ser simétrica, es decir \\(g(\\theta_{a}|\\theta_b) = g(\\theta_{b}|\\theta_a)\\) para todo \\(\\theta_{a}\\), \\(\\theta_{b}\\). Metrópolis-Hastings generaliza Metrópolis, eliminando la restricción de simetría en la distribución propuesta \\(g(\\theta_{a}|\\theta_b)\\), sin embargo para corregir por esta asimetría debemos calcular \\(\\alpha\\) como sigue: \\[\\alpha=\\min\\bigg\\{ \\frac{p(\\theta^*)}{g(\\theta^*|\\theta^{(i)})} \\cdot \\frac{g(\\theta^{(i)}|\\theta^*)}{p(\\theta^{(i)})},1\\bigg\\}\\] La generalización de Metrópolis-Hastings puede resultar en algoritmos más veloces. Se puede ver Gibbs como una generalización de Metrópolis-Hastings, cuando estamos actualizando un componente de los parámetros, la distribución propuesta es la distribución posterior para ese parámetro, por tanto siempre es aceptado. Comparado con Metrópolis, Gibbs tiene la ventaja de que no se necesita afinar los parámetros de una distribución propuesta (o seleccionar siquiera una distribución propuesta). Además que no hay pérdida de simulaciones debido a rechazo. Por su parte, la desventaja debemos conocer las distribuciones condicionales y poder simular de ellas. En el caso de modelos complicados se utilizan combinaciones de Gibbs y Metrópolis. Cuando se consideran estos dos algoritmos Gibbs es un método más simple y es la primera opción para modelos condicionalmente conjugados. Sí solo podemos simular de un subconjunto de las distribuciones condicionales posteriores, entonces podemos usar Gibbs siempre que se pueda y Metrópolis unidimensional para el resto, o de manera más general separamos en bloques, un bloque se actualiza con Gibbs y otro con Metrópolis. El algoritmo de Gibbs puede atorarse cuando hay correlación alta entre los parámetros, reparametrizar puede ayudar, o se pueden usar otros algoritmos. JAGS (Just Another Gibbs Sampler), WinBUGS y OpenBUGS son programas que implementan métodos MCMC para generar simulaciones de distribuciones posteriores. Los paquetes rjags y R2jags permiten ajustar modelos en JAGS desde R. Es muy fácil utilizar estos programas pues uno simplemente debe especificar las distribuciones iniciales, la verosimilitud y los datos observados. Para aprender a usar JAGS se puede revisar la sección correspondiente en las notas de 2018, ahora nos concentraremos en el uso de Stan. HMC y Stan It appears to be quite a general principle that, whenever there is a randomized way of doing something, then there is a nonrandomized way that delivers better performance but requires more thought. -E.T. Jaynes Stan es un programa para generar muestras de una distribución posterior de los parámetros de un modelo, el nombre del programa hace referencia a Stanislaw Ulam (1904-1984) que fue pionero en los métodos de Monte Carlo. A diferencia de JAGS y BUGS, los pasos de la cadena de Markov se generan con un método llamado Monte Carlo Hamiltoniano (HMC). HMC es computacionalmente más costoso que Metrópolis o Gibbs, sin embargo, sus propuestas suelen ser más eficientes, y por consiguiente no necesita muestras tan grandes. En particular cuando se ajustan modelos grandes y complejos (por ejemplo, con variables con correlación alta) HMC supera a otros. Diagnósticos generales para MCMC Cuando generamos una muestra de la distribución posterior usando MCMC, sin importar el método (Metrópolis, Gibbs, HMC), buscamos que: Los valores simulados sean representativos de la distribución posterior. Esto implica que no deben estar influenciados por el valor inicial (arbitrario) y deben explorar todo el rango de la posterior, con suficientes retornos para evaluar cuánta masa hay en cada región. Debemos tener suficientes simulaciones de tal manera que las estimaciones sean precisas y estables. Queremos tener un método eficiente para generar las simulaciones. En la práctica intentamos cumplir lo más posible estos objetivos, pues aunque en principio los métodos MCMC garantizan que una cadena infinitamente larga logrará una representación perfecta, siempre debemos tener un criterio para cortar la cadena y evaluar la calidad de las simulaciones. Representatividad Burn-in e iteraciones iniciales- En primer lugar, en muchas ocasiones las condiciones iniciales de las cadenas están en partes del espacio de parámetros que son “atípicos” en términos de la posterior. Así que es común quitar algunas observaciones iniciales (iteraciones de burn-in) para minimizar su efecto en resúmenes posteriores. Por ejemplo, para el ejemplo de los cantantes, podemos ver que las iteraciones iniciales tienen como función principal llegar a las regiones de probabilidad posterior alta: log_p &lt;- crear_log_posterior_norm(cantantes$estatura_cm, mu_0, n_0, a, b) log_post &lt;- function(pars) { log_p(pars[1], pars[2]) } set.seed(823) metro_normal &lt;- crear_metropolis(log_post, sigma_salto = 0.5) sim_tbl &lt;- metro_normal(c(mu = 162, sigma = 1), 5000) ggplot(sim_tbl %&gt;% filter(iter_num &lt; 500), aes(x = mu, y = sigma)) + geom_path(alpha = 0.5) + geom_point(aes(colour = iter_num)) De modo que puede ser buena idea eliminar las primeras iteraciones. En teoría, no es necesario hacer esto si hacemos suficientes iteraciones, pues la cadena va a terminar en su estado estable explorando la posterior. En la práctica, y con pocas iteraciones, puede ayudar un poco a mejorar la precisión numérica de las cantidades que queramos calcular. sim_g &lt;- sim_tbl %&gt;% pivot_longer(-iter_num, names_to = &quot;parametro&quot;, values_to = &quot;valor&quot;) todas &lt;- ggplot(sim_g, aes(x = iter_num, y = valor)) + geom_line(alpha = 0.5) + facet_wrap(~ parametro, ncol = 1, scales = &quot;free_y&quot;) + labs(subtitle = &quot;Todas las simulaciones&quot;) sin_burnin &lt;- sim_g %&gt;% filter(iter_num &gt; 200) %&gt;% ggplot(aes(x = iter_num, y = valor)) + geom_line(alpha = 0.5) + facet_wrap(~ parametro, ncol = 1, scales = &quot;free_y&quot;) + labs(subtitle = &quot;Quitando 200 de burn-in&quot;) todas + sin_burnin Convergencia a estado límite. Para determinar la convergencia es conveniente realizar más de una cadena: buscamos ver si realmente se ha olvidado el estado inicial, si las distribuciones de cada cadena son consistentes unas con otras, y revisar que algunas cadenas no hayan quedado atoradas en regiones inusuales del espacio de parámetros. Inicializamos las cadenas con valores al azar en rangos razonables (por ejemplo simulando de la inicial): set.seed(8513) valores_iniciales &lt;- tibble(mu_0 = rnorm(4, 160, 20), sigma_0 = runif(4, 0, 20), cadena = 1:4) sims_tbl &lt;- valores_iniciales %&gt;% mutate(sims = map2(mu_0, sigma_0, ~ metro_normal(c(mu = .x, sigma = .y), 300) )) %&gt;% unnest(sims) ggplot(sims_tbl, aes(x = iter_num, y = sigma, colour = factor(cadena))) + geom_line() Y este es un ejemplo donde claramente las cadenas no han alcanzado un estado estable: tienen muy distintas medias y varianzas. Por ejemplo: set.seed(83243) sims_tbl &lt;- valores_iniciales %&gt;% mutate(sims = map2(mu_0, sigma_0, ~ metro_normal(c(mu = .x, sigma = .y), 20000) )) %&gt;% unnest(sims) ggplot(sims_tbl, aes(x = iter_num, y = sigma, colour = factor(cadena))) + geom_line() Y este resultado se ve mejor. La parte transición hacia las zonas de alta probabilidad pasa antes de unas 1000 iteraciones. Podemos hacer más simulaciones, o eliminar como burn-in las primiras iteraciones: media_g &lt;- ggplot(sims_tbl %&gt;% filter(iter_num &gt; 2000), aes(x = iter_num, y = mu, colour = factor(cadena))) + geom_line() sigma_g &lt;- ggplot(sims_tbl %&gt;% filter(iter_num &gt; 2000), aes(x = iter_num, y = sigma, colour = factor(cadena))) + geom_line() media_g / sigma_g Las gráficas anteriores nos ayudan a determinar si elegimos un periodo de calentamiento adecuado o si alguna cadena está alejada del resto. Una vez que las cadenas están en estado estable, podemos usar todas las simulaciones juntas para resumir: head(sims_tbl) ## # A tibble: 6 × 6 ## mu_0 sigma_0 cadena iter_num mu sigma ## &lt;dbl&gt; &lt;dbl&gt; &lt;int&gt; &lt;int&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 155. 3.16 1 1 155. 3.16 ## 2 155. 3.16 1 2 155. 3.16 ## 3 155. 3.16 1 3 155. 3.16 ## 4 155. 3.16 1 4 155. 3.16 ## 5 155. 3.16 1 5 155. 3.50 ## 6 155. 3.16 1 6 155. 3.81 # medias posteriores sims_tbl %&gt;% summarise(mu = mean(mu), sigma = mean(sigma)) ## # A tibble: 1 × 2 ## mu sigma ## &lt;dbl&gt; &lt;dbl&gt; ## 1 176. 6.77 Además de realizar gráficas podemos usar la medida de convergencia \\(\\hat{R}\\). La medida \\(\\hat{R}\\) se conoce como el factor de reducción potencial de escala o diagnóstico de convergencia de Gelman-Rubin, esta es una estimación de la posible reducción en la longitud de un intervalo de confianza si las simulaciones continuaran infinitamente. \\(\\hat{R}\\) es aproximadamente la raíz cuadrada de la varianza de todas las cadenas juntas dividida entre la varianza dentro de cada cadena. Si \\(\\hat{R}\\) es mucho mayor a 1 esto indica que las cadenas no se han mezclado bien. Una regla usual es iterar hasta alcanzar un valor \\(\\hat{R} \\leq 1.1\\) para todos los parámetros. \\[\\hat{R} \\approx \\sqrt{\\frac{\\hat{V}}{W}}\\] donde \\(B\\) es la varianza entre las cadenas, \\(W\\) es la varianza dentro de las cadenas \\[B = \\frac{N}{M-1}\\sum_m (\\hat{\\theta}_m - \\hat{\\theta})^2\\] \\[W = \\frac{1}{M}\\sum_m \\hat{\\sigma}_m^2\\] Y \\(\\hat{V}\\) es una estimación del varianza de posterior de \\(\\theta\\): \\[\\hat{V} = \\frac{N-1}{N}W + \\frac{M+1}{MN}B\\] #### Ejemplo {-} En nuestro ejemplo anterior, tenemos sims_tbl %&gt;% pivot_longer(mu:sigma, names_to = &quot;parametro&quot;, values_to = &quot;valor&quot;) %&gt;% group_by(parametro, cadena) %&gt;% summarise(media = mean(valor), num = n(), sigma2 = var(valor)) %&gt;% summarise(N = first(num), M = n_distinct(cadena), B = N * var(media), W = mean(sigma2), V_hat = ((N - 1) / N) * W + (M + 1)/(M * N) * B, R_hat = sqrt(V_hat / W)) ## # A tibble: 2 × 7 ## parametro N M B W V_hat R_hat ## &lt;chr&gt; &lt;int&gt; &lt;int&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 mu 20000 4 1281. 4.29 4.37 1.01 ## 2 sigma 20000 4 121. 1.31 1.32 1.00 Y verificamos que los valores de \\(\\hat{R}\\) son cercanos a uno, lo cual indica que este diagnóstico es aceptable. Si hubiéramos trabajado con las primeras 300 iteraciones sims_tbl %&gt;% filter(iter_num &lt; 300) %&gt;% pivot_longer(mu:sigma, names_to = &quot;parametro&quot;, values_to = &quot;valor&quot;) %&gt;% group_by(parametro, cadena) %&gt;% summarise(media = mean(valor), num = n(), sigma2 = var(valor)) %&gt;% summarise(N = first(num), M = n_distinct(cadena), B = N * var(media), W = mean(sigma2), V_hat = ((N - 1) / N) * W + (M + 1)/(M * N) * B, R_hat = sqrt(V_hat / W)) ## # A tibble: 2 × 7 ## parametro N M B W V_hat R_hat ## &lt;chr&gt; &lt;int&gt; &lt;int&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; ## 1 mu 299 4 32334. 40.4 175. 2.08 ## 2 sigma 299 4 7394. 11.9 42.8 1.89 Y estos valores indican problemas en la convergencia de las cadenas. Es necesario diagnosticar el problema, que en este caso resolvemos incrementando el número de iteraciones. Precisión Una vez que tenemos una muestra representativa de la distribución posterior, nuestro objetivo es asegurarnos de que la muestra es lo suficientemente grande para producir estimaciones estables y precisas de la distribución. Para ello usaremos el tamaño efectivo de muestra, Si las simulaciones fueran independientes \\(N_{eff}\\) sería el número total de simulaciones; sin embargo, las simulaciones de MCMC suelen estar correlacionadas, de modo que cada iteración de MCMC es menos informativa que si fueran independientes. Ejemplo: Si graficaramos simulaciones independientes, esperaríamos valores de autocorrelación chicos: acf(rgamma(1000,1,1)) Sin embargo, los valores que simulamos tienen el siguiente perfil de autocorrelación: sigma_metro_sims &lt;- sims_tbl %&gt;% filter(cadena==4) %&gt;% pull(mu) acf(sigma_metro_sims) El tamaño efectivo de muestra nos dice qué tamaño de muestra de observaciones independientes nos daría la misma información que las simulaciones de la cadena. Una manera de manera relativamente simple de estimarlo es: \\[N_{eff} = \\frac{N}{1+2\\sum_{k=1}^\\infty ACF(k)} \\] Usualmente nos gustaría obtener un tamaño efectivo de al menos \\(100\\) (para cálculo de medias y varianzas posteriores). Esta cantidad usualmente se reporta en el software (con mejores estimaciones que la de la fórmula de arriba), y es necesario checarlo. En nuestro ejemplo hacemos una aproximación como sigue: calc_acf &lt;- function(x){ valores_acf &lt;- acf(x, lag.max = 1000, plot = FALSE)$acf %&gt;% as.numeric() valores_acf[-1] } acf_tbl &lt;- sims_tbl %&gt;% pivot_longer(mu:sigma, names_to = &quot;parametro&quot;, values_to = &quot;valor&quot;) %&gt;% group_by(parametro, cadena) %&gt;% summarise(N = n_distinct(iter_num), k = 1:1000, acf = calc_acf(valor)) %&gt;% summarise(N = first(N), N_eff = N / (1 + 2 * sum(acf))) acf_tbl ## # A tibble: 8 × 4 ## # Groups: parametro [2] ## parametro cadena N N_eff ## &lt;chr&gt; &lt;int&gt; &lt;int&gt; &lt;dbl&gt; ## 1 mu 1 20000 251. ## 2 mu 2 20000 700. ## 3 mu 3 20000 104. ## 4 mu 4 20000 394. ## 5 sigma 1 20000 421. ## 6 sigma 2 20000 411. ## 7 sigma 3 20000 93.9 ## 8 sigma 4 20000 724. Nótese que algunas cadenas tienen un tamaño efectivo de muestra relativamente bajo para el número de iteraciones que hicimos. De cualquier forma, el agregado sobre todas las cadenas es suficientemente grande para calcular resúmenes básicos: acf_tbl %&gt;% group_by(parametro) %&gt;% summarise(N = sum(N), N_eff = sum(N_eff)) ## # A tibble: 2 × 3 ## parametro N N_eff ## &lt;chr&gt; &lt;int&gt; &lt;dbl&gt; ## 1 mu 80000 1450. ## 2 sigma 80000 1650. Sin embargo, podemos hacer más simulaciones si es necesario, por ejemplo para aproximar de manera apropiada percentiles en las colas. Eficiencia Hay varias maneras para mejorar la eficiencia de un proceso MCMC: Paralelizar, no disminuimos el número de pasos en las simulaciones pero podemos disminuir el tiempo que tarda en correr. Cambiar la parametrización del modelo o transformar los datos. Adelgazar la muestra cuando tenemos problemas de uso de memoria, consiste en guardar únicamente los \\(k\\)-ésimos pasos de la cadena y resulta en cadenas con menos autocorrelación . Recomendaciones generales Gelman and Hill (2006) recomienda los siguientes pasos cuando uno esta simulando de la posterior: Cuando definimos un modelo por primera vez establecemos un valor bajo para el número de iteraciones. La razón es que la mayor parte de las veces los modelos no funcionan a la primera por lo que sería pérdida de tiempo dejarlo correr mucho tiempo antes de descubrir el problema. Si las simulaciones no han alcanzado convergencia aumentamos las iteraciones a \\(500\\) ó \\(1000\\) de tal forma que las corridas tarden segundos o unos cuantos minutos. Si tarda más que unos cuantos minutos (para problemas del tamaño que veremos en la clase) y aún así no alcanza convergencia entonces juega un poco con el modelo (por ejemplo intenta transformaciones lineales), para JAGS Gelman sugiere más técnicas para acelerar la convergencia en el capitulo \\(19\\) del libro Data Analysis Using Regression and Multilevel/Hierarchical models. En el caso de Stan veremos ejemplos de reparametrización, y se puede leer más en la guía. Otra técnica conveniente cuando se trabaja con bases de datos grandes (sobre todo en la parte exploratoria) es trabajar con un subconjunto de los datos, quizá la mitad o una quinta parte. Referencias "],["apéndice-principios-de-visualizacion.html", "Apéndice: Principios de visualizacion Introducción Visualización popular de datos Teoría de visualización de datos Ejemplo: gráfica de Minard", " Apéndice: Principios de visualizacion “The simple graph has brought more information to the data analyst’s mind than any other device.” — John Tukey El cuarteto de Anscombe En 1971 un estadístico llamado Frank Anscombe (fundador del departamento de Estadística de la Universidad de Yale) publicó cuatro conjuntos de dato. Cada uno consiste de 11 observaciones. La peculariedad de estos conjuntos es que tienen las mismas propiedades estadísticas. Sin embargo, cuando analizamos los datos de manera gráfica en un histograma encontramos rápidamente que los conjuntos de datos son muy distintos. Media de \\(x\\): 9 Varianza muestral de \\(x\\): 11 Media de \\(y\\): 7.50 Varianza muestral de \\(y\\): 4.12 Correlación entre \\(x\\) y \\(y\\): 0.816 Línea de regresión lineal: \\(y = 3.00 + 0.500x\\) En la gráfica del primer conjunto de datos, se ve clara una relación lineal simple con un modelo que cumple los supuestos de normalidad. La segunda gráfica (arriba a la derecha) muestra unos datos que tienen una asociación pero definitivamente no es lineal. En la tercera gráfica (abajo a la izquierda) están puntos alineados perfectamente en una línea recta, excepto por uno de ellos. En la última gráfica podemos ver un ejemplo en el cual basta tener una observación atípica para que se produzca un coeficiente de correlación alto aún cuando en realidad no existe una asociación lineal entre las dos variables. El cuarteto de Ascombe inspiró una técnica reciente para crear datos que comparten las mismas propiedades estadísticas al igual que en el cuarteto, pero que producen gráficas muy distintas (Matejka, Fitzmaurice). Introducción La visualización de datos no trata de hacer gráficas “bonitas” o “divertidas”, ni de simplificar lo complejo o ayudar a una persona “que no entiende mucho” a entender ideas complejas. Más bien, trata de aprovechar nuestra gran capacidad de procesamiento visual para exhibir de manera clara aspectos importantes de los datos. El siguiente ejemplo de (Tufte 2006), ilustra claramente la diferencia entre estos dos enfoques. A la izquierda están gráficas (más o menos típicas de Powerpoint) basadas en la filosofía de simplificar, de intentar no “ahogar” al lector con datos. El resultado es una colección incoherente, de bajo contenido, que no tiene mucho qué decir y que es, “indeferente al contenido y la evidencia”. A la derecha está una variación del rediseño de Tufte en forma de tabla, que en este caso particular es una manera eficiente de mostrar claramente los patrones que hay en este conjunto simple de datos. ¿Qué principios son los que soportan la efectividad de esta tabla sobre la gráfica de la derecha? Veremos que hay dos conjuntos de principios importantes: unos relacionados con el diseño y otros con la naturaleza del análisis de datos, independientemente del método de visualización. Visualización popular de datos Publicaciones populares (periódicos, revistas, sitios internet) muchas veces incluyen visualización de datos como parte de sus artículos o reportajes. En general siguen el mismo patrón que en la visión tradicionalista de la estadística: sirven más para divertir que para explicar, tienden a explicar ideas simples y conjuntos chicos de datos, y se consideran como una “ayuda” para los “lectores menos sofisticados”. Casi siempre se trata de gráficas triviales (muchas veces con errores graves) que no aportan mucho a artículos que tienen un nivel de complejidad mucho mayor (es la filosofía: lo escrito para el adulto, lo graficado para el niño). Teoría de visualización de datos Existe teoría fundamentada acerca de la visualización. Después del trabajo pionero de Tukey, los principios e indicadores de Tufte se basan en un estudio de la historia de la graficación y ejercicios de muestreo de la práctica gráfica a lo largo de varias disciplinas (¿cuáles son las mejores gráficas? ¿por qué?) El trabajo de Cleveland es orientado a la práctica del análisis de datos (¿cuáles gráficas nos han ayudado a mostrar claramente los resultados del análisis?), por una parte, y a algunos estudios de percepción visual. En resumen, hablaremos de las siguientes guías: Principios generales del diseño analítico Aplicables a una presentación o análisis completos, y como guía para construir nuevas visualizaciones (Tufte 2006). Principio 1. Muestra comparaciones, contrastes, diferencias. Principio 2. Muestra causalidad, mecanismo, explicación, estructura sistemática. Principio 3. Muestra datos multivariados, es decir, más de una o dos variables. Principio 4. Integra palabras, números, imágenes y diagramas. Principio 5. Describe la totalidad de la evidencia. Muestra fuentes usadas y problemas relevantes. Principio 6. Las presentaciones analíticas, a fin de cuentas, se sostienen o caen dependiendo de la calidad, relevancia e integridad de su contenido. Técnicas de visualización Esta categoría incluye técnicas específicas que dependen de la forma de nuestros datos y el tipo de pregunta que queremos investigar (Tukey (1977), William S. Cleveland (1993), W. S. Cleveland (1994), Tufte (2006)). Tipos de gráficas: cuantiles, histogramas, caja y brazos, gráficas de dispersión, puntos/barras/ líneas, series de tiempo. Técnicas para mejorar gráficas: Transformación de datos, transparencia, vibración, banking 45, suavizamiento y bandas de confianza. Pequeños múltiplos Indicadores de calidad gráfica Aplicables a cualquier gráfica en particular. Estas son guías concretas y relativamente objetivas para evaluar la calidad de una gráfica (Tufte 1986). Integridad Gráfica. El factor de engaño, es decir, la distorsión gráfica de las cantidades representadas, debe ser mínimo. Chartjunk. Minimizar el uso de decoración gráfica que interfiera con la interpretación de los datos: 3D, rejillas, rellenos con patrones. Tinta de datos. Maximizar la proporción de tinta de datos vs. tinta total de la gráfica. For non-data- ink, less is more. For data-ink, less is a bore. Densidad de datos. Las mejores gráficas tienen mayor densidad de datos, que es la razón entre el tamaño del conjunto de datos y el área de la gráfica. Las gráficas se pueden encoger mucho. Percepción visual. Algunas tareas son más fáciles para el ojo humano que otras (W. S. Cleveland 1994). Factor de engaño y Chartjunk El factor de engaño es el cociente entre el efecto mostrado en una gráfica y el efecto correspondiente en los datos. Idealmente, el factor de engaño debe ser 1 (ninguna distorsión). El chartjunk son aquellos elementos gráficos que no corresponden a variación de datos, o que entorpecen la interpretación de una gráfica. Estos son los indicadores de calidad más fáciles de entender y aplicar, y afortunadamente cada vez son menos comunes. Un diseño popular que califica como chartjunk y además introduce factores de engaño es el pie de 3D. En la gráfica de la derecha, podemos ver como la rebanada C se ve más grande que la rebanada A, aunque claramente ese no es el caso (factor de engaño). La razón es la variación en la perspectiva que no corresponde a variación en los datos (chartjunk). Crítica gráfica: Gráfica de pie Todavía elementos que pueden mejorar la comprensión de nuestra gráfica de pie: se trata de la decodificiación que hay que hacer categoría - color - cuantificación. Podemos agregar las etiquetas como se muestra en la serie de la derecha, pero entonces: ¿por qué no mostrar simplemente la tabla de datos? ¿qué agrega el pie a la interpretación? La deficiencias en el pie se pueden ver claramente al intentar graficar más categorías (13) . En el primer pie no podemos distinguir realmente cuáles son las categorías grandes y cuáles las chicas, y es muy difícil tener una imagen mental clara de estos datos. Agregar los porcentajes ayuda, pero entonces, otra vez, preguntamos cuál es el propósito del pie. La tabla de la izquierda hace todo el trabajo (una vez que ordenamos las categrías de la más grande a la más chica). Es posible hacer una gráfica de barras como la de abajo a la izquierda. Hay otros tipos de chartjunk comunes: uno es la textura de barras, por ejemplo. El efecto es la producción de un efecto moiré que es desagradable y quita la atención de los datos, como en la gráfica de barras de abajo. Otro común son las rejillas, como mostramos en las gráficas de la izquierda. Nótese como en estos casos hay efectos ópticos no planeados que degradan la percepción de los patrones en los datos. Pequeños múltiplos y densidad gráfica La densidad de una gráfica es el tamaño del conjunto de datos que se grafica comparado con el área total de la gráfica. En el siguiente ejemplo, graficamos en logaritmo-10 de millones de cabezas de ganado en Francia (cerdos, res, ovejas y caballos). La gráfica de la izquierda es pobre en densidad pues sólo representa 4 datos. La manera más fácil de mejorar la densidad es hacer más chica la gráfica: La razón de este encogimiento es una que tiene qué ver con las oportunidades perdidas de una gráfica grande. Si repetimos este mismo patrón (misma escala, mismos tipos de ganado) para distintos países obtenemos la siguiente gráfica: Esta es una gráfica de puntos. Es útil como sustituto de una gráfica de barras, y es superior en el sentido de que una mayor proporción de la tinta que se usa es tinta de datos. Otra vez, mayor proporción de tinta de datos representa más oportunidades que se pueden capitalizar. Más pequeños múltiplos Los pequeños múltiplos presentan oportunidades para mostrar más acerca de nuestro problema de interés. Consideramos por ejemplo la relación de radiación solar y niveles de ozono: En el ejemplo anterior incluyendo una variable adicional (velocidad del viento) podemos entender más acerca de la relación de radiación solar y niveles de ozono: Tinta de datos Maximizar la proporción de tinta de datos en nuestras gráficas tiene beneficios inmediatos. La regla es: si hay tinta que no representa variación en los datos, o la eliminación de esa tinta no representa pérdidas de significado, esa tinta debe ser eliminada. El ejemplo más claro es el de las rejillas en gráficas y tablas: ¿Por qué usar grises en lugar de negros? La respuesta tiene qué ver con el principio de tinta de datos: si marcamos las diferencias sutil pero claramente, tenemos más oportunidades abiertas para hacer énfasis en lo que nos interesa: a una gráfica o tabla saturada no se le puede hacer más - es difícil agregar elementos adicionales que ayuden a la comprensión. Si comenzamos marcando con sutileza, entonces se puede hacer más. Los mapas geográficos son un buen ejemplo de este principio. El espacio en blanco es suficientemente bueno para indicar las fronteras en una tabla, y facilita la lectura: Para un ejemplo del proceso de rediseño de una tabla, ver aquí. Finalmente, podemos ver un ejemplo que intenta incorporar los elementos del diseño analítico, incluyendo pequeños múltiplos: Decoración Percepción de escala Entre la percepción visual y la interpretación de una gráfica están implícitas tareas visuales específicas que las personas debemos realizar para ver correctamente la gráfica. En la década de los ochenta, William S. Cleveland y Robert McGill realizaron algunos experimentos identificando y clasificando estas tareas para diferentes tipos de gráficos (William S. Cleveland and McGill 1984). En estos, se le pregunta a la persona que compare dos valores dentro de una gráfica, por ejemplo, en dos barras en una gráfica de barras, o dos rebanadas de una gráfica de pie. Los resultados de Cleveland y McGill fueron replicados por Heer y Bostock en 2010 y los resultados se muestran en las gráficas de la derecha: Ejemplo: gráfica de Minard Concluimos esta sección con una gráfica que, aunque poco común, ejemplifica los principios de una buena gráfica, y es reconocida como una de las mejores visualizaciones de la historia. Una gráfica excelente, presenta datos interesantes de forma bien diseñada: es una cuestión de fondo, de diseño, y estadística… [Se] compone de ideas complejas comunicadas con claridad, precisión y eficiencia. … [Es] lo que da al espectador la mayor cantidad de ideas, en el menor tiempo, con la menor cantidad de tinta, y en el espacio más pequeño. … Es casi siempre multivariado. … Una excelente gráfica debe decir la verdad acerca de los datos. (Tufte, 1983) La famosa visualización de Charles Joseph Minard de la marcha de Napoleón sobre Moscú, ilustra los principios de una buena gráfica. Tufte señala que esta imagen “bien podría ser el mejor gráfico estadístico jamás dibujado”, y sostiene que “cuenta una historia rica y coherente con sus datos multivariados, mucho más esclarecedora que un solo número que rebota en el tiempo”. Se representan seis variables: el tamaño del ejército, su ubicación en una superficie bidimensional, la dirección del movimiento del ejército y la temperatura en varias fechas durante la retirada de Moscú”. Hoy en día Minard es reconocido como uno de los principales contribuyentes a la teoría de análisis de datos y creación de infografías con un fundamento estadístico. Se grafican 6 variables: el número de tropas de Napoleón, la distancia, la temperatura, la latitud y la longitud, la dirección en que viajaban las tropas y la localización relativa a fechas específicas. La gráfica de Minard, como la describe E.J. Marey, parece “desafiar la pluma del historiador con su brutal elocuencia”, la combinación de datos del mapa, y la serie de tiempo, dibujados en 1869, “retratan una secuencia de pérdidas devastadoras que sufrieron las tropas de Napoleón en 1812”. Comienza en la izquierda, en la frontera de Polonia y Rusia, cerca del río Niemen. La línea gruesa dorada muestra el tamaño de la Gran Armada (422,000) en el momento en que invadía Rusia en junio de 1812. El ancho de esta banda indica el tamaño de la armada en cada punto del mapa. En septiembre, la armada llegó a Moscú, que ya había sido saqueada y dejada desértica, con sólo 100,000 hombres. El camino del retiro de Napoleón desde Moscú está representado por la línea oscura (gris) que está en la parte inferior, que está relacionada a su vez con la temperatura y las fechas en el diagrama de abajo. Fue un invierno muy frío, y muchos se congelaron en su salida de Rusia. Como se muestra en el mapa, cruzar el río Berezina fue un desastre, y el ejército de Napoleón logró regresar a Polonia con tan sólo 10,000 hombres. También se muestran los movimientos de las tropas auxiliaries, que buscaban proteger por atrás y por la delantera mientras la armada avanzaba hacia Moscú. La gráfica de Minard cuenta una historia rica y cohesiva, coherente con datos multivariados y con los hechos históricos, y que puede ser más ilustrativa que tan sólo representar un número rebotando a lo largo del tiempo. Referencias "],["apéndice-transformaciones.html", "Apéndice: Transformaciones", " Apéndice: Transformaciones En ocasiones es conveniente transformar los datos para el análisis, el objetivo de los ajustes es simplificar la interpretación y el análisis al eliminar fuentes de variación conocidas, también es común realizan transformaciones para simplificar los patrones. Algunos ejemplos donde eliminamos efectos conocidos: Cuando analizamos el precio de venta de las casas podemos eliminar la variación debida al tamaño de las casas al pasar de precio de venta a precio de venta por metro cuadrado. De manera similar al analizar las propinas puede convenir considerar la propina como porcentaje de la cuenta. En series de tiempo cuando los datos están relacionados con el tamaño de la población podemos ajustar a mediciones per capita (en series de tiempo PIB). También es común ajustar por inflación, o poner cantidades monetarias en valor presente. mex_dat &lt;- global_economy |&gt; filter(Code == &quot;MEX&quot;) pib &lt;- ggplot(mex_dat, aes(x = Year, y = GDP / 1e6)) + geom_line() pib_pc &lt;- ggplot(mex_dat, aes(x = Year, y = GDP / Population)) + geom_line() pib / pib_pc Adicionalmente podemos recurrir a otras transformaciones matemáticas (e.g. logaritmo, raíz cuadrada) que simplifiquen el patrón en los datos y la interpretación. Veamos un ejemplo donde es apropiado la transformación logaritmo. Usamos los datos Animals con información de peso corporal promedio y peso cerebral promedio para 28 especies. Buscamos entender la relación entre estas dos variables, e inspeccionar que especies se desvían (residuales) del esperado. Comenzamos con un diagrama de dispersión usando las unidades originales animals_tbl &lt;- as_tibble(Animals, rownames = &quot;animal&quot;) p1 &lt;- ggplot(animals_tbl, aes(x = body, y = brain, label = animal)) + geom_point() + labs(subtitle = &quot;Unidades originales&quot;) p2 &lt;- ggplot(animals_tbl, aes(x = body, y = brain, label = animal)) + geom_point() + xlim(0, 500) + ylim(0, 1500) + geom_text_repel() + labs(subtitle = &quot;Unidades originales, eliminando &#39;grandes&#39;&quot;) (p1 + p2) Incluso cuando nos limitamos a especies de menos de 500 kg de masa corporal, la relación no es fácil de descrubir.En la suguiente gráfica hacemos la transformación logaritmo y obtenemos una gráfica más fácil de leer, además los datos se modelarán con más facilidad. p3 &lt;- ggplot(animals_tbl, aes(x = log(body), y = log(brain), label = animal)) + geom_smooth(method = &quot;lm&quot;, se = FALSE, color = &quot;red&quot;) + geom_point() + geom_text_repel() + stat_poly_eq(use_label(c(&quot;eq&quot;))) p3 ## `geom_smooth()` using formula = &#39;y ~ x&#39; La transformación logaritmo tiene también ventajas en interpretación, para diferencias chicas en escala log, las diferencias corresponden a diferencias porcentuales en la escala original, por ejempo consideremos la diferencia entre el peso en escala log de humano y borrego: 4.13 - 4.02 = 0.11. Confirmamos que el humano es aproximadamente 11% más pesado que el borrego en la escala original: 62/55.5 - 1 = 0.12 animals_tbl &lt;- animals_tbl |&gt; mutate(log_body = log(body), log_brain = log(brain)) animals_tbl |&gt; filter(animal == &quot;Human&quot; | animal == &quot;Sheep&quot;) |&gt; arrange(body) |&gt; gt::gt() |&gt; gt::fmt_number() #ociqvanemt table { font-family: system-ui, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji'; -webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; } #ociqvanemt thead, #ociqvanemt tbody, #ociqvanemt tfoot, #ociqvanemt tr, #ociqvanemt td, #ociqvanemt th { border-style: none; } #ociqvanemt p { margin: 0; padding: 0; } #ociqvanemt .gt_table { display: table; border-collapse: collapse; line-height: normal; margin-left: auto; margin-right: auto; color: #333333; font-size: 16px; font-weight: normal; font-style: normal; background-color: #FFFFFF; width: auto; border-top-style: solid; border-top-width: 2px; border-top-color: #A8A8A8; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #A8A8A8; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; } #ociqvanemt .gt_caption { padding-top: 4px; padding-bottom: 4px; } #ociqvanemt .gt_title { color: #333333; font-size: 125%; font-weight: initial; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; border-bottom-color: #FFFFFF; border-bottom-width: 0; } #ociqvanemt .gt_subtitle { color: #333333; font-size: 85%; font-weight: initial; padding-top: 3px; padding-bottom: 5px; padding-left: 5px; padding-right: 5px; border-top-color: #FFFFFF; border-top-width: 0; } #ociqvanemt .gt_heading { background-color: #FFFFFF; text-align: center; border-bottom-color: #FFFFFF; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #ociqvanemt .gt_bottom_border { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #ociqvanemt .gt_col_headings { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; } #ociqvanemt .gt_col_heading { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 6px; padding-left: 5px; padding-right: 5px; overflow-x: hidden; } #ociqvanemt .gt_column_spanner_outer { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: normal; text-transform: inherit; padding-top: 0; padding-bottom: 0; padding-left: 4px; padding-right: 4px; } #ociqvanemt .gt_column_spanner_outer:first-child { padding-left: 0; } #ociqvanemt .gt_column_spanner_outer:last-child { padding-right: 0; } #ociqvanemt .gt_column_spanner { border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: bottom; padding-top: 5px; padding-bottom: 5px; overflow-x: hidden; display: inline-block; width: 100%; } #ociqvanemt .gt_spanner_row { border-bottom-style: hidden; } #ociqvanemt .gt_group_heading { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; text-align: left; } #ociqvanemt .gt_empty_group_heading { padding: 0.5px; color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; vertical-align: middle; } #ociqvanemt .gt_from_md > :first-child { margin-top: 0; } #ociqvanemt .gt_from_md > :last-child { margin-bottom: 0; } #ociqvanemt .gt_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; margin: 10px; border-top-style: solid; border-top-width: 1px; border-top-color: #D3D3D3; border-left-style: none; border-left-width: 1px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 1px; border-right-color: #D3D3D3; vertical-align: middle; overflow-x: hidden; } #ociqvanemt .gt_stub { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; } #ociqvanemt .gt_stub_row_group { color: #333333; background-color: #FFFFFF; font-size: 100%; font-weight: initial; text-transform: inherit; border-right-style: solid; border-right-width: 2px; border-right-color: #D3D3D3; padding-left: 5px; padding-right: 5px; vertical-align: top; } #ociqvanemt .gt_row_group_first td { border-top-width: 2px; } #ociqvanemt .gt_row_group_first th { border-top-width: 2px; } #ociqvanemt .gt_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #ociqvanemt .gt_first_summary_row { border-top-style: solid; border-top-color: #D3D3D3; } #ociqvanemt .gt_first_summary_row.thick { border-top-width: 2px; } #ociqvanemt .gt_last_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #ociqvanemt .gt_grand_summary_row { color: #333333; background-color: #FFFFFF; text-transform: inherit; padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; } #ociqvanemt .gt_first_grand_summary_row { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-top-style: double; border-top-width: 6px; border-top-color: #D3D3D3; } #ociqvanemt .gt_last_grand_summary_row_top { padding-top: 8px; padding-bottom: 8px; padding-left: 5px; padding-right: 5px; border-bottom-style: double; border-bottom-width: 6px; border-bottom-color: #D3D3D3; } #ociqvanemt .gt_striped { background-color: rgba(128, 128, 128, 0.05); } #ociqvanemt .gt_table_body { border-top-style: solid; border-top-width: 2px; border-top-color: #D3D3D3; border-bottom-style: solid; border-bottom-width: 2px; border-bottom-color: #D3D3D3; } #ociqvanemt .gt_footnotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #ociqvanemt .gt_footnote { margin: 0px; font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #ociqvanemt .gt_sourcenotes { color: #333333; background-color: #FFFFFF; border-bottom-style: none; border-bottom-width: 2px; border-bottom-color: #D3D3D3; border-left-style: none; border-left-width: 2px; border-left-color: #D3D3D3; border-right-style: none; border-right-width: 2px; border-right-color: #D3D3D3; } #ociqvanemt .gt_sourcenote { font-size: 90%; padding-top: 4px; padding-bottom: 4px; padding-left: 5px; padding-right: 5px; } #ociqvanemt .gt_left { text-align: left; } #ociqvanemt .gt_center { text-align: center; } #ociqvanemt .gt_right { text-align: right; font-variant-numeric: tabular-nums; } #ociqvanemt .gt_font_normal { font-weight: normal; } #ociqvanemt .gt_font_bold { font-weight: bold; } #ociqvanemt .gt_font_italic { font-style: italic; } #ociqvanemt .gt_super { font-size: 65%; } #ociqvanemt .gt_footnote_marks { font-size: 75%; vertical-align: 0.4em; position: initial; } #ociqvanemt .gt_asterisk { font-size: 100%; vertical-align: 0; } #ociqvanemt .gt_indent_1 { text-indent: 5px; } #ociqvanemt .gt_indent_2 { text-indent: 10px; } #ociqvanemt .gt_indent_3 { text-indent: 15px; } #ociqvanemt .gt_indent_4 { text-indent: 20px; } #ociqvanemt .gt_indent_5 { text-indent: 25px; } #ociqvanemt .katex-display { display: inline-flex !important; margin-bottom: 0.75em !important; } #ociqvanemt div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { height: 0px !important; } animal body brain log_body log_brain Sheep 55.50 175.00 4.02 5.16 Human 62.00 1,320.00 4.13 7.19 Y podemos usarlo también para interpretar la recta de referencia \\(y = 2.55 + 0.5 x\\) , para cambios chicos: Un incremento de 10% en masa total corresponde en un incremento de 5% en masa cerebral. El coeficiente de la regresión log-log, en nuestro ejemplo 0.5, es la elasticidad y es un concepto común en economía. Justificación Para entender la interpretación como cambio porcentual recordemos primero que la representación con series de Taylor de la función exponencial es: \\[e^x = \\sum_{n=0}^\\infty \\frac{x^n}{n!}\\] Más aún podemos tener una aproximación usando polinomios de Taylor, en el caso de la exponencial el \\(k\\)-ésimo polinomio de Taylor está dado por: \\[e^\\delta \\approx 1 + \\delta + \\frac{1}{2!}\\delta^2 + \\dots + \\frac{1}{k!}\\delta^k\\] y si \\(\\delta\\) es chica (digamos menor a 0.15), entonces la aproximación de primer grado es razonable y tenemos: \\[Ae^{\\delta} \\approx A(1+\\delta)\\] dat &lt;- tibble(delta = seq(0, 1, 0.01), exp_delta = exp(delta), uno_mas_delta = 1 + delta) ggplot(dat, aes(x = uno_mas_delta, y = exp_delta)) + geom_line() + geom_abline(color = &quot;red&quot;) + annotate(&quot;text&quot;, x = 1.20, y = 1.18, label = &quot;y = x&quot;, color = &quot;red&quot;, size = 6) "],["referencias.html", "Referencias", " Referencias "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/temario.html b/temario.html
index 37a7353..ada47e6 100644
--- a/temario.html
+++ b/temario.html
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>
diff --git a/tipos-de-estudio-y-experimentos.html b/tipos-de-estudio-y-experimentos.html
index 8e61054..ad85efb 100644
--- a/tipos-de-estudio-y-experimentos.html
+++ b/tipos-de-estudio-y-experimentos.html
@@ -412,7 +412,7 @@
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-varias-pruebas-independientes"><i class="fa fa-check"></i>Ejemplo: varias pruebas independientes</a></li>
 </ul></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#simulando-de-la-posterior"><i class="fa fa-check"></i>Simulando de la posterior</a></li>
-<li class="chapter" data-level="12.1" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i><b>12.1</b> Ejemplo de islas</a></li>
+<li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ejemplo-de-islas"><i class="fa fa-check"></i>Ejemplo de islas</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#por-qué-funciona-metrópolis"><i class="fa fa-check"></i>¿Por qué funciona Metrópolis?</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#método-de-metrópolis"><i class="fa fa-check"></i>Método de Metrópolis</a></li>
 <li class="chapter" data-level="" data-path="métodos-de-cadenas-de-markov-monte-carlo.html"><a href="métodos-de-cadenas-de-markov-monte-carlo.html#ajustando-el-tamaño-de-salto"><i class="fa fa-check"></i>Ajustando el tamaño de salto</a>